United States Office of Air Quality EPA-450/3-83-011a
Environmental Protection Planning and Standards April 1983
Agency Research Triangle Park NC 27711
Air
Inorganic Arsenic Draft
Emissions EIS
from Glass
Manufacturing
Plants -
Background
Information for
Proposed Standards
-------
EPA-450/3-83-011a
Inorganic Arsenic Emissions from
Glass Manufacturing Plants -
Background Information
for Proposed Standards
Emission Standards and Engineering Division
U.S ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
April 1983
-------
This report has been reviewed by the Emission Standards and Engineering Division of the Office of Air Quality Planning
and Standards, EPA, and approved for publication. Mention of trade names or commercial products is not intended to
constitute endorsement or recommendation for use. Copies of this report are available through the Library Services
Office (MD-35), U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711; or, for a fee, from
the National Technical Information Services, 5285 Port Royal Road, Springfield, Virginia 22161.
Publication No. EPA-450/3-83-011a
-------
ENVIRONMENTAL PROTECTION AGENCY
Background Information
and Draft Environmental Impact Statement
for Inorganic Arsenic Emissions from
Glass Manufacturing Plants
Prepared by:
f f
/Jack R. Farmer (Date)
Director, Emission Standards and Engineering Division
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
1. The proposed standards of performance would limit emissions of
inorganic arsenic from existing and new glass manufacturing plants.
The proposed standards implement Section 112 of the Clean Air Act
and are based on the Administrator's determination of June 5, 1980,
(44 FR 37886) that inorganic 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. States within the Northeast United States would be
particularly affected.
2. Copies of this document have been sent to the following Federal
Departments: Labor, Health and Human Services, Defense,
Transportation, Agriculture, Commerce, Interior, and Energy; the
National Science Foundation; the Council on Environmental Quality;
members of the State and Territorial Air Pollution Program
Administrators; the Association of Local Air Pollution Control
Officials; EPA Regional Administrators; and other interested parties.
3. The comment period for review of this document is 60 days. Mr. James U.
Crowder may be contacted regarding the date of the comment period.
4. For additional information contact:
Mr. Gene W. Smith
Standards Development Branch (MD-13)
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
Telephone: (919) 541-5624
5. Copies of this document may be obtained from:
U. S. EPA Library (MD-35)
Research Triangle Park, NC 27711
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
m
-------
TABLE OF CONTENTS
LIST OF TABLES v
LIST OF FIGURES ix
CHAPTER 1 - SUMMARY 1-1
1.1 Regulatory Alternatives 1-1
1.2 Environmental Impact 1-1
1.3 Economic Impact 1-3
CHAPTER 2 - THE GLASS MANUFACTURING INDUSTRY 2-1
2.1 Industry Description 2-1
2.1.1 Flat Glass Industry 2-2
2.1.2 Container Glass Industry 2-3
2.1.3 Pressed and Blown (N.E.C.) Industry 2-3
2.1.4 Wool Fiberglass Industry 2-5
2.2 Use of Arsenic in Glass Manufacturing 2-5
2.3 Glass Manufacturing Processes 2-8
2.3.1 Basic Process 2-8
2.3.1.1 Raw Material Handling and Mixing 2-8
2.3.1.2 The Melting Process 2-10
2.3.1.3 Forming and Finishing 2-14
2.4 Factors Affecting Arsenic Emissions from Glass Melting
Furnaces 2-14
2.4.1 Glass Type 2-15
2.4.2 Furnace Operation 2-15
2.5 Inorganic Arsenic, Emissions Occurring Under
Existing Regulations 2-16
2.5.1 Existing Regulations 2-16
2.5.2 Baseline Inorganic Arsenic Emissions and
Controls 2-19
2.6 References 2-23
CHAPTER 3 - EMISSION CONTROL TECHNIQUES 3-1
3.1 Process Modifications 3-1
3.1.1 Batch Formulation Alterations 3-1
3.1.2 Electric Boosting 3-2
3.1.3 All-Electric Melters 3-3
3.2 Add-on Control Techniques 3-4
3.2.1 Fabric Filters 3-4
3.2.2 Electrostatic Precipitators 3-7
3.2.3 Scrubber Systems 3-10
3.3 Effect of Gas Cooling on Arsenic Emissions Control. . . . 3-13
3.4 References 3-18
-------
TABLE OF CONTENTS (Continued)
CHAPTER 4 - MODEL FURNACES AND REGULATORY ALTERNATIVES 4-1
4.1 Model Furnaces 4-1
4.2 Regulatory Alternatives 4-3
4.2.1 Regulatory Alternative 1 4-5
4.2.2 Regulatory Alternative 2 4-5
4.2.3 Regulatory Alternative 3 4-5
4.3 References 4-6
CHAPTER 5 - ENVIRONMENTAL IMPACT 5-1
5.1 Air Pollution Impact 5-1
5.1.1 Model Furnace Emissions 5-1
5.1.2 Nationwide Emissions 5-3
5.2 Water Pollution Impact 5-3
5.3 Solid Waste Impact 5-3
5.4 Energy Impact 5-
7.5 References 5-6
CHAPTER 6 - COST ANALYSIS 6-1
6.1 Cost Analysis of Regulatory Alternatives for Model
Furnaces 6-1
6.1.1 Capital Costs 6-2
6.1.2 Annualized Costs 6-5
6.2 Nationwide Cost Impacts 6-11
6.3 Other Cost Considerations 6-11
6.3.1 Costs Associated With Monitoring 6-11
6.4 References 6-12
CHAPTER 7 - ECONOMIC IMPACT 7-1
7.1 Industry Economic Profile 7-1
7.1.1 Introduction 7-1
7.1.2 Machine-Made Pressed and Blown Consumer
Glassware 7-4
7.1.3 Handmade Pressed and Blown Glassware 7-9
7.1.4 Glass Tubing for Fluorescent and Neon Lighting . . 7-12
7.1.5 Incandescent Light Bulbs Blanks 7-13
7.1.6 Television Picture Tube Envelopes 7-14
7.1.7 Scientific and Technical Glass Tubing 7-15
7.1.8 Optical Glass 7-16
7.1.9 Glass Products Than Have Contained Arsenic .... 7-17
VI
-------
TABLE OF CONTENTS (Continued)
7.2 Economic Analysis 7-23
7.2.1 Introduction 7-23
7.2.2 Summary 7-23
7.2.3 Methodology 7-24
7.2.4 Results 7-25
7.3 Socio-Economic Impact Assessment 7-48
7.3.1 Executive Order 12291 7-48
7.3.2 Regulatory Flexibility 7-49
APPENDIX A A-l
APPENDIX B B-l
APPENDIX C C-l
APPENDIX D D-l
APPENDIX E E-l
VII
-------
LIST OF TABLES
Table Page
1-1 Environmental and Economic Impacts of Regulatory
Alternatives 1-2
2-1 Products Made by Furnaces That Emit Inorganic Arsenic. . . 2-5
2-2 Theoretical Arsenic Input/Retention Data 2-15
2-3 State Particulate Regulations for Existing Stationary
Sources 2-17
2-4 New Source Performance Standard for Particulate Emissions
from Glass Furnaces 2-19
2-5 Available Inorganic Arsenic Emissions Data for Existing
Arsenic Glass Plants 2-20
3-1 Effect of Electric Boosting on Arsenic Emissions from
Glass Furnaces 3-3
3-2 Performance of Fabric Filters on Arsenic Emissions from
Glass Furnaces 3-7
3-3 Performance of Electrostatic Precipitators on Arsenic
Emissions from Glass Furnaces 3-11
3-4 Arsenic Trioxide (Arsenolite) Vapor Pressure Data 3-14
4-1 Glass Manufacturing Model Furnace Parameters 4-2
4-2 Stack Parameters Used for Risk Analysis 4-4
4-3 Regulatory Alternatives for Control of Inorganic Arsenic
Emissions from Glass Manufacturing 4-5
5-1 Arsenic Emissions from Model Glass Furnaces Under the
Three Regulatory Alternatives 5-2
5-2 Estimated First Year Nationwide Arsenic Emissions from
Glass Furnaces Under Each Regulatory Alternative .... 5-4
6-1 Regulatory Alternatives for Controlling Arsenic Emissions
from Glass Manufacturing Furnaces 6-1
6-2 Bases for Capital Cost Estimates 6-3
vm
-------
LIST OF TABLES
Table Page
6-3 Component of Capital Costs for Electrostatic
Precipitators 6-4
6-4 Capital Cost Estimates for Model Furnaces 6-7
6-5 Bases for Annualized Costs 6-8
6-6 Annualized Costs for Model Furnaces 6-10
6-7 Nationwide First-Year Cost Impacts 6-11
7-1 Value of Shipments in the Glass Industry, 1980 7-2
7-2 Number of Companies and Employment in the Glass
Industry, 1980 7.3
7-3 Glass Manufacturing Plants Which Add Arsenic to Raw
Materials 7_5
7-4 Shipments of Consumer, Scientific, Technical, and
Industrial Glassware, 1970 to 1981 7-6
7-5 Shipments, Output and Prices of Machine-Made Consumer
Glassware, 1970 to 1981 7-8
7-6 Shipments, Output and Prices of Hand-Made Consumer
Glass, 1970 to 1981 7-11
7-7 Shipments, Output, and Prices of Flat Glass,
1970 to 1981 7_19
7-8 Shipments, Output, and Price of Containers,
1970 to 1981 7_22
7-9 Maximum Percent Price Increases 7-27
7-10 Financial Characteristics of Corning Glass Works,
1978 to 1981 7-3?
7-11 Financial Characteristics of RCA, 1978 to 1981 7-34
7-12 Financial Characteristics of Anchor Hocking Corporation,
1978 to 1981 7-35
-------
LIST OF TABLES
Table Page
7-13 Financial Characteristics of GTE, 1978 to 1981 7-36
7-14 Financial Characteristics of Owens-Illinois,
1978 to 1981 7-37
7-15 Financial Characteristics of General Electric,
1979 to 1981 7-38
7-16 Financial Characteristics of PPG, 1978 to 1981 7-39
7-17 Financial Characteristics of Libbey-Owens-Ford,
1978 to 1981 7-40
7-18 Summary of Financial Ratios for Profit Impacts 7-41
7-19 Profit Impacts: Percent Change in Before Tax Profits
on Sales - After Controls 7-42
7-20 Ratio of Long-Term Debt to Total Capitalization
(Pre-Control) 7-46
7-21 Ratio of Long-Term Debt to Total Capitalization
(Post Control) 7-47
7-22 Pressed and Blown Glass NEC Manufacturing Plants
in Four States 7-51
C-l Summary of Uncontrolled Arsenic Emissions (ESP) C-3
C-2 Summary of Controlled Arsenic Emissions (ESP) C-5
C-3 Summary of Inlet Arsenic Emissions (Fabric Filter). . . . C-7
C-4 Summary of Outlet Arsenic Emissions (Fabric Filter) . . . C-8
E-l Identification of Glass Manufacturing Plants E-14
E-2 Input Data to Exposure Model Glass Manufacturing E-15
E-3 Total Exposure and Number of People Exposed
(Glass Manufacturing Plants) E-16
E-4 Public Exposure for Glass Manufacturing Plants
as Produced the Human Exposure Model E-17
-------
Table
LIST OF TABLES
E-5 Maximum Lifetime Risk and Cancer Incidence for
Glass Manufacturing Plants (Assuming Baseline
Controls) E-22
-------
LIST OF FIGURES
Figure Pa9e
2-1 Process Flow Diagram 2-9
2-2 Typical Side-Port Furnace and End-Port Furnace 2-13
3-1 A Simple Two Cell Inside Out Baghouse Equipped for
Shake Cleaning 3-5
3-2 Conventional and Needle Type Electrostatic
Precipitators 3-8
3-3 Typical Scrubber System 3-12
3-4 Effect of Temperature on the Performance of Arsenic
Emissions Control Devices on Copper Smelters 3-16
6-1 Reported Installed Costs of Electrostatic Precipitator
Control Systems Compared With Estimated Cost Curve
Used in This Study 6-6
Xll
-------
1.0 SUMMARY
1.1 REGULATORY ALTERNATIVES
National emissions standards for hazardous air pollutants (NESHAP) for
control of inorganic arsenic emissions from glass manufacturing plants are
being developed under the authority of Section 112 of the Clean Air Act.
These standards would affect furnaces in the glass industry that melt
arsenic containing glasses.
Three regulatory alternatives were considered. Regulatory
Alternative 1 is the baseline alternative and represents the level of
control that would exist in the absence of any NESHAP regulations.
Regulatory Alternative 2 represents the control level achievable by a fabric
filter or an electrostatic precipitator. Regulatory Alternative 3 is the
most restrictive and would require zero emissions, that is complete
elimination of the use of arsenic in glass manufacturing.
1.2 ENVIRONMENTAL IMPACT
Although emissions data are not available for all arsenic emitting
furnaces in the glass manufacturing industry, available data indicate that
current annual arsenic emissions are at least about 37 Mg. This is the
emissions level that would exist under Regulatory Alternative 1. The
emissions level under Regulatory Alternatives 2 and 3 would be 4.3 Mg/year
and zero, respectively. This represents an industry-wide 88 percent
reduction for Alternative 2 and 100 percent reduction for Alternative 3.
There would not be any water pollution impact under any of the
regulatory alternatives. There will be a relatively small negative solid
waste impact under Regulatory Alternative 2. There will be a relatively
small increase in energy use under Regulatory Alternative 2 and a small
energy savings under Regulatory Alternative 3. A summary of the
environmental and energy impacts associated with the three regulatory
alternatives is shown in Table 1-1.
1-1
-------
TABLE 1-1. ENVIRONMENTAL AND ECONOMIC IMPACTS OF REGULATORY ALTERNATIVES
Administrative
Action
Regulatory
Alternative I
(No Action)
Regulatory
Alternative II
Regulatory
Alternative III
Air
Impact
0
+4**
+4**
Water
Impact
0
0
0
Solid
Waste Energy Noise
Impact Impact Impact
ooo
_!** _i** ~i**
+\** +1** 0
Economic
Impact
0
-I**
_4**
KEY: + Beneficial impact
- Adverse impact
0 No impact
1 Negligible impact
2 Small impact
3 Moderate impact
4 Large impact
* Short-term impact
** Long-term impact
*** Irreversible impact
-------
1.3 ECONOMIC IMPACT
Regulatory Alternative 2 is expected to result in a total industry
capital cost of about 30 million dollars and an annualized cost of about
5.4 million dollars. There would be no capital or annualized control costs
associated with Regulatory Alternative 3. However, there will be a negative
economic impact under this alternative due to elimination of the production
of arsenic-containing glasses and possibly the shut down of glass melting
furnaces that produce arsenic-containing glasses. A detailed cost analysis
is presented in Chapter 6 of this document. An economic analysis is
presented in Chapter 7. Table 1-1 shows the economic impacts associated
with the three regulatory alternatives.
1-3
-------
2.0 THE GLASS MANUFACTURING INDUSTRY
This chapter presents general information on the glass manufacturing
industry including a description of the glass producing processes and a
discussion of atmospheric emissions.
2.1 INDUSTRY DESCRIPTION
The glass manufacturing industry is classified in accordance with the
industry definitions embodied in the Standard Industrial Classification
(SIC) system. Under this system of classification, an industry is generally
defined as a group of establishments producing a single product or a more or
less closely related group of products. Accordingly, for the glass industry
there are four SIC codes:
SIC 3211 - Flat glass
SIC 3221 - Container glass
SIC 3229 - Pressed and blown glass, not elsewhere
classified (N.E.C.)
SIC 3296 - Wool fiberglass
The products produced by the glass industry are extremely diverse in
nature. Although the major basic manufacturing operations are common in the
various industry segments, the chemical composition and corresponding
properties of the products may vary significantly between products of the
various segments as well as between different products produced within a
given segment. The use of arsenic in the glass industry is mainly
concentrated in the manufacture of certain products in the pressed and
blown, n.e.c. segment. The flat glass and container glass segments have
used arsenic in the past but are now reported to have virtually eliminated
the use of the substance. There has not been any known use of arsenic in
the wool fiberglass segment. The following description of the various glass
industry segments, therefore, concentrates on the pressed and blown sector.
2-1
-------
2.1.1 Flat Glass Industry1
This industry contains establishments primarily engaged in
manufacturing flat glass as well as some laminated and tempered glass.
Almost 100 percent of the flat glass produced is soda/lime glass. The basic
ingredients of soda/lime flat glass are sand (SiO^), soda ash (Na2C03), and
limestone (CaC03 plus some MgC03). Typical composition of soda/lime
flat glass is as follows:
Si02 71 to 74%
A1203 0 to 2%
Na20 12 to 15%
CaO 8 to
The major products shipped by the flat glass industry are window glass,
plate and float glass, rolled and wire glass, tempered glass, and laminated
glass.
Four flat glass products -- float, sheet, rolled, and plate -- are
manufactured in the United States. Of these, float glass accounts for more
than 90 percent of the total flat glass production. Float glass is made by
floating molten gla.ss from the melting furnace on a bath of molten tin until
the glass hardens. This glass, with its high optical quality, has replaced
plate glass, which required grinding and polishing to produce a smooth
surface. It is used for automobile windows and large picture windows.
Average thickness ranges from 3.2 to 6.4 mm. Sheet glass is made by drawing
molten glass upward from the melt. It is thinner than float glass (1.6 to
3.2 mm) and is used for windows in residential construction. Rolled or
patterned glass is formed by drawing molten glass through rollers with
patterns impressed on them. This decorative glass is used for special
purposes such as shower doors and partitions. Plate glass is made by
drawing molten glass through smooth rollers and then grinding and polishing
both glass surfaces to a smooth finish.
The use of arsenic in the flat glass segment is reported to have been
eliminated in the past few years. Arsenic can potentially add some
desirable properties to some specialty flat glass products such as
2-2
-------
chemically strengthenable flat glass used in aircraft and spacecraft
windows and solar collectors. However, there are no flat glass plants
currently known to be using arsenic.
2.1.2 Container Glass Industry
Glass containers (SIC-3221) is the largest of the three major segments
of the glass industry. It includes the manufacture of narrow-neck and
wide-mouth glass containers for foods, beverages, medicines, toiletries, and
cosmetics. Three general types of container glass are produced: amber,
green, and clear.
Soda/lime is the major type of glass produced in the container glass
segment. The basic raw materials for soda/lime container glass are silica
sand, soda ash (Na3C03), and limestone (primarily CaC03, plus some MgCO,
in dolomitic limestones). Feldspathic minerals (anhydrous aluminosilicates
containing potassium, sodium, and calcium in varying ratios) are also
utilized as sources of alumina and alkali. Minor amounts of other oxides
exist as impurities, and additional minor ingredients are added for specific
purposes. A typical soda/lime glass-batch composition is:
Silica sand 55%
Soda ash 19%
Feldspar 7%
Limestone 18%
Salt cake (Na2S04) 0.5%
According to industry sources, the use of arsenic in the container
glass industry has been virtually eliminated. However, there is at least
one specialty container glass plants in the United States that is known
currently to use arsenic.
2.1.3 Pressed and Blown (N.E.C.) Industry5
The pressed and blown glassware industry, as represented by SIC-3229,
essentially includes all industrial establishments primarily engaged in
manufacturing glass and glassware that is pressed, blown, or shaped from
glass produced in the same establishment. It consists of every category of
2-3
-------
glass or glassware except flat glass (SIC-3211), glass containers (SIC-3221)
and wool fiberglass (SIC-3296). Establishments include those manufacturing
textile glass fibers; lighting, electronic, and technical ware; and machine-
made and handmade table, kitchen, and art-ware glass products. By far, the
major use of arsenic in the glass industry is in the pressed and blown glass
segment. Compositionally, the four important categories of glass
manufactured by the pressed and blown glass industry are as follows:
Estimated percent of
Glass category total production
Soda/lime 77
Borosilicate 11
Lead silicate 5
Opal 7
Soda/lime glasses are overwhelmingly the most important type of glass
in terms of variety of use as well as in tonnage melted. The combination of
silica sand, soda ash, and limestone produces a glass that is easily melted
and shaped and that has good chemical durability. Primary pressed and blown
products employing this type of glass are incandescent lamps, tubing, and
tableware.
Borosilicate glasses are basically a combination of silica sand, boric
oxide, and soda ash. The borosilicate glasses have excellent chemical
durability and electrical properties, and their low thermal expansion yields
a product with high resistance to thermal shock. These combined properties
make them ideal for demanding industrial and domestic uses such as chemical
laboratory ware, cookware, pharmaceutical ware, and some lens reflectors and
R R
lamp envelopes. Pyrex , produced by Corning Glass Works, and Kimax ,
produced by the Kimble Division of Owens-Illinois, Inc., are examples of
products made from borosilicate glasses.
Lead silicate glasses are composed of silica, lead oxide, and
significant amounts of alkali oxide. The lead glasses are characterized by
high electrical resistivity, high refractive index, and slow rate of
increase in viscosity with decreasing temperature. This viscosity
2-4
-------
characteristic makes them particularly well suited to hand fabrication.
Lead glasses are used in high-quality art and tableware, for special
electrical applications, optical glasses, fluorescent lamp envelopes, and
X-ray, gamma-ray, and neutron radiation shielding windows.
Opal glasses are translucent and may be colored. Commercial products
of opal glass include lighting globes, ointment jars, dinnerware, and wall
paneling. The translucency or opacity of opal glasses is produced by
multiple scattering of light inside the glass. This scattering is achieved
by the precipitation of crystals (or an immiscible amorphous phase) with an
index of refraction different from that of the base glass. Commercial opal
glasses commonly employ fluorine additions to yield opacifying crystals of
sodium or calcium fluoride.
2.1.4 Wool Fiberglass Industry
Wool fiberglass, the product of SIC-3296 segment, is used primarily as
building insulation, accoustical ceiling tiles, heating and cooling pipe and
duct insulation, and in process equipment and appliance insulation. There
are no known wool fiberglass plants that have used arsenic in the past or
are using the substance currently.
2.2 USE OF ARSENIC IN GLASS MANUFACTURING
Available information indicates that at least 15 glass plants are
currently using inorganic arsenic in glass products. Except for some
specialty container glasses, all the arsenic containing glass products made
by these plants are pressed and blown glass products. Table 2-1 lists the
arsenic containing glass products made by the 15 plants.
TABLE 2-1. PRODUCTS MADE BY FURNACES THAT EMIT INORGANIC ARSENIC
PRIMARY PRODUCTS
TV Picture Tube Components
Glass Tubing
Tableware Glass
Specialty Container Glass
Heat Resistant Globes
Electric Light Covers
Lead Glasses
Optical Glasses
Lead Crystal Glass
2-5
-------
In recent years, the use of arsenic in many glass products has been
either completely eliminated or has been reduced to the minimum amount that
is technically necessary. This was accomplished by isolated investigations
by individual companies to replace arsenic with other substances. Such
substitution must practically be determined individually for each type of
glass. These research efforts often exceed the means of small and medium
size glass companies and companies producing many different types of
glasses. Therefore, the development occurred in the portion of the glass
industry which almost exclusively produces only one kind of glass, i.e.,
soda/lime glass produced by the flat glass and container glass segments.
Substitution was possible for soda/lime glass and the use of arsenic in this
large area of glass production was eliminated by most companies.
Although arsenic is still being used in some soda/lime glass products
in the pressed and blown glass sector, the primary types of glass where
arsenic is used are lead, opal, lead silicate, borosilicate and
aluminosilicate-ceramic. Inorganic arsenic compounds are used in glass
manufacturing for a combination of reasons depending on the particular glass
being produced. In the majority of cases, arsenic compounds act as fining
or clarification agents. During the melting of the glass batch raw
materials, gaseous reaction products such as oxygen, nitrogen, and carbon
dioxide are evolved and rise through the glass melt in the form of bubbles.
These bubbles greatly reduce the overall quality of the glass. The addition
of the inorganic arsenic material causes the bubbles to rise more rapidly to
the melt surface and dissipate. It also appears that chemical reactions
brought about by the use of the arsenic reduce the release of some bubbles
p
caused by nitrogen and carbon dioxide. Fining agents act in various
ways. As a result of their ready conversion between various oxidation
stages, they can both take up and release oxygen. Arsenic is added to the
glass batch in the form of powdered arsenic trioxide (As?03) or liquid
arsenic acid (H-^AsO^). As^Og with its trivalent and pentavalent
oxidation stages is one of the classical fining agents. The overall formula
for this redox equilibrium is as follows:
As203 + 02 >- As2°5
2-6
-------
The initial formation of the pentavalent stage is necessary and is brought
about by other oxidizing agents, e.g., nitrates. (The use of arsenic acid
essentially changes the input form to the pentavalent stage.) This
pentavalent stage spontaneously releases the oxygen again at elevated
temperatures. As a result, the bubbles which form as a function of the
batch reaction become enlarged and rise more rapidly to the surface.
A second function of arsenic in glass is to act as a decolorizing
agent. The effectiveness of arsenic trioxide in this use is again based on
the ease of interconversion of the various oxidation stages. This
interconversion helps establish a concentration of arsenic pentoxide under
the equilibrium conditions of the melt. Arsenic pentoxide oxidizes divalent
iron impurities in the melt (which impart a greenish color to the glass) to
a trivalent iron, which results in a yellowish color glass. The
simultaneous addition of other elements such as nickel oxide, cobalt oxide,
and rare earth oxides provides balancing colors which produce a colorless
finished glass.
Inorganic arsenic is also used in some instances in special glass types
to impart particular properties that are needed for the end use of the
glass. For example, arsenic can provide stable fixation of certain colors
for optical glass by stabilizing selenium, provide high glass permeability
to infrared light for camera lenses, and provide a high degree of energy
I O
transmission for solar collector glass.
With the advent of environmental and occupational health laws for
inorganic arsenic in the late sixties and early seventies, glass companies
began reducing arsenic usage and initiating research into arsenic
substitutes. In 1968 approximately 3,900 Mg (4,300 tons) of arsenic
(elemental) were used, while in 1981 total usage was estimated at 730 Mg
(800 tons). ' Five percent of the total arsenic consumed in the United
States goes into glass production. Several manufacturers of television
picture tube funnel glass have totally eliminated inorganic arsenic
compounds from the feed batch raw materials.16'17 Another manufacturer
2-7
-------
reports that since 1978 arsenic usage was reduced 50 percent in a fluoride
18
opal glass and 30 percent in a borosilicate glass.
2.3 GLASS MANUFACTURING PROCESSES
Although there are numerous unit operations used in the manufacturing
of glass, most key processing steps, which generate the largest amounts of
atmospheric emissions, are common throughout the industry. In this
subsection, the basic operations are identified and discussed briefly.
2.3.1 Basic Process
Glass is manufactured in a high temperature conversion of raw materials
into a homogeneous melt capable of fabrication into useful articles. This
process can be broken down into three sub-processes: raw material handling
and mixing; melting; and forming and finishing. Figure 2-1 gives a typical
19
flow diagram for the manufacture of soda-lime glass; however, it has
general application to other commercial glass formulations.
The production of an arsenic glass involves melting a uniform mixture
of raw materials in a furnace to obtain a homogeneous mass. Typical
materials include sand, limestone, soda ash, feldspar, sodium sulfate and
nitrate, anhydrous borax, potassium carbonate, and arsenic trioxide or
arsenic acid. Arsenic compounds may be introduced into the batch as either
arsenic trioxide powder or liquid arsenic acid with no effect on the overall
glass making process. Inorganic arsenic is also introduced into the batch
feed as a constituent of the return cullet or scrap glass. The level of
inorganic arsenic in the cullet is the same as the percent retained in the
20
glass of the total batch fill.
2.3.1.1 Raw Material Handling and Mixing. The raw materials are
received in packages or in bulk and are unloaded by hand, vibrator-gravity,
drag shovels, or vacuum systems. Cullet, crushed recycled glass, must be
segregated and transferred to storage bins by various processes including
the utilization of bucket elevators, belt conveyors, or screw conveyors. In
addition to bulk raw materials, certain minor constituents are packaged and
stored in their original containers until mixed with the batch.
2-8
-------
Clan land
Si02 £ 991
to ywM S»0j
crushed, w*th*d
and screened
to — 20-100
mtth
t—- — — -"I^K— —— '
Soda. MgO.
ZnO. Bad. PbO.
etc andtnoMtor
hmng, o»di*mg.
COtefinj. and-
decotofiimg
Side-port
continuous tank.
looking down
through top
Submerged
throat in
bridgewal!
Packing, warerxxning.
and shipping
Figure 2-1. Process flow diagram.
2-9
-------
Prior to being fed into the melting unit, the raw materials are mixed
according to the desired product recipe. Weighing and mixing operations may
be automated or carried out by hand depending on the size or specialty of
the operation. The melters themselves are charged manually or automati-
cally, usually through screw or reciprocating-type feeders.
A potential source of inorganic arsenic emissions in the raw materials
handling part of the glass plant is fugitive emissions from arsenic trioxide
handling. Fugitive arsenic dust emissions from raw materials handling are
highly controlled due to OSHA regulations for airborne inorganic arsenic in
the workplace. The glass industry uniformly controls these fugitive arsenic
dust emissions by enclosing the unloading, conveying, and storage areas and
21 22
venting them to fabric filters. ' This type of control method is
effective in lowering arsenic emissions to negligible levels.
The OSHA regulations are the primary reason that some glass companies
have switched from using powdered arsenic trioxide to liquid arsenic acid.
By using the liquid arsenic acid as a batch raw material, a minimal amount
released into the workplace. One manufacturer has indicated that it now
uses only liquid arsenic acid for their glass manufacturing in order to
23
protect their workers and comply with the OSHA standard. Because this
source of potential arsenic emissions is already well controlled and the
arsenic emissions are orders of magnitude less than controlled process
arsenic emissions, it will not be discussed further in this chapter.
2.3.1.2 The Melting Process. From the handling area, the weighed
raw materials are delivered into the furnace where they are transformed
through a sequence of chemical reactions into glass. In operation, the raw
materials float on the bed of molten glass until they dissolve. Mixing in
the molten glass bed is caused by gases evolved in chemical reactions and by
natural convection currents in the molten glass bed. In addition, some
furnaces have air injected in the bottom of the bed to augment ebullient
mixing.
Within the temperature range of the furnace (1,500°C to 1,700°C), the
glass exists as a liquid free of crystalline matter with a viscosity of
2
10 Newton-seconds per square meter (N-s/m ) (^ 100 poise). Because the
2-10
-------
viscosity of the glass exiting the furnace must be compatible with the
forming operations, the temperature of the molten bed is decreased gradually
to a point until the viscosity of the glass is about 100 to 1,000 N-s/m2
(10 -10 poise). In addition to cooling as it flows through the
furnace, the glass is retained in the furnace long enough for gaseous
inclusions to be removed. From the introduction of raw materials to the
extraction of a homogeneous melt suitably ready for forming, a furnace
accomplishes three functions in glassmaking: to bring raw materials
together to react; to hold the molten glass until it is free of bubbles and
inclusions; and to condition the glass for forming.
Energy required for melting glass is supplied by burning either natural
gas or fuel oil and sometimes by augmenting the energy produced from these
fossil fuels with electricity which is converted to heat within the liquid
glass bed. These three energy sources provide over 99 percent of the energy
consumed in the industry. Consumption of energy in the furnace constitutes
the major comsumption of energy in the glass industry. Natural gas has been
the preferred primary fuel with fuel oil used when natural gas supplies are
curtailed. Coal is not presently burned or gasified to provide energy for
the glass industry.
There are three types of fossil fuel-fired melting units used in the
glass industry: day pots, day tanks, and continuous tanks. Typically, day
pots are used where other larger tanks are not economically justifiable
because of limited production of special compositions of glass. The range
of the capacities of day pots varies from 9 kilograms (20 pounds) to
1,800 kilograms (2 tons) with these quantities melted in 24-hour batches.
Although the capacity of a typical day tank is slightly larger than that of
a typical day pot, the primary distinction between the two units is the
material of construction of the vessel walls.
Although day pots and day tanks are used to produce glass, most glass
tonnage is melted in larger capacity, continuously operating regenerative or
recuperative furnaces. Generally, regenerative furnaces maintain a larger
production rate than recuperative furnaces. These types of furnaces differ
2-11
-------
in the types and modes of operation of the heat exchangers used to recover
heat from the furnace exhaust gases.
Regenerative furnaces utilize two chambers of refractory called
checker-works in the following manner. At any one time, while combustion
flue gases heat the refractory in one checkerwork chamber, the other
checkerwork preheats combustion air; then after intervals ranging from 10 to
30 minutes, this gas flow is diverted so that combustion air is drawn
through the chamber previously heated by flue gases. The flue gases heat
the refractory in the other chamber previously used to preheat combustion
air. Regenerative furnaces, themselves, are divided into two functional
categories - side-fired and end-fired furnaces depending on the furnace
flame firing pattern. Figure 2-2 illustrates a typical side-fired furnace
and an end-fired furnace.
Recuperative furnaces employ one continuously operating shell and tube
type heat exchanger to preheat combustion air instead of the checkerwork
heat exchangers used in regenerative furnaces.
Regenerative and recuperative furnaces, which are fossil fuel-fired,
are being modified and in some cases replaced by systems using electrical
current to melt glass. Electric boosting is the term applied to the furnace
modification in which an electric current is used to augment furnace firing
of gas or oil. Electrical energy is converted to heat because of the high
electrical resistance of the molten glass. The effect of electric boosting
is to decrease the required furnace bridgewall temperature, which in turn
decreases the fuel consumption rate, thereby decreasing the pollutant
25
emission rate.
In some limited applications, more traditional regenerative furnaces
have been totally replaced by systems known as all-electric melters.
All-electric melters produce less than 10 percent of the glass in the United
?fi
States and none of the units produce more than 136 Mg (150 tons)/day.
The surface of the melter in a cold top all-electric furnace is maintained
at ambient temperature, and fresh batch raw materials are continuously fed
over the entire surface. Because energy is supplied internally to the
glass, a higher percentage of the total energy expended can be converted
2-12
-------
END-FIRED BOX TYPE REGENERATIVE
GLASS FURNACE
SIDE-FIRED BOX TYPE REGENERATIVE
GLASS FURNACE
Figure 2-2. Typical side-port furnace and end-port furnace.
2-13
-------
27
into usable heat for melting than with fossil fuel-fired melters.
Pollutant emissions from all-electric melters are virtually
nil.28'29'30'31 The main limitation with these systems is that they
cannot be used to produce all varieties of glass. Not all glasses posess
the electrical properties required for these melters and some glass
formulations actually corrode the electrodes presently used in the
32
all-electric melters.
The melting process is the major source of arsenic emissions in the
glass industry. However, the majority of the input inorganic arsenic is
permanently fixed in the molten glass and is not emitted from the furnace.
The remaining arsenic material is evaporated from the furnace melt and is
either eventually emitted as arsenic vapor or it condenses as submicron
^30 *5/l OC
arsenic particles or on other particulate matter. ' ' The majority of
the arsenic particulate matter is entrained in the exhaust flue gas,
however, a small portion can be deposited as particle fallout in the
og 37
refractory checkerwork of a regenerative furnace. ' The extent to
which arsenic may be removed in the checkerwork has not been quantified.
Recuperative furnaces would, of course, not have a similar arsenic removal
capability.
38
2.3.1.3 Forming and Finishing. In the forming and finishing step,
the molten glass is extracted from the furnace, shaped to the desired form,
and then annealed at high temperature. The final product is then either
inspected and shipped or sent for further finishing such as tempering or
decorating.
In practice, the molten glass, while at a yellow-orange temperature, is
drawn quickly from the furnace and worked in forming machines by a variety
of methods: pressing; blowing in molds; and drawing, rolling, and casting.
Immediately, this formed glass is conveyed to continuous annealing ovens to
remove internal stresses in the glass by controlled cooling.
The forming and finishing process is not a source of arsenic emissions.
2.4 FACTORS AFFECTING ARSENIC EMISSIONS FROM GLASS MELTING FURNACES
The two major factors affecting arsenic emissions from glass furnaces
include (1) the glass type and (2) furnace operation.
2-14
-------
2.4.1 Glass Type
The most important factor affecting arsenic emissions in glass plants
is the type of glass being produced. In general, the amount of arsenic
input to the glass batch will depend on the specific glass type. In
addition, different glass types have different arsenic retention properties.
One company has provided data on theoretical arsenic input and retention
amounts for the various types of glasses it produces.39 The company also
indicated that its glasses are very similar in physical and chemical
properties to the glasses produced by its competitors. Based on this
information, the theoretical arsenic input and retention data for the major
types of glasses using arsenic in the pressed and blown segment would be as
shown in Table 2-2.
TABLE 2-2. THEORETICAL ARSENIC INPUT/RETENTION DATA
TheoreticalTheoretical ArsenicTheoretical
Arsenic Input, Retained in Batch, Arsenic Lost.,
Glass Type kg/Mg Filled0 kg/Mg Filled3 kg/Mg Filled0
Lead
Lead Silicate
Fluoride Opal
Alumino Silicate
Borosilicate
1.9
1.0
3.8
6.3
6.8
1.8
0.7
3.7
5.7
6.2
0.1
0.3
0.1
0.6
0.6
Amount of glass pulled = 0.85 to 0.93 depending upon the glass type.
Amount of glass filled
Not all theoretical arsenic lost may go out of the stack. Some may be
recovered in the slag.
As seen from Table 2-2, percent retention of arsenic ranges from 70 to
99 percent.
2.4.2 Furnace Operation
Another important factor affecting arsenic emissions is the furnace
operation. Furnace operation primarily affects particulate emissions from
glass furnaces which may impact arsenic emissions. The key operating
2-15
-------
parameters affecting emissions are the furnace (or bridgewall) temperature,
the amount of cullet in the batch, the surface area of the molten glass bed,
the production (or pull rate) of glass, and the type of fuel being burned.
Of these operating parameters, temperature is the most important.
Increasing the temperature over the melt vaporizes more of the volatile
materials than at lower temperatures. Maintaining high temperature requires
more fuel to be consumed and, therefore, increases the level of pollutants
derived from fossil fuels.
Other parameters influence pollutant emission levels by changing the
temperature required to maintain production. For example, increasing the
cullet proportions in the raw batch lowers bridgewall temperature, thereby,
lowering emissions. The amount of surface area of molten glass exposed to
combustion gases has been shown to affect particulate emissions. With all
other parameters constant, a larger exposed area generates more particulate
40
than a smaller area.
For a furnace producing a single type of glass, increasing the pull
rate requires more energy, which if supplied by the combustion of fossil
fuels, causes an increase in furnace temperature with a corresponding
increase in emissions. As the pull rate increases, the emissions increase
at a decreasing rate. In the limiting case of no pull rate, data have been
published which show that particulates are still emitted from the molten
glass bed.41 For a soda/lime glass, the emission levels at zero pull rate
were found to be roughly 20 percent of those at the normal pull rate with
both measurements being taken at the same temperature.
2.5 INORGANIC ARSENIC EMISSIONS OCCURRING UNDER EXISTING REGULATIONS
2.5.1 Existing Regulations
There are currently no regulations at the Federal or State level for
arsenic emissions from glass plants. Atmospheric inorganic arsenic
emissions from glass plants are presently being controlled indirectly as a
result of State and Federal particulate matter regulations. Table 2-3 lists
particulate compliance limits for various glass production rates as allowed
by states in which most of the glass manufacturing facilities are located.
2-16
-------
TABLE 2-3. STATE PARTICIPATE REGULATIONS FOR EXISTING STATIONARY SOURCES
PO
Pull
tote
Tons/My
24
50
75
100
150
200
250
300
Pull
tott
Tons/Mr
1
2.1
3.1
4.2
6.3
8.3
10.4
12.5
Process
Weight
Tons /Mr
1.2
2.5
3.7
4.9
7.4
9.8
12.3
14.3
Indiana, Ohio
Oklahoma, and
Illinois
kg/hr
2.1
3.4
4.5
5.4
7.1
8.6
10.0
11.3
(1b/nr)
( «•«)
( 7.5)
(9.8)
(H.9)
(15.6)
(18.9)
(22.0)
(24.8)
Texas
kg/hr
1.7
3.4
5.1
6.8
10.1
13.4
16.7
20.0
(Ib/hr)
( 3.7)
( 7.5)
(11.2)
(14.9)
(22.3)
(29.5)
(36.8)
(44.0)
California
South
Coast Air Qua-
lity Manaoe-
•ent District
kg/hr
1.8
»2.5
3.1
3.6
4.3
5.0
5.4
5.8
(Ib/hr)
(<-0)
( 5-5)
( 6.7)
(8.4)
(9.5)
(10.9)
(11.8)
02.8)
New Jersey
kg/hr
2.5
2.8
3.0
3.3
3.8
4.3
4.8
5.3
(Ib/hr)
( 5.5)
( 6-1)
( 6.6)
(7.2)
(8-3)
( 9.4)
(10.5)
(11.6)
New York
kg/hr
1.9
3.1
4.1
4.9
6.4
7.8
9.0
10.2
(Ib/hr)
(«.2)
( 6.8)
( 8-9)
(10.8)
(M.2)
(17.2)
(19.9)
(22.5)
West Virginia
kg/hr
1.1
2.2
3.4
3.8
5.8
7.2
7.5
8.6
(Ib/hr)
(2.«)
( 4-9)
( 7.4)
(8.3)
(12.B)
(15.8)
(16.5)
(19.0)
-------
The table entries are calculated for existing Container Glass furnaces
assuming that 85 percent of process weight rate is transformed into glass
production, which corresponds to the normal 15 to 20 percent cullet
usage.
The limits of Illinois, Indiana, New York, Ohio, Oklahoma, and Texas
are formulated on a mass basis with an exponential dependence on process
weight. California limits are represented by the allowable particulate
emissions of the South Coast Air Quality Management District. The
New Jersey regulation for glass plants is also determined on a mass basis
but with a linear dependence on process weight and with allowances made for
increased cullet utilization. The New Jersey Department of Environmental
Protection makes an exception to the process weight limits for the case of
furnaces producing lead glass where a compliance schedule of 0.02 Gr/SCF
applies. West Virginia compliance regulations are interpolated from a table
based on process weight. Pennsylvania maintains a concentration basis
regulation for flat and container glass and mass basis regulation for
pressed, blown, or spun glass.
There is also a new source performance standard (NSPS) for control of
particulate emissions from glass melting furnaces. The NSPS emission limits
are shown in Table 2-4. These limits are met by installation of add-on
control devices such as fabric filters and electrostatic precipitators. The
State particulate matter regulations are met by installing an add-on control
device, installing an electric boosting process modification, modifying
batch formulations, or collecting particulate matter as fallout in the
furnace checkerwork. Of these particulate control methods, the use of
add-on control devices does the most effective job of reducing arsenic
emissions, while the checkerwork fallout technique does the least effective.
2-18
-------
TABLE 2-4. NEW SOURCE PERFORMANCE STANDARD FOR
PARTICULATE EMISSIONS FROM GLASS FURNACES
Emission Limit, g of Participate/
kg of Glass Produced
Furnace Furnace
Glass Manufacturing Fired With Fired With
Plant Industry Sector Gaseous Fuel Liquid Fuel
Container glass 0.100 0.130
(a) Borosilicate recipes
(b) Soda/lime and lead recipes
(c) Other than borosilicate, soda/lime,
and lead recipes (including opal,
fluoride, and other recipes) ,
Wool fiberglass
Flat glass
0.500
0.100
0.250
0.250
0.225
0.650
0.130
0.325
0.325
0.225
2.5.2 Baseline Inorganic Arsenic Emissions and Controls
The baseline arsenic emissions are estimated from the available arsenic
emissions data for glass furnaces. These data do not represent emissions
from all arsenic emitting furnaces in the nation and, therefore, understate
the baseline emissions. However, the furnaces not included in the data base
are expected to be small furnaces, mainly in the hand pressed and blown
glass sector, with relatively low annual arsenic emissions. The baseline
emission estimates, therefore, account for most of the total arsenic
emissions from glass furnaces. The available estimates of hourly and annual
inorganic arsenic emissions from existing furnaces known to produce
arsenic-containing glasses are given in Table 2-5. These emissions data are
from 32 furnaces at 15 plants. All 32 furnaces meet the applicable state
participate limits. Of the 32 furnaces, 13 have add-on control devices.
These include two fabric filters and 11 electrostatic precipitators. Two of
the 19 furnaces without add-on control devices have electric boosting. The
majority of the arsenic emission estimates were provided by the companies
involved and are based on actual measurements in some cases and on material
2-19
-------
TABLE 2-5. AVAILABLE INORGANIC ARSENIC EMISSIONS DATA
FOR EXISTING ARSENIC-USING GLASS PLANTS3
Plant Furnace
1 A
B
2 ' A
3 A
4 A
B
C
D
E
5 A
B
6 A
B
C
7 A
8 A
9 A
10 A )
B 1
c !
D ,
11 A
12 A (2
B (2
C
Existing PM Controls that
Reduce Arsenic Emissions
ESP preceded by evaporative cooler
ES? preceded by evaporative cooler
ESP
ESP
None
None
None
None
None
None
None
None
None
None
FF
ESP
EB
i Total
f from None
4 furnaces
1
ESP preceded by an evaporative
cooler
stacks) None
stacks) None
ESP preceded by evaporative cooler
Baseline Arsenic Emissions
kg/hr
0.0031
0.0007
0.0082
0.011
0.37
0.42
0.09
0.06
0.06
0.15
0.11
0.05
0.22
0.37
0.023
0.009
1.81
0.0045
0.0027
0.227
0.273
0.018
Ib/hr
(0.0069)
(0.0015)
(0.018)
(Q.025)
(0.81)
(0.93)
(0.19)
(0.14)
(0.14)
(0.33)
(0.24)
(0.12)
(0.48)
(0.81)
(0.051)
(0.02)
(4.0)
(0.01)
(0.006)
(0.5)
(0.6)
(0.02)
Mg/yr
0.027
0.006
0.069
0.089
3.09
3.53
0.73
0.55
0.55
1.27
0.91
0.45
1.82
3.09
0.19
0.076
15.20
0.038
0.023
1.99
1.83
0.073
(tons/yr)
(0.03)
(0.0066)
(0.076)
(0.101)
(3.4)
(.3.9)
(0.8)
(0.6)
(0.6)
(1.4)
(1-0)
(0.5)
(2.0)
(3.4)
(0.21)d
(0.084)
(16.8)d
(0.042)d
(0.025)b
(2.19)
(2.01)
(0.08)
and a cyclone
2-20
-------
TABLE 2-5. (Continued)
Baseline Arsenic Emissions3
FvUt-inr| DM rnnt-^s that
Plant Furnace Reduce Arsenic Emissions kg/hr Ib/hr
13 A "1
B 1 Common ESP preceded by evaporative 0.005 (0.01)
f Stack cooler
C
D •) ESP preceded by evaporative 0.0227 (0.05)
cooler
14 A EB ' 0.091 (0.2)
8 FF 0.004 (0.01)
15 A None 0.014 (0.03)
Total
Mg/yr
0.038
0.191
0.76
0.038
0.118
36.70
(tons/yr)
(0.042)
(0.21)
(0.84)
(0.042)
(0.126)d
(40.50)
bFF = fabric filter
ESP = electrostatic precipitator
EB = electric boosting
PM = particulate matter
Emissions expressed as total elemental arsenic.
For furnaces where no information was available on the exact hours of operation, 8,400 hr/yr was assumed.
2-21
-------
balance calculations in others. Several of these furnaces produce many
different types of glasses, only some of which contain arsenic. The annual
emissions shown for these furnaces take into account the production
schedules of arsenic-containing glasses. Total existing arsenic emissions
from the 32 furnaces are about 37 Mg (41 tons)/yr.
2-22
-------
2.6 REFERENCES
1. Spinosa, E. D., et al. (Battelle Columbus Laboratories.) Summary
Report on Emissions from the Glass Manufacturing Industry. (Prepared
for the U. S. Environmental Protection Agency.) Cincinnati, Ohio.
Publication No. EPA-600/2-79-101. April 1979. 49 p.
2. Bauer, R. J. Arsenic: Glass Industry Requirements. In: Arsenic:
Proceedings of the Arsenic Symposium, Lederer, W. H. and
Fensterheim, R. J. (ed.) Gaithersberg, Maryland, Van Nostrand Reinhold
Company. 1983. p. 47.
3. Peters, A. The Use of Arsenic with Particular Reference to the Glass
Industry. Glastechn. Ber. (Mainz, Germany). J50(12):328-335. 1977.
4. Reference 1, p. 7.
5. Reference 1, pp. 10-16.
6. U. S. Environmental Protection Agency. Glass Manufacturing Plants
Background Information: Proposed Standards of Performance. Research
Triangle Park, N. C. Publication No. EPA-450/3-79-005a. June 1979.
p. 8-20.
7. Reference 3.
8. Reference 3.
9. Reference 3.
10. Reference 3.
11. Reference 3.
12. Reference 3.
13. Crose, P. (Acurex Corporation.) Arsenic: An Environmental Materials
Balance- Draft Final Report. (Prepared for U. S. Environmental
Protection Agency.) Washington, D. C. EPA Contract No. 68-01-6017
March 1981. pp. 3-3, 3-14.
14. PEDCo Environmental, Inc. Extended Source Survey Report for Arsenic.
(Prepared for U. S. -Environmental Protection Agency.) Research
Triangle Park, N. C. EPA Contract No. 68-02-3173. May 1982. pp. 3-31
\r \J O "* O O •
15. Reference 6. p. 3-10.
2-23
-------
16. Letter from Cherill, J., Corning Glass Works to Hellwig, G. V., PEDCo
Environmental. June 1, 1981. 2 p. Information on Coming's glass
composition changes since 1978.
17. Telecon. Armstrong, F., RCA Corporation, with Brooks, G., Radian
Corporation. February 19, 1982. Conversation concerning arsenic usage
by RCA.
18. Reference 16.
19. Reference 6. p. 3-5.
20. Letter from Mosely, G., Corning Glass Works to O'Connor, J., EPA:MDAD.
August 28, 1978. 4 p. Information on arsenic in glass and arsenic
emissions from glass melting units.
21. Reference 6. p. 3-10.
22. Schorr, J. R., et al. (Battelle-Columbus Laboratories.) Source
Assessment: Pressed and Blown Glass Manufacturing Plants. (Prepared
for U. S. Environmental Protection Agency.) Research Triangle Park,
N. C. Publication No. EPA-600/2-77-005. January 1977. p. 34.
23. Reference 22.
24. Reference 6. pp. 3-6 to 3-8.
25. Reference 6. p. 4-4.
26. Reference 22.
27. Reference 22.
28. Reference 22.
29. Memo from Cuffe, S., EPA-.ESED, to O'Connor, J., EPA:MDAD. July 17,
1978. 2 p. Emissions of arsenic from glass melting furnaces.
30. Memo from Herring, W., EPA, to Cuffe, S., EPArESED. December 27, 1978.
2 p. Arsenic emissions from glass manufacturing plants.
31. Telecon. Cherill, J., Corning Glass Works, with Brooks, G., Radian
Corporation. January 29, 1982. Conversation about Coming's use of
arsenic.
32. Reference 6. p. 4-6 to 4-7.
33. Reference 3.
2-24
-------
34. Reference 6. p. 3-16.
35. Suta, B. E. (SRI International.) Human Exposures to Atmospheric
Arsenic. (Prepared for U. S. Environmental Protection Agency.)
Research Triangle Park, N. C. EPA Contract Nos. 68-01-4314 and
68-02-2835. May 1980. p. 97.
36. Reference 20.
37. Letter and attachments from Swander, T., RCA Corporation, to
Goodwin, D., EPAtESED. September 27, 1978. 8 p. Information on
arsenic in glass and arsenic emissions from glass manufacturing units.
38. Reference 6, pp. 3-8 to 3-10.
39. Reference 20.
40. Reference 6, pp. 3-17 to 3-18.
41. Reference 22, pp. 46 - 48.
42. Reference 6, p. 3-23.
43. Memo from Shareef, S. A., Radian Corporation, to file. April 7, 1983.
1 p. Emissions data on glass furnaces at RCA Corporation.
44. Letter from Goebel, G., Kentucky Bureau of Environmental Protection, to
Brooks, G., Radian Corporation. February 11, 1982. 2 p. Information
on inorganic arsenic emissions from Corning Glass Works plant in
Danvile.
45. Telecon. Iden, C., Owens-Illinois, with Brooks, G., Radian
Corporation. March 3, 1982. Conversation about Owens-Illinois arsenic
emissions.
46. Letter and attachments from Murray, D. E., Anchor Hocking Corporation,
to Shareef, S. A., Radian Corporation. March 14, 1983. 5 p.
Information on inorganic arsenic emissions from Anchor Hocking glass
plants. (Project Confidential Files.)
47. Letter and attachments from Cherill, J., Corning Glass Works, to
Brooks, G., Radian Corporation. April 7, 1982. 19 p. Information on
inorganic arsenic emissions from Corning Glass Works plants. (Project
Confidential Files.)
2-25
-------
3.0 EMISSION CONTROL TECHNIQUES
As discussed in Chapter 2 of this document, the arsenic material that
does not get permanently fixed in the molten glass or in the slag, is
evaporated from the furnace melt. The arsenic is either eventually emitted
as vapor, as condensed submicron arsenic particles, or as arsenic condensed
on other particulate matter. Control techniques that reduce particulate
emissions would, therefore, reduce emissions of arsenic. The overall
reduction in arsenic emissions would depend on the effectiveness of the
particulate control device as well as the relative quantities of vapor phase
and solid phase arsenic in the flue gas stream at the inlet to the
particulate control device. The two major methods applicable to particulate
and, therefore, arsenic control in the glass manufacturing industry are
(1) process modifications and (2) add-on control techniques.
3.1 PROCESS MODIFICATIONS
*
Process modifications lower glass melting furnace emissions either by
altering raw material recipes or by modifying furnace equipment. These
modifications are discussed below.
3.1.1 Batch Formulation Alterations
The glass industry has greatly reduced, in the past years, the amount
of arsenic used in glass products. The flat glass industry reports that the
use of arsenic has been eliminated completely in this segment of the
12345
industry. ' ' ' ' Arsenic use has also been almost eliminated in the
container glass industry. However, at present there is at least one
speciality container glass plant known to be using arsenic. One source
reports that the various oxidation states of sulfur make it the most
commonly used fining agent for soda-lime glasses which dominate the flat and
container glass segments.
3-1
-------
The pressed and blown glass industry is the largest user of arsenic in
the glass industry. However, arsenic use by this segment of the industry
has also been completely eliminated for some glass products. In many other
products arsenic input is reported to have already been reduced to the
extent possible. The most commonly used substitute is antimony. As an
oxygen generator, it is a very effective fining agent for lead glasses at
temperatures below 1400°C (2550°F). Antimony is used in some glasses as
antimony trioxide, and in most commercial high volume production as sodium
antimonate. The use of arsenic and antimony together is considered to be
more effective than when either is used alone. However, antimony is not
considered to be an effective fining agent in many of the opal and ceramic
o
glasses either alone or in combination with other agents.
The halogens, especially chloride and fluoride, are used in many
glasses as fining agents. Chloride fining, usually as sodium and potassium
salts, is used in the hard borosilicate glasses. However, it does not
control redox and can, with water and alkali-boron balances in the process,
9
create particulate air pollution problems. Fluoride can also cause air
pollution and is also ineffective at high temperatures.
Other metal oxides, especially tin when used with antimony oxides, have
shown some desirable attributes in fining and redox control with some
glasses. However, this combination has not yet been shown to produce an
equivalent product to that produced using arsenic and is not considered a
replacement.
No data are available to quantify the reduction in arsenic emissions
that can be achieved due to batch formulation alterations. However, the use
of arsenic in the glass industry decreased by more than 80 percent between
1968 and 1981.11>12
3.1.2 Electric Boosting
Electric boosting is the term applied to the technique of augmenting
melting in a fossil-fuel-furnace by dissipating electrical current through
the molten glass. Electrical energy is converted to heat because of the
high electrical resistance of the molten glass. For a fixed furnace
throughput, utilizing electric boosting decreases the required bridgewall
3-2
-------
temperature, decreasing the fuel consumption rate, and thereby decreasing
both particulate and gaseous pollutant levels. Boosting has normally been
used to increase production rate since it does not require substantial
modifications of the furnace. However, as mentioned in Chapter 2, not all
glasses possess the electrical properties required for these melters and the
use of electric boosting is limited.
There are presently two furnaces known to be producing
arsenic-containing glasses that use electric boosting. Emissions data
provided by the plants indicate that arsenic emissions were reduced by 45
13
and 60 percent by the use of electric boosting. Table 3-1 shows the
uncontrolled and controlled emission rates for the two furnaces using
electric boosting.
TABLE 3-1. EFFECT OF ELECTRIC BOOSTING ON ARSENIC EMISSIONS
FROM GLASS FURNACES
UNCONTROLLED ARSENIC CONTROLLED ARSENIC
EMISSIONS EMISSIONS
PLANT FURNACE kg/hr (Ib/hr) kg/hr (Ib/hr)
9 A 4.55 (10.0) 1.81 (4.0)
14 A 0.17 (0.36) 0.091 (0.2)
PERCENT
REDUCTION
60
45
3.1.3 All-Electric Melters14
In contrast to conventional fuel-fired furnaces, the surface of the
melter in a cold top all-electric furnace is maintained at ambient
temperature, and fresh raw batch materials are fed continuously over the
entire surface. As molten glass is withdrawn from the melter, raw batch
drops in the melter, gradually heats up, and finally reacts in the liquid
phase. This processing minimizes losses from vaporization. The gases
discharged through the batch crust consist of carbon dioxide and water
vapor.
3-3
-------
Design objectives for all-electric melters have not been based
primarily on emission control, but rather on efficient melting and product
control. Construction is less expensive than that for fossil fuel furnaces
since there are no regenerator chambers, port necks, checkers, flues, or
reversing valves, and stacks can be eliminated. Additionally, there is no
need for ductwork, combustion blowers, fans, extra piping, burners, or
special refractory shapes.
Accomplishment of design objectives results in a low surface
temperature and a finer control on the glass melt formulation and,
therefore, relatively low levels of emissions. Presently no data are
available for arsenic emissions from all-electric glass furnaces. As with
electric boosting the potential for all electric melting is limited by the
required electrical properties of the glass.
3.2 ADD-ON CONTROL TECHNIQUES
Add-on control techniques applicable to glass furnaces include fabric
filters (FF), electrostatic precipitators (ESP), and scrubber systems. Of
the 32 arsenic emitting furnaces listed in Table 2-5, 11 are equipped with
electrostatic precipitators and two have fabric filters. The three add-on
control techniques are discussed below.
3.2.1 Fabric Filters15
Several glass manufacturing facilities utilize fabric filter systems to
collect particulates in the glass melting furnace exhaust. In these
systems, the furnace exhaust is first cooled and then passed through a
fabric filter which retains particulate, and allows the gases to vent to the
atmosphere. The physical characteristics of the filtering fabrics and the
agglomerating tendency of submicron particles have made the fabric filter
systems viable control techniques for the collection of glass melting
furnace particulates.
Figure 3-1 illustrates a typical baghouse system. In operation, a fan
pulls the furnace gases through devices which cool the gases to a
temperature compatible with the filter material. Cooling is accomplished by
duct cooling, dilution air addition, or water injection. The gases are then
3-4
-------
s-c
•6uLuee[D a^eqs aoj. paddinba
9snoi|6eq }no apisui [[ao OM^ a[duiLS \/ '!-
1.3 INI
1J.SIC1
-------
forced through the filter bags. Periodic cleaning of the bags is necessary
to maintain high collection efficiencies. Filter bags are cleaned through
shaking or reverse air pulsations. Conveyors transfer the collected dusts
to hoppers for disposal.
Fabric filter systems are claimed to have the advantages of: high
collection efficiency; low pressure drop across the system; and, low energy
requirements. Collection efficiencies are not affected by the electrical
resistivity of the particles. Bag life is up to 2 years depending on the
bag construction material.
There are certain disadvantages to the application of fabric filters to
glass melting furnace gases. The temperature of gases entering the fabric
filter must be below a maximum value to minimize thermal degradation and
prevent melting of the fabric as well as above a minimum value to prevent
condensation of sulfur trioxides. In addition, a high moisture content of
the gases can form an irremovable plug within a filter bag by blinding.
Fabric filters have been successfully demonstrated to control arsenic
emissions from glass plants. Arsenic emissions testing was performed by the
Environmental Protection Agency on a fabric filter installed on a glass
furnace. The testing method employed was "Reference Method for the
Determination of Particulate and Gaseous Arsenic Emissions from Non-Ferrous
Smelters", Method No. 108, with some approved modifications. The samples
collected were analyzed for arsenic using atomic absorption spectrophoto-
metry. The overall arsenic control efficiency of the fabric filter was
about 93 percent. Detailed results of this test are presented in
Appendix C. Emissions data provided by another plant also showed an
estimated 93 percent control of arsenic emissions by a fabric filter.
Table 3-2 shows the uncontrolled and controlled arsenic emissions from the
two glass furnaces equipped with fabric filters.
3-6
-------
TABLE 3-2. PERFORMANCE OF FABRIC FILTERS ON ARSENIC EMISSIONS
FROM GLASS FURNACES
UNCONTROLLED ARSENIC
EMISSIONS
PLANT
7
14
FURNACE
A
B
kg/hr
0.291
0.068
Ob/hr)
(0.65)
(0.15)
CONTROLLED ARSENIC
EMISSIONS
kg/hr
0.023
0.004*
Ob/hr)
(0.051)
(0.01)
PERCENT
REDUCTION
93
93
*Estimated emissions.
3.2.2 Electrostatic Precipitators18
Electrostatic precipitators are the most widely used participate
control technique in the glass manufacturing industry. The fundamental
steps of electrostatic precipitation are particle charging, collection, and
removal and disposal of the collected material. Particulate charging is
accomplished by generating charge carriers which are driven to the
particulates by an electric field. Collection occurs as the charged
particulates migrate to electrodes to which the charged particles adhere.
Applying a mechanical force to the collection electrodes dislodges the
collected material which then falls into hoppers. Effective transfer of
dust to the hopper depends on the formation of agglomerations of dust which
fall with a minimum of reentrainment.
There are two types of electrostatic precipitators used in the glass
industry. Both types are shown in Figure 3-2. One type consists of a large
rectangular chamber divided by a number of parallel rows of collection
plates that form gas flow ducts. Between these plates are hung a number of
small diameter wires which are connected to a high voltage direct current
potential forming a corona discharge around the wire. This corona generates
electrons which migrate into the incoming gas stream to form gas ions which
attach to these particles. The charged particles, in turn, are collected by
the grounded collection plates.
3-7
-------
J**5P
1 "'Collecting Electrodes
Discharge Electrodes
Charging Electrode Weights
Charging
Electrodes (-)
Positive Grounded
Collector Plates (+)
Conventional Electrostatic Precipitator
Positive
Charging
Needles
Positive
Electrode
Plates
egative Collecting
Plate Electrodes
Negative Grounded
Collecting Plate
Non-Uniform
Electric
Section
UnL
Electric
Section
Positive Plate
Electrodes
om
Collecting
Section
Needle Type Electrostatic Precipitator
Figure 3-2. Conventional and needle type electrostatic precipitators.
3-8
-------
The other type of ESP has a multitude of stainless steel needles
fastened to the leading and trailing edge of the discharge plates. This
design configuration requires a low voltage which allows close spacing
between the two collecting surfaces in each field: the positively charged
discharge plates, which have the attached needles; and, the grounded
collector plates. This close plate spacing permits short collecting
sections and relatively high flow velocities. Additionally, the regions
between the needles exhibit a uniform electric field which aid particle
agglomeration. Dust is retained on both the collector plates and discharge
plates.
Resistivity of the particulate is a determining design parameter. If
the resistivity of the particles is very high (>_ 2 x 10 ohm-cm), the
dust layer accumulating on the collection plates must not be allowed to
build up to the thickness typical for a lower resistivity particle. If this
occurs, excessive sparking might result. Under these conditions, the
applied voltage might have to be reduced, and the resulting decrease in both
corona current and electrical field would lower the collection efficiency.
Particles with very low resistivity will also adversely affect the ESP
performance. If the resistivity of the particles is less then
2 x 10 ohm-cm, the electrical forces holding the dust cake onto the
collecting plates are weaker than with particles of higher resistivity. As
a result, more of the particles will be reentrained and the emissions will
be higher. Some typical resistivity figures for various types of glass are:
12
Borosilicate glass 10 ohm - cm
Lead glass 10 ohm - cm
Soda-lime glass 10 to 10 ohm - cm
(Depending on temperature and
moisture content)
Electrical resistivity decreases as temperature increases. Depending upon
the glass type, the operating temperature may be increased or decreased to
obtain the optimal performance from an electrostatic precipitator.
Electrostatic precipitators are being used on several furnaces melting
arsenic containing glasses and have shown high control efficiencies on
3-9
-------
arsenic. The Environmental Protection Agency has performed arsenic emission
19
test on one glass furnace equipped with an electrostatic precipitator.
The testing method employed was "Reference Method for the Determination of
Participate and Gaseous Arsenic Emissions from Non-Ferrous Smelters", Method
No. 108, with some approved modifications. The samples collected were
analyzed for arsenic using atomic absorption spectrophotometry. Due to the
lack of test ports at the inlet to the electrostatic precipitator, the
uncontrolled emissions were measured at the outlet with the electric current
to the precipitator turned off. The electrostatic precipitator was found to
reduce arsenic emissions by about 99 percent. Detailed results of this test
are presented in Appendix C.
Several other electrostatic precipitators are also Installed on
furnaces producing arsenic containing glasses. Emissions data provided by
plants for these furnaces indicate that electrostatic precipitators reduced
arsenic emissions by 90 to 99 percent. One other plant reported 95 percent
20
reduction on particulate emissions. In general, the reduction
efficiency of particulate control devices on arsenic can be expected to be
similar to that on total particulates. One company reports that its
electrostatic precipitators perform highly selectively on arsenic and lead,
25
and achieve higher retention than on total particulates. The
uncontrolled and controlled arsenic emissions data for electrostatic
precipitators are presented in Table 3-3.
3.2.3 Scrubber Systems26
Although scrubber systems have been built to control particulate
emissions in the glass industry, presently only a few devices are in use to
control container glass emissions. The most common system in operation is
the venturi scrubber. A typical venturi scrubber is shown in Figure 3-3.
In a venturi scrubber, particle-laden gases are accelerated through a
restriction in the ducting where water is injected into the gas stream. The
velocity of the gas stream provides the dual function of atomizing the
scrubbing fluid while at the same time providing a differential velocity
between particles and the resulting liquid droplets. By utilizing high
3-10
-------
TABLE 3-3. PERFORMANCE OE ELECTROSTATIC PRECIPITATORS ON
ARSENIC EMISSIONS FROM GLASS FURNACES3
PLANT
1
1
3
8
11
12
13
13
UNCONTROLLED ARSENIC
EMISSIONS
FURNACE kg/hr (Ib/hr)
A
B
A
A
A
C
Al
1
CJ
)
D
0.07
0.02
0.37
0.30
0.79
1.80
> Common 0.05
Stack
0.378
(0.144)
(0.0514)
(0.81)
(0.66)
(1.74)
(3.96)
(0.11)
(0.83)
CONTROLLED ARSENIC
EMISSIONS PERCENT
kg/hr (Ib/hr) REDUCTION
0.0031
0.0007
0.011
0.009
0.0027
0.018
0.005
0.0227
(0.0069)
(0.0015)
(0.025)
(0.02)
(0.006)
(0.04)
(0.01)
(0.05)
95
97
97
97
99
99
90
94
'References: 21, 22, 23, ?4.
3-11
-------
CO
I
ro
Stack
ID Fan
Venturi Feed Pump
Cyclonic Separator
Dual-Throat Scrubber
Packed Bed Preconditioner
Electrical Controls
/Agitator
Overflow to Disposal
Makeup Water
Recirculation Tank
Gate Valves for Tank Cleanout
>- Packed Bed Preconditioner Feed Pump
Figure 3-3. Typical scrubber system.
-------
power fans to accelerate the gas stream, it is possible to generate high gas
velocities at the throat of the venturi. Since the participates are mostly
water soluble, the scrubber provides a means of removing these emissions.
Additionally, some gases are absorbed as condensables.
The scrubber liquor is acidic due to the absorbed acid gases. Before
being recycled to the venturi, the pH of the liquor is controlled by caustic
solution injection. A bleed stream and makeup water addition insure that
the scrubber liquor is not saturated. Typically, a bleed rate of 1.3 x
10" m /s (2 GPM) is discharged for a 2.1 kg/s (200 TPD) container glass
plant. Even for a larger furnace, the bleed rate would be expected to be
less than 3.2 x 10"4 m3/s (5 GPM).
The pressure drop to obtain high velocities in the throat of a scrubber
is directly proportional to the gas velocity squared and the liquid-to-gas
ratio; therefore, high velocities are possible only at substantial pressure
drops which result in high fan energy expenditures. Typical pressure drops
are approximately 7,500 Pa (30 inches of water).
None of the scrubber systems in the glass industry is currently used on
furnaces melting arsenic glasses.
3.3 EFFECT OF GAS COOLING ON ARSENIC EMISSIONS CONTROL
As discussed in Chapter 2, inorganic arsenic compounds are used in
glass manufacturing for a combination of reasons depending on the particular
glass being produced. In the majority of cases arsenic compounds act as
fining (clarification) or decolorizing agents. Arsenic is added to the
glass batch in the form of powdered arsenic trioxide (As?07) or liquid
C~ O
arsenic acid (H^AsO.).
From the theoretical considerations presented in Section 2.2, it can be
expected that essentially all the arsenic in the batch eventual I/ converts
to the trivalent stage, AsgOg. Therefore, the emissions of arsenic
would be in the form of As^. The relative quantities of As203 in
vapor and solid phase would depend on the temperature of the flue gas.
Table 3-4 shows the vapor pressure data for arsenic trioxide. Due to the
high volatility of arsenic trioxide, a large portion may exist in the vapor
phase in the flue gas from glass furnaces. As the gas is cooled and
3-13
-------
97
TABLE 3-4. ARSENIC TRIOXIDE (ARSENOLITE) VAPOR PRESSURE DATA^'»
Vapor
Pressure
(mm of Hg)
_7
2.4 x 10 '
2.5 x 10":
— Li.
4.6 x 10 I
1.9 x 10"?
2.2 x 10 ^
2.6 x 10",
1.0 x 10,
2.7 x loir
1.0 x lOp
5.0 x lOV
1.0 x I0j
2.0 x 10;
4.0 x 10,1
6.0 x 10,
1.0 x 10,
2.0 x 10,
4.0 x 10,
7.6 x 10
Equilibrium
Vapor Phase
Concentration
(mg As/m )
0
3.9 x 10 r1
4.0 x 10"1
1 1
7.4'x lO"
3.1 x 10}
3.5 x lOi
4.2 x 10p
1.6 x 10,
4.3 x lOJf
1.6 x 10J
8.1 X 10r
1.6 x lOj
3.2 x 10^
6.4 x 10?
9.7 x 10g
1.6 x 10,-
3 2 x 10
•J • L- A X\J/~
6.4 x 107
1.2 x 10'
Temperature of Arsenic Trioxide
60-61
81-86
101-105
117-124
119-124
149-152
153.5
165
212.5
242.6
259.7
279.2
299.2
310.3
332.5
370.0
412.2
457.2
3-14
-------
saturation (equilibrium vapor phase concentration) is achieved, some arsenic
trioxide condenses to the solid phase. Theoretically, this cooling is
essential for the particulate control device to be effective in controlling
arsenic emissions. Figure 3-4 shows that in tests performed on various
arsenic sources at copper smelters, cooling the gas resulted in condensation
of arsenic to final vapor concentration close to the saturation line. This
indicates that arsenic emissions from copper smelters are in the form of
arsenic trioxide.
However, in spite of the theoretical considerations for glass furnaces
(which also suggest arsenic trioxide emissions), available test data provide
conflicting information. For example, EPA tests on the ESP at furnace 11A
showed the inlet concentration of total arsenic (vapor and solid phase
o oq
arsenic) to the ESP was about 0.014 gm/m at 207°C. The vapor phase
equilibrium concentration at this temperature is about 16 gm/m3,
indicating that all the arsenic in the flue gas could be accomodated in the
vapor phase. However, the ESP outlet concentration showed 99 percent
reduction in total arsenic emissions. This level of control indicates that
almost all the arsenic was in solid phase. This was further verified by the
fact that essentially all of the arsenic was collected in the front half of
the stack sampling train (probe and filter) which was maintained at 121°C
*5
(250°F). At this temperature about 0.035 gm/m of arsenic (or all of the
3
0.014 gm/m of total arsenic in the gas) is predicted to be in the vapor
phase and should have passed through the filter to be collected in the back
half impingers.
Similarly, the test results on the fabric filter at furnace 7A indicate
that the total inlet arsenic concentration to the fabric filter (at about
238°C) was only about 6 percent of the equilibrium vapor phase concentration
30
at this temperature. This indicates that all the arsenic in the gas
could be accomodated in the vapor phase. However, the stack sampling train
collected about 93 percent of the total arsenic in the front half filter.
The total arsenic emission reduction achieved by the fabric filter was also
about 93 percent.
3-15
-------
00
o
V
c
V
u
<
10
10
10
» 3
A I
10
-1
10"
Legend
Symbol Device
£ Electrostatic Precipitator
Q Fabric Filter
O Venturi Scrubber
O Acid Plaat
• Blackened symbols represent inlet
conditions to the control devices.
• Open symbols represent outlet conditions
from the control devices. 2
• Superscript numbers designate the
inlet/outlet pairs.
• Solid line corresponds to saturated As,0.
vapor.
• 8
• Asterisk (*) indicates that a spray chamber unit
cooled the inlet gas prior to treatment by the
control device.
B 9
1.65 1.8
2.0
2.2
2.4
2.6
2.8
3.0
3.2
3.4
3 o-l
10 Z Reciprocal of Offgas Temperature, K
300
250 200
150
100 80 60
20
Offgas Temperature. °C
Figure 3-4. Effect of temperature on the performance of arsenic
emissions control devices on copper smelters.J1
3-16
-------
The test data, therefore, indicate that at about 200 - 250°C
temperature range, greater than 90 percent of the arsenic emissions are in
the solid phase. In addition, at these temperatures particulate control
devices achieve greater than 90 percent reduction in total arsenic
emissions.
The total concentration of arsenic at the above temperature levels was
lower than that necessary to cause saturation of arsenic trioxide. There
are several possible reasons why, in spite of the low concentrations,
greater than 90 percent of the arsenic was in the solid phase. One possible
reason may be that the presence of other cations and anions in the batch
interfere with the physical behavior of pure arsenic trioxide. Another
possible reason may be that the final form of the arsenic emissions is not
in the form of arsenic trioxide. The presence of various substances such as
nitrates, chlorides, and fluorides may cause formation of one or more stable
arsenic compounds that are in the solid phase at the 200 - 250°C temperature
range. In particular, the final form of arsenic emissions may be in the
form of arsenic pentoxide which is considerably less volatile than arsenic
trioxide. Yet another possible reason may be the adsorption of arsenic
trioxide on other particulate matter.
In general, based on emission factors for various glass types presented
in Table 2-2,'the concentration of arsenic in flue gases from glass furnaces
will be below the saturation line shown in Figure 3-4. Due to practical
considerations such as acid dew point, glass furnace flue gases may not be
cooled much below 121°C (250°F). Even at this temperature level, arsenic
concentrations will not be high enough to cause saturation if the arsenic
compound is arsenic trioxide. Therefore, cooling the gas from glass
furnaces would not have the same beneficial effect in reducing arsenic
emissions (even if all arsenic was in the vapor phase) as in the case of
copper smelters. In any case, since due to one or more of several possible
reasons discussed earlier, greater than 90 percent of the total arsenic is
in the solid phase, the theoretical maximum condensation due to cooling is
limited to less than 10 percent of the total arsenic.
3-17
-------
3.4 REFERENCES
1. Telecon. Koralewski, T., Libby-Owens-Ford Company, with Fidler, K.,
Radian Corporation. February 7, 1983. Conversation about flat glass
plant.
2. Telecon. Leroy, A., Fourco Glass Company, with Fidler, K., Radian
Corporation. February 7, 1983. Conversation about flat glass plant.
3. Telecon. Havran, D., Combustion Engineering (Glass Division), with
Fidler, K., Radian Corporation. February 8, 1983. Conversation about
flat glass plant.
4. Telecon. DeNormandie, R. L., AFG Industries, Inc., with Fidler, K.,
Radian Corporation. February 8, 1983. Conversation about flat glass
plant.
5. Telecon. Sherron, T., Ford Motor Company (Glass Division), with
Fidler, K., Radian Corporation. February 8, 1983. Conversation about
flat glass plant.
6. Bauer, R. J. Arsenic: Glass Industry Requirements. In: Arsenic:
Proceedings of the Arsenic Symposium, Lederer, W. H. and
Fensterheim, R. J. (ed.) Gaithersberg, Maryland, Van Nostrand Reinhold
Company. 1983. p. 52.
7. Draft trip report. S. A. Shareef, Radian Corporation, to file.
March 8, 1983. 4 p. Report of March 2, 1983 visit to Corning Glass
Works in Corning, New York.
8. Reference 6, p. 51 - 52.
9. Reference 6, p. 52.
10. Reference 6, p. 52.
11. Crose, P. (Acurex Corporation.) Arsenic: An Environmental Materials
Balance - Draft Final Report. (Prepared for U. S. Environmental
Protection Agency.) Washington, D. C. EPA Contract No. 68-01-6107.
March 1981. pp. 3-3, 3-14.
12. PEDCo Environmental, Inc. Extended Source Survey Report for Arsenic.
(Prepared for U. S. Environmental Protection Agency.) Research
Triangle Park, N. C. EPA Contract No. 68-02-3173. May 1982. pp. 3-31
to 3-33.
13. Letter and attachments from Cherill, J., Corning Glass Works, to
Brooks, G., Radian Corporation. April 7, 1982. 19 p. Information on
inorganic arsenic emissions from Corning Glass Works plants.
3-18
-------
14. U. S. Environmental Protection Agency. Glass Manufacturing Background
Information: Proposed Standards of Performance. Research Triangle
Park, N. C. Publication No. EPA-450/3-79-005a. June 1979. pp. 4-6 to
4-8.
15. Reference 14, pp. 4-9 to 4-13.
16. Thalman*, M. ?.*, et al. (Monsanto Research Corporation.) Arsenic Glass
Manufacturing Emission Test Report at Corning Glass Works, Central
Falls, Rhode Island. (Prepared for U. S. Environmental Protection
Agency.) Research Triangle Park, N. C. Publication No. EMB Report
78-GLS-3. February 1979.
17. Reference 13.
18. Reference 14, pp. 4-19 to 4-21.
19. Thalman, M. T., et al. (Monsanto Research Corporation.) Arsenic Glass
Manufacturing Emission Test Report at Corning Glass Works, State
College, Pennsylvania. (Prepared for U. S. Environmental Protection
Agency.) Research Triangle Park, N. C. Publication No. EMB Report
78-GLS-4. February 1979.
20. Letter from Goebel, G., Kentucky Bureau of Environmental Protection, to
Brooks, G., Radian Corporation. February 11, 1982. 2 p. Information
on inorganic arsenic emissions from Corning Glass Works plant in
Danville, Kentucky.
21. Memo from Shareef, S. A., Radian Corporation, to file. April 7, 1983.
1 p. Emissions data on glass furnaces at RCA Corporation.
22. Telecon. Iden, C., Owens-Illinois, with Brooks, G., Radian
Corporation. March 3, 1982. Conversation about inorganic arsenic
emissions from Owens-Illinois glass plants.
23. Reference 13.
24. Reference 19.
25. Letter from Armstrong, F. L., FCA Corporation, to Brooks, G., Radian
Corporation. March 8, 1982. 1 p. Information on corrected estimate
of arsenic emissions.
26. Reference 14, pp. 4-14 to 4-16.
27. u. S. Environmental Protection Agency. Arsenic Emissions from Primary
Copper Smelters - Background Information for Proposed Standards.
Research Triangle Park, N. C. Draft EIS. February 1981. p. 4-2
3-19
-------
28. Reimers, J. W. and Associates, Ltd. Study of the Emission Control
Technology for Arsenic in the Nonferrous Metallurgical Industry.
(Prepared for the Air Pollution Control Directorate - Environment
Canada.) January 1977. p. 24.
29. Reference 19.
30. Reference 16.
31. Radian Corporation. Preliminary Study of Sources of Inorganic Arsenic.
(Prepared for U. S. Environmental Protection Agency.) Research
Triangle Park, N. C. Publication No. EPA-450/5-82-005. August 1982.
p. 11.
3-20
-------
4.0 MODEL FURNACES AND REGULATORY ALTERNATIVES
Model furnaces for the glass manufacturing source category and
regulatory alternatives for the control of arsenic emissions are described
in this chapter. Model furnaces are developed in order to provide estimates
of the environmental, cost, and economic impacts of the regulatory
alternatives on individual typical furnaces. The nationwide environmental,
cost, and economic impacts which are presented in the subsequent chapters
are based on the actual existing furnaces.
The model furnaces are based on data from the background information
document (BID) for the glass manufacturing new source performance standard
(NSPS) except for the 4 and 23 Mg/day furnaces which were based on the
actual plant data submitted by glass manufacturing companies.2'3'4'5'6 It
should be noted that in order to obtain conservative estimates for control
costs, the gas flow data assumed in the NSPS analysis were on the high side.
In addition, the gas flows estimated from actual plant data were based on
very high stack oxygen level (18%). It is expected that due to energy
conservation considerations, the actual gas flows and, therefore, the actual
control costs may be somewhat lower. Section 4.1 presents the model
furnaces in terms of production capacity, glass industry segment, stack
parameters, and uncontrolled arsenic emissions. Section 4.2 describes the
regulatory alternatives for control of arsenic emissions from glass plants.
4.1 MODEL FURNACES
The model furnaces for the glass manufacturing source category are
defined to be various sizes of a single regenerative-type, fossil fuel-fired
glass melting furnace with no add-on control device. The model furnace
parameters are shown in Table 4-1. No specific glass type (e.g., lead,
borosilicate, soda-lime, etc.) is assumed to be manufactured by the model
furnaces. Instead, each model furnace is characterized as being in a
4-1
-------
TABLE 4-1. GLASS MANUFACTURING MODEL FURNACE PARAMETERS
.£»
I
PRODUCTION CAPACITY
Mg/day (tons/day) glass pulled
Glass industry segment
Arsenic emissions, kg/hr
(Ib/hr)
Stack Height, m (ft)
Stack Diameter, m (ft)
Exit Temp., °K (°F)
Flow Rate, m /sec (acfm)
Flow Velocity, m/sec (fps)
4.5
(5)
Pressed
and
bl own
0.046
(0.1)
25.9 (85)
0.4 (1.36)
480 (405)
1.14 (2,400)
9.1 (30)
?3
(25)
Pressed
and
blown
0.24
(0.52)
25.9 (85)
0.89 (3.04)
480 (405)
5.7 (12,000)
9.1 (30)
45
(50)
Pressed
and
b 1 own
0.47
(1.04)
25.9 (85)
1.0 (3.4)
480 (405)
7.6 (16,000)
9.1 (30)
91
(100)
Pressed
and
blown
0.95
(2.08)
25.9 (85)
1.4 (4.75)
480 (405)
15.1 (32,000)
9.1 (30)
131
(200)
Pressed
and
blown,
container, and
wool fiberglass
1.89
(4.17)
25.9 (85)
2.1 (6.7)
480 (405)
30.2 (64,000)
9.1 (30)
636
(700)
Flat
6.63
(14.58)
25.9 (85)
4.4 (14.4)
480 (405)
138.3 (293,000)
9.1 (30)
-------
general glass industry segment such as pressed and blown, container, etc.
It should be noted that the glass type does not affect the flue gas flow
rates from the furnaces. However, as shown in Table 2-2, the emission
factors are affected by the type of glass produced. Since due to
proprietary considerations, the model furnaces are not characterized as
producing specific types of glasses, a mid-point emission factor of
0.25 kg/Mg glass pulled (0.5 Ib/ton) is assumed for all model furnaces.
The stack parameters (gas temperature, stack height, stack diameter,
gas flow rate, and gas flow velocity) are based on information from the NSPS
background information document. Stack parameter information is used for
the risk analysis presented in Appendix E. The risk analysis is performed
for the actual existing furnaces and the actual stack parameters are used,
when available. For furnaces with no stack parameter information, the model
plant stack parameters were used. Table 4-2 presents the stack parameters
used for the risk analysis.
4.2 REGULATORY ALTERNATIVES
The purpose of developing regulatory alternatives is to provide a basis
for determining the air-quality and nonair-quality environmental impacts,
energy requirements, and the costs associated with varying degrees of
arsenic emissions reduction. The regulatory alternatives selected for
analysis include "no NESHAP regulation" and two increasingly restrictive
levels of control requirements. The two increasingly restrictive control
requirements allow for analysis of the impacts of varying degrees of
emissions reduction. The three regulatory alternatives are shown in
Table 4-3 and are discussed below.
4-3
-------
TABLE 4-2. STACK PARAMETERS USED FOR RISK ANALYSIS
Plant Furnace
i A
B
2 A
3 A
4 A
B
f*
L,
D
E
5 A
8
6 A
B
C
7 A
8 A
9 A
10 A-.
B ("Total From
C (4 Furnaces
oJ
11 A
12 A (? Stacks)
B (2 Stacks)
C
13 A-)
B [Common
C | Stack
oJ
14 A
B
15 A
Stack
Height
M(Ft)
22.3 (73)
9.75 (32)
15.5 (50.7)
21.3 (70)
44.2 (145)
36 (118)
38.1 (125)
19.2 (63)
21.9 (72)
39.6 (130)
29 (95)
15.2 (50)
15.2 (50)
36.6 (120)
18.3 (60)
41.1 (135)
41.1 (135)
15.2 (50)
45.7 (150)
9.1 (30)
7.6 (25)
22.9 (75)b
15.2 (50)
9.1 (30)
22.9 (75)b
30.5 (100)
25.9 (85)
27.9 (90)
27.9 (90)
29.0 (95)
Vertical
Stack Cross-Sectional
Diameter ,Area,
M(Ft) VF (Ft2)
1.52 (5)
0.95 (3)
0.70 (2.3)
1.58 (5.2)
1.78 (5.8)
1.4 (4.58)
0.99 (3.25)
0.8 (2.67)
1.83 (6)
1.07 (3.5)
1.52 (5)
2.4 (8)
1.37 (4.5)
1.52 (5)
0.91 (3)
1.8 (5.9)
2.4 (8.0)
4.98(16)
1.52 (5)
0.76 (2.5)
0.76 (2.5)
0.91 (3.0)a
0.82 (2.7)
0.76 (2.5)
1.22 (4)b
1.2 (4)
1.2 (4)
1.1 (3.7)
1.2 (3.9)
0.45 (1.5)
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
^3,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8.073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
(8,073)
Gas Velocity
M/Sec (Ft/Sec)
Reg. Alt. 1
14.0
9.32
29.3
12.7
6.0
10
9.8
18
4.9
8.56
3.0
3.1
3.1
6.2
7.5
3.0
3.4
1.5
15.2
13.1
' 18.3
29
18.6
17.4
22.9
14.0
15.2
12.2
(46)
(31)
(96)
(41.6)
(19.8)
(33)
(32.2)
(60)
(16.1)
(28)
(10)
(1.0)
(10)
(20.2)
(25)
(10)
(11)
(5)
(50)
(43)
(60)
(95)
(61)
(57)
(75)
(46)
(50)
(40)
Reg. A
14.0
9.32
29.3
12.7
3.67
6.6
6.7
13.7
3.5
6.2
2.4
10
2.26
4.2
7.6
3.0
2.6
1.5
15.2
15.2
15.2
17.4
22.9
8.9
15.2
12.2
TtTT
(46)
(31)
(96)
(41.6)
(12)
(21.7)
(22)
(45.1)
(11.5)
(20.4)
(7.8)
(33)
(7.4)
(13.8)
(25)
(10)
(8.6)
(5)
(50)
(50)
(50)
(57)
(75)
(29.1)
(50)
(40)
Gas Temperature
°K (°F)
Reg. AH. 1
505
436
494
450
783
725
700
636
669
655
610
310
644
700
410
440
666
311
477
672
755
644
544
461
477
755
477
644
K450)
(325)
(430)
(350)
(950)
(845)
(800)
(685)
(745)
(720)
(640)
(100)
(700)
(800)
(280)
(333)
(700)
(100)
(100)
(750)
(90C)
(700)
(520)
(370)
(400)
(900)
(400)
(700)
Reg. Alt.
505 (450)
436 (325)
494 (430)
450 (350)
477 (400)
477 (400)
477 (400)
477 (400)
477 (400)
477 (400)
477 (400)
477 (400)
477 (400)
477 (400)
410 (280)
440 (333)
477 (400)
311 (100)
477 (400)
477 (400)
477 (400;
461 (370'
477 (400;
477 (400
477 (400
644 (700
Vertical cross sectional area of the
bTwo stacks assumed to be replaced by
building perpendicular to the wind.
single stack under Reg. Alt. 2.
4-4
-------
TABLE 4-3. REGULATORY ALTERNATIVES FOR CONTROL OF INORGANIC
ARSENIC EMISSIONS FROM GLASS MANUFACTURING
Regulatory Alternative 1: -No NESHAP regulation.
Regulatory Alternative 2: Installation of electrostatic precipitators
or fabric filters.
Regulatory Alternative 3: Complete ban on inorganic arsenic emissions
from glass melting furnaces.
4.2.1 Regulatory Alternative 1
Alternative 1 represents the general level of control that would exist
in the absence of establishing any arsenic NESHAP regulations. None of the
existing glass furnaces are currently required to control arsenic emissions
at the Federal or State level. However, arsenic emissions from several
existing glass furnaces are controlled indirectly as a result of State
participate matter regulations. The particulate matter regulations are met
by installing an add-on control device, installing an electric boosting
process modification, or collecting particulate matter as fallout in the
furnace checkerwork. As shown in Table 2-2, 13 of the 32 existing arsenic
emitting furnaces have add-on control devices (a fabric filter or an
electrostatic precipitator).
4.2.2 Regulatory Alternative 2
This alternative represents the level of control achievable by
installation of an electrostatic precipitator or a fabric filter.
4.2.3 Regulatory Alternative 3
This is the most restrictive alternative for control of arsenic
emissions from glass furnaces. This alternative would require a complete
elimination of arsenic emissions from glass furnaces.
4-5
-------
4.3 REFERENCES
1. U. S. Environmental Protection Agency. Glass Manufacturing Background
Information: Proposed Standards of Performance. Research Triangle
Park, N. C. Publication No. EPA-450/3-79-005a. June 1979. pp. 6-8 to
6-14.
2. Letter from Armstrong, F. L., RCA Corporation, to Brooks, G., Radian
Corporation. March 8, 1982. 1 p. Information on corrected estimate
of arsenic emissions.
3. Letter from Goebel, G., Kentucky Bureau of Environmental Protection, to
Brooks, G., Radian Corporation. February 11, 1982. 2 p. Information
on inorganic arsenic emissions from Corning Glass Works plant in
Danville.
4. Telecon. Iden, C., Owens-Illinois, with Brooks, G., Radian
Corporation. March 3, 1982. Conversation about Owens-Illinois arsenic
emissions.
5. Letter and attachments from Murray, D. E., Anchor Hocking Corporation,
to Shareef, S. A., Radian Corporation. March 14, 1983. 5 p.
Information on inorganic arsenic emissions from Anchor Hocking glass
plants.
6. Letter and attachments from Cherill, J., Corning Glass Works, to
Brooks, G., Radian Corporation. April 7, 1982. 19 p. Information on
inorganic arsenic emissions from Corning Glass Works plants.
4-6
-------
5.0 ENVIRONMENTAL IMPACT
The environmental impact resulting from the implementation of the three
regulatory alternatives identified in Chapter 4 are discussed in this
chapter. As discussed in Chapter 4, the impacts on model furnaces are
presented for the purpose of providing estimates of the environmental impact
on an individual furnace. The nationwide impacts are estimated by using the
available emissions data for the existing furnaces shown in Table 2-5.
Section 5.1 presents the air pollution impacts. The air emissions of
arsenic from each model furnace are presented for the three regulatory
alternatives. The total nationwide air emissions are based on the actual
existing glass furnaces with available arsenic emissions data. Sections 5.2
through 5.4 present the water pollution, solid waste, and energy impacts.
5.1 AIR POLLUTION IMPACT
In this section the impact of each regulatory alternative on air
pollution is considered. The impact of implementing the regulatory
alternatives is calculated for each model furnace. The total nationwide
impact is calculated for the actual furnaces.
5.1.1 Model Furnace Emissions
The arsenic emissions from each model furnace under the three
regulatory alternatives are presented in Table 5-1. Under Regulatory
Alternative 1, no NESHAP regulation, the arsenic emissions are the same as
the baseline emissions presented in Chapter 4.
Regulatory Alternative 2 represents installation of a fabric filter or
an electrostatic precipitator. As presented in Chapter 3, these devices
have demonstrated reduction of arsenic emissions ranging from 90 to
99 percent. To be conservative, a reduction level of 90 percent is assumed
for both fabric filters and electrostatic precipitators. The arsenic
5-1
-------
TABLE 5-1. ARSENIC EMISSIONS FROM MODEL GLASS FURNACES UNDER THE THREE REGULATORY ALTERNATIVES
en
l
ro
Arsenic Emissions, kg/hr (Ib/hr)
1.
2.
3.
REGULATORY ALTERNATIVE
No NESHAP regulation.
Installation of a fabric
filter or an electrostatic
precipitator.
Complete ban on arsenic
emissions from glass
melting furnaces.
4.5
(5)
0.046
(0.1)
0.0046
(0.01)
0.0
(0.0)
Model
23
(25)
0.24
(0.52)
0.024
(0.052)
0.0
(0.0)
Furnace Mg/day (tons/day)
45
(250)
0.47
(1.04)
0.047
(0.104)
0.0
(0.0)
91
(100)
0.95
(2.08)
0.095
(0.208)
0.0
(0.0)
181
(200)
1.89
(4.17)
0.189
(0.417)
0.0
(0.0)
Percent
,.,.. Reduction
Over
Baseline
636 Arsenic
(700) Emissions
6.63
(14.58) 0
0.663
(1.458) 90
0.0
(0.0) 100
Baseline is Regulatory Alternative 1.
-------
emission levels shown for Regulatory Alternative 2, therefore, represent
90 percent reduction over baseline emissions.
Regulatory Alternative 3 requires complete elimination of arsenic
emissions from glass melting furnaces. This would result in 100 percent
reduction in arsenic emissions.
5.1.2 Nationwide Emissions
The total nationwide annual impact in the first year is calculated
based on the available emissions data for arsenic emitting glass furnaces.
As discussed in Chapter 2, there may be several other furnaces in the glass
industry with arsenic emissions. Therefore, the total environmental impact
is probably understated. Table 5-2 presents the nationwide arsenic
emissions estimates for the first year under each regulatory alternative.
The nationwide arsenic emissions under Regulatory Alternative 1 are the same
as the baseline emissions presented in Table 2-2 and are about 36.7 Mg
(40.5 tons). Emissions under Regulatory Alternative 2 are calculated based
on installation of a fabric filter or an electrostatic precipitator on all
existing furnaces that do not currently have add-on control devices. The
arsenic control efficiency is assumed to be 90 percent. The nationwide
emissions under Regulatory Alternative 2 are estimated to be 4.33 Mg
(4.76 tons). This represents an annual emissions reduction of 32.37 Mg
(35.61 tons) over baseline arsenic emissions.
Regulatory Alternative 3 would result in zero arsenic emissions. This
represents a reduction of 36.7 Mg (40.5 tons) over Regulatory Alternative 1
and 4.33 Mg (4.76 tons) over Regulatory Alternative 2.
5.2 WATER POLLUTION IMPACT
There would be no water pollution impact due to implementation of any
of the regulatory alternatives.
5.3 SOLID WASTE IMPACT
There would be no significant solid waste impact due to implementation
of the regulatory alternatives. The catch from particulate control devices
in glass plants are generally recycled to the batch feed.1'2'3 One
5-3
-------
TABLE 5-2. ESTIMATED FIRST YEAR NATIONWIDE ARSENIC EMISSIONS FROM
GLASS FURNACES UNDER EACH REGULATORY ALTERNATIVE9
Existing PM Controls that
Plant Furnace Reduce Arsenic Emissions
1 . A
B
2 A
3 A
4 A
8
C
0
E
5 A
B
6 A
B
C
7 A
8 A
9 A
10 A
B
C
D
11 A
12 A
B
C
ESP preceded fay evaporative cooler
ESP preceded by evaporative cooler
ESP
ESP
None
None
None
None
None
None
None
None
None
None
FF
ESP
EB
1
1 Total
> from None
4 furnaces
J
ESP preceded by an evaporative
cooler
(2 stacks) None
(2 stacks) None
ESP preceded by evaporative cooler
and a cyclone
Arsenic Emissions
Reg. Alt
0.
0.
0.
0.
3.
3.
0.
027
006
069
089
09
53
73
0.55
0.55
1.27
0.91
0.45
1.82
3
0
0
15
0
0
1
1
.09
.19
.076
.20
.038
.023
.99
.83
0.073
(0.
(0.
(0.
(0.
(3.
(3.
(0.
. 1
035)
0056)
076)
101)
4)
9)
8)
(0.6)
(0.6)
(1.4)
(1.0)
(0.5)
(2.0)
(3.4)
(0.21)
(0.084)
(16.8)
(0
(0
(2
.042)
.025)
.19)
(2.01)
(0.08)
Reg. AH.
0.
0.
0.
0.
0.
0.
0.
0.
0.
027
006
069
089
309
353
073
055
055
0.127
0.091
0.045
0.182
0.309
0
0
1
0
0
0
0
.19
.076
.52
.0038
.023
.199
.183
0.073
, Mg/yr (Tons/yr)
2 Reg. AH. 3
(0.035)
(0.
(0.
(0.
(0.
(0.
(0.
(0.
(0.
0066)
076)
101)
034)
39)
08)
06)
06)
(0.14)
(0.1)
(0.05)
(0.2)
(0.34)
(0.21)
(0.084)
(1.68)
(0
.0042)
(0.025)
(0
(0
.219)
.201)
(0.08)
5-4
-------
TABLE 5-2. (Continued)
Existing PM Controls that
Plant Furnace Reduce Arsenic Emissions
13 A "
' B Common ESP preceded by evaporative 0
f Stack cooler
ESP preceded by evaporative 0
cooler
14 A EB 0
B FF 0
15 A None 0
Arsenic
Reg.
AH. 1
.038 (0.
.191
.76
.038
.118
Total 36.70
(0.
(0.
(0.
(0.
(40.
042)
21)
84)
042)
126)
50)
0
0
0
0
0
4
Emissions, Mg/yr (Tons/yr)
Reg. Alt. 2 Reg. Alt. 3
.038
.191
.076
.038
.0118
.33
(0
(0
(0
(0
(0
(4
.042)
.21)
.084)
.042)
.0126)
.76) 0.0 (0.0)
bFF =• fabric filter
ESP = electrostatic precipitator
EB * electric boosting
PM « participate matter
5-5
-------
company reports that about 10 percent of the collected dust cannot be
recycled and, therefore, must be disposed of. This dust is classified as
a hazardous waste.
5.4 ENERGY IMPACT
As in the case of the new source performance standard for the glass
industry, the energy impact of the regulatory alternatives are not
significant. Regulatory Alternative 1 will have no energy impact. There
will be a minor negative energy impact under Regulatory Alternative 2. The
energy required to operate fabric filters and electrostatic precipitators
ranges from 0.006 to 0.03 kw-hr/kg of glass produced, depending upon the
glass type. Using published values for the container glass, flat glass, and
pressed and blown glass categories, the energy required to operate these
control systems for these categories ranges from 0.2 to 2 percent,
0.2 percent, and from 0.1 to 0.5 percent of the energy needed to produce
glass, respectively. As discussed in Chapter 2, it may not be technically
feasible to substitute arsenic with other agents in the glasses that
currently use arsenic. Therefore, it may be assumed that the production of
all arsenic glasses will be eliminated under Regulatory Alternative 3, and,
in the worst case, that existing furnaces that produce arsenic-containing
glasses will close. Assuming that the 32 existing arsenic-using furnaces
close, there will be a small positive energy inpact of about 580,000 Mw-hr
per year, representing the energy saved by not operating these furnaces.
5-6
-------
5.5 REFERENCES
1. Letter and attachments from Swander, T., RCA Corporation to
Goodwin, D., EPArESED. September 27, 1978. 8 p. Information on
arsenic in glass and arsenic emissions from glass manufacturing units.
2. Telecon. Iden, C., Owens-Illinois, with Brooks, G., Radian
Corporation. March 3, 1982. Conversation about Owens-Illinois arsenic
emissions.
3. Telecon. Biles, R., GTE, with Shareef, S. A., Radian Corporation.
March 1, 1983. Conversation about inorganic arsenic emissions from GTE
glass furnaces.
4. Draft trip report. S. A. Shareef, Radian Corporation, to file
March 8, 1983. 4 p. Report of March 2, 1983 visit to Corning Glass
Works in Corning, New York.
5-7
-------
6.0 COST ANALYSIS
This chapter presents a cost analysis of the regulatory alternatives
described in Chapter 4. The cost impacts for individual model furnaces are
estimated in Section 6.1. The nationwide cost impacts are estimated by
calculating costs for the actual existing furnaces and are presented in
Section 6.2.
6.1 COST ANALYSIS OF REGULATORY ALTERNATIVES FOR MODEL FURNACES
The three regulatory alternatives for controlling arsenic emissions
from glass manufacturing furnaces were discussed in Chapter 4 and are listed
again in Table 6-1.
TABLE 6-1. REGULATORY ALTERNATIVES FOR CONTROLLING ARSENIC
EMISSIONS FROM GLASS MANUFACTURING FURNACES
Regulatory Description
Alternative
1 No NESHAP regulation.
2 Installation of electrostatic
precipitators or fabric filters.
3 Complete ban on inorganic arsenic
emissions from glass melting furnaces,
Regulatory Alternatives 1 and 3 have no capital costs associated with them
and, thus, are not considered in this chapter. Costs associated with
Regulatory Alternative 2 are presented in this section for the six model
furnaces. The economic impacts of Regulatory Alternatives 2 and 3 are
presented in Chapter 7.
6-1
-------
6.1.1 Capital Costs
Installed capital costs of fabric filter and electrostatic precipitator
(ESP) systems were estimated for each of the six model furnaces. The bases
for the capital cost estimates are summarized in Table 6-2. The major
components of the control system include the particulate device and an
evaporative cooler (e.g., a spray chamber). The evaporative cooler is
assumed to cool the gas from 207°C (405°F) to 121°C (250°F) in order to
operate a cool side ESP or a baghouse.
For the fabric filter systems, costs of the control device, spray
chamber, and other auxiliary equipment were determined from standard cost
1 2
estimating references. ' Equipment costs were updated to October 1982
dollars using Chemical Engineering Plant Cost Indices. ' Additional
capital costs for instruments and control and taxes and freight were based
on a percentage of control device and auxiliary equipment costs. Direct
and indirect installation costs for the fabric filter systems were estimated
as a factor of total purchased equipment costs. Finally, in order to
compute capital costs for existing furnaces, a retrofit allowance equivalent
to 100 percent of installed capital costs was applied.
Capital cost estimates for ESP systems were based on the cost analysis
of ESP systems performed in the background information document (BID) for
New Source Performance Standards (NSPS) for the glass manufacturing
industry. In the NSPS cost analysis, the capital cost of the ESP systems
was based on the purchase price of the control device and auxiliary
equipment (except a spray chamber), the cost of site preparation, equipment
installation, design engineering, the contractor's fee, and interest during
construction. The cost components used in the NSPS document are shown in
Table 6-3. Estimated costs were provided by glass manufacturers for actual
retrofit installations and were based on the use of heavy-duty equipment,
liberal installation allowances, and 20 percent contingency. Not included
are costs of research and development, cost of land, possible loss of
production during equipment installation, or losses during startup.
Equipment cost of ESP's was based on estimates relating plate area to cost
for conventional plate and wire precipitators. Installation costs and
6-2
-------
TABLE 6-2. BASES FOR CAPITAL COST ESTIMATES
Capital Cost Item
Cost Basis
Reference
Fabric Filter System
fabric Filter
Spray Chamber
Suction Blower
Pump
Motors
Ductwork, Expansion
Joints, and Dampers
Screw Conveyor
Instruments and
Control
Taxes and Freight
Air-to-cloth ratio • 0.61 m /min/m (?. cfm/ft' )
Filtering media - Nomex. Reverse air cleaning,
continuous duty, suction blower, insulated. Net
cloth area =• inlet flow » air-to-cloth ratio.
Baghouse costs based on net cloth area. Filter
costs based on gross cloth area.
Inlet gas temperature = 207°C (405°F). Outlet
gas temperature = 121°C (250°F). Water temperature
• 16"C (60°F). Inlet flow rate is the furnace out-
let flow rate. Outlet flow rate is equivalent to
the inlet flow rate plus the volume of added water
vapor. Spray chamber costs based on Inlet flow rate.
Gas temperature » 121°C (250°F). Actual iP « 25 4 cm
(10 in., H-0. 4P at standard conditions * 34 cm
(13.4 in.) H,0. Flow rate is the fabric filter outlet
flow rate. Backwardly curved fan costs based on iP at
standard conditions and flow rate.
Spr|y chamber water requirements • 0.024 l/m3 (1.8 x
10 gal/ft ) of gas through spray chamber. Pump
costs based on water requirements.
Totally enclosed motor costs based on RPM and horse-
power requirements for fan and pump operation.
Duct length = 30.5 m (100 ft) per fan. Duct diameter *
spray chamber outlet flow rate t 1,372 m/min (4,£00 fpm)
(rnnimum velocity to prevent fallout). Duct material -'
0.635 cm (0.25 In.) carbon steel. Two expansion joints
per fan. Two circular dampers per fan.
Screw conveyor required on model plants > 91 |Mg/day
(100 tons/day). Conveyor length = 15 m T50 ft)
Screw diameter * 23 cm (9 in.) for model plants
< 181 Mg/day (200 tons/day). Screw diameter *
30 cm (12 in.) for 635 Mg/day (700 tons/day) model
plant.
10 percent of fabric filter, spray chamber, and
auxiliary equipment costs.
8 percent of fabric filter, spray chamber, and
auxiliary equipment costs.
117 percent of purchased equipment costs.
Direct and Indirect
Installation Costs
Retrofit Allowance !00 percent of Installed capital costs.
Electrostatic Preeipitator System
Installed costs for
control device and
auxiliary equipment
(except spray chamber)
Spray Chamber
Installed costs for control device and auxiliary
equipment (except spray chamber) based on cost
estimates presented In the background information
document for HSPS for the glass manufacturing
industry. Specific collection area •< 3.28 m? per
m /mm (1,000 ftVl.OOO ACFM) for pressed aod blown
and wooj fiberglass furnaces, 1.08 mz per nvVmln
(330 ftYl,000 acfm) for flat nlass furnaces
and 1.31 m1- per m /mln (400 fr/1,000 acfm) for
container glass furnaces. Installed costs follow
log/log relationship with the following data points:
279 m 13,000 ft') predpl tator - 5816,405
5,574 m reO.OOO ft") precipitator - $3,973,915
Same as spray chamber cost estimates developed for
fabric filter systems.
Ref. 1, Fig. 5-13
Table 5.1
Table 5.2
P-ef. 1, Fig. 4-23
Ref. 1, F
-------
TABLE 6-3. COMPONENT OF CAPITAL COSTS FOR ELECTROSTATIC PRECIPITATORS
DIRECT COST COMPONENTS9
Equipment
Instrumentation
Piping
Electrical
Foundation
Structural
Sitework
Insulation
Paintings
Buildings
INDIRECT COST COMPONENTS9
Field overhead
Contractor's fee
Engineering
Freight
Offsite
Taxes (5% of material)
Allowable for shakedown
Spares
Contingency (20% of total)
Interest during construction
aEach component cost estimated separately depending on equipment involved.
6-4
-------
indirect costs were based on published information and engineering judgment.
Total installed costs calculated in this manner for ESP systems are shown in
Figure 6-1. Figure 6-1 also shows actual control system costs as reported
by industry in response to EPA inquiries for the NSPS study. It can be seen
from Figure 6-1 that the relationship between cost and the size of the ESP
system follows an increasing logarithmic relationship such that a doubling
of plate area results in a cost increase of about 1.5 times. For purpose of
this analysis, the NSPS cost estimates were updated to October 1982 dollars
using Chemical Engineering Plant Indices and the installed cost of a spray
p Q in r j
chamber was added. 'y'lu
Table 6-4 presents the installed capital costs of fabric filter and ESP
systems for the six model furnaces.
6.1.2 Annualized Costs
Total annualized costs for fabric filter and ESP systems include direct
operating costs and capital recovery costs. The bases used in determining
annualized costs for the six model furnaces using fabric filter and ESP
systems are summarized in Table 6-5. Direct operating costs include
operating labor, maintenance, overhead, replacement materials, utilities,
waste disposal, property taxes, insurance, and administration.
The number of operator man-hours required for ESP systems was estimated
for the six model furnaces using the following equation:
man-hours/yr = 485 Inx - 1,221 (Ref. 11)
where: x = kg/day of dust collected = 4.75 x glass production, Mg/day
The equation was derived such that operator man-hour requirements determined
for the smallest and largest model furnaces would represent the range of
values recommended for ESP systems in Reference I.12 Operator man-hours
for fabric filter systems were estimated to be double the requirements for
ESP systems for the individual model furnaces.13 Maintenance labor costs
were based on estimated maintenance man-hour requirements on a per-shift
basis. Maintenance requirements were estimated to be one man-hour per shift
for fabric filter systems and 0.5 man-hour per shift for ESP systems.14
6-5
-------
CTv
10
VI c
«= 5
10
5
10'
1000
I
o ESTIMATED VALUES
o REPORTED VALUES
10.000
100,000
TOTAL PLATE AREA, ft'
Figure 6-1 Reported installed costs of electrostatic precipitator control
systems compared with estimated cost curve used in this study.
-------
TABLE 6-4. CAPITAL COST ESTIMATES FOR MODEL FURNACES
(Thousands of October 1982 Dollars)
Capital Cost Item
Fabric Filter System
Fabric filter
Spray chamber
Auxiliary equipment
Instruments and control
Taxes and freight
Direct and indirect installation costs
Installed Capital Costs
Retrofit Allowance
Total Installed Capital Costs
Electrostatic Precipitator System
Total Installed Capital Costs
4.5
(5)
65.9
62.7
12.0
14.1
11.2
194
360
360
720
666
Model
23
(25)
104
65.9
25.5
19.5
15.6
269
500
500
1,000
1,530
Furnace Size, Mg/day (Tons/day)
45
(50)
120
67.3
22.7
21.0
16.8
290
538
538
1,076
1,813
91
(100)
179
72.7
37.3
28.9
23.1
399
740
740
1,480
2,615
181
(200)
296 1
83.5
56.3
43.6
34.9
602 2
1,116 3
1,116 3
2,232 7
3,772 4
(pressed and blown
and wool fiberglass
furnaces)
2,335
(container glass
furnaces)
636
(700)
,148
161
175
148
119
,049
,800
,800
,600
,751
-------
TABLE 6-5. BASES FOR ANNUALIZED COSTS
Arnualized Cost Item
Cost Basis
Fabric Filter
Electrostatic Precipitator
Operating Labor
(a) operator
(b) supervisor
Maintenance
(a) labor
(b) materials
Overhead
Replacement Parts
(a) materials
(b) labor
Utilities
(a) electricity
(b) water
Waste Disposal
Property Tax, Insurance
and Administration
Capital Recovery
hrs/yr = 2 (485 Inx - 1,221) where x =
kg/day of dust collected and kg dust
collected/Mg of glass = 4.75 (9.5 Ib/ton).
$11.33/hr Bureau of Labor Statistics (BLS)
average wage plus 20 percent fringes for glass
industry workers
15 percent of operator cost
1 hr/shift, $12.52/hr (hourly rite of
10 percent premium over operating labor)
100 percent of maintenance labor cost
80 percent of labor charges for operation
and maintenance
Cost of bags/useful life of 1.5 yrs
100 percent of replacement materials cost
$0.0525/kWh
SO. 066/1 ($0.25/1,000 gal)
($40/ton)
kg dust collected/Mg of glass = 4.75
(9.5 Ib/ton)
4 percent of installed capital cost
10 percent interest over 20 year equipmert
life (CRF = 11.746 percent of installed
capital cost)
hrs/yr = 485 Inx - 1,221
9 511.38/hr
15 'percent of operator cost
0.5 hr/shift, $12.52/hr
100 percent of maintenance
labor cost
80 percent of labor charges
for operation and maintenance
S0.0525AWH
$C.066/1 ($0.25/1,000 gal)
$44/Mg ($40/ton)
kg dust collected/Mg of
glass = 4.75 (9.5 Ib/ton)
4 percent of installed
capital cost
10 percent interest over
20 year equipment life (CRF
11.746 percent of installed
capital cost)
NA » Not applicable.
6-8
-------
Overhead costs were based on a factor of total labor charges for operation
and maintenance.
Replacement parts for fabric filter systems consist of filter bags.
The annualized cost of replacing fabric filter bags is determined by
dividing the purchased cost of the bags by their estimated useful life of
1 f\
1.5 years. Labor costs associated with replacement of fabric filter
bags were based on 100 percent of replacement materials cost.17
Utility requirements for the fabric filter and spray cooler systems
include electricity for fan, pump, and reverse air cleaning operations and
water for the spray chamber. Electricity requirements for fabric filters
were estimated from the appropriate equations and charts presented in the
standard cost estimating references.18 ESP systems require electricity to
meet precipitator power requirements of 10.76 watts/m2 (1,000 watts/1,000 ft2)
of precipitator and additional fan power requirements.19 Waste disposal
costs were based on estimated disposal requirements of 4.75 kg dust/Mg of
glass produced. Although the amount of waste disposed of is expected to be
0 to 10 percent of the dust collected, it is conservatively assumed that all
dust is disposed of in an offsite landfill. Property taxes, insurance, and
administration costs were estimated as a factor of total installed capital
costs.20
Capital recovery costs were based on 10 percent interest over 20 year
equipment life for both fabric filter and ESP systems. The capital recovery
factor (CRF) is calculated using the following equation:
CRF = i (1 + i)n
(1 + i)n - 1
where: i = annual interest rate
n = equipment life (years)
Table 6-6 presents annualized costs for fabric filter and ESP systems
for the six model plants.
6-9
-------
TABLE 6-6. ANNUALIZED COSTS FOR MODEL FURNACES1
(Thousands of October 1982 Dollars)
Model Plant Size,
Annualized Cost Item
4,5
(5)
23
(25)
45
(50)
Mg/day (Tons/day)
31
(100)
181
(200)
636
(700)
Fabric Filter System
'
Operating labor
Malntenanc*
Overhead
Replacement parts
Utilities
Waste disposal
Property tax, insurance,
and administration
Capital recovery (Wf, interest,
20 year li'e, CRF « 0.1176)
Total Annualized Costs
6.99
26.3
16.1
4.15
2.93
0.33
28.8
84.5
170
27.4
26.3
32.5
15.6
12.5
1.66
40.0
117
273
36.2
26.3
39.5
20.8
14.9
3.32
43.0
126
310
45.0
26.3
46.5
34.6
28.0
6.64
59.1
174
420
53.8
26.3
53.6
64.7
53.7
13.3
89.3
262
617
69.7
25.3
66.3
268
265
46.6
304
893
1,939
Electrostatic Precipitator System
Operating labor
Maintenance
Overhead
Utilities
Waste disposal
Property tax, insurance, and
administration
Capital recovery (Wf, Interest,
20 year life, CRF - 0.11746)
Total Annualized Costs
4.02
13.1
8.47
1.24
0.33
26.5
78.2
132
13.7
13.1
16.2
6,00
1.66
61.2
180
292
18.1
13.1
19.8
8.29
3.32
72.5
213
348
22.5
13.1
23.3
16.6
6.6
105
307
494
26-91;
n.:b
13.1?
26 -8c
13 "^
93 '.4C
443b
274C
707^
467C
34.9
13.1
33.2
78.9
46.6
190
558
955
Based on 8,400 operating hours/yr.
Pressed and blown and wool fiberglass furnaces.
""Container glass furnaces.
6-10
-------
6.2 NATIONWIDE COST IMPACTS
The nationwide cost impacts are estimated for the actual existing
arsenic emitting furnaces without add-on particulate devices listed in
Table 3-5. The capital costs are for electrostatic precipitator systems and
are calculated for the actual gas flows from the existing glass furnaces.
The cost bases used are the same as presented for electrostatic
preclpitators in Tables 6-2 and 6-5. The nationwide capital cost estimates,
the first year annualized costs, and the first year arsenic emission
reductions are presented in Table 6-7.
TABLE 6-7. NATIONWIDE FIRST YEAR COST IMPACTS
(Last Quarter 1982 Dollars)
Capital Costs (Million dollars) 29.84
Annualized Costs (Million dollars) 5.39
Emission Reductions (Mg) 32.37
6.3 OTHER COST CONSIDERATIONS
6.3.1 Costs Associated with Monitoring
The costs associated with monitoring include costs for continuous
opacity monitoring devices. Capital and annualized costs for continuous
opacity monitoring include equipment and installation costs for a
transmissometer, recorder, and other associated equipment and are estimated
to be about $25,000 per site. Annualized costs include preventative
maintenance, calibration, recording and reducing data, general maintenance
and capital recovery and are estimated to be about $14,000 per source.21
6-11
-------
6.4 REFERENCES
1. Neveril, R. B. (GARD, Incorporated.) Capital and Operating Costs of
Selected Air Pollution Control Systems. (Prepared for the U. S.
Environmental Protection Agency.) Research Triangle Park, N. C.
Publication No. EPA-450/5-80-002. December 1978.
2. Richardson Engineering Services, Inc. Process Plant Construction
Estimating Standards, Volume 4. San Marcos, California, 1982. Work
account pp. 100-300 and 100-651.
3. Economic Indicators. Chemical Engineering. 9CK3):7- February 7,
1983.
4. Reference 1, Table 3.1.
5. Reference 1, Table 3.3.
6. Reference 5.
7. U. S. Environmental Protection Agency. Glass Manufacturing Background
Information: Proposed Standards of Performance. Research Triangle
Park, N. C. Publication No. EPA-450/3-79-005a. June 1979. pp. 8-52
to 8-62.
8. Reference 3.
9. Reference 4.
10. Reference 1, Figure 4-23.
11. Letter and attachments from Myers, R. E., EPA:ISB to Jenkins, R.,
EPA:EAB. December 7, 1981. 2 p. Revising annual control costs for
glass manufacturing plants.
12. Reference 1, Table 3.5.
13. Reference 11.
14. Reference 12.
15. Reference 1, Table 3.4.
16. Reference 1, Table 3.6.
17. Reference 15.
18. Reference 1, p. 3-17.
6-12
-------
19. Letter and attachments from Myers, R. E., EPA:ISB, to file. April 21,
1983. Model furnace control cost for ESP systems.
20. Reference 15.
21. Memo from Fidler, K., Radian Corporation, to file. March 30, 1983.
2 p. Updated costs of opacity monitoring.
6-13
-------
7.0 ECONOMIC IMPACT
This chapter first presents an economic profile of the glass industry
(Section 7.1). The data presented in the economic profile is then used in
an analysis of the economic impacts of alternatives to regulating inorganic
arsenic from the industry (Sections 7.2 and 7.3). The economic profile
focuses on several glass industry characteristics, such as: number
and location of plants, glass supply, glass demand, prices, value of
shipments, competition, and employment.
7.1 Industry Economic Profile
7.1.1 Introduction
Four major segments of the glass industry are included in this pro-
file: (1) pressed and blown glass not elsewhere classified, which is Standard
Industrial Classification (SIC) 3229; (2) flat glass, which is SIC 3211; (3)
container glass, which is SIC 3221; and (4) wool fiberglass, which is part of
SIC 3296 (Mineral Wool) .l
The products of these industries are briefly discussed. Pressed and
blown glassware includes textile fiberglass (a fiberglass reinforced plastic
used primarily in automobiles) and the following other industries: (1)
machine-made consumer pressed and blown glassware (tumblers, stemware, and
tableware); (2) handmade consumer pressed and blown glassware; (3) glass
tubing for fluorescent and neon lighting, (4) incandescent light bulb blanks;
(5) television picture tube envelopes; (6) scientific and technical glass;
and (7) optical glass. Major attention is given to these seven industries in
the pressed and blown category. Flat glass is basically window glass and is
used in the construction and automobile industries. Container glass is
bottles. Wool fiberglass is primarily used in the thermal insulation of
residential and nonresidential structures.
Table 7-1 summarizes the value of shipments in the four segments of
the glass industry in 1980, and Table 7-2 summarizes the number of companies
and employment in the same industries. From the first table, it can be seen
that these four industries together accounted for 0.4 percent of GNP in
1980.
7-1
-------
Table 7-1
VALUE OF SHIPMENTS IN THE GLASS INDUSTRY. 19802*3
Value of
Value of Shipments
SIC Shipments as percent
Code Industry (million dollars) of GNP
3229a Pressed and Blown 2,847.5 .108
Glass not else-
where classified
3211
3221
Part of
3296^
TOTAL
Flat Glass
Container Gl ass
Wool Fiberglass
1,497.6
4,486.3
1,340.6
10,172.0
.057
.170
.051
.386
alncludes textile fiberglass shipments of 681.0 million
dollars.
bWool fiberglass accounts for 63 percent of SIC 3296,
Mineral Wool. SIC 3296 had a value of shipments of
2,128.0.
7-2
-------
Table 7-2
NUMBER OF COMPANIES AND EMPLOYMENT IN THE GLASS INDUSTRY, 19804'6
SIC
Code
3229b
3211
3221
Part of
3296C
Industry
Pressed and Blown
61 ass not el se-
where cl assified
Flat Glass
Container Glass
Wool Fiberglass
TOTAL
Number of
Companies3
325
43
31
4
403
Total Employment
(1,000)
45.7
17.3
64.5
13.9
141.4
Employment
of Production
Workers
(1,000)
37.1
13.8
56.5
11.2
118.6
aData on the number of companies is from 1977.
^Includes textile fiberglass for which no separate figures are re-
ported for employment.
cWool fiberglass accounts for 63 percent of SIC 3296, Mineral Wool.
SIC 3296 had a total employment of 22,000 and employed 17,700 pro-
duction workers in 1980.
7-3
-------
Five firms in SIC 3229 which have been identified by EPA as using
arsenic in glass melting furnaces at 15 plants are of principal interest
for the purpose of the impact analysis in Section 7.2. The five firms are
Anchor Hocking, Corning Glass Works, GTE, Owens-Illinois, and RCA.7 The
plants and products are in Table 7-3. SIC 3211 and SIC 3221 are included
in this profile because they have used arsenic in the past and/or may elect
to do so in the future. Textile fiberglass (part of SIC 3229) and wool
fiberglass (part of SIC 3296) are not known to contain arsenic. Therefore
they will not be discussed in this chapter.
Table 9-4 summarizes the value of shipments for the seven industries
in SIC 3229 that have been identified as of principal interest. As this
table shows, there were marked differences in the growth in value of
shipments. Glass for fluorescent lighting, for machine-made pressed and
blown glassware, and for scientific purposes (including opthalmic lens
blanks) grew more than 10 percent per year between 1970 and 1981. Handmade
pressed and blown glass, on the other hand, grew only 5.6 percent per year.
When it is used in glass manufacturing, arsenic performs several
functions. Its major functions are to prevent the formation of bubbles and
to act as a decolorizing agent. It is also used in some instances in
special glass types to impart other particular properties needed for the
end use of the glass. For example, arsenic can provide stable fixation of
certain colors for optical glass by stabilizing selenium, provide high
glass permeability to infrared light for camera lenses, and provide a high
degree of energy transmission for solar collector glass.8
The basic raw materials for most glass production include sand in
the form of silica dioxide, soda ash, and limestone. Additional materials
such as boric acid are added to the basic raw materials to obtain a variety
of glass types, such as borosilicate glass or lead crystal. The raw
materials for glass production are relatively abundant so that none is in
short supply.9
7.1.2 Machine-Made Pressed And Blown Consumer Glassware
7.1.2.1 Introduction. In 1979 there were seven major firms who
produced machine-pressed and blown consumerware in 13 plants in 7 states of
the United States. Four plants were located in the Central part of the
country, one on the West Coast, and eight in the Eastern Sector. There were
7-4
-------
Table 7-3
GLASS MANUFACTURING PLANTS WHICH ADD ARSENIC TO RAW MATERIALS10'14
Plant
1
2
3
4
5
6
7
8
9
10
11
12
13
Furnace
A-B
A
A
A-E
A-B
A-C
A
A
A
A-D
A
A-C
A-D
Primary Products3
TV Picture Tube Components
Glass Tubing
TV Picture Tube Components
Tableware Glass
Specialty Container
Tableware Glass
Heat Resistant Globes, Electric
Light Covers
Lead Glasses
Tableware Glass
Optical Glasses
TV Picture Tube Components
Tableware Gl ass
Scientific and Technical Glass
Tubing
14 A-B Aluminosilicate and Lead
Glasses, Scientific
Glassware
15 A Hand Blown Lead Crystal Glass
aAll products are machine-made except where noted
7-5
-------
Table 7-4
SHIPMENTS OF CONSUMER. SCIENTIFIC. TECHNICAL. AND
INDUSTRIAL GLASSWARE, 1970 TO 198115"24
(Thousands of Dollars)
SIC Product
Annual Compound
Growth Rates
Code Description 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1970-1981 19/b-1981
3229121 Machine-made 286,749 326,291 380,022 418.674 462,408 625.664 554.085 686,581 732,399 758,399 808,605 912,700 11.1 8.7
to pressed and
3229128 blown glass-
ware
3229131 Handmade pressed 31,450 32,577 36,502 37,980 40,126 38,046 36,863 42,890 46,662 47,696 55,125 56,992 5.6 7.5
to and blown glass-
3229133 ware
3229227 Tubing and cane 37,744 38,481 40,151 45,386 42,958 50,102 40,608 56.630 65,952 84,811 103,607 109,195 10.1 17.9
for fluorescent
and neon light-
ing
3229225 Incandescent 214,136 211,752 268,399 295,922 235,000 298,163 236,377 311,085 333,103 372,655 426,815 441,284 6.8 11.0
and electric bulb
3229235 blanks and tele-
vision tube blanks
3229423 Scientific, 156,424 170,116 D" D° 260,480 270,293 246,455 295,168 353.233 401.279 438.500 476,235 10.7 11.6
~^ to laboratory, 1n-
CJT, 3229427 dustrial and
technical glass-
ware; ophthalmic
lens blanks and
optical Instru-
ment lens blanks
All otherC 183,888 198,470 428,833 491,410 257,536 314,375 256,238 438,829 523,750 587,677 686,571 732,868 13.4 19.1
TOTAL 910,391 977.687 1.153,907 1.289,372 1.298.508 1,596,643 1.370.626 1.831,183 2.055,099 2.252,517 2.519,223 2,729.274 10J 12^.2
aCorapound growth rates «re calculated using two equations:
(1) G = (Xp/Xi)1/"
where G = compound growth multiplier; Xn = the value of X In the time period
n; Xi= the value of X in tlfie period 1; n « the length of time period; and
(2) The compound growth rate (C) can be found by: C = G - 1.
bD = withheld to avoid disclosing figures of individual companies.
'includes substantial interplant transfers of partially fabricated products.
-------
also fourteen plants that are primarily in the hand-pressed and blown
sector but which have been identified as having machine-pressed and blown
capabilities.25
7.1.2.2 Market Concentration. The sector is highly concentrated.
Currently, seven firms predominate: Corning Glass, Owens-Illinois, Anchor
Hocking, Brockway, Bart!ett-Collins, Jeanette Corp., and J. G. Durand Inter-
national , Inc.26
7.1.2.3 Supply. The most noteworthy aspect of supply in this
industry is that imports from a large number of foreign countries play a
substantial role in domestic supply. Table and kitchenware represent the
largest category of imports in this category of the pressed and blown segment
of the glass industry.2*>
7.1.2.4 Demand. In this sector, consumer taste and relative price
between similar items are the major determinants of demand. Corning Glass
and Anchor Hocking produce large lines of similar heat-resistant table and
kitchenware. Owens-Illinois and Anchor Hocking produce similar tumbler and
stemware items. Brand loyalty plays a large part in which firm's products
are chosen by the consumer.2?
The availability of a wide variety of substitutes from the import
market, especially in the tableware, tumbler, and stemware categories,
provides a highly competitive structure in this sector. The impact of
plastics and paper tableware products cannot be assessed.2?
Table 7-5 summarizes output, shipments and price for machine-made
consumer glassware between 1970 and 1981. As this table shows, output (in
1,000 dozen) for tumblers and stemware declined at an average compound
rate of 1.5 percent between 1970 and 1981; output of tableware and cookware
during the same period of time grew 1.5 percent a year. Both sectors have
experienced declines in 1980 and 1981 due to the recession.
Value of shipments rose throughout the same time period. This is
because prices rose at a substantial rate.
7.1.2.5 Prices. Prices for tumblers and stemware are typically
measured as prices per 1,000 dozen, and prices for tableware and cookware are
7-7
-------
—1
oo
Table 7-5
SHIPMENTS, OUTPUT AND PRICES OF MACHINE-MADE
CONSUMER GLASSWARE, 1970 TO
Tumblers and Stemware
1
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Annual Compound
Growth Rates
1970-1981
1975-1981
Uiitput
,000 Dozen
84,819
91,660
92,308
82,908
78,649
87,277
93,405
90,898
86,112
75,160
68,084
72,172
-1.5
-3.1
Shipments
$1,000
121,966
135,619
156,543
156,229
164,375
206,885
239,825
245,383
256,735
259,206
286,869
325,433
9.3
7.8
Price Per
1,000 Dozen
1,438
1,480
1,696
1,884
2,090
2,370
2,568
2,700
2,981
3,449
4,213
4,509
10.9
11.3
198115-24
Tableware and Cookware
Output
1,000 Pieces
324,852
395,884
430,270
458,498
466,455
474,668
494,533
504,965
523,542
519,047
377,000
381,156
1.5
-3.6
bhipments
$1,000
131,051
151,914
176,800
208,394
235,689
278,704
312,067
355,595
382,484
402,942
415,456
473,033
12.4
9.2
Knee Ker
1,000 Pieces
403
384
411
455
505
587
631
704
731
776
1,102
1,241
10.8
13.3
-------
typically measured as prices per 1,000 pieces. As Table 7-5 shows, the price
of tumblers and stemware per 1,000 dozen in 1981 was $4,509. The price of
tableware and cookware in the same year was $1,241 per 1,000 pieces. Assuming
that one pound of glass is required to produce between 2 and 6 pieces of
tumblers and stemware, the price per unit weight would be between $1.65/kg
and $4.96/kg ($0.75/lb and $2.25/lb). Assuming that tableware and cookware
require production of between 1 and 4 Ib each, the price per unit weight
would be between $2.76/kg and $11.02/kg ($1.25/lb and $5.00/lb).
Obviously it is difficult to obtain an estimate of the price per unit
weight of all machine-made pressed and blown consumer glassware. In August,
1978, an industry source estimated it to be $3.218/kg ($1.46/lb).28 This
estimate can be updated to October, 1982, using an average of the producer
price indexes for flat glass and container glass.29 This results in an
estimate of $4.386/kg ($1.99/lb) for October, 1982.
7.1.3 Handmade Pressed And Blown Glassware
7.1.3.1 Introduction. In 1979 there were 84 firms that were
identified as producing handmade pressed and blown products in 90 plants in
17 states. The greatest concentration of plants is located in the Ohio-
Pennsylvania-West Virginia area where 53 of the participating plants are
located. West Virginia has 28 firms producing in 30 plants.30 Currently,
it is estimated there are 80 firms, 30 of which are in the handmade pressed
sector and 50 in the handmade blown sector.31
7.1.3.2 Market Concentration. There is little apparent concentra-
tion in this industry. Many of the firms participating in this sector are
small and privately owned, producing in one plant only.30 About half of
the firms in this industry are estimated to have less than 100 employees.31
Manufacturers own their own molds and select their own designs or
patterns. In the past, customers placed special orders, and products were
made to order. The more recent trend is toward annual contract production.
The sector is characterized by entreprenuership.32
7.1.3.3 Supply. The major aspect of supply in this industry is the
role of imports. Imports are typically large and have grown in recent
years.32
7-9
-------
7.1.3.4 Demand. Hand-pressed and blown glassware supplies that
portion of the market that demands uniqueness and craftmanship. Because
large capital investment in machines and in multiple molds is not justified
in this sector, the volume of production and the variety of items produced
are low. Products are valued by the consumer because they are in relatively
short supply and because of the quality of individual craftsmanship which is
attached to them.33
Although machine-pressed and blown products are not perfect substi-
tutes for hand-processed items, both machine-made products and imported
hand-processed items are partial substitutes for U.S. hand-blown glassware.31
(Plastics have made some impact on the novelty glass portion of the market.
To the extent that uniqueness and the value of craftsmanship determine demand
for hand-processed products, plastics are not perfect substitutes.)33
Hand-pressed and blown products represent luxury spending from per-
sonal disposable income. They may represent major expenditures. In periods
of economic uncertainty, consumers may defer purchases of these items or may
purchase a lower-cost, though imperfect, substitute. Demand appears to be
highly price-elastic.33
Table 7-6 summarizes output, value of shipments, and prices for
hand-made consumer glassware. As the data in this table shows, output of
tumblers and tableware has declined 7.3 percent per year between 1970 and
1981. Output of stemware declined 4.2 percent during the same period of
time. It should be noted that stemware output has declined 14.8 percent per
year since 1975 and that this segment has been particularly hard-hit by the
current recession. The current level of capacity utilization in this industry
is said to be only 50 percent.31
7.1.3.5 Prices. Prices of handmade tumblers and tableware are
typically measured as price per 1,000 pieces, and prices of stemware are
measured as prices per 1,000 dozen. As Table 7-6 shows, the price of
hand-made tumblers and tableware in 1981 was $3,270 per 1,000 pieces, and the
price of handmade stemware in the same year was $30,092 per 1,000 dozen, the
same problem exists here as with machine-made pressed and blown products in
arriving at a price per unit weight.
In 1978, EPA used an estimated price of $6.061/kg ($2.75/lb).34
Using the producer price index for container and flat glass, this figure can
be used to arrive at an estimate for October, 1982. The October, 1982,
estimated price for all hand-made glassware is $8.419/kg ($3.82/lb).
7-10
-------
Table 7-6
SHIPMENTS, OUTPUT AND PRICES OF HANDMADE
CONSUMER GLASSWARE, 1970 TO 198115-24
Tumblers and Tableware
Output Shipments
1,000 Pieces $1,000
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Annual Compound
Growth Rates
1970-1981
1975-1981
18,510
17,189
18,457
16,875
15,507
11,194
8,963
8,330
9,010
9,073
7,906
8,061
-7.3
-5.3
16,180
15,817
17,582
16,083
19,010
17,059
17,527
18,732
20,681
21,486
24,252
26,358
4.5
7.5
Price Per
1,000 Pieces
874
920
953
953
1,226
1,524
1,955
2,249
2,295
2,368
3,068
3,270
12.7
13.6
Output
1,000 Dozen
1,641
1,927
2,069
3,094
2,897
2,666
2,756
2,017
2,088
2,326
966
1,018
-4.2
-14.8
Stemware
Shipments
$1,000
15,270
16,760
18,920
21,897
21,116
19,804
20,519
24,158
25,981
26,210
30,873
30,634
6.5
7.5
Price Per
1,000 Dozen
9,305
8,697
9,145
7,077
7,289
7,428
7,445
11,977
12,443
11,268
31,960
30,092
11.3
26.3
-------
7.1.4 Glass Tubing For Fluorescent And Neon Lighting
7.1.4.1 Introduction. There are currently only three firms that
manufacture glass tubing for fluorescent and neon lighting.35
7.1.4.2 Market Concentration. This industry is highly concen-
trated. The three fims are: General Electric, GTE Sylvania, and North
American Phillips.35 These firms produce mul ti-product lines and are
integrated in areas other than glass production as well.36
Investment capital requirements for equipment in this category is so
large, as is volume demanded, that small flourescent and neon light manufac-
turers find it more economically feasible to buy from a large firm than to
produce glass tubing themselves. The major producers are large multi-product
firms, and light tubing manufacture does not account for a significant
portion of their sales volume.36
7.1.4.3 Supply. A high percentage of the output of this industry
is used captively by the firms that produce glass tubing. Neither imports
nor exports appear to influence the domestic market for this tubing.37
7.1.4.4 Demand. Fluorescent and neon lights are the end-products.
Demand for the product is largely from the non-residential lighting market
and is determined by the need for the product. Nothing substitutes for glass
in lighting, and as a result demand appears to be relatively price-inelastic.37
The demand for fluorescent lighting does, however, depend on the state
of the economy, because plants which are shut down (or only partially used)
do not need to replace lights. As a result, it is estimated the current
capacity utilization in the industry is only 50 percent.35
7.1.4.5 Prices. Prices are not available from published documents.
A 1978 estimate of the price is $1.289/kg ($0.585/lb).34 Using producer
price indexes for flat and container glass, one can estimate the October,
1982 price. The estimate is $1.790/kg ($0.812/lb).
7-12
-------
7.1.5 Incandescent Light Bulb Blanks
7.1.5.1 Introduction. In 1979 two firms produced incandescent
light bulb blanks in 7 plants located in five states. All plants were
located in the Eastern region of the United States.38
7.1.5.2 Market Concentration. In 1979, Corning Glass and General
Electric produced virtually all of the incandescent light bulb blanks manu-
factured in the United States for both residential and non-residential use.
Both firms are large, integrated, multi-product, multi-national companies.38
7.1.5.3 Supply. As in other areas of lighting manufacture, capi-
tal investment requirements for entry into this market are huge. These
light bulb blanks are produced on the Corning ribbon machine, the cost of
which is estimated at $25 million in 1978 dollars. The very large inventory
of finished goods that must be maintained further limits available capital.
The characteristics of this category suggest a highly oligopolistic industry
in which few firms participate and the capital requirements are so vast as to
preclude the entry of small firms.38 Imports play a role in this industry,
but there are no estimates of the relative importance of imports in the
domestic market.38
7.1.5.4 Demand. There are no viable substitutes for an incandescent
bulb blank except insofar as consumers switch to fluorescent lighting.
Demand for the product is consistent with demand for electrical lighting as a
way of life.39
Consumer taste plays some very small role as a demand determinant in
the end product market. Preference for other forms of lighting, such as
fluorescent or neon tubing for residential use, influences demand. Mercury
vapor and high-pressure sodium vapor lights, superior to incandescent lights
in the amount of energy consumed, are new products being developed. To the
extent that product substitutes are limited, demand appears to be relatively
price-inelastic.3^
Demand does depend on the state of the economy, as it does for fluor-
escent lights. At present, the estimated capacity utilization rate in this
industry is 50 percent.35
7-13
-------
7.1.5.5 Prices. The price per unit weight of incandescent light
bulb blanks is difficult to estimate. The estimated price in 1978 was
$0.766/kg ($0.352/lb),34 Using the producer price indexes for flat and
container glass, an estimate of the October, 1982, price can be made. The
estimate is $1.076/kg ($0.488/lb).
7.1.6 Television Picture Tube Envelopes
7.1.6.1 Introduction. Three firms produce color TV picture tube
envelopes in five plants. One firm (Corning Glass) has announced plans to
close one of its plants.40 In addition, Owens-Illinois produces black and
white TV envelope tubes.41
7.1.6.2 Market Concentration. The industry is highly concentrated.
Corning Glass, Owens-Illinois and RCA, are the only firms participating in
this category of the industry. RCA produces for its own consumption.40
7.1.6.3 Supply. The major factor influencing the supply of tele-
vision envelope tubes is the size of imports of the end-product, television
sets. Imports of television sets are significant, particularly from Japan.
It is estimated that imports control the black and white TV market and the
smaller (15 to 16 inch) colored TV market. The 19 inch market is highly
competitive between U.S. producers, on the one hand, and foreign producers,
on the other. Domestic producers control the market for larger sets.40
7.1.6.4 Demand. Demand for TV tube envelopes is derived from
demand for TV sets. The market for new TV sets accounts for 90 to 95 percent
of demand for TV envelope picture tubes, the remaining 5-10 percent being the
replacement market.40
Television sets are a luxury item representing a large expenditure
from personal disposable income. The state of the general economy is an
indicator of how demand for a.luxury item will rise or fall, depending on
whether consumers defer purchases until a more economically auspicious
time.42 The industry is currently operating at a capacity utilization rate
of 67 percent.40
7-14
-------
7.1.6.5 Prices. The estimated 1982 price of TV envelope tubes is
$1.102/kg ($0.500/lb).40 This price per unit weight can be expressed in
October, 1982, dollars using the producer price indexes for flat and container
glass. In that case, price is $1.107/kg ($0.503/lb).
7.1.7 Scientific And Technical Glass Tubing
7.1.7.1 Introduction. In 1979, tubing for medical and pharmaceuti-
cal uses was produced by four major firms in ten plants located in seven
states. With the exception of one plant located in Illinois, all of the
plants were located in the Eastern section of the U.S.43
7.1.7.2 Market Concentration. The industry is highly concentrated.
In 1979, the four firms in this industry were: Owens-Illinois, Corning
Glass, Schott Optical Glass, and Wheaton Glass.43
7.1.7.3 Supply. Very little is known about supply conditions
in the industry. Scientific and technical tubing are imported into the U.S.
The extent to which they penetrate the market for medical-pharmaceutical
products cannot be estimated from available public data.44
7.1.7.4 Demand. Demand in this category is influenced by the
specialized nature of the products being manufactured and is relatively
price-inelastic. Growth and advances in technology in the medical and
health-related industries will influence demand. Viable substitutes from
some products in this area include plastics and ceramic glass. In that both
Owens-Illinois and Corning Glass maintain full production capabilities in
both of these areas, substitutability of other goods is not anticipated to
influence demand in terms of Owens-Illinois and Corning Glass in this market.45
7.1.7.5 Prices. It is difficult to arrive at an estimate of the
price per unit weight for this product category. An August, 1978, estimate
was $1.168/kg {$0.530/1b),28 Using the producer price indexes for container
and flat glass, the October, 1982, estimate is $1.596/kg ($0.724/lb).
7-15
-------
7.1.8 Optical Glass.
7.1.8.1 Introduction. In 1979, five firms in this category pro-
duced optical glass products in 8 plants located in New York, Massachusetts,
Virginia, and Pennsylvania.46
7.1.8.2 Market Concentration. The presence of only five major
firms in the industry indicates a high degree of concentration. Corning
Glass, American Optical, Bausch and Lomb, Eastman Kodak, and Schott Optical
Glass predominate in the industry.46
7.1.8.3 Supply. The major firms in this sector of the industry
are also engaged in end-product manufacture of a wide range of analytical ,
technical, electronic, and health-related diagnostic instruments. The sector
is highly specialized with products being produced largely by contract
arrangement or to fill intra-company needs. To that extent, small inven-
tories exist and no real impact is experienced from imported end-products.46
No significant export market exists for the glass itself, except
to an existing foreign subsidiary. End-product export reflects the special-
ized contract nature of this category and is relatively insignificant.46
7.1.8.4 Demand. Optical glass and instruments are a highly spec-
ialized, high-priced, low-volume category. Demand is influenced by any
growth of the electronic market, and by an expansion in government scientific
and research programs. Optical glass for use in analytical instruments
experiences increased demand when environmental, industrial, and aerospace
programs are initiated or expanded. Any advance in biomedical research
techniques brings about increased demand in this category.47
Plastics have become viable substitutes in some areas of ophthalmic
and instrument lens production. Plastics are, however, more expensive than
glass in a sector where prices are high for glass products. In uses where
weight is a factor, plastics have become a viable substitute.47
The highly specialized, high-priced, low-volume nature of this cate-
gory of the industry appears to indicate that price elasticity of demand is
relatively inelastic.47
7-16
-------
7.1.8.5 Prices. The prices per unit weight of optical glass was
estimated to be $9.014/kg ($4.09/lb) in August, 1978.28 Using the producer
price indexes for flat and container glass, an estimate of the October, 1982,
price can be made. The estimate is $12.320/kg ($5.59/lb).
7.1.9 Glass Products That Have Contained Arsenic
There are two glass products that have contained arsenic in the past.
They are flat glass and container glass. Each is discussed in turn.
7.1.9.1 Flat Glass
7.1.9.1.1 Introduction. Flat glass is window glass primarily used in
the automobile and construction industries.^8 in 1979, there were 27
manufacturing plants operating in 14 different states. The plants were,
however, concentrated in 4 states: California, Pennsylvania, Tennessee, and
West Virginia.49
7.1.9.1.2 Market concentration. The flat glass industry is highly
concentrated, with only 8 companies currently participating in the primary
manufacturing sector: PPG Industries, Inc.; Libbey-Owens-Ford Co. (LOF);
Ford Motor Co., Glass Division; Guardian Industries Corp.; AFG Industries,
Inc.; Hortis Brothers; Jeanette Flat Glass; and West Virginia Flat Glass
Company.50 The four largest companies (PPG, LOF, Ford, and Guardian)
controlled 86.7 percent of total domestic production capacity in 1977. PPG
dominates the construction market, and Libbey-Owens-Ford dominates the
automotive market.51
7.1.9.1.3 Total supply
Domestic supply. The plants in the flat glass industry have operated
at less than full capacity in recent years because of the slumps in domestic
automobile production and housing. New office building construction has,
however, continued to be a major source of demand and has raised operating
rates from their otherwise depressed levels. In addition, truck and bus
production has increased slightly.48
7-17
-------
Imports and exports. In 1982, exports were estimated to be 8.7
percent of industry shipments, and imports were estimated to be 3.9 percent
of apparent consumption (the sum of product shipments plus imports less
exports). The current situation in which exports exceed imports by a
comfortable margin is an improvement that has occurred during the last
decade. This change has been attributed to the industry's shift to float-
glass production techniques (which reduced unit costs) and to relatively low
costs for natural gas. As natural gas prices increase in the coming years,
U.S. producers may lose their present competitive advantage in world markets.48
The principal market for exports and imports of flat glass is Canada.52
Other important export markets are Mexico, Venezuela, and Australia. In
addition to Canada, West Germany and Japan are leading suppliers to the
U.S.48
7.1.9.1.4 Total demand. The demand for flat glass depends on (1)
the demand for new automobiles and new residential and nonresidential con-
struction, (2) the demand for the retrofitting of additional thermal glass to
existing structures, (3) the repairing of existing glass in automobiles and
construction, and (4) the size of automobiles and houses.
Table 7-7 presents the output, shipments, and average price paid for
flat glass from 1970 to 1981. From 1977 to 1981, the flat glass industry
experienced an annual compound decline in demand (output in 1,000 sq.
ft.) of 2.2 percent. This decline has been attributed to two factors:
(1) the cyclically depressed state of the domestic markets for new auto-
mobiles and houses, and (2) the longer, secular shift to smaller automobiles
and smaller houses, which require less glass per unit.53 Output of flat
glass is expected to recover as the housing and construction industries
improve. It has been estimated that the industry will grow at an annual
compound rate of 5 percent during the 1982-1987 period.53
Although physical output declined, Table 7-7 shows that value of
shipments rose from 1977 to 1981 by 6.5 percent. This was due to the increase
in the average price paid of 8.9 percent during the same period of time.
7.1.9.1.5 Prices. Prices of flat glass are typically measured as
price per 1,000 square feet. As Table 7-7 shows, the price of flat glass in
7-18
-------
Table 7-7
SHIPMENTS. OUTPUT. AND PRICES OF FLAT GLASS, 1970 TO 198154-60
•
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Annual Compound
Growth Rates
1970 to 1981
1977 to 1981
Quantity
U.UUU sq. ft.)
NAa
NA
NA
NA
NA
NA
NA
2,897,025
3,099,841
2,974,460
2,749,948
2,652,609
NA
-2.2
Flat Glass
Value of
Shipments
($1,000)
384,790
464,674
544,875
597,645
543,382
467,994
644,751
739,919
829,449
858,130
868,459
952,283
8.6
6.5
Price
1$ Her 1,000
sq. ft)
NA
NA
NA
NA
NA
NA
NA
255
268
288
316
359
NA
8.9
aNA = not available.
7-19
-------
1981 was $359 per 1,000 square feet. If prices are measured in price per
unit weight, flat glass is estimated to have sold at $0.456/kg ($0.207/lb) in
the second quarter of 1982.61
This data can be expressed in October, 1982, prices by using the
producer price index for flat glass. In October, 1982, dollars the price
of flat glass was $0.461/kg ($0.209/lb).
7.1.9.2 Container Glass
7.1.9.2.1 Introduction. The beer, food products, liquor/wine, and
soft drink beverage markets accounted for over 90 percent of glass container
production in 1981.62 Other markets served were medicinal and health
supplies, toiletries and cosmetics, and industrial products.62
In 1977, there were 123 container glass manufacturing plants located
in 30 states.63 Currently, it is estimated there are 24 to 29 companies
and 110 plants in operation.64
7.1.9.2.2 Market concentration. The container glass industry is con-
centrated. In 1979, the four largest companies accounted for 56 percent of
sales, with the next four largest companies accounting for an additional 21
percent of sales.63
In 1976, Owens-Illinois, Inc., was reported to have a 28.9 percent
share of the container glass market, with Brockway Company, Inc. having the
next largest share, 11.4 percent. Other companies ranking among the first 10
in 1976 market share included: Anchor Hocking, Thatcher Glass, Glass Con-
tainers Corp., Kerr Glass, Indian Head, Ball Corp., Chattanooga Glass, and
Midland Glass.63
7.1.9.2.3 Total supply. Because glass containers have a high
weight to volume, they are located as close as possible to the market they
are intended to serve. Plants are sometimes dedicated to servicing the needs
of one nearby customer (for example, a brewery).63
The desire of container glass manufacturers to relocate production
closer to its customers led to a wave of expansions in the years 1972-
1977.65 This has led to some overcapacity in the industry because of three
7-20
-------
factors: (1) increased competition from substitutes, especially plastic
bottles and metal cans; (2) decreased consumption of bottled products due to
the current recession; and (3) a tendency to use lighter weight bottles.66
Exports and imports. International trade is insignificant in the
glass container industry, chiefly because glass containers have a high
weight to volume. Exports were only .9 percent of shipments in 1981, and
imports were only 2.1 percent of apparent consumption.67
7.1.9.2.4 Total demand. The demand for container glass depends on
(1) the demand for its end products, chiefly beer, food products, and soft
drink beverages, (2) the existence and use of substitutes for glass in con-
tainers, and (3) the amount of glass recycling. Glass container manufacturers
face competition from metal cans and from plastic bottles. The industry is
currently said to be especially concerned about the substitution of plastic
bottles for glass containers in the wine and soft drink beverage market.68
Table 7-8 summarizes the growth in output, shipments, and price for
all containers from 1970 to 1981. From 1970 to 1981, the container glass
industry experienced an annual compound growth rate in physical output (1,000
gross) of 1.6 percent. Although output has increased only 1.6 percent since
1970, value of shipment has increased 9.7 percent. This higher growth rate
in value of shipments is due to the increase in the average price paid for
containers of 8.0 percent.
7.1.9.2.5 Prices. Prices of container glass are typically measured in
price per 1,000 gross. As Table 7-8 shows, the price of container glass in
1981 was $15,871 per 1,000 gross. If prices are measured in price per unit
weight, it is estimated container glass sold at $0.419/kg ($0.190/lb) in 1981.69
This figure can be updated to an October, 1982, price by using the pro-
ducer price index for container glass. The estimate is $0.456/kg ($0.207/lb).
7-21
-------
Table 7-8
SHIPMENTS, OUTPUT AND PRICE OF CONTAINERS. 1970 TO 19817°-75
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Domestic
Quantity
(1,000 Gross)
266,031
255,261
265,981
274,295
273,709
279,022
292,345
304,114
317,440
319,842
323,899
316,408
Glass Contai
Value of
Shipments (
($1,000)
1,815,456
1,907,550
2,062,403
2,208,245
2,475,168
2,929,814
3,294,089
3,570,147
4,023,458
4,234,851
4,540,764
5,021,865
ners
Price
$ Per l.OUU
Gross)
6,824
7,473
7,754
8,051
9,043
10,500
11,268
10,740
12,675
13,240
14,019
15,871
Annual Compound
Growth Rates
1970 to 1981 1.6 9.7 8.0
1975 to 1981 2.1 9.4 7.1
7-22
-------
7.2 Economic Analysis
7.2.1 Introduction
Section 7.2 presents the analysis of the economic impacts of alterna-
tives to regulating inorganic arsenic emissions from the glass industry.
Three principal economic impacts are analyzed. First, Section 7.2.4.1
analyzes the impact on prices. Second, Section 7.2.4.2 analyzes the impact
on profitability. Third, Section 7.2.4.3 analyzes the ability of firms to
obtain the capital required to meet the standards. Two regulatory alterna-
tives are evaluated: (1) standards based on the use of an evaporative cooler
followed by an electrostatic precipitator (ESP) or a fabric filter (FF)
and (2) standards that ban arsenic emissions. Section 7.2.4.4 discusses the
economic impact of banning arsenic emissions in glass manufacturing plants.
7.2.2 Summary
There are several results of the evaluation of price impacts, profit
impacts, and capital availability that should be mentioned briefly. It
should be noted that the conclusions discussed below are generally the same
for the ESP and FF options, exceptions being flat and container glass. Also,
the price and profit impact results were not evaluated for the actual furnaces
at the 15 plants that are likely to be affected, but rather for hypothetical
furnaces of a particular size and producing a particular product. Only some
of these "model plant" furnaces are representative of the furnaces at the 15
plants identified as producing arsenic-containing glass. Where model plant
furnace impacts do not depict impacts expected at actual plants (because no
arsenic is used), this is noted in the discussion.
Six model plant sizes and two or three levels of capacity utilization
are evaluated. In this analysis the assumption is made in cases of 25 and 50
percent capacity utilization that all control costs can be passed on only to
the arsenic-containing products and not to all products produced in the same
furnace. An alternative assumption is that control costs are distributed to
the entire product line produced by the furnace (including products that do
not contain arsenic). If the latter occurs, price and profit impacts are
lower than those reported here.
7-23
-------
In the case of price impacts, the larger size furnaces (45 megagrams per
day (Mg/D) through 636 Mg/D) have modest potential price increases, whereas
the smaller furnaces (4.5 Mg/D) and 23 Mg/D) have more significant increases.
Maximum price increases are particularly large for borosilicate incandescent
bulb blanks, borosilicate tubing, and TV envelope tubes, reaching 11.6, 8.2,
and 13.3 percent respectively. In the case of TV envelope tubes, it is more
likely that the control costs will be absorbed by the producers, rather than
passed on to consumers, because of the competitive role of imports. No
information is available about the likely absorption of costs for the other
two products. However, none of the actual 19 furnaces that are expected to
have to install add-on controls as a result of the regulation produce these
three products.
Profit impacts also vary substantially with furnace size. At the
smallest furnace size (4.5 Mg/D or 23 Mg/D) and 25 percent capacity utiliza-
tion, the results for all products but optical glass indicate a furnace
closure or a product elimination as a potential result. On the other hand,
for large furnaces (45 Mg/D through 636 Mg/D ), the profit impacts are
generally small. Just as with the case of maximum price increases, three
products show the greatest potential for closure at small furnace sizes.
They are borosilicate incandescent bulb blanks, borosilicate tubing, and TV
envelope tubes. However, none of the actual 19 furnaces that are expected to
have to install add-on controls as a result of regulations produce these
three products.
There are no major problems with raising the necessary capital for
either of the two firms requiring controls. Although control capital costs
are not trivial sums, the affected firms are large and should be able to
raise the necessary capital.
7.2.3 Methodology
Several aspects of the methodology for the evaluation of impacts
should be mentioned. One is that prices per unit weight are difficult to
establish for the products in SIC 3229. Prices per unit weight are not only
not published in standard references such as Current Industrial Reports or
the American Glass Review, but they are not readily available from industry
sources. With the exception of TV envelope tubes, prices per unit weight in
SIC 3229 are based on 1978 information and updated to October, 1982, by using
the producer price indexes for container and flat glass.
7-24
-------
A second aspect of the methodology that should be noted is that all of
the companies involved in the analysis are large firms with multiple products.
As a result, pre-control financial ratios for individual products are not
available on an isolated basis. The analysis of profit impacts is, there-
fore, based on business segments containing the affected product but also
other products.
Third, several products may be produced by a glass melting furnace,
however not all of these products contain arsenic. For example, in the case
of handmade consumer glass, crystal stemware typically uses arsenic, but
other glass products produced in the same furnace do not. As a result, this
analysis uses capacity utilization rates of 25 and 50 percent (as well as
100%) to take into account the fact that not all products produced in a given
furnace use arsenic. The assumption is made in these cases of 25 and 50
percent capacity utilization that all of the control costs will be assigned
only to the arsenic-containing products and not to all products manufactured
in the same furnace. An alternative assumption is that control costs may be
distributed among all products produced in the same furnace whether or not
they use arsenic. This alternative assumption would reduce the price and
profit impacts reported in Sections 7.2.4.1 and 7.2.4.2.
A fourth aspect of the methodology is that model plant furnaces are
used to evaluate the economic impact of the regulatory alternatives. Although
model plant furnaces include characteristics that represent worst case situa-
tions of actual furnaces, they are not designed to duplicate exactly any
actual furnace. In fact, as mentioned previously, only some of the model
furnaces are representative of the 32 furnaces that have been identified as
producing arsenic-containing glass.
Finally, data on specific companies have been used to derive pre-control
financial characteristics of the industry for purposes of the impact analysis.
Standard references for industry financial data, such as Robert Morris Associ-
ates, are not available for the glass industry. As a result it is not possible
to cross check information to the degree that would typically be desirable.
7.2.4 Results
7.2.4.1 Price Impacts. In order to evaluate the maximum impact that
control costs can have on prices, a simplifying assumption is made whereby
7-25
-------
the plants increase their prices of arsenic containing glass to the point
where all control costs are passed through to their customers. Maximum price
increases are thus a "worst case" from the point of view of the consumer.
One can approximate the maximum percentage increase in price by expressing
the annualized costs of control as a percent of sales (revenues) before
controls. A complete pass forward of control costs may not be possible in
every case, and later this assumption is relaxed when profit impacts are
evaluated. However, assuming a complete pass forward is possible in every
case introduces a common reference point, which then facilitates comparisons
of various alternatives and scenarios.
The above approach does not consider the investment tax credit, and
therefore is a conservative approach that will tend to overstate the impact
of the control costs. Other approaches can also be used to determine price
increases. For example, a net present value (NPV) approach could be used.
In an NPV approach the revenue increase necessary to exactly offset the
control costs is determined such that the NPV of the plant remains constant.
An NPV analysis can also take into account the investment tax credit, depreci-
ation over the applicable time period, income taxes, the time value of money
as well as capital costs, and operating and maintenance (O&M) costs. Al-
\
though the NPV approach is a more complicated calculation that takes into
account more variables than the simplified approach described above, the two
approaches yield similar results. The method described above is preferable
in this case due to its straightforward approach, ease of presentation, and
reasonable results.
Table 7-9 summarizes the maximum percent price increases for all
products and each control option of the regulatory alternative that calls for
standards based on the use of an ESP or fabric filter. The effects on prices
of alternative furnace sizes and rates of capacity utilization are also
presented. The results shown on Table 7-9 are discussed on an individual
product basis and in more detail in the sections below.
7.2.4.1.1 Soda lime incandescent bulb blanks. The maximum price
increases for soda lime incandescent bulb blanks are below 2 percent for the
7-26
-------
—I
I
Table 7-9
MAXIMUM PERCENT PRICE INCREASES
Model Plant Size, Mg/Day (Tons/Day)
Control
Product Option3
Incandescent bulb
blanks (soda lime)
Incandescent bulb
blanks (borosilicate)
Optical glass
Handmade consumer
ware
Machine-made
consumer ware
Tubing (borosilicate)
TV envelope tubes
Flat glass
Container glass
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
4.5
(5)
23
(25)
Capacity Capacity
Utilization0 at Utilization at
25% 50% 100% 25% 50% 100%
_c
11.
10.
2.7 1.3 0.7 1.
3.5 1.7 0.9 1.
3.9 2.0 1.0 1.
5.1 2.5 1.3 1.
3.
3.
8.
7.
13.
12.
-
-
-
— — — -
6
9
2
1
7
6
3
1
2
7
3
4
-
5.8
5.5
0.6
0.6
0.9
0.8
1.7
1.6
4.1
3.8
6.6
6.2
-
-
_
-
-
2.9
2.7
0.3
0.3
0.4
0.4
0.8
0.8
2.1
1.9
3.3
3.1
_
-
_
-
45
(50)
Capacity
Utiliza-
tion at
70% 100%
3.5
3.1
2.5
2.2
0.3
0.2
0.4
0.3
0.7
0.6
1.7
1.6
2.8
2.5
_
-
_
-
2.5
2.2
1.7
1.5
0.2
0.2
0.3
0.2
0.5
0.4
1.2
1.1
2.0
1.8
_
-
_
-
91
(100)
181
(200)
Capacity Capacity
Utiliza- Utiliza-
tion at tion at
70% 100% 70% 100%
2.5
2.1
1.8
1.5
0.2
0.2
0.3
0.2
0.5
0.4
1.2
1.1
2.0
1.7
—
-
_
-
1.
1.
1.
1.
0.
0.
0.
0.
n.
0.
0.
0.
i.
i.
—
-
—
7
5
2
0
1
1
2
2 - -
4 0.5 0.3
3 0.3 0.2
9
7 - -
4
2
_
_ _
3.5 2.4
3.0 2.1
636
(700)
Capacity
Utiliza-
tion at
70% 100%
-
_
— —
— _
_ —
_ (—
1.4 1.0
2.7 1.9
. „__ cooler followed by a fabric fil
an electrostatic precipitator (ESP).
b!00 percent capacity utilization is 8,400 hours/year
c- = not applicable.
ter (FF). ESP = evaporative gas cooler followed by
, or 350 days/year.
-------
91 Mg/D furnace at 100 percent capacity utilization, and 2.1 to 3.5 percent
in all other cases. It is likely that the control costs will be passed
forward to the consumer, although there is a small but growing import
market for soda lime incandescent bulb blanks. None of the furnaces that
have been identified as using arsenic are known to produce soda lime in-
candescent bulb blanks.
7.2.4.1.2 Borosilicate Incandescent bulb blanks. The maximum price
increases for borosilicate incandescent bulb blanks vary from a low of 1.0
percent (at 91 Mg/D and 100 percent capacity utilization) to a high of 11.6
percent (23 Mg/D and 25 percent capacity utilization). The price impacts for
the 91 Mg/D furnace, the 45 Mg/D furnace and the 23 Mg/D furnace at 100
percent capacity utilization are less than 3 percent, whereas those for the
23 Mg/D furnace at 50 or 25 percent capacity utilization vary from 5.5
percent to 11.6 percent. It is unknown how likely it is for these costs to
be passed forward to the consumer because the specialized nature of the
product makes it impossible to evaluate the industry from published data.
There is, for example, no information available on the role of imports.
Borosilicate bulb blanks include high and low pressure street lamps and
mercury vapor lamps for automobiles.^ These wide differences in maximum
percent price increases for 91 Mg/D versus 23 Mg/D may mean the 23 Mg/D plant
will have to absorb costs to compete with 91 Mg/D furnace. Of the 19 actual
furnaces that are expected to have to install controls (13 already have
add-on controls in place), none produces borosilicate bulb blanks.
7.2.4.1.3 Optical glass. The maximum price increases for optical
glass are less than 2 percent except for the cases of the smallest furnace
sizes. In the case of the 4.5 Mg/D furnace and 25 percent capacity utiliza-
tion, the maximum price increases vary from 2.7 (ESP) to 3.5 (FF) percent.
Because this industry contains only a few firms, and the product is specialized,
it is likely these price increases will be passed on to the consumer.
Mitigating this conclusion is the fact that plastics are partial substitutes
for optical glass.
7.2.4.1.4 Handmade consumer glass. The maximum price increases for
handmade consumer glassware are also less than 2 percent except for the cases of
7-28
-------
the smallest furnace sizes. In the case of a 4.5 Mg/D furnace at 25 percent
capacity utilization, the maximum price increase is 3.9 percent (ESP) or 5.1
percent (FF). It is unlikely that all of this price increase will be passed
on to consumers because the domestic industry is highly competitive. In
addition, handmade consumer glassware manufacturers are threatened by competi-
tion from machine-made glassware and from imports. The likelihood of pass-
through versus absorption influences an evaluation of percent price increase.
7.2.4.1.5 Machine-made consumer glass. The maximum price increases
for machine-made consumerware are less than 2 percent for the larger furnaces
(45 Mg/D through 181 Mg/D)), but they are of some importance in the case of
the 23 Mg/D furnace operating at 25 percent of capacity utilization, for
which price increases vary from 3.1 (FF) percent to 3.3 (ESP) percent. It is
unlikely that all of control costs will be passed on to the consumer or
applied to only the arsenic requiring glass, however, because of the role of
imports. Most of the existing furnaces that have been identified as using
arsenic produce machine-made consumer glass.
7.2.4.1.6 Borosi1icate tubing. The maximum price increases for
borosilicate tubing vary from 0.7 percent at 91 Mg/D furnaces at 100 percent
capacity utilization to as much as 8.2 (ESP) percent or 7.7 (FF) percent at 23
Mg/D and 25 percent capacity utilization. It is impossible to evaluate the
likelihood of the control costs being passed on to the consumer, because
borosilicate tubing is said to cover hundreds of end-products for which data
are unavailable. Among the uses of borosilicate tubing are: scientific
glassware; pharmaceutical glassware; process tubing for the food and dairy
industry; bulbs for microwave ovens; electronic tubes; encapsulators for
semiconductors; photomultiplier tubes; ultraviolet light transmitting tubing;
photoflash bulbs; blueprint cylinders; exhaust tubing for seal-beam headlights;
and control rods for nuclear reactors.77 The wide differences in the
percent increase for 91 Mg/D versus 23 Mg/D may mean the 23 Mg/D furnaces
will have to absorb control costs to compete with the 91 Mg/D furnace. Of
the 19 actual furnaces that are expected to have to install controls (13
already have add-on controls in place), none produces borosilicate tubing.
7-29
-------
7.2.4.1.7 TV envelope tubes. The maximum price increases for TV
envelope tubes are 2 percent or less for the 45 Mg/D and 91 Mg/D furnace but
they may be as high as 13.3 percent (ESP) or 12.4 percent (FF) for the 23
Mg/D furnace at 25 percent capacity utilization. It is unlikely the costs of
control will be passed on to the consumer because of the substantial role of
imports, particularly from Japan. These wide differences in the percent
increase for the 91 Mg/D furnace and the 23 Mg/D furnace may mean the 23 Mg/D
furnace will have to absorb control costs or distribute costs to the furnaces
entire product line to compete with the 91 Mg/D furnace. Of the 19 actual
furnaces that are expected to install controls (13 already have add-on
controls in place), none produces TV envelope tubes.
7.2.4.1.8 Flat glass. The maximum price increases are less than 2
percent for flat glass at 100 percent capacity utilization and for the ESP
option at 70 percent but are 2.7 percent at 70 percent capacity utilization
with the FF option. It is unlikely the costs of control will be passed on to
the consumer because imports play a role in this market. However, because
arsenic is not now used in the production of flat glass, any arsenic emission
regulation would not affect this industry segment.
7.2.4.1.9 Container glass. The maximum price increases for container
glass vary from a low of 2.1 percent for the FF option at 100 percent
capacity utilization to a high of 3.5 percent for the ESP option at 70
percent capacity utilization. It is likely the costs of control will be
passed on to the consumer, because imports do not play a role in this market.
None of the 32 furnaces identified as using arsenic produces container
glass.
7.2.4.2 Profit Impacts. The estimation of maximum profit impacts is
based on the opposite assumption of the maximum price increase; namely, that
plants are unable to increase their prices in order to cover the control
costs. It is thus a "worst case" from the point of view of the firm. In
this situation all control costs are treated as additions to baseline oper-
ating expenses. If prices are not increased, net sales remain constant
resulting in a reduction of profit before taxes.
The extent to which profit is reduced can be seen by comparing the
baseline before tax return on sales with before tax return on sales after
controls. Return on sales is defined as profit before tax divided by total
7-30
-------
revenue. Before tax profits are used rather than after tax profits to avoid
the complication of varying tax rates.
Table 7-10 to 7-17 presents background financial data for eight firms:
Corning Glass, RCA, Anchor Hocking, GTE, Owens-Illinois, General Electric,
PPG and Libbey-Owens-Ford. Although General Electric is not of principal
concern, it is included here because it is a major manufacturer of incan-
descent bulb blanks and therefore its financial results are likely to add
perspective to the analysis. PPG and Libbey-Owens-Ford are included to
provide financial data on the flat glass industry.
Table 7-18 summarizes Table 7-10 to 7-17 about the existing ratios of
before tax profits to sales for each product category. The assumptions were
developed by combining the appropriate business segments in Tables 7-10 to
7-17. For example, the ratio of profits to sales for machine-made consumer
glassware is an average of the ratio of profits to sales for that business
segment at Anchor Hocking and at Corning Glass (using both the unadjusted and
adjusted figures). The averages are unweighted due to data limitations.
Several years (3 or 4) are used to be more representative. Business segment
information is the best financial information available and is believed to be
quite reasonable. However, other products in addition to the one under
analysis are included in business segments.
Table 7-19 presents the results of the profit imports, namely, the
percent change in before tax profits to sales for all products and each
control option of the regulatory alternative that calls for standards based
on the use of an ESP or fabric filter. The effects on profits of alternative
furnace sizes and rates of capacity utilization are also presented. The
results shown on Table 7-19 are discussed on an individual product basis and
in more detail in the sections below.
7.2.4.2.1 Soda lime incandescent bulb blanks. The profit impacts for
soda lime incandescent bulb blanks vary from 11.1 percent for the 91 Mg/D
furnace at 100 percent capacity utilization (FF) to 26.4 percent (ESP) for
the 45 Mg/D furnace at 70 percent capacity utilization. As indicated in the
discussion of price impacts, it is more likely that control costs will be
passed on to the consumer rather than being absorbed by the firm although
there is currently a small market for imported bulb blanks. None of the
furnaces that have been identified as using arsenic are known to produce soda
lime incandescent bulb blanks.
7-31
-------
Table 7-10
FINANCIAL CHARACTERISTICS OF CORNING GLASS WORKS, 1978 TO 198174»75
($ 106)
Year
Company
Total3
Machine-made
and Handmade
Consumer Glassware1'
Picture Tube Envelopes
and Incandescent
Light Bulb Blanks
Optical Glass
Unadjusted Adjusted0 Unadjusted Adjusted0 Unadjusted Adjusted0
Sales
1978
1979
1980
1981
1,251.7
1,421.6
1,529.7
1,598.5
331.1
357.1
374.5
428.8
373.7
423.6
443.7
416.8
245.4
291.9
350.6
374.4
Profits Before Taxes
1978
1979
1980
1981
120.6
114.4
85.1
57.6
63.8
46.6
32.5
57.5
31.1
16.4
-.29
17.1
91.5
85.6
8.42
65.3
54.6
49.8
45.4
26.0
31.8
49.4
60.7
66.9
7.6
24.7
30.0
31.6
1978
1979
1980
1981
1,224.4
1,385.2
1,499.6
1,613.4
167.5
197.6
224.6
241.6
317.5
357.7
383.9
422.9
Assets3
186.3
181.0
199.1
204.2
355.6
371.0
387.8
380.4
148.4
158.8
204.8
228.6
259.6
289.7
353.9
386.8
7-32
-------
Table 7-10
(Continued)
Year
Company
Total3
Machine-made
and Handmade
Consumer Glassware13
Unadjusted Adjusted0
Picture Tube Envelopes
and Incandescent
Light Bulb Blanks
Unadjusted Adjusted0
Optical
Unadjusted
Glass
Adjusted0
1978
1979
1980
1981
Average
9.63
8.05
5.56
3.60
6.71
19.27
13.05
8.68
13.41
13.60
9.4
4.6
-0.08
3.98
4.48
Return on Sales
(percent)
24.48
20.21
18-98
15.67
19.84
14.61
11.75
10.22
6.24
10.71
12.96
16.92
17.31
17.87
16.27
3.08
8.45
8.55
8.44
7.13
Before Tax Return on Investment
1978
1979
1980
1981
Average
9.85
8.26
67
57
38.09
23.58
14.47
23.80
6.84 24.99
9.81
4.60
-0.08
4.04
4.59
(percent)
49.11
47.29
42.29
31.98
42.67
15.35
13.42
11.70
6.83
11.84
21.43
31.11
29.64
29.27
27.86
2.92
8.52
8.47
8.17
7.02
Sales to Assets
(ratio)
1978
1979
1980
1981
Average
1.02
1.03
1.02
.99
1.02
1.98
1.81
1.67
1.77
1.81
1.04
1.00
1.00
1.01
1.01
2.01
2.34
2.23
2.04
2.16
1.05
1.14
1.14
1.10
1.11
1.65
1.84
1.71
1.64
1.71
.95
1.01
.99
1.00
1.00
aBusiness segments shown do not sum to the company total because not all of the
company's segments are shown here.
bBased on the business segment which includes the product under analysis as well
as other products.
°In the adjusted figures, the general corporate income from operations and assets
are distributed among business segments based on their proportion of sales.
7-33
-------
Table 7-11
FINANCIAL CHARACTERISTICS OF RCA. 1978 TO
($ 106)
Year
1978
1979
1980
1981
TV
Company Total Tube
Sales
6,600.6
7,454.6 1,
8,001.3 1,
8,004.8 1,
Picture
Envelopes
962.8
113.0
280.5
196.4
Profits Before Taxes
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
514.6
471.5
507.3
98.3
Assets
N/Aa
6,058.3
7,147.6
7,856.7
Return on Sales
(percent)
7.80
6.32
6.33
1.23
5.42
Before Tax Return
on Investment
(percent)
N/A
7.78
7.10
1.25
5.38
Sales to Assets
(ratio)
N/A
1.23
1.12
1.02
1.12
103.4
97.1
95.7
-55.7
N/A
N/A
N/A
N/A
10.74
8.72
7.47
-4.66
5.57
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
aN/A = not available.
7-34
-------
Table 7-12
FINANCIAL CHARACTERISTICS OF ANCHOR HOCKING
Year
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
CORPORATION,
($
Company Total
716.8
785.2
857.5
953.4
63.9
53.7
44.6
48.9
473.5
517.0
559.1
584.4
8.91
6.84
5.20
5.13
6.52
13.50
10.39
7.98
8.37
10.06
1.51
1.52
1.53
1.63
1.55
1978 TO 198183'83
106)
Machine-Made
Consumer Glass
Sales
220.3
269.7
326.5
364.8
Profits Before Taxes
25.4
34.5
38.6
31.1
Assets
143.4
207.7
246.0
259.8
Return on Sales
(percent)
11.53
12.79
11.82
8.53
11.17
Before Tax Return
on Investment
(percent)
17.71
16.61
15.69
11.97
15.50
Sales to Assets
(ratio)
1.54
1.30
1.33
1.40
1.39
Containers
391.4
411.6
441.0
485.8
39.4
27.0
22.0
30.4
209.8
216.4
231.9
227.8
10.07
6.56
4.99
6.26
6.97
18.78
12.48
9.49
13.35
13.53
1.87
1.90
1.90
2.13
1.95
7-35
-------
Table 7-13
FINANCIAL CHARACTERISTICS OF GTE. 1978 TO 198184
{$ 106)
Year
1978
1979
1980
1981
Tubing for
Company Total Fluorescent Lights
Sales
N/Aa
N/A
9,979
11,026
1,395
1,632
1,905
1,830
Profits Before Taxes
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
Average
N/A
N/A
1,964
2,229
Assets
N/A
N/A
19,525
21,113
Return on Sales
(percent)
N/A
N/A
19.68
20.22
19.95
211
233
243
197
1,029
1,178
1,306
1,348
15.13
14.28
12.76
10.77
13.24
Before Tax Return
on Investment
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
aN/A = not available.
(percent)
N/A
N/A
10.06
10.56
20.51
19.78
18.61
14.61
10.31
18.38
Sales to Assets
(ratio)
N/A
N/A
.51
.52
1.36
1.39
1.46
1.36
.51
1.39
7-36
-------
Table 7-14
M
FINANCIAL CHARACTERISTICS OF OWENS-ILLINOIS, 1978 TO 198185>86
Year
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
($
Company Total
3,111.7
3,504.3
3,905.7
3,943.3
166.5
206.1
261.0
271.9
2,600
2,910
3,066
3,072
5.35
5.88
6.68
6.90
6.20
6.40
7.08
8.51
8.85
7.71
1.20
1.20
1.27
1.28
1.24
106)
TV Picture
Tube Envelopes
Sales
246
288
368
378
Profits Before Taxes
19.7
32.5
48.7
37.7
Assets
277
297
346
375
Return on Sales
(percent)
8.01
11.28
13.23
9.97
10.62
Before Tax Return
on Investment
(percent)
7.11
10.94
14.08
10.05
10.55
Sales to Assets
(ratio)
.89
.97
1.06
1.01
.98
Containers
2,481
2,767
3,010
3,091
155.2
186.2
253.0
266.1
1,775
1,998
1,997
2,060
6.26
6.73
8.41
8.61
7.50
8.74
9.32
12.67
12.92
10.91
, 1.39
1.38
1.51
1.50
1.45
7-37
-------
Table 7-15
FINANCIAL CHARACTERISTICS OF GENERAL ELECTRIC,
Year
1979
1980
1981
1979 TO 198187
($ 106)
Incandescent
Company Total Bulb Blanks
Sales
22,980 5
25,523 6
27,854 6
,990
,342
,643
Profits Before Taxes
1979
1980
1981
1979
1980
1981
1979
1980
1981
Average
2,391
2,493
2,660
Assets
16,644 2
18,511 2
20,942 2
Return on Sales
(percent)
10.4
9.0
9.6
9.7
617
615
549
,500
,656
,926
10.3
9.7
8.3
9.4
Before Tax Return
on Investment
1979
1980
1981
Average
(percent)
14.4
8.7
12.7
11.9
24.7
23.2
18.8
22.2
Sales to Assets
1979
1980
1981
Average
(ratio)
1.38
1.38
1.33
1.36
2.40
2.38
2.27
2.35
7-38
-------
Table 7-16
FINANCIAL CHARACTERISTICS OF PPG 1978-198188»89
($ 106)
Year
1978
1979
1980
1981
Company Total Flat Glass
Sales
2,794
3,092 1
3,158 1
3,354 1
995
,097
,174
,210
Profits Before Taxes
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979"
1980
1981
Average
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
244
391
371
352
Assets
2,324
2,955
2,826
2,637
Before Tax
Return on Sales
(percent)
8.7
12.7
11.8
10.5
10.9
Before Tax Return
on Investment
(percent)
10.5
13.2
13.1
13.4
12.6
Sales to Assets
(ratio)
1.20
1.05
1.12
1.27
1.16
167
174
143
113
659
724
879
897
16.8
15.9
12.2
9.3
13.6
25.3
24.0
16.3
12.6
19.6
1.51
1.52
1.34
1.35
1.43
7-39
-------
Table 7-17
FINANCIAL CHARACTERISTICS OF LIBBEY-OWENS-FORD 1978-198190
($ 106)
Year
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
1978
1979
1980
1981
Average
Company Total Flat Glass
Sales
1,107.1 617.9
1,208.1 630.5
1,159.9 559.6
1,226.5 598.7
Profits Before Taxes
112.3 49.2
112.4 44.7
67.9 -6.2
73.8 -14.0
Assets
726.6 338.7
846.0 424.3
903.8 425.9
893.0 389.1
Before Tax
Return on Sales
(percent)
10.1 8.0
9.3 7.1
5.9 -1.1
6.0 -2.3
7.8 2.9
Before Tax Return
on Investment
(percent)
15.5 14.5
13.3 10.5
7.5 -1.2
8.3 -3.6
11.2 5.1
Sales to Assets
(ratio)
1.52 1.82
1.43 1.49
1,28 1.31
1.37 1.54
1.40 1.54
7-40
-------
Table 7-18
SUMMARY OF FINANCIAL RATIOS FOR PROFIT IMPACTS
Before Tax
Profits
to Sales3
Product (Percent)
Incandescent bulb blanks 13.3
(soda lime)
Incandescent bulb blanks 13.3
(borosilicate)
Optical glass 11.7
Handmade consumer ware 9.0
Machine-made consumer 9.8
ware
Tubing (borosilicate) 6.7
TV envelope tubes 11.7
Flat glass 8.3
Container glass 7.2
aWith one exception, the financial ratios are
averages for the appropriate business segments
in the firms producing that product. In the
case of borosilicate tubing, however, the
ratios are averages for Corning Glass as a
whole, because borosilicate tubing is produced
in each of Coming's business segments.
7-41
-------
I
J2,
ro
Table 7-19
PROFIT IMPACTS: PERCENT CHANGE IN BEFORE TAX PROFITS ON SALES - AFTER CONTROLS
Model Plant Size, Mg/Day (Tons/Day)
4.5 23
(5) (25)
Product
Incandescent bulb blanks
(soda Hme)
Incandescent bulb blanks
(borasilicate)
Optical glass
Handmade consumer ware
Machine-made consumer
ware
Tubing (borasilicate)
TV envelope tubes
Flat glass
Container glass
Control
Option9
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
ESP
FF
Capacity Capacity
Utilization atb Utilization at
25* 50* 100% 2S%
_c
_
87.2
81.9
23.1 11.1 6.0 10.3
29.9 14.5 7.7 9.4
43.3 22.2 11.1 20.0
56.7 27.8 14.4 18.9
33.7
31.6
- 122.4
- 113.4
- 113.7
- 106.0
.
.
.
50*
-
-
43.5
40.6
5.1
5.2
10.0
8.9
17.3
16.3
61.2
58.2
56.4
53.0
_
_
-
100*
-
-
21.8
20.3
2.6
2.6
4.4
4.4
8.2
8.2
31.3
30.0
28.2
26.5
-
-
-
45
(50)
91
(100)
Capacity util-
ization at
70S
26.4
23.4
18.8
16.5
1.7
1.7
4.4
3.3
7.1
6.1
25.4
23.9
23.9
20.5
_
-
-
100*
18.5
16.4
12.8
11.3
1.7
1.7
3.3
2.2
5.1
4.1
17.9
16.4
17.1
15.4
_
-
-
Capacity util-
ization at
70*
18.7
15.9
13.5
11.3
1.5
0.9
3.3
2.2
5.1
4.1
17.9
16.4
17.1
14.5
_
-
-
100*
13.1
11.1
9.8
8.3
0.9
0.9
1.1
2.2
3.1
3.1
13.4
10.4
12.0
10.3
„
-
-
181
(200)
Capacity util-
ization at
70*
-
-
-
-
-
-
-
5.1
3.1
-
-
-
-
_
48.6
41.7
100*
-
-
-
-
-
-
-
4.1
2.0
-
-
-
-
-
33.3
29.2
636
(700)
Capacity Util-
ization at
70* 100*
-
_ —
-
- -
-
.
— "•
-
- -
-
— **
-
- -
16.9 12.0
32.5 22.9
-
*• **
aFF = evaporative gas cooler followed by a fabric filter
tator (ESP).
b!00 percent capcity utilization is 8,400 hours/year, or 350 days/year.
c- = not applicable.
(FF). ESP = evaporative gas cooler followed by an electrostatic preclpi-
-------
7.2.4.2.2 Borosilicate bulb blanks. The profit impacts on borosili-
cate bulb blanks vary from 8.3 percent for the 91 Mg/D furnace at 100 percent
capacity utilization (FF) to 87.2 percent for the 23 Mg/D furnace at 25
percent capacity utilization (ESP). As indicated in the discussion of price
impacts, it is impossible to evaluate the likelihood of profit absorption in
the product category, because no information is available on the role of
imports. The situation with respect to furnaces sizes is similar to that for
price impacts. The wide differences in profits impacts for the 91 Mg/D
furnace versus the 23 Mg/D furnace may put the 23 Mg/D furnace at a competi-
tive disadvantage. Of the 19 actual furnaces that are expected to install
controls (13 already have add-on controls in place), none produces borosi-
licate bulb blanks.
7.2.4.2.3 Optical glass. The profit impacts vary from 0.9 percent
for the 91 Mg/D furnace at 100 percent capacity utilization to 29.9 percent
for the 4.5 Mg/D furnace at 25 percent capacity utilization. As indicated
in the discussion of price impacts, it is more likely that control costs will
be passed on to the consumer rather than being absorbed by the firm although
the role of plastic substitutes cannot be discounted.
7.2.4.2.4 Handmade consumer glass. The profit impact for handmade
consumer glass varies from 1.1 percent for the 91 Mg/D furnace at 100 percent
capacity utilization to 43.3 percent (ESP) and 56.7 percent (FF) for the 4.5
Mg/D furnace at 25 percent capacity utilization. As indicated in the section
on price impacts, it is likely that the costs of control will be absorbed by
producers because of competition from imports and from machine-made consumer
ware.
7.2.4.2.5 Machine-made consumerware. The profit impacts on machine-
made consumerware manufacturers vary from 2.0 percent for 181 Mg/D furnaces
at 100 percent capacity utilization to 33.7 percent for the 23 Mg/D furnace
at 25 percent capacity utilization. As indicated in the section on prices,
it is likely producers will have to absorb costs of control because of the
role of imports in this market. Most of the existing furnaces that have been
identified as using arsenic produce machine-made consumer glass.
7-43
-------
7.2.4.2.6 Borosi11cate tubing. The profit impacts for borosilicate
tubing varies from 10.4 percent for the 91 Mg/D furnace at 100 percent
capacity utilization to 113.4 (FF) and 122.4 (ESD) percent for the 23 Mg/D
plant at 25 percent capacity utilization. For the latter cases, closures are
almost certain because when profit declines more than 100 percent, that
means a net loss. As indicated in the section on price impacts, it is
impossible to evaluate whether control costs will be passed on to the con-
sumer or absorbed by the firm because of lack of information on the indivi-
dual products in the borosilicate tubing category. The situation with
respect to furnace sizes is similar to that for price impacts. The wide
differences in profit impacts for the 45 Mg/D and 91 Mg/D furnace and the 23
Mg/D furnace may put the 23 Mg/D furnace at a competitive disadvantage. Of
the 19 actual furnaces that are expected to have to install controls (13
already have add-on controls in place), none produces borosilicate tubing.
*
7.2.4.2.7 TV envelope tubes. The impact on profits varies from 10.3
percent for the 91 Mg/D furnace at 100 percent capacity utilization to 113.7
percent for the 23 Mg/D at 25 percent capacity utilization. When profits
decline more than 100 percent, that means there is a net loss from opera-
tions, hence, closure is almost certain. As indicated in the discussion of
price impacts, absorption of control costs by TV envelope tube producers is
likely because of the role of imports. The situation with respect to furnace
sizes is similar to that for price impacts. The wide differences in profit
impacts for the 91 Mg/D furnace and the 23 Mg/D furnace may put the 23 Mg/D
furnace at a competitive disadvantage. Of the 19 actual furnaces that are
expected to install controls (13 already have add-on controls in place), none
produces TV envelope tubes.
7.2.4.2.8 Flat glass. The profit impact on flat glass for the 636
Mg/D furnace with the ESP option varies from 12.0 to 16.9 percent, whereas
the profit impact with the FF option varies from 22.9 to 32.5 percent.
However, it is expected that the least expensive control device would be
installed. As indicated in the section on prices, it is likely these profit
impacts will have to be absorbed by producers because of the role of imports.
However, because arsenic is not now used in the production of flat glass, the
regulation would not affect this industry segment.
7-44
-------
7.2.4.2.9 Container glass. The profit impacts for the 181 Mg/D
furnace vary from 29.2 percent (ESP) at 100 percent capacity utilization to
48.6 percent (ESP) at 70 percent capacity utilization. It is unlikely,
however, that these control costs will be absorbed by producers since the
high weight to volume of containers precludes imports. None of the 32
furnaces identified as using arsenic produces container glass.
7.2.4.3 Capital Availability. The final topic that must be consid-
ered is the ability of the plants to finance the capital costs associated
with the installation of add-on controls. In general, profitability will
determine capital availability, but an explicit discussion of capital avail-
ability provides additional insight.
An examination of the long term debt to total capitalization (long
term debt plus equity) ratio for the plant provides a measure of capital
availability. A high ratio usually indicates that further debt financing may
not be available. This percentage can then be reviewed for the firm over a
period of several years as well as compared with debt ratios of similar firms
in the industry. If it is assumed that the control costs are financed solely
with debt, then the impact of control costs can be evaluated by adding the
control costs to both the numerator and denominator. This calculation
results in the debt ratio after control costs. This ratio can then be
compared to the debt ratio before control costs and an evaluation made with
respect to capital availability problems.
Table 7-20 summarizes the pre-control ratio of long-term debt to total
capitalization for the five firms identified as currently producing glass
using arsenic. All of the five firms are major publicly-held corporations.
As this table shows, the precontrol ratio varies from 16.2 percent for
Corning Glass to 61.6 percent for GTE. Only two of these firms, Corning
Glass and Anchor Hocking, have arsenic-using furnaces that would need to
install add-on controls.
Table 7-21 presents the post-control situation for Corning and Anchor
Hocking based on controls costs estimated for a "worst case" (largest
capital cost), namely, that all furnaces are large (91 Mg/D). Even in the
case of Anchor Hocking, however, capital availability does not appear to be
overly burdensome factor for two reasons: (1) it has a low ratio of long-
term debt to total capitalization even with controls, and (2) the ratio
after controls (27.7% and 25.7%) is below three other companies in the
industry.
7-45
-------
Table 7-20
RATIO OF LONG-TERM DEBT TO TOTAL CAPITALIZATION (PRE-CONTROL)91"9?
Corning
Glass
Owens-
Illinois
RCA
Anchor
GTE Hocking
Long-Term Debt
1978
1979
1980
1981
Average
163
147
153
200
166
.4
.1
.6
.1
.0
691.
697.
687.
618.
673.
3
($
1
1
1
1
106)
N/Aa
,474.
,771.
,855.
,700.
2
2
8
4
7
7
7
N/A
N/A
,468.
,978.
,723.
N/A
86.5
87.5
85.0
86.
3
Total Capitalization
(Long-Term Debt Plus Equity)
1978
1979
1980
1981
Average
905
975
1,066
1,165
1,028
.1
.1
.3
.5
.0
1,797.
1,898.
2,003.
2,033.
1,932.
8
Ratio of
Total
3
4
4
3
N/A
,234.
,182.
,095.
,837.
0
6
7
4
12
13
N/A
N/A
,016.
,049.
12,532.5
N/A
363.6
380.6
393.4
379.
2
Long Term Debt to
Capital ization
(Percent)
1978
1979
1980
1981
Average
18
15
14
17
16
.1
.1
.4
.2
.2
38.
36.
34.
30.
34.
5
7
3
4
8
N/A
45.
42.
45.
44.
6
4
3
3
N/A
N/A
62.2
61.1
61.6
N/A
23.8
23.0
21.6
22.
8
aN/A - not available.
7-46
-------
Table 7-21
RATIO OF LONG TERM DEBT TO TOTAL CAPITALIZATION (POST CONTROL)
Company
Anchor Hocking
Corning Glass
GTE
Owens-Illinois
RCA
Number of
Affected Furnaces
10
9
0
0
0
Ratio of Long
Before Centre
22.8
16.2
61.6
34.8
44.3
Term Debt to Total
After Controls3
)ls ESP FF
27.7 25.7
18.0 17.2
NAb NA
NA NA
NA NA
Capitalization
Percent Change
ESP FF
21.5 12.7
11.1 6.2
NA NA
NA NA
NA NA
aAssumes the worst case (largest capital cost), namely, all plants are large
(91 Mg/D).
bNot Applicable
7-47
-------
7.2.4.4 Impacts of Ban on Arsenic Emissions. The third regulatory
alternative, that of setting a standard that would ban arsenic emissions, was
evaluated also. This alternative would have an impact on the value of
shipments and employment in SIC 3229 if the 15 plants closed down as a result
of the ban. This is a worst case, because it is possible a plant would
delete an arsenic product but continue to produce other products (non-arsenic).
It is difficult to arrive at an estimate of the impact of the closures
of these 15 plants on the value of shipments in SIC 3229, because there is
insufficient information available on the capacity of each of these plants.
For purposes of arriving at an estimate, it is assumed each of the 32 furnaces
in the 15 plants has a capacity of 45 Mg/D and revenues are calculated for 45
Mg/D furnaces at 100 percent capacity utilization. This capacity means there
is $1,900.6 million in revenues (value of shipments) in the plants. As Table
7-1 showed, SIC 3229 had $2,847.5 million dollars of shipments in 1980.
Under the assumptions discussed above, value of shipments in SIC 3229 would
be reduced by 66.7 percent as a result of the closures.
Assuming a proportional relationship between employment and value of
shipments, the effect of the plant closures on SIC 3229 can also be estimated.
As Table 7-2 presented, employment in this industry was 45,700 employees. A
66.7 percent decline in employment would reduce employment by 30,482 employees
(or almost 2,000 per plant on an average basis).
If there is a ban on arsenic emissions from glass manufacturing and
arsenic containing glass products is discontinued, imports can be expected to
increase their market share. This is particularly likely in the handmade
glass sector where imports are already highly competitive with domestically
produced handmade crystal.
7.3 Socio-economic Impact Assessment
7.3.1 Executive Order 12291
The purpose of Section 9.3.1 is to address those tests of macro-economic
impact presented in Executive Order 12291, and, more generally, to assess any
other significant macroeconomic impacts that may result from the NESHAP.
Executive Order 12291 stipulates as "major rules" those that are projected to
have any of the following impacts:
o An annual effect on the economy of $100 million or more.
7-48
-------
o A major Increase in costs or prices for consumers; individual
industries; Federal, State, or local government agencies; or
geographic regions.
o Significant adverse effects on competition, employment, invest-
ment, productivity, innovation, or the ability of U.S.-based
enterprises to compete with foreign-based enterprises in domestic
or export markets.
The fifth year annualized costs associated with a regulation that is
based on the use of add-on controls for a "worst case" of 91 Mg/D furnaces is
$9.4 million. This cost estiamte assumes that 19 of the 32 existing arsenic-
using furnaces would install ESP"s. If the FF option is used as an add-on
control device, the fifth year annual ized costs of 19 furnaces at 91 Mg/D
capacity is $8.0 million. Hence, the costs of compliance are well below
the $100 million which, according to Executive Order 12291, signifies a major
rule.
The potential effects on consumers of a regulation that is based on
the use of add-on controls were addressed in Section 7.2.4.1, and the potential
effects on producers were addressed in Section 7.2.4.2. As the information
in those sections makes clear, there may be a substantial impact on producers
of TV envelope tubes. In the cases of borosilicate tubing and borosilicate
incandescent lights, it is unclear whether the impact will fall on consumers
or producers. However, none of the actual 19 furnaces that are expected to
have to install add-on controls as a result of the regulation produce these
products.
Based on the potential adverse impacts of furnace closure resulting
from a regulation that would ban arsenic emissions, such a regulation would
likely be considered a major rule.
7.3.2 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 measures must be
met, substantial numbers of small businesses and significant impacts, to
require an analysis. If either measure is not met, then no analysis is
required.
7-49
-------
The Small Business Administration (SBA) definition of a small
business for SIC 3229 is a firm that employs 750 persons or less.98
Table 7-22 shows the number of glass plants by number of employees in
the four States of West Virginia, Pennsylvania, Ohio, and Virginia.
Of the 106 plants in these four States, at least 73 or about 70 percent
are considered by SBA's definition to be small firms.
Arsenic usage by these small firms is not known. However, it is
expected that at least some do use arsenic in their batch material and,
therefore, have arsenic emissions. The EPA guidelines state that a
regulatory flexibility analysis is required if it is expected that a
regulation will have a significant impact on more than 20 percent of the
small firms being affected by the regulation. Because the number of
firms that may be affected and the differential impacts that might occur
will depend on the form of the standard, the analysis cannot be presented
in this publication. However, the analysis performed will be included
in the docket for the proposed standards for arsenic emissions from glass
manufacturing plants (A-83-08).
7-50
-------
Table 7-22
PRESSED AND BLOWN GLASS NEC OQ
MANUFACTURING PLANTS IN FOUR STATES*jy
Number of
Empl oyees
1 -
10 -
20 -
50 -
100 -
500 -
>1000
9
19
49
99
499
999
Unknown
Number of
Plants
24
8
12
10
27
13
3
9
Number of Plants Owned
by Small Firmsb
24
8
12
9
20
l*c
0
4C
TOTAL
106
78
Four States are West Virginia, Pennsylvania, Ohio, and Virginia, where
small glassmaking firms are chiefly located.
A small firm is one with less than 750 employees.
It is not known whether these plants are owned by small firms.
7-51
-------
7.4 References
1. U.S. Department of Commerce. Bureau of the Census. Standard Indus-
trial Classification Manual. Washington, D.C. 1972. pp. 136-137,
144.
2. U.S. Department of Commerce. Bureau of the Census. 1980 Annual
Survey of Manufacturers: Value of Product Shipments, p. 18.
3. U.S. Council of Economic Advisors. Economic Report of the President:
1983. Washington, D.C. p. 163.
4. U.S. Department of Commerce. Bureau of the Census. 1977 Census of
Manufacturers: Glass Products. Washington, D.C. p. 32A-5.
5. Goldfarb, J. Owens-Corning Fiberglas. Merill Lynch, Pierce, Fenner
and Smith. September 28, 1982. p. 8.
6. U.S. Department of Commerce. Bureau of the Census. 1980 Annual
Survey of Manufacturers: Statistics for Industry Groups and Indus-
tries. Washington, D.C. p. 16.
7. Radian Corporation. Preliminary Study of Sources of Inorganic Arsenic:
Final. A report prepared for the U.S. Environmental Protection Agency
(EPA 68-02-3513). p. 180.
8. Reference 7. pp. 180, 182.
9. U.S. International Trade Commission. Household Glass. USITC Publi-
cation 841. Washington, D.C. p. 2.
10. Shareef, S.A., Radian Corporation. Memo to files. Emissions data on
glass furnaces at RCA Corporation.
11. Letter from Goebel, G., Kentucky Bureau of Environmental Protection to
Brooks, G. W., Radian Corporation. February 11, 1982.
12. Telecon. Brooks, G. W. Radian Corporation, with Iden, C., Owens-Illinois.
March 3, 1982. Conversation concerning Owens-Illinois Arsenic Emissions.
13. Letter and attachments from Murray, D.E., Anchor Hocking Corporation
to S. A. Shareef, Radian Corporation. March 14, 1983. (Project
Confidential File).
14. Letter and attachments from Cherill, J., Corning Glass Works to G. W.
Brooks, Radian Corporation. April 7, 1982. (Project Confidential
Files).
7-52
-------
15. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1981. p. 2.
16. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1980. p. 3.
17. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1979. p. 3.
18. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1978. p. 3.
19. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1976. p. 4.
20. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1975. p. 4.
21. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1974. p. 2.
22. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1973. p. 2.
23. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1972. p. 2.
24. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Consumer, Scientific, and Technical Glassware. MA-32E.
1971. p. 2.
25. U.S. Environmental Protection Agency. Office of Air Quality Planning
and Standards. Glass Manufacturing Plants: Background Information:
Proposed Standards of Performance. EPA-450/3-79-005a. p. 8-32.
26. Reference 25. p. 8-33.
27. Reference 25. pp. 8-33 to 8-34.
28. Telecon. Mosley, G., Corning Glass, with Timothy, A., JACA. August 17,
1978. Data on SIC 3229.
7-53
-------
29. Photocopies. Lasarski, R., U.S. Bureau of Labor Statistics, to Ando,
F. H.j JACA. February 22, 1983. BLS price indexes for flat and
container glass.
30. Reference 25. p. 8-34.
31. Telecon. Kastner, F., Smith Glass, with Ando, F. H., JACA. February
28, 1983. Handmade pressed and blown consumer glassware.
32. Reference 25. p. 8-35.
33. Reference 25. p. 8-36.
34. Reference 25. pp. 8-96 to 8-98.
35. Telecon. Nelson, J., GTE, with Ando, F., JACA. March 1, 1983. Glass
tubing for fluorescent lighting.
36. Reference 25. p. 8-37.
37. Reference 25. p. 8-38.
38. Reference 25. p. 8-39.
39. Reference 25. p. 8-40.
40. Telecon. Wiley, L., RCA, with Ando, F., JACA. February 25, 1983. TV
picture tube envelopes.
41. Telecon. Sprouse, J., Owens-Illinois, with Ando, F., JACA. March 8,
1983. TV picture tube envelopes.
42. Reference 25. p. 8-41.
43. Reference 25. p. 8-42.
44. Reference 25. p. 8-43.
45. Reference 25. pp. 8-43 to 8-44.
46. Reference 25. p. 8-44.
47. Reference 25. p. 8-45.
48. U.S. Department of Commerce. Bureau of Industrial Economics. 1983
U.S. Industrial Outlook. Washington, D.C. p. 2-5.
49. Reference 25. p. 8-3.
50. Telecon. Harris, J. M., U.S. Department of Commerce, with Ando, F.
H., JACA. February 28, 1983. Companies in the flat glass industry.
7-54
-------
51. Reference 25. pp. 8-3 to 8-4.
52. Reference 25. p. 8-5.
53. Reference 48. p. 2-6.
54. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Report: Flat Glass. MQ-32A. 1981. p. 2.
55. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Report: Flat Glass. MQ-32A. 1979. p. 3.
56. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Report: Flat Glass. MQ-32A. 1978. p. 3.
57. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Report: Flat Glass. MQ-32A. 1977. pp. 2, 3.
58. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Report: Flat Glass. MQ-32A. 1974. p. 2.
59. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Report: Flat Glass. MQ-32A. 1973. p. 2.
60. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Report: Flat Glass. MQ-32A. 1971. p. 2.
61. Telecon. Harris, J. M. U.S. Department of Commerce, with Ando, F.
H., JACA. February 22, 1983. Price of flat glass.
62. U.S. Department of Commerce. Bureau of the Census. Current Indus-
trial Reports: Glass Containers. M32G. 1981.
63. Reference 25. p. 8-10.
64. Telecon. Roland, G., Glass Packaging Institute, with Ando, F. H.,
JACA. March 1, 1983. Container Industry.
65. Reference 25. p. 8-17.
66. Reference 48. p. 6-5.
67. Reference 48. p. 6-4.
68. Reference 48. p. 6-6.
69. Telecon. Lofquist, W. S., U.S. Department of Commerce, with Ando, F.,
H., JACA. February 22, 1982. Price of Container Glass.
70. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Glass Containers. M32G. 1981. p. 3.
71. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Glass Containers. M32G. 1980. p. 4.
7-55
-------
72. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Glass Containers. M32G. 1978. p. 4.
73. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Glass Containers. M32G. 1979. p. 4.
74. U.S. Department of Commerce. Bureau of the Census. Current Industrial
Report: Glass Containers. M32G. 1977. p. 4.
75. Telecon. Hanks, R., U.S. Bureau of the Census, with Miller, L. March 1,
1983. Container shipments and quantity data, 1970 to 1975.
76. Telecon. Cherill, J., Corning Glass, with Ando, F. H., JACA. March
16, 1983. Borasilicate light bulb blanks.
77. Telecon. Cherill, J., Corning Glass, with Ando, F. H., JACA. March
16, 1983. Borasilicate tubing.
78. Annual Report for Corning Glass for the Fiscal Year ending December
31, 1981. p. 29.
79. Annual Report for Corning Glass for the Fiscal Year ending December
31, 1980. p. 30.
80. Annual Report for RCA for the Fiscal Year ending December 31, 1981.
p. 38.
81. Annual Report for RCA for the Fiscal Year ending December 31, 1980.
p. 36.
82. Annual Report for Anchor Hocking for the Fiscal Year ending December
31, 1981. p. 25.
83. Annual Report for Anchor Hocking for the Fiscal Year ending December
31, 1980. p. 32.
84. Annual Report for GTE for the Fiscal Year ending December 31, 1981.
p. 25.
85. Annual Report for Owens-Illinois for the Fiscal Year ending December
31, 1981. p. 38.
86. Annual Report for Owens-Illinois for the Fiscal Year ending December
31, 1980. p. 36.
87. Annual Report for General Electric for the Fiscal Year ending December
31, 1981. p. 49.
88. Annual Report for PPG for the Fiscal Year ending December 31, 1981.
pp. 17-18.
7-56
-------
89. Annual Report for PPG for the Fiscal Year ending December 31, 1980.
p. 20.
90. Annual Report for Libbey-Owens-Ford for the fiscal year ending December
31, 1981. p. 21.
91. Reference 74. p. 42.
92. Reference 81. p. 40.
93. Reference 78. p. 27.
94. Reference 79. p. 29.
95. Reference 76. p. 41.
96. Reference 77. p. 39.
97. Reference 80. p. 26.
98. Telecon. Bronstein, H., U.S. Small Business Administration (SBA)
with Ando, F. H., JACA. March 21, 1983. Size standards for SIC
3229.
99. Electronic Yellow Pages - Manufactures Directory. Market Data
Retrieval, Inc., Westport, Connecticut. April 29, 1983.
7-57
-------
APPENDIX A
EVOLUTION OF THE BACKGROUND INFORMATION DOCUMENT
DATE
NATURE OF ACTION
1/12/83
1/26/83
1/28/83
2/7/83
2/7/83
2/7/83
3/2/83
3/31/83
Effective date of court order to develop
inorganic arsenic NESHAP regulations.
Selection of source category.
Begin literature review.
Working group meeting.
Begin telephone contacts with flat glass
industry.
Begin telephone contacts to update
arsenic emissions data from glass
furnaces.
Visits to glass plants with add-on
particulate control devices.
NAPCTAC package mail out to industry
consisting of Preliminary Study of
Sources of Arsenic Emissions.
A-l
-------
APPENDIX B
INDEX TO ENVIRONMENTAL CONSIDERATIONS
This appendix consists of a reference system which is cross-indexed
with the October 21, 1974 Federal Register (39 FR 37419) containing the
Agency guidelines for the preparation of Environmental Impact Statements.
This index can be used to identify sections of the document which contain
data and information germane to any portion of the Federal Register
guidelines.
B-l
-------
APPENDIX B
INDEX TO ENVIRONMENTAL IMPACT CONSIDERATIONS
Agency Guidelines for
Preparing Regulatory
Action Environmental
Impact Statements
(39 FR 37419)
Location Within the Background
Information Document (BID)
1. Background and Description
Summary of the Regulatory
Alternatives
Industry Affected
Sources Affected
Availability of Control
Technology
2. Regulatory Alternatives
Regulatory Alternative 1
No Action (Baseline)
Environmental Impacts
Costs
The regulatory alternatives are
summarized in Chapter 1,
Section 1.2.
A description of the industry to be
affected is given in Chapter 7,
Section 7.1.
Descriptions of the various sources
to be affected are given in
Chapter 2, Section 2.2.
Information on the availability of
control technology is given in
Chapter 3.
Environmental effects of Regulatory
Alternative I are considered in
Chapter 5.
No capital costs are associated with
Regulatory Alternative 1, and thus
a cost analysis is not included.
B-2
-------
INDEX TO ENVIRONMENTAL IMPACT CONSIDERATIONS (Continued)
Agency Guidelines for
Preparing Regulatory
Action Environmental
Impact Statements
(39 FR 37419)
Location Within the Background
Information Document (BID)
Regulatory Alternative II •
Environmental Impacts
Costs
Regulatory Alternative III
Environmental Impacts
Costs
Environmental effects associated
with Regulatory Alternative II
emission control systems are
considered in Chapter 5.
The cost impact of Regulatory
Alternative II emission control
systems is considered in Chapter 6,
The implementation of this
alternative would require a ban on
arsenic emissions from glass
furnaces. Arsenic emissions
would be zero.
This alternative could not be
implemented without closing glass
furnaces which presently use
arsenic. This alternative would
not require capital costs, and
hence costs were not evaluated.
B-3
-------
APPENDIX C
EMISSION SOURCE TEST DATA
The purpose of this appendix is to present arsenic emissions test data
used in the development of this background information document. Arsenic
emission test results were available for two glass furnaces and are
presented in this appendix. Both tests were conducted by the U. S.
Environmental Protection Agency. One test was performed on a furnace at
Corning Glass Works' State College, Pennsylvania plant. This furnace is
equipped with an electrostatic precipitator. The other test was performed
on a furnace at Corning Glass Works' Central Fall, Rhode Island plant. This
furnace is equipped with a fabric filter. The results of the two tests are
described below.
C.I CORNING GLASS WORKS, STATE COLLEGE, PENNSYLVANIA
The Corning Glass Works plant in State College, Pennsylvania, was
tested by the Emission Measurement Branch, Field Testing Section of the
Environmental Protection Agency (EPA). The purpose of testing the Corning
Glass plant was to gather data that could possibly be used to support the
setting of standards of performance for the glass industry. The furnace
tested attains a temperature of approximately 1200°C. The greatest level of
pollution from the plant comes from the melt tank emissions which are
directed through the plant's control equipment (an ESP) and eventually out
the stack.
Corning Glass Works feeds two types of batch material into the melt
tanks. One type is the raw material for glass made up of approximately
50 percent sand and smaller concentrations of arsenic, lead, silicon,
fluorine, aluminum oxide, sodium nitrate, boric acid, and anhydrous borax.
The lesser concentrations vary depending on the batch and the type of glass
being manufactured. The other component of the melt tank feed is previously
rejected glass from the earlier melts. The rejected glass is called
C-l
-------
cullet glass. The raw material and the cullet were mixed in a 1:1
ratio (during the emissions tests) and fed into the melt tank for
approximately 48 hours. The conditioned glass is then removed from the melt
tank and continues on to the molding process.
The gases from this melting process flow from the melt tank through a
spray tower and on to the electrostatic precipitator (ESP). From the ESP
the gases are directed to the outlet stack. The water spray in the spray
tower is used only when an equipment design temperature is exceeded.
The control device was sampled, generally, in accordance with the draft
Method 108. Since Method 108 is designed for use at non-ferrous smelters
where high levels of SO,, are expected, impingers containing solutions of
H202 are utilized. However, at the Corning Glass Works only very small
amounts of S02 were present, therefore, the impinger solutions were
changed to contain only H20 to trap the gaseous arsenic and moisture in
the gas stream. Cleanup of the impinger portion of the train was done using
a solution of NaOH in accordance with Method 108. All other requirements of
sampling using Method 108 were used along with any Federal Register methods
needed to perform Method 108.
During the arsenic emission tests an integrated gas sample was taken at
a single point according to Method 3 and analyzed by Orsat to determine the
stack gas molecular weight and excess air. Also run simultaneously with the
sampling runs for arsenic was Method 9, an opacity determination by a single
observer, started 30 minutes before and ended 30 minutes after the runs.
Grab samples of ESP dust, feed material, slag and product were
collected during each run. These samples were analyzed for arsenic content.
All arsenic analysis was performed by Atomic Absorption Spectrophotometry
(AA). Analysis was performed directly on the liquid samples and on the
solid samples after being digested.
Samples showing high arsenic levels were quantitated using Flame Atomic
Absorption, while the low-level samples were quantitated by Atomic
Absorption using the hydride generation technique.
A total of six arsenic emission tests were conducted at the ESP outlet
stack of the furnace. Table C-l summarizes the arsenic data under
C-2
-------
TABLE C-l. SUMMARY OF UNCONTROLLED ARSENIC EMISSIONS (ESP)
Run Number
Date
Volume of gas sampled - DNCMa
Percent moisture by volume
Average stack temperature - °C
Stack volumetric flow rate - DNCMPM&
Stack volumetric flow rate - ACMPM
Percent isokinetic
Duration of run - minutes
Arsenic loading
Front half - mg
- g/DNCM
- fcg/hr j
Back half - mg
- g/DNCM
- kg/hr
Total - mg
- g/DNCM
- kg/hr
2
9/19/78
1.49
12.13
207
865
1674
97.0
120
21.10
0.0141
0.732
0.000
0.0000
0.000
21.10
0.0141
0.732
4
9/20/78
1.57
11.57
210
885
1711
100.1
120
22.80
0.0145
0.767
0.10
0.0001
0.004
22.90
0.0146
0.771
6
9/21/78
1.54
13.51
209
852
1677
101.7
120
26.20
0.0170
O.R68
0.000
0.0000
0.000
26.20
0.0170
0.868
o
I
CO
Dry normal cubic meters at 20°C, 760 mm Hg
Dry normal cubic meters per minute
-------
uncontrolled conditions, when the electrical current to the ESP was off.
Table C-2 summarizes the arsenic data during controlled conditions, when the
electrical current to the ESP was on. All arsenic data is reported as
elemental arsenic.
C.2 CORNING GLASS WORKS, CENTRAL FALLS, RHODE ISLAND
The corning Glass Works plant in Central Falls, Rhode Island, was
tested by the Emission Measurement Branch, Field Testing Section of the
Environmental Protection Agency (EPA). The purpose of testing the Corning
Glass plant was to gather data that could possibly be used to. support the
setting of standards of performance for the glass industry.
The furnace attains a temperature of approximately 1400°C. The
emissions from the furnace are directed through the central equipment (a
fabric filter) and eventually out the stack. Corning Glass Works mixes feed
batch and feed cullet with hot cullet return in the melt tank. Also mixed
with these raw materials is a solution of arsenic acid. From the melting
tank, product is sent to molding and cutting machines. Exces^ glass from
molding forms and any glass produced during stoppages of the molding
production line is recycled as hot cullet return.
The gas stream leaves the melting tank and enters the regenerators
where slag is removed. Regenerator exhaust enters the baghouse where
particulate is collected. Gas from the baghouse is vented to the
atmosphere.
The baghouse was sampled generally in accordance with the draft
Method 108. Method 108 is designed for use at nonferrous smelters where
high levels of S02 are expected. At this furnace, only very small amount
of S02 were present, therefore, impinger solutions were changed from
H?0? to H^O to trap the gaseous arsenic and moisture in the gas
stream. Cleanup of the impinger portion of the train was done using a
solution of NaOH in accordance with Method 108. All other requirements of
sampling using Method 108 were used along with any Federal Register methods
needed to perform Method 108.
C-4
-------
TABLE C-2. SUMMARY OF CONTROLLED ARSENIC EMISSIONS (ESP)
Run Number
Date
Volume of gas sampled - DNCMa
Percent moisture by volume
Average stack temperature - *C
Stack volumetric flow rate - DNCMPM&
Stack volumetric flow rate - ACMPM
Percent isokinetic
Duration of run - minutes
Arsenic loading
cFront half - mg
- g/DNCMc
- kg/hr
Back half - mg
- g/DNCNF
- kg/hr
Total - mg
- g/DNCMC
- kg/hr
1
9/19/7R
1.62
13.96
210
840
1669
108.4
120
0.049
0.0000
0.002
0.003
0.0000
0.0000
0.05
0.0000
0.002
3
9/20/78
1.60
12.22
210
891
1736
101.3
120
0.042
0.0000
0.001
0.001
0.0000
0.000
0.04
0.0000
0.001
5
9/21/78
1.59
12.41
208
877
1701
102.2
120
0.15
0.0001
0.005
0.000
0.0000
0.0000
0.15
0.0001
0.005
Dry normal cubic meters at 20°C, 760 mm Hg
?Dry normal cubic meters per minute
•»
T.rnms per dry normal cubic meter
-------
During the arsenic emission tests, an integrated gas sample was taken
at a single point according to Method 3 and analyzed by Orsat to determine
the stack gas molecular weight and excess air. Also run simultaneously with
the sampling runs for arsenic was Method 9, an opacity determination by a
single observer, started 30 minutes before and ended 30 minutes after the
runs.
Grab samples of baghouse dust, feed material, slag and product were
collected during each run. All arsenic analysis was performed by Atomic
Absorption Spectrophotometry (AA). Analysis was performed directly on the
liquid samples and on the solid samples after being digested. Samples
showing high arsenic levels were quantitated using Flame Atomic Absorption,
while the low-level samples were quantitated by Atomic Absorption using the
hydride generation technique.
A total of eight arsenic emission tests were performed at the fabric
filter inlet and outlet stack. Table C-3 summarizes the arsenic data at the
inlet. Table C-4 summarizes the arsenic data at the outlet. All arsenic
data are reported as elemental arsenic.
C-6
-------
TABLE C-3. SUMMARY OF INLET ARSENIC EMISSIONS (FABRIC FILTER)
Run Number
Date
Volume of gas sampled - DNCMa
Percent moisture by volume
Average stack temperature - °C
Stack volumetric flow rate - DNCMPM&
Stack volumetric flow rate - ACMPM
Percent isokinetic
Duration of run - minutes
Arsenic loading
Front half - mg
- g/DNCM
- kg/hr
Back half - mg
- g/DNCM
- kg/hr
Total - mg
- g/DNCM
- kg/hr
1-1
10/10/78
2.55
3.24
238
189
346
97.5
128
68.30
0.0267
0.303
4.70
0.0019
0.021
73.00
0.0286
0.324
1-2
10/11/78
2.58
3.50
235
180
327
103.9
128
59.80
0.0231
0.250
4.70
0.0018
0.019
64.50
0.0249
0.269
T-3
10/11/78
2.68
3.71
240
189
150
102.5
128
62.50
0.0233
0.264
5.90
0.0022
0.025
fifi.40
0.0255
0.289
1-4
10/1^/78
2 47
1.51
235
173
124
103.0
128
62.40
0.0253
0.263
7. fin
n.nn-in
n m?
7n. nn
n.n2m
0.295
o
I
Dry normal cubic meters at 20°C, 760 mm Hg
Dry normal cubic meters per minute
-------
TABLE C-4. SUMMARY OF OUTLET ARSENIC EMISSIONS (FABRIC FILTER)
o
I
CD
Run Number
Date
Volume of gas sampled - DNCM*
Percent moisture by volume
Average stack temperature - °C
Stack volumetric flow rate - DNCMPM°
Stack volumetric flow rate - ACMPM
Percent isokinetic
Duration of run — minutes
Arsenic loading
Front half - mg
- q/DNCM
- kq/hr
Back half - mg
- q/DNCM
- kq/hr
Total - mg
- g/DNCM
- kg/hr
0-1
10/10/78
3.63
145
222
326
97.1
120
0.26
0.0001
0.002
2.41
0.0013
0.017
2.67
Ono i A
. UU J. 1
0.019
n-2
10/11/78
1 .Qfl
3.44
138
232
333
95.8
120
0.09
0.0000
0.001
2.51
0.0013
0.017
2.60
n nm ^
1 U « uU J. J
0.018
n-i
10/11/78
1 ,01
3.70
135
222
137
L 9§.6
120
0.05
0.0000
0.0000
3.13
0.0017
0.022
3-18
0 0017
0.022
4.43
1 17
215
11 0
101.0
120
0-07
o.onno
o.oono
4-15
0.0022^
4.72
n.nn52
0.028
*Dry normal cubic meters at 20°C, 760 mm Hg
JDry normal cubic meters per minute
-------
APPENDIX D
D-l EMISSION MEASUREMENT METHODS
At the beginning of the testing program, a literature search was
conducted to identify available sampling and analytical techniques for
determining arsenic emissions. The search revealed that most arsenic
emissions are in the form of arsenic trioxide and arsenic pentoxide.
According to the literature, the most commonly used arsenic sampling method
has been filtration; however, a number of reports have indicated that
filtration alone is not adequate, even at ambient temperatures, because
arsenic trioxide is a potentially volatile material. Since it was decided
to determine the amount of arsenic collected as a particulate, the Method 5
train, with back-up impinger collectors, was chosen as the starting point
for the arsenic sampling system. Based on the available information, a
dilute sodium hydroxide solution was chosen as a collecting solution for the
impingers. This, however, presented a problem since many of the gas streams
to be sampled had very high concentrations of sulfur dioxide (S02), some
as high as 3.5 percent. Therefore, a series of impingers containing
hydrogen peroxide was placed between the filter and the first impinger
containing sodium hydroxide to remove the S02. This was the configuration
for the "working train" used during the first four field tests.
Analytical methods for arsenic were better defined in the literature.
The most commonly-used procedure is a wet chemical method based on arsine
generation, but certain metals including copper are interfering agents with
this method. Instrumental techniques include atomic absorption, neutron
activation, and x-ray fluorescence. Atomic absorption spectrophotometry
(AAS) was chosen as the most promising technique because of its ready
availability, familiarity, and low cost; however, arsenic absorbs weakly and
only in the extreme ultra-violet area of the spectrum (193.7 nm). At that
D-l
-------
wavelength, molecular absorption by flame gases and solution species can
interfere with arsenic detection. Despite this, conventional AAS can still
be used, provided that: (1) the fuel and combustion gas are carefully
chosen and nonatomic background correction is used; and (2) arsenic
concentrations are relatively high. However, for lower arsenic
concentrations, the interference effects necessitate the use of special,
more sensitive technique, such as the hydride generator or the carbon rod
(flameless) system. Before testing began, both conventional and special AAS
methods were compared and evaluated, in terms of their accuracy, precision,
and sensitivity.
During the first two field tests, samples were collected with the
working train and analyzed either by conventional or carbon rod AAS
depending on the arsenic concentration. The analytical results showed that
95 to 100 percent of the arsenic was collected ahead of the NaOH impingers.
In the course of analyzing these samples, the following detailed sample
preparation procedure was developed. Solid samples were digested with 0.1 N
sodium hydroxide, extracted with concentrated nitric acid, evaporated to
dryness, and then redissolved in dilute nitric acid. Liquid samples were
treated similarly except that there was no need for the sodium hydroxide
digestion step. Advantages of the sample preparation procedure include:
(1) reduction of the level of the collected sulfuric acid in the liquid
sample fraction; (2) dissolution of the arsenic in the solid samples; and
(3) production of a similar solution matrix for all the different sample
fractions.
After the second test, questions were raised about the sampling and
analytical procedures. First of all, laboratory studies of vaporized
arsenic trioxide showed no difference in the arsenic collection efficiency
of 0.1 N sodium hydroxide and pure water. These results indicated that the
arsenic collection mechanism is condensation and that any condenser would be
an effective collector. Consequently, the conventional Method 5 train (with
H?0 impingers) was suggested as an alternate to the working train and
simultaneous testing of the two trains was planned for the next facility.
D-2
-------
Second, an evaluation of the different AAS techniques for
low-concentration uncovered some precision and accuracy problems with the
carbon rod method when large quantities of dissolved solids (particularly
sulfates) are present. The hydride generator technique, it was found, gives
much more precise and accurate results in the presence of dissolved solids.
In view of this, it was decided that all future low-concentration arsenic
samples (i.e., too low for conventional AA analysis) would be analyzed by
the hydride generator method.
Third, concern was expressed that arsenic was being lost in the
evaporation step of sample preparation. To investigate this, recovery
studies were performed on standard samples. These studies showed that there
is no significant loss of arsenic during the evaporation step.
Fourth, additional studies showed that while arsenic trioxide is
soluble in alkaline, acid, and neutral solutions, its rate of dissolution is
slow except in alkaline solutions. Therefore, the clean-up procedure for
future test was modified, to require that the train be rinsed with 0.1 N
sodium hydroxide to insure removal of condensed arsenic.
Fifth, a comparison of arsenic extraction techniques indicated that
higher arsenic yields (by up to 200 percent) can be obtained from smelter
particulate when a method capable of dissolving the entire sample is used
instead of the less rigorous acid extraction procedure. As a result, it was
decided that in future tests, filters would be analyzed by both methods,
until more conclusive filter extraction data were obtained.
During the third and fourth field tests, the working train was used for
sampling, but additional runs were taken during the fifth test using paired
trains of the working and alternate procedures. Analysis of the samples
from the paired tests showed no significant difference in collection
efficiency. Therefore, the final recommendation was to use the alternate
train, since it is easier to operate and analyze. During the fifth and
final field test, the alternate train was used.
Filters from the third, fourth, and fifth field tests were extracted,
using both the total dissolution and acid extraction procedure. The results
showed that filters extracted by the less rigorous method could in some
D-3
-------
cases yield 25 percent less arsenic than if totally dissolved. Based upon
these results, the final recommendation was to extract the filters first by
the simple acid extraction; the, if any undissolved sample remained, to
extract the undissolved solids by the total dissolution method.
D-2 CONTINUOUS MONITORING
There is currently no available method for continuously monitoring
arsenic emissions. For purposes of demonstrating proper operation and
maintenance of control devices, continuous monitors are available for
measuring opacity from baghouses or electrostatic precipitators, and
measuring pressure drop across scrubbers. However, these measurements are
not necessarily indicators of the magnitude of arsenic emissions and should
not be used for compliance determinations. In addition, opacity may not be
applicable as an indicator of proper operation and maintenance where
baghouses and precipitators are used to control captured fugitive emissions
because of the uncontrolled particulate is very low in concentration.
The recommended monitoring program for continually assessing arsenic
emissions is a periodic application of the performance test Method 108 as
recommended in Part D-3 below. This is the only method evaluated at this
time for demonstration of compliance with arsenic emissions.
D-3 PERFORMANCE TEST METHODS
The recommended performance test method for arsenic is Method 108.
Based on the development work already discussed, the method uses the
Method 5 train for sampling, 0.1 N sodium hydroxide for cleanup, and either
conventional or hydride generator AAS for sample analysis. In order to
perform Method 108, Methods 1 through 4 must also be used. Subpart A or
40 CFR 60 requires that facilities subject to standards of performance for
new stationary sources be constructed so as to provide sampling ports
adequate for the applicable test methods, and platforms, access, and
utilities necessary to perform testing at those ports.
D-4
-------
Sampling costs for performing a test consisting of three Method 108
runs is estimated to range from $10,000 to $14,000. If in-plant personnel
are used to conduct tests, the costs will be somewhat less.
D-4 BIBLIOGRAPHY
1. Hefflefinger, R. E. and D. L. Chase (Battelle). Analysis of Copper
Smelter Samples for Arsenic Content. Prepared for U. S. Environmental
Protection Agency. Research Triangle Park, N. C. April 1977. 14 p.
2. Haile, D. M. (Monsanto Research Corporation). Final Report on the
Development of Analytical Procedures for the Determination of Arsenic from
Primary Copper Smelters. Prepared for U. S. Environmental Protection
Agency. Research Triangle Park, N. C. February 1978. 27 p.
3. Harris, D. L. (Monsanto Research Corporation). Particulate and
Arsenic Emission Measurements from a Copper Smelter. Prepared for the
U. S. Environmental Protection Agency. Research Triangle Park, N. C.
77-CUS-5. April 1977. 48 p.
4. Harris, D. L. (Monsanto Research Corporation). Particulate and
Arsenic Emission Measurements from a Copper Smelter. Prepared for the
U. S. Environmental Protection Agency. Research Triangle Park, N. C.
77-CUS-6. June 1977. 276 p.
5. TRW, Inc. Emission Testing of Asarco Copper Smelter. Prepared for
the U. S. Environmental Protection Agency. Research Triangle Park, N. C.
77-CUS-7. April 1978. 150 p.
D-5
-------
APPENDIX E
QUANTITATIVE EXPRESSIONS OF PUBLIC CANCER RISKS FROM EMISSIONS OF
INORGANIC ARSENIC FROM GLASS MANUFACTURING PLANTS
-------
QUANTITATIVE EXPRESSIONS OF PUBLIC CANCER RISKS FROM THE EMISSIONS OF
INORGANIC ARSENIC FROM GLASS MANUFACTURING PLANTS
E.I INTRODUCTION
E .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.
E.I.2 The Relationship of Exposure to Cancer Risk
The relationship of exposure to the risk of getting lung cancer is
derived from epidemiological studies in occupational settings rather than
from studies of excess cancer incidence among the public. The epidemiological
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 epidemiological
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
E-2
-------
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 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
E-3
-------
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 of public exposure is the same as that between cancer
risks and levels of occupational exposure. EPA believes that this assumption
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 model 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. 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 continuously from birth throughout their lifetimes (about 70
years) to a concentration of one yg/m3 of the agent in the air which they
E-4
-------
breathe. Unit risk estimates are used for two purposes: (1) to 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 comparative potency of different
agents is more reliable when the comparison is based on studies of like
populations and on the same route of exposure, preferably inhalation.
The unit risk estimate for inorganic arsenic that is used in this
appendix was prepared by combining the three different exposure-risk
numerical constants developed from three occupational studies.2 The unit risk
estimate is expressed as a range that reflects the statistical uncertainty
associated with combining the three exposure-risk relationships. The
methodology used to develop the unit risk estimate is described in E.2
below. EPA is updating its health effects assessment document for inorganic
arsenic. A preliminary estimate by EPA's health scientists is that the
-*
unit risk estimate may change.
E.I.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., of 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 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
E-5
-------
necessity and difficulty of dealing with long-term exposures because of
"...the long latent period required for the development and expression
of neoplasia [cancer]..."
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 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 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 E.3
below.
E.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 dispersion 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 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 20 kilometers) of a source and is customarily summed
for all the sources in a particular category. The aggregate risk is expressed
as incidences of cancer among all of the exposed population after 70 years
E-6
-------
of exposure; for statistical convenience, it is often divided by 70 and
expressed as cancer incidences per year. These calculations are described
in more detail in E.4 below.
There are also risks of nonfatal cancer and of serious genetic effects,
depending on which organs receive the exposure. No 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.
»
E.2 THE UNIT RISK ESTIMATE FOR INORGANIC ARSENIC?
E'2J The Li"ear No-Threshold Model for Estimation of Unit Risk Based on
Human Data (General)4
Very little information exists that can be utilized to extrapolate
from high exposure occupational studies to low environmental levels.
However, if a number of simplifying assumptions are made, it is possible
to construct a crude dose-response model whose parameters can be estimated
using vital statistics, epidemiologic studies, and estimates of worker
exposures. In human studies, the response is measured in terms of the
relative risk of the exposed cohort of individuals compared to the control
group. The mathematical model employed assumes that for low exposures the
lifetime probability of death from lung cancer (or any cancer), P, may be
represented by the linear equation
P = A + BHx (!)
where A is the lifetime probability of cancer in the absence of the agent, x
is the average lifetime exposure to environmental levels in micrograms per
cubic meter (,g/m3) and BH is the increased probability of cancer associated
with each Mg/m3 increase of the agent in air.
If we make the assumption that R, the relative risk of lung cancer for
exposed workers, compared to the general population, is independent of the length
E-7
-------
or age of exposure but depends only upon the average lifetime exposure, it
follows that
P A + BH (x0 + X!)
R= = (2)
PO A + BH (x0)
or
RP0 = A + BH (XQ + X]) (3)
where XQ = lifetime average exposure to the agent for the general popu-
lation, xi = lifetime average exposure to the agent in the occupational
setting, and PQ = lifetime probability of respiratory cancer applicable with
no or negligible arsenic exposure. Substituting PQ = A + BH XQ and rearranging
gives
BH = P0 (R - 1)/X! (4)
To use this model, estimates of R and X] must be obtained from the epidemio-
logic studies. The value PQ is derived from the age-cause-specific death
rates for combined males found in 1976 U.S. Vital Statistics tables using
the life table methodology. For lung cancer the estimate of PQ is 0.036.
E.2.2 The Unit Risk Estimate for Inorganic Arsenic^
As noted in the health effects assessment document*3 for inorganic
arsenic, there are numerous occupational studies which relate increased
cancer rates to arsenic exposure. Based on these studies, it is concluded
in the health assessment document that there is substantial evidence that
inorganic arsenic is a human carcinogen. However, many of these studies
are inappropriate for use in developing a unit risk estimate for inorganic
arsenic because the route of exposure was not by inhalation or because it
was impossible to make a reasonable estimate of the population's lifetime
average exposure.
E-8
-------
Three studies, Lee and Fraumeni (1969), Ott et al. (1974), and Pinto
et al. (1977), contained enough pertinent information to make independent
quantitative estimates of human cancer risks due to human exposures to
atmospheric arsenic. The crudeness of the exposure estimates in those
studies is due to such factors as high variability in the chemical measurement
of arsenic, a scarcity of monitoring data, and the necessity of working
from summarized data tables presented in the literature rather than complete
data on all individuals.' However, by accepting the data in spite of its
recognized limitations, and making a number of simplifying assumptions
concerning dose-response relationships and exposure patterns, it was possible
to estimate the carcinogenic potency of arsenic. Using a linear model, it
was estimated that the increase in the lung cancer rate per increase of 1
ug/m3 of atmospheric arsenic was 9.4% (Pinto et al.), 17.0% (Ott et al.),
and 3.3% (Lee and Fraumeni). The consistency of these estimates is very
good considering the relative crudeness of the data upon which they are
based. The geometric mean of the rate estimates from the three studies was
calculated to be 8.1%. Using this value as a best estimate and applying
equation 4, one calculates the unit risk estimate of 2.95 x 10~3 per
ug/m3.
If we assume that the linear model and exposure estimates are correct,
so that the only source of uncertainty is from combining results from the
three different studies, a 95% confidence interval for the above unit risk
estimate may be obtained. Upper and lower 95% confidence limits can be
obtained by multiplying the unit risk estimate by about 4 and 0.25 respect-
ively. Thus, the 95% statistical confidence limits for the unit risk estimate
range from 7.5 x lO'4 to 1.2 x 10'2.
E-9
-------
E.3 QUANTITATIVE EXPRESSIONS OF PUBLIC EXPOSURE TO INORGANIC ARSENIC
EMITTED FROM LOW-ARSENIC PRIMARY COPPER SMELTERS
E.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 atmospheric
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 20 kilometers from the source. If other radial distances are preferred,
an over-ride feature allows the user to select the distance desired. The
selection of 20 kilometers as the programmed distance is based on modeling
considerations, not on health effects criteria or EPA policy. The dispersion
model contained in HEM is felt to be reasonably accurate within 20 kilometers.
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 concentrations if they are put into an appropriate
format.
Based on the radial distance specified, HEM combines numerically the
distributions of pollutant concentrations and people to produce quantitative
expressions of public exposure to the pollutant.
E-10
-------
E.3.1.1 Pollutant Concentrations Near a Source
The dispersion model within the HEM is a gaussian diffusion model that
uses the same basic dispersion algorithm as EPA's Climatological Dispersion
Model.6 The algorithm has been simplified to improve computational efficiency.7
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 meteorological input to the HEM dispersion model. STAR data are
standard Climatological frequency-of-occurence summaries formulated for use
in EPA models and available for major '.!.S. meteorological monitoring sites
from the National Climatic Center, Asheville, N.C. A STAR summary is a
joint frequency-of-occurence of wind speed, atmospheric stability, and wind
direction, classified according to Pasquill's categories. The STAR summaries
in HEM usually reflect five years of meteorological data for each of 309
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, 0.3, 0.5, 0.7, 1.0, 2.0, 5.0, 10.0, 15.0, and 20.0 kilometers. The
center of the receptor grid for each plant is assumed to be the plant
center.
E.3.1.2 The People Living Near A Source
To estimate the number and distribution of people residing within 20
kilometers of each plant, the model contains a slightly modified version of
the "Master Enumeration District List—Extended" (MED-X) data base.
E-ll
-------
The data base is broken down into enumeration district/block group (ED/8G)
values. MED-X contains the population centroid coordinates (latitude and
longitude) and the 1970 population of each ED/BG in the United States (50
States plus the District of Columbia). For human exposure estimates, MED-X
has been reduced from its complete form (including descriptive and summary
data) to produce a computer file of the data necessary for the estimation.
A separate file of county-level growth factors, based on 1978 estimates of
the 1970 to 1980 growth factor at the county level, has been used to estimate
the 1980 population for each ED/BG. HEM identifies the population around
each plant by using the geographical coordinates of the plant. The HEM
identifies, selects, and stores for later use those ED/BGs with coordinates
falling within 20 kilometers of plant center.
E.3.1.3 Exposure?
The Human Exposure Model (HEM) uses the estimated ground level concen-
trations of a pollutant together with population data to calculate public
exposure. For each of 160 receptors located around a plant, the concentra-
tion 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 is classified by the region in which its centroid lies.
E-12
-------
Population exposure is calculated differently for the ED/BG's located
within each region. For ED/BG centroids located between 0.1 km and 2.8 km
from the emission source, populations are divided between neighboring
concentration grid points. There are 96 (6 x 16) polar grid points within
this range. Each grid point has a polar sector defined by two concentric
arcs and two wind direction radials. Each of these grid points and respec-
tive concentrations are assigned to the nearest ED/BG centroid identified
from MED-X. 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 2.8 km and 20 km from the source, a
concentration grid cell, the area approximating a rectangular shape bounded
by four receptors, is much larger than the area of a typical ED/BG. 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.
Concentration estimates for 80 (5 x 16) grid cell receptors at 2.0, 5.0,
10.0, 15.0, and 20.0 km from the source along each of 16 wind directions
are used as reference points for this interpolation.
In summary, two approaches are used to arrive at coincident
concentration/population data points. For the 96 concentration points
within 2.8 km of the source, the pairing occurs at the polar grid points
using an apportionment of ED/BG population by land area. For the remaining
portions of the grid, pairing occurs at the ED/BG centroids themselves
E-13
-------
through the use of log-log and linear interpolation. (For a more detailed
discussion of the model used to estimate exposure, see Reference 7.)
E-3»2 Public Exposure to Inorganic Arsenic Emissions from Glass Manufacturing
Plants *
E.3.2.1 Source Data
Fifteen glass manufacturing plants are included in the analysis.
Table E.I lists the names and addresses of the plants considered, and Table
E.2 lists the plant data used as input to the Human Exposure Model (HEM).
E.3.2.2 Exposure Data
Table E.3 lists, on a plant-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 concentration
to which they are exposed, as calculated by HEM. Table E.4 sums, for the
entire source category (15 plants), the numbers of people exposed to various
ambient concentrations, as calculated by HEM.
E-14
-------
TABLE E-l
IDENTIFICATION OF GLASS MANUFACTURING PLANTS
Plant Number Code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Plant Name and Address
RCA, Circleville, OH
GTE, Versailles, KY
Owens-Illinois, Columbus, OH
a
a
a
a
a
a
a
a
a
a
a
a
a Companies requested confidential treatment of plant identity
associated with input data used in developing risk estimates,
E-15
-------
Table E-2 Input Data to Exposure Model Glass Manufacturing
Plant Latitude
(Degrees
Minutes
(Furnace) Seconds)
1 1 393358
2
2 380246
3 395555
4 1 394232
2
3
4
5
b 1 391554
2
6 1 391718
2
3
7 41 5346
8 374518
9 392613
0 363500
. 1 405006
1 2 1 40081 7
2
3
4
5
i3 420835
14 1 420842
2
Longtitude
(Degrees
Minutes
Seconds)
0825718
0844504
0825823
. 0823441
0763748
0802102
0712320
0844948
0775910
0792400
0774728
0795346
0770242
0770322
Emission
Rate
(Kg/yr)
30.5
6.9
68.6
95.3
3048.
3538.
725.
544.
544.
1270.
907.
454.
1814.
3084.
194.3
49.5
15240.7
38.1
22.9
952.5
952.5
1714.6
571.5
76.2
228.6
76.2
38.1
Emission
Point
Elevation
(Meters)
22.3
9.75
15.5
21.3
44.2
36.0
38.1
19.2
21.9
39.6
29.0
15.0
15.24
36.6
18.3
41.1
41.1
15.2
45.7
9.1
7.6
15.2
9.1
30.5
25.9
27.9
27.9
Emission
Point
Diameter
(Meters)
1.52
0.95
0.70
1.58
1.78
1.4
0.99
0.80
1.83
1.07
1.52
2.4
1.37
1.52
0.91
1.8
2.4
4.80
1.52
0.76
0.76
0.82
0.76
1.2
1.2
1.1
1.2
Emission*
Point
Cross
Sectional
Area (m2)
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
750
Emission
Point Gas
Exit
Velocity
m/sec
14.02
9.32
29.3
12.7
6.0
10.0
9.8
18.0
4.9
8.56
3.0
0.30
3.05
6.15
7.6
3.0
3.4
1.5
15.2
13.1
18.3
29.0
18.6
17.4
22.9
14.0
15.2
Emission
Point Gas
Temp.
(°K)
505
436
494
450
783
725
700
636
669
655
610
310
644
700
410
440
666
311
477
672
755
644
544
461
477
755
477
Emission
Point ,
Type
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Vent
Stack
Stack
Stack
Stack
Stack
Vent
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
420838
0770235
114.3
29.0
0.45
750
12.2
644
Stack
'Assumed 750 m2 for all glass manufacturing plants, this is the vertical cross sectional area of the emission point
to the mean wind direction for purpose of calculating downwash.
E-16
-------
TABLE E-3 TOTAL EXPOSURE AND NUMBER OF PEOPLE EXPOSED
(GLASS MANUFACTURING PLANTS)*
Total Total
Number of Exposure
Plant People Exposed (People - yg/m3)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
41,000
99,000
866,000
79,000
1,530,000
82,000
708,000
48,000
60,000
82,000
88,000
204,000
96,000
85,000
97,000
2
4
72
850
2300
1200
190
3
1200
23
1
1100
13
55
20
* A 20-kilometer radius was used for the analysis of glass manufacturing
plants.
E-17
-------
TABLE E-4
Public Exposure for Glass Manufacturing Plants
as Produced the Human Exposure Model
Concentration Population Exposure
Level ( ug/m^) Exposed (Persons-
(Persons)*
0.863 15 14
0.5 62 44
0.25 212 93
0.1 1755 314
0.05 13517 1010
0.025 , 49475 2210
0.01 140160 3620
0.005 271060 4540
0.0025 502235 5340
0.001 1106248 6290
0.0005 1719070 6730
0.00025 2212152 6910
0.0001 2657137 6980
0.00005 3331006 7030
0.000025 4048413 7050
0.00001 4144677 7050
0.0000005 4166149 7050
0.00000025 4166150 7050
0.000000147 4166152 7050
*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.
E-18
-------
E.4 QUANTITATIVE EXPRESSIONS OF PUBLIC CANCER RISKS FROM INORGANIC ARSENIC
EMITTED FROM GLASS MANUFACTURING PLANTS
E.4.1 Methodology (General)
E.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
estimated maximum ambient air concentration for 70 years.
E.4.1.2 ThejCaJculation 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 (pici)
E-19
-------
where
I = summation over all grid points where exposure is calculated
P.J = 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 20 kilometers of each source.
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 yg/m3, and the assumed
exposures are:
ambient air number of people exposed
concentrations to given concentration
2 yg/m3
1 Ug/m3
0.5 yg/m3 10°>000
The probability of getting cancer if continuously exposed to the assumed
concentrations for 70 years is given by:
concentration unit- jri.sk. probability of cancer
2 yg/m3 x 3 x IQ-3-1 = 6 x 10'3
1 ug/m3 x 3 x 10'3 " = 3 * ™-3
0.5 yg/m3 x 3 x 10"3 " = 1.5 x 10'3
E-20
-------
The 70 year cancer incidence among the people exposed to these concentrations
is given by:
cancer incidences
probability of cancer number of people at after 70 years
at each exposure level each exposure level of exposu re
6
30
150
6
3
1.5
x lO-3
x lO-3
x TO'3
x
x
x
1
10
100
,000
,000
,000
TOTAL = 186
The aggregate risk, or total cancer incidence, is 186 and, expressed
as cancer incidence per year, is 186 * 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.
E.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 yg/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 = 1 pg/m^
which people are exposed
E.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 plants, assuming
a baseline level of emissions. A baseline level of emissions means the
E-21
-------
level of emissions after the application of controls either currently in
place or required to be in place to comply with curent state or Federal
regulations but before application of controls that would be required by a
NESHAP.
Table E-5 summarizes the calculated risks. To understand the relevance
of these numbers, one should refer to the analytical uncertainties discussed
in section E.5 below.
E-22
-------
Plant
TABLE E-b MAXIMUM LIFETIME RISK AND CANCER
INCIDENCE FOR GLASS MANUFACTURING PLANTS
(Assuming Baseline Controls)
Maximum Lifetime Risk
Cancer Incidences Per Year
Cancer Incidence
(one case 1n [x] years)
1 3.7 x 10-7 . 5.9 x 10-6 i.6 x 10-5 . 2.6 x 10"4
2 1.3 x 10-6 _ 2.] x io-5 4i2 x 10'5 - 6.7 x 10'4
3 5,0 x 10"7 •• 8.0 x 10-6 7.5 x 10"4 - 1.2 x 10'2
4 ' 6.2 x 10~5 - 1,0 x 10-3 8.8 x 10"3 - 1.4 x 10"1
5 7A x 10-5 _ io2 x ID"3 2.4 x 10"2 - 3.9 x 11 x 10-3
14 3.9 x 10-6 . 6.2 x 10"5 5.7 x 10"4 - 9.1 x 10'3
15 4.2 x 10-6 „ 6.7 x 10"5 2.1 x 10'4 - 3 3 x 10'3
TOTALS FOR THIS SOURCE CATGEGORY
1 In 60,000 yrs. - 1 In 4,000 yrs.
1 1n 20,000 yrs. - 1 1n 1000 yrs.
1 1n 1000 yrs. - 1 1n 100 yrs.
1 In 100 yrs. - 1 1n 7 yrs.
1 1n 40 yrs. - 1 1n 3 yrs.
1 1n 80 yrs. - 1 1n 5 yrs.
1 1n 500 yrs. - 1 1n 30 yrs.
1 in 30,000 yrs. - 1 in 2000 yrs.
1 in 80 yrs. - 1 in 5 yrs.
1 in 4,000 yrs. - 1 in 300 yrs.
1 in 100,000 yrs. - 1 in 9000 yrs.
1 in 90 yrs. - 1 in 6 yrs.
1 in 8000 yrs. - 1 in 500 yrs.
1 In 2000 yrs. - 1 1n 100 yrs.
1 in 5000 yrs. - 1 in 300 yrs.
Number Total Number
of of People Exposed
Plants (within 20 km)
15
Highest Individual Risk
Cancer Incidences
per year
one case in [x] years
4J66.000
6.4 x 10-4 - 1.0 x 10-2
(For Plant 6)
0.07 - 1.2
1 in 10 yrs. - 1 in 1 yr.
NOTE: The ranges in these quantifications of public health impacts reflect the uncertainty of combining the three
different dose-response relationships relevant to the three occupational studies which EPA used as the basis
for the development of unit risk estimate.2
E-23
-------
E.5 ANALYTICAL UNCERTAINTIES APPLICABLE TO THE CALCULATIONS OF PUBLIC
HEALTH RISKS CONTAINED IN THIS APPENDIX
E.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.
E.5.2 Public Exposure
E.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 rather than a realistic premise.
E-24
-------
It is not known whether this results in an over-estimation or an under-
estimation of public exposure.
E.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.
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, they are "located" in the Bureau of
Census data for 1970 by population centroids of census districts. Further,
the locations of these centroids has not been changed to reflect the 1980
census. The effect is that the actual locations of residences with respect
to the estimated ambient air concentrations is not known and that the relative
locations used in the exposure model have changed since the 1970 census.
E-25
-------
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.
E.5.2.3. The Ambient Air Concentrations
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.
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. With few exceptions, the arsenic emission rates are based on
assumptions rather than on emission tests. See the BID for details on each
source.
E-26
-------
United States
Environmental Protection
Agency
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park NC 27711
Official Business
Penalty for Private Use
$300
Publication No EPA-450/3-83-011a
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
If your address is incorrect, please change on the above label,
tear off. and return to the above address
If you do not desire to continue receiving this technical report
series, CHECK HERE D , tear off label, and return it to the
above address
------- |