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

-------