Economic Impact Analysis of Final Iron and
Steel Foundries NESHAP
Final Report

-------
                                                          EPA-452/R-03-012
                                                                August 2003
Economic Impact Analysis of Final Iron and Steel Foundries NESHAP
                              By:

                       RTI International*
              Health, Social, and Economics Research
           Research Triangle Park, North Carolina 27711
                          Prepared for:

              U.S. Environmental Protection Agency
            Office of Air Quality Planning and Standards
       Innovative Strategies and Economics Group, C339-01
                Research Triangle Park, NC 27711
                             *RTI International is a trade name of Research Triangle Institute.

-------
This report has been reviewed by the Emission Standards Division of the Office of Air
Quality Planning and Standards of the United States Environmental Protection Agency
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 (MD-35), U.S. Environmental Protection Agency,
Research Triangle Park, NC 27711, or from the National Technical Information Services
5285 Port Royal Road, Springfield, VA 22161.

-------
                                     CONTENTS

Section                                                                          Page

    1      Introduction  	1-1

          1.1    Agency Requirements for an EIA  	1-1

          1.2    Overview of Coke, Iron and Steel, and Foundry Industries	1-2

          1.3    Summary of EIA Results	1-2

          1.4    Organization of this Report	1-4

    2      Industry Profile	2-1

          2.1    Overview of Production Process	2-1
                 2.1.1   Mold and Core Making 	2-3
                        2.1.1.1  Sand Casting Process 	2-3
                        2.1.1.2  Permanent Mold Casting Process	2-4
                        2.1.1.3  Investment Casting Process  	2-5
                 2.1.2   Metal Melting	2-5
                        2.1.2.1  Cupola Furnace	2-6
                        2.1.2.2  Electric Arc Furnace (EAF)  	2-6
                        2.1.2.3  Induction Furnace	2-8
                 2.1.3   Pouring and Shakeout 	2-9
                 2.1.4   Finishing and Cleaning 	2-10
                 2.1.5   Residuals and By-products	2-11
                        2.1.5.1  Particulate Matter	2-12
                        2.1.5.2  Slag	2-13
                 2.1.6   Production Costs  	2-13
                 2.1.7   Metal Types	2-14

          2.2    Industry Organization	2-17
                 2.2.1   Manufacturing Plants	2-17
                 2.2.2   Companies	2-18
                 2.2.3   Industry Trends 	2-23
                 2.2.4   Markets  	2-23
                                         111

-------
       2.3    Historical Industry Data	2-25
             2.3.1   Domestic Production	2-25
             2.3.2   Foreign Trade and Apparent Consumption	2-25
             2.3.3   Market Prices	2-25

       2.4    Market Shipments and Future Projections	2-26

3      Engineering Cost Analysis	3-1

       3.1    Overview of Emissions from Iron Foundries	3-1

       3.2    General Approach for Estimating Compliance Costs  	3-3

       3.3    Cupola Melting Furnace Control Systems  	3-3
             3.3.1   Baghouse Control Costs for Cupola Melting Furnaces	3-4
             3.3.2   Venturi Scrubber Control Costs for Cupola Melting
                    Furnaces	3-6
             3.3.3   Net Metal HAP Control Cost for Cupola Melting Furnaces ... 3-6
             3.3.4   Sample Calculation of Metal HAP Control Cost for Cupola
                    Melting Furnaces	3-7
             3.3.5   Afterburning Control Cost for Cupola Melting Furnaces  	3-7
             3.3.6   Sample Calculation of Organic HAP Control Cost for Cupola
                    Melting Furnaces	3-9

       3.4    Electric Induction, Scrap Preheater, and Pouring Station
             Control Systems  	3-10
             3.4.1   Baghouse Control Costs for EIFs and Scrap Preheaters  	3-10
             3.4.2   Baghouse Control Costs for Pouring Stations	3-13

       3.5    Mold- and Core-Making Control Systems	3-16

       3.6    Monitoring, Reporting, and Recordkeeping  	3-17
             3.6.1   Bag Leak Detection Systems	3-17
             3.6.2   Parameter Monitoring Systems  	3-18
                    3.6.2.1  Parameter Monitoring Systems for Venturi (PM)
                            Wet Scrubbers	3-18
                    3.6.2.2  Parameter Monitoring Systems for Acid/Wet
                            Scrubbing Systems 	3-18
             3.6.3   Foundry Recordkeeping, Reporting, and Compliance Costs ..  3-19
                    3.6.3.1  Performance Tests	3-20
                    3.6.3.2  Scrap Selection and Inspection	3-20
                    3.6.3.3  Start-up, Shutdown, and Malfunction Plan	3-21
                    3.6.3.4  Operating and Maintenance Plan  	3-21
                    3.6.3.5  Miscellaneous Recordkeeping and Reporting Costs .  3-21
                                     IV

-------
          3.7    Total Nationwide Costs	3-22

   4      Economic Impact Analysis  	4-1

          4.1    Market Characterization	4-1
                4.1.1  Regulatory Control Costs	4-2

          4.2    EIA Methodology Summary  	4-4

          4.3    Economic Impact Results	4-5
                4.3.1  Market-Level Impacts  	4-7
                4.3.2  Industry-Level Impacts	4-9
                4.3.3  Social Cost	4-9

   5      Small Business Impacts	5-1

          5.1    Identifying Small Businesses	5-1

          5.2    Screening-Level Analysis  	5-2

          5.3    Assessment	5-2

   References 	R-l

Appendixes

   A     Economic Impact Analysis Methodology 	  A-l

   B      Development of Coke Battery Cost Functions	B-l

   C      Econometric Estimation of the Demand Elasticity for Iron Castings  	C-l

-------
                                 LIST OF FIGURES

Number                                                                        Page

    1-1    Interactions between Source Categories and Markets	1-3

    2-1    Overview of the Foundry Casting Processes	2-2
    2-2    The Cupola Furnace	2-7
    2-3    Indirect and Direct Electric Arc Furnaces	2-8
    2-4    Coreless and Channel Electric Induction Furnaces	2-9
    2-5    Number of U.S. Iron and Steel Foundries by State:  1997	2-18
    2-6    Possible Ownership Configurations for U.S. Iron and Steel Foundries	2-20
    2-7    Distribution of Affected U.S. Companies by Size:  1997 	2-21

    3-1    Control Cost Curves for Cupola Afterburners	3-9
    3-2    Control Cost Curves for EIF/Scrap Preheater Baghouses	3-11
    3-3    Control Cost Curves for Pouring Station Baghouses	3-15

    4-1    Market Linkages Modeled in the Economic Impact Analysis	4-3
    4-2    Distribution of Plant-level Compliance Costs-to-Sales Ratios: 2000	4-4
    4-3    Market Equilibrium without and with Regulation	4-6
                                         VI

-------
                                 LIST OF TABLES

Number                                                                      Page

   2-1    Distribution of Iron and Steel Foundry Furnaces by Type: 1997	2-6
   2-2    Summary of Production Costs by Foundry Type (1996)	2-13
   2-3    Distribution of Material Costs by Foundry Type (1992)	2-15
   2-4    Summary of Labor Statistics for SIC Code 332, Iron and Steel Foundries:
          1980-1997  	2-16
   2-5    Shipments of Iron and Steel Castings by Type (1997)	2-17
   2-6    Iron and Steel Foundry Data by Type of Producer: Iron-Making
          Furnaces (1997)	2-19
   2-7    Iron and Steel Foundry Data by Type of Producer: Steel-Making
          Furnaces (1997)	2-19
   2-8    Distribution of Companies by Number of Foundries:  1997	2-21
   2-9    Summary of Iron and Steel Foundry Operations by Firm Size
          Category:  1997 	2-22
   2-10   U.S. Shipments of Iron and Steel Castings by Market and Captive:
          1997 (103 short tons)  	2-24
   2-11   U.S. Shipments of Iron and Steel Castings: 1987-1997 (103 short tons)  ....2-26
   2-12   U.S. Production, Foreign Trade, and Apparent Consumption of Iron and
          Steel Castings: 1994 (103 short tons)  	2-27
   2-13   Foreign Trade Concentration Ratios for Iron and Steel Castings by
          Type: 1994  	2-27
   2-14   Market Prices for Iron and Steel Castings by Type: 1987-1997	2-28
   2-15   Projected U.S. Shipments of Iron and Steel Castings by Type:  1997, 2000,
          and 2005 (103 short tons)	2-29
   2-16   Iron Castings Shipments by Market:  1987-2005 (103 short tons)	2-30
   2-17   Steel Castings Shipments  by Market: 1987-2005 (103 short tons)	2-32

   3-1    Summary of Control Costs for Baghouses and Wet Scrubbers: 1998$  	3-5
   3-2    Estimating Exhaust Air Flow Rates for Control Costs Estimates	3-5
   3-3    Summary of Control Costs for Acid/Wet Scrubbing Systems: 1998$	3-17
   3-4    Nationwide Cost Estimates for Iron Foundry MACT: 1998$	3-22

   4-1    Baseline Market Data Set: 2000	4-2
   4-2    Supply and Demand Elasticities Used in Analysis 	4-7
                                        vn

-------
4-3    Market-Level Impacts of the Iron and Steel Foundries
       MACT:  2000	4-8
4-4    National-Level Industry Impacts of the Iron and Steel Foundries
       MACT:  2000	4-9
4-5    Distribution of the Social Costs of the Iron and Steel Foundry
       MACT:  2000	4-11

5-1    Summary Statistics for SBREFA Screening Analysis: 2000  	5-3
                                    Vlll

-------
                                      SECTION 1

                                   INTRODUCTION
       The U.S. Environmental Protection Agency (EPA) is developing a maximum
achievable control technology (MACT) standard to reduce hazardous air pollutants (HAPs)
from the iron and steel foundries source category.  To support this rulemaking, EPA's
Innovative Strategies and Economics Group (ISEG) has conducted an economic impact
analysis (EIA) to assess the potential costs of the rule.  This report documents the methods
and results of this EIA.  In 1997, the United States shipped over 12 million short tons of iron
and steel castings.  These castings are used in a variety of industries including automotive,
aerospace, construction, appliances, and hardware.

1.1    Agency Requirements for an EIA

       Congress and the Executive Office have imposed statutory and administrative
requirements for conducting economic analyses to accompany regulatory actions. Section
317 of the CAA specifically requires estimation of the cost and economic impacts for
specific regulations and standards proposed under the authority of the Act.1 EPA's
Economic Analysis Resource Document provides detailed guidelines and expectations for
economic analyses that support MACT rulemaking (EPA, 1999). In the case of the iron and
steel foundry MACT, these requirements are fulfilled by examining the following:

       •   facility-level impacts (e.g., changes in output rates, profitability, and facility
           closures),
       •   market-level impacts (e.g., changes in market prices,  domestic production, and
           imports),
'In addition, Executive Order (EO) 12866 requires a more comprehensive analysis of benefits and costs for
   proposed significant regulatory actions. Office of Management and Budget (OMB) guidance under EO
   12866 stipulates that a full benefit-cost analysis is required only when the regulatory action has an annual
   effect on the economy of $100 million or more. Other statutory and administrative requirements include
   examination of the composition and distribution of benefits and costs. For example, the Regulatory
   Flexibility Act (RFA),  as amended by the Small Business Regulatory Enforcement and Fairness Act of 1996
   (SBREFA), requires EPA to consider the economic impacts of regulatory actions on small entities.

                                           1-1

-------
       •   industry-level impacts (e.g., changes in revenue, costs, and employment), and
       •   societal-level impacts (e.g., estimates of the consumer burden as a result of higher
          prices and reduced consumption levels and changes in domestic and foreign
          profitability).
1.2    Overview of Coke, Iron and Steel, and Foundry Industries

       Iron and steel foundries produce castings that are either used internally (i.e., captive
foundries vertically integrated into firms such as automobile manufacturers) or sold on the
open market (i.e., merchant foundries).  Although most steel foundries use electric arc or
electric induction processes to melt metal, some foundries employ cupola furnaces that use
foundry coke as a fuel to melt metal.  Figure 1-1 summarizes these interactions between
source categories and markets within the broader iron and steel industry.

       The EIA models the specific links between these product markets.  The analysis to
support the iron and steel foundry EIA focuses on three specific markets:

       •   iron and steel castings and

       •   foundry coke.

Changes in price and quantity in these markets are used to estimate the market, industry, and
social impacts of the iron and steel foundry MACT regulation.

1.3    Summary of EIA Results

       The rule requires some iron and steel foundries to implement pollution control
methods that will increase the costs of producing iron and steel foundries at affected
facilities.  The increased production costs primarily affect iron foundries and will lead to
economic impacts in the form of very small increases in market prices and decreases in
domestic iron castings production.  The impacts of these price increases will be borne largely
by affected iron foundries that use cupola furnaces as well as consumers of iron foundry
products.  Unaffected domestic foundries and foreign producers of coke will earn slightly
higher profits. Key results of the EIA for the iron and steel foundry MACT are as follows:

       •   Engineering Costs: The engineering analysis estimates annual costs for existing
          sources of $21.23 million.
                                          1-2

-------
 55
era'
 I
 I
 3
 O
 CB
 cr
 CD
 CD
 CD
 OJ
 O

 I
 O
1
 CD'
                  Markets for Finished Steel Products
                                   Finished Steel Products
 Foreign
(Imports)
Nonintegrated Steel
  Mills (including
    minimills)
                         Scrap -
                                                    Integrated Iron and
                                                        Steel Mffls
Finishing Mills
                                                     Molten
                                                      Steel
                                  Steelmaking
                                    Furnace
                                                                     Pig Iron
                                                                        Blast Furnace
                                                                    Furnace
                                                                      Coke
                                                                          Captive
                                                                        Coke Plants
                                                          f Markets for
                                                          V  Iron Castings
                                                                                              Furnace
                                                                                               Coke
                                                                                           Furnace, Foundry, and Other
                                                                                                 Coke Markets
                                                                                                                                            Other Coke
-/"' '
k
Merchant
Coke Plants

>
k
Foreign
(Imports)

-------
       •  Price and Quantity Impacts:  The EIA model predicts the following:

          — The market price for iron castings is projected to increase by 0.1 percent
              ($1.12/short ton), and domestic iron castings production is projected to
              decrease by 0.08 percent (8,400 tons/year).

          — The market price for steel castings is projected to increase by less than 0.05
              percent ($0.45/short ton).

          — The market price for foundry coke is projected to remain unchanged, and
              domestic foundry coke production is projected to decrease by less than 0.1
              percent.

       •  Small Businesses:  The Agency has determined that 20 small businesses within
          this source category would be subject to this rule.  The average cost-to-sales ratio
          (CSR) for these firms is 0.40  percent. One small company is projected to have a
          CSR between 1 and 3 percent. However, no small firms are projected to have
          CSRs greater than 3 percent.

       •  Social Costs:  The annual social costs are projected to be approximately $21.22
          million.

          — The consumer burden as a result of higher prices and reduced consumption
              levels is $13.2 million annually.

          — The aggregate producer profits are expected to decrease by $8.0 million.

              /^ The profit losses are $12.1 million annually for affected domestic iron and
                 steel foundries.

              /^ The profit increases are $3.1 million annually for unaffected domestic iron
                 and steel foundries.

              /" Foreign producer profits increase by $1.0 million due to higher prices.

1.4    Organization of this Report

       The remainder of this report supports and details the methodology and the results of
the EIA of the iron and steel foundry MACT.

       •  Section 2 presents a profile of the iron and steel foundry industry.

       •  Section 3 describes the regulatory controls and presents engineering cost
          estimates for the regulation.

       •  Section 4 reports market-, industry-,  and societal-level impacts.


                                          1-4

-------
Section 5 presents the small business screening analysis.

Appendix A describes the EIA methodology.

Appendix B describes the development of the coke battery cost functions.

Appendix C includes the econometric estimation of the demand elasticity for iron
and steel casting products.
                               1-5

-------
                                     SECTION 2

                                INDUSTRY PROFILE
       This section provides a summary profile of the iron and steel castings industry in the
United States. The profile provides background on the technical and economic aspects of the
industry used to support the EIA.  The manufacture of iron and steel castings is included
under Standard Industrial Classification (SIC) codes 3321—Gray and Ductile Iron
Foundries; 3322—Malleable Iron Foundries; 3324—Steel Investment Foundries; and
3325—Steel Foundries, Not Elsewhere Classified.1 Iron and steel castings are used in the
production of over 90 percent of all manufactured durable goods and almost all industrial
equipment (DOE, 1996). Therefore, the demand for iron and steel castings is a derived
demand that depends on a diverse base of consumer products. In 1997, the United States
produced  12 million short tons of iron and steel castings.

       Section 2.1 provides an overview of the production processes and the resulting types
of castings.  Section 2.2 summarizes the organization of the U.S. iron and steel castings
industry, including a description of the U.S. iron and steel foundries, the companies that own
these facilities, and the markets for foundry products. Lastly, Section 2.3 presents historical
data and future projections of the iron and steel foundry industry, including U.S. production
and shipments.

2.1     Overview of Production Process

       A casting is a "metal object obtained by allowing molten metal to solidify in a mold"
(SFSA, 1998). Foundries manufacture castings by pouring metal melted in a furnace into a
mold of a  desired, and potentially intricate, shape. Achieving the same detail of form as a
casting would require extensive tooling and shaping of metal from a mill.  Creating some
very small and precise castings is  impossible by other means than casting.

       The production of castings at foundries involves four distinct processes (see
Figure 2-1). The first process is to make the molds and cores that will shape the casting.
'These SICs correspond to the following North American Industrial Classification System (NAICS) codes:
   331511-Iron Foundries; 331512-Steel Investment Foundries; and 331513-Steel Foundries, except
   Investment.

                                          2-1

-------
                              Mold and Core Materials
                              (Sand, Metal, Plaster, etc.)
                               Mold and Core Making
        Scrap Metal
             Ingots
              Flux
                                         Molds and Cores
Metal Melting
->Slag
 Sand
Metal
                                         Molten Metal
                               Pouring and Shakeout
                            Mold Material
                         ->• Plaster, etc.
                            Sand
                                         Casting
                               Finishing and Cleaning
                         ->• Chemicals
                                 Finished Casting
Figure 2-1. Overview of the Foundry Casting Processes
Foundries use many types of molds, depending on the type, quality, and quantity of castings
required. The two most common mold types are the sand mold and the permanent mold.
Once the mold has been made, the second process involves melting the iron or steel, which is
done almost exclusively by cupolas, electric arc furnaces, and electric induction furnaces.
Once the steel is melted, the third process is pouring the steel into the mold.  After sufficient
cooling time, the casting and mold are separated.  The fourth process is finishing the casting,
which requires smoothing, mechanical cleaning, and, in some cases, coating with a
protective material such as paint or varnish.
                                          2-2

-------
2.1.1   Mold and Core Making

       Iron and steel castings can range in size from ounces to tons, but all molds possess a
few important features. All molds have a vertical channel, called a sprue, through which the
molten metal is poured. Molten metal might flow to many sprues if the mold is designed to
make multiple castings at a time. From the bottom of the sprue, channels direct the metal
into points at the bottom of the mold so that the mold is filled from the bottom up. At the top
of the mold, vertical channels called risers collect excess metal, gases that do not escape
through the mold, and loose sand and other debris that is picked by the molten  metal (SFSA,
1998). As the metal cools, it shrinks somewhat.  For every foot, aluminum alloys shrink
5/32 inch, cast iron shrinks 4/32 inch, and stainless steel shrinks 8/32 inch (LaRue, 1989).
Mold makers must take this shrinkage into account and make the mold slightly oversized.
The risers also serve to compensate for shrinkage by serving as reservoirs of extra molten
metal that can flow back into the mold when the metal begins to shrink.

2.1.1.1 Sand Casting Process

       Most iron foundries pour metal into molds that are made primarily out of sand. The
outer shapes of sand molds are typically made by forming sand into two  halves that are
subsequently joined together.  The inner shapes of the mold that cannot be directly
configured into the mold halves are created by inserting separately made components called
cores, which are  also made of sand. Sand cores are also required in many permanent mold
and centrifugal casting operations.

       Silica sand is  the most commonly used granular refractory material in sand molding.
Other more expensive granular refractory materials are used for specialized applications.
Some of these materials are zircon, olivine, chromite, mullite,  and carbon sands (Schleg and
Kanicki, 1998).  Olivine, for example, is more resistant to fracture than silica sand and
exhibits less thermal  expansion than silica sand (LaRue, 1989).  Sand can be molded to very
precise specifications, and, after solidification by compaction or chemical reaction, sand
molds have sufficient strength to contain a significant volume  of molten metal.

       Ninety percent of all castings are done with green sand (EPA, 1998a). Green sand is
a combination of roughly 85 to 95 percent sand, 4 to 10 percent bentonite clay, 2  to 10
percent carbonaceous materials, and 2 to 5 percent water.  The composition of green sand is
chosen so that the sand will form a stable shape when compacted under pressure, maintain
that shape when heated by the molten metal poured, and separate easily from the  solidified
metal casting. The clay and water bind the  sand together. The carbonaceous materials
                                         2-3

-------
partially volatilize when molten metal is poured into the mold, which serves to create a
reducing atmosphere that prevents the surface of the casting from oxidizing while it
solidifies.  Addition of these materials also helps to control expansion of the mold.
Commonly used materials are powdered coal (commonly called sea coal), petroleum
products, corn starch, wood flour, and cereal (LaRue, 1989; EPA, 1998a).

       Once the green sand is formed around the pattern, the pattern can either be removed,
or additional steps can be taken to improve the quality of the mold. In the skin drying
technique, the outer layer of the mold is dried and coated with a fine layer of crushed
refractory material such as silica, zircon, chromite, or mullite.  This coating provides
resistance to the high temperatures of the hot metal and easier separation of the casting from
the mold. In dry sand molding, foundries bake the green sand mold. A petroleum binder
may be added to the mold before baking to increase the strength of the baked mold.  Baked
molds are stronger than standard green sand molds, and they also produce a smoother finish.

       Chemical binder systems are used when the shape of the mold or core cannot be
made from  green sand or when strength and dimensional stability requirements are too
stringent for green sand to provide. Chemically bonded sand moldings are stronger than
green sand  moldings. The traditional method is to mix sand with a resin or oil and then to
bake the mold.  In the shell process, foundry workers coat sand with a plastic resin and then
blow the sand onto a metal pattern that has been heated to at least 450°F.  The shell process
can be time intensive because the mold or core sometimes must be  slowly heated
progressively from one end to the other. After curing, the chemically bonded molds are
often coated with a finely ground refractory material to provide a smoother surface finish on
the casting.

       Other less often used methods include forming molds by combining sand with
expendable polystyrene (EPS) beads, and vacuum molding  by shaping unbonded sand
around a pattern with a vacuum, which holds the mold in the desired shape.

2.1.1.2 Permanent Mold Casting Process

       Permanent molds must themselves be cast, tooled, and machined, but once the initial
time and expense are invested, foundries can use the mold thousands of times. The most
common metal fashioned into permanent molds is gray iron. Other materials, such as  steel,
copper, and aluminum, can also be used. Permanent molds can be made out of graphite,
which has a chilling effect that enhances particular characteristics of the casting. Molds are
typically hinged to open. Permanent molds may also have water cooling channels and
                                         2-4

-------
ejector pins. The molds do wear out over time and must eventually be replaced. Permanent
molds are most appropriate for large quantities of uniform castings as well as smooth surface
finish and intricate details.

2.1.1.3 Investment Casting Process

       A third, less common casting process is investment casting. In this process, workers
dip wax or plastic molds into a vat of liquid ceramic.  Foundries use wax or plastic so that
the entire pattern can be melted away from the finished mold. The plaster that workers use is
typically gypsum (calcium sulfate) mixed with fibrous talcs, finely ground  silica, pumice
stone, clay, and/or graphite.  Plaster can be 50 percent sand (EPA, 1998a).  Foundries cover
the coated pattern with a layer of refractory material.  Workers may repeat  this process
several times to achieve a mold of desired thickness.  The foundry then heats the mold to
about 1,800°F to harden the mold and burn out the pattern. These molds are best suited for
metals containing titanium and other super-alloys that do not react well with green sand.

2.7.2   Metal Melting

       The primary source  of iron and steel for foundries is scrap.  Workers must  sort scrap,
cut it to fit the furnaces, and clean it.  Scrap cannot have any rust and is cleaned by using
solvents or a precombustion step to burn off residues  (EPA,  1998a). Metal ingot is a
secondary source of iron and steel for foundries. Foundry returns consisting of sprues,
runners, and risers separated from previous castings may contribute a significant share of
input metal.  Foundries can also purchase directly reduced iron (DRI) to employ as an iron
source. Pig iron and DRI dilute the alloy content of the  scrap metal. Foundries add flux
material, typically chloride  or fluoride salts, to the furnace to combine with the impurities in
molten metal in the furnace, forming a dross or slag.  This dross or slag separates from the
molten metal and is removed from the metal before workers tap the furnace.

       Foundries use various alloy metals, such as aluminum, magnesium, nickel,
chromium, zinc, and lead, to alter the metallurgical properties of the resulting product.
Foundries add graphite for carbon content in the production of ductile iron. Twenty percent
of the carbon in ductile iron must come from graphite (Ductile Iron Society, 1998). These
materials may then be melted in furnaces ranging from cupolas to electric induction.

       Table 2-1 provides the number and share of the primary furnaces used by iron and
steel foundries in 1997.  As shown, over 80 percent of furnaces in the industry are electric
induction. Electric arc and  electric induction furnaces use electrical energy to create heat
that melts the metal.  Cupola furnaces use foundry coke  as fuel to melt the  metal.  Coke is

                                         2-5

-------
Table 2-1. Distribution of Iron and Steel Foundry Furnaces by Type:  1997
             Furnace Type
Number
Share (%)
 Cupola
 Electric arc
 Electric induction
 Other
    155
    131
  1,292
     12
    9.8%
    8.2%
   81.3%
    0.8%
 Total
  1,590
  100.0%
Source: U.S. Environmental Protection Agency (EPA). 1998b. Foundry Industry Responses to Information
       Collection Request (ICR) Survey. Database prepared for EPA's Office of Air Quality Planning and
       Standards. Research Triangle Park, NC.

made from metallurgical coal and is purchased by the foundries from merchant coke
producers. The burning coke removes some contaminants and also raises the carbon content
of the metal.  Other furnaces include reverbertory and crucible  types that represent less than
1 percent of furnaces in use during 1997.

2.1.2.1 Cupola Furnace

       As Figure 2-2 illustrates, the cupola furnace is a hollow vertical cylinder that is lined
with refractory material and has doors at the bottom. The charging process begins by laying
sand in the bottom of the furnace and topping the sand with coke, which is ignited.  Next,
workers add carefully measured alternating layers of metal, flux,  and coke until the furnace
is full to the charging door. Air is forced through tuyeres, which  are the holes at the base of
the furnace. The metal melts and drips to the base of the furnace. A tap hole near the top of
the sand layer allows workers to remove the molten metal. Foundries remove slag either
through a slightly higher slag spout or through the tap spout with the metal and separated by
a specially designed spout (LaRue, 1989). As the metal and slag  are removed, additional
layers of charge can be added to the furnace to maintain continuous production. When the
furnace needs to be cleaned and emptied, the doors at the bottom swing open and drop the
contents on a bed of sand.

2.1.2.2 Electric Arc Furnace (EAF)

       EAFs have a rounded, shorter shape than cupola furnaces. Workers charge the
furnace with metal, and carbon electrodes create an arc of electric current. If the arc passes
through the metal, it is considered a direct arc furnace. If the arc  passes above the metal, it is
                                          2-6

-------
                       Spark
                      arrester
                         Lining -"
                          Shell
                      Coke
                    charges


                      Metal
                    charges
                      Lining
                       shelf

                      Coke
                        bed

                       Slag
                      spout

                     Bottom
                     door in
                    dropped
                    position
       Tapping spout
   Bottom plate

Sand bottom
Figure 2-2. The Cupola Furnace

Source: LaRue, James P.  1989. Basic Metalcasting. Des Plaines, IL: American Foundrymen's Society.
                                           2-7

-------
considered an indirect arc furnace. The two models are shown in Figure 2-3.  The electric
arc creates heat, which melts the metal. As the metal melts, workers adjust the electrodes to
maintain their relative position to the top of the charge. Once the metal has melted, workers
insert flux to combine with the impurities, and the metal can be supplemented with alloy
materials.  Doors opposite the spout where the metal is poured out allow workers to remove
the slag. Workers remove the metal by tipping the furnace to pour the liquid.
Indirect
Direct
                        Electrodes
                                                     Spout
                                                                              — 3U- Door
                                                          Furnace tilted to pour

                                                       - Ladle
Figure 2-3. Indirect and Direct Electric Arc Furnaces

Source: LaRue, James P. 1989. Basic Metalcasting. Des Plaines, IL: American Foundrymen's Society.


2.1.2.3 Induction Furnace

       Induction furnaces, or electric induction furnaces as they are sometimes called,
generate heat by passing an electric current through a coil either around or below the hearth.
Furnaces with the coil around the hearth are called coreless induction furnaces, and those
with the coil below the hearth are called channel induction furnaces. Both types are shown
in Figure 2-4.  The electric current generates a magnetic field.  The magnetic field creates
voltage, which moves across the hearth and through the charged metal. As the electric
current attempts to pass through the metal,  it meets resistance, which produces heat to melt
the metal.  Typically, the coils carrying the  electric current are cooled with water.  Induction
furnaces are designed in various shapes and sizes so the tapping and slag removal varies.
                                          2-8

-------
Coreless
 Special cement
Metal bath
                         Asbestos liner
                         Water-cooled copper coil
                         Protective coil liner
                         Monolithic lining
                         Asbestos liner

                         Fire brick base
Channel
  Pouring
   spout
                                                  Air-cooled
                                                   inductor
                                                                  Melting
                                                                  channels
                                                               Circulation
                                                               of melt
                                                               . Blower
                                                               " motor
                                                            Clean-out
                                                            plug
Figure 2-4. Coreless and Channel Electric Induction Furnaces

Source: LaRue, James P. 1989. Basic Metalcasting. Des Plaines, IL: American Foundrymen's Society.


Induction furnaces require cleaner scrap input than EAFs, but induction furnaces make
possible more precise adjustments to the metallurgical properties of the metal (EPA, 1998a).

2.1.3  Pouring and Shakeout

       Workers transport liquid iron and steel directly from the foundry furnaces to the
molds to maintain the liquid state. Some molds, particularly green sand and vacuum molds,
cannot be stored long before they are used. Typically, foundries start the process of melting
before the molds are finished. Permanent molds and chemical molds with strong binders can
be stored for a considerable period without losing their shape.  Carbonaceous material in the
mold burns, creating a reducing atmosphere and prevents the oxidation of the hot metal. In
the vacuum process, a vacuum inside the mold sucks the molten metal up into the mold. The
vacuum pressure is maintained until the casting has solidified. In the lost foam process, the
foam pattern is still inside the mold. As the metal is poured into the mold, the foam
vaporizes and leaves the mold. Certain molds, particularly intricate designs, require pressure
to force the molten metal into all areas of the mold. Some techniques used with permanent
molds require centrifugal casting machines to spin the mold at high speeds. The pressure
                                          2-9

-------
holds refractory material on the walls of the mold and forces the metal into the mold to
eliminate empty spaces.

       Spills of molten metal are called runouts.  Workers must be ready to cover runouts
with sand and to use sand to block the flow of metal if the mold begins to overflow, because
fires can result. No molten metal should be allowed to solidify in the crucible or ladle, so a
standby container such as ingot molds must be ready to receive any excess metal.  The
crucible must be quickly cleaned to prevent build-up.

       After pouring, castings are allowed to cool within the mold.  Rapid cooling increases
casting hardness.  Workers can manually separate castings from the mold, although some
large foundries have vibrating grids to shake the sand off the casting. Certain permanent
molds have ejector pins to push the casting out of the mold.

2.1.4  Finishing and Cleaning

       When castings emerge from the mold,  they are typically hard and brittle. To improve
the metallurgical properties of castings, they are frequently put through heat treatment. Heat
treatment refines the grain of the metal and relieves internal stresses in addition to improving
the metal's properties (Lankford et al., 1985).  Heat treatment must be done with care,
because it can potentially warp or crack the casting. The standard heat treatment is
annealing.  For annealing, foundries place the  casting in a furnace and raise the temperature
slowly, with the target temperature depending on the metal type. For carbon steel, the
temperature is about 1,650°F.  Operators can manipulate the properties of a particular area or
part of a casting by directly applying a flame to the casting.

       New castings require processing after removal from the mold; sometimes the
processing is extensive and is used to modify the basic shape.  Workers remove unwanted
structures with hammers, saws, flame devices such as oxyacetylene torches, and grinders.
Workers can also add structures to the casting at this point, typically by welding.  In any
case, the surface may be rough and contain unwanted substances such as rust, oxides, oil,
grease, and dust.  Foundries typically clean and smooth the casting surface by sand or steel
shot blasting.

       Workers cool and rinse castings with water. The water may contain chemical
additives to prevent oxidation of the casting. Chemical cleaning of the casting can be done
with organic solvents such as chlorinated solvents, naphtha, methanol, and toluene.
Emulsifiers, pressurized water, abrasives, and alkaline agents such as caustic soda, soda ash,
                                         2-10

-------
alkaline silicates, and phosphates are also used for cleaning. Castings may also undergo acid
pickling using hydrochloric, sulfuric, or nitric acid (EPA, 1998a).

       Coatings are used to inhibit oxidation and corrosion, to alter mechanical and
metallurgical properties, and to improve surface finish and appearance. Some coating
processes include painting, electroplating, electroless nickel plating, hard facing, hot dipping,
thermal spraying, diffusion, conversion, porcelain enameling, and organic or fused dry-resin
coating (EPA, 1998a).

       Foundries can typically reuse sand from molds numerous times, although a portion
must be disposed of each time to eliminate the sand that has been worn very fine. Some sand
can be used in construction as filler and for the production of portland cement, concrete, and
asphalt. Much sand from foundries contains chemical binders, and about 2  percent of
foundry waste sand is considered hazardous waste, which requires expensive disposal (EPA,
1998a). Core sand is the most likely sand to be disposed of because it requires the strongest
binders.  To reuse sand, it must be cleaned. Metal particulates must be separated from the
sand.  Various machines are used to break apart sand clumps and grind the binder off the
sand.  Heat can  also be used to break down the resins on sand.

2.7.5   Residuals and By-products

       Resins and binders are left in spills, containers, and outdated materials. Other
residuals include gaseous emissions such as carbon monoxide, volatile organic compounds
(VOCs), and  other HAPs. HAP emissions can occur during mold and core making, melting,
pouring, cooling, and shakeout (PCS) (EPA, 1998a).  Foundries scrub offgases from core-
making processes that use triethylamine gas as a catalyst with acidic solution. Scrubbing
gases generate sludges or liquors, which must be adjusted for pH so that they can be released
as nonhazardous waste.  Sulfur dioxide can be controlled with amine scrubbers that convert
the sulfur dioxide to hydrogen sulfide.  Cleaning  solvents such as methanol,
trichloroethylene, and xylenes are also toxic residuals (EPA, 1998a). The making of
expendable polystyrene patterns leaves chemicals, and the use of lost foam casting generates
organic vapors that may contain HAPs.

       All furnaces emit hydrocarbons, while cupolas and EAFs also emit sulfur dioxide,
nitrogen oxides, chlorides, and fluorides, and cupolas emit carbon monoxide (EPA, 1998a).
Melting furnaces also emit metallic fumes.  The composition of fumes emitted depends on
the type of scrap used as input. Galvanized steel leads to high zinc emissions, and stainless
steel produces greater nickel and chromium emissions than standard carbon steel. Whenever
                                        2-11

-------
the furnace is open, such as for tapping or charging, emissions are highest.  Hoods over the
doors and spouts of the furnaces and near pouring areas capture emissions.

       The water used to cool and rinse the castings picks up lubricants, cleansers, mill
scale, mold coatings, and acids. Water used to cool chemically bonded molds can pick up
chemicals. A sludge may form that contains metals such as cadmium, chromium, and lead.
Foundries are able  to recover acids to be used again.  Iron chloride, when removed from the
acids, is a saleable  product. The water in the acids is recovered by evaporation and
condensation to be re-used for rinsing and cooling. This procedure is less expensive than
transporting and disposing of the acid.  Some foundries are also recovering fluoride from
spent pickling acids in the form of calcium fluoride, which can be used as a flux material that
is more effective than purchased fluorspar (EPA, 1998a).

2.1.5.1 Paniculate Matter

       Particulates are emitted by cupolas and EAFs and to a lesser extent by induction
furnaces.  The emissions of foundry furnaces typically are cleaned with fabric filters
(baghouses), which collect particulates,  or wet scrubbers, which produce waste water and
sludge.

       EAFs release 1 to 2 percent of their charge as dust or fumes.  Lead and cadmium can
be reclaimed if their contents are significantly high. Some techniques send the dust back
through the furnace after the metal has melted so that the dust captures more metal
particulates such as zinc to increase the zinc content above 15 percent.  On  site, foundries
can pelletize EAF dust to be reused in the furnace.  This method is not frequently cost-
effective at the foundry and may be more efficient off site. Some techniques recycle EAF
dust directly back into the furnace, but this approach requires low impurity  content for the
dust.

       The vigorous shakeout operations generate metal and other types of dust. In addition,
as permanent molds gradually wear out they produce metallic particles.  The dust requires air
filtering by using electrostatic precipitators, baghouses, or wet scrubbers. Dust from sand
systems can be used by cement companies and can potentially supply 5 to 10 percent of the
raw material used by cement manufacturers (EPA,  1998a).
                                         2-12

-------
2.7.5.2 Slag
       Slag can have a complex composition at foundries. Foundry slag may contain metal
oxides, melted refractory materials, sand, coke ash, and other impurities. If the slag contains
enough toxic metals such as lead, cadmium, or chromium, the slag will be classified as a
hazardous material. Some foundries making ductile iron use calcium carbide as a flux for
desulfurization, resulting in slag that is classified as a reactive waste because it is potentially
explosive (EPA, 1998a). Metal can be reclaimed by allowing the slag to solidify and then
crushing it.  Metal can be extracted from crushed slag by hand or with magnets. Reusing
slag in different iron production lines can sometimes reduce the hazardous content of the
slag.

2.1.6   Production Costs

       Table 2-2 shows production costs for foundries by type. Total costs are greatest for
gray and ductile iron foundries across all categories due to higher production volume.
Table 2-2 also shows the average variable cost and average total cost by type of foundry.
Gray and ductile iron foundries have the lowest costs per short ton, while the per-ton cost of
steel castings is more than three times that of gray and ductile iron.

Table 2-2. Summary of Production Costs by Foundry Type (1996)

Cost Element
Variable inputs ($106)
Production labor
Materials
Fuels and electricity
General and administrative costs ($106)
Capital expenditures ($106)
Total costs C$106)
AVC ($/short ton)
ATC ($/short ton)
Gray and Ductile Iron
Foundries
$7,857.8
$2,261.1
$5,040.6
$556.1
$1,677.1
$515.7
$10.050.6
$747.65
$956.29
Malleable
Foundries
$230.6
$92.2
$115.3
$23.1
$65.7
$15.8
$312.1
$876.81
$1,186.69
Steel
Foundries
$2,911.2
$956.2
$1,757.1
$197.9
$789.3
$151.4
$3.851.9
$2,290.48
$3,030.61
AVC = average variable cost
ATC = average total cost
Source: U.S. Department of Commerce. February 1998. 1996 Annual Survey of Manufactures:  Statistics for
       Industry Groups and Industries.  M96(AS)-1. Washington, DC:  U.S. Government Printing Office.
                                         2-13

-------
       Table 2-3 displays costs of materials and their shares by type of casting.  Gray and
ductile iron foundries use scrap as a greater share of their input costs compared to malleable
iron and steel. The ability to use large amounts of low-cost scrap contributes to the low price
of gray and ductile iron castings.

       Table 2-4 shows employees and earnings for iron and steel foundries.  The number of
employees and production workers decreased until the early 1990s as shipments decreased.
The number of employees, including production workers, has increased from the lows of the
1980s, but not to the highs of the early 1980s. Hourly earnings have consistently increased
every year since 1980.

2.7.7   Metal Types

       The most basic variation in castings stems from manipulating the charge material.
Four basic types of metal are melted in foundry furnaces: gray iron, ductile iron, malleable
iron, and steel. Each type of iron and steel has a general range of characteristics. Further
variation in mechanical properties of the casting can be achieved during cooling and
finishing operations. Table 2-5 shows the volume of the iron and steel castings in 1997 by
casting type.  The majority of all ferrous castings in 1997 was gray iron, followed by ductile
iron.

       Because gray iron was the first cast iron, some people use the term cast iron to refer
to gray iron. Gray iron received its name from the color of the graphite flakes dispersed
throughout the silicon iron matrix of the metal. The graphite flakes do not contribute
strength or hardness, but they can have some positive  benefits such as dimensional stability
under differential heating and high vibration damping. Gray iron has the greatest damping
capacity, followed by ductile iron, then malleable, and finally steel (Foti et al., 1998).
Foundries can add alloys to gray iron to increase the hardness and employ heat treatments to
soften gray iron, making it easier to form but decreasing its strength.

       Ductile iron was invented in the 1940s.  It is similar to gray iron, although the
graphite is in spheroids or spherulites rather than flakes. Because of the spheres, ductile iron
is sometimes called nodular iron.  The spheroids are created by adding a controlled amount
of magnesium to the molten iron, which alters the way the graphite is formed. The formation
of graphite prevents ductile iron from shrinking when it solidifies, as occurs in malleable iron
and cast steel. Ductile iron is known for being capable of a wide range of yield strengths,
high ductility, and  ease of being shaped.
                                         2-14

-------
Table 2-3. Distribution of Material Costs by Foundry Type (1992)
Gray and Ductile
Iron Foundries

Material
Ferrous and nonferrous shapes and forms
Purchased scrap
Steel, clay, glass, and concrete products
Industrial patterns, dies, molds, and other
machinery and equipment
Sand
All other materials
Total3
Delivered
Cost
($106)
$314.7
$785.1
$86.7
$76.2
$110.2
$1,258.5
$2,637.4

Share
11.9
29.8
3.3
2.9
4.2
47.7
100.0
Malleable Foundries
Delivered
Cost
($106)
$2.2
$13.9
$0.3
NA
$0.5
$53.1
$72.6

Share
3.0
19.1
0.4
NA
0.7
73.1
100.0
Steel Foundries
Delivered
Cost
($106)
$260.9
$137.0
$71.0
$37.4
$34.3
$521.6
$1,062.4

Share
24.6
12.9
6.7
3.5
3.2
49.1
100.0
NA = not available
a Totals may not sum due to undisclosed costs for some categories.
Source:  U.S. Department of Commerce. 1995. 7992 Census of Manufactures: Industry Series—Blast Furnaces, Steel Works, and Rolling and
        Finishing Mills Industry.  Washington, DC: U.S. Government Printing Office.

-------
Table 2-4. Summary of Labor Statistics for SIC Code 332, Iron and Steel Foundries:
1980-1997
Production Workers
All
Year Employees (103)
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
208.8
200.8
158.6
139.0
148.6
141.4
130.9
129.8
136.4
137.3
132.4
125.8
120.2
119.0
125.1
131.1
128.5
130.0
Production
Workers
(103)
167.3
159.9
121.8
106.3
117.5
111.6
102.9
102.4
109.4
109.7
105.3
99.5
96.1
94.8
101.4
107.1
105.2
106.7
Average
Weekly Average Weekly
Earnings Hours
328.00
354.99
353.98
397.79
421.45
427.76
438.01
460.53
477.86
475.42
486.26
491.71
522.09
555.89
608.72
597.19
604.78
636.64
40.00
39.4
37.3
40.1
41.4
40.7
41.4
42.8
43.6
42.6
42.1
41.6
42.9
44.4
45.7
44.5
44.6
46.1
Average
Hourly
Earnings ($)
8.20
9.01
9.49
9.92
10.18
10.51
10.58
10.76
10.96
11.16
11.55
11.82
12.17
12.52
13.32
13.42
13.56
13.81
Source: U.S. Department of Labor, Bureau of Labor Statistics.  BLS LABSTAT Database: Employment and
       Earnings, SIC 33.  . Obtained in March 2002.
       Malleable iron is the result of heat treating iron over an extended period. Similar to
ductile iron, the majority of the carbon content in malleable iron is in nodules. As suggested
by the name, malleable iron is soft and can be bent without immediately breaking.
                                         2-16

-------
Table 2-5. Shipments of Iron and Steel Castings by Type (1997)

Iron
Ductile iron
Gray iron
Malleable iron
Steel
Total
Volume
(103 short tons)
10,790
4,333
6,186
271
1,217
12,007
Share
(%)
89.9%
36.1%
51.5%
2.3%
10.1%
100.0%
Source: U.S. Environmental Protection Agency (EPA). 1998b. Foundry Industry Responses to Information
       Collection Request (ICR) Survey. Database prepared for EPA's Office of Air Quality Planning and
       Standards. Research Triangle Park, NC.

       Steel products made by casting processes have mechanical properties inferior to those
of steel products manufactured by steel mills. The advantage of using the casting process to
make steel products is that casting is the most direct method for making products of a
specific shape.
2.2    Industry Organization

       This section presents information on the manufacturing plants within this source
category and the companies that own and operate these foundries.

2.2.1  Manufacturing Plants

       Figure 2-5 identifies the number of U.S. iron and steel foundries by state. Iron and
steel foundries are located in nearly every state, and Ohio has the most for a single state, with
79 iron and steel foundries. The remainder of this section characterizes these foundries using
facility responses to EPA's industry survey and industry data sources.

       Tables 2-6 and 2-7 present summary data by type of producer, merchant or captive.
Merchant producers are foundries that purchase their inputs and sell their products on the
open market.  Captive foundries are vertically integrated with iron and steel and/or coke
producers. As of 1997, the United States had 860 reported iron-making furnaces and 730
reported steel-making furnaces.  In U.S. foundries, iron melting capacity is nearly ten times
the steel melting capacity. Most furnaces for iron and steel making are electric induction.
For the 545 iron and steel foundries that reported the relevant information of the total 798
                                          2-17

-------
Figure 2-5.  Number of U.S. Iron and Steel Foundries by State:  1997
affected iron and steel foundries, total hourly capacity in 1997 for iron melting was 41,298
tons and for steel melting was 4,737 tons.

2.2.2   Companies

       The National Emission Standard for Hazardous Air Pollutants (NESHAP) will
potentially affect business entities that own iron and steel foundry facilities. Facilities
comprise a site of land with plant and equipment that combine inputs (raw materials, energy,
labor) to produce outputs (castings). Companies that own these facilities are legal business
entities that have capacity to conduct business transactions and make business decisions that
affect the facility. The terms facility, establishment, plant, and mill are synonymous in this
analysis and refer to the physical location where products are manufactured.  Likewise, the
terms company and firm are synonymous and refer to the legal business entity that owns one
or more facilities. Figure 2-6 shows the possible chains of foundry ownership.
                                        2-18

-------
Table 2-6. Iron and Steel Foundry Data by Type of Producer:  Iron-Making Furnaces
(1997)

Iron-making furnaces
Number (#)
Cupola
Electric arc
Electric induction
Other
Total
Capacity (short tons/hour)
Cupola
Electric arc
Electric induction
Other
Total
Merchant
Foundries


35
0
170
0
205

1,139
0
2,154
0
3.293
Captive Alla
Foundries Foundries


107
2
493
5
607

10,132
48
27,433
206
37.819
a Not all survey respondents identified their production by type. Therefore, merchant and captive
data do add to totals shown for all foundries.
Table 2-7. Iron and Steel Foundry
(1997)

Steel-making furnaces
Number (#)
Electric arc
Electric induction
Other
Total
Capacity (short tons/hour)
Electric arc
Electric induction
Other
Total
Data by Type of
Merchant
Foundries


69
341
5
415

772
1,078
4
1.854
Producer: Steel-Making


155
2
698
5
860

11,307
48
29,737
206
41.298
foundries
Furnaces
Captive Alla
Foundries Foundries


31
176
2
209

269
1,830
1
2.100


129
594
7
730

1,342
3,390
5
4.737
  Not all survey respondents identified their production by type. Therefore, merchant and captive foundries
  data do add to totals shown for all foundries.
                                           2-19

-------
Parent Company
(Direct Owner)
i
k
Facility

Parent Company
i

Subsidiary
Company
(Direct Owner)
i

Facility

Parent Company
i
t
Other Companies
or Legal Entities
>
t
Subsidiary
Company
(Direct Owner)
>

Facility
Figure 2-6. Possible Ownership Configurations for U.S. Iron and Steel Foundries
       The Small Business Administration (SBA) defines small businesses based on size
standards developed for North American Industrial Classification System (NAICS).  The
SBA defines firms owning iron and/or steel foundries as small if they have 500 or fewer
employees. As shown in Figure 2-7,78 percent of affected U.S. companies with available
data meet the small business definition. Table 2-8 shows the distribution of companies by
the number of foundries owned: 6 percent of small companies own more than one foundry,
while 34 percent of large companies own more than one foundry. Table 2-9 summarizes
foundry operations by firm size. Even though there are nearly three times as many reporting
small companies as there are large companies, the number of iron-making and steel-making
furnaces is near the number owned by small companies. The mean number of furnaces for
large companies versus small reflects the distribution of furnaces between the two groups.
                                       2-20

-------
                      Large
                      22.1%
                                                              Small
                                                              77.9%
Figure 2-7. Distribution of Affected U.S. Companies by Size:  1997a

a Reflects distribution for only those 584 companies owning U.S. iron and steel foundries with data allowing
  identification as small or large business.
Table 2-8.  Distribution of Companies by Number of Foundries:  1997a

Company Size
Category
Small
Large
All companies
Number of Foundries Owned per Company

1
299
74
373

2
17
19
36

3
3
7
10

4
0
4
4

5 or more
0
8
8

Total
319
112
431
  Data reported for only those foundries with complete responses to EPA industry survey.
                                           2-21

-------
Table 2-9.  Summary of Iron and Steel Foundry Operations by Firm Size Category:  1997a
Foundries (#)b
Company
Size
Category

Small
Large
Total

Small
Large
Total
Number of
Companies Total

319
112
431

319
112
431

342
215
557

1.05
1.81
1.20
Iron
Making

245
146
391

0.54
1.13
0.59
Steel
Making

152
79
231

0.33
0.61
0.35
Iron-Making Furnaces
Capacity
(short
Number tons/hour)
Sample
447
408
855
Sample
0.98
3.16
1.29
Totals
31,132
10,163
41,295
Means
68.42
78.78
62.10
Annual
Production
(103 short
tons)

2,268.4
12,351.6
14,620.0

5.0
95.7
22.0
Steel-Making Furnaces
Number

399
329
728

0.88
2.55
1.10
Capacity
(short
tons/hour)

1,794
2,924
4,718

3.94
22.67
7.12
Annual
Production
(103 short
tons)

380.7
1,371.2
1,751.9

0.8
10.6
2.7
a  Data reported for only those foundries with complete responses to U.S. EPA industry survey.
b  Foundries that produce iron and steel shown once in each column.

Source:  U.S. Environmental Protection Agency (EPA).  1998b. Foundry Industry Responses to Information Collection Request (ICR) Survey. Database
        prepared for EPA's Office of Air Quality Planning and Standards. Research Triangle Park, NC.

-------
2.2.3   Industry Trends

       The number of metal casting foundries in the United States has dropped by nearly
half since 1955 (Heil and Peck, 1998). During the 1970s, orders for foundry castings
exceeded annual capacity. Profit margins were high, and shipments were often late due to
the excess demand. In the presence of excess demand, the foundry industry did not
experience pressure to improve casting quality. Foreign producers gained a foothold during
the recession of the 1980s and 1990s. In addition to lower prices, foreign producers had
more modern equipment than U.S. producers, which allowed foreign producers to have
higher quality castings. The number of U.S. foundries steadily dropped and capacity
utilization was below 50 percent in the mid-1980s. Even though the number of foundries has
not increased to the highs of the 1970s and 1980s, output per producer has risen (Heil and
Peck, 1998).

       U.S. iron and steel foundry production has increased since the lows at the beginning
of the 1990s. Gray iron castings production has increased with the health of the economy,
while ductile iron saw greater gains because in some applications it replaced steel castings
and forgings, as well as malleable iron castings. Malleable iron castings production has
decreased, because more than one-third of malleable iron foundries in the United States have
closed since the 1980s; this trend is expected to continue. Malleable iron castings production
has declined to mostly small custom orders and captive operations (Heil and Peck, 1998).

       Similar to gray and ductile iron, steel castings production has increased since the
lows of the early 1990s. The primary issue of concern in the 1990s for steel castings has
been the  cleanliness of the steel (Tardiff, 1998). Steel with low impurities has superior
mechanical properties, improving the position of steel against possible substitutes such as
aluminum and reinforced plastics. A general trend among ferrous castings is demand for
low-weight parts, particularly among the transportation industry as it seeks greater fuel
efficiency.  New casting techniques allow metals to be cast with thinner dimensions,
reducing overall weight.

2.2.4   Markets

       The markets for the various types of iron and steel castings overlap but are not
identical, because the properties and costs of each type vary.  During the 1970s and 1980s,
many iron and steel foundries were captively owned. As end product production dropped,
many captive foundries were left with excess capacity. To avoid the fixed costs of idle
foundries, companies shut down captive foundries or sold them to produce for sale directly to
                                         2-23

-------
the market.  Of those foundries that are still captive, the majority are iron-producing.
Nineteen percent of 1997 iron casting shipments were captive, while only 3.5 percent of
1997 steel castings were captive (DOC, 1997). Table 2-10 shows the market and captive
shares for iron and steel castings.

Table 2-10. U.S. Shipments of Iron and Steel Castings by Market and Captive:  1997
(103 short tons)
Market

Casting
Gray
Ductile
Malleable
Steel

Volume
4,693
3,925
121
1,174
Share
(%)
75.9%
90.6%
44.6%
96.5%
Captive

Volume
1,493
408
150
43
Share
(%)
24.1%
9.4%
55.4%
3.5%
Total
Volume
6,186
4,333
271
1,217
Sources:  U.S. Department of Commerce, Bureau of the Census. 1988-1997.  Current Industrial Reports.
        Washington, DC: U.S. Government Printing Office.
        U.S. Department of Energy.  1996. "Trends Effecting [sic] R&D in the Metalcasting Industry."
        Prepared by BCS Incorporated for Office of Industrial Technologies, Washington, DC.

       Automotive and aerospace have traditionally been the largest consumers of gray and
ductile iron castings.  Substitutes such as composites and aluminum have gained share in
these markets due to reduced weight and corrosion resistance.  Pipes and pipe fittings are
another major market for gray and ductile iron castings.  Improvements in ductile iron that
increased strength and durability have allowed it to be a reduced-cost substitute for forged
and cast steel in some applications.

       Appliances, hardware, aerospace, and automotive components have been the major
uses for malleable iron castings.  Plastics, nonferrous metals, as well as other types of iron
and steel castings have displaced malleable iron from many applications (Heil and Peck,
1998). Ductile iron castings are  responsible for the majority of malleable iron castings
displacement, particularly in plumbing and electrical.

       Steel castings are used in many of the same markets as iron castings, including
automotive, aerospace, and construction. Steel castings are also extensively used in the
railroad industry (BTA, 1996). In addition, steel investment castings  are used in a diverse
range of industries employing small or very thin castings, including jewelry.
                                         2-24

-------
2.3    Historical Industry Data

       This section presents domestic production, imports, exports, and apparent
consumption. We also present historic market price. Finally, this section shows past iron
and steel castings shipments by end-user market and discusses various projections for the
future of the shipments in the next decade.

2.3.1   Domestic Production

       Table 2-11 shows iron and steel castings shipments.  Shipments were at their lowest
over the 10-year period in 1991 for all types except malleable iron.  Shipments for gray and
ductile iron, as well as steel castings, have increased over 25 percent since the lows of 1991.
Malleable iron castings shipments were lowest in 1992, and although shipments increased in
the mid-1990s along with the other types of iron and steel castings, malleable iron castings
shipments have declined. Gray iron castings shipments have also declined slightly, while
ductile iron castings shipments have consistently increased every year since 1991.

2.3.2   Foreign Trade and Apparent Consumption

       The only year for which import and export data are available for iron and steel
castings is 1994. We used the import and export data to derive apparent consumption by
subtracting exports and adding imports to production.  Table 2-12 presents these data.
Table 2-13 provides the export and import concentration ratios for the types of iron and steel
castings. Export and import concentration ratios represent the share of U.S. production
expected and the share of apparent consumption imported. Concentration ratios for iron and
steel castings are typically around 7 percent. Foreign producers of iron and steel castings
gained a foothold in the 1980s and early 1990s when foreign prices  were lower than those of
U.S. castings, and foreign quality was equal or better (Heil and Peck, 1998).

2.3.3   Market Prices

       We derived prices for iron and steel castings by dividing the quantity of shipments by
the value of shipments, generating an average price. Table 2-14 shows the prices from 1987
through 1997. Gray iron castings are consistently the lowest priced, which explains the
steady share of castings maintained by gray iron. Ductile iron castings are consistently lower
priced than malleable iron castings. Ductile iron castings are displacing malleable iron
castings for many applications because of their lower price,  strength, and durability.
                                         2-25

-------
Table 2-11.  U.S. Shipments of Iron and Steel Castings:  1987-1997 (103 short tons)
Iron Castings
Year
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997

1987-1997
1987-1992
1992-1997
Ductile
3,044
3,210
3,321
3,186
2,789
3,051
3,267
3,709
4,304
4,312
4,333

4.2%
0.0%
8.4%
Gray
5,701
5,941
5,638
5,073
4,609
4,800
5,215
6,401
6,260
6,198
6,186
Average
0.9%
-3.2%
5.8%
Malleable
321
323
299
290
262
253
278
300
293
263
271
Annual Growth
-1.6%
-4.2%
1.4%
Total
9,066
9,474
9,258
8,549
7,660
8,104
8,760
10,410
10,857
10,773
10,790
Rates
1.9%
-2.1%
6.6%
Steel
Castings
1,013
1,187
1,184
1,133
957
986
1,021
1,039
1,160
1,271
1,217

2.0%
-0.5%
4.7%
Total
10,079
10,661
10,442
9,682
8,617
9,090
9,781
11,449
12,017
12,044
12,007

1.9%
-2.0%
6.4%
Source: U.S. Department of Commerce, Bureau of the Census. 1988-1997. Current Industrial Reports.
       Washington, DC: U.S. Government Printing Office.

2.4    Market Shipments and Future Projections

       Future projections for iron and steel castings take into account the strength of the
economy, the strength of the  U.S. dollar, interest rates, end-user product markets, input
supply, and development of substitutes. The American Foundrymen's Society (AFS)
projects that the metal casting industry in general will experience declines until 2002 and
then increases until 2004, which AFS expects could possibly be the strongest year for
castings in the past two decades (AFS, 1998). AFS expects gray and ductile iron castings
shipments to do well early in the next decade because it will be the peak period for baby
boomers to purchase vehicles, although the share of gray iron per vehicle will continue to
drop.  A short-term downturn in the strength of the economy, followed by an expansion from
2002 through 2008, should maintain gray and ductile iron castings shipments for farm and
construction equipment and tools. AFS projects that malleable iron castings shipments will
continue a rapid decline.

                                        2-26

-------
Table 2-12. U.S. Production, Foreign Trade, and Apparent Consumption of Iron and
Steel Castings: 1994 (103 short tons)
Type
Iron castings
Ductile iron
Gray iron
Malleable iron
Steel castings
Steel investment
Steel castings, n.e.c.
Iron and steel castings
U.S. Production
10,411
3,710
6,401
300
1,129
91
1,038
11,540
Exports
741
276
442
23
96
NA
NA
837
Imports
831
213
579
39
113
NA
NA
944
Apparent
Consumption3
10,501
3,647
6,538
314
1,146
NA
NA
11,647
NA = not available
a Apparent consumption is equal to U.S. production minus exports plus imports.
Sources:  U.S. Department of Commerce, Bureau of the Census.  1988-1997. Current Industrial Reports.
         Washington, DC: U.S. Government Printing Office.
         U.S. Department of Energy.  1996. "Trends Effecting [sic] R&D in the Metalcasting Industry."
         Prepared by BCS Incorporated for Office of Industrial Technologies, Washington, DC.
Table 2-13. Foreign Trade Concentration Ratios for Iron and Steel Castings by Type:
1994


                                   Export Concentration Ratio3    Import Concentration Ratiob
	Type	(%)	(%)	
Iron castings                                   7.1%                          7.9%
    Ductile iron                                 7.4%                          5.8%
    Gray iron                                  6.9%                          8.9%
    Malleable iron                              7.7%                         12.3%
Steel castings                                  8.5%                          9.9%
Iron and steel castings (combined)                  7.3%                          8.1%

NA = not available
a  Measured as export share of U.S. production.
b  Measured as import share of U.S. apparent consumption.
Source:  U.S. Department of Energy.  1996. "Trends Effecting [sic] R&D in the Metalcasting Industry."
        Prepared by BCS Incorporated for Office of Industrial Technologies, Washington, DC.
                                            2-27

-------
Table 2-14.  Market Prices for Iron and Steel Castings by Type:  1987-1997
Iron Castings
Year

1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997

1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
Ductile

$624.05
$640.22
$681.51
$725.46
$751.31
$700.26
$777.39
$800.22
$851.43
$921.04
$957.81

3,044
3,210
3,321
3,186
2,789
3,051
3,749
3,709
4,304
4,312
4,333
Gray

$529.92
$567.26
$602.31
$641.59
$641.96
$662.77
$669.35
$600.00
$719.33
$720.06
$759.87

5,701
5,941
5,638
5,073
4,609
4,800
9,128
6,401
6,260
6,198
6,186
Malleable
Prices ($/short ton)
$885.05
$1,040.56
$947.16
$1,057.59
$1,062.21
$869.96
$923.81
$923.42
$924.59
$1,011.72
$1,006.33
Quantities (103 short tons)
321
323
299
290
262
253
292
300
293
263
271
Steel

$5,311.45
$5,192.42
$2,793.67
$2,863.64
$3,125.71
$2,780.22
$3,990.01
$5,816.15
$5,472.04
$5,253.90
$5,159.39

1,013
1,187
1,184
1,133
957
986
1,461
1,039
1,160
1,271
1,217
All Castings

$1,050.23
$1,118.53
$885.85
$941.68
$965.97
$910.80
$1,033.73
$1,146.70
$1,230.43
$1,276.84
$1,282.79

10,079
10,661
10,442
9,682
8,617
9,090
14,630
11,449
12,017
12,044
12,007
Source: U.S. Department of Commerce, Bureau of the Census. 1988-1997. Current Industrial Reports.
       Washington, DC: U.S. Government Printing Office.
                                           2-28

-------
       Table 2-15 provides projections for castings shipments from a different group,
Business Trend Analysts (BTA).  BTA expects shipments to increase consistently until 2005
for all types except malleable iron.

Table 2-15. Projected U.S. Shipments of Iron and Steel Castings by Type:  1997, 2000,
and 2005 (103 short tons)

Year
1997
2000
2005

Ductile
3,289.6
3,395.6
3,505.6
Iron
Gray
5,420.5
5,577.7
5,614.6
Castings
Malleable
203.5
232.4
194.0
Steel Castings
Total
8,913.6
9,205.7
9,314.2
Average Annual Growth
1997-2005
1997-2000
2000-2005
0.8%
1.1%
0.6%
0.4%
1.0%
0.1%
-0.6%
4.7%
-3.3%
0.6%
1.1%
0.2%
Investment
40.3
41.6
45.2
Rates
1.5%
1.1%
1.7%
All Other
1,557.8
1,784.9
1,945.0

3.1%
4.9%
1.8%
Total
1,598.1
1,826.5
1,990.2

3.1%
4.8%
1.8%
Source: Business Trend Analysts. 1996. "Foundry Products and Markets in the U.S.—Company Profiles and
       Ferrous Castings."
       BTA separates projected castings shipments by market, as displayed for iron in
Table 2-16. The greatest decreases in shipments have been for soil pipe (shown on continued
page), and BTA expects these decreases to accelerate as iron pipe is replaced by PVC pipe.
Displacement by PVC will reduce the annual growth rate for iron pressure pipe, but the
growth rate is expected to stay positive. From 1987 to 1997, machinery was a strong and
growing area for iron castings, and BTA expects this trend to continue.  BTA projects that
the relatively small market of railroad equipment will have the strongest growth rate as
ductile iron replaces some steel castings.

       Table 2-17 shows historical shipments and BTA projections for steel castings. BTA
expects growth rates for nearly all markets to decrease over the next decade from the growth
rates of the 1990s.  Motor vehicles, defense, and aerospace are exceptions, which BTA
projects will climb back to positive growth rates. Railroad equipment has been and will
continue to be the largest and fastest growing market for steel  castings, although BTA
projects the growth rate to decrease, as ductile iron replaces  steel castings.
                                        2-29

-------
Table 2-16. Iron Castings Shipments by Market: 1987-2005 (103 short tons)

Year
1987
1990
1992
1995
1996
1997"
2000a
2005a

1987-2005
1987-1997
1997-2005


Motor Valves and
Vehicles Fittings
2,623.8
2,323.7
2,180.8
2,629.2
2,381.5
2,325.5
2,386.1
2,294.0

-0.7%
-1.1%
-0.2%
337.8
402.4
403.6
452.7
458.9
460.2
511.7
536.1

3.3%
3.6%
2.1%


Construction Railroad
Machinery Equipment
425.0
479.7
418.9
534.4
582.8
568.5
607.8
625.0

2.6%
3.4%
1.2%
14.0
21.8
22.6
27.2
23.3
24.9
29.6
32.0
Average Annual
7.1%
7.8%
3.6%


Mining
Engines Equipment
542.4
464.1
516.4
584.3
484.7
504.2
533.9
557.0
Growth Rates
0.1%
-0.7%
1.3%
8.0
8.3
6.9
8.2
8.4
8.7
9.6
9.9

1.3%
0.9%
1.7%

Hardware
9.0
9.4
8.9
9.4
9.5
9.7
9.9
10.0

0.6%
0.8%
0.4%

Pressure
Pipe
1,200.0
1,565.0
1,449.1
1,487.0
1,516.7
1,586.0
1,606.9
1,650.6

2.1%
3.2%
0.5%
Farm
Machinery
and
Equipment
546.8
695.0
801.1
815.5
774.8
783.2
855.1
917.4

3.8%
4.3%
2.1%
                                                                                                 (continued)

-------
      Table 2-16. Iron Castings Shipments by Market: 1987-2005 (103 short tons) (Continued)
K>
Metalworking
and Industrial
Year Machinery
1987
1990
1992
1995
1996
1997"
2000a
2005a

1987-2005
1987-1997
1997-2005
485.4
575.9
533.7
618.0
647.6
668.4
716.1
774.9

3.3%
3.8%
2.0%
Ingot
Mold
476.0
330.0
310.0
219.4
172.9
151.0
147.0
139.4

-3.9%
-6.8%
-1.0%
Soil
Pipe
378.0
325.7
273.4
234.4
222.7
211.6
145.6
106.9

-4.0%
-44%
-6.2%
Municipal
550.0
552.2
523.2
553.8
560.5
567.6
570.0
575.0
Average Annual
0.3%
0.3%
0.2%
Power
HVAC Compressors Transmission
142.0
154.9
138.3
137.0
133.7
134.0
145.5
146.8
Growth Rates
0.2%
-0.6%
1.2%
192.0
228.7
230.2
245.7
250.3
251.6
258.2
265.3

2.1%
3.1%
0.7%
108.0
109.6
115.2
115.8
117.5
119.3
121.1
122.9

0.8%
1.0%
0.4%
Other
427.0
441.6
421.0
430.0
431.5
437.8
436.8
435.8

0.1%
0.3%
-0.1%
Total
8,465.2
8,688.0
8,353.3
9,102.0
8,777.3
8,812.2
9,090.9
9,199.0

0.5%
0.4%
0.5%
      a Forecasts
      Source: Business Trend Analysts. 1996. "Foundry Products and Markets in the U.S.—Company Profiles and Ferrous Castings."

-------
Table 2-17.  Steel Castings Shipments by Market:  1987-2005 (103 short tons)
General and
Railroad Construction
Year Equipment Equipment
1987
1990
1992
1995
1996
1997a
2000a
2005"

1987-2005
1987-1997
1997-2005
630.0
956.0
811.0
1,102.5
1,050.0
952.0
1,130.5
1,260.0

5.6%
5.1%
4.0%
59.2
69.6
58.4
91.2
85.4
83.7
99.1
106.6

4.4%
4.1%
3.4%
Valves Special
Mining Motor and Industrial Metalworking
Machinery Vehicles Fittings Machinery Machinery
149.0
154.2
128.9
152.3
157.2
162.2
178.3
184.0

1.3%
0.9%
1.7%
88.8
91.9
76.8
90.8
72.6
75.7
86.2
89.6
Average
0.1%
-1.5%
2.3%
72.6
86.5
87.1
100.4
102.2
102.7
105.1
111.7
Annual Growth
3.0%
4.1%
1.1%
64.0
76.2
76.7
88.5
90.1
90.5
92.7
98.5
Rates
3.0%
4.1%
1.1%
34.0
44.3
41.1
42.1
43.4
44.3
46.3
48.6

2.4%
3.0%
1.2%
Defense
and
Aerospace
20.7
25.0
15.0
12.9
13.0
13.2
14.2
16.7

-1.1%
-3.6%
3.3%
Other
68.1
71.8
68.4
71.2
72.4
73.8
74.1
74.5

0.5%
0.8%
0.1%
Total
1,186.4
1,575.5
1,363.4
1,751.9
1,686.3
1,598.1
1,826.5
1,990.2

3.8%
3.5%
3.1%
a  Forecasts
Source: Business Trend Analysts.  1996. "Foundry Products and Markets in the U.S.—Company Profiles and Ferrous Castings."

-------
                                    SECTION 3

                         ENGINEERING COST ANALYSIS
       Control measures implemented to comply with the MACT standard will impose
regulatory costs on iron and steel foundries. This section presents compliance costs for
affected foundries and the national estimate of compliance costs associated with the rule.
These engineering costs are defined as the annual capital and operating and maintenance
costs, assuming no behavioral market adjustment by producers or consumers. For input to
the EIA, engineering costs are expressed per unit of iron casting production and used to shift
the individual supply functions in the market model.

       The MACT standard covers the iron foundry and steel foundry source categories.
The processes covered by the regulation include the melting furnace; scrap preheating;
pouring, cooling, and shakeout (PCS); mold and core coating; and mold and core making.
EPA estimates that approximately 100 iron and steel foundries are major sources of HAP
emissions.  For the purposes of developing an estimate of the environmental and economic
impacts, EPA identified 98 specific foundries that are projected to be major sources of HAP
emissions.  Consequently, the economic impacts of the MACT standard are based on the
compliance costs projected for these 98 foundries. Capital, operating, and maintenance and
monitoring costs were estimated for each plant.  New or upgraded control equipment will be
required at 52 of the foundries, while all  98 foundries will be required to install additional
monitoring equipment.

3.1    Overview of Emissions from Iron Foundries

       A variety of metal HAPs are contained in the particulate matter (PM) emitted from
iron foundries, primarily from the furnace melting operations. The primary metal HAPs
emitted from iron and steel foundries are manganese and lead, with concentrations typically
ranging from 1 to 5 percent of PM. Trace quantities (generally less than 0.1 percent of PM)
of antimony, arsenic, beryllium, cadmium, chromium, cobalt, mercury, nickel, and selenium
are also present. Some iron and steel foundries produce iron alloys that increase the
concentration of certain metal HAPs in the emitted PM, most commonly chromium and/or
nickel. By controlling PM emissions, the foundry effectively controls its metal HAP
emissions.  The  PM control devices most commonly used at foundries are  baghouses and
venturi scrubbers.
                                        3-1

-------
       Organic HAP compounds are released from a variety of sources. The organic HAP
emission source common to nearly all foundries is the PCS lines, where organic material in
the mold or core is vaporized or partially combusted after the hot metal is poured into the
mold. The organic HAPs from PCS lines include benzene, toluene, formaldehyde,
acetaldehyde, and polycyclic organic matter (POM); the primary POM is naphthalene.
Organic HAP emission controls, such as incineration and carbon adsorption, are used for a
few selected PCS lines, comprising roughly 1 percent of the total number of PCS lines
operated by iron and steel foundries.

       Chemical additives (or "binders")  are commonly added to the sand cores and
occasionally added to the sand molds to increase the strength of the cores and molds.
Depending on the composition of the binder systems employed, significant quantities of
organic HAPs may be released during the mold- and core-making processes. The organic
HAPs released are specific to the type of binder system employed, but important organic
HAPs for mold- and core-making operations include cumene, formaldehyde, methanol,
naphthalene (a POM), phenol, and  triethylamine (TEA). Binder systems using TEA gas are
commonly controlled by acid/wet scrubbing. Except for the  TEA gas binder systems, the
only available organic HAP emission control technique appears to be binder reformulations
with non-HAP or reduced-HAP binder systems.

       Melting furnaces, primarily cupolas and scrap preheaters, have  a potential to emit
trace amounts of organic HAPs including POM (such as polynuclear aromatic hydrocarbons
and chlorinated dibenzodioxins and furans), and volatile organics such as benzene, carbon
disulfide, toluene, and xylene. Organic HAP emissions for these sources typically are
controlled through afterburning or direct-flame incineration.

       Overall, the iron and steel foundries MACT standard is expected to reduce metal
HAP emissions by 102 tons per year (tpy) and organic HAP emissions  by 720 tpy.  The
standard is also expected to reduce PM emissions by 1,780 tpy and volatile organic
compound (VOC) emissions by 770 tpy. The emission reductions will result from replacing
existing venturi scrubbers with new baghouse control systems on cupolas; installing
afterburners in cupolas that do not currently use afterburning; installing PM control systems
for currently uncontrolled electric induction furnaces (EIFs) and pouring stations; installing
new acid/wet scrubbers at foundries that currently have uncontrolled TEA gas binder
systems; and setting binder and coating formulation limitations for mold- and core-making
lines.
                                        3-2

-------
3.2    General Approach for Estimating Compliance Costs

       EPA conducted a detailed survey of the iron and steel industries in 1998 to gather
information regarding the types of processes and control devices currently used by each
foundry. This survey information was used to identify the specific processes within each
foundry that would need to be upgraded or to have new control equipment added. The
control costs were estimated using the cost algorithms described in the OAQPS Control Cost
Manual (EPA, 1991c) and the Handbook: Control Technologies for Hazardous Air
Pollutants (EPA, 1991b). The control costs were estimated in fourth-quarter 1998 dollars.1
Costs of the control systems are driven primarily by the flow rate of the exhaust gas
requiring treatment. Typical vent stream characteristics (e.g., flow rates per unit capacity or
throughput, temperature) were developed from data reported in response to the detailed
questionnaires. Costs also were included for monitoring devices associated with the control
equipment, such as temperature monitors, pressure monitors, flow rate monitors, and bag
leak detection systems. Finally, costs were included for recordkeeping and reporting
requirements.  More details regarding the control costs estimated for specific processes are
provided in the following sections.

3.3    Cupola Melting Furnace Control Systems

       The MACT standard establishes PM emission limits (as a surrogate for metal HAP)
for cupola melting furnaces based on baghouse control systems. Venturi scrubbers are not
expected to be able to meet the PM limits when used to control cupola emissions.
Consequently, foundries whose cupolas currently are controlled using venturi scrubbers are
expected to have to replace  their existing venturi scrubbing control systems with baghouses.
To estimate the costs of replacing venturi scrubbers with baghouses, essentially two cost
estimates were made.  First, the capital investment costs and the annual operating and
maintenance costs (AOCs) of a new baghouse system were estimated.  Second, the AOCs of
a venturi wet scrubber system were estimated because these costs were already being
incurred by the foundries and  offset the operating cost of the baghouse. Additionally, the
MACT standard establishes organic HAP emission limits for cupolas.  Therefore, costs were
also developed for installing an afterburner to cupola furnace exhaust streams that currently
do not use afterburning.
1 Cost estimates were calculated in 1998 dollars because the detailed industry survey provided a snapshot of the
industry in 1998.

                                         3-3

-------
3.3.1   Baghouse Control Costs for Cupola Melting Furnaces

       Baghouse or fabric filter control costs were estimated using the CostAir program
(EPA, 1991c). The fabric filters were designed as pulse-jet modular systems with an air-to-
cloth ratio of 2.37 ft per minute using Nomex bags. Auxiliary equipment included the cost
of a new fan, motor, two dampers, 300 ft of ductwork, and new stack.  The additional damper
and ductwork (300 ft versus a typical value of 100  ft) and an additional system pressure drop
of 8 in of water were included in the cost analyses to roughly simulate the  added capital and
operating costs associated with cooling the cupola exhaust stream prior to  the baghouse by
using the hot exhaust gases to preheat the cupola blast air.  These costs would be incurred
because a baghouse control system cannot operate at as high an inlet gas temperature as a
venturi scrubber control system. A retrofit cost factor of 2.2 was applied to the total capital
investment cost estimate to capture costs of removing existing control equipment and to
include additional costs associated with the exhaust stream cooling system and site-specific
difficulties anticipated with a system retrofit of this nature. All cost values were calculated
in fourth-quarter 1998 dollars (Vatavuk Air Pollution Control Cost Index = 110.9).

       Control costs for six different sizes of baghouses were calculated based on the
anticipated range of vent stream flow rates. The baghouse flow rates considered ranged from
20,000 to 280,000 actual cubic feet per minute (acfm), which approximately covers a range
of cupola melting furnace capacities from 10 to 140 tons per hour (tph). The total capital
investment and the annual  operating and maintenance costs for these model baghouse
systems are summarized in Table 3-1.  The calculated control costs for these systems were
essentially linear over the flow rates investigated; a linear regression analysis of the capital
and the operating and maintenance control costs had R2 values  of 0.999 and 0.993,
respectively. Consequently, a simple linear expression was derived to estimate the total
capital investment (TCI) and the AOC  based on the system exhaust flow rate as follows:

                             TCIBH  = 510,100 +32.90 QBH                         (3.1)

                             AOCBH = 95,820 + 4.703 CU                         (3.2)

where

       TCIBH  = total capital investment for a baghouse, $ (fourth quarter 1998);

       AOCBH = annual operating and maintenance cost for a baghouse,  1998 $/yr; and

       QEH   = design exhaust vent flow rate based on cupola-baghouse, acfm.
                                         3-4

-------
     Table 3-1.  Summary of Control Costs for Baghouses and Wet Scrubbers:  1998
Flow Rate Through
Control
Device — Baghouse
(Wet Scrubber), acfm
20,000 (15,400)
40,000 (30,800)
80,000 (61,500)
120,000 (92,300)
200,000 (154,000)
280,000 (215,000)
Baghouse
Total Capital
($103)
$1,064
$1,814
$3,208
$4,542
$7,126
$9,648
Baghouse
Annual
Capital
($103)a
$100
$171
$303
$429
$673
$911
Baghouse
Annual
Operating
($103/yr)
$167
$259
$509
$687
$1,041
$1,392
Baghouse
Total
Annual
($ioV)
$267
$430
$812
$1,116
$1,714
$2,303
Wet Scrubber
Annual
Operating
($10Vyr)
$217
$343
$599
$855
$1,365
$1,867
 ' Reflects capital recovery based on a 20-year life and 7 percent interest rate.
The TCI and AOC for each cupola baghouse were calculated using these equations and the
maximum anticipated flow rate based on the cupola melt capacity and the cupola exhaust
system design (above or below gas takeoff).  Table 3-2 provides the flow rate factors used to
estimate the exhaust stream flow rate based on the cupola melting capacity and exhaust
system design.

      Table 3-2. Estimating Exhaust Air Flow Rates for Control Costs Estimates
 Cupola Charge Position and Type of Air Pollution Control Device    Flow Rate Factor (acfm/tph)a
 Above-Charge Takeoff // Fabric Filter
 Above-Charge Takeoff // Wet Scrubber
 Below-Charge Takeoff// Fabric Filter
 Below-Charge Takeoff// Wet Scrubber
3,000
2,200
2,500
1,800
 ' Adjusted for typical operating temperatures of approximately 500 °F.
       A capital recovery factor (CRF) of 0.0944 was used for baghouses to annualize the
capital investment on the basis of a 20-year equipment life and an annual interest rate of
7 percent. The total annualized cost (TAG) was calculated as the sum of the annualized
capital investment cost and the annual operating and maintenance cost (e.g., TCI x CRF +
AOC).
                                          3-5

-------
3.3.2   Venturi Scrubber Control Costs for Cupola Melting Furnaces

       The costs of operating a venturi scrubber with a pressure drop of 40 in of water were
estimated using the cost algorithms described in the Handbook: Control Technologies for
Hazardous Air Pollutants (EPA, 1991b).  The cost of wastewater disposal was assumed to be
approximately the same as the water consumption cost, an assumption that likely understates
the operating cost of the venturi scrubber.  The control costs were converted from 1989 to
1998 dollars using the Chemical Engineering Plant Cost Index (from 355 for base year to
389.5).  As with baghouses, the operating costs for venturi scrubbers were calculated for six
different exhaust vent flow rates, and a linear regression analysis was performed.  Because
the operating and maintenance costs for venturi scrubbers are driven by the fan electrical
usage, the water consumption, and the wastewater treatment costs, the AOC for venturi
scrubbers is linear with exhaust stream flow rate (R2 = 0.99999).  The resulting AOC
equation for venturi scrubbers is

                             AOCVS = 90,250 +  8.262 Qvs                         (3.3)

where

       AOCVS =  annual operating and maintenance cost for a venturi scrubber, 1998 $/yr,
                 and

       Qvs    =  design exhaust vent flow rate based on cupola- venturi scrubber, acfm.

As shown in Table 3-2, the average flow rate per  furnace capacity is approximately 30
percent higher when baghouse systems are employed than when venturi scrubbers are used.
This is thought to be caused primarily by additional air sucked into the exhaust system (the
vent is at a negative pressure with respect to atmospheric) when the exhaust stream is cooled
prior to the baghouse. Consequently, the operating  and maintenance costs for venturi
scrubbers are provided in Table 3-1 for flow rates ranging from 15,400 to 215,000 acfm,
because these costs are more comparable to the baghouse costs reported in Table 3-1 on the
basis of cupola melting capacity.

3.3.3   Net Metal HAP Control Cost for Cupola Melting Furnaces

       The net control costs for replacing a venturi scrubber with a baghouse control system
were calculated using the following equations:
                                                                                (3.4)

                                      = AOCBH - AOCVS                         (3.5)

                                 = CRF x TCIvs.BH + AOCVS.BH                   (3.6)
                                         3-6

-------
where
       TCIVS,BH  =  total capital investment for replacing a venturi scrubber with a
                    baghouse, 1998 $;
       AOCVS,BH =  net annual operating and maintenance cost for replacing a venturi
                    scrubber with a baghouse, 1998 $/yr;

       TACVS,BH =  total annualized cost for replacing a venturi scrubber with a baghouse,
                    1998 $/yr; and
       CRF      =  capital recovery factor = 0.0944 (20 years; 7 percent interest).
3.3.4  Sample Calculation of Metal HAP Control Cost for Cupola Melting Furnaces
       The annual operating and maintenance cost of an existing venturi scrubber system
was first calculated based on the melting capacity of the furnace. For example, given an
above-charge takeoff cupola with a melting capacity of 50 tph, the design vent stream flow
rate for a cupola-venturi scrubber system was calculated to be Qvs = 50 tph x 2,200 acfm/tph
(flow rate factor from Table 3-2) = 110,000 acfm. The annual operating and maintenance
cost of the existing venturi scrubber system was then calculated using Eq. (3.3) to yield an
AOCVS = $999,000/yr.

       The design exhaust flow rate of the new fabric filter system was then calculated as
QBH = 50 tph x 3,000 acfm/tph (factor from Table 3-2) = 150,000 acfm.  The control costs for
the new fabric filter system were then calculated using this revised design exhaust flow rate
as shown in Eqs. (3.1) and (3.2), yielding TCIBH = $5,445,000 and AOCBH = $801,000/yr.
       The net control costs were  then calculated using Eqs. (3.4) and (3.5) to yield TCIVS,BH
= $5,445,000; AOCVS.BH = ($198,000/yr);  and TACVS.BH = $316,000/yr.
3.3.5  Afterburning Control Cost for Cupola Melting Furnaces
       Afterburning control costs  were estimated using the CostAir program for incinerator
systems (EPA, 1991c). The incinerators were designed to operate at a minimum of 1,300 °F.
From data collected during EPA source tests, it was assumed that the temperature of the
cupola exhaust stream entering the incinerator/afterburner was 500 °F. This inlet gas stream
was assumed to contain adequate oxygen,  coming from air entering  the cupola exhaust
stream through the charge door opening. The CO concentration after dilution with the
charge door ventilation air was assumed to be 5 percent. The incinerator/afterburner was
assumed to operate without heat recovery, and a retrofit cost factor of 1.2 was applied to  the
total capital investment cost estimate.
                                         3-7

-------
       Control costs for 10 different sizes of incinerators were calculated based on the
anticipated range of vent stream flow rates because of a shift in the cost curves identified for
gas flows less than 40,000 acfm. The  shift in the cost curve function can be seen in
Figure 3-1. Subsequently, two control cost equations were developed for each cost
parameter (total capital investment and the annual operating and maintenance cost): one for
systems of less than 40,000 acfm and one for systems of 40,000 acfm or more. A log-log
correlation was used for the total capital investment cost curves.  The calculated control cost
equations are:

For systems O^ < 40,000 acfm

                      TClAB = 1,000 x exp[2.997 + 0.2355 InCQ^)]                   (3.7)

                              AOCAB = 36,360 + 2.1130^                          (3.8)
For systems O^ > 40,000 acfm

                      TClAB = 1,000 x exp[3.339 + 0.2355 InCQ^)]                   (3.9)

                              AOCAB = 60,430 + 2.040 Q^                        (3.10)

where

       TCI^     =   total capital investment for an afterburner, $ (fourth quarter 1998);

       AOC^   =   annual operating and maintenance cost for an afterburner, 1998 $/yr;
                     and

       QAB       =   design exhaust vent flow rate at afterburner inlet, acfm.

A linear regression analysis of these control cost equations had R2 values of 0.9999 or
greater.

       The cupola inlet gas flow rate was estimated using the flow rate factors presented in
Table 3-2.  These  flow rate factors were developed from systems that had afterburners, but
the flow rate measurements were made downstream of the cupola afterburner. As such, use
of the flow rate factors in Table 3-2 is expected to yield cost estimates that are biased high.
Nonetheless, because cupola afterburners generally operate with a minimum of auxiliary fuel
(CO in the exhaust stream being the primary fuel), applying the flow rate factors in Table 3-2
should result in a reasonable estimate of the flow rate at the inlet to the afterburner.
                                         3-8

-------
             20,000    40,000    60,000   80,000   100,000  120,000   140,000   160,000   180,000  200,000
                                      Row rate (acfrri)
                 Figure 3-1.  Control Cost Curves for Cupola Afterburners
       A capital recovery factor of 0.1424 was used for baghouses to annualize the capital
investment on the basis of a 10-year equipment life and an annual interest rate of 7 percent.
The total annualized cost was calculated as the sum of the annualized capital investment cost
and the annual operating and maintenance cost (e.g., TCI x CRF + AOC).

3.3.6   Sample Calculation of Organic HAP Control Cost for Cupola Melting Furnaces

       Continuing the example of an above-charge takeoff cupola with a melting capacity of
50 tph, the design exhaust flow rate was estimated as QAB = 50 tph x 3,000 acfm/tph (factor
from Table 3-2) = 150,000 acfm. The control costs for the new afterburner system were then
calculated using Eqs. (3.9) and (3.10), yielding TCI^ = $467,000 and AOCBH = $366,000/yr.
Using the capital recovery factor of 0.1424, TAC^ = $433,000/yr.
                                         3-9

-------
3.4    Electric Induction, Scrap Preheater, and Pouring Station Control Systems

       The MACT standard also establishes PM emission limits (as a surrogate for metal
HAP) for other types of melting furnaces commonly used at foundries (i.e., electric induction
furnaces and electric arc furnaces [EAFs]), as well as scrap preheaters and pouring stations.
These emission limits are again based primarily on baghouse control systems. All EAFs at
the 98 foundries expected to be major sources of HAP emissions were controlled using
baghouses.  Some EIFs, scrap preheaters, and pouring stations employed venturi wet
scrubbers. However, these emission sources do not have the same level of uncontrolled
emissions as cupolas or EAFs, and well-operated and maintained venturi scrubbers should be
able to meet the PM limits for existing EIFs, scrap preheaters, and pouring stations. As such,
costs were only estimated for adding new PM control systems (assumed to be baghouses for
costing purposes) for sources that currently do not operate a control system.

       Additionally, the MACT standard establishes requirements for scrap preheaters. All
scrap preheaters must either use direct-gas  flame scrap preheating or meet an organic HAP
emission limit. The existing scrap preheaters, as operated at the 98 foundries projected to be
major sources of HAP emissions, are expected to meet the MACT requirements without
additional control costs. Consequently, this section summarizes control costs for baghouses
for EIFs, scrap preheaters, and pouring stations.

3.4.1   Baghouse Control Costs for EIFs and Scrap Preheaters

       As with the baghouse costs developed for cupolas, baghouse control costs for the
control of EIFs and scrap preheater PM emissions were estimated using the CostAir program
(EPA, 1991c). However, the fabric filters in service for these emission sources, based on the
information in the detailed industry survey, generally operate at much lower temperatures
and at significantly higher air-to-cloth ratios than cupola baghouses. The  EIF/scrap preheater
fabric filters were designed as pulse-jet modular systems with an air-to-cloth ratio of
5.1 acfm/ft2 using polyester bags.  Auxiliary equipment included the cost of a new fan,
motor, one damper, 40 ft of ductwork, and new stack. A retrofit cost factor of 1.2 was
applied to the total capital investment cost to estimate the retrofit costs for all baghouse
systems installed to control emissions from scrap preheaters and from EIFs that already have
a capture system (but no control device).2  As before, all cost values were calculated in fourth
quarter 1998 dollars (Vatavuk Air Pollution Control Cost Index = 110.9).

       Control costs for 10 different sizes of baghouses were calculated based on the
anticipated range of vent stream flow rates. There  is a noticeable shift in the operating cost
2EIFs with no capture systems were assumed to elect to meet the opacity limit for the foundry rather than install
a capture and control system for their EIF.

                                         3-10

-------
curve for gas flows between 40,000 and 50,000 acfm (see Figure 3-2). Subsequently, two
control cost equations were developed for each cost parameter (total capital investment and
the annual operating and maintenance cost): one for systems of less than 50,000 acfm and
one for systems of 50,000 acfm or more.  Overall, the baghouse flow rates considered ranged
from 5,000 to 180,000 acfm.  A linear regression analysis of the capital and the operating and
maintenance control costs resulted in R2 values exceeding 0.999 for each size range for each
cost parameter.
    1,600
    1,400
        0      23,000    40,000    60,000    80,000    100,000    123,000   140,000   160,000    180,000   200,000
                                       Row Rate (acfm)

           Figure 3-2. Control Cost Curves for EIF/Scrap Preheater Baghouses
       From the linear regression analysis, the following control cost equations were
developed:

For systems QETE/gPIJ < 50,000 acfm
                           TCIEiF/SpH = 63,840 + 8.162 QEIF/SPH
                           AOCEIF/SPH = 63,870 + 1.458 QEIF/SPH
(3.11)

(3.12)
                                         3-11

-------
For systems QEiF/SpH ^ 50,000 acfm
                           TCIEIF/SPH = 99,230 +7.437 QEIF/SPH

                          AOCEIF/SPH = 133,900 + 1.429 QEIF/SPH

where
                                                                  (3.13)

                                                                  (3.14)
       TCI
           EIF/SPH
       AOC
            EIF/SPH
       Q
        -EIF/SPH
             total capital investment for an EIF/scrap preheater baghouse,
             1998 $;

             annual operating and maintenance costs for an EIF/scrap
             preheater baghouse, 1998 $/yr; and

             design exhaust vent flow rate based on EIF/scrap preheater,
             acfm.
Again, a capital recovery factor of 0.0944 was used for baghouses to annualize the capital
investment on the basis of a 20-year equipment life and an annual interest rate of 7 percent.

       The design exhaust flow rate of the EIF control system was estimated based on the
melting capacity of the EIF. The design exhaust flow rate for a scrap preheater control
system was estimated based on the number of scrap preheaters requiring control.  If a
foundry needed to add controls for both its EIFs and its scrap preheaters, then a single
baghouse would need to be designed to control the combined system flow rate. Therefore,
the EIF/scrap preheater control system flow rate was calculated as
                                 Q,
                                   EIF/SPH
                          = QEIF + Q;
                                                  SPH
(3.15)
where
       QEIF

       QsPH
=  design exhaust vent flow rate based on EIF capacity, acfm and
=  design exhaust vent flow rate based on number of scrap preheaters, acfm.
       The EIF control system flow rate was calculated from EIF melting rate capacity.
Additionally, if pouring station emission controls were also required at the foundry, then the
EIF control system would need to be designed to include the flow rate from the pouring
station capture system, as well.  Specifically, the EIF exhaust flow rate was estimated as
                   QEIF = 5,000 x NumEIF x (MeltCapEIF)°6o7 + Q
                                                               PourSt
                                                                  (3.16)
where
       NumEIF     =  number of EIF of given melting rate capacity;

       MeltCapEIF   =  melting rate capacity of the EIF, tons/hr; and

                                         3-12

-------
       Qpourst        =  design exhaust vent flow rate based on number of pouring stations,
                       acfm (see Section 3.4.2).

The factor of 5,000 was assigned based on a 5 ft x 5 ft canopy hood with an entrance design
velocity of 200 ft/min for a 1 tph EIF. The capture system exhaust flow rate from other EIF
melting furnaces was assumed to be proportional to the cross-sectional area of the furnace
(or to the 2/3 power of the capacity of the furnace). If pouring stations also required control
at the foundry, the exhaust from the pouring station capture systems was assumed to be
added to the EIF exhaust stream prior to the control device.

       The flow rate of a scrap preheater control system was calculated as a simple function
of the number of scrap preheaters requiring PM controls, as follows:

   Number of Scrap Preheaters Requiring Control:          Exhaust Flow Rate:
                          1                                QSPH =20,000 acfm
                      2, 3, or 4                             QSPH = 60,000 acfm
                      5 or more                             QSPH = 100,000 acfm

These scrap preheater exhaust flow rates were used directly as the control system flow rates
in Eqs. (3.11) through (3.14) if the EIFs at a given foundry did not need control. That is,
pouring station exhaust flows were only combined  with EIF/scrap preheater control systems
when the EIFs required control. Otherwise, costs for separate control systems were
developed when a foundry required control of a scrap preheater and a pouring station but not
an EIF.

3.4.2   Baghouse Control Costs for Pouring Stations

       Baghouse control costs for the controlling PM emissions from pouring stations were
again estimated using the CostAir program (EPA, 1991c). The design used for pouring
station control systems was based on baghouses used to control PM emissions from pouring,
cooling, and shakeout lines.  As such, the cost curves presented in this section can be used to
estimate baghouse costs for controlling pouring, cooling, or shakeout PM emissions.
However, these equations were employed only for pouring station emission control and then
only when no additional EIF emission control was required at the foundry. The pouring
station baghouses were designed as pulse-jet modular systems with an air-to-cloth ratio of
6.8 acfm/ft2. Auxiliary equipment included the cost of a new fan, motor, one damper, 40 ft
of ductwork, and new stack.  A retrofit cost factor of 1.2 was applied to the total capital
investment cost estimate. Control system costs were only estimated for pouring stations with
existing capture systems but no additional PM control system. Pouring stations with no
                                        3-13

-------
capture systems reported were assumed to elect to meet the opacity limit for the foundry
rather than install a capture and control system for their pouring station.  Again, all cost
values were calculated in fourth-quarter 1998 dollars (Vatavuk Air Pollution Control Cost
Index = 110.9).

       Control costs for nine different sizes of baghouses were calculated based on the
anticipated range of vent stream flow rates.  As with the EIF/scrap preheater operating cost
curve,  there is a noticeable shift in the operating cost curve for gas flows between 40,000 and
50,000 acfm (see Figure 3-3). Subsequently, two control cost equations were developed for
each cost parameter (total capital investment and the annual operating and maintenance
cost):  one for systems less than 50,000 acfm and one for systems of 50,000 acfm or more.
Overall, the baghouse flow rates for which direct cost estimates were developed ranged from
5,000 to 180,000 acfm. A linear regression analysis of the capital and the operating and
maintenance control costs resulted in R2 values exceeding 0.999 for both size ranges and for
each cost parameter evaluated. The resulting control cost equations follow.

For systems Qp01]rSt < 50,000 acfm

                            TCIPourSt = 63,360 + 6.732 QPourSt                        (3.17)

                            AOCPourSt = 63,720 +  1.412 QPourSt                       (3.18)

For systems QEIF/SPH ^ 50,000 acfm
                            TCIPourSt = 99,100 + 5.892 QPomSt                        (3.19)

                           AOCPourSt = 133,900 + 1.378 QPourSt                      (3.20)

where

       TCIPourSt   =  total capital investment for pouring station baghouse, 1998$;

       AOCPomSt =  annual operating and maintenance cost for pouring station baghouse,
                     1998 $/yr; and

       Qpourst     =  design exhaust vent flow rate based on number of pouring stations
                     requiring additional control, acfm.

Again, a capital recovery factor of 0.0944 was used for baghouses to annualize the capital
investment on the basis of a 20-year equipment life  and an annual interest rate of 7 percent.
                                         3-14

-------
     1,200
     1,000
      800
   o
   o
   o
   O
   O
      200
         0     20,000    40,000   60,000    80,000   100,000   120,000   140,000   160,000   180,000   200,000
                                        Row Rate (acfrr)
              Figure 3-3. Control Cost Curves for Pouring Station Baghouses
       The flow rates for the pouring station control systems were calculated assuming each
pouring station capture system was a 5 ft x 4 ft canopy hood with an entrance design velocity
of 200 ft/min, so that each pouring station requiring control contributed 4,000 acfm to the
pouring station control system.  However, if only one pouring station control system required
control at a foundry, the pouring station baghouse was designed for a flow rate of 6,000 acfm
(essentially a 6 ft x 5 ft canopy hood with an entrance design velocity of 200 ft/min).
    Number of Pouring Stations Requiring
                   Control:
                       1
                   2 or more
     Exhaust Flow Rate:
      Qpourst = 6,000 acfm
Qpourst  = 4,000 acfm x # Pouring
           Stations
                                         3-15

-------
       As discussed in Section 3.4.1, the pouring station emissions were assumed to be
combined with the EIF emissions if the foundry was required to add a control system for the
EIFs. Even though the cost of an EIF/SPH baghouse at any given flow rate is higher than the
cost of a similar-sized pouring station baghouse (because of the different air-to-cloth ratios
assumed for these systems), it is still more cost-effective to install a single control system at
the lower air-to-cloth ratio than to install two separate control systems. It is also likely that
foundries that have to control both scrap preheater and pouring station emissions will install
a single control system for both of these emission sources to save on costs.  However,  based
on the logic used in the control cost model, separate control systems were designed for these
emission sources if no additional control was required for the EIFs.

3.5    Mold- and Core-Making Control Systems

       The MACT standard establishes two emission reduction measures for mold- and
core-making lines. For binder systems that employ a triethylamine gas catalyst, the emission
reduction method is the installation of an acid/wet (absorptive) scrubber. Five of the 96
foundries had uncontrolled TEA emissions from their mold- and core-making TEA gas
binder systems. The costs associated with the acid/wet scrubber for TEA emissions controls
are described in this section.  For furan warm-box systems, there are also restrictions on the
methanol content of the catalyst portion of the binder system. Alternative water-based
catalyst formulations, however, appear to be available that can meet the restrictions
pertaining to methanol in the furan warm-box catalyst formulation with no further costs
associated with the adaptation.

       Costs associated with the installation and operation of acid/wet scrubbers to control
emissions of TEA were calculated using the cost algorithms reported in the OAQPS Control
Cost Manual, 5th Edition  (EPA, 1991c). Based on the TEA usage rates at the five foundries
with uncontrolled TEA mold- and core-making lines, three scrubber control systems were
sized based on removing 10 tpy, 25 tpy and 125 tpy. These model acid/wet scrubber systems
were then assigned to each of the uncontrolled TEA mold- and core-making lines based on
the current TEA usage rates.  The model scrubbers generally had a 25 to 50 percent excess
capacity compared to the projected usage rates for any given uncontrolled TEA mold-  and
core-making line for which it was assigned. From the available source test data, TEA  inlet
(uncontrolled) concentrations ranged from 10  to 130 ppm. However, the systems with the
highest flow rates also had the highest TEA concentrations. Therefore, the small- and
medium-sized scrubbers (10 and 25 tpy of TEA systems) were assumed to operate 4,000
hrs/yr with an inlet TEA concentration of 50 parts per million by volume (ppmv; median
value from the test data).  The larger scrubber, capable of removing 125 tpy of TEA, was
assumed to operate 4,000  hrs/yr and at an inlet TEA concentration of 100 ppmv.
                                        3-16

-------
       The cost functions presented in the OAQPS Control Cost Manual are provided in
third-quarter 1991 dollars. These costs were scaled to 1998 dollars by using the Chemical
Engineering Plant Cost Index (using 361 for 1991 and 389.5 for 1998). The calculated
control costs for the three acid/wet scrubbers are as provided in Table 3-3.

     Table 3-3. Summary of Control Costs for Acid/Wet Scrubbing Systems: 1998
Model Scrubber
Scrubber 1 (lOtpy)
Scrubber 2 (25 tpy)
Scrubber 3 ( 125 tpy)
Total Capital
($103)
$87
$157
$309
Annual
Capital3
$12.3
$22.3
$44.1
Annual Operating
($103/yr)
$25.0
$31.6
$50.8
Total Annual
($103/yr)
$37.3
$53.9
$94.9
a Reflects capital recovery factor of 0.1424 based on a 10-year life and a 7 percent interest rate.

3.6    Monitoring, Reporting, and Recordkeeping

       Most of the monitoring requirements in the iron and steel foundries MACT standard
are continuous parameter monitoring requirements. The standard requires each baghouse to
be equipped with a bag leak detection system.  Foundries with TEA scrubber systems are
also required  to install and operate a pH monitoring system and gas and liquid flow rate
monitors. For venturi wet scrubbers used to meet a PM emission limit, the monitored
parameters include the pressure drop and gas and liquid flow rates.  Finally, some
recordkeeping and reporting costs were estimated for implementing a scrap selection and
inspection program; conducting performance tests; preparing a start-up, shut-down, and
malfunction plan; preparing operating and maintenance records; and maintaining records of
HAP usage rates associated with coatings and chemical binders used for mold- and core-
making.  These costs are described in the following sections. All capital costs for monitoring
and recordkeeping equipment were annualized using a capital recovery factor of 0.1424
based on 10-year equipment life and 7 percent interest rate.

3.6.1   Bag Leak Detection Systems

       Each baghouse will need to be equipped with a bag leak detection system. These
systems will have an installed capital cost of $9,000 each, with an annual operating cost  of
$500/yr (EPA, 1998a).  There are a total of 288 baghouses, either existing or required to  be
installed, at the 98 major source iron and steel foundries. Consequently, the total capital cost
for bag leak detectors was calculated as $2.59 million, with an annual operating cost of
$144,000/yr, resulting in a total annualized cost for bag leak detectors of $513,000/yr.
                                         3-17

-------
3.6.2   Parameter Monitoring Systems

       The costs for parameter monitoring systems were estimated using a generic parameter
monitoring system. Monitoring system costs were evaluated from control equipment supply
company catalogues for pH, pressure, and flow measurement systems and associated
electronic recording systems. These costs ranged from $1,500 to $3,000 per monitoring and
data recording system (in 1998 dollars).  Therefore, the general cost of equipment for any
parameter monitoring system was estimated to be $2,500.  It was estimated that the
installation, calibration, troubleshooting, training, and quality assurance procedure
development costs for a new monitoring system would be $5,000, so that the total installed
cost per new monitoring system would be $7,500.

       The annual operating costs were estimated to be $2,000/yr.  These costs are largely
for calibration and maintenance of the equipment, but they also include summarizing and
annual reporting of the data.

       Costs were not separately included for temperature monitoring systems on thermal
destruction devices. These monitoring devices are integral to the control device and must be
properly maintained for proper control device performance. Thus, costs for temperature
monitors are included in the operating and maintenance costs  of the new afterburners, and
foundries currently operating thermal combustion systems are assumed to already have and
maintain the required temperature monitors.

3.6.2.1 Parameter Monitoring Systems for Venturi (PM) Wet Scrubbers

       Existing venturi wet scrubbing systems are expected to be able to meet PM emission
limits from EIFs, scrap preheaters, and pouring stations. If a venturi scrubbing system is
employed, both the pressure drop and scrubbing liquid flow rate must be monitored.  Both of
these monitoring systems were assumed to be in place at each venturi scrubber control
device.  Annual operating costs for both parameter monitoring systems were assumed to be
$4,000/yr ($2,000 x 2).  Eighteen existing venturi wet scrubbing systems were associated
with EIFs, scrap preheaters, and pouring stations at the major source foundries,  so the total
annualized  monitoring costs for these systems would be $72,000.

3.6.2.2 Parameter Monitoring Systems for Acid/Wet Scrubbing Systems

       The acid/wet scrubbing systems used to control TEA emissions from mold- and core-
making operations are required to monitor flow rate and pH of the scrubbing liquor. Because
the equipment costs for new scrubbers typically includes these monitors, it was  assumed that
all acid/wet scrubbing systems have flow monitors already in place, and it was assumed that
all systems  would have to have a pH monitor installed. Consequently, the monitoring costs
                                        3-18

-------
per scrubbing system would be $7,500 capital costs and $4,000 annual operating and
maintenance costs.  These costs were projected for 46 TEA acid/wet scrubbing systems.

3.6.3   Foundry Recordkeeping, Reporting, and Compliance Costs

       Several work practices in the MACT standard require foundries to prepare plans and
maintain records of certain emissions-reducing activities.  It was assumed that all foundries
routinely provided an ignition source to mold vents after pouring if these vents did not auto-
ignite. Therefore, no control costs were attributed to these activities; however, costs were
attributed to performing an initial assessment of which mold vents were required to be
ignited under the regulation. Costs were also estimated for developing a scrap selection and
inspection plan, conducting performance tests, and maintaining records of HAP usage rates
in coating materials and binder formulations.  These costs were calculated primarily based on
an estimate of the technical person-hours required to complete each activity and the
frequency of occurrence. Additionally, the increased cost of certain scrap material based on
the requirements of the scrap  selection and inspection program was also estimated.

       Labor rates and associated costs were based on Bureau of Labor  Statistics (BLS) data.
Technical, management, and clerical average hourly rates  for civilian workers were taken
from the March 2002 Employment Cost Trends (http://stats.bls.gov). Wages for civilian
workers (white-collar occupations) were used as the basis for the labor rates  with a total
compensation of $28.49/hr for technical, $42.20/hr for managerial, and $18.41/hr for clerical.
These rates represent salaries plus fringe benefits and do not include the cost of overhead.
An overhead rate of 110 percent was used to account for these costs. The fully burdened
wage rates used to represent respondent labor costs were technical at $59.83, management at
$88.62, and clerical at $38.66.

       The number of technical hours needed for each compliance activity was first
estimated. For each technical hour needed, 0.05 managerial hours and 0.10 clerical hours
were also assumed to be required. Consequently, the total labor cost, including technical,
managerial,  and clerical labor, for a compliance activity would be $68.125 per technical hour
expended.3  The 2002 labor costs were multiplied by 0.8837 (139.8/158.2) to estimate the
compliance  costs in fourth-quarter 1998 dollars. Therefore, the compliance costs were
calculated using $60.20 per technical hour expended.
3 As stated, these labor rates are based on March 2002 statistics. According to BLS data, the Employment Cost
Index (ECI) for civilian workers in March 2002 was 158.2, while the ECI for civilian workers in December
1998 was 139.8.

                                         3-19

-------
3.6.3.1 Performance Tests

       A total of 70 technical hours was estimated for each performance test. This included
time to prepare a site-specific QA test plan, conduct the performance test, and prepare the
final source test report. The total number of stacks requiring performance testing was
estimated to be 377.

       The performance tests are required once every 5 years. The compliance period is 3
years. Therefore, the required performance tests can be fairly evenly distributed across the
compliance period.  The average annual compliance cost associated with conducting
performance tests at a given foundry was calculated as the total costs for conducting all
required performance tests at that foundry and distributing those costs evenly over 5 years.
Thus, 5,278 technical hrs/yr were estimated for conducting performance tests, for a total
annual cost of $318,000/yr.

3.6.3.2 Scrap Selection and Inspection

       Most of the 98 foundries had a scrap selection and inspection program; however, it is
anticipated that many foundries will have increased the number of technical hours spent on
scrap selection and inspection to comply with requirements in the MACT standard. It was
assumed that the scrap inspection requirements would increase a typical foundry's inspection
process by 0.5 hr/day or 175 hrs/yr (assuming 350 operating days/yr). A one-time scrap
selection plan must be prepared and communicated within the foundry.  Because most
foundries do not operate 350 days per year, this activity was assumed to be included in the
175 technical hrs/yr per foundry labor estimate. Of the 98 facilities considered to be major
sources of HAP emissions, two of these foundries each operate two adjacent plants.
Although they are considered a single facility under the Clean Air Act, these adjacent plants
have separate scrap-receiving areas that would require inspection.  Thus, the scrap selection
and inspection costs were estimated based on 100 scrap-receiving areas.  The total
nationwide cost for the scrap selection and inspection program was estimated to be $1.05
million/yr.

       The scrap selection and inspection program also includes restrictions on the use of
scrap metal known to contain mercury switches or lead components, which is primarily
contained in automotive body scrap. The MACT standard will require foundries to purchase
only automotive body scrap that has had the mercury switches and lead components
removed. It is estimated that this will increase the cost of automotive body scrap by
$1.60/ton of scrap. It is estimated that automotive scrap is approximately 10 percent of the
total nationwide scrap supply. Based on the estimated production capacity of major source
iron and steel foundries of 16.1 million tons/yr, the scrap selection and inspection program is
estimated to cost foundries an additional $2.58 million/yr in increased scrap costs. Together

                                         3-20

-------
with the inspections, the total cost of the scrap selection and inspection program is estimated
to be $3.63 million/yr.

3.6.3.3 Start-up, Shutdown, and Malfunction Plan
       Each foundry is required to develop a start-up, shutdown, and malfunction (SSM)
plan. Eighty technical hours were estimated for preparing the SSM plan. This plan is
predominately a one-time requirement, but it is likely that the plan will be reviewed and
updated once per Title V permit period (i.e., once every 5 years). Assuming re-evaluations
are performed once every 5 years, the average annual technical hours required to complete
these evaluations would be 16 hrs/yr per foundry. Based on 98 major source iron and steel
foundries, the total nationwide annualized cost for preparing  the SSM plan was estimated to
be $94,400/yr.

3.6.3.4 Operating and Maintenance Plan

       Each foundry is required to develop an operating and maintenance (O&M) plan.  This
plan includes an assessment of ignitability for mold vents and operating and maintenance
requirements for capture systems and emission control devices.  One hundred and twenty
technical hours were estimated for preparing the O&M plan and performing the mold vent
ignitability determination. Again, the development of the O&M plan is predominately a one-
time requirement, but it is likely that the plan will be reviewed and updated once per Title V
permit period (i.e., once every 5 years). Therefore, the annualized costs for the preparation
of the O&M plan were estimated based on 24 hrs/yr per foundry, resulting in a total
nationwide annualized cost for the 98 major source iron and steel foundries of $142,000/yr.

3.6.3.5 Miscellaneous Recordkeeping and Reporting Costs

       Each foundry is required to maintain certain records of its monitored parameters.
Costs for data logging instruments were included in the cost of the monitoring equipment.
Additionally, the scrap inspection cost estimate included the effort to complete an inspection
record. Costs associated with maintaining these types of records at the foundry level  were
estimated based on two filing cabinets (for capital equipment costs of $400, which were
annualized over 10 years) and a pack of 10 writeable CDs (for annual operating costs of
$32). In addition, each foundry is required to maintain records of the annual HAP usage
rates associated with the foundry's coating materials and chemical binder formulations.
Compiling and filing these records were estimated to  require 8 hrs/yr per foundry or
$47,000/yr nationwide.

       Each foundry is also required to submit a semiannual report to document compliance
with the MACT standard.  Each report was estimated to require 20 technical hours to
prepare, so that the annual labor estimate for this requirement is  40 hrs/yr per foundry.

                                        3-21

-------
These reports are required for 98 major source iron and steel foundries so that the total
nationwide costs for reporting were estimated to be $236,000/yr.

3.7    Total Nationwide Costs

       The total nationwide costs for each of the major control or monitoring systems are
provided in Table 3-4. Table 3-4 also summarizes the estimated recordkeeping and reporting
costs. The total annual nationwide cost of the MACT standard for iron and steel foundries is
projected to be $21.0 million/yr.

        Table 3-4.  Nationwide Cost Estimates for Iron Foundry MACT: 1998$

Source
Baghouse replacement of cupola venturi
scrubbers
Cupola afterburners
Baghouses on EIF and scrap preheaters
Baghouses on pouring stations
Acid/wet scrubber systems for TEA control
Total Emission Control Costs
Bag leak detection systems
Venturi scrubber monitoring systems
Acid/wet scrubber parameter monitoring systems
Performance tests
Scrap selection and inspection
Start-up, shutdown, and malfunction plan
Operating and maintenance plan
Other recordkeeping and reporting costs
Total Monitoring, Recordkeeping and
Reporting Costs
Total Engineering Control Costs
Total
Capital
($1,000)
175,217
555
5,721
2,882
949
185,324
2,592
0
345
0
0
0
0
39
2,976
188,300

Annual
Capital3
16,540
79
540
272
135
17,567
369
0
49
0
0
0
0
6
176
17,743
Annual
Operating
($l,000/yr)
(6,134)
228
2,126
1,715
183
(1,882)
144
72
184
318
3,630
94
142
286
4,870
2,988
Total
Annual
($l,000/yr)
10,406
307
2,666
1,987
318
15,685
513
72
233
318
3,630
94
142
292
5,294
20,979
  Reflects capital recovery based on a 20-year life and 7 percent interest rate for baghouse emission controls
  and a 10-year life and 7 percent interest rate for cupola afterburners, TEA scrubbers, and all monitoring
  equipment.
                                          3-22

-------
                                    SECTION 4

                         ECONOMIC IMPACT ANALYSIS
       The rule to control the release of HAPs from iron and steel foundry operations will
directly (through imposition of compliance costs) or indirectly (through changes in market
prices) affect the entire U.S. iron and steel industry. The response by these producers to
these additional costs will determine the economic impacts of the regulation.  Specifically,
the impacts will be distributed across producers and consumers of iron castings and foundry
coke through changes in prices and quantities in the affected markets. This section presents
estimates of the economic impacts of the iron and steel foundry MACT using an economic
model that captures the linkages between the iron castings and foundry coke markets.

       This section describes the  data and approach used to estimate the economic impacts
of this rule for the baseline year of 2000.  Section 4.1 presents the inputs for the economic
analysis, including characterization of producers, markets, and the costs of compliance.
Section 4.2 summarizes the conceptual approach to estimating the economic impacts on the
affected industries. A fully detailed description of the economic impact methodology is
provided in Appendix A. Lastly,  Section 4.3 provides the results of the economic impact
analysis.

4.1    Market Characterization

       EPA estimated changes in the equilibrium price and quantity due to control costs on
iron and steel foundries using the  following three linked markets:

       •   market for iron castings,
       •   market for steel castings, and
       •   market for foundry coke.
EPA collected data for the castings market using the Current Industrial Reports series (U.S.
Bureau of the Census, 2001). Data on the foundry market were obtained from the economic
impact analysis of the final Coke Ovens NESHAP (EPA, 2002). Table 4-1 reports the
baseline data used in the market model.
                                         4-1

-------
Table 4-1.  Baseline Market Data Set:  2000
                        Market                                         Value
 Iron Castings
     Market price ($/short ton)                                           $ 1,029
     Market output (103 tpy)                                             11,345.7
         Domestic                                                     10,507.0
            Affected                                                   7,985.3
            Unaffected                                                 2,521.7
         Imports                                                         838.7
 Steel Castings
     Market price ($/short ton)                                           $3,761
     Market output (103 tpy)                                              1,144.1
         Domestic                                                      1,040.0
            Affected                                                     436.8
            Unaffected                                                   603.2
         Imports                                                         104.1
 Foundry Coke
     Market price ($/short ton)                                             $ 161
     Market output (103 tpy)                                              1,385
         Domestic                                                      1,238
         Imports                                                         147
4.1.1  Regulatory Control Costs
       As shown in Section 3, the Agency developed compliance cost estimates for model
plants that can be mapped to each supply segment affected by the rule. These estimates
reflect the "most reasonable" scenario for this industry. These cost estimates serve as inputs
to the economic analysis and affect the supply decisions of casting producers (see
Figure 4-1). The  total annual nationwide cost of the MACT for iron and steel foundries is
projected to be $21.23 million as expressed in 2000 dollars.1
'The baseline year of the economic analysis is 2000.  Therefore, engineering compliance cost estimates (as
   presented in Section 3) were adjusted using the Chemical Engineering plant cost index (394.1/389.5 =
   1.012).

                                             4-2

-------
               Imports

               Exports  •<
                              Consumers of
                              Iron and Steel
                                Castings
                                Market for
                              Foundry Coke
Figure 4-1.  Market Linkages Modeled in the Economic Impact Analysis
       EPA compared each individual plant's total annual compliance costs with an estimate
of baseline plant-level revenue (cost-to-sales ratio [CSR]).  The results (see Figure 4-2) show
the following:

       •   No affected plant is projected to have a CSR exceeding 2 percent.

       •   Over 90 percent of the affected plants have CSRs below 0.5 percent.
                                         4-3

-------
    100
     90
     80
 ^  70
equency
     60
     40
     30
     20
     10
             0%
>0-0.5%     >0.5-1%      >1-2%       2-4%
          Costs-to-Sales Ratio Range
Figure 4-2. Distribution of Plant-level Compliance Costs-to-Sales Ratios: 2000
4.2    EIA Methodology Summary

       In general, the EIA methodology needs to allow EPA to consider the effect of the
different regulatory alternatives.  Several types of economic impact modeling approaches
have been developed to support regulatory development. These approaches can be viewed as
varying along two modeling dimensions:

       •   the scope of economic decision making accounted for in the model, and
       •   the scope of interaction between different segments of the economy.
Each of these dimensions was considered in selecting the approach used to model the
economic impact of the coke regulation.

       To conduct the analysis for the iron and steel foundry MACT, the Agency used a
market modeling approach that incorporates behavioral responses in a multiple-market
partial equilibrium model. Multiple-market partial equilibrium analysis provides a
                                        4-4

-------
manageable approach to incorporate interactions between the foundry coke and iron castings
markets into the EIA to better estimate the regulation's impact. The multiple-market partial
equilibrium approach represents an intermediate step between a simple, single-market partial
equilibrium approach and a full general equilibrium approach.  The modeling technique is to
link a series of standard partial equilibrium models by specifying the interactions between
the supply and demand for products and then solving for changes in prices and quantities
across all markets simultaneously.  The EIA methodology is fully detailed in Appendix A.

       The Agency's methodology is soundly based on standard microeconomic theory
relying heavily on previous economic analyses, employs a comparative static approach, and
assumes certainty in relevant markets.  For this analysis, prices and quantities are determined
in perfectly competitive markets for iron castings and foundry coke. The competitive model
of price formation, as shown in Figure 4-3 (a), posits that market prices and quantities are
determined by the intersection of market supply and demand curves. Under the baseline
scenario, a market price and quantity (P, Q) are determined by the downward-sloping market
demand curve (DM) and the up ward-sloping market supply curve  (SM) that reflects the
horizontal summation of the individual supply curves of directly affected and indirectly
affected facilities that produce a given product.

       With the regulation, the cost of production increases for directly affected producers.
The imposition of the compliance costs is represented as an upward shift in the supply curve
for each affected facility from Sa to Sa'. As a result, the market supply curve shifts upward to
SM, as shown in Figure 4-3(b), reflecting the increased costs of production at these facilities.
In the baseline scenario without the standards, the industry would produce total output, Q, at
the price, P, with affected facilities producing the amount qa and unaffected facilities
accounting for Q minus qa, or qu. At the new equilibrium with the regulation, the market
price increases from P to P' and market output (as determined from the market demand
curve, DM) declines from Q to Q'. This reduction in market output  is the net result from
reductions at affected facilities and increases at unaffected facilities.

4.3    Economic Impact Results

       Based on the simple analytics presented above, when faced with higher costs of
production, producers will attempt to mitigate the impacts by making adjustments to shift as
much of the burden on other economic agents as market conditions allow. We would expect
upward pressure on prices as producers reduce output rates in response to higher costs.
Higher prices reduce quantity demanded and output for each market product, leading to
changes in profitability of foundries, batteries, and firms. These market and industry

                                         4-5

-------
                             +  p
                   =  p
           Affected Facilities
Unaffected Facilities
                                a) Baseline Equilibrium
Market
      P'
      P
              S'
                       p'
                       p
           Affected Facilities
Unaffected Facilities
                            b) With-Regulation Equilibrium
                                                                              DM
                                                                           Q
   Q' Q
Market
Figure 4-3.  Market Equilibrium without and with Regulation

adjustments determine the social costs of the regulation and the distribution of costs across
stakeholders (producers and consumers).

       To estimate these impacts, the economic modeling approach described in Appendix A
was operationalized in a multiple spreadsheet model.  This model characterizes those
producers and consumers identified in Figure 4-1 and their behavioral responses to the
imposition of the regulatory compliance costs.  These costs are expressed per ton of casting
                                         4-6

-------
product and serve as the input to the economic model, or "cost-shifters" of the baseline
supply curves at affected facilities.

       In addition to the "cost-shifters" the other major factors that influence behavior
adjustments in the model are the supply and demand elasticities of producers and consumers.
Table 4-2 presents the key elasticity parameters used in the model.  Specific functional forms
are presented in Appendix A.

Table 4-2. Supply and Demand Elasticities Used in Analysis
Market
Iron Castings
Domestic
Foreign
Steel Castings
Domestic
Foreign
Foundry Coke
Domestic
Foreign
Supply Elasticity

1.0a
1.0a

1.0a
1.0a

i.r
3.0d
Demand Elasticity

-0.58b
-1.0a

-0.59b
-1.0a

Derived demand
-0.3d
a Assumed value.
b Weighted average of product demand elasticities estimated in econometric analysis (see Appendix C).
c Estimate based on individual battery production costs and output.
d Graham, Paul, Sally Thorpe, and Lindsay Hogan.  1999. "Non-competitive Market Behavior in the
  International Coking Coal Market." Energy Economics 21:195-212.
       Given these costs and supply and demand elasticities, the model determines a new
equilibrium solution in a comparative static approach.  The following sections provide the
Agency's estimates of the resulting economic impacts for the rule.

4.3.1  Market-Level Impacts

       The increased cost of production due to the regulation is expected to slightly increase
the price of castings products and reduce their production and consumption from 2000
baseline levels.  As shown in Table 4-3, the regulation is projected to increase the price of
iron casting by less than 0.2 percent, or $1.12 per short ton.  Similarly, we project the price
of steel casting will increase by less than 0.1 percent, or $0.45 per short ton. The regulation
results in output declines of less than 0.1 percent in both castings markets.  Iron castings
output declines by 7,500 short tons per year, and steel castings decline by less than 500 short
                                           4-7

-------
Table 4-3. Market-Level Impacts of the Iron and Steel Foundries MACT: 2000
Market
Iron Castings
Market price ($/short ton)
Market output (103tpy)
Domestic
Affected
Unaffected
Imports
Steel Castings
Market price ($/short ton)
Market output (103tpy)
Domestic
Affected
Unaffected
Imports
Foundry Coke
Market price ($/short ton)
Market output (103tpy)
Domestic
Imports
Baseline

$1,029
11,345.7
10,507.0
7,985.3
2,521.7
838.7

$3,761
1,144.1
1,040.0
436.8
603.2
104.1

$161
1,385
1,238
147
Changes
Absolute

$1.12
-7.5
-8.4
-11.1
2.7
0.9

$0.45
-0.1
-0.1
-0.2
0.1
0.0

$0.00
<0.01
<0.01
0
from Baseline
Percent

0.109%
-0.066%
-0.080%
-0.140%
0.109%
0.109%

0.012%
-0.007%
-0.009%
-0.038%
0.012%
0.012%

0.00%
-0.01%
-0.01%
0.00%
tons per year. Market output declines are the net effect of directly affected castings output
declines across affected producers (11,200 tpy) and small increases in supply (3,700 tpy)
from unaffected domestic and foreign producers not subject to the regulation.

       Given that the change in casting output is so small and the change in demand for
foundry coke is also very small (less than 10,000 tpy), the entire market impact can be
absorbed by a single foundry battery that is assumed to have a constant marginal cost.  As a
result, the market price is unchanged.2 This in turn leads to no change in the level of imports
(or exports) of foundry coke.
2See Appendix B for a detailed description of the step-wise supply function for the foundry coke market.

                                         4-8

-------
4.3.2  Industry-Level Impacts

       Industry profitability changes as prices and production levels adjust to increased
production costs. As shown in Table 4-4, the economic model projects that profits for
affected foundries will decrease by $12.1. In addition, the Agency projects no change in
profits for furnace coke plants because the small reduction in output comes from the
marginal coke battery,  which by assumption has zero profit in baseline.  Those domestic
suppliers not subject to the regulation experience small gains ($3.1 million); profits for
foreign foundries increase by $0.9 million.

Table 4-4. National-Level Industry Impacts of the Iron and Steel Foundries MACT:
2000

                        Market                                      Value
 Iron Castings
     Change in operating profit ($106/yr)                                   -$7.7
        Domestic                                                     -$8.6
            Affected                                                 -$11.5
            Unaffected                                                 $2.8
        Imports                                                       $0.9
 Steel Castings
     Change in operating profit (106/yr)                                    -$0.3
        Domestic                                                     -$0.4
            Affected                                                  -$0.6
            Unaffected                                                 $0.3
        Imports                                                       $0.0
 Foundry Coke
     Change in operating profit (106/yr)                                     $0
        Domestic                                                      $0
        Imports                                                       $0
4.3.3  Social Cost

       The social impact of a regulatory action is traditionally measured by the change in
economic welfare that it generates.  The social costs of the rule will be distributed across
consumers and producers alike.  Consumers experience welfare impacts due to changes in
market prices and consumption levels associated with the rule. Producers experience welfare
                                          4-9

-------
impacts resulting from changes in profits corresponding with the changes in production
levels and market prices. However, it is important to emphasize that this measure does not
include benefits that occur outside the market, that is, the value of reduced levels of air
pollution with the regulation.

       The national compliance cost estimates are often used as an approximation of the
social cost of the rule. The engineering analysis estimated annual costs of $21.23 million. In
this case, the burden of the regulation falls solely on the affected facilities that experience a
profit loss exactly equal to these cost estimates. Thus, the entire loss is a change in producer
surplus with no change (by assumption) in consumer surplus. This is typically referred to as
a "full-cost absorption" scenario in which all factors of production are assumed to be fixed
and firms are unable to adjust their output levels when faced with additional costs.

       In contrast, the economic analysis accounts for behavioral responses by producers
and consumers to the regulation (i.e., shifting costs to other economic agents). This
approach may result in a social cost estimate that differs from the engineering estimate and
also provides insights on how the regulatory burden is distributed across stakeholders. As
shown in Table 4-5, the  economic model estimates the total social cost of the  rule to be
$21.22 million. Although society reallocates resources as a result of the increased cost of
iron castings production, the social cost estimate is only slightly smaller than the engineering
cost estimate (approximately $10,000).

       In the castings markets, higher market prices lead to consumer losses of $13.2
million. A significant share  of these losses occurs in the iron castings market (90 percent, or
$11.9 million). Although foundries are able to pass on a limited amount of cost increases to
their final consumers (e.g., automotive manufacturers, appliance and hardware producers,
and the construction industry), the increased costs result in a net decline in producer profits
of $8.0 million. The model projects affected foundries experience $12.1 million in losses,
while unaffected domestic and foreign producers experience gains ($3.1 million and $0.9
million). These producers benefit from higher  market prices  for castings without additional
control costs.  In the coke industry, foundry coke profits at merchant plants  are projected to
remain unchanged, because reductions in output come from the marginal merchant furnace
coke battery.
                                         4-10

-------
Table 4-5. Distribution of the Social Costs of the Iron and Steel Foundry MACT: 2000

                        Market                                          Value
 Change in Consumer Surplus ($106/yr)                                     -$13.2
     Iron castings consumers
         Domestic                                                        -$11.9
         Foreign                                                           -$0.8
     Steel castings consumers
         Domestic                                                         -$0.5
         Foreign                                                           -$0.0
 Change in Producer Surplus ($106/yr)                                       -$8.0
     Iron casting producers
         Domestic                                                         -$8.6
             Affected                                                     -$11.5
             Unaffected                                                     $2.8
         Foreign                                                            $0.9
     Steel casting producers
         Domestic                                                         -$0.4
             Affected                                                      -$0.6
             Unaffected                                                     $0.3
         Foreign                                                            $0.0
 Change in Total Surplus ($106/yr)                                         -$21.22a

a  The negative change in total surplus indicates the social cost of the regulation is $21.22 million.
                                             4-11

-------
                                    SECTION 5

                           SMALL BUSINESS IMPACTS
       This regulatory action will potentially affect the economic welfare of owners of iron and
steel foundries.  These individuals may be owners/operators who directly conduct the business of
the firm or, more commonly, investors or stockholders who employ others to conduct the business
of the firm on their behalf through privately held or publicly traded corporations. The legal and
financial responsibility for compliance with a regulatory action ultimately rests with plant managers,
but the owners must bear the financial consequences of the decisions. Although environmental
regulations can affect all businesses, small businesses may have special problems complying with
such regulations.

       The Regulatory Flexibility Act (RFA) of 1980 requires that special consideration be given
to small entities affected by federal regulations. The RFA was amended in 1996 by the Small
Business Regulatory Enforcement Fairness Act (SBREFA) to strengthen its analytical and
procedural requirements.  Under SBREFA, the Agency must perform a regulatory flexibility
analysis for rules that will have a significant impact on a substantial number of small entities.

       This section focuses on the compliance burden of small businesses within the iron and steel
foundry industry and provides a screening analysis to determine whether this rule is likely to impose
a significant impact on a substantial number of the small entities (SISNOSE) within this  industry.
The screening analysis employed here is a "sales test" that computes the annualized compliance
costs as a share of sales for each company. In addition, it provides information about the impacts
on small businesses after accounting for producer responses to the rule and the resulting changes in
market prices and output.

5.1    Identifying Small Businesses

       For purposes of assessing the impacts of the rule on small entities, a small entity is defined
as (1) a small business according to SBA size standards for NAICS code 331511 (Iron
Foundries), 331512 (Steel Investment Foundries), or 331513 (Steel Foundries, Except Investment)
                                        5-1

-------
of 500 or fewer employees; (2) a small governmental jurisdiction that is a government of a city,
county, town, school district, or special district with a population of less than 50,000; and (3) a
small organization that is any not-for-profit enterprise that is independently owned and operated
and is not dominant in its field.  Based on these SBA size definitions for the affected industries and
reported sales and employment data, the Agency has identified 20 small businesses directly affected
by the rule.1

5.2    Screening-Level Analysis

       To assess the potential impact of this rule on small businesses, the Agency calculated the
share of annual compliance costs relative to baseline sales for each company.  When a company
owns more than one affected facility, EPA combined the costs for each facility for the numerator of
the test ratio. Annual compliance costs include annualized capital costs and operating and
maintenance costs imposed on these companies. They do not include changes in production or
market adjustments.

       Small businesses within the source category are expected to incur approximately
15 percent of the total industry compliance cost of $21.23 million (see Table 5-1). The average
total annual compliance cost is projected to be $163,000 for small companies, while the average
for large companies is projected to be approximately $418,000 per company.  The average CSR
for small firms is 0.40 percent.  One small company is projected to have a CSR between 1 and 3
percent. No small firms are projected to have CSRs greater than 3  percent.

5.3    Assessment

       The Agency's analysis indicates no significant impacts on small firms' ability to continue
operations and remain profitable. The screening analysis shows all small firms' cost-to-sales ratios
are below the average return to sales for all reporting companies reported by statistical publications.
The Quarterly Financial Report (QFR) from the U.S.  Bureau of the Census reports that the
average return on sales  for the iron and steel industry ranged from 3.2 to 4.6 percent (U.S. Bureau
of the Census, 1998).2  In addition, Dun & Bradstreet reports
!EPA updated sales and employment data for affected companies reported in Section 2 to reflect more recent
   financial data (2002).

furthermore, the QFR reports that companies within the iron and steel industry with less than $25 million in
   assets reported an average return to sales ranging from 6.8 to 9.8 percent.

                                         5-2

-------
Table 5-1. Summary Statistics for SBREFA Screening Analysis:  2000

Total Number of Affected Companies
Total Annual Compliance Costs (TACC) (S103/yr)
Average TACC per company ($103/yr)
Compliance Cost-to-Sales Ratios
Average
Median
Maximum
Minimum

Compliance costs are < 1 % of sales
Compliance costs are > 1 to 3% of sales
Compliance costs are >3% of sales
Small
20
$3,269
S163

0.40%
0.26%
1.04%
0.04%
Number Share
19 95%
1 5%
0 0%
Large
43
$17,961
$418

0.13%
0.07%
1.92%
0.00%
Number Share
42 98%
1 2%
0 0%
All Companies
63
$21,230
$337

0.22%
0.09%
1.92%
0.00%
Number
61
2
0








Share
97%
3%
0%
Note: Assumes no market responses (i.e., price and output adjustments) by regulated entities.

-------
the median return on sales as 3.4 percent for SIC 3321—Gray and Ductile Foundries; 4.3 percent
for SIC 3324—Steel Investment Foundries; and 4.8 percent for SIC 3325—Steel Foundries,
NEC (Dun & Bradstreet, 1997).
                                        5-4

-------
                                  REFERENCES
American Foundrymen's Society (AFS).  1998.  1999 AFS Metalcasting Forecast and
       Trends. Des Plaines, IL: American Foundrymen's Society.

Bingham, T.H., and TJ. Fox. 1999.  "Model Complexity and Scope for Policy Analysis."
       Public Administration Quarterly 23(3).

Business Trend Analysts (BTA).  1996.  "Foundry Products and Markets in the
       U.S.—Company Profiles and Ferrous Castings."

Ductile Iron Society. "Ductile Iron Data For Design Engineers." .
       As obtained on September 24, 1998.

Dun & Bradstreet. 1997. Industry Norms & Key Business Ratios: Desk-Top Edition 1996-
       97. Murray Hill, NJ: Dun & Bradstreet.

Engineering and Mining Journal.  1997.  "A New U.S. Trend? No Leftovers—No
       Problems."  198(10):WW15.
Foti, Ross, Michael Lessiter, and Alfred Spada, eds.  1998.  "Mechanical Properties of Gray
       Iron." Casting Source Directory 1998-1999.

Graham, Paul, Sally Thorpe, and Lindsay Hogan. 1999. "Non-competitive Market
       Behaviour in the International Coking Coal Market." Energy Economics 21:195-
       212.

Heil, Scott and Terrance W. Peck, eds. 1998.  Encyclopedia of American Industries. New
       York:  Gale.

Ho, M., and D. Jorgenson. 1998. "Modeling Trade Policies and U.S. Growth: Some
       Methodological Issues." Presented at USITC Conference on Evaluating APEC Trade
       Liberalization: Tariff and Nontariff Barriers. September 11-12,  1997.

Hogan, William T., and Frank T. Koelble. 1996. "Steel's Coke Deficit: 5.6 Million Tons
       and Growing." New Steel  12(12):50-59.

Lankford, William T., Norman L. Samways, Robert F. Craven, and Harold E. McGannon,
       eds. 1985. The Making, Shaping and Treating of Steel. Pittsburgh: United States
       Steel, Herbick & Held.

LaRue, James P.  1989.  Basic Metalcasting. Des Plaines, IL: American Foundrymen's
       Society.
                                       R-l

-------
Schleg, Fred, and David P. Kanicki.  1998.  "Guide to Selecting Casting and Molding
      Processes."  Casting Source Directory 1998-1999 8th Edition.  Des Plaines, IL:
      American Foundrymen's Society.

Steel Founders' Society of America (SFSA). "Glossary of Foundry Terms."
      . Accessed September 3, 1998.

Tardiff, Joseph C., ed.  1998.  U.S. Industry Profiles: The Leading 100.  New York:  Gale.

U.S. Bureau of the Census. 1998. Quarterly Financial Report for Manufacturing, Mining,
      and Trade Corporations.  First Quarter, Series QFR 98-1. Washington, DC:
      Government Printing Office.

U.S. Bureau of the Census. 2001. Iron and Steel Castings. MA331A(00)-1. Washington,
      DC: U.S. Department of Commerce, U.S. Census Bureau.
U.S. Department of Commerce.  1995. 7992 Census of Manufactures: Industry
      Series—Blast Furnaces, Steel Works, and Rolling and Finishing Mills Industry.
      Washington, DC: U.S. Government Printing Office.

U.S. Department of Commerce.  February 1998.  1996 Annual Survey of Manufactures :
      Statistics for Industry Groups and Industries. M96(AS)-1. Washington, DC: U.S.
      Government Printing Office.

U.S. Department of Commerce, Bureau of the Census.  1988-1997. Current Industrial
      Reports. Washington, DC: U.S. Government Printing Office.

U.S. Department of Labor, Bureau of Labor Statistics.  BLS LABSTAT Database:
      Employment and Earnings, SIC 33.  .  Obtained in March 2002.
U.S. Department of Energy. 1996. "Trends Effecting [sic] R&D in the Metalcasting
      Industry." Prepared by BCS Incorporated for Office of Industrial Technologies,
      Washington, DC.

U.S. Environmental Protection Agency.  1979. Technical Approach for a Coke Production
      Cost Model. Prepared by Research Triangle Institute, Research Triangle Park, N.C.

U.S. Environmental Protection Agency.  1988. Benzene Emissions form Coke By-Product
      Recovery Plants-BackGround Information for Proposed Standards. Office of Air
      Quality Planning and Standards, Research Triangle Park, NC.

U.S. Environmental Protection Agency.  199 la. Controlling Emissions from By-Product
      Coke Oven Charging, Door Leaks, and Topside Leaks:  An Economic Impacts
      Analysis.  Prepared by Research Triangle Institute, Research Triangle Park, NC.

U.S. Environmental Protection Agency (EPA). 1991b. Handbook: Control Technologies
      for Hazardous Air Pollutants (EPA/625/6-91/014).  Research Triangle Park, NC:
      U.S. Environmental Protection Agency.
                                       R-2

-------
U.S. Environmental Protection Agency (EPA).  1991c.  OAQPS Control Cost Manual
       (EPA/453/B-96-01). Washington, DC:  U.S. Environmental Protection Agency.

U.S. Environmental Protection Agency (EPA).  1998a.  EPA Office of Compliance Sector
       Notebook Project: Profile of the Metal Casting Industry. Washington, DC.

U.S. Environmental Protection Agency (EPA).  1998b.  Foundry Industry Responses to
       Information Collection Request (ICR) Survey. Database prepared for EPA's Office of
       Air Quality Planning and Standards. Research Triangle Park, NC.

U.S. Environmental Protection Agency (EPA).  1999.  OAQPS Economic Analysis Resource
       Document. Durham, NC: Innovative Strategies and Economics Group.

U.S. Environmental Protection Agency (EPA).  2002. Economic Impact Analysis of Final
       Coke Ovens NESHAP. Research Triangle Park, NC: U.S. Environmental Protection
       Agency.

U.S. International Trade Commission (USITC).  1994.  Metallurgical Coke: Baseline
       Analysis of the U.S. Industry and Imports. Publication No. 2745. Washington, DC:
       U.S. International Trade Commission.

U.S. International Trade Commission (USITC).  July 2000. "Foundry Coke:  A Review of
       the Industries in the United States and China."
       .

U.S. International Trade Commission (USITC).  2001c. "Blast Furnace Coke from China
       and Japan." Investigations Nos. 73l-TA-951-952 (Preliminary) Publication 3444;
       August 2001. .
                                       R-3

-------
                                   APPENDIX A

               ECONOMIC IMPACT ANALYSIS METHODOLOGY
       This appendix provides the methodology for analyzing the economic impacts of the
coke ovens, integrated iron and steel, and iron and steel foundries MACT standards to ensure
consistency across the EIAs. Implementation of this methodology provided the economic
data and supporting information that EPA requires to support its regulatory determination.
This approach is firmly rooted in microeconomic theory and the methods developed for
earlier EPA studies to operationalize this theory. The Agency employed a computerized
market model of the coke, steel mill products, and iron and steel castings industries to
estimate the behavioral responses to the imposition of regulatory costs and, thus, the
economic impacts of the standard. The market model captures the linkages between these
industries through changes in equilibrium prices and quantities.

       This methodology section describes the conceptual approach selected for this EIA.
For each product market included in the analysis, EPA derived facility-level supply functions
and demand functions that are able to account for the behavioral responses of producers and
consumers and market implications of the regulatory costs. Finally, this appendix presents
an overview of the specific functional forms that constitute the Agency's computerized
market model.

A.I    Overview of Economic Modeling Approach

       In general, the EIA methodology needs to allow EPA to consider the effect of the
different regulatory alternatives.  Several types of economic impact modeling approaches
have been developed to support regulatory development. These approaches can be viewed as
varying along two modeling dimensions:

       •   the scope of economic decision making accounted for in the model, and
       •   the scope of interaction between different segments of the economy.
Each of these dimensions was considered in selecting the approach used to model the
economic impact of the regulation. Bingham and Fox (1999) provide a useful summary of
these dimensions as they relate to modeling the outcomes of environmental regulations.
                                        A-l

-------
       For this analysis, prices and quantities are determined in perfectly competitive
markets for furnace coke, foundry coke, steel mill products, and iron castings. The Agency
analyzed the impact of the regulation using a market modeling approach that incorporates
behavioral responses in a multiple-market partial equilibrium model.  Multiple-market partial
equilibrium analysis accounts for the interactions between coke, steel mill product, and iron
and steel castings markets into the EIA to better estimate the regulation's impact. The
modeling technique is to link a series of standard partial equilibrium models by specifying
the interactions between the supply and demand for products and then solving for changes in
prices and quantities across all markets simultaneously.

       Figure A-l  summarizes the market interactions included in the Agency's EIA
modeling approach. Changes in the equilibrium  price and quantity due to control costs
associated with individual MACTs were  estimated simultaneously in four linked markets:

       •  market for furnace coke,
       •  market for foundry coke,
       •  market for steel mill products, and

       •  markets for iron and steel castings.

       As described in Section 2 of this EIA report, many captive coke plants supply their
excess furnace coke to the market. Merchant coke plants and foreign imports account for the
remaining supply to the furnace coke market. Furnace coke produced at captive coke  plants
and shipped directly to integrated iron and steel mills owned by their parent companies does
not directly enter the market for furnace coke.  However, compliance costs incurred by these
captive, or "in-house," furnace coke batteries indirectly affect the furnace coke market
through price and output changes in the steel mill products market.

       The market demand for furnace coke is derived from integrated mills producing steel
mill products. Integrated iron and steel mills that need more coke than their captive batteries
can produce will purchase furnace coke from the market.  Integrated mills' market demand
for furnace coke  depends on their production levels as influenced by the market for steel mill
products.  Steel mill products are supplied by three sources:  integrated iron and steel mills,
nonintegrated steel mills (primarily minimills), and imports.  Domestic consumers of steel
mill products and exports account for the market demand.
                                         A-2

-------
                         Consumers of Steel
                           Mill Products
                Imports
                Exports
Imports
Exports
'
Integrated Iron
and Steel Mills
t

• Finishing Mills '
• Steelmaking Furnace |^"-»
• Blast Furnace I
1
i
Captive Coke Plants f
1

                                            Imports
                                         Exports
[ Iron and Steel )<•
  Castings


Electric
Furnaces


Cupola Furnaces
Figure A-l. Market Linkages Modeled in the Economic Impact Analysis
       Domestic merchant plants are the primary suppliers of foundry coke to the market.
However, the U.S. International Trade Commission (2000) has documented an increasing
trend in foreign imports of foundry coke from China. Therefore, we have included a single
import supply curve to characterize this supply segment.

       In addition to furnace and foundry coke, merchant and captive coke plants sell a by-
product referred to as "other coke" that is purchased as a fuel input by cement plants,
chemical plants, and nonferrous smelters. Because "other coke" is a by-product and
represented only 2 percent of U.S. coke production in 1997, it is not formally characterized
by supply and demand in the market model. Revenues from this product are accounted for
by assuming its volume is a constant proportion of the total amount of coke produced by a
battery and sold at a constant price.
                                         A-3

-------
A.2    Conceptual Market Modeling Approach

       This section examines the impact of the regulations on the production costs for
affected facilities, both merchant and captive.  It provides an overview of the basic economic
theory of the effect of regulations on facility production decisions and the concomitant effect
on market outcomes. Following the OAQPS Economic Analysis Resource Document (EPA,
1999), we employed standard concepts in microeconomics to model the supply of affected
products and the impacts of the regulations on production costs  and the operating decisions.
The approach relies heavily on previous economic analyses, employs a comparative static
approach, and assumes certainty in relevant markets. The three  main elements of the
analysis are regulatory effects on the manufacturing facility, market responses, and
facility-market interactions.  The remainder of this section describes each of these main
elements.

A.2.1  Facility-level Responses to Control Costs

       Individual plant-level production decisions were modeled to  develop the market
supply and demand for key industry segments in the analysis. Production decisions were
modeled as intermediate-run decisions, assuming that the plant size, equipment, and
technologies are fixed.  For example, the production decision typically involves (1) whether
a firm with plant and equipment already in place purchases inputs to produce output and (2)
at what capacity utilization the plant should operate.  A profit-maximizing firm will operate
existing capital as long as the market price for its output exceeds its per-unit variable
production costs, since the facility will cover not only the cost of its  variable inputs but also
part of its capital costs. Thus, in the short run, a profit-maximizing firm will not pass up an
opportunity to recover even part of its fixed investment in plant  and equipment.

       The existence of fixed production factors gives rise to diminishing returns to those
fixed factors and, along with the terms under which variable inputs are purchased, defines
the up ward-sloping form of the marginal cost (supply) curve employed for this analysis.
Figure A-2 illustrates this derivation of the supply function at an individual mill based on the
classical U-shaped cost structure.  The MC curve is the marginal cost of production, which
intersects the facility's average variable (avoidable) cost curve (AVC) and its average total
cost curve (ATC) at their respective minimum points. The supply function is that portion of
the marginal cost curve bounded by the minimum economically feasible production rate (qm)
and the technical capacity (qM). A profit-maximizing producer will select the output rate
where marginal revenue equals price, that is, at [P*, q*]. If market price falls below ATC,
                                         A-4

-------
                  $/q
                                                                       q/t
Figure A-2. Product Supply Function at Facility

then the firm's best response is to cease production because total revenue does not cover total
costs of production.

       Now consider the effect of the regulation and the associated compliance costs. These
fall into one of two categories:  avoidable variable and avoidable nonvariable.  These final
costs are characterized as avoidable because a firm can choose to cease operation of the
facility and, thus, avoid incurring the costs of compliance. The variable control costs include
the operating and maintenance  costs of the controls, while the nonvariable costs include
compliance capital equipment.  Figure A-3 illustrates the effect of these additional costs on
the facility supply function. The facility's AVC and MC curves shift upward (to AVC' and
MC') by the per-unit variable compliance costs.  In addition, the nonvariable compliance
costs increase total avoidable costs and, thus, the vertical distance between ATC' and AVC'.
The facility's supply curve shifts upward with marginal costs, and the new (higher)
minimum  operating level (q) is determined by a new (higher) ps.

       Next consider the effect of compliance costs on the derived demand for inputs at the
regulated facility. Integrated iron and steel mills are market demanders of furnace coke,
while foundries with cupola furnaces are market demanders of foundry coke. We employ
similar neoclassical analysis to that above to demonstrate the effect of the regulation on the
                                         A-5

-------
           $/q
          p m'
          pm
                                   qm  qm
                                                 ,M
                                                                        q/t
Figure A-3. Effect of Compliance Costs on Product Supply Function at Facility
demand for market coke inputs, both furnace and foundry.  Figure A-4 illustrates the derived
demand curve for coke inputs. Each point on the derived demand curve equals the
willingness to pay for the corresponding marginal input.  This is typically referred to as the
input's value of marginal product (VMP), which is equal to the price of the output (P) less
the per-unit compliance cost (c) times the input's "marginal physical product" (MPP), which
is the incremental output attributable to the incremental inputs.  If, as assumed in this
analysis, the input-output relationship between the market coke input and the final product
(steel mill products or iron castings) is strictly fixed, then the VMP of the market coke is
constant and the derived demand curve is horizontal with the constant VMP as the vertical
intercept, as shown in Figure A-4.  Ignoring any effect on the output price for now, an
increase in regulatory costs will lower the VMP of all inputs leading  to a downward shift in
the derived demand in Figure A-4 from D to  Dv .
                                         A-6

-------
                       $/qv
                                                        D,
                                                       D
Figure A-4. Derived Demand Curve for Coke Inputs

A.2.2  Market Effects

       To evaluate the market impacts, the economic analysis assumes that prices and
quantities are determined in a competitive market (i.e., individual facilities have negligible
power over the market price and thus take the price as "given" by the market). As shown in
Figure A-5(a), under perfect competition, market prices and quantities are determined by the
intersection of market supply and demand curves.  The initial baseline scenario consists of a
market price and quantity (P, Q) that is determined by the downward-sloping market demand
curve (DM) and the upward-sloping market supply curve (SM) that reflects the horizontal
summation of the individual producers' supply curves.

       Now consider the effect of the regulation on the baseline scenario as shown in
Figure A-5(b). In the baseline scenario without the standards, at the projected price, P, the
industry would produce total output, Q, with affected facilities producing the amount qa and
unaffected facilities accounting for Q minus qa, or qu. The regulation raises the production
costs at affected facilities, causing their supply curves to shift upward from Sa to Sa' and the
market supply curve to shift upward to SM/.  At the new with-regulation equilibrium, the
market price increases from P to P' and market output (as determined from the market
demand curve, DM) declines from Q to Q'.  This reduction in market output is the  net result
from reductions at affected facilities and increases  at unaffected facilities.
                                         A-7

-------
           Affected Facilities
      P'
      P
              S'
           Affected Facilities
                             +  p
                    =  p
                                                                            Q
Unaffected Facilities
                                a) Baseline Equilibrium
Market
                       P'
                       P
Unaffected Facilities
                             b) With-Regulation Equilibrium
   Q' Q
Market
Figure A-5.  Market Equilibrium without and with Regulation
                                          A-8

-------
Unaffected facilities do not incur the increased costs due to regulation so their response to
higher product prices is to increase production. Foreign suppliers (i.e., imports), which also
do not face higher costs, will respond in the same manner as these unaffected producers.

       The above description is typical of the expected market effects for final product
markets. The regulations would potentially affect the costs of producing steel mill products
through additional control costs and increases in the market price of furnace coke and the
cost of producing captive furnace coke. The increase in control costs, the market price, and
captive production costs for furnace coke result in an upward shift in the supply functions of
integrated iron and steel mills, while nonintegrated  and foreign suppliers are unaffected.
Additionally, the regulations would potentially affect the costs of producing iron castings
through additional control costs and changes in the  market price of foundry coke. This
results in an upward shift in supply functions of foundries operating cupola furnaces, while
foundries operating  electric furnaces are only affected to the extent they are subject to
additional control costs.

       However, there are additional impacts on the furnace and foundry coke markets
related to their derived demand as inputs to either the production of steel mill products or
iron castings. Figure A-6 illustrates, under perfect competition, the baseline scenario where
the market quantity  and price of the final steel mill  product or iron and steel casting, Qx(Qx0,
Px0), are determined by the intersection of the market demand curve (Dx) and the market
supply curve (Sx), and the  market quantity and price of furnace or foundry coke, Qy(Qy0, Py0),
are determined by the intersection of the market demand curve (Dy) and market supply curve
(Sy). Given the derived demand for coke, the demanders of coke, Qy, are the individual
facilities that purchase coke for producing their final products (i.e., integrated steel mills in
the case of furnace coke or foundries with cupola furnaces in the case of foundry coke).

       Imposing the regulations increases the costs of producing coke and, thus, the final
product, shifting the market supply functions for both commodities  upward to Sx' and Sy',
respectively. The supply shift in the final product market causes the market quantity to fall
to Qxj and the market price to rise to Pxl in the new  equilibrium.  In  the market for coke, the
reduced production  of the  final product causes a downward shift in the demand curve (Dy)
with an unambiguous reduction in coke production, but the direction of the change in market
price is determined by the  relative magnitude of the demand and supply shift. If the
downward demand effect dominates, the price will  fall (e.g., Pyl); however, if the upward
supply effect dominates, the price will rise (e.g., Py2). Otherwise, if the effects just offset
each other, the price remains unchanged (e.g., Py3 = Py0).
                                         A-9

-------
            $/Qv
                                                    0x1  QxO


                (a) Market for single steel mill product or iron and steel casting, Q
cyt
            $/Qu
              y2

         P  - P
         ry3 ~ ryO
Qy1   Qy:
                                                         Q,,,
Q/t
                                   (b) Market for coke input, Q
Figure A-6.  Market Equilibria With and Without Compliance Costs
                                         A-10

-------
A.2.3  Facility-Level Responses to Compliance Costs and New Market Prices

       In evaluating the market effects, we must distinguish between the initial effect of the
regulations and the net effect after all markets have adjusted.  The profit-maximizing
behavior of firms, as described above, may lead to changes in output that, when aggregated
across all producers, lead to changes in the market-clearing price and feedback on the firms
to alter their decisions. These adjustments are characterized as a simultaneous interaction of
producers, consumers, and markets.  Thus, to evaluate the facility-market outcomes, the
analysis must go beyond the initial effect of the regulation and estimate the net effect after
markets have fully adjusted.

       Given changes in the market prices and costs, each facility will elect to either

       •  continue to operate, adjusting production and input use based on new revenues
          and costs, or
       •  cease production at the facility if total revenues do not exceed total costs.
This decision can be extended to those facilities with multiple product lines or operations
(e.g., coke batteries, blast furnaces, cupolas).  If product revenues are less than product-
specific costs, then these product lines or operations may be closed.

       Therefore, after accounting for the facility-market interaction, the operating decisions
at each individual facility can be derived.  These operating decisions include whether to
continue to operate the facility (i.e., closure) and, if so, the optimal production level based on
compliance costs and new market prices.  The approach to modeling the facility closure
decision is based on conventional microeconomic theory.  This approach compares the
ATC—which includes all cost components that fall to zero when production
discontinues—to the expected post-regulatory price.  Figure A-3 illustrates this comparison.
If price falls below the ATC, total revenue would be less than the total costs.  In this
situation,  the owner's cost-minimizing response is to close the facility.  Therefore, as long as
there is some return to the fixed factors of production—that is, some positive level of
profits—the firm is expected to continue to operate the facility.

       If the firm decides to continue operations, then the facility's decision turns to the
optimal output rate.  Facility and product-line closures, of course, directly translate  into
reductions in output. However, the output of facilities that continue to operate will also
change depending on the relative impact of compliance costs and higher market prices.
Increases  in costs will tend to reduce producers' output rates; however, some of this effect is
mitigated when prices are increased. If the market price increase more than offsets  the

                                         A-ll

-------
increase in unit costs, then even some affected facilities could respond by increasing their
production.  Similarly, supply from unaffected domestic producers and foreign sources will
respond positively to changes in market prices.

A.3    Operational Economic Model

       Implementation of the MACT standards will affect the costs of production for plants
across the United States subject to the rule. Responses at the facility level to these additional
costs will collectively determine the market impacts of the rule.  Specifically, the cost of the
regulation may induce some facilities to alter their current level of production or to cease
operations. These choices affect and, in turn are affected  by, the market price of each
product.  As described above, the Agency has employed standard microeconomic concepts to
model the supply and demand of each product and the impacts of the regulation on
production costs and the output decisions of facilities. The main elements of the analysis are
to

       •   characterize production of each product at the  individual supplier and market
          levels,
       •   characterize the demand for each product, and
       •   develop the solution algorithm to determine the new with-regulation equilibrium.

The following sections provide the supply and demand specifications for each product
market as implemented in the EIA model and summarize  the model's solution algorithm.
Supply and demand elasticities used in the model  are presented in Table A-l.

A.3.1  Furnace Coke Market

       The market for furnace coke consists of supply from domestic coke plants, both
merchant and captive,  and foreign imports and of  demand from integrated steel mills and
foreign exports.  The domestic supply for furnace  coke is  modeled as a step-wise supply
function developed from the marginal cost of production at individual furnace coke batteries.
The domestic demand is derived from iron and steel production at integrated mills as
determined through the market for steel mill products and coking rates for individual
batteries. The following section details the market supply and demand components for this
analysis.
                                        A-12

-------
Table A-l. Supply and Demand Elasticities Used in Analysis
Market
Furnace Coke
Domestic
Foreign
Foundry Coke
Domestic
Foreign
Steel Mill Products
Domestic
Foreign
Iron Castings
Domestic
Foreign
Steel Castings
Domestic
Foreign
Supply Elasticity

2 la
3.0b

i.r
3.0b

3.5C
1.5C

1.0f
1.0f

1.0f
1.0f
Demand Elasticity

Derived demand
-0.3b

Derived demand
-0.3b

-0.59d
-1.25e

-0.58d
-1.0f

-0.59d
-1.0f
  Estimate based on individual battery production costs and output.
  Graham, Paul, Sally Thorpe, and Lindsay Hogan. 1999. "Non-competitive Market Behaviour in the
  International Coking Coal Market." Energy Economics 21:195-212.
  U.S. International Trade Commission (USITC). 2001a. Memorandum to the Commission from Craig
  Thomsen, John Giamalua, John Benedetto, and Joshua Level, International Economists. Investigation
  No. TA-201-73:  STEEL—Remedy Memorandum.  November 21, 2001.
  Econometric analysis (see Appendix C for details).
  Ho, M., and D. Jorgenson. 1998. "Modeling Trade Policies and U.S. Growth: Some Methodological
  Issues."  Presented at USITC Conference on Evaluating APEC Trade Liberalization: Tariff and Nontariff
  Barriers. September 11-12, 1997.
  Assumed value.
A.3.1.1  Market Supply of Furnace Coke

        The market supply for furnace coke, QSc, is the sum of coke production from
merchant facilities, excess production from captive facilities (coke produced at captive
batteries less coke consumed for internal production on steel mill products), and foreign
imports, i.e.,

                              rvSc  _ nSc     Sc      Sc                               (\ 1\
                              V    ~ 4M     4i   +  4p                                \A-i)
                                            A-13

-------
where

        q Sc  = furnace coke supply from merchant plants,

        q^6  = furnace coke supply from integrated steel mills, and

        qpC  = furnace coke supply from foreign sources (imports).

       Supply from Merchant and Captive Coke Plants. The domestic supply of furnace
coke is composed of the supply from merchant and captive coke plants reflecting plant-level
production decisions for individual coke batteries.  For merchant coke plants the supply is
characterized as


                                 
-------
              =      supply of steel mill product from integrated mill (1).
The MAX function in Eq. (A.3) indicates that if the total captive production of furnace coke
at an integrated mill is greater than the amount of furnace coke consumption required to
produce steel mill products, then supply to the furnace coke market will equal the difference;
otherwise, the mill's supply to the furnace coke market will be zero (i.e., it only satisfies
internal requirements from its captive operations).

       As stated above, the domestic supply of furnace coke is developed from plant-level
production decisions for individual coke batteries. For an individual coke battery the
marginal cost was assumed to be constant. Thus, merchant batteries  supply 100 percent of a
battery's capacity to the market if the battery's marginal cost (MC) is below the market price
for furnace coke (pc), or zero if MC exceeds pc.  Captive batteries first supply the furnace
coke demanded by their internal steel-making requirements. Any excess capacity will then
supply the furnace  coke market if the remaining captive battery's MC is below the market
price.

       Marginal cost curves were developed for all furnace coke batteries at merchant and
captive plants in the United States as detailed in Appendix B. Production costs for a single
battery are characterized by constant marginal cost throughout the capacity range of the
battery. This yields the inverted L-shaped supply function shown in Figure  A-7(a).  In this
case, marginal cost (MC) equals average variable cost (AVC) and is constant up to the
production capacity given by q. The supply function becomes vertical at q because
increasing production beyond this point is not possible. The minimum economically
achievable price level is equal to p*.  Below this price level, p* is less than AVC, and the
supplier would choose to shut down rather than to continue to produce coke.

       A step-wise supply function can be created for each facility with multiple batteries by
ordering  production from least to highest MC batteries (see Figure A-7[b]).  For captive coke
plants, the lowest cost batteries are assumed to supply internal demand, leaving the higher
cost battery(ies) to  supply the market if MC


-------
             $/q
                                                AVC = MC
                                                               q/t
                  (a) Inverted L-Shaped Supply Function at Single-Battery Plant
             $/q
                                                MC battery 1

                                                MC battery 2
                                                               q/t
                  (b) Inverted L-Shaped Supply Functions at Multibattery Plant
             $/q
                                             Eq
                             (c) Step-wise Aggregate Supply Curve

Figure A-7. Facility-Level Supply Functions for Coke
                                                               q/t
                                         A-16

-------
       Foreign Supply of Furnace Coke.  Foreign supply of furnace coke (qFSc) is expressed
as

                                qFSc =  AFC (p «)«                                (A.4)
where

       Ap  =  multiplicative parameter for the foreign furnace coke supply equation and

       g£   =  foreign supply elasticity for furnace coke.

The multiplicative parameter (Ap) calibrates the foreign coke supply equation to replicate
the observed 2000 level of furnace coke imports based on the market price and the foreign
supply elasticity.

A. 3.1.2  Market Demand for Furnace Coke

       Market demand for furnace coke (QDc) is the sum of domestic demand from
integrated steel mills and foreign demand  (exports), i.e.,

                                Q00 =  qi°C +  
-------
                     DC
                        =  MAX

(A.6)
       rim   = ^ co^e rate ^or mte§rate(l steel mill (1)> which specifies the amount of
               furnace coke input per unit of final steel mill product;
             = supply of steel mill product from integrated mill (1); and
       qAc   = the furnace coke production from captive battery (j) at integrated steel
               mill (1).
The MAX function in Eq. (A.3) indicates that if the amount of furnace coke consumption
required by an integrated mill to produce steel mill products is greater than its total captive
production, then demand from the furnace coke market will equal the difference; otherwise,
the mill's demand from the furnace coke market will be zero (i.e., it fully satisfies internal
requirements from its captive operations).

       Increases in the price for furnace coke will increase the per-unit costs of final steel
products and thereby shift upward the integrated mill's supply curve  for steel mill products.
The shift in the supply curve decreases the market quantity of finished steel products
produced, which subsequently reduces the quantity of furnace coke consumed at integrated
mills and shifts their demand curve downward in the furnace coke market.

       Foreign Demand for Furnace Coke (Exports).  Foreign demand for furnace coke is
expressed as

                                qf =  BFC (p c)^                                (A.7)
where

       Bp    = multiplicative demand parameter for the foreign furnace coke demand
               equation and
       •qp    = foreign demand elasticity for furnace coke.

The multiplicative demand parameter, 3^ calibrates the foreign coke demand equation to
replicate the observed 2000 level of foreign exports based on the market price and the
foreign demand elasticity.
                                        A-18

-------
A.3.2  Market for Steel Mill Products

       The market for steel mill products consists of supply from domestic mills and foreign
imports and of demand from domestic and foreign consumers.  Steel mill products are
modeled as a single commodity market.  The domestic supply for steel mill products includes
production from integrated mills operating blast furnaces that require furnace coke and from
nonintegrated mills that operate electric arc furnaces that do not. The coke oven NESHAP is
expected to increase the cost of furnace coke inputs. In addition, the integrated iron and steel
NESHAP will also increase the costs of production leading to similar impacts. This will
increase the cost of production at integrated mills and thereby shift their supply curves
upward and increase the price of steel mill products.

A.3.2. 1   Market Supply of Steel Mill Products

       The market supply for steel mill products (QSs) is defined as the sum of the supply
from integrated iron and steel mills, nonintegrated mills, and foreign imports, i.e.,

                             QSs  = qiss +  <*J +  
-------
the production of steel mill products at each facility.  This technology is appropriate, given
the fixed-proportion material input of coke and the variable-proportion inputs of labor,
energy, and raw materials. The generalized Leontief supply function is
                              Ss
                                         B
                                                                               (A.10)
where ps is the market price for the steel product, YI and p are model parameters, and 1
indexes affected integrated mills. The theoretical restrictions on the model parameters that
ensure upward-sloping supply curves are YI > 0 and p < 0.

       Figure A-8 illustrates the theoretical supply function of Eq. (A. 10).  As shown, the
up ward-sloping supply curve is specified over a productive range with a lower bound of zero
                                             R2
that corresponds with a shutdown price equal to -1-— and an upper bound given by the
                                             4Y?
productive capacity of qf that is approximated by the supply parameter YI-  The curvature of
the supply function is determined by the p parameter.
Figure A-8. Theoretical Supply Function for Integrated Facilities and Foundries
                                        A-20

-------
       To specify the supply function of Eq. (A. 10) for this analysis, the p parameter was
computed by substituting a market supply elasticity for the product (£), the market price of
the product (p), and the average annual production level across mills (q) into the following
equation:
p=  -
                                                                                (A. 11)
The p parameter was calculated by incorporating market price and elasticity of supply values
into Eq. (A. 11).

       The intercept of the supply function, yb approximates the productive capacity and
varies across products at each facility. This parameter does not influence the facility's
production responsiveness to price changes as does the p parameter. Thus, the parameter YI
is used to calibrate the economic model so that each individual facility's supply equation
matches its baseline production data from 2000.

       Modeling the Impact of Compliance Costs. The effect of the coke oven NESHAP is
to increase the MC of producing furnace coke by the compliance costs. These costs include
the variable component consisting of the operating and maintenance costs and the
nonvariable component consisting of the control equipment required for the regulatory
option. Regulatory control costs will shift the supply curve upward for each affected facility
by the annualized compliance cost (operating and maintenance plus annualized capital)
expressed per unit of coke production. Computing the supply shift in this way treats
compliance costs as the conceptual equivalent of a unit tax on output.  For coke facilities, the
horizontal portion of its supply curve will rise by the per-unit total compliance costs.  In this
case, the  MC curve will shift by this amount to allow the new higher reservation price for the
coke battery to appropriately reflect the fixed costs of compliance in the operating decision.
At a multiple-battery facility, the change in each battery's MC may cause a reordering of the
steps because the compliance costs vary due to the technology, age, and existing controls of
individual batteries.

       Compliance costs on captive furnace coke batteries will directly affect production
decisions at integrated mills, while compliance costs on merchant furnace coke batteries will
indirectly affect these decisions through the change in the market price of furnace coke. In
addition, direct compliance costs associated with the integrated iron and steel NESHAP will
directly affect production decisions at these mills.  Both of these  impacts were modeled as
reducing the net price integrated mills receive for steel mill products. Returning to the

                                         A-21

-------
integrated mill's supply function presented in Eq. (A.10), the mill's production quantity with
compliance costs is expressed as
                                                                                 (A. 12)
Ss
Y
1(1)
J5
1+2"
1


where

       rim    =  ^e co^e rate ^or ifltegrated steel mill (1), which specifies the amount of
                 furnace coke input per unit of steel mill product;
       «!     =  the share of integrated steel mill 1's furnace coke provided by captive
                 batteries;
       A:f   =  change in per-unit cost of captive coke production at integrated steel mill 1;
       (1-cci)  =  share of integrated steel mill 1's furnace coke provided by the market;
       Apc    =  change in the market price for furnace coke; and
       A^l   =  change in per-unit compliance cost at integrated steel mill 1.
The bracketed term in the denominator represents the increased costs due to the coke ovens
NESHAP and integrated iron and steel NESHAP (i.e., both the direct and indirect effects).
The coke oven NESHAP compliance costs,  ^c and Apc, are expressed per ton of furnace
                                            1
coke and weighted to reflect each integrated mill's reliance on captive versus market furnace
coke.2 The change in the cost per ton of furnace coke due to the regulation is then multiplied
by the mill's coke rate to obtain the change in the cost per ton of steel mill product.  The
integrated iron and steel NESHAP compliance costs, ^s, are also expressed in cost per ton
of steel mill product. These changes in the cost per ton of steel  mill product correspond to
the shift  in the affected facility supply curve shown in Figure A-5b.

       Supply from Nonintegrated Mitts. The supply of steel mill products from domestic
nonintegrated mills is specified as
2The captive versus market furnace coke weights are endogenous in the model because integrated mills exhaust
   their captive supply of coke first; hence, changes in coke consumption typically come from changes in
   market purchases, while captive consumption remains relatively constant.

                                         A-22

-------
                               qM =  A^(ps)§-                               (A.13)
where

       ANI =  multiplicative parameter for nonintegrated mill supply equation and

       £NI  =  ^ nonmtegrated mill supply elasticity for steel mill products.

The multiplicative supply parameter is determined by backsolving Eq. (A.13), given baseline
values of the market price, supply elasticities, and quantities supplied by nonintegrated mills
and foreign mills.

       Foreign Supply (Imports).  The supply of steel mill products from foreign suppliers
(imports) is specified as

                               qFSs  = AFs(ps/F                               (A.14)
where

       Ap =  multiplicative parameter for foreign supply equation and

       £F  =  the foreign supply elasticity for steel mill products.

The multiplicative supply parameters are determined by backsolving Eq. (A.14), given
baseline values of the market price, supply elasticity, and level of imports.

A. 3.2.2 Market Demand for Steel Mill Products

       The market demand for steel mill products, QDs, is the  sum of domestic and foreign
demand, i.e.,

                               QDs  = q? + qFDS                               (A.15)
where

       q°s =  domestic demand for steel mill products and

       qPs =  foreign demand for steel mill products (exports).
                                        A-23

-------
       Domestic Demand for Steel Mitt Products. The domestic demand for steel mill
products is expressed as
where

       Bp  = multiplicative parameter for domestic steel mill products demand equation
              and
       Tjp  = domestic demand elasticity for steel mill products.

The multiplicative demand parameter calibrates the domestic demand equation given
baseline data on price and demand elasticity to replicate the observed 2000 level of domestic
consumption.

       Foreign Demand for Steel Mill Products (Exports). Foreign demand (exports) for
steel mill products is expressed as
                               qFDs=
where
       BFS   = multiplicative demand parameter for foreign steel mill products' demand
              equation and
       •qp   = foreign (export) demand elasticity for steel mill products.

The multiplicative demand parameter calibrates the foreign demand equation given data on
price and demand elasticities to replicate the observed 2000 level of foreign exports.

A.3.3  Market for Foundry Coke

       The market for furnace coke consists of supply from domestic merchant coke plants
and imports and demand from foundries operating cupola furnaces. The domestic supply for
foundry coke is modeled as a step-wise supply function developed from the marginal cost of
production at individual foundry coke batteries.  Imports are modeled using a representative
supply curve. The domestic demand is derived from iron castings production at foundries
operating cupola furnaces (domestic and foreign) as determined through the market for iron
castings and coking rates. The following section details the market supply and demand
components for this analysis.
                                        A-24

-------
A. 3.3.1   Market Supply of Foundry Coke

       The market supply of foundry coke, Qsk, is composed of the supply from domestic
merchant plants reflecting plant-level production decisions for individual merchant coke
batteries, and a single representative foreign supply curve, i.e.,
                         Qsk =
                              Merchant  i
where

       1       = plants,
       j       = batteries,
        Ii5n ••>   = suPply °f foundry coke from coke battery (j) at merchant plant (1), and
       qpk     = foundry coke supply from imports.

As was the case for furnace coke batteries, the marginal cost for an individual foundry coke
battery is assumed to be constant reflecting a fixed-coefficient technology.  Marginal cost
curves were developed for all foundry coke batteries at merchant plants in the United States
as detailed in Appendix B.

       Foundry coke production decisions are based on the same approach used to model
furnace coke production decisions. Thus, as illustrated previously in Figure A-7, the
production decision is determined by an inverted L-shaped supply curve that is perfectly
elastic to the capacity level of production and perfectly inelastic thereafter.  Foundry coke
batteries will supply 100 percent of capacity if its marginal cost is less than market price;
otherwise, it will cease production. The regulatory costs shift each affected battery's
marginal cost upward, affecting facilities' decision to operate or shut down individual
batteries.

       Foreign Supply of Foundry Coke. Foreign supply of foundry coke (qpsk) is expressed
as
                                         A-25

-------
where

       Ap  = multiplicative parameter for the foreign foundry coke supply equation, and

       j-J;   = foreign supply elasticity for foundry coke.

The multiplicative parameter (Ap) calibrates the foreign coke supply equation to replicate
the observed 2000 level of foundry coke imports based on the market price and the foreign
supply elasticity.
A.3.3.2  Market Demand for Foundry Coke

       The market demand for foundry coke, QDk, is composed of domestic and foreign
demand by foundries operating cupola furnaces. Therefore, the foundry coke demand is
derived from the production of iron castings from cupola furnaces.  Increases in the price of
foundry coke due to the regulation will lead to decreases in production of iron castings at
foundries operating cupola furnaces.  The demand function for foundry coke is expressed  as
follows:

where

                   r->Dk_  Dk .^Dk _  i     Si  ,^Dk                     f A  9fA
                   ^c   —4CF   4CFF—  CF 4CF   4CFF                   ^JT..^U;

            = derived demand for foundry coke from domestic cupola foundries;
       QCFF  = demand for foundry coke from foreign cupola foundries;

       r'    = the coke rate for cupola foundries, which specifies the amount of foundry
        i_/.r
               coke input per unit output; and
             = quantity of iron castings produced at domestic cupola foundries.

Changes in production at foundries using electric arc and electric induction furnaces to
produce iron castings do not affect the demand for foundry coke.

       Foreign Demand for Foundry Coke (Exports).  Foreign demand for foundry coke is
expressed as
                                        A-26

-------
                               qFDk=  BFk(p^

where

       Bp   = multiplicative demand parameter for the foreign foundry coke demand
               equation and
       rjp    = foreign demand elasticity for foundry coke.

The multiplicative demand parameter,  Bp, calibrates the foreign coke demand equation to
replicate the observed 2000 level of foreign exports based on the market price and the
foreign demand elasticity.

A.3.4  Markets for Iron and Steel Castings

       The model includes two markets for this industry: iron castings and steel castings.
Each market consists of supply from domestic foundries and foreign imports and of demand
from domestic and foreign consumers.  The rule is expected to increase production costs for
selected cupola and electric foundries and thereby shift their supply curves upward and
increase the prices.

A.3.4.1 Market Supply

       The market supply for castings  market i, Qst, is defined as the sum of the supply from
domestic and foreign foundries.

                             Qsi - nSi +  nSi  + n Si                            (A. 99"i
                                ~ IAF    *IUF    HF                             (/\.zz;
where

        •Si   = quantity of castings produced at affected domestic foundries,
             = suPply from unaffected domestic foundries, and

       qp51   = supply from foreign foundries.

       The functional form of the supply curve for domestic foundries is specified as

                                                                               (A.24)
                                        A-27

-------
where

       Ap  = multiplicative parameter for foundries supply equation,

       Ac  = per-unit direct compliance costs of casting production3, and

       g   =  foundries supply elasticity.

The multiplicative supply parameter, AF> is determined by backsolving Eq. (A.24), given
baseline values of the market price, supply elasticity, and quantity supplied.

       Foreign Supply (Imports).  The functional form of the foreign supply curve is
specified as

                                q/ = Ap1 (p ')*                                (A.25)
where

       Ap  = multiplicative parameter for foreign supply equation and

       £'   = foreign supply elasticity.

The multiplicative supply parameter, AF> is determined by backsolving Eq. (A.25), given
baseline values of the market price, supply elasticity, and level of imports.

A.3.4.2  Market Demand

       The market demand for castings market i, (QDl), is the sum of domestic and foreign
demand, and it is expressed as a function of the price of castings:

                                QDi  =  q?  + qFDi                                (A.26)
where

       qP>  = domestic demand for castings and
3The economic model projects the foundry coke price remains unchanged after regulation. Therefore, there is
   no indirect effect of the regulation associated with changes in foundry coke prices.

                                         A-28

-------
       q°'  = foreign demand (exports) for castings.

       Domestic Demand. The domestic demand for castings is expressed as

                               q? =  B^p')1*                               (A.27)
where

       B '  =  multiplicative parameter for domestic demand equation and

       «i  =  domestic demand elasticity.

The multiplicative demand parameter calibrates the domestic demand equation given
baseline data on price and demand elasticity to replicate the observed 2000 level of domestic
consumption.

       Foreign Demand.  Foreign demand (exports) is expressed as

                               qFDi =  B^p^                               (A.28)
where

       B   =  multiplicative demand parameter for demand equation and
           =  foreign (export) demand elasticity.

The multiplicative demand parameter, Bp, is determined by backsolving Eq. (A.28), given
baseline values of market price, demand elasticity, and level of exports.

A.3.5  Post-regulatory Market Equilibrium Determination

       Integrated steel mills and iron foundries with cupola furnaces must determine output
given the market prices for their finished products, which in turn determines their furnace
and foundry coke requirements. The optimal output of steel mill products at integrated mills
also depends on the cost of producing captive furnace coke and the market price of furnace
coke; whereas iron and steel foundries with cupolas depend on only the market price of
foundry coke because they have no captive operations. Excess production of captive furnace
coke at integrated mills will spill over into the furnace coke market; whereas an excess
                                        A-29

-------
demand will cause the mill to demand furnace coke from the market. For merchant coke
plants, the optimal market supply of furnace and/or foundry coke will be determined by the
market price of each coke product.

       Facility responses and market adjustments can be conceptualized as an interactive
feedback process.  Facilities face increased costs from the regulation, which initially reduce
output. The cumulative effect of these individual changes leads to an increase in the market
price that all producers (affected and unaffected) and consumers face, which leads to further
responses by producers (affected and unaffected) as well as consumers and thus new market
prices, and so on. The new equilibrium after imposing the regulation is the result of a series
of iterations between producer and consumer responses and market adjustments until a stable
market price arises where market supply equals market demand for each product, i.e., Qs =
QD-
       The Agency employed a Walrasian auctioneer process to determine equilibrium price
(and output) associated with the increased production costs of the regulation. The auctioneer
calls out a market price for each product and evaluates the reactions by all  participants
(producers and consumers), comparing total quantities supplied and demanded to determine
the next price that will guide the market closer to equilibrium (i.e., where market supply
equals market demand). Decision rules are established to ensure that the process will
converge to an equilibrium, in addition to specifying the conditions for equilibrium.  The
result of this approach is a vector of prices with the regulation that equilibrates supply and
demand for each product.

       The algorithm for deriving the with-regulation equilibria in all markets can be
generalized to five recursive steps:

       1.   Impose the control costs for each affected facility, thereby affecting their supply
           decisions.
       2.   Recalculate the production decisions for coke products and both final steel mill
           products and iron castings across all affected facilities. The adjusted production
           of steel mill products from integrated steel mills and iron castings from foundries
           with cupola furnaces determines the derived demand for  furnace and foundry
           coke through the input ratios. Therefore, the domestic demand for furnace and
           foundry coke is simultaneously determined with the domestic supply of final steel
           mill products and iron castings from these suppliers. After accounting  for these
           adjustments, recalculate the market supply of all products by aggregating across
           all producers, affected and unaffected.
                                         A-30

-------
       3.  Determine the new prices via a price revision rule for all product markets.
       4.  Recalculate the supply functions of all facilities with the new prices, resulting in a
          new market supply of each product, in addition to derived (domestic) demand for
          furnace and foundry coke. Evaluate domestic demand for final steel mill products
          and iron castings, as well as import supply and export demand for appropriate
          products given the new prices.
       5.  Go to Step #3, resulting in new prices for each product.  Repeat until equilibrium
          conditions are satisfied in all markets (i.e., the ratio of supply to demand is
          approximately one for each and every product).
A.3.6  Economic Welfare Impacts

       The economic welfare implications of the market price and output changes with the
regulation can be examined using two slightly different tactics, each giving a somewhat
different insight but the same implications: changes in the net benefits of consumers and
producers based on the price changes and changes in the total  benefits and costs of these
products based on the quantity changes. This analysis focuses on the first measure—the
changes in the net benefits of consumers and producers.  Figure A-9 depicts the change in
economic welfare by first measuring the change in consumer surplus and then the change in
producer surplus.  In  essence, the demand and supply curves previously used as predictive
devices are now being used as a valuation tool.

       This method of estimating the change in economic welfare with the regulation
divides society into consumers  and producers. In a market environment, consumers and
producers of the good or service derive welfare from a market transaction. The difference
between the maximum price consumers are willing to pay for  a good and the price they
actually pay is referred to as "consumer surplus."  Consumer surplus is measured as the area
under the demand curve and above the price of the product.  Similarly, the difference
between the minimum price producers are willing to accept  for a good and the price they
actually receive is referred to as "producer surplus" or profits. Producer surplus is measured
as the area above the  supply curve and below the price of the product. These areas can be
thought of as consumers'  net benefits of consumption and producers' net benefits of
production, respectively.

       In Figure A-9, baseline equilibrium occurs at the intersection of the demand curve, D,
and supply curve, S.  Price is Pj with quantity Q. The increased cost of production with the
regulation will cause  the market supply curve to shift upward to  S'.  The new equilibrium
price of the product is P2. With a higher price for the product, there is less consumer welfare,

                                        A-31

-------
                    $/Q
                                            Q
               Q/t
                       (a) Change in Consumer Surplus with Regulation

                    $/Q                                    s'
                                           Q
               Q/t
                       (b) Change in Producer Surplus with Regulation
                    $/Q
Q2  Q,
                                                          Q/t
                       (c) Net Change in Economic Welfare with Regulation
Figure A-9. Economic Welfare Changes with Regulation: Consumer and Producer
Surplus
                                      A-32

-------
all else being unchanged as real incomes are reduced. In Figure A-9(a), area A represents the
dollar value of the annual net loss in consumers' benefits with the increased price. The
rectangular portion represents the loss in consumer surplus on the quantity still consumed,
Q2, while the triangular area represents the foregone surplus resulting from the reduced
quantity consumed, Qi-Q2.

       In addition to the changes in consumer welfare, producer welfare also changes with
the regulation. With the increase in market price, producers receive higher revenues on the
quantity still purchased, Q2.  In Figure A-9(b), area B represents the increase in revenues due
to this increase in price. The difference in the area under the supply curve up to the original
market price, area C, measures the loss in producer surplus, which includes the loss
associated with the quantity no longer produced. The net change in producer welfare is
represented by area B-C.

       The change in economic welfare attributable to the compliance costs  of the regulation
is the sum of consumer and producer surplus changes, that is, - (A) + (B-C).  Figure A-9(c)
shows the net (negative) change in economic welfare associated with the regulation as  area
D. However, this analysis does not include the benefits that occur outside the market (i.e.,
the value of the reduced levels of air pollution with the regulation).  Including this benefit
may reduce the net cost of the regulation or even make it positive.
                                         A-33

-------
                                   APPENDIX B

            DEVELOPMENT OF COKE BATTERY COST FUNCTIONS
       This appendix outlines EPA's method for estimating 2000 baseline production costs
for coke batteries. The Agency used a coke production cost model developed in support of
the 1993 MACT on coke ovens. EPA's Technical Approach for a Coke Production Cost
Model (EPA,  1979) provides a more detailed description of this model.  For this analysis, the
model was updated with reported technical characteristics of coke batteries from the
Information Collection Request (ICR) survey responses and available price data (see
Table B-l). In addition, the Agency incorporated estimates of MACT pollution abatement
costs developed for the  1993 MACT on coke ovens (EPA, 1991b).

B.I    Variable Costs

       Coke batteries use four variable inputs during the manufacturing process—
metallurgical  coal, labor, energy, and other materials/supplies. Metallurgical coal is
essentially the only raw material used in the production of coke. Labor transports and
delivers the raw materials as well as final products.  Coke ovens and auxiliary equipment
consume energy and supplies during the production process and periodic maintenance and
repair of the coke  batteries.

       Coke production requires a fixed amount of each variable input per ton of coke,  and
these inputs are not substitutable.  Accordingly, the  total variable cost function is linear in the
output and input prices, or, in other words, the average variable cost function is independent
of output.  Therefore, the average variable cost function (expressed in dollars per short ton of
coke) can  be written as

                 AVC = AV_CI«PC + AV_LI«w + AV_EI»Pe + AV_OI«P0            (B.I)

where AV_CI, AV_LI, AV_EI, and AV_OI are the fixed requirements per ton of coke of
metallurgical  coal, labor, energy, and other material and supplies. Pc, w, Pe, and P0 are the
prices of each variable input, respectively.  As shown above, the contribution of each
variable input to the per-unit coke cost is equal to the average variable input (fixed
requirement of the input per ton of coke) times the price of the input.  For  example, the
                                        B-l

-------
 Table B-l. Key Parameter Updates for Coke Production Cost Model: 2000a
Variable
Rl
R2
R3
R4
R7
R8
R9
RIO
Rll
R12
R13
R14
R14*
R15
R16
R17
R18
R19
R20
R21
R22
R23
R25

Description
Steam Cost
Cooling Water
Electricity
Underfire Gas
Calcium Hydroxide
Sulfuric Acid
Sodium Carbonate
Sodium Hydroxide
Coal Tar Credit
Crude Light Oil
BTX Credit
Ammonium Sulfate Credit
Anhydrous Ammonia Credit
Elemental Sulfur Credit
Sodium Phenolate Credit
Benzene Credit
Toluene Credit
Xylene Credit
Naphalene Credit
Coke Breeze Credit
Solvent Naptha Credit
Wash Oil Cost
Phosphoric Acid (commercial)
Industrial Coke Price
Units
$71,000 Ib steam
$71,000 gal
$7kWh
$71 03 eft
$/ton
$/ton
$/ton
$/ton
$/gal
$/gal
$/gal
$/ton
$/ton
$/ton
$/ton
$/gal
$/gal
$/gal
$/lb
$/ton
$/gal
$/gal
$/ton
$/ton
2000
8.97
0.26
Varies by state
1.06
74.00
79.00
537.00
315.00
0.82
1.27
0.94
40.04
239.21
287.48
864.12
1.21
0.85
0.75
0.27
45.62
0.88
1.29
711.31
112.00
 This table provides price updates for the coke production cost model (EPA, 1979, Table 2-3).
contribution of labor to the cost per ton of coke (AV_LI) is equal to the labor requirement
per ton of coke times the price of labor (w).

       The variable costs above include those costs associated with by- and co-product
recovery operations associated with the coke battery.  To more accurately reflect the costs
specific to coke production, the Agency subtracted by- and co-product revenues/credits from
Eq. (B.I). By-products include tar and coke oven gas among others, while co-products
include coke breeze and other industrial coke.  Following the same fixed coefficient
                                         B-2

-------
approach, these revenues or credits (expressed per ton of coke) are derived for each
recovered product at the coke battery by multiplying the appropriate yield (recovered product
per ton of coke) by its price or value. The variable cost components and by-/co-product
credits are identified below.

B.1.1  Metallurgical Coal (AVCI, Pc)

       The ICR survey responses provided the fixed input requirement for metallurgical coal
at each battery. Based on the responses from the survey, U.S. coke producers require an
average of 1.36 tons of coal per ton of coke produced. This fixed input varies by type of
producer. Integrated, or captive, producers require an average of 1.38 tons of coal per ton of
coke produced, while merchant producers require an average of 1.31 tons of coal per ton of
coke produced. The U.S. Department of Energy provides state-level coal price data for
metallurgical coal. For each coke battery, EPA computed the cost of coal per short ton  of
coke by multiplying its input ratio times the appropriate state or regional price. As shown in
Table B-2, the average cost of metallurgical coal per ton of coke in 2000 was $61.23 for
captive producers and $57.98 for merchant producers.

  Table B-2.  Metallurgical Coal Costs by Producer Type:  2000 ($/ton of coke)

Number of batteries
Average
Minimum
Maximum
Captive
40
$61.23
$56.21
$71.98
Merchant
18
$57.98
$52.17
$68.39
All Coke Batteries
58
$60.22
$52.17
$71.98
B.I.2  Labor (AVLI, w)

       The cost model provides an estimate of the fixed labor requirement for operation,
maintenance, and supervision labor at each battery.  The Agency used these estimates to
derive the average variable labor cost for each individual battery given its technical
characteristics and the appropriate state-level wage rates obtained from the U.S. Bureau of
Labor Statistics (2002). As shown in Table B-3, average labor costs per ton of coke are
significantly lower for captive producers (e.g., $17.18 per ton of coke)  relative to merchant
                                         B-3

-------
 Table B-3.  Labor Costs by Producer Type: 2000 ($/ton of coke)

Number of batteries
Average
Minimum
Maximum
Captive
40
$17.18
$9.19
$38.35
Merchant
18
$28.95
$11.07
$44.63
All Coke Batteries
58
$20.83
$9.19
$44.63
producers (e.g., $28.95 per ton of coke). Captive batteries are typically larger capacity
batteries and therefore require fewer person-hours per ton of coke.

B.I.3  Energy (AVEI, Pe)

       The cost model estimates the fixed energy requirements (i.e., electricity, steam, and
water) for each battery.  These estimates are used to derive the energy costs per ton of coke
for each battery. Captive producers have a lower electricity requirement (i.e., 47.58 kWh per
ton of coke) relative to merchant producers (i.e., 50.96 kWh per ton of coke). As shown in
Table B-4, the average energy cost per ton of coke across all coke batteries is $5.77.
Average energy costs per ton of coke are lower for captive producers (e.g., $5.51 per ton of
coke) relative to merchant producers (e.g., $6.34 per ton of coke).  This difference reflects
lower state/regional electricity prices in regions where  captive batteries produce coke.

 Table B-4.  Energy Costs by Producer Type: 2000  ($/ton of coke)
                             Captive
                   Merchant
                All Coke Batteries
 Number of batteries
 Average
 Minimum
 Maximum
  40
$5.51
$3.91
$16.11
  18
$6.34
$4.31
$15.41
  58
$5.77
$3.91
$16.11
                                         B-4

-------
B.1.4  Other Materials and Supplies (A VOI, P0)

       The fixed requirements for other materials and supplies associated with the
production of coke include

       •   chemicals,
       •   maintenance materials,
       •   safety and clothing, and
       •   laboratory and miscellaneous supplies.
As shown in Table B-5, the cost model estimates the average cost for these items across all
coke batteries is $4.76 per short ton of coke, ranging from $3.26 to $7.69 per ton of coke.
These costs vary by producer type, with merchant producers averaging $5.53 per ton of coke
versus captive producers who average $4.42 per ton of coke.

 Table B-5. Other Costs by Producer Type:  2000 ($/ton of coke)

Number of batteries
Average
Minimum
Maximum
Captive
40
$4.42
$3.27
$7.69
Merchant
18
$5.53
$3.26
$7.42
All Coke Batteries
58
$4.76
$3.26
$7.69
B.1.5  By- and Co-product Credits

       In addition to the variable cost inputs described above, by- and co-products are
associated with the manufacture of coke products. Therefore, the Agency modified Eq. (B.I)
by subtracting (1) revenues generated from the sale of by-/co-products and (2) credits
associated with using coke oven gas as an energy input in the production process. The
following cost function adjustments were made to the engineering model to incorporate by-
and co-products into the coke-making cost function:

       •   Coke breeze—ICR survey responses provided coke breeze output per ton of coke
          for each battery.
                                         B-5

-------
       •  Other industrial coke—ICR survey responses provided other industrial coke
          output per ton of coke for each battery.

       •  Coke oven gas—Based on secondary sources and discussions with engineers,
          furnace coke producers were assumed to produce 8,500 ft3 per ton of coal, and
          foundry producers were assumed to produce 11,700 ft3 per ton of coal (Lankford
          et al., 1985; EPA, 1988).
As shown in Table B-6, the average by-/co-product credit is $19.54 per ton of coke for
captive producers and $24.05 per ton of coke for merchant producers.

 Table B-6.  By-/Co-Product Credits by Producer Type: 2000 ($/ton of coke)

Number of batteries
Average
Minimum
Maximum
Captive
40
$19.54
$16.09
$35.99
Merchant
18
$24.05
$10.69
$51.78
All Coke Batteries
58
$20.94
$10.69
$51.78
B.2    MACT/LAER Pollution Abatement Costs

       The 1990 Clean Air Act Amendments mandated two levels of control for emissions
from coke ovens.  The first control level, referred to as MACT, specified limits for leaking
doors, lids, offtakes, and time of charge.  This level of control was to be attained by 1995.
The second level of control, Lowest Achievable Emissions Rate (LAER), specified more
stringent limits for leaking doors and offtakes. Estimates of the MACT and LAER costs
associated with these controls were developed for EPA's Controlling Emissions from By-
Product Coke Oven Charging, Door Leaks, and Topside Leaks: An Economic Impacts
Analysis  (EPA, 1991a).1 Table B-7 provides summary statistics for the projected costs
associated with each level of control.  However, the Agency determined that industry actions
undertaken in the interim period to comply with the MACT limits have enabled them to also
meet the  LAER limits.  Therefore, only the MACT-related pollution abatement costs have
'The Agency estimated costs for the LAER control level using two scenarios. The first (LAER-MIN) assumed
   all batteries will require new doors and jambs.  The second (LAER-MAX) also assumed all batteries will
   require new doors and jambs and in addition assumed batteries with the most serious door leak problems
   would be rebuilt. This analysis reports cost estimates for the LAER-MIN scenario.

                                         B-6

-------
 Table B-7. Pollution Abatement Costs by Producer Type: 2000 ($/ton of coke)

Number of batteries
MACT
Average
Minimum
Maximum
LAER
Average
Minimum
Maximum
Captive
40

$0.83
$0.00
$2.59

$1.64
$0.07
$2.63
Merchant
18

$2.34
$0.00
$11.14

$2.44
$0.94
$6.07
All Coke Batteries
58

$1.30
$0.00
$11.14

$1.88
$0.07
$6.07
been incorporated to determine the appropriate baseline costs for the 2000 economic model.
As shown in Table B-7, the average MACT pollution abatement cost across all coke batteries
is $1.30 per short ton of coke. The projected costs for captive producers range from zero to
$2.59 per ton of coke, while projected costs for merchant producers range from zero to
$11.14 per ton of coke.

B.3    Fixed Costs

       Production of coke requires the combination of variable inputs outlined above with
fixed capital equipment (e.g., coke ovens and auxiliary equipment).  It also includes other
overhead and administrative expenses. For each coke battery, the average fixed costs per ton
of coke can be obtained by dividing the total fixed costs (TFC) estimated by the coke model
by total battery coke production. Therefore, the average fixed cost function (expressed in
dollars per ton of coke) can be written as
                        AFC = (PTI + ASF +PYOH+ PLOH)/Q
(B.2)
where
          property taxes and insurance (PTI) = (0.02)«($225«Coke Capacity). This category
          accounts for the fixed costs associated with property taxes and insurance for the
          battery. The cost model estimates this component as 2 percent of capital cost.
          Capital costs are estimated to be $225 per annual short ton of capacity based on
          reported estimates of capital investment cost of a rebuilt by-product coke-making
          facility (USITC,  1994). As shown in Table B-8, the average PTI cost across all
          batteries is $4.47 per ton of coke.
                                        B-7

-------
Table B-8. Average Fixed Costs by Producer Type:  2000 ($/ton of coke)

Number of batteries
Property taxes and insurance
Average
Minimum
Maximum
Administrative and sales expense
Average
Minimum
Maximum
Payroll overhead
Average
Minimum
Maximum
Plant overhead
Average
Minimum
Maximum
Captive
40

$4.41
$3.20
$6.78

$4.96
$3.60
$7.63

$3.44
$1.84
$7.67

$10.18
$5.73
$21.83
Merchant
18

$4.58
$3.55
$6.11

$5.16
$4.00
$6.87

$5.79
$2.21
$8.93

$18.91
$7.92
$28.62
All Coke Batteries
58

$4.47
$3.20
$6.78

$5.02
$3.60
$7.63

$4.17
$1.84
$8.93

$12.89
$5.73
$28.62
         administration and sales expense (ASE) = (0.02)«($225«Coke capacity).  This
         category accounts for the fixed costs associated with administrative and sales
         expenses for the coke battery.  The cost model also calculates this component as 2
         percent of capital cost.  As shown in Table B-8, the average cost across all coke
         batteries for ASE is $5.02 per ton of coke.

         payroll overhead (PYOH) = (0.2)» (Total labor costs). Payroll overhead is
         modified as 20 percent of total labor costs. Payroll overhead is used to capture
         fringe benefits because wage rates obtained from the Bureau of Labor Statistics
         exclude fringe benefits. As shown in Table B-8, the  average payroll overhead is
         $3.44 per ton of coke for captive producers and $5.79 per ton of coke for
         merchant producers, reflecting the different labor requirements by producer type.

         plant overhead (PLOH) = (0.5)»(Total payroll + Total other expenses). The cost
         model computes plant overhead as 50  percent of total payroll and total other
         expenses by producer type. As shown in Table B-8,  the average plant overhead
         cost is $10.18 for captive producers and $18.91 for merchant producers.  As with
                                       B-8

-------
          payroll overhead, this difference reflects differences in labor requirements for
          captive and merchant producers.
B.4    Summary of Results

       Table B-9 summarizes each cost component and aggregates them to estimate the
average total costs per ton of coke by producer type. As shown, the average total cost (ATC)
across all coke batteries is $98.49 per short ton of coke. The ATC for captive producers is
$92.62 per short ton of coke and is significantly lower than the ATC for merchant producers
at $111.52. This difference reflects both economies of scale and lower production costs
associated with the production of furnace coke.  These differences are also consistent with
observed market prices for furnace coke of $112 (produced mainly by captive producers) and
for foundry coke of $161 (produced solely by merchant producers with some furnace coke)
(USITC, 2001b, 2001c). A correlation analysis of these cost estimates shows that ATC is
negatively correlated with coke battery capacity (correlation coefficient of -0.70) and
start/rebuild date (correlation coefficient of -0.63).  Therefore, average total costs are lower
for larger coke batteries and those that are new or recently rebuilt.  Tables B-10 and B-l 1
present cost estimates for individual captive and merchant coke batteries, respectively.

B.5    Nonrecovery Coke Making

       Several substitute technologies for by-product coke making have been developed in
the United States and abroad.  In the United States, the nonrecovery method is the only
substitute that has a significant share of the coke market. This technology is relatively new,
and, as a result, the original coke production cost model did not include estimates for these
types of coke-making batteries.  The nonrecovery process is less costly than the by-product
process because of the absence of recovery operations and a lower labor input requirement
per ton of coke. Therefore, the Agency modified the model to reflect these cost advantages
in the following manner:

       •   No expenses/credits associated with by- and co-product recovery.
       •   Reduced labor input—labor requirement estimates generated by the model were
          multiplied by a factor of 0.11, which represents the ratio of employment per ton
          of coke at merchant batteries to employment per ton of coke at nonrecovery
          batteries.
                                         B-9

-------
 Table B-9. Cost Summary by Producer Type: 2000 ($/ton of coke)

Number of batteries
Average variable costa
Average
Minimum
Maximum
MACT
Average
Minimum
Maximum
Average fixed cost
Average
Minimum
Maximum
Average total cost
Average
Minimum
Maximum
Captive
40

$68.80
$57.95
$82.94

$0.83
$0.00
$2.59

$22.99
$15.61
$43.91

$92.62
$73.87
$127.07
Merchant
18

$74.74
$39.80
$91.00

$2.34
$0.00
$11.14

$34.44
$17.91
$48.34

$111.52
$69.92
$141.84
All Coke Batteries
58

$70.64
$39.80
$91.00

$1.30
$0.00
$11.14

$26.55
$15.61
$48.34

$98.49
$69.92
$141.84
 ""Includes by-/co-product credits.

       •   Exceed current standards of pollution abatement (Engineering and Mining
          Journal, 1997)—MACT compliance costs were excluded.
As shown in Table B-12, the ATC for nonrecovery coke-making facilities is $69.25 per ton
of coke, which is significantly lower than the average ATC of captive and merchant
producers. These costs vary slightly across these batteries ranging from $67.51 to $70.12 per
ton of coke.  Table B-13 presents cost estimates for individual nonrecovery coke-making
batteries.
                                        B-10

-------
Table B-10. Cost Data Summary for Captive Coke Batteries: 2000
Facility Name
Acme Steel
Acme Steel
AK Steel
AK Steel
AK Steel
Bethlehem Steel
Bethlehem Steel
Bethlehem Steel
Bethlehem Steel
Geneva Steel
Geneva Steel
Geneva Steel
Geneva Steel
Gulf States Steel
Gulf States Steel
LTV Steel
LTV Steel
National Steel
National Steel
National Steel
Producer
Location Type3
Chicago, IL
Chicago, IL
Ashland, KY
Ashland, KY
Middletown, OH
Burns Harbor, IN
Burns Harbor, IN
Lackawanna, NY
Lackawanna, NY
Provo, UT
Provo, UT
Provo, UT
Provo, UT
Gadsden, AL
Gadsden, AL
Chicago, IL
Warren, OH
Ecorse, MI
Granite City, IL
Granite City, IL
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
Capacity
Coke (short
Typeb tons/yr)
1 250
1 250
1 634
1 366
1 429
1 948
1 929
1 375
1 375
1 200
1 200
1 200
1 200
1 250
1 250
1 615
1 549
1 924
1 300
1 300
,000
,000
,000
,000
,901
,000
,000
,000
,000
,000
,000
,000
,000
,000
,000
,000
,000
,839
,931
,931
Start/
Rebuild
Date
1979
1978
1978
1953
1952
1972
1983
1962
1952
1944
1944
1944
1944
1942
1965
1982
1979
1992
1982
1980
AVCC
($/short
ton)
$74
$74
$66,
$69,
$74,
$58
$59
$65
$65
$77
$78
$78
$82
$75
$74
$63
$69
$78
$69,
$69,
.41
.26
.88
.25
.42
.99
.27
.66
.65
.49
.44
.41
.94
.28
.47
.79
.00
.68
.93
.93
MACT
($/short
ton)
$1.02
$1.02
$1.28
$1.02
$1.23
$0.72
$0.71
$1.78
$1.83
$0.27
$0.27
$0.22
$0.22
$1.71
$2.59
$0.36
$0.04
$0.27
$0.68
$0.68
AFC
($/short
ton)
$20.69
$20
$18
$21
$23
$18
$18
$21
$21
$28
$30
$26
$43
$27
$19
$18
$22
$17
$21
$21
.69
.88
.15
.62
.11
.68
.41
.23
.62
.92
.47
.91
.56
.44
.38
.18
.44
.26
.26
ATC
($/short
ton)
$96.13
$95.97
$87.05
$91.42
$99.27
$77.82
$78.66
$88.86
$88.71
$106.38
$109.62
$105.10
$127.07
$104.55
$96.51
$82.52
$91.22
$96.38
$91.87
$91.88
                                                                                               (continued)

-------
Table B-10. Cost Data Summary for Captive Coke Batteries:  2000 (continued)


Facility Name
USX
USX
USX
USX
USX
USX
USX
USX
USX
USX
USX
USX
USX
USX
USX
USX
Wheeling-Pitt
Wheeling-Pitt
Wheeling-Pitt
Wheeling-Pitt
aC = Captive; M =
bl = Furnace; 2 =


Location
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Clairton, PA
Gary, IN
Gary, IN
Gary, IN
Gary, IN
Follansbee, WV
Follansbee, WV
Follansbee, WV
Follansbee, WV
= Merchant.
Foundry; 3 = Both.

Producer Coke
Type" Type"
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1
C 1


Capacity
(short
tons/yr)
844,610
668,680
668,680
373,395
373,395
373,395
378,505
378,505
378,505
378,505
378,505
378,505
827,820
827,820
297,110
297,110
782,000
163,000
151,000
151,000


Start/
Rebuild
Date
1982
1976
1978
1989
1989
1979
1955
1955
1955
1954
1954
1954
1976
1975
1954
1954
1977
1964
1955
1953


AVCC
($/short
ton)
$59.24
$60.62
$60.62
$63.33
$63.33
$63.33
$65.43
$65.43
$65.43
$66.39
$66.39
$66.39
$65.47
$66.41
$72.99
$73.22
$57.95
$73.58
$74.69
$74.69


MACT
($/short
ton)
$0.72
$0.00
$0.00
$0.00
$0.00
$1.04
$1.04
$1.09
$1.09
$1.09
$1.04
$0.00
$0.65
$0.65
$1.51
$1.51
$0.31
$1.36
$1.11
$1.11


AFC
($/short
ton)
$15.75
$20.32
$20.32
$21.71
$21.71
$21.71
$22.73
$22.73
$22.73
$22.46
$22.46
$22.46
$23.24
$22.60
$24.76
$25.94
$15.61
$30.00
$29.28
$29.28


ATC
($/short
ton)
$75.71
$80.94
$80.94
$85.03
$85.03
$86.07
$89.20
$89.25
$89.25
$89.94
$89.89
$88.85
$89.36
$89.67
$99.26
$100.67
$73.87
$104.93
$105.07
$105.07



-------
Table B-11. Cost Data Summary for Merchant Coke Batteries: 2000
Facility Name
ABC Coke
ABC Coke
ABC Coke
Citizens Gas
Citizens Gas
Citizens Gas
Empire Coke
Empire Coke
Erie Coke
Erie Coke
Koppers
Koppers
New Boston
Shenango
Sloss Industries
Sloss Industries
Sloss Industries
Tonawanda
aC = Captive; M =
bl = Furnace; 2 =
Location
Tarrant, AL
Tarrant, AL
Tarrant, AL
Indianapolis, IN
Indianapolis, IN
Indianapolis, IN
Holt, AL
Holt, AL
Erie, PA
Erie, PA
Monessen, PA
Monessen, PA
Portsmouth, OH
Pittsburgh, PA
Birmingham, AL
Birmingham, AL
Birmingham, AL
Buffalo, NY
= Merchant.
Foundry; 3 = Both.
Producer
Type3
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M

Coke
Type"
2
3
3
3
2
2
2
2
2
2
1
1
1
1
3
1
1
2

Capacity
(short
tons/yr)
490,528
112,477
96,962
389,116
128,970
116,845
108,026
54,013
130,073
84,878
245,815
126,766
346,126
514,779
184,086
133,931
133,931
268,964

Start/
Rebuild
Date
1968
1951
1941
1979
1946
1941
1978
1978
1943
1952
1981
1980
1964
1983
1959
1952
1956
1962

AVCC
($/short
ton)
$66.46
$81.68
$86.10
$47.46
$79.85
$84.51
$88.52
$90.09
$73.99
$75.12
$79.25
$91.00
$78.73
$78.87
$44.32
$79.78
$79.78
$39.80

MACT
($/short
ton)
$1.22
$2.69
$2.56
$1.05
$2.02
$2.13
$7.38
$11.14
$1.73
$1.48
$0.12
$0.36
$1.35
$0.00
$1.61
$1.61
$1.61
$2.03

AFC
($/short
ton)
$17.91
$32.48
$36.12
$21.41
$43.85
$48.34
$38.11
$40.61
$46.76
$48.19
$30.25
$39.67
$27.76
$28.29
$25.59
$30.30
$30.30
$34.09

ATC
($/short
ton)
$85.59
$116.85
$124.78
$69.92
$125.72
$134.98
$134.01
$141.84
$122.48
$124.78
$109.63
$131.03
$107.84
$107.16
$71.52
$111.69
$111.69
$75.92


-------
Table B-12. Cost Summary for Nonrecovery Coke Batteries:  2000 ($/ton of coke)
Nonrecovery
Number of batteries
Metallurgical coal
Average
Minimum
Maximum
Labor
Average
Minimum
Maximum
Energy
Average
Minimum
Maximum
Other
Average
Minimum
Maximum
Average fixed cost
Average
Minimum
Maximum
Average total cost
Average
Minimum
Maximum
8

$47.58
$46.95
$48.21

$2.07
$1.47
$2.68

$6.45
$6.25
$6.71

$2.53
$2.44
$2.66

$10.62
$10.07
$11.13

$69.25
$67.51
$70.12
                                    B-14

-------
Table B-13.  Cost Data Summary for Nonrecovery Coke Batteries:  1997
Facility Name
Jewell Coke
and Coal
Jewell Coke
and Coal
Jewell Coke
and Coal
Jewell Coke
and Coal
Indiana Harbor
Coke Co
Indiana Harbor
Coke Co
Indiana Harbor
Coke Co
Indiana Harbor
Coke Co
Location
Vansant, VA
Vansant, VA
Vansant, VA
Vansant, VA
East Chicago, IN
East Chicago, IN
East Chicago, IN
East Chicago, IN
Producer Coke
Type3 Typeb
M 1
M 1
M 1
M 1
M 1
M 1
M 1
M 1
Capacity
(short
tons/yr)
197,000
164,000
124,000
164,000
325,000
325,000
325,000
325,000
Start/
Rebuild
Date
1966
1983
1989
1990
1998
1998
1998
1998
AVCC
($/short
ton)
$58.59
$59.31
$59.98
$59.31
$62.36
$62.36
$62.36
$62.36
MACT
($/short
ton)
$0.00
$0.00
$0.00
$0.00
$0.00
$0.00
$0.00
$0.00
AFC
($/short
ton)
$9.90
$10.38
$10.85
$10.38
$10.52
$10.52
$10.52
$10.52
ATC
($/short
ton)
$68.49
$69.69
$70.83
$69.69
$72.88
$72.88
$72.88
$72.88
aC = Captive; M = Merchant.
bl = Furnace; 2 = Foundry; 3 = Both.
Includes by-/co-product credits.

-------
                                   APPENDIX C

      ECONOMETRIC ESTIMATION OF THE DEMAND ELASTICITY FOR
                                 IRON CASTINGS
       In this appendix, we summarize the econometric procedure used to estimate demand
elasticities and present demand elasticity estimates for iron and steel castings.  Elasticity
estimates are based on national-level annual sales and price data. In addition, individual
demand elasticity estimates are developed for three subcategories of iron castings:

       •  gray iron castings,
       •  ductile iron castings, and
       •  malleable iron castings.
C.I    Econometric Model

       A partial equilibrium market supply/demand model is used to simulate the interaction
of producers and consumers in the iron and steel casting markets.  The model consists of a
system of interdependent equations in which the price and output of a product are
simultaneously determined. This class of model is referred to as a simultaneous equation
model.

       In simultaneous equation models, where variables in  one equation feed back into
variables in another equation, the error terms in each equation are correlated with the
endogenous variables (price and output). In this case, single-equation ordinary least squares
(OLS) estimation of individual equations will lead to biased  and inconsistent parameter
estimates.

       We therefore use a two-stage least squares (2SLS) approach to correct for the
correlation between the error term and the endogenous variables. The 2SLS approach
requires that each equation be  identified through the inclusion of exogenous variables to
control for shifts in the supply and demand curves over time.

       Exogenous variables influencing the demand for iron and steel castings include
measures of economic activity such as U.S. gross domestic production, the number of motor
                                        C-l

-------
vehicle sales, and the price of substitute products such as plastics, nonferrous castings and
forgings, and steel mill products (typically proxied by the appropriate producer price
indices). Exogenous variables influencing the level of supply include measures of the
change in the costs of iron and steel castings production caused by changes in prices of key
inputs such as raw materials, fuel, and labor (typically proxied by the producer price index
for iron ore, coke, fuel, and electricity as well as the average hourly earnings for the
industry's production workers).

       The supply/demand system for a particular iron or steel casting over time (t) is
defined as follows:

                                    Qtd = f(Pt,Zt) + ut                               (C.I)

                                    Qts = g(Pt,Wt) + vt                               (C.2)

                                        Qtd = Qts                                   (C.3)
Eq. (C.I) represents  quantity demanded,  Qtd in year t as a function of price, Pt, and other
demand factors, Zt(e.g., measures of economic activity and prices of substitute products),
and an error term, ut. Eq. (C.2) represents quantity supplied, Qts, in year t as a function of
price and other supply factors, Wt (e.g., wage rate and other input prices), and an error term,
vt. Eq. (C.3) specifies the equilibrium condition, where quantity supplied equals quantity
demanded in year t.  Eq.  (C.3) creates a system of three equations in three variables.  Solving
the system generates equilibrium values for the variables Pt* and Qt*=Qtd*=Qts*.

       We use a 2SLS regression procedure to estimate the parameters and obtain the
demand elasticities.1  In the first stage of the 2SLS procedure,  the observed price is regressed
against the supply and demand "shifter" variables that are exogenous to the system.  The first
stage produces fitted (or  imputed) values for the price variable that are, by definition, highly
correlated with the true endogenous variable (the observed price) and uncorrelated with the
error term. In the second stage, these fitted values are then employed as explanatory
variables of the right-hand side in the demand function. The imputed value is uncorrelated
with the error term by construction and thus does not incur the endogeneity bias.
'The 2SLS approach was selected over the three-stage least squares (3SLS) approach because of the limited
   number of observations available for the regression analysis. The 3SLS approach requires more degrees of
   freedom for the estimation procedure.

                                          C-2

-------
       The logarithm of the quantity demanded is modeled as a linear function of the
logarithm of the commodity price. This specification enables us to interpret the price
variable coefficient as a constant elasticity of demand.

C.2    Econometric Results

       Demand elasticities are estimated based on commodity data from the U.S.
Department of Commerce, U.S. Bureau of Labor Statistics, and  other government sources.
The average prices for iron and steel commodities are calculated based on value of shipments
data from 1987 through 1997.  Prior to estimating demand elasticities, all prices are deflated
by the gross domestic product (GDP) implicit price deflator to reflect real rather than
nominal prices.

       Table C-l provides demand elasticity estimates for iron and steel castings. The
coefficients on the price variables, In (price), are the estimates of the demand elasticity.
Demand elasticity reflects how responsive consumers are to changes in the price of a
product.  For normal goods, consumption decreases as price increases, and this negative
relationship is shown by a negative price variable coefficient. As economic theory predicts,
our estimated coefficients on the price variables are negative.

       As shown in Table C-l, all of the individual elasticity estimates are inelastic,
implying that a 1 percent increase in price results in a less than 1 percent decrease in
consumption. Individual demand elasticity estimates for the iron casting subcategories range
from -0.41 for malleable iron castings to -0.67 for gray iron castings.  As shown in
Table C-l, the econometrically determined demand elasticity for all iron castings  was -0.58.
Similarly, the demand elasticity for steel castings was -0.59. Both estimates are significant
at the 95 percent or higher confidence level.
                                         C-3

-------
      Table C-l.  Two Stage Least Squares Regression Estimation of Iron and Steel Castings Demand Equations
n
Dependent Variables
Iron Castings
Independent Variables
Constant

In(price)

In(gdpd)

In(motor)

ln(PPI_plast_parts_trans)

ln(PPI_nonferr_forge)

ln(PPI_nonferr_foundry)

ln(PPI_plast_parts_mfg)

ln(pipe_price)a

R-Squared
Adjusted R-Squared
F Value
Observations
Degrees of Freedom
Steel Castings
-37.35
(-1.59)
-0.59
(-2.26)**
2.75
(1.82)
—

2.18
(1.00)
—

2.92
(2.09)*
—

—

0.68
0.52
4.23**
13
4
Gray Iron
0.81
(0.43)
-0.67
(-2.80)**
—

0.91
(9.97)***
0.09
(0.26)
0.50
(1.37)
—

—

0.16
(0.76)
0.97
0.94
33.90***
12
5
Ductile Iron
0.82
(0.20)
-0.42
(-1.89)*
—

1.01
(4.62)***
—

—

1.83
(1.88)*
-0.90
(-1.22)
-0.57
(-0.95)
0.92
0.87
17.08***
13
5
Malleable Iron
-3.12
(-1.04)
-0.41
(-1.51)
—

0.61
(3.79)***
—

0.04
(0.07)
—

1.07
(3.48)***
0.14
(0.41)
0.89
0.81
11.49***
13
5
All Iron
^2.90
(-8.15)***
-0.58
(-2.52)**
5.17
(11.10)***
—

—

-2.57
(-6.33)***
—

4.58
(7.97)***
0.23
(0.95)
0.97
0.94
38.46***
12
5
      Note: T-statistics of parameter estimates are in parentheses. The F test analyzes the usefulness of the model. Asterisks indicate significance levels for these tests as
      follows:  * = 90%, ** = 95%, *** = 99%
      Trice of corresponding casting.
      Variable Descriptions:
      In(gdp)                  real gross domestic product
      In(motor)                U.S. motor vehicle production
      ln(PPI_plast_parts_trans)   real producer price index for
ln(PPI_nonferr_forge)
ln(PPI_plast_parts_mfg)
ln(pipe_price)
real producer price index for nonferrous metal forge shop products
real producer price index for parts and components for manufacturing
real producer of steel mill pipe and tube products
                              plastic parts for transportation    ln(PPI_nonferr_foundry)    real producer price index for nonferrous foundry shop products

-------
                                     TECHNICAL REPORT DATA
                                (Please read Instructions on reverse before completing)
  i. REPORT NO.
   EPA-452/R-03-012
                                                                     3. RECIPIENT'S ACCESSION NO.
 4. TITLE AND SUBTITLE
 Economic Impact Analysis of Final Iron and Steel Foundries
 NESHAP
                                                                     5. REPORT DATE
                                                                       August 2003
                  6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)
 Michael P. Gallaher, Brooks M. Depro, and Laurel Clayton, RTI
 International
                  8. PERFORMING ORGANIZATION REPORT NO.
                  RTI Project Number 7647-004-390
 9. PERFORMING ORGANIZATION NAME AND ADDRESS
                                                                     10. PROGRAM ELEMENT NO.
 RTI International
 Center for Regulatory Economics and Policy Research, Hobbs Bldg.
 Research Triangle Park, NC 27709
                  11. CONTRACT/GRANT NO.
                  68-D-99-024
  12. SPONSORING AGENCY NAME AND ADDRESS
                                                                     13. TYPE OF REPORT AND PERIOD COVERED
                                                                     Final
   Office of Air Quality Planning and Standards
   Office of Air and Radiation
   U.S. Environmental Protection Agency
   Research Triangle Park, NC 27711	
                  14. SPONSORING AGENCY CODE
                  EPA/200/04
  15. SUPPLEMENTARY NOTES
  16. ABSTRACT
 This report evaluates the economic impacts of the final NESHAP for melting furnace; scrap preheating;
 pouring, cooling and shakeout (PCS); mold and core coating; and mold and core making operations at iron
 foundries. The social costs of the rule are estimated by incorporating the expected costs of compliance to a
 partial equilibrium model and projecting the market impacts for iron and steel castings and related products.
 The report also provides the screening analysis for small business impacts.
  17.
                                       KEY WORDS AND DOCUMENT ANALYSIS
                    DESCRIPTORS
                                                   b. IDENTIFIERS/OPEN ENDED TERMS
                                                                                        c. COSATI Field/Group
 economic impacts
 small business impacts
 social costs
Air Pollution Control
Economic Impact Analysis
Regulatory Flexibility Analysis
  18. DISTRIBUTION STATEMENT
   Release Unlimited
                                                   19. SECURITY CLASS (Report)
                                                     Unclassified
                                     21. NO. OF PAGES
                                     139
                                                   20. SECURITY CLASS (Page)
                                                     Unclassified
                                                                                        22. PRICE
EPA Form 2220-1 (Rev. 4-77)   PREVIOUS EDITION IS OBSOLETE

-------
United States                                Office of Air Quality Planning and Standards                          Publication No. EPA-452/R-03-012
Environmental Protection                     Air Quality Strategies and Standards Division                         August 2003
Agency                                     Research Triangle Park, NC

-------