Section 112 requires us to establish standards that are no less stringent than a certain minimum baseline, we refer to this as the ``MACT floor.'' For new sources, the standards for a source category or subcategory cannot be less stringent than the emission control that is achieved in practice by the bestcontrolled similar source. The standards for existing sources can be less stringent than the standards for new sources, but they cannot be less stringent than the average emission limitation achieved by the bestperforming 12 percent of existing sources (excluding certain sources) for categories and subcategories with 30 or more sources. For categories and subcategories with fewer than 30 sources, the standards cannot be less stringent than the average emission limitation achieved by the bestperforming five sources.

We may take alternative approaches to establishing the MACT floor, depending on the type, quality, and applicability of available data. The three approaches most commonly used involve reliance on State regulations or permit limits, source test data that characterize actual emissions, and use of a technology floor with an accompanying demonstrated achievable emission level that accounts for process and/or air pollution control device variability.

Section 112(d) allows us to distinguish among classes, types, and sizes of sources within a category or subcategory. For example, we can establish classes of sources within a category or subcategory based on size and establish a different emission standard for each class, provided both standards are at least as stringent as the MACT floor for that class of sources.

We evaluate several alternatives (which may be different levels of emission control or different levels of applicability or both) to select the one that best reflects the appropriate MACT level. The selected alternative may be more stringent than the MACT floor, but the control level selected must be technically achievable. In selecting an alternative, we consider the achievable HAP emission reduction (and possibly other pollutants that are cocontrolled), cost and economic impacts, energy impacts, and other environmental impacts. The objective is to achieve the maximum degree of emission reduction without unreasonable economic or other impacts. The regulatory alternatives selected for new and existing sources may be different because of different MACT floors, and separate regulatory decisions may be made for new and existing sources.

We then translate the selected regulatory alternative into a proposed rule. The public is invited to comment on the proposal during the public comment period. Based on an evaluation of these comments, we reach a final decision and promulgate the standards.

C. What Source Category Is Affected by This Proposed Rule?

Section 112(c) of the CAA requires us to list all categories of major and area
[[Page 36838]]
sources of HAP for which we will develop national emission standards. We published the initial list of source categories on July 16, 1992 (57 FR 31576). ``Integrated Iron and Steel Manufacturing'' is one the source categories on the initial list. The listing was based on our determination that integrated iron and steel manufacturing facilities may reasonably be anticipated to emit a variety of HAP listed in section 112(b) in quantities sufficient to be major sources.

An integrated iron and steel manufacturing facility produces steel from iron ore. The integrated iron and steel manufacturing source category includes sinter production, iron production, and steel production.
D. What Processes Are Used at Integrated Iron and Steel Manufacturing Facilities?

The primary processes of interest because of their potential to generate HAP emissions include sinter plants, blast furnaces that produce iron, and basic oxygen process furnaces (BOPF) that produce steel. There are also several ancillary processes, including hot metal transfer, desulfurization, slag skimming, and ladle metallurgy. Iron and steel are produced at 20 plant sites in the United States (U.S.) that have a total of 39 blast furnaces, 50 BOPF, and 9 sinter plants. Integrated iron and steel plants are located in ten States; however, the majority of the iron and steel is produced in Indiana, Ohio, and Illinois.

The sintering process converts finesized raw materials, including iron ore, coke breeze, limestone, mill scale, and flue dust, into an agglomerated product (sinter) of suitable size for charging into the blast furnace. The raw materials are mixed with water to provide a cohesive matrix and then placed on a continuous traveling grate called the sinter strand. A burner hood at the beginning of the sinter strand ignites the coke in the mixture, after which the combustion is self supporting and provides sufficient heat (2,400 to 2,700 degrees Fahrenheit) to cause surface melting and agglomeration of the mix. On the underside of the sinter strand are a series of windboxes that draw combusted air down through the material bed into a common duct leading to a gas cleaning device (either a venturi scrubber or a baghouse).

The fused sinter is discharged at the end of the sinter strand where it is crushed and screened. Undersize sinter is recycled to the mixing mill and back to the strand. The remaining sinter product is cooled in open air or in a circular cooler with mechanical fans. The cooled sinter is crushed and screened for a final time, then the fines are recycled and the product is sent to be charged to the blast furnace. Generally, 2.5 tons of raw materials, including water and fuel, are required to produce 1 ton of product sinter.

Iron is produced in blast furnaces by the reduction of iron bearing materials with a hot gas. The large, refractory lined furnace is charged through its top with iron ore, iron ore pellets, sinter, flux (limestone and dolomite), and coke, which provides fuel and forms a reducing atmosphere in the furnace. Many modern blast furnaces also inject pulverized coal to reduce the quantity of coke required. Iron oxides, coke, coal, and fluxes react with the heated blast air injected into the bottom of the furnace to form molten reduced iron, carbon monoxide (CO), and slag. The molten iron and slag collect in the hearth at the base of the furnace. The byproduct gas is collected through offtakes located at the top of the furnace and is recovered for use as fuel.

The molten iron and slag are removed, or cast, from the furnace periodically. The casting process begins with drilling a hole, called the taphole, into the clayfilled iron notch at the base of the hearth. During casting, molten iron flows into runners that lead to transport ladles. Slag also flows from the furnace and is directed through separate runners to a slag pit adjacent to the casthouse, or into slag pots for transport to a remote slag pit. At the conclusion of the cast, the taphole is replugged with clay. The area around the base of the furnace, including all iron and slag runners, is enclosed by a casthouse.

The blast furnace byproduct gas, which is collected from the furnace top, contains CO and particulate matter (PM). As a fuel, the blast furnace gas has a low heating value, about 75 to 90 British thermal units per cubic foot (Btu/ft³). Before it can be efficiently burned, the PM must be removed from the gas. Initially, the gases pass through a settling chamber or dry cyclone to remove about 60 percent of the particulate. Next, the gases undergo a one or two stage cleaning operation. The primary cleaner is normally a wet scrubber, which removes about 90 percent of the remaining particulate. The secondary cleaner is a highenergy wet scrubber (usually a venturi) which removes up to 90 percent of the particulate that eludes the primary cleaner. Together, these control devices provide a clean fuel with less than 0.02 grains per dry standard cubic foot (gr/dscf) of PM. A portion of this gas is fired in the blast furnace stoves to preheat the blast air, and the rest is used in other plant operations.

After the molten iron (called ``hot metal'') is produced in the blast furnace, it is transferred to the BOPF shop. Bricklined torpedo cars are used because their insulating properties lower heat loss from the iron. Hot metal transfer occurs when the molten iron is transferred (``reladled'') from the torpedo car to the BOPF shop ladle.

Hot metal is desulfurized by adding various reagents such as soda ash, lime, and magnesium. The reagents are usually injected pneumatically with either dry air or nitrogen. Following
desulfurization, any slag formed is skimmed from the ladle and the hot metal is transferred to a BOPF.

In the BOPF, molten iron from a blast furnace and iron scrap are refined by lancing (or injecting) highpurity oxygen. The input material is typically 70 percent hot metal and 30 percent scrap metal. The oxygen reacts with carbon and other impurities to remove them from the metal. Because the reactions are exothermic, no external heat source is necessary to melt the scrap and to raise the temperature of the metal to the desired range for tapping. For a BOPF, tapping begins when the furnace is tilted to remove steel and slag and ends when the furnace returns to an upright position. The large quantities of CO produced by the reactions in the BOPF can be controlled by combustion at the mouth of the furnace and then vented to gas cleaning devices, as with open hoods, or combustion can be suppressed at the furnace mouth, as with closed hoods.

The BOPF is a large (up to 400ton capacity) refractory lined pear shaped furnace. There are two major variations of the process. Conventional BOPF have oxygen blown into the top of the furnace through a watercooled lance (topblown). In the newer Quelle Basic Oxygen process (QBOP), oxygen is injected through tuyeres located in the bottom of the furnace (bottomblown). A typical BOPF cycle consists of the scrap charge, hot metal charge, oxygen blow (refining) period, testing for temperature and chemical composition of the steel, alloy additions and reblows (if necessary), tapping, and slagging. The full furnace cycle typically ranges from 25 to 45 minutes.

Ladle metallurgy is a secondary step of the steelmaking process performed in a ladle after the initial refining process in the BOPF is completed. The purpose of ladle metallurgy (also referred to as secondary steelmaking) is to produce steel that satisfies the many stringent requirements associated with surface and internal quality as well as
[[Page 36839]]
mechanical properties. Nearly all of the integrated iron and steel facilities perform some type of ladle metallurgy, such as vacuum degassing, ladle refining, reheating, alloy addition, argon/oxygen decarburization, argon stirring, and lance powder injection.

After the steel has been refined in the BOPF and ladle metallurgy operations, the molten metal is transferred to a continuous casting operation where it is cast and subsequently rolled into a semifinished product, such as a bloom, billet, or slab.
E. What HAP Are Emitted and How Are They Controlled?

1. Sinter Plants

The primary source of HAP emissions from sinter plants (over 40 percent) is the windbox exhaust. The windbox exhaust is a high volume stream of hot gases on the order of 300,000 to 600,000 dscfm. Control devices applied include baghouses and venturi scrubbers. The HAP emissions include HAP metal compounds, primarily lead and manganese, which comprise about 3 percent of the total PM. Organic HAP compounds, including both volatile and semivolatile HAP such as polycyclic organic matter, are also emitted. The organic compounds are formed from oily materials, mostly rolling mill scale, that are used in the sinter feed. Most plants minimize emissions of organic compounds by carefully monitoring and limiting the quantity of oil introduced with the sinter feed.

The discharge end emission points include the crusher, hot screen and various transfer points as the hot sinter is conveyed to the cooler. The sinter cooler stack is also an emission point. These operations are a source of PM emissions from the dusty sinter product and account for only 7 percent of the HAP emissions from the sinter plant. The most significant HAP found in emissions from the discharge and sinter cooler is manganese, which comprises only about 0.75 percent of the PM.

2. Blast Furnace Casthouse

The primary source of blast furnace emissions is the casting operation. Particulate emissions are generated when the molten iron and slag contact air above their surface. Casting emissions are also generated by drilling and plugging the taphole. The occasional use of an oxygen lance to open a clogged taphole can increase emissions. During the casting operation, iron oxides, magnesium oxide and carbonaceous compounds are generated as PM. The only significant HAP found in the PM is manganese, which comprises about 0.6 percent of the PM.

Casting emissions are controlled by evacuation through capture hoods to a baghouse or by suppression techniques. The basic concept of suppression techniques that use steam or inert gas is to prevent the formation of pollutants by preventing ambient air from contacting the molten surfaces. Newer furnaces have been constructed with evacuated covered runners and local hooding ducted to a baghouse.

3. Hot Metal Transfer, Desulfurization, and Slag Skimming

Hot metal transfer from the torpedo car into the BOPF shop ladle is accompanied by the emissions of kish, a mixture of fine iron oxide particles together with larger graphite particles. The reladling generally takes place under a hood to capture these emissions. Emissions during desulfurization are created by both the reaction of the reagents injected into the metal and the turbulence during injection. The pollutants emitted are mostly iron oxides, calcium oxides, and oxides of the compound injected. The sulfur reacts with the reagents and is skimmed off as slag.

The emissions generated from desulfurization and slag skimming are usually collected by a hood positioned over the ladle and vented to a baghouse. Many plants perform hot metal transfer, desulfurization, and slag skimming at the same station to take advantage of a single capture and control system. Manganese is the predominant HAP in the PM emissions. The level of manganese is expected to be comparable to that of PM from the casthouse (on the order of 0.6 percent).

4. Basic Oxygen Process Furnace

Emissions from the BOPF occur during charging, the oxygen blow and tapping. Fugitive emissions escape through the BOPF shop roof monitor, and stack emissions are released through primary and secondary control systems. The predominant compounds emitted are iron oxides, and the most significant HAP is manganese. Manganese comprises about 1 percent of the particulate, which is more than all of the other HAP metals combined.

Emissions during oxygen blow periods are controlled using a primary hood capture system located directly over the open mouth of the furnaces. Two types of capture systems are used to collect exhaust gas as it leaves the furnace mouth: a closed hood design that suppresses combustion, and an open hood design that promotes combustion. A closed hood fits snugly against the furnace mouth, ducting all PM and CO to a venturi scrubber. The CO is flared at the scrubber outlet stack. The open hood design allows combustion air to be drawn into the hood, thus burning the CO. Electrostatic precipitators (ESP) and venturi scrubbers are used as the primary controls for open hood BOPF.

Charging and tapping emissions are controlled by a variety of evacuation systems and operating practices. Charging hoods, tapside enclosures, and full furnace enclosures are used to capture these emissions and send them either to the primary control device or to a secondary device, usually a baghouse. Almost all closed hood BOPF have a secondary capture and control system, whereas many open hood BOPF rely on the primary system for capture and control of fugitive emissions.

5. Ladle Metallurgy

Most BOPF shops have a ladle metallurgy station where various adjustments are made to the steel's physical and chemical properties. Almost all ladle metallurgy stations are enclosed or hooded, and any fume from the vessel is ducted to a baghouse. There are few data on the HAP composition of ladle metallurgy emissions; however, the composition should be similar to that of emissions from the BOPF (primarily manganese).
F. What Are the Health Effects Associated With Emissions From Integrated Iron and Steel Manufacturing Processes?

There are a variety of metal HAP contained in the PM emitted from iron and steel manufacturing processes. These include primarily manganese and lead, with much smaller quantities of antimony, arsenic, beryllium, cadmium, chromium, cobalt, mercury, nickel, and selenium. Organic HAP compounds are released in trace amounts from the sinter plant windbox exhaust and include polycyclic organic matter (such as polynuclear aromatic hydrocarbons and chlorinated dibenzodioxins and furans), and volatile organics such as benzene, carbon disulfide, toluene, and xylene. These HAP are associated with a variety of adverse health effects including chronic and acute disorders of the blood, heart, kidneys, reproductive system, and central nervous system.

Manganese and lead comprise the majority of the metal HAP emissions. Health effects in humans have been associated with both deficiencies and excess intakes of manganese. Chronic exposure to low levels of manganese in the diet is considered to be nutritionally essential in humans, with a
[[Page 36840]]
recommended daily allowance of 2 to 5 milligrams per day. Chronic exposure to high levels of manganese by inhalation in humans results primarily in central nervous system (CNS) effects. Visual reaction time, hand steadiness, and eyehand coordination were affected in chronicallyexposed workers. Manganism, characterized by feelings of weakness and lethargy, tremors, a masklike face, and psychological disturbances, may result from chronic exposure to higher levels. Impotence and loss of libido have been noted in male workers afflicted with manganism attributed to inhalation exposures. We have classified manganese in Group D, not classifiable as to carcinogenicity in humans.

Lead is a very toxic element, causing a variety of effects at low dose levels. Brain damage, kidney damage, and gastrointestinal distress may occur from acute exposure to high levels of lead in humans. Chronic exposure to lead in humans results in effects on the blood, CNS, blood pressure, and kidneys. Children are particularly sensitive to the chronic effects of lead, with slowed cognitive development, reduced growth and other effects reported. Reproductive effects, such as decreased sperm count in men and spontaneous abortions in women, have been associated with lead exposure. The developing fetus is at particular risk from maternal lead exposure, with low birth weight and slowed postnatal neurobehavioral development noted. Human studies are inconclusive regarding lead exposure and cancer, while animal studies have reported an increase in kidney cancer from lead exposure by the oral route. We have classified lead as a Group B2, probable human carcinogen.

Trace quantities of organic HAP, such as chlorinated dibenzodioxins and furans (CDD/F) and benzene, have been detected in the windbox exhaust at sinter plants. One CDD/F compound, 2,3,7,8
tetrachlorodibenzopdioxin (2,3,7,8TCDD, commonly called ``dioxin'') is listed singly as a HAP. Other CDD/F compounds, many of which cause adverse health effects in the same way as dioxin, are HAP under the definition of polycyclic organic matter. Exposure to CDD/F mixtures causes chloracne, a severe acnelike condition, and has been shown to be extremely toxic in animal studies. Dioxin itself is known to be a developmental toxicant in animals, causing skeletal deformities, kidney defects, and weakened immune responses in the offspring of animals exposed during pregnancy. Human studies have shown an association between dioxin and softtissue sarcomas, lymphomas, and stomach carcinomas. We have classified dioxin as a probable human carcinogen (Group B2).

Acute inhalation exposure of humans to benzene may cause drowsiness, dizziness, headaches, as well as eye, skin, and respiratory tract irritation, and, at high levels, unconsciousness. Chronic inhalation exposure has caused various disorders in the blood, including reduced numbers of red blood cells and aplastic anemia, in occupational settings. Reproductive effects have been reported for women exposed by inhalation to high levels, and adverse effects on the developing fetus have been observed in animal tests. Increased incidence of leukemia (cancer of the tissues that form white blood cells) has been observed in humans occupationally exposed to benzene. We have classified benzene as a Group A, known human carcinogen.

In addition to HAP, the proposed rule also would reduce PM emissions, which are controlled under national ambient air quality standards. Briefly, emissions of PM have been associated with aggravation of existing respiratory and cardiovascular disease and increased risk of premature death.

We recognize that the degree of adverse effects to health experienced by exposed individuals can range from mild to severe. The extent and degree to which the health effects may be experienced depends on:

A. What Are the Affected Sources and Emission Points?

The affected sources are each new and existing sinter plant, blast furnace, and BOPF shop at an integrated iron and steel manufacturing facility that is a major source. A new affected source is one constructed or reconstructed after July 13, 2001. An existing affected source is one constructed or reconstructed on or before July 13, 2001. The proposed rule covers emissions from the sinter plant windbox exhaust, discharge end, and sinter cooler; the blast furnace casthouse; and the BOPF, BOPF shop roof monitor, and BOPF ancillary operations (hot metal transfer, hot metal desulfurization, slag skimming, and ladle metallurgy).

B. What Are the Emission Limitations?

The proposed rule includes PM emission limits and opacity limits as well as operating limits for capture systems and control devices. Particulate matter and opacity serve as a surrogate measures of HAP emissions.

1. Sinter Plants

The proposed PM emission limit for the windbox exhaust stream, 0.3 pounds per ton (lb/ton) of product sinter, is the same for existing and new sinter plants. The proposed rule limits PM emissions from a discharge end to 0.02 gr/dscf for an existing plant and 0.01 gr/dscf at a new plant. The discharge end PM limit is a flowweighted average for one or more control devices that operate in parallel. A 20 percent opacity limit is proposed for secondary emissions from a discharge end at an existing sinter plant; a 10 percent opacity limit is proposed for a new sinter plant. The proposed PM emission limits for sinter cooler stacks are 0.03 gr/dscf for an existing plant and 0.01 gr/dscf for a new plant.

2. Blast Furnaces

The proposed PM emission limit for a control device applied to emissions from a casthouse is 0.009 gr/dscf for the casthouse at a new or existing blast furnace. The proposed opacity limits are 20 percent for a casthouse at an existing blast furnace and 15 percent for a casthouse at a new blast furnace (both 6minute averages).

3. Basic Oxygen Process Furnaces

For primary emissions from BOPF, we are proposing different PM emission limits based on hood system (closed or open). For BOPF with closed hood systems, we are proposing a PM emission limit of 0.024 gr/ dscf which would apply only during periods of primary oxygen blow. For BOPF with open hood systems, we are proposing a PM emission limit of 0.019 gr/dscf which would apply during all periods of the steel production cycle. The primary oxygen blow is the period in which oxygen is initially blown into the furnace and does not include any subsequent reblows. The steel production cycle begins when the furnace is first charged with either scrap or hot metal and ends 3 minutes after slag is removed. The PM emission limits are the same for BOPF at new and [[Page 36841]]
existing BOPF shops. The proposed PM emission limits for a control device applied solely to secondary emissions from a BOPF are 0.01 gr/ dscf for an existing BOPF shop and 0.0052 gr/dscf for a new BOPF shop. Secondary emissions are those not controlled by the primary emission control system, including emissions that escape from open and closed hoods and openings in the ductwork to the primary control system.

For the BOPF shop, a PM emission limit of 0.007 gr/dscf is proposed for a control device applied to emissions from ancillary operations (hot metal transfer, skimming, desulfurization, or ladle metallurgy) at a new or existing BOPF shop. For the BOPF roof monitor, a 20 percent opacity limit is proposed for secondary emissions from the BOPF or BOPF shop operations in an existing BOPF shop. This opacity limit is based on 3minute averages. For a new BOPF shop housing a bottomblown furnace, a 10 percent opacity limit is proposed (6minute average) except that one 6minute period not to exceed 20 percent may occur once during each steel production cycle. For a new BOPF shop housing a top blown furnace, a 10 percent opacity limit is proposed (3minute average) except that one 3minute period greater than 10 percent but less than 20 percent may occur once during each steel production cycle.

For capture systems applied to emissions from a sinter plant discharge end or blast furnace casthouse, the proposed rule provides two options: maintain the hourly average volumetric flow rate through each separately ducted hood at or above the level established during the performance test, or maintain the total hourly average volumetric flow rate at the control device inlet at or above the level established during the performance test with all capture system dampers in the same positions as during the performance test.

The same options are available in the operating limits proposed for capture systems applied to secondary emissions from a BOPF. However, the averaging period is the steel production cycle rather than each 1 hour period.

The proposed operating limit for baghouses requires that the bag leak detection system alarm not sound for more than 5 percent of the total operating time in a semiannual reporting period. For a venturi scrubber, the hourly average pressure drop and scrubber water flow rate must remain at or above the level established during the initial performance test. For an ESP, the hourly average opacity must remain at or below the level established during the initial performance test. The proposed rule requires plants to submit information on monitoring parameters if another type of control device is used.

The proposed rule also requires sinter plants to maintain the oil content of the feedstock at or below 0.025 percent. This limit is based on a 30day rolling average.

C. What Are the Operation and Maintenance Requirements?

All plants subject to the proposed rule would be required to prepare and implement a written startup, shutdown, and malfunction plan according to the requirements in Sec. 63.6(e) of the NESHAP General Provisions. A written operation and maintenance plan is also required for capture systems and control devices subject to an operating limit. This plan must describe procedures for monthly inspections of capture systems, preventative maintenance requirements for control devices, and corrective action requirements for baghouses. In the event of a bag leak detection system alarm, the plan must include specific requirements for initiating corrective action to determine the cause of the problem within 1 hour, initiating corrective action to fix the problem within 24 hours, and completing all corrective actions needed to fix the problem as soon as practicable.

D. What Are the Initial Compliance Requirements?

The proposed rule requires performance tests to demonstrate that each affected source meets all applicable emission and opacity limits. The PM concentration would be measured using EPA Method 5, 5D, or 17 in 40 CFR part 60, appendix A. The proposed rule also allows plants to use a method developed by the American Society for Testing and Materials (ASTM), Standard Test Method for HighVolume Sampling for Solid Particulate Matter and Determination of Particulate Emissions (ASTM D453696). Plants may use this method instead of the sampling equipment and procedures required by EPA Method 5 or 17 when testing a positive pressure baghouse, but must use the sample traverse location and number of sampling locations required by EPA Method 5D. The EPA Method 9 in 40 CFR part 60, appendix A, is proposed for determining the opacity of emissions, with special instructions for computing 3minute averages. The proposed testing requirements also include procedures for establishing sitespecific operating limits for capture systems and control devices and for revising the limits, if needed, after the performance test.

The proposed rule also requires a performance test to demonstrate initial compliance with the operating limit for the oil content of the sinter plant feedstock. This test would require measurements of the oil content using EPA Method 9071B (Revision 2, April 1998) for 30 consecutive days and computing the 30day rolling average. To demonstrate initial compliance with the proposed operation and maintenance requirements, plants would certify in their notification of compliance status that they have prepared the written plans and will operate capture systems and control devices according to the procedures in the plan.

E. What Are the Continuous Compliance Requirements?

The proposed rule would require plants to conduct performance tests at least twice during each title V operating permit term (at midterm and renewal) to demonstrate continuous compliance with the emission and opacity limits. Plants also would be required to monitor operating parameters for capture systems and control devices subject to operating limits and carry out the procedures in their operation and maintenance plan.

For capture systems, a continuous parameter monitoring system (CPMS) is required to measure and record the volumetric flow rate through each separately ducted hood or the total volumetric flow rate at the control device inlet. Plants electing to monitor the total volumetric flow rate also must check the capture system dampers at least once a day (every 24 hours) to verify that all dampers are in the same position as during the initial performance test. To demonstrate continuous compliance, plants must keep records documenting compliance with the rule requirements for monitoring, the operation and maintenance plan, and installation, operation, and maintenance of CPMS.

For baghouses, plants would be required to monitor the relative change in PM loading using a bag leak detection system and make inspections at specified intervals. The bag leak detection system must be installed and operated according to the EPA guidance document ``Fabric Filter Bag Leak Detection Guidance,'' EPA 454/R98015, September 1997. The document is available on the TTN at
http:www.epa.gov/ ttnemc01/cem/tribo.pdf. If the system does not work based on the triboelectric effect, it must be installed and operated consistent
[[Page 36842]]
with the manufacturer's written specifications and recommendations. The basic inspection requirements include daily, weekly, monthly, or quarterly inspections of specified parameters or mechanisms with monitoring of bag cleaning cycles by an appropriate method.

To demonstrate continuous compliance, the proposed rule requires records of bag leak detection system alarms and records documenting conformance with the operation and maintenance plan, as well as the inspection and maintenance procedures.

For venturi scrubbers, plants would be required to use CPMS to measure and record the hourly average pressure drop and scrubber water flow rate. To demonstrate continuous compliance, plants would keep records documenting conformance with the monitoring requirements and the installation, operation, and maintenance requirements for CPMS.

For ESP, plants would be required to use a continuous opacity monitoring system (COMS) to measure and record the average hourly opacity of emissions exiting each stack of the control device. Plants must operate and maintain the COMS according to the requirements in Sec. 63.8 of the NESHAP General Provisions and Performance Specification 1 in 40 CFR part 60, appendix B. These requirements include a quality control program including a daily calibration drift assessment, quarterly performance audit, and annual zero alignment.

To demonstrate continuous compliance with the operating limit for the sinter plant feedstock, plants would be required to determine the oil content every 24 hours (from the composite of three samples taken at 8hour intervals) and compute and record the 30day rolling average oil content for each operating day.
F. What Are the Notification, Recordkeeping, and Reporting Requirements?

The proposed notification, recordkeeping, and reporting requirements rely on the NESHAP General Provisions in 40 CFR part 63, subpart A. Table 4 to proposed subpart FFFFF lists each of the requirements in the General Provisions (Secs. 63.2 through 63.15) with an indication of whether they do or do not apply.

The plant owner or operator would be required to submit each initial notification required in the NESHAP General Provisions that applies to their facility. These include an initial notification of applicability with general information about the facility and notifications of performance tests and compliance status.

Plants would be required to maintain the records required by the NESHAP General Provisions that are needed to document compliance, such as performance test results; copies of startup, shutdown, and malfunction plans and associated corrective action records; monitoring data; and inspection records. Except for the operation and maintenance plan for capture systems and control devices, all records must be kept for a total of 5 years, with the records from the most recent 2 years kept onsite. The proposed rule requires that the operation and maintenance plan for capture systems and control devices subject to an operating limit be kept onsite and available for inspection upon request for the life of the affected source or until the affected source is no longer subject to the rule requirements.

Semiannual reports are required for any deviation from an emission limitation, including an operating limit. Each report would be due no later than 30 days after the end of the reporting period. If no deviation occurred, only a summary report would be required. If a deviation did occur, more detailed information would be required.

An immediate report would be required if there were actions taken during a startup, shutdown, or malfunction that were not consistent with the startup, shutdown, and malfunction plan. Deviations that occur during a period of startup, shutdown, or malfunction are not violations if the owner or operator demonstrates to the authority with delegation for enforcement that the source was operating in accordance with the startup, shutdown, and malfunction plan.

G. What Are the Compliance Deadlines?

The owner or operator of an existing affected source would have to comply within [24 MONTHS OF PUBLICATION OF THE FINAL RULE IN THE Federal Register]. New or reconstructed sources that startup on or before the effective date of the final rule must comply by the effective date of the final rule. New or reconstructed sources that startup after the effective date of the final rule must comply upon initial startup.
III. Rationale for Selecting the Proposed Standards

A. How Did We Select the Affected Sources?

Affected source means the collection of equipment and processes in the source category or subcategory to which the emission limitations, work practice standards, and other regulatory requirements apply. The affected source may be the same collection of equipment and processes as the source category or it may be a subset of the source category. For each rule, we must decide which individual pieces of equipment and processes warrant separate standards in the context of the CAA section 112 requirements and the industry operating practices.

We considered three different approaches for designating the affected source: the entire integrated iron and steel manufacturing facility, groups of emission points, and individual emission points. In selecting the affected sources for regulation, we identified the HAP emitting operations, the HAP emitted, and the quantity of HAP emissions from the individual or groups of emissions points. We concluded that designating the group of emission points associated with each of the major processes as the affected source is the most appropriate approach. The major processes include sinter production in a sinter plant, iron production in a blast furnace, and steel production in a BOPF shop. Consequently, we selected the sinter plant, blast furnace, and BOPF shop as the affected sources. The proposed rule includes requirements for the control of emissions from the windbox exhaust, discharge end, and cooler at sinter plants; the blast furnace casthouse; the BOPF shop including both primary and secondary emissions from the furnace; and the ancillary operations in the BOPF shop (hot metal transfer, desulfurization, slag skimming, and ladle metallurgy). B. How Did We Select the Pollutants?

For the proposed rule, we decided that it is not practical to establish individual standards for each specific type of metallic HAP that could be present in the various processes (e.g., separate standards for manganese emissions, separate standards for lead emissions, and so forth for each of the metals listed as HAP and potentially could be present). When released, each of the metallic HAP compounds behave as PM. As a result, strong correlations exist between air emissions of PM and emissions of the individual metallic HAP compounds. The control technologies used for the control of PM emissions achieve comparable levels of performance on metallic HAP emissions. Therefore, standards requiring good control of PM will also achieve good control of metallic HAP emissions. Therefore, we decided to establish standards for total PM as a surrogate pollutant for the individual
[[Page 36843]]
types of metallic HAP. In addition, establishing separate standards for each individual type of metallic HAP would impose costly and significantly more complex compliance and monitoring requirements and achieve little, if any, HAP emissions reductions beyond what would be achieved using the surrogate pollutant approach based on total PM.

For stack discharges, we have traditionally relied on setting numerical emission limits, sometimes coupled with limits on opacity. In the case of fugitive emissions, we have traditionally relied on setting visible emission standards, typically expressed as opacity limits. C. How did we determine the bases and levels of the proposed standards?

There are nine sinter plants in the U.S.; however, only seven are currently operating. The windbox exhaust is controlled by a baghouse at four plants and by a venturi scrubber at five plants. Useful test data on actual emissions are available for six of the nine plants, two equipped with baghouses and four equipped with venturi scrubbers. In each case, the data reflect the results of performance tests comprised of the average of three test runs, expressed in terms of total PM.

An initial characterization of achievable performance based on concentration (gr/dscf) suggested that baghouses perform substantially better than do scrubbers. Concentration values recorded for the two baghouses are two to nearly four times lower than those recorded for the four scrubbers. Upon closer scrutiny, we determined that much of the difference in perceived performance is due to the fact that baghouses require the addition of relatively large quantities of ambient air to cool the hot windbox exhaust gases prior to control, whereas scrubbers do not. To correct for this difference, we transformed the test results into a pounds of PM emissions per ton of sinter format. The test results expressed in terms of the hourly mass rate were converted to annual emissions assuming 8,760 hours per operating year. The resultant annual emissions were then divided by a best estimate of annual sinter production for each plant (average for the 5year period from 1995 through 1999). The results range from 0.26 to 0.33 lb PM/ton of sinter. Averaging the results for the top five performers produces a MACT floor value of 0.29 lb PM/ton of sinter. Relying on the median value produces a MACT floor value of 0.30 lb/ton. Included among the top five performers are two baghouses and three venturi scrubbers, which indicates that both control devices are capable of achieving the MACT floor level of control as expressed in the lb/ton format.

The windbox exhaust gas can contain appreciable quantities of organic HAP, including both volatile and semivolatile compounds. There is strong evidence that demonstrates that the quantity of organic HAP emitted is directly related to the quantity and oil content of the mill scale component of the sinter feed. United States sinter plants limit organic emissions by carefully monitoring and limiting the oil content of the sinter feed. This pollution prevention control measure is an effective method for preventing, and thus reducing, emissions of organic HAP. Two plants in Indiana have performed testing to relate oil content with emissions of volatile organic compounds (VOC). The test results show a strong correlation between oil content and potential VOC emissions.

One of the organic pollutants of concern that has been related to oil content is a family of compounds called polychlorinated dibenzodioxins and furans (D/F). A 1994 paper \1\ identified sinter plants in Germany as one of the most important industrial sources of D/F. Tests showed an average concentration in the windbox exhaust of 47 nanograms (ng) expressed in toxic equivalency (TEQ)/per cubic meter (m\3\) and annual emissions of 122 grams (g) TEQ. The D/F emissions were attributed to high levels of oils and chlorinated organics in the waste materials recycled to the sinter plant.
\1\ Lahl, Uwe. Sintering Plants of Steel IndustryPCDD/F Emission Status and Perspective. In Chemosphere, vol. 29, nos. 911, pages 19391945. 1994.

We decided to perform testing at two representative facilities to characterize D/F emissions from U.S. sinter plants, one that uses a venturi scrubber as the windbox control device and one that uses a baghouse. The tests were performed in 1997 on the venturi scrubber in East Chicago, IN and on the baghouse in Youngstown, OH. These plants routinely monitor the oil content of their sinter feed, which averages 0.014 percent oil at the East Chicago, IN facility and 0.025 percent oil at the Youngstown, OH facility. The average D/F concentration from three 4hour runs at each plant ranged from 0.2 ng TEQ/m\3\ at the East Chicago, IN facility to 0.8 ng TEQ/m\3\ at the Youngstown, OH facility, both far below the levels reported for the German sinter plant. Assuming typical operation of each plant (310 days/yr), annual emissions would range from 0.7 to 2.8 g TEQ/yr, well below the levels indicated by the German data. Based upon emission factors derived from these test results, we estimate nationwide emissions from all U.S. sinter plants to be 26 g TEQ/yr, which corresponds to less than 1 percent of current estimates of the national inventory from all sources.

We surveyed the operators of all seven active sinter plants, as well as the two inactive plants, to obtain information on the oil content of their sinter feed. Four of the active plants provided data that ranged in magnitude from 976 samples collected over 1 year (sampling about three times per day) to 14 samples collected over 14 months (monthly sampling). All four plants carefully monitor their sinter feed for oil to minimize emissions of VOC. In addition, plants with baghouses are motivated to limit oil content due to concerns over blinding of bags and possible fire hazards. The other three active plants and the two inactive plants provided little data since none routinely monitor oil content. The four plants providing data reported longterm averages of 0.014, 0.02, 0.02, and 0.025 percent, respectively. We conclude that limiting substantially the oil content in the sinter feed represents the MACT floor for organic HAP in the windbox exhaust.

We know of no control devices besides venturi scrubbers and baghouses that can achieve better emissions reductions than that indicated by the level of performance selected as the MACT floor. As a result, we are selecting 0.3 lb/ton as the standard. We selected 0.3 lb/ton as opposed to either 0.29 or 0.30 lb/ton to provide a modest but warranted margin of safety given the relatively limited data available for this standard setting and the inherent uncertainty associated with the needed transformations of the test data from mass rate to mass per ton.

For the PM limit, we also considered setting alternative concentration limits that would be tailored to each type of control devicebaghouses and venturi scrubbers. Concentration limits (e.g., gr/dscf) have several advantages over a
lb/ton format when determining compliance. A lb/ton format requires that three measurements be made very accurately: The concentration of PM in the exhaust gas, the volumetric flow rate of exhaust gas, and the sinter production rate. Concentration is directly measured by EPA reference methods (such as Method 5), and there is no uncertainty introduced by additional measurements or calculations. The concentration limit is a direct and accurate measure of how [[Page 36844]]

well the emission control device is performing.

The two plants with baghouses averaged 0.007 and 0.009 gr/dscf when meeting the 0.3 lb/ton MACT floor level of control. Individual runs ranged from 0.004 to 0.01 gr/dscf. Considering the runtorun variability, we conclude that an appropriate alternative concentration limit for baghouses used for the control of windbox exhaust gases would be on the order of 0.01 gr/dscf. As noted previously, plants with baghouses introduce large volumes of tempering air to cool the windbox exhaust gas prior to entering the baghouse, whereas plants with venturi scrubbers do not. Consequently, a concentration limit for scrubbers, reflecting an equivalent level of control as baghouses, would of necessity be higher than one for baghouses. The four plants equipped with scrubbers recorded average concentration values of 0.017, 0.017, 0.025, and 0.026 gr/dscf when meeting the 0.3 lb/ton MACT floor level of control. Individual runs ranged from 0.014 to 0.029 gr/dscf. Since all four of these scrubbers represent MACT, an alternative concentration limit for scrubbers would be on the order of 0.03 gr/dscf considering runtorun variability. We request comments on both the appropriateness of setting concentration limits in addition or instead of a lb/ton limit and on the suggested values for these limits.

Relative to sinter feed oil content, we know of no control measures beyond this pollution prevention measure which would be more effective in limiting HAP organic emissions from sinter plant windboxes. Based on our review of the data obtained through our survey on oil content, we select a limit of 0.025 percent oil in sinter feed as representative of the MACT floor. Although 0.025 percent is the highest average value reported by the four plants, all of the averages are low, all are indicative of careful control of oil content, and for all intents and purposes are indistinguishable.

The sinter plant discharge end is comprised of sinter breakers (crushers), hot screens, conveyors, and transfer points that are designed to separate undersize sinter and to transfer the hot sinter to the cooler. In most cases, these discharge end operations are housed in a building. Emissions are usually controlled by local hooding and ventilation to one or more baghouses or wet scrubbers. Seven plants use baghouses and two plants use wet scrubbers.

Existing State regulations include both building opacity standards to limit releases of fugitive emissions (those escaping capture) and PM emission standards assigned to control devices. Five of the seven operating sinter plants are subject to a building opacity limit. One plant is subject to a 10 percent limit (6minute average), and four plants are subject to 20 percent limits (6minute average). The PM limits for control devices vary substantially from plant to plant both in terms of format and numerical values. Four plants have concentration limits for total PM (0.01, 0.02, 0.02, and 0.03 gr/dscf), one has concentration limits for PM10, and three have mass rate limits (42.9, 50, and 50 lb/hr).

We have credible source test data on actual emissions from only one plantthe refurbished sinter plant in Youngstown, OH. Captured emissions from the discharge end are ventilated to a relatively new baghouse (1991) for control. We have no data from any source on the opacity of fugitive emissions that escape capture from the discharge end.

In selecting the MACT floor for the discharge end, we evaluated all of the available information on control measures, State regulations, and actual emissions. Due to the limited information on actual emissions available, we concluded that the available information on State regulations provided the best and most complete information for establishing floor conditions for both the discharge end building and control devices. We believe that these State limits are in fact a reasonable representation of what is actually achieved in practice and are, therefore, suitable proxies for establishing MACT floor conditions. The existing State emission limits reflect a level of performance which we would expect from the capture systems and control devices which are currently applied to the control of emissions from sinter plant discharge ends.

As noted above, five plants are subject to State standards that limit the opacity of visible emissions released from the discharge end building. These range from 10 percent (one plant) to 20 percent (four plants). We chose the median value as the MACT floor, which is 20 percent opacity based on a 6minute average.

For control devices, we examined the top five most stringent existing State permit limits for total PM emissions. These include the four concentration limits cited above and a fifth value derived from the lowest mass rate limit to which a plant is subject (42.9 lb/hr), which is equivalent to 0.02 gr/dscf. The resulting five most stringent limits are 0.01, 0.02, 0.02, 0.02 and 0.03 gr/dscf. Averaging these five values produces a MACT floor limit of 0.02 gr/dscf.

We examined options to go beyond the floor level of control. One option is a concentration limit lower than the floor level of 0.02 gr/ dscf. For example, the installation of a new pulse jet baghouse could conceivably achieve a concentration limit of 0.01 gr/dscf. We estimate the capital cost of a new pulse jet baghouse designed for a flow rate of 120,000 dscfm (typical for discharge ends) to be $3.5 million and the total annual cost to be $840,000 per year. We estimate the corresponding reduction in HAP metals achieved by reducing the PM concentration from 0.02 to 0.01 gr/dscf (for 120,000 dscfm and 0.75 percent metal HAP in the PM) to be 0.34 tons per year. The cost per ton of HAP is $2.5 million. We believe that the high cost, coupled with the small reduction in HAP emissions, does not justify this beyond the floor alternative. We could not identify any other beyond the floor alternatives. Consequently, we chose the floor level of control (0.02 gr/dscf) as MACT.

For new source MACT, we chose an opacity limit of 10 percent (6 minute average) based on the most stringent emission limit currently in place (Sparrows Point, MD). For control devices used on the discharge end, we relied on test data for the baghouse at the Youngstown, OH sinter plant. We believe this baghouse represents the best controlled similar source among the seven operating plants. It is a relatively recent installation (1991) and is a stateoftheart pulse jet unit. The discharge end at this facility is comprised of a sinter breaker, single deck hot screen, fourstack sinter cooler, and a double deck cold screen. Capture systems are used for the breaker, hot screen, cold screen, and about 40 transfer points. The capture system is ventilated to a four compartment pulse jet baghouse with polyester bags at a rate of 140,000 dscfm.

Three test runs were conducted in 1991. The runs range from 0.005 to 0.006 gr/dscf and average 0.006 gr/dscf. Rounding the results of this single performance test (average of three runs) would support a new source MACT concentration limit of 0.01 gr/dscf. We believe that rounding from 0.006 to 0.01 is justified given the data are limited to the one performance test conducted in 1991.

The numerical limit selected for the standard is the same as that established for MACT: (1) An opacity limit of 20 percent (6minute average) for the building and a concentration limit of
[[Page 36845]]
0.02 gr/dscf for control devices for existing sinter plants, and (2) an opacity limit of 10 percent (6minute average) and a concentration limit of 0.01 gr/dscf for new sinter plants.

For compliance demonstration purposes, we are proposing a flow weighted average for emission control devices on the discharge end. Some plants employ multiple control devices applied to the several emission points that comprise the discharge end (crushers, screens, conveyor transfer points). For example, one plant routes emissions from the crusher to one baghouse, and emissions from screens and conveyors are sent to a second baghouse. Averaging emissions across multiple control devices provides flexibility and enhances achievability. With this approach, some air pollution control devices may under perform and others may over perform provided that the average concentration weighted by volumetric flow rate meets the concentration limit for the discharge end.

Sinter plant coolers are large diameter circular tables through which ambient air is drawn to cool the hot sinter after screening. Seven plants operate sinter coolers to cool the sinter product prior to storage. Two plants that are not currently operating have no cooler and stockpile hot sinte

FOR FURTHER INFORMATION CONTACT Phil Mulrine, Metals Group, Emission Standards Division (MD13), U.S. EPA, Research Triangle Park, NC 27711, telephone number (919) 5415289, electronic mail address:
mulrine.phil@epa.gov.