DIALOG Accession Number 00456644 Title National Emission Standards for Hazardous Air Pollutants for Source Categories; Organic Hazardous Air Pollutants from the Synthetic Organic Chemical Manufacturing Industry and Seven Other Processes Vol. 57,No. 252 Part II 57 FR 62608 Thursday, December 31, 1992 ENVIRONMENTAL PROTECTION AGENCY Proposed Rules 40 CFR Part 63 AD-FRL-4535-5 DATES: Comments. Comments must be received on or before March 31, 1993. Public Hearing. If anyone contacts EPA to speak at a public hearing by January 21, 1993, a public hearing will be held on February 25, 1993 beginning at 10 a.m. Persons interested in attending the hearing should call Ms. Julia Stevens at the address below by January 21, 1993. Request to Speak at Hearing. Persons wishing to present oral testimony must contact EPA by January 21, 1993 (contact Ms. Julia Stevens at 919 541-5578). CONTACT: For general information on the proposed rule and information on the equipment leak standard, contact Dr. Janet S. Meyer, Standards Development Branch, Emission Standards Division (MD-13), U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards, Research Triangle Park, North Carolina 27711, telephone number (919) 541-5254. For information on emissions averaging, contact Ms. Daphne L. McMurrer, Standards Development Branch, at the same address, telephone number (919) 541-0248. For technical information on wastewater handling operations, contact Mr. K. C. Hustvedt, Chief, Petroleum Section, Chemicals and Petroleum Branch, at the same address, telephone number (919) 541-5395. For technical information on the other kinds of emission points, contact Mr. Robert E. Rosensteel, Chief, Chemical Manufacturing Section, Chemicals and Petroleum Branch, at the same address, telephone number (919) 541- 5608. ADDRESS: Comments. Comments should be submitted (in duplicate if possible) to the EPA's Air Docket (LE-131), Attn: Docket Number (see list following address), room M1500, U. S. Environmental Protection Agency, 401 M Street, SW., Washington, DC 20460. Comments that address areas pertinent to the proposed rule as a whole or that are applicable to more than one kind of emission point, such as general policy or legal comments, comments on the overall impacts of the standards, and comments on test methods should be marked Attn: Docket Number A-90-19. Technical comments specific to process vents should be marked Attn: Docket Number A-90-19; technical comments about equipment leaks and any other comments about the negotiated regulation for equipment leaks should be marked Attn: Docket Number A- 90-20; technical comments about storage vessels should be marked Attn: Docket Number A-90-21; technical comments about transfer operations should be marked Attn: Docket Number A-90-22; and comments specific to wastewater operations should be marked Attn: Docket Number A-90-23. Public Hearing. If anyone contacts EPA requesting to present oral testimony on the rule being proposed today, a public hearing will be held at the EPA's Office of Administration Auditorium, Research Triangle Park, North Carolina. Persons interested in attending the hearing or wishing to present oral testimony should notify Ms. Julia Stevens, Standards Development Branch, Emission Standards Division, U. S. Environmental Protection Agency, Office of Air Quality Planning and Standards (MD- 13), Research Triangle Park, North Carolina 27711, telephone number (919) 5410 Background Information Document. The background information document (BID) for this rulemaking may be obtained from the U. S. EPA Library (MD-35), Research Triangle Park, North Carolina 27711, telephone number (919) 541-2777. Refer to EPA-453/D- 92-016a, b, and c. The HON BID comprises three volumes. Persons requesting copies of the BID should specify the volume(s) required. For information on the methodology and results of the analysis of national impacts, request BID Volume 1A. For information on emission control technologies and cost procedures, request BID Volume 1B. For information on the development of models for the five kinds of emission points, request BID Volume 1C. Dockets. The dockets listed above under ADDRESSES contain supporting information used in developing the proposed rule. Supporting information used in developing the negotiated standard for equipment leaks is available in Docket Number A-89-10. These dockets are available for public inspection and copying between 8:30 a.m. and 3:30 p.m., Monday through Friday, at the EPA's Air Docket Section, Waterside Mall, room M1500, U.S. Environmental Protection Agency, 401 M Street, SW., Washington, DC 20460. A reasonable fee may be charged for copying. ACTION: Proposed rule and notice of public hearing. SUMMARY: The EPA is proposing to regulate the emissions of certain organic hazardous air pollutants from synthetic organic chemical manufacturing industry (SOCMI) production processes which are part of major sources under section 112 of the Clean Air Act as amended in 1990 (the Act). The proposed rule, referred to as the hazardous organic NESHAP or the HON, would require sources to achieve emission limits reflecting the application of the maximum achievable control technology consistent with section 112(d) of the Act. The proposed rule would reduce the emissions of 149 of the organic chemicals identified in the Act's list of 189 hazardous air pollutants at both new and existing SOCMI sources and from equipment leaks at sources in the following processes: Styrene/butadiene rubber production; polybutadiene production; chlorine production; pesticide production; chlorinated hydrocarbon use; pharmaceutical production; and miscellaneous butadiene use. The EPA is also proposing Methods 304 and 305 with the standard. These methods can be used to demonstrate compliance with control requirements for wastewater streams. A public hearing will be held, if requested, to provide interested persons with an opportunity for oral presentation of data, views, or arguments concerning the proposed rule. 265,475 TEXT: SUPPLEMENTARY INFORMATION: The following outline is provided to aid in reading the preamble to the proposed standards. I. Definitions, Acronyms, and Abbreviations A. Definitions B. Acronyms C. Abbreviations II. Policy Approach A. Background B. Overview of the Proposed Rule C. Legal Framework D. Policy Goals E. Major Policy Decisions III. Summary of Proposed Rule A. Summary of Subpart F B. Summary of Subpart G C. Summary of Subpart H IV. Summary of Impacts of Proposed Rule A. Environmental Impacts B. Energy Impacts C. Cost Impacts D. Economic Impacts V. Emissions and Impacts Estimation Methodology A. Overview B. Control Technologies for Impacts Estimation C. National Emissions and Control Cost Calculations VI. Rationale for Provisions in Subpart F A. Selection of Source Categories B. Selection of Emission Points VII. Rationale for Provisions in Subpart G A. Selection of Emission Control Requirements B. Selection of Process Vents Provisions C. Selection of Storage Vessel Provisions D. Selection of Transfer Loading Operations Provisions E. Selection of Wastewater Collection and Treatment Operations Provisions F. Selection of Emissions Averaging Provisions G. Selection of Reporting and Recordkeeping Requirements H. Selection of Compliance Provisions VIII. Rationale for Provisions in Subpart H A. Background B. Scope and Applicability C. Background Information on Equipment Leaks D. Development of Framework and Selection of Maximum Achievable Control Technology E. Selection of Format of Standards F. Selection of Emission Limits and Work Practice Requirements G. Test Methods and Procedures H. Recordkeeping and Reporting IX. Administrative Requirements A. Coordination with Other Clean Air Act Requirements B. Executive Order 12291 C. Paperwork Reduction Act D. Regulatory Flexibility Act E. Review I. Definitions, Acronyms, and Abbreviations The following lists of definitions, acronyms, and units of measure are provided to aid in reading the preamble to the proposed rule. Additional definitions are provided near the beginning of the proposed subparts F, G, and H. A. Definitions The following definitions were developed for use in preparing and describing the proposed rule. Control device means any equipment used for recovering or oxidizing organic hazardous air pollutant vapors. Such equipment includes, but is not limited to, absorbers, carbon absorbers, condensers, incinerators, flares, boilers, and process heaters. For process vents, recovery devices are not considered control devices. Discount factor is a specified percentage used to reduce the value of emission credits. A discount factor of 20 percent reduces 10 Mg of potential emission credits to 8 Mg of actual emission credits that could be used to balance an emissions debit. For regulatory purposes, a 20 percent discount factor is represented as 0.8 in credit estimation equations. Emissions averaging is a means of complying with subpart G of this proposed rule. Emissions averaging allows a source to create emission credits by reducing emissions from specific points to a level below that required by subpart G. Those credits are used to offset emission debits from points that are not controlled to the level required by subpart G. Emission credits are excess emission reductions above those required by subpart G that are used to offset emission debits in emissions averaging. Emission debits are increased emissions that result when a source elects not to control a Group 1 emission point to the level required by subpart G. Emission point means an individual process vent, storage vessel, transfer rack, wastewater stream, or equipment leak. Group 1 emission point means an individual process vent, storage vessel, transfer rack, or wastewater stream that satisfies the applicability criteria for the control requirements of subpart G. Group 2 emission point means an individual process vent, storage vessel, transfer rack, or wastewater stream that does not satisfy the applicability criteria for the control requirements of subpart G. Halogenated vent stream or halogenated stream means a vent stream from a process vent or transfer operation determined to have a total concentration of halogen atoms (by volume) contained in organic compounds of 200 parts per million by volume or greater. Hazardous Air Pollutant or HAP means any air pollutant listed under section 112(b) of the Clean Air Act. Plant site means all contiguous or adjoining property that is under common ownership or control, including properties that are separated only by a road or other public right-of-way. Common ownership or control includes properties that are owned, leased, or operated by the same entity, parent entity, subsidiary, or any combination thereof. Reference control technology means a device or devices that can be used to comply with the control requirements in subpart G. Subpart G specifies the reference control technologies for each kind of emission point and establishes a control efficiency that the devices should achieve when being used to comply with this rule. Very volatile hazardous air pollutant or very volatile HAP means one of the chemicals listed in Table 8 of subpart G. Volatile organics or VO refers to the portion of organic compounds (including both hazardous air pollutant and non-hazardous air pollutant organic compounds) in a wastewater stream that is measured by Method 25D, as found in 40 CFR part 60, appendix A. Volatile organic hazardous air pollutant or VOHAP means the volatile portion of an individually- speciated organic hazardous air pollutant in a wastewater stream or a residual that is measured by proposed Method 305. Waste management unit means any component, piece of equipment, structure, or transport mechanism used in conveying, storing, treating, or disposing of any waste, including a wastewater stream or a residual. Wastewater tanks are an example of a waste management unit. Wastewater means organic hazardous air pollutant-containing water or process fluid discharged into an individual drain system and includes process wastewater, maintenance-turnaround wastewater, and maintenance wastewater. Organic hazardous air pollutant- containing water or process fluids contain at least 5 parts per million by weight total organic hazardous air pollutant and have a flow rate of 0.02 liter per minute, or greater, or a concentration of at least 10,000 parts per million by weight and any flow rate. B. Acronyms B. Acronyms Acronyms BD Term butadiene. Acronyms BID Term background information document. Acronyms CFR Term Code of Federal Regulations. Acronyms CMA Term Chemical Manufacturers Association. Acronyms CO Term carbon monoxide. Acronyms CTG Term control techniques guideline. Acronyms CWA Term Clean Water Act. Acronyms DMS Term dual mechanical seal. cronyms DOT Term Department of Transportation. Acronyms EB/S Term ethylbenzene/styrene. Acronyms EO Term ethylene oxide. Acronyms E.O. Term Executive Order. Acronyms EPA Term Environmental Protection Agency. Acronyms FR Term Federal Register . Acronyms HAP Term hazardous air pollutant. Acronyms HON Term hazardous organic national emission standards for hazardous air pollutants. Acronyms LDAR Term leak detection and repair. Acronyms MACT Term maximum achievable control technology. Acronyms NESHAP Term national emission standards for hazardous air pollutants. Acronyms NO sub x Term nitrogen oxides. Acronyms NSPS Term new source performance standards. Acronyms OMB Term Office of Management and Budget. Acronyms OSHA Term Occupational Safety and Health Administration. Acronyms P.L. Term Public Law. Acronyms QIP Term quality improvement program. Acronyms RCRA Term Resource Conservation and Recovery Act. Acronyms RIA Term Regulatory Impact Analysis. Acronyms SIP Term State Implementation Plan. Acronyms SMS Term single mechanical seal. Acronyms SOCMI Term synthetic organic chemical manufacturing industry. Acronyms TOC Term total organic compound. Acronyms TRE Term total resource effectiveness. Acronyms TACB Term Texas Air Control Board. Acronyms TSDF Term treatment, storage, and disposal facility. Acronyms VHAP Term volatile hazardous air pollutant. Acronyms VO Term volatile organics measurable by Method 25D. Acronyms VOC Term volatile organic compound. Acronyms VOHAP Term volatile organic hazardous air pollutant. C. Abbreviations C. Abbreviations Abbreviation bbl Unit of measure barrel. Abbreviation BOE Unit of measure barrels of oil equivalent. Abbreviation Btu Unit of measure British thermal unit. Abbreviation Btu/kW-hr Unit of measure British thermal unit per kilowatt-hour. Abbreviation degrees C Unit of measure degrees Celsius. Abbreviation degrees F Unit of measure degrees Fahrenheit. Abbreviation gal Unit of measure gallon. Abbreviation hr Unit of measure hour. Abbreviation kPa Unit of measure kilopascals. Abbreviation kW-hr/yr Unit of measure kilowatt-hour per year. Abbreviation l pm Unit of measure liters per minute. Abbreviation gal Unit of measure gallons. Abbreviation m sup 3 Unit of measure cubic meters. Abbreviation Mg Unit of measure megagrams. Abbreviation mg Unit of measure milligrams. Abbreviation mg/dscm Unit of measure milligram per dry standard cubic meter. Abbreviation MW Unit of measure megawatts. Abbreviation ppb Unit of measure parts per billion. Abbreviation ppm Unit of measure parts per million. Abbreviation ppmv Unit of measure parts per million by volume. Abbreviation ppmw Unit of measure parts per million by weight. Abbreviation psia Unit of measure pounds per square inch absolute. Abbreviation scm/min Unit of measure standard cubic meter per minute. Abbreviation TJ Unit of measure terajoules. Abbreviation yr Unit of measure year. II. Policy Approach This section provides background about the legal and policy criteria that the Administrator took into consideration in selecting the provisions of this proposed rule. It is included to give the reader a sense of the rule as a whole. To that end, the section includes background about the rule, a brief overview of the rule, some statutory history, a summary of the current statutory requirements for standards developed under Section 112 of the Act, and a discussion of the Agency's policy goals. This section concludes with a short discussion of the major policy decisions that the Administrator made to structure this proposed rule in such a way that it meets the statutory criteria and the Agency's policy goals. A. Background The regulation being proposed today, under Section 112 of the Act, is known as the hazardous organic NESHAP, or HON. The HON, as proposed, would set MACT for one source category and seven processes in other source categories. The entire SOCMI source category and equipment leaks from seven non- SOCMI processes are to be regulated. The Act, as amended in 1990, requires that EPA promulgate standards for 40 source categories or subcategories emitting HAP's within 2 years of its enactment. In the Statement of the Managers accompanying the final bill, as enacted, Congress indicated that EPA should fulfill this initial statutory requirement by regulating the priority elements of the HON as it was under development during consideration of the bill. 136 Cong. Rec. H13198 (October 26, 1990). This proposed rule would cover the HON, as Congress described it, and more. As such, this rulemaking, while regulating less than 40 source categories, will fulfill Congressional intent concerning what should have been regulated within 2 years of enactment. The SOCMI, as a source category, emits a large volume and variety of HAP's relative to other source categories. In addition, individual SOCMI sources tend to be located in close proximity to populations. As such, components of SOCMI sources have already been subject to various Federal, State, and local air pollution control rules. However, the existing rules, even when considered together, do not comprehensively regulate emissions of all the organic HAP's emitted from all the emission points at both new and existing sources. The HON, as proposed today, reflects the EPA's regulatory experience from previous NESHAP and NSPS rulemakings involving similar kinds of sources and emission points. Information on control technology applicability, performance, and costs was developed to support these NESHAP and NSPS. This information was carefully reconsidered in light of the 1990 amendments to the Act and used in the selection of MACT and the other provisions of the proposed rule, such as monitoring, recordkeeping, and reporting requirements. The EPA has promulgated NSPS for SOCMI air oxidation and distillation process vents, SOCMI equipment leaks, petroleum refinery equipment leaks, and VOC emissions from volatile organic liquid storage vessels. The EPA has also promulgated NESHAP for benzene transfer operations, storage vessels, and waste operations, and benzene equipment leaks. The vinyl chloride NESHAP establishes standards for emission points at vinyl chloride and ethylene dichloride production processes. Although these existing rules will remain in effect, the HON would provide comprehensive coverage of the SOCMI by regulating the organic HAP emissions from five kinds of emission points at each affected SOCMI source. B. Overview of the Proposed Rule This section of the notice provides an overview of the proposed rule. A more detailed summary of the proposed rule is provided in section III and the rationale for the provisions in the proposed rule is provided in sections VI through VIII. The proposed rule comprises three subparts to be included in 40 CFR part 63. Subpart F provides the applicability criteria and general compliance requirements for the rule. Subparts G and H provide the control, monitoring, recordkeeping, and reporting requirements for the five kinds of emission points. 1. Subpart F: Applicability of the HON The HON would regulate certain components of new and existing major sources, as defined by section 112(a) of the Act, for the SOCMI and seven non-SOCMI processes. To define the SOCMI source category, subpart F includes a list of organic HAP's and a list of 396 synthetic organic chemicals produced by the SOCMI as commercial products. The ''chemical manufacturing processes'' used to produce these 396 chemicals can, but do not always, result in organic HAP emissions depending on whether HAP's are used or produced. Only those processes that use as a reactant or produce as a product, by- product, or co-product one or more organic HAP's would be subject to the proposed rule. As proposed, subpart F defines ''source'' for the SOCMI source category as the set of process vents, storage vessels, transfer racks, wastewater streams, and equipment leaks in the organic HAP-emitting chemical manufacturing processes that are subject to the HON. To be subject to the HON, a chemical manufacturing process must be used to produce one or more of the 396 SOCMI chemicals listed in subpart F, and have an organic HAP as either: (1) A product, by-product, co-product, or intermediate; or (2) A reactant. To be part of the same source, chemical manufacturing processes that are subject to the HON must also be located within a contiguous plant site under common control.{pg 62611} Subpart G would apply to the following kinds of emission points in SOCMI chemical manufacturing processes: process vents, wastewater operations, storage vessels, and transfer operations. Subpart H would apply to the equipment leaks in SOCMI and seven non-SOCMI chemical manufacturing processes. The following non-SOCMI equipment leak processes are subject to subpart H: Styrene/butadiene rubber production; polybutadiene production; chlorine production; pesticide production; chlorinated hydrocarbon use; pharmaceutical production; and miscellaneous butadiene use. 2. Subpart G: Provisions for Process Vents, Wastewater, Storage Vessels and Transfer Operations Subpart G of the proposed rule would require the owner or operator of a source to limit source-wide emissions of HAP's. Subpart G provides specific instructions for determining how much emission reduction must be achieved at each source. The required emission reduction is determined by how much emissions would be reduced if a ''reference control technology'' were applied to each ''Group 1'' emission point in the source. The proposed rule specifies the reference control technology for each kind of point. Group 1 points are those points that meet the applicability criteria included in the control requirements for the proposed rule. The reference control technologies and the characteristics of Group 1 points are specified in subpart G of the proposed rule. The owner or operator of a source can use two methods to comply with the emission reduction requirement. Either method can be used exclusively, or the two can be combined. The first method is to apply the reference control technology, or an equivalent technology, to Group 1 emission points; thereby achieving some part of the required emission reduction at each Group 1 point that is controlled. The second method is to average emissions from two or more emission points such that the overall required emission reduction is achieved. With the second method, emissions averaging, the owner or operator does not have to apply the reference control technology to each Group 1 point as long as an equivalent or greater emissions reduction is achieved elsewhere in the source. Although equipment leaks are included in the definition of source for the SOCMI source category, equipment leaks cannot be included in the emissions averages because: (1) The equipment leaks standard has no fixed performance level; and (2) no method currently exists for determining the magnitude of allowable emissions to assign equipment leaks for purposes of emissions averaging. When this methodology is developed, EPA will consider allowing equipment leak emissions to be included in emissions averages. 3. Subpart H: Provisions for Equipment Leaks The provisions in subpart H of the proposed rule were developed using regulatory negotiation and represent an extension of existing equipment leak control procedures and techniques to the processes regulated by today's proposal. Subpart H proposes work practice requirements to reduce emissions from equipment leaks for equipment in volatile HAP service for 300 or more hours per year. To be in volatile HAP service is to be in contact with or containing process fluid that contains total 5 percent or more total HAP. The following types of equipment are subject to Subpart H: valves, pumps, connectors, compressors, pressure relief devices, open-ended lines, sampling connection systems, instrumentation systems, agitators, product accumulator vessels, and closed-vent systems and control devices. C. Legal Framework This section provides a brief history of section 112 of the Act and background regarding the definition of source categories and source for section 112 standards. This information is included to give the reader a sense of the statutory, judicial, and Congressional guidance that the Administrator took into consideration in developing the source category and source definitions for the HON. 1. Statutory Background Prior to the 1990 Amendments, section 112 of the Act required the Administrator to list air pollutants for which he intended to establish NESHAP. Then, within 180 days of such listing, the Administrator was required to propose regulations for each listed pollutant. He was also required to issue final regulations within another 180 days. Thus, once the Administrator added a pollutant to the section 112 list, a final NESHAP for that pollutant had to be issued within 1 year. The statute itself did not contain a list of pollutants. Section 112 also provided that the Administrator must establish NESHAP at the level which ''in his judgment provides an ample margin of safety to protect the public health from such hazardous air pollutant.'' Section 112(b)(1)(B). As a result of this language, EPA conducted risk assessments to determine which pollutants should be regulated and to what level. Because of the substantial length of time required to complete a risk assessment study, EPA generally did not list an air pollutant until the proposed regulations were well underway. In 1987, the legal framework for setting NESHAP was further defined when the D.C. Circuit handed down an en banc decision in Natural Resources Defense Council, Inc., versus EPA, 824 F.2d 1146 (D.C. Cir. 1987), (hereafter referred to as Vinyl Chloride). In that decision, the court set out a two-step process for EPA to follow in setting standards: (1) Determine a ''safe'' or ''acceptable'' risk level, and (2) set the standard at the level- which may be lower but not higher than the ''safe'' or ''acceptable'' level-that protects public health with an ample margin of safety. Following the 1987 Vinyl Chloride decision, the EPA promulgated NESHAP for several source categories of benzene and radionuclide emissions using the two- step process mandated by the court. A ''safe'' level was determined in each instance through risk assessment, then a standard providing protection with an ''ample margin of safety'' was set after consideration of factors such as cost and feasibility. 2. Current Statutory Requirements The 1990 Amendments altered the preexisting scheme of section 112 fundamentally. Instead of requiring the EPA to determine which air pollutants should be listed and regulated as HAP's, Congress provided a list of 189 HAP's in the statute itself and directed EPA to develop rules to control HAP emissions. The Act requires that the rules be established for categories of sources of the emissions, rather than being set by pollutant. In addition, the Act sets out specific criteria for establishing a minimum level of control, and criteria to be considered in evaluating control options more stringent than the minimum control level. For most of these rules, assessment and control of any remaining unacceptable health risk is to occur 8 years after they are promulgated. However, for the rules required to be promulgated in the first 2 years after enactment, EPA is not required to conduct this assessment until 9 years after promulgation. Specifically, section 112(c), as amended, directs the Administrator to develop a list of all categories or {pg 62612} subcategories of major and area sources, as defined in section 112(a), emitting significant amounts of the HAP's listed in section 112(b). Section 112(d) directs the Administrator to promulgate emission standards for each listed category or subcategory of HAP sources. Such standards will be applicable to both new and existing sources and shall require: the maximum degree of reduction in emissions of the hazardous air pollutants subject to this section (including a prohibition on such emissions, where achievable) that the Administrator, taking into consideration the cost of achieving such emission reduction, and any non-air quality health and environmental impacts and energy requirements, determines is achievable for new and existing sources in the category or subcategory to which such emission standard applies 42 U.S.C. 7412(d)(2). The Amendments further provide that ''the maximum degree of reduction in emissions that is deemed achievable'' shall be subject to a ''floor'' which is determined differently for new and existing sources. For new sources the standards set shall not be any less stringent than ''the emission control that is achieved in practice by the best controlled similar source.'' For existing sources, the standards may not be less stringent than the average emission limitation achieved by the best performing 12 percent of existing sources in each category or subcategory of 30 or more sources. (Smaller categories or subcategories are limited to the average of the best five performing sources in the category or subcategory.) 3. The Definition of Source Category and Source The definition of source is an important element of Section 112 standards because it describes the emission points to which each standard applies. The definition of source is fundamental to the determination of the MACT floor and the evaluation of regulatory options more stringent than the floor. This section describes some of the factors that the Administrator took into consideration in selecting the definition of source for the SOCMI source category. Section 112(c) directs the Administrator to create a list of source categories for MACT standards such that '' t o the extent practicable, the categories and subcategories listed under this subsection shall be consistent with the list of source categories established pursuant to section 111 and Part C.'' As is clear from a review of those existing lists, the categories listed are generally broadly drawn. For example, the Part C list includes fossil-fuel fired steam electric plants of more than 250,000,000 Btu/hr heat input, iron and steel mill plants, petroleum refineries and chemical process plants and the section 111 list includes petroleum manufacturing and marketing, plywood manufacture, glass and crude oil and natural gas production. Listing the SOCMI as a category on the section 112(c) list (57 FR 31576, July 16, 1992) is consistent with the general broad categorization of the section 111 and part C lists. Section 112(d) directs the Administrator to set standards for all ''major sources'' within every listed category. Area sources meeting the requirements of sections 112(c)(3) or 112(k) must also be regulated. Major sources are ''stationary sources,'' or groups of stationary sources, of a given size, as defined in section 112(a)(1). The definition of ''stationary source'' included in section 112 is identical to the definition used in section 111(a) which is ''any building, structure, facility, or installation which emits or may emit any air pollutant.'' 42 U.S.C. 7411(a). However, section 112 as amended does not require that the standards set under section 112(d) be set for the same components of the categories as was done under section 111. Thus, there is no requirement that section 112(d) standards for sources in the SOCMI be set for precisely the same portions of the industry as NSPS. As the Supreme Court has recognized in Chevron, USA, Inc., v. Natural Resources Defense Council, 467 U.S. 837 (1984) (hereafter referred to as Chevron), EPA has broad discretion to define ''source.'' The Court recognized in Chevron that if any Congressional intent can be discerned from the statutory language of section 111(a)(3) (the definition of source that is used in section 112), ''the listing of overlapping, illustrative terms was intended to enlarge, rather than confine, the scope of the EPA's power to regulate particular sources in order to best effectuate the policies of the Act.'' Chevron. Thus, the court found that a ''source'' can encompass ''any discrete, but integrated operation, which pollutes.'' Chevron. As such, it could also encompass an entire plant and EPA has flexibility, within the broad definition of ''stationary source,'' to define the source for each section 112(d) standard as broadly or narrowly as is appropriate for the particular industry being regulated. For the HON, EPA is proposing to define ''source'' for the SOCMI source category as the process vents, storage vessels, transfer racks, wastewater collection and treatment operations, and equipment leaks in the organic HAP emitting chemical manufacturing processes that are located in a single facility covering a contiguous area under common control. With this definition of source, all SOCMI plant sites that are major sources under section 112, approximately 350, will be subject to the standard. One of the implications of the definition of source proposed for the SOCMI source category is that a single ''floor,'' as defined in section 112(d)(3), is applicable to the entire SOCMI operation regulated by the HON. Thus, in setting MACT for the SOCMI, EPA had to first determine the floor for new and existing SOCMI sources. To determine the floor for existing sources, EPA assessed the average emission limitation achieved by the best performing 12 percent of existing SOCMI sources. To determine the floor for new sources, EPA assessed the emissions limitations achieved by the best performing existing source in the source category. The EPA then evaluated the costs and non-air quality impacts of control before arriving at new and existing standards at least as stringent as the new and existing floors for the SOCMI source category. 4. Authority for Emissions Averaging Under Section 112, the Administrator has legal authority to permit affected sources to comply with the standard through emissions averaging. Section 112(d) provides that standards are to be established for each category or subcategory of sources listed by the Administrator, and that such standards shall be applicable to sources within those categories or subcategories. The statute does not define source category, nor, as explained above, does it impose precise limits on how the Administrator may define source. Thus, the Administrator has the discretion to define the source category and the source either narrowly or broadly. In this case, the Administrator is proposing to exercise that discretion to define source broadly to include all the emission points relating to SOCMI production. In setting the standard, the Administrator is required to determine a floor for the entire category or subcategory, and then set a standard applicable to each source within that category that is at least as stringent as the floor. In determining whether the standard should be more stringent than the floor and by how much, the Administrator is to consider, among other things, the cost of achieving such additional reductions. The statutory provisions do not limit how the {pg 62613} standard is to be set beyond requiring that it be applicable to all sources in a category and be at least as stringent as the floor. Therefore, the relevant statutory language does not prohibit EPA from allowing a source to meet MACT through use of emissions averaging as long as every source in the category is required to comply, averaging does not cross source boundaries, and the standard is set at a level at least as stringent as the floor. As further discussed in section VII.F of this notice, the EPA is seeking comment on a complementary legal interpretation of sections 112(d) and 112(i) of the Act. In addition, it should be noted that Congress explicitly provided that cost should be considered in setting the standard. Emissions averaging is a means of achieving the reductions required by the standard in a cost effective way, and is thus clearly in line with Congressional intent. D. Policy Goals The SOCMI component of the HON is expected to result in the greatest emissions reduction likely to be achieved by any single source category being regulated under section 112. As such, regulating this industry represents a significant first step toward fulfilling the mandate of section 112 to reduce emissions of HAP's. In addition, SOCMI facilities, or sources, tend to be large individual emitters of HAP's, which are generally thought to pose potential health hazards at the local level. The EPA recognizes that the HAP's covered by the HON represent a wide range of toxicities, a variety of potential toxic effects, and a variety of exposure levels. The EPA also recognizes clear public interest in reducing HAP emissions from the SOCMI as much as is achievable, based upon the potential for health and environmental benefits from HAP emission reductions of this magnitude. Aside from the general goal of maximum achievable emissions reduction, the EPA has endeavored to structure the proposed rule to incorporate several other goals: overall administrative simplicity, allowing flexibility in implementation in order to reduce costs, encouraging pollution prevention, and ensuring enforceability. Some goals such as flexibility and encouraging pollution prevention reinforce each other, while other goals such as flexibility and enforceability may seem contradictory. The EPA has striven to find a workable balance among the potentially contradictory goals, and is requesting comment on the proposed solutions in this notice. E. Major Policy Decisions This section provides a discussion of the major policy decisions that provide the framework for this rule, and the context in which those decisions were reached. 1. The Definition of Source For the SOCMI source category, ''source'' is defined as the set of emission points in the organic HAP-emitting processes used to produce synthetic organic chemicals that are in a contiguous area under common control. The Administrator carefully considered the previously described statutory requirements, legal history, and Congressional intent in selecting the definition of source for the SOCMI source category. In addition, the Administrator considered the technical concerns regarding implementation and enforcement of the rule. Specifically, the Administrator wanted to select a definition of source that would provide flexibility in compliance with the rule, while maintaining enforceability. Thus, the principal rationale for adopting this definition of source is to allow flexibility in compliance, specifically to facilitate use of emissions averaging as a means of compliance with the rule. The Administrator considered this flexibility for compliance important because of the diversity among SOCMI facilities. Although the rule being proposed today sets national standards, it is based on a model analysis because data on all the plant configurations in this industry are not available. While this model analysis may reflect the characteristics of the industry as whole, it does not account for unique operational scenarios at individual sources. Emissions averaging allows the owners or operators of SOCMI sources to seek the least costly way for their individual sources to meet the allowable emissions level in this Federal rule. 2. The Floor In setting MACT standards, the EPA must establish the floor for a source category because the Act specifies that each standard be at least as stringent as the floor for the relevant source category. However, EPA did not have sourcewide data to determine the floor for the source as defined for the SOCMI source category. As a result, EPA examined available data on each kind of emission point included in the source to determine the emission reductions achieved by the best- performing 12 percent of each for existing sources; and, for new sources, the best- controlled similar emission point. Existing Federal and State regulations were used to determine current control levels on the emission points regulated by subpart G of the HON. This approach was necessary because of the difficulty of conducting a costly and time-consuming data collection effort in the time allocated for developing the rule. The EPA does not believe the results from this approach are significantly different from what they would be if source-specific data had been collected. Using this process to establish a floor for the emission points regulated by subpart G ensures that the control level of the standard will be equivalent to the emission control level for the best-controlled 12 percent of SOCMI facilities. For subpart H, the negotiating committee agreed that the requirements of the negotiated rule constitute MACT for equipment leaks. Further rationale for the negotiated regulation for equipment leaks is given in section VIII of this preamble. 3. The Control Requirements Once the floor level of control was established, EPA considered the options for control requirements more stringent than the floor. As previously noted, the Act specifies that EPA must choose control requirements as stringent as, or more stringent than, the floor level of control. As required by the statute, when considering control requirements beyond the floor, EPA considered the relative cost of achieving different levels of emissions reductions, non-air quality health and environmental impacts, and the energy requirements of the controls. While non-air quality health and environmental impacts have not been quantified for this proposed rulemaking, the control levels proposed in this rule reflect a recognition of the large magnitude of emissions that can be controlled at relatively reasonable costs on a national level, and the magnitude and variety of HAP's emitted from large individual sources. The Administrator believes that the proposed rule represents the maximum degree of emissions reductions achievable with reasonably cost- effective controls for most kinds of equipment covered by this rulemaking, given the large size of most major sources, as well as the magnitude and variety of their emissions. 4. The Form of the Standard The proposed HON establishes a control requirement for each kind of emission point regulated by subparts G and H. To facilitate emissions averaging, the Administrator chose to have the subpart G standard also establish an {pg 62614} overall allowable emissions level for the source as a whole. The allowable emissions would be equal to the sum of the emissions from each point in the source, excluding equipment leaks, if the required controls were applied. As such, the allowable emissions level is set for a given mix of emission points, and the allowable emissions will change as the number or kind of emission points in the source changes. Though the form of the standard is an allowable emissions level, compliance can be determined on a point- by-point basis for emission points not included in emissions averages. Point-by-point compliance is determined based on the types of controls in place on individual emission points, their performance, and their operating conditions. 5. Emissions Averaging With emissions averaging, if an owner or operator does not wish to control a particular emission point, the uncontrolled emissions from that point can be offset by emissions below what is required by subpart G at one or more other points. Although subpart G allows for extensive use of emissions averaging, it is expected that it will only be used for a limited number of emission points at any one source. Section III.B.6 provides a more detailed discussion of how emissions averaging is designed to work and section VII.F provides a detailed discussion of the issues considered by the EPA in the process of developing the emissions averaging policy. In section VII.F, the EPA seeks comment on numerous aspects of the emissions averaging policy included in the proposed rule. III. Summary of Proposed Rule This section of the notice summarizes the proposed rule. For an explanation of the process used to select these requirements and the rationale for specific provisions, see sections VI, VII, and VIII. The proposed rule consists of three subparts in 40 CFR part 63. Subpart F provides the applicability criteria for the rule, requires that owners and operators of SOCMI sources comply with Subparts G and H, and specifies general recordkeeping and reporting requirements. Subparts G and H provide the specific control, monitoring, reporting, and recordkeeping requirements for the respective kinds of emission points. A. Summary of Subpart F Subpart F lists the HAP's regulated by this rule, specifies what is included in the SOCMI source category, and details the seven non-SOCMI processes that are also subject to subpart H. In addition, subpart F presents definitions and general information on compliance, reporting, and recordkeeping requirements that are applicable for sources subject to subpart G or H. 1. Regulated Pollutants Subpart F lists 112 organic HAP's that EPA has determined may be emitted from SOCMI processes because they are either produced as a product or used as a reactant. The emissions of these 112 organic chemicals are regulated by subparts F and G. In addition to these 112 organic HAP's, 37 other organic HAP's are regulated by subparts F and H. The complete list of 149 organic HAP's is presented in subpart H. 2. Definition of Source Category and Source Sources in the SOCMI source category and seven non-SOCMI processes would be subject to the proposed rule. To define the SOCMI source category for purposes of the HON, subpart F lists 396 chemicals considered SOCMI products. The EPA has determined that the production of these chemicals may result in organic HAP emissions. As a result, chemical manufacturing processes used to produce one of these 396 SOCMI chemicals as a product are in the SOCMI source category. These processes would make up the SOCMI sources that would be subject to subparts F, G, and H of this rule. If a process produces one of the 396 listed chemicals but does not use a HAP as a reactant or produce a HAP as a product, by-product, or co-product, it would not be subject to this proposed rule. For the SOCMI source category, a source comprises all the SOCMI chemical manufacturing processes that are subject to the rule and located at one contiguous geographic site under common control. Subpart F defines the SOCMI source as the collection of process vents, storage vessels, transfer racks, wastewater streams (and associated residuals), and equipment leaks in the relevant chemical manufacturing processes. As listed above, the first four of the five kinds of emission points in a SOCMI source would be subject to subparts F and G. However, SOCMI equipment leaks will be subject to subparts F and H. As such, a SOCMI source is subject to all three of the HON's subparts. In contrast to the sources in the SOCMI source category, sources in the seven non-SOCMI processes would be covered by subparts F and H only. For these processes, the source would include only equipment leaks. As explained in the draft schedule for the promulgation of emission standards (57 FR 44147), EPA is considering regulating the other kinds of emission points in these seven processes in future section 112 standards. The seven processes subject to subpart H of the HON are included in 20 different source categories or subsets of source categories. The exact relationship of the HON's seven equipment leak processes to the source categories listed for section 112 standards is specified in Table 1 of the draft schedule for the promulgation of emission standards. 3. Other Provisions The proposed subpart F establishes the compliance dates for new and existing sources and requires the source be properly operated and maintained at all times. As part of proper operation and maintenance provisions, sources are required to include procedures for managing wastewaters generated during maintenance turnarounds and emptying and purging of equipment during routine maintenance in the startup, shutdown, and malfunction plan. Monitoring of cooling water is also required to detect leaks in heat exchange equipment. If a leak is detected, the heat exchanger must be repaired. Procedures for obtaining permission to use an alternative means of emission reduction are included in the proposed subpart F. The applicability of the General Provisions in subpart A to sources subject to subparts F, G, and H is clarified. General performance test requirements are specified, including the provision that performance tests be conducted under representative operating conditions. The General Reporting and Recordkeeping Provisions of the proposed subpart F include the requirement that required records and reports must be maintained for 5 years, and specify where reports must be sent. B. Summary of Subpart G 1. Overview The proposed subpart G requires that organic HAP emissions be limited to the level that could be achieved by application of a reference control technology to each Group 1 emission point in the source. Although controls are not required for Group 2 emission points, both Group 1 emission points and Group 2 emission points are included in the equation defining the source's allowable emissions level. However, emission points associated with equipment that is no longer operational are not to be included in the {pg 62615} calculation of the allowable emissions because these points are not subject to this proposed rule. Though subpart G is structured as an allowable emissions level, EPA does not anticipate that any owner or operator would actually calculate emissions estimates for every emission point at the source in order to comply with the HON. Actual emissions estimates would be required, by the HON, for only those emission points that are included in emissions averages. The owner or operator can utilize two methods to demonstrate compliance with the HON. The first method is application of the reference control technologies to achieve the required level of emission reduction at Group 1 emission points. This compliance approach is described in sections 2 through 5 below. The second compliance approach is emissions averaging. Emissions averaging is described in section 6 below. Section 7 describes the HON's recordkeeping and reporting provisions. 2. Process Vent Provisions A process vent means a gas stream that is continuously discharged during the operation of the unit from an air oxidation process unit, reactor process unit, or distillation operation within a SOCMI chemical manufacturing process. Process vents include gas streams that are discharged directly to the atmosphere and gas streams discharged to the atmosphere after diversion through a product recovery device. The proposed rule would apply only to process vents that are associated with continuous (nonbatch) processes and emit vent streams containing more than 0.005 weight- percent HAP. The process vent provisions do not apply to vents from control devices installed to comply with the wastewater provisions of subpart G. (Air emissions from control devices installed to remove HAP's from the wastewater streams are required by the wastewater provisions to be ducted to a 95-percent efficient air emissions control device.) Process vents exclude relief valve discharges and leaks from equipment regulated under subpart H, but include vents from product accumulator vessels. Vents from product accumulator vessels that are complying with the process vent provisions of subpart G are not subject to the equipment leak provisions in subpart H. The proposed process vent provisions require the owner or operator to calculate a TRE index value to determine whether each process vent is a Group 1 or Group 2 vent, except that the owner or operator can elect to comply with the control requirements without calculating the TRE index. The TRE index value is determined after the last recovery device in the process or prior to venting to the atmosphere. The TRE calculation involves an emissions test or engineering assessment and use of the TRE equations in the proposed regulation. Process vents with a TRE index equal to or less than 1.0 would be Group 1 vents and must be controlled to the level of the reference control technology, 98 percent HAP reduction or a reduction to 20 ppmv of HAP, using control devices. The proposed rule encourages use of recovery devices for additional product recovery because an owner or operator of a Group 1 process vent may add product recovery devices or otherwise reduce emissions to the extent that the TRE becomes greater than 1.0 and the Group 1 vent becomes a Group 2 vent. No additional control is required for Group 2 process vents but the TRE must be maintained above 1.0. Performance test provisions are included for Group 1 process vents to verify that the control device achieves the required performance. Halogenated streams which use a combustion device to comply with 98 percent or 20 ppmv HAP emission {pg 62616} reduction must vent the emissions from the combustor to an acid gas scrubber prior to venting to the atmosphere. The scrubber must reduce the overall emissions of hydrogen halides and halogens by 99 percent or reduce the outlet concentration of each individual hydrogen halide or halogen to 0.5 mg/dscm or less. Monitoring, reporting, and recordkeeping provisions necessary to demonstrate compliance are also included in the proposed process vent provisions. 3. Storage Vessel Provisions A storage vessel means a tank or other vessel storing the feed or product for a SOCMI chemical manufacturing process if the liquid is on the list of organic HAP's in subpart F. The storage vessel provisions do not apply to the following: (1) Vessels permanently attached to motor vehicles, (2) pressure vessels designed to operate in excess of 204.9 kPa (29.7 psia), (3) vessels with capacities smaller than 38 m sup 3 (10,000 gal), (4) product accumulator vessels, (5) wastewater tanks, and (6) vessels storing liquids that contain organic HAP's only as impurities. An impurity is produced coincidentally with another chemical substance and is processed, used, or distributed with it. The proposed storage provisions require that one of the following control systems be applied to Group 1 storage vessels: (1) An internal floating roof with proper seals and fittings, (2) an external floating roof with proper seals and fittings, (3) an external floating roof converted to an internal floating roof with proper seals and fittings, or (4) a closed vent system with a 95- percent efficient control device. The storage provisions give details on the types of seals and fittings allowed. Monitoring and compliance provisions include periodic visual inspections of vessels, roof seals, and fittings, as well as internal inspections. If a control device is used, the owner or operator must establish appropriate monitoring procedures. Reports and records of inspections, repairs, and other information necessary to determine compliance are also required by the proposed storage provisions. No controls are required for Group 2 storage vessels. 4. Transfer Operations Provisions Transfer operations are defined as the loading of liquid products that are on the list of organic HAP's in subpart F from a transfer rack within a SOCMI chemical manufacturing process into a tank truck or railcar. Transfer rack means the total of loading arms, pumps, meters, shutoff valves, relief valves, and other piping and valves necessary to load tank trucks or railcars. The transfer provisions do not apply to the loading of liquid organic HAP's at an operating pressure in excess of 204.9 kPa (29.7 psia); loading of marine vessels; racks loading liquids that contain organic HAP's only as impurities; or racks loading liquid organic HAP's if emissions are returned to a storage vessel in a vapor balancing system. The proposed transfer provisions require control of Group 1 transfer racks to achieve 98 percent organic HAP reduction or an outlet concentration of 20 ppmv. Combustion devices or product recovery devices may be used to comply with this requirement. Alternatively, vapor balancing systems may be used. The transfer provisions include design specifications for vapor collection systems. Specifically, vapor collection systems are required to route the organic vapors to a control device or to a vapor balancing system and are required to operate without detectable emissions. In addition, the proposed provisions require that liquid organic HAP's be loaded only into DOT certified vehicles or vehicles that have been determined to be vapor tight according to Method 27 of 40 CFR part 60, appendix A. Halogenated streams which use a combustion device to comply with 98 percent or 20 ppmv HAP emission reduction must vent the emissions from the combustor to an acid gas scrubber prior to venting to the atmosphere. The scrubber must reduce the overall emissions of hydrogen halides and halogens by 99 percent or reduce the outlet concentration of each individual hydrogen halide or halogen to 0.5 mg/dscm or less. Initial performance tests of control device efficiency are required, and monitoring, reporting, and recordkeeping provisions are specified. Controls are not required for Group 2 racks. 5. Wastewater Provisions The wastewater streams to which the proposed standard applies are any organic HAP-containing water or process fluid discharged into an individual drain system and includes process wastewater, maintenance-turnaround wastewater, and routine maintenance wastewater. These provisions also apply to organic HAP- containing materials (i.e., residuals) separated from wastewater. The characteristics of the process wastewater stream (e.g., flow rate, VOHAP concentration) are determined for the point of generation. Examples of a process wastewater stream include, but are not limited to, wastewater streams from process equipment, product or feed tank drawdown, cooling tower blowdown, steam trap condensate, reflux, and fluids drained into and material recovered from waste management units. Examples of maintenance- turnaround wastewater streams are those generated by descaling of heat exchanger tubing bundles, cleaning of distillation column traps, and draining of pumps into an individual drain system. For purposes of the proposed standard, an organic HAP- containing wastewater stream is any wastewater stream that has a HAP concentration of 5 ppmw or greater and a flow rate of 0.02 l pm or greater. The proposed process wastewater provisions include detailed flow charts to assist in determining applicability and control requirements. Controls must be applied to Group 1 wastewater streams, unless the source complies with the source- wide mass flow rate provisions of sections 63.132(c)(5) through (c)(7) of subpart G or if the Group 1 stream is returned to a process. Controls are not required for Group 2 wastewater streams. The proposed process wastewater provisions include equipment and work practice provisions for the transport and handling of wastewater streams between the point of generation and the wastewater treatment processes. These provisions include use of covers and enclosures and closed vent systems to route organic HAP vapors from the transport and handling equipment. The proposed provisions also include requirements for reduction of VOHAP concentration in wastewater streams. The required removal efficiencies are based on steam stripping. A variety of formats (e.g., percent reduction, effluent concentration, mass removal) are proposed to provide flexibility as described in section VII.E of this notice. Finally, air emissions routed through closed-vent systems from covers, enclosures, and treatment processes must be reduced by 95 percent or to a level of 20 ppmv. This reduction could be achieved using a combustion or recovery device. For demonstrating compliance with the various requirements, owners or operators have a choice of conducting performance tests or documenting engineering calculations. Appropriate compliance, monitoring, reporting, and recordkeeping provisions are included in the regulation. 6. Emissions Averaging Under the proposed subpart G, owners or operators may seek approval to comply by emissions averaging with any process vents, storage vessels, transfer racks, or wastewater streams. Equipment leaks are regulated under a separate subpart and may not be included in an emissions average at this time. a. Credit/debit system. To utilize emissions averaging under the proposed rule, the owner or operator must identify all the emission points that would be included in an average and estimate their allowable and actual emissions. The EPA has established a control efficiency, or percent emissions reduction, for the reference control technology for each kind of emission point. The owner or operator would use these reference control technologies to estimate the allowable emissions for each emission point. For each Group 1 point, the allowable emissions level is the residual emissions after application of a reference control technology. As a result, all Group 1 emission points that are not being controlled with the reference control technology or an equivalent are emitting more than their allowable emissions. These points are generating emission ''debits.'' Emission debits are calculated by subtracting the amount of emissions allowed by the standard for a given emission point from the amount of actual emissions for that point. If a Group 1 emission point is controlled by a device or a pollution prevention measure that does not achieve the control level of the reference control technology, the amount of emission debits will be based on the difference between the actual control level being achieved and what the reference control would have achieved. For example, if a pollution prevention measure that achieves a 70 percent reduction in emissions is used on a Group 1 wastewater operation, and the reference control would have achieved 98 percent emissions reduction on that emission point, then the debit would be equal to the difference, 28 percent of the uncontrolled emissions. The owner or operator must control other emission points to a level more stringent than what is required for that kind of point to generate emission ''credits.'' Emission credits are calculated by subtracting the amount of emissions that actually exist for a given emission point from the amount of emissions that would be allowed under subpart G, and then possibly applying a discount factor. The EPA is soliciting comments on use of a discount factor for emissions credits, and is proposing a range from 0 to 20 percent. To be in compliance, the owner or operator must be able to show that the source's emission credits were greater than or equal to its emission debits. Credits may come from: (1) Control of Group 1 emission points using technologies that EPA has rated as being more effective than the appropriate reference control technology; (2) Control of Group 2 emission points; and (3) Pollution prevention projects that result in control levels more stringent than what the standard requires for the relevant point or points. With the exception of some storage vessels and process vents, use of the reference control technology at a level more stringent than its assigned efficiency would not generate credits. If EPA cannot or has not determined the control efficiency of a control technology or work practice, it cannot be used to generate credits. The EPA will assign control efficiencies to new control devices or practices upon request. Specific restrictions on what can be counted as a credit are discussed in section VII.F of this preamble. Equations for calculating debits and credits are provided in section 63.150 of the proposed subpart G. b. Compliance. The proposed rule requires that emission averaging plans be reviewed as part of a source's Implementation Plan or operating permit application. The controls in the {pg 62617} averaging plan would be cited in a source's Implementation Plan or operating permit. Thus, to show compliance using emissions averaging, the owner or operator must prove both: (1) The appropriate controls have been applied and maintained; and (2) That the amount of emission credits and debits meet certain quarterly and annual requirements. 7. Recordkeeping and Reporting The proposed rule requires sources complying with subpart G to keep records and submit reports of information necessary to document compliance. Records must be kept for 5 years. The following five types of reports must be submitted to the Administrator: (1) Initial Notification, (2) Implementation Plan (if an operating permit application has not been submitted), (3) Notification of Compliance Status, (4) Periodic Reports, and (5) other reports. The requirements for each of the five types of reports are summarized below. a. Initial Notification. The Initial Notification is due 120 days after the date of promulgation for existing sources. For new sources, it is due 180 days before commencement of construction or reconstruction, or 45 days after promulgation of subpart G, whichever is later. The notification must list the chemical manufacturing processes that are subject to subpart G, and which provisions may apply (e.g., process vents, transfer operations, storage vessels, and/or wastewater provisions). A detailed identification of emission points is not necessary for the Initial Notification. However, the notification must include a statement of whether the source expects that it can achieve compliance by the specified compliance date. b. Implementation Plan. The Implementation Plan details how the source plans to comply with subpart G. An Implementation Plan would be required only for sources that have not yet submitted an operating permit application. Existing sources must submit the Implementation Plan at different times for emission points included in averages and emission points not included in averages. The Implementation Plan for emission points included in the average would be due 18 months prior to the date of compliance. The Implementation Plan for emission points not included in an emissions average would be due 12 months prior to the date of compliance. For new sources, Implementation Plans would be submitted with the Notification of Compliance Status. The information in the Implementation Plan should be incorporated into the source's operating permit application. The terms and conditions of the plan, as approved by the permit authority, would then be incorporated into the operating permit. For points included in emissions averaging, the Implementation Plan would include: An identification of all points in the average and whether they are Group 1 or Group 2 points; the specific control technique or pollution prevention measure that will be applied to each point; the control efficiency for each control used in the average; the projected credit or debit generated by each point; and the overall expected credits and debits. The plan must also certify that the same types of testing, monitoring, reporting, and recordkeeping that are required by the proposed rules for Group 1 points will be done for all points (both Group 1 and Group 2) included in an emissions average. If a source requests approval to monitor a unique parameter or use a unique recordkeeping and reporting system, a rationale must be included in the Implementation Plan. For emission points not included in an average, the Implementation Plan would include a list of emission points subject to the process vents, storage vessels, transfer operations, and wastewater provisions and whether each point is Group 1 or Group 2. The control technology or method of compliance planned for each Group 1 point must be specified. The plan must also certify that appropriate testing, monitoring, reporting, and recordkeeping will be done for each Group 1 point. If a source requests approval to monitor a unique parameter, a rationale must be included. c. Notification of Compliance Status. The Notification of Compliance Status would be submitted 150 days after the source's compliance date. It contains the information for Group 1 points and for all points in emissions averages, necessary to demonstrate that compliance has been achieved, such as: The results of any performance tests for process vents, transfer operations, and wastewater emission points; one complete test report for each test method used for a particular kind of emission point; TRE determinations for process vents; design analyses for storage vessels and wastewater emission points; site-specific ranges for each monitored parameter for each emission point and the rationale for the range; and values of all parameters used to calculate emission credits and debits for emissions averaging. d. Periodic Reports. Generally, Periodic Reports would be submitted semiannually. However, there are two exceptions. Quarterly reports must be submitted for all points included in an emissions average. In addition, if monitoring results show that the parameter values for an emission point are outside the established range for more than 1 percent of the operating time in a reporting period, or the monitoring system is out of service for more than 5 percent of the time, the regulatory authority may request that the owner or operator submit quarterly reports for that emission point. After 1 year, semiannual reporting can be resumed, unless the regulatory authority requests continuation of quarterly reports. All Periodic Reports would include information required to be reported under the recordkeeping and reporting provisions for each emission point. For emission points involved in emissions averages, the report would include the results of the calculations of credits and debits for each month and for the quarter. For continuously monitored parameters, the data on those periods when the parameters are outside their established ranges are included in the reports. Periodic Reports would also include results of any performance tests conducted during the reporting period and instances when required inspections revealed problems. Additional information the source is required to report under its operating permit or Implementation Plan would also be described in Periodic Reports. e. Other reports. Other reports would be submitted as required by the provisions for each kind of point. Other reports include: Reports of startup, shutdown, and malfunction; process changes that change the compliance status of process vents; and requests for extensions of repair and notifications of inspections for storage vessels and wastewater. C. Summary of Subpart H The following is a general summary of the requirements and concepts of the negotiated regulation. The reader is referred to the proposed standard for detail of specific provisions. 1. Applicability The standards would apply to equipment in VHAP service 300 or more hours per year associated with a production process manufacturing any of the 396 chemicals listed in the proposed standard that make or use as a reactant one of the organic VHAP's listed in Sec. 63.183 of the regulation. They would also apply to equipment {pg 62618} handling specific chemicals for a limited number of listed non-SOCMI processes. Petroleum refinery processes will not be covered by the proposed standard; a separate rulemaking will be conducted for those processes. It should be noted that although refinery processes would not be affected by this standard, organic chemical manufacturing units (e.g., benzene units) located on refinery property would be affected. The equipment subject to the proposed standard includes valves, pumps, connectors, compressors, pressure relief devices, open-ended lines, sampling connection systems, instrumentation systems, agitators, product accumulator vessels, and closed-vent systems and control devices. ''In VHAP service'' means that equipment contains or contacts a fluid that is 5 percent or greater VHAP's. The standards would also split the covered processes into five distinct groups to which the regulation would apply over time. The rule would apply to the first group 6 months after promulgation. Thereafter, the rule would become applicable to another group every 3 months until all the processes were covered. a. Pumps and valves. The regulation is structured similarly for pumps and valves. Standards for both would be implemented in three phases and both standards have associated QIP's. The first and second phases for both types of equipment consist of an LDAR program, with lower leak definitions in the second phase. The LDAR program involves a periodic check for organic vapor leaks with a portable instrument; if leaks are found, they must be repaired within a certain period of time. In the third phase, the periodic monitoring (a work practice standard) would be coupled with a base performance level (i.e., allowable percent leaking components). As part of the base program, pumps would require monthly monitoring using an instrument and weekly visual inspection. Valves would initially require quarterly monitoring, but the length of time between monitoring could be increased if the percent leaking valves demonstrate incrementally better performance, as specified in the rule, over the base performance level. Special provisions apply to pumps in food/medical service, pumps in polymerizing monomer service, ''leakless'' pumps, and unsafe- and difficult-to- monitor valves. Plants with less than 250 valves in VHAP service are subject only to LDAR and not the base performance level. If the base performance levels for a type of equipment are not achieved, based on a rolling average of monitoring results, owners or operators must, in the case of pumps, enter into a QIP, and in the case of valves may either enter into a QIP or implement monthly LDAR. The QIP is a concept that enables plants exceeding the base performance levels to eventually achieve the desired levels without incurring penalty or being in a noncompliance status. As long as the requirements of the QIP are met, the plant is in compliance. The basic QIP consists of information gathering, determining superior performing technologies, and replacing poorer performers with the superior technologies until the base performance levels are achieved. b. Connectors. The rule also provides for performance standards for connectors in terms of percent leaking connectors in each process unit. The negotiated standard for connectors is not phased in, i.e., the performance level applies as soon as the rule is effective for the process unit. Consistent achievement of the base performance level would result in monitoring being required less frequently. Failure to achieve the base performance level would cause the plant to remain in an annual monitoring cycle. Special provisions would apply to certain existing screwed connectors and to connectors that are inaccessible or unsafe to monitor or repair. c. Other equipment. Standards for compressors, open-ended lines, pressure relief devices, sampling connection systems, and closed vent systems and control devices remain essentially unchanged from existing regulations (see 40 CFR part 61, subpart V). Agitators must meet LDAR requirements, but not base performance levels. Pumps, valves, connectors, and agitators in heavy liquid service; instrumentation systems; and pressure relief devices in liquid service are subject to instrument monitoring only if evidence of a potential leak is found through sight, sound, or smell. Instrumentation systems consist of smaller pipes and tubing that carry samples of process fluids to be analyzed to determine process operating conditions. 2. Delay of Repair Under certain conditions delay of repair beyond the required 15 days may be acceptable. Examples of these situations include where: (1) A piece of equipment cannot be repaired without a process unit shutdown, (2) equipment is taken out of VHAP service, (3) emissions from repair will exceed emissions from delay of repair until the next shutdown, (4) pumps with SMS are replaced with DMS, and (5) valves assembly supplies have been depleted from stocks. 3. Alternative Standards Generally, an alternative means of emission limitation may be used if an owner or operator can demonstrate emission reductions equal to or better than that required by the standards. Specific alternative standards have been written for batch processes and enclosed buildings. Batch processes can choose either to meet similar standards to those for continuous processes, with monitoring frequency prorated to time in use of VHAP, or to periodically pressure test the entire system. Enclosed buildings may forego monitoring if the building is kept under a negative pressure and all emissions are routed through a closed vent system to an approved control device. 4. Test Methods and Procedures The standards would retain the use of Method 21 to detect leaks. Method 21 requires a portable organic vapor analyzer to monitor for leaks from equipment in use. A ''leak'' is a concentration specified in the regulation for the type of equipment being monitored and is based on the instrument response to methane (the calibration gas) in air. The observed screening value may require adjustment for response factor relative to methane if the weighted response factor of the stream exceeds a specified multiplier. Method 18 is to be used to determine organic content of a process stream. Test procedures using either a gas or a liquid for pressure testing the batch system are specified to detect for leaks. 5. Recordkeeping The standards would require a readily accessible recordkeeping system. Records required include identification of equipment that would be covered by the standards, identification of equipment that is found to be leaking during a monitoring period and when it is repaired, testing associated with batch processes, design specifications of closed vent systems and control devices, test results from performance tests or testing process streams for organic content, and information required by equipment in QIP. Other recordkeeping requirements also apply, and the reader is referred to Section 63.181. 6. Reporting Owners and operators would be required to submit an initial report that describes the source and all equipment {pg 62619} subject to these standards. Every 6 months, a report must be submitted that summarizes the results of monitoring and performance tests conducted during that period, changes to the process unit, changes in monitoring frequency or monitoring alternatives, and/or initiation of a QIP. Reports can be submitted on electronic media that is compatible with the system used by the Administrator or the State permitting authority. IV. SUMMARY OF IMPACTS OF PROPOSED RULE This section presents the environmental, energy, cost, and economic impacts resulting from the control of HAP emissions under the proposed rule. It is estimated that approximately 370 sources and 1,050 chemical manufacturing processes would be required to apply controls by the proposed standards. The analysis of impacts was performed assuming the requirements of the rule would be met through point-by-point compliance instead of emissions averaging. Under emissions averaging, the emission reductions would be approximately the same or greater, but presumably, since emission averaging is voluntary, the costs of control would be lower. It is not possible to quantify the potential cost savings from emissions averaging because the savings will depend on how many sources use emissions averaging and the mix of emission points and controls that are included in emissions averages. At this time, the EPA does not have the data necessary to estimate these parameters. The EPA requests comments on the potential savings from the proposed emissions averaging provisions. Impacts are presented relative to a baseline set at the level of control in the absence of the proposed rule. The estimates include the impacts of applying control to: (1) Existing emission points and (2) additional emission points from SOCMI process units that are expected to begin operation over a 5-year period. Thus, the estimates represent annual impacts occurring in the fifth year. Assuming a SOCMI- wide growth rate of 3 percent each year over a 5-year period, national impacts for the emission points that will be added in the first 5 years of the rule are estimated to be 16 percent of total national impacts in the fifth year. The environmental, energy, cost, and economic impacts are discussed in greater detail in the BID, Volumes 1A, 1B, and 1C. Specifically, the impacts estimation methodology is discussed in Volume 1A; the performance and costing methodologies of the evaluated control technologies are discussed in Volume 1B; and the sources of other environmental and energy impacts are discussed in Volume 1C. A. Environmental Impacts Environmental impacts include the reduction of HAP and VOC emissions, increases in other air pollutants, and decreases in water pollution and solid waste resulting from the proposed rule. Under the proposed rule, it is estimated that emissions of HAP would be reduced by 475,000 Mg/yr (522,500 tons/yr) and the emissions of VOC's would be reduced by 986,000 Mg/yr (1,085,000 tons/yr) (see Table 1). TABLE 1.-National Primary Air Pollution Impacts in the Fifth Year sup a Emission points Equipment leaks Baseline emissions (Mg/yr) HAP 66,000 VOC sup b 84,000 Emission reductions (Mg/yr) HAP 53,000 VOC sup b 68,000 (percent) HAP 80 VOC sup b 81 Emission points Process vents Baseline emissions (Mg/yr) HAP 317,000 VOC sup b 551,000 Emission reductions (Mg/yr) HAP 292,000 VOC sup b 460,000 (percent) HAP 92 VOC sup b 83 Emission points Storage vessels Baseline emissions (Mg/yr) HAP 15,200 VOC sup b 15,200 Emission reductions (Mg/yr) HAP 5,560 VOC sup b 5,560 (percent) HAP 37 VOC sup b 37 Emission points Wastewater collection and treatment operations Baseline emissions (Mg/yr) HAP 198,000 VOC sup b 728,000 Emission reductions (Mg/yr) HAP 124,000 VOC sup b 452,000 (percent) HAP 63 VOC sup b 62 Emission points Transfer loading operations Baseline emissions (Mg/yr) HAP 900 VOC sup b 900 Emission reductions (Mg/yr) HAP 500 VOC sup b 500 (percent) HAP 56 VOC sup b 56 Total Baseline emissions (Mg/yr) HAP 597,000 VOC sup b 1,380,000 Emission reductions (Mg/yr) HAP 475,000 VOC sup b 986,000 (percent) HAP 80 VOC sup b 71 sup a These numbers represent estimated values for the fifth year. Existing emission points contribute 84 percent of the total. Emission points associated with chemical manufacturing process equipment built in the first 5 years of the standard contribute 16 percent of the total. sup b The VOC estimates consist of the sum of the HAP estimates and the non-HAP VOC estimates. Estimates of baseline emissions are presented in conjunction with emissions reductions estimates to better illustrate the level of control being achieved by the rule. Baseline emissions take into account the current estimated level of emissions control, based on State and Federal regulations, for each SOCMI emission point. As a result, baseline emissions reflect the level of control that would be achieved in the absence of the proposed rule. The baseline emission estimates in Table 1 include the extrapolation of estimates for well-characterized processes to account for processes that could not be characterized. Consequently, the Table 1 estimates contain considerable uncertainty and are presented only to provide an estimate of the total nationwide impact of the proposed rule. Regulatory alternatives were developed using information only for the well-characterized processes and are discussed in section VII.A.2 of this preamble. On average, SOCMI sources generate over twice as much VOC emissions as HAP emissions. Although the intent of the proposed rule is to reduce HAP emissions, the control of HAP's also results in the control of non-HAP VOC's. The control requirements of the HON would result in reduction of 80 percent of HAP emissions and 71 percent of VOC emissions beyond the baseline control level. There would be a very slight increase in emissions of CO and NO sub x, relative to other sources of these pollutants, resulting from the on-site combustion of fossil fuels as part of control device operations. Additional emissions of NO sub x and CO resulting from increased electricity demand are not included in the impacts presented. Under the proposed rule, estimates of increased emissions of CO and NO sub x are 1,570 Mg/yr (1,730 tons/yr) and 15,700 Mg/yr (17,300 tons/yr), respectively (see Table 2). Table 2.- National CO and NO sub x Emissions Impacts in the Fifth Year sup a Emission points Equipment leaks CO Emissions sup b (Mg/yr) 0 NO sub x Emissions sup b (Mg/yr) 0 Emission points Process vents sup c CO Emissions sup b (Mg/yr) 1,490 NO sub x Emissions sup b (Mg/yr) 15,100 Emission points Storage vessels CO Emissions sup b (Mg/yr) 0 NO sub x Emissions sup b (Mg/yr) 0 Emission points Wastewater collection and treatment operations sup d CO Emissions sup b (Mg/yr) 80 NO sub x Emissions sup b (Mg/yr) 600 Emission points Transfer loading operations sup c CO Emissions sup b (Mg/yr) sup e NO sub x Emissions sup b (Mg/yr) sup e Total CO Emissions sup b (Mg/yr) 1,570 NO sub x Emissions sup b (Mg/yr) 15,700 sup sup a These numbers represent estimated values for the fifth year. Existing emission points contribute 84 percent of the total. Emission points associated with chemical manufacturing process equipment built in the first 5 years of the standard contribute 16 percent of the total. sup b Emissions of these criteria pollutants are caused by operation of control devices. sup c Emissions result from the combustion of natural gas along with the organic HAP emission streams in incinerators and flares. sup d Emissions result from the combustion of various fossil fuels to generate steam for use in a steam stripper. sup e Emissions are less than 5 Mg/yr. The impacts for process vents and transfer operations are based on the assumptions that incinerators or flares are used to combust emission streams. To the extent non-combustion controls are used to achieve compliance with the standards, the actual CO and NO sub x emissions would be lower. Impacts for water pollution and solid waste were judged to be negligible and were not quantified as part of the impacts analysis. The basis for judging these impacts to be negligible is discussed in chapter 5.0 of BID Volume 1A. B. Energy Impacts Increases in energy use were estimated for steam, natural gas, and electricity. These three types of energy were compared and totaled on a BOE basis. Table 3 shows the estimated individual and total energy use increases. Table 3. National Energy Impacts in the Fifth Year sup a Emission points Equipment Leaks Electricity sup b (10 sup 6 kw/hr/yr) 0 (10 sup 3 BOE/yr) 0 Natural gas sup c (10 sup 9 Btu/yr) 0 (10 sup 3 BOE/yr) 0 Steam sup c (10 sup 9 Btu/yr) 0 (10 sup 3 BOE/yr) 0 Total sup d (10 sup 3 BOE/yr) 0 (TJ) 0 Process Vents Electricity sup b (10 sup 6 kw/hr/yr) 240 (10 sup 3 BOE/yr) 400 Natural gas sup c (10 sup 9 Btu/yr) 6,600 (10 sup 3 BOE/yr) 1,090 Steam sup c (10 sup 9 Btu/yr) 0 (10 sup 3 BOE/yr) 0 Total sup d (10 sup 3 BOE/yr) 1,500 (TJ) 9,600 Storage Vessels Electricity sup b (10 sup 6 kw/hr/yr) 15 (10 sup 3 BOE/yr) 25 Natural gas sup c (10 sup 9 Btu/yr) 0 (10 sup 3 BOE/yr) 0 Steam sup c (10 sup 9 Btu/yr) 0 (10 sup 3 BOE/yr) 0 Total sup d (10 sup 3 BOE/yr) 25 (TJ) 160 Wastewater Collection and Treatment Electricity sup b (10 sup 6 kw/hr/yr) 6 (10 sup 3 BOE/yr) 10 Natural gas sup c (10 sup 9 Btu/yr) 0 (10 sup 3 BOE/yr) 0 Steam sup c (10 sup 9 Btu/yr) 5,300 (10 sup 3 BOE/yr) 880 Total sup d (10 sup 3 BOE/yr) 890 (TJ) 5,700 Transfer Loading Operations Electricity sup b (10 sup 6 kw/hr/yr) sup e (10 sup 3 BOE/yr) 0 Natural gas sup c (10 sup 9 Btu/yr) 50 (10 sup 3 BOE/yr) 10 Steam sup c (10 sup 9 Btu/yr) 0 (10 sup 3 BOE/yr) 0 Total sup d (10 sup 3 BOE/yr) 10 (TJ) 60 Total Electricity sup b (10 sup 6 kw/hr/yr) 260 (10 sup 3 BOE/yr) 440 Natural gas sup c (10 sup 9 Btu/yr) 6,650 (10 sup 3 BOE/yr) 1,100 Steam sup c (10 sup 9 Btu/yr) 5,300 (10 sup 3 BOE/yr) 880 Total sup d (10 sup 3 BOE/yr) 2,420 (TJ) 15,500 sup a These numbers represent estimated values for the fifth year. Existing emission points contribute 84 percent of the total. Emission points associated with chemical manufacturing process equipment built in the first 5 years of the standard contribute 16 percent of the total. sup b Conversion to BOE assumed a power plant heat rate of 10,000 Btu/kw-hr, heating value for oil of 144,400 Btu/gal, and 42 gal/bbl. sup c Conversion to BOE assumed a heating value for oil of 144,400 Btu/gal and 42 gal/bbl. sup d Due to rounding error, column totals may be slightly different. sup e Electricity usage is less than 1 * 10 fn 6 kw-hr/yr. Under the proposed rule, estimates for total energy use are 260 million kw- hr/yr of electricity, 6,650 billion Btu/yr of natural gas, and 5,300 billion Btu/yr of steam. This equates to 15,500 TJ/yr (2.4 million BOE/yr). C. Cost Impacts Cost impacts include the capital costs of new control equipment, the cost of energy (supplemental fuel, steam, and electricity) required to operate control equipment, and operation and maintenance costs. Generally, cost impacts also include cost savings generated by reducing the loss of valuable product in the form of emissions. Average cost effectiveness ($/Mg of pollutant removed) is also presented as part of cost impacts. Average cost effectiveness is determined by dividing the annual cost by the annual emission reduction. Under the proposed rule, it is estimated that total capital costs would be $347 million (1989 dollars), and total annual costs, excluding the cost savings attributable to equipment leaks, would be $134 million (1989 dollars) per year (see Table 4). The impacts presented for the annual costs of controlling emissions from equipment leaks reveal a cost savings. By avoiding losses from equipment leaks, product is saved. The impacts analysis indicates that the value of the product that is saved is higher than the costs incurred from applying the control required by the rule. However, due to uncertainty about the true nature and magnitude of the cost savings from controlling equipment leaks, the total impacts estimate for the rule does not reflect this estimated cost savings. Instead, the total national annual control cost estimate, as presented in Table 4, has no cost or cost savings element for equipment leaks. Table 4 . National Control Cost Impacts in the Fifth Year sup a Emission points Equipment Leaks Total capital costs (10 fn 6 $) 110 Total annual costs (10 fn 6 $/yr) (1) Average HAP cost effectiveness sup b ($/Mg HAP) (20) Average VOC cost effectiveness sup b ($/Mg VOC) (10) Process Vents Total capital costs (10 fn 6 $) 92 Total annual costs (10 fn 6 $/yr) 75 Average HAP cost effectiveness sup b ($/Mg HAP) 260 Average VOC cost effectiveness sup b ($/Mg VOC) 160 Storage Vessels Total capital costs (10 fn 6 $) 49 Total annual costs (10 fn 6 $/yr) 19 Average HAP cost effectiveness sup b ($/Mg HAP) 3,400 Average VOC cost effectiveness sup b ($/Mg VOC) 3,400 Wastewater Collection and Treatment Operations Total capital costs (10 fn 6 $) 86 Total annual costs (10 fn 6 $/yr) 35 Average HAP cost effectiveness sup b ($/Mg HAP) 280 Average VOC cost effectiveness sup b ($/Mg VOC) 80 Transfer Loading Operations Total capital costs (10 fn 6 $) 10 Total annual costs (10 fn 6 $/yr) 5 Average HAP cost effectiveness sup b ($/Mg HAP) 10,000 Average VOC cost effectiveness sup b ($/Mg VOC) 10,000 Total sup c Total capital costs (10 fn 6 $) 347 Total annual costs (10 fn 6 $/yr) 134 Average HAP cost effectiveness sup b ($/Mg HAP) 280 Average VOC cost effectiveness sup b ($/Mg VOC) 140 sup a These numbers represent estimated values for the fifth year. Existing emission points contribute 84 percent of the total. Emission points associated with chemical manufacturing process equipment built in the first 5 years of the standard contribute 16 percent of the total. sup b Average cost effectiveness values are determined by dividing total annual costs by total annual emission reduction. sup c Except for the Total Capital Costs column, the total figures do not include an element for equipment leaks because the analysis of equipment leak requirements indicated a cost savings. It is expected that the actual compliance cost impacts of the proposed rule would be less than those presented, but it is not possible to quantify the amount. This is because cost estimates for some kinds of emission points were made assuming a separate control device would be constructed for each emission point. In reality, some operators will duct emissions from several of these emission points to a common control device, upgrade an existing control device, use other less expensive control technologies, implement pollution prevention technologies, or use emissions averaging. All of these options would reduce the estimated costs while achieving the same emission reductions. The effect of such practices on the national costs could not be estimated because the ability to use any of these practices is highly site-specific and data were not available to estimate how often the lower cost compliance practices could be utilized. D. Economic Impacts The economic impact analysis assumed that controls would be applied to all emission points and did not consider the selected applicability criteria. The fifth year annualized costs used in the economic impact analysis are $359 million per year. This is about 270 percent greater than the estimated costs for the proposed rule, which are $134 million per year. Therefore, the economic impacts calculated in this analysis are greater than the impacts associated with the proposed rule. Because many SOCMI chemicals are used as raw materials in the production of other SOCMI chemicals, the economic impact analysis looked at cumulative costs of control for each of the over 400 SOCMI chemicals listed in Subparts F and H. About 88 percent of the chemicals are estimated to have a cost increase of less than 10 percent; more than 75 percent have cost increases less than 3 percent. Approximately 12 percent of the chemicals analyzed incur a cost increase of over 10 percent. All but 5 of these chemicals have annual national production of less than 10 million kilograms (11,000 tons) and are therefore low volume chemicals. Two- thirds of the SOCMI chemicals have production over 10 million kilograms (11,000 tons). Market analyses for a subset of 20 of the chemicals estimated price increases from 0.3 percent to 4.8 percent and quantity decreases from 0.1 percent to 4 percent. The market analyses lead to the conclusion that percentage quantity decreases will be less than the percentage cost increases due to the regulation. The market analyses indicate that severe disruption of the industry is an unlikely result. Even for the total control cost scenario analyzed, which has a total cost of over 50 percent more than the anticipated costs associated with the proposed standard, significant numbers of business closures are not expected. The diversity of chemical producers (most sources are involved in the production of several chemicals) decreases the likelihood of plant closure as a result of the regulation. A more likely consequence of the regulation is a change from a chemical manufacturing process with a higher cumulative control cost to a process with a lower control cost. The impact for the low volume chemicals is the most uncertain. The cost estimates for these chemicals involve more uncertainty and, in many cases, industry profile information specific to the manufacturers of these chemicals was not available. Many of the low volume chemicals can be considered specialty chemicals. Generally, there is a lack of viable substitutes for specialty chemicals. In addition, the cost of specialty chemicals is usually only a small portion of the cost of the final good made with the specialty chemical. For these two reasons, a price increase for a specialty chemical is less likely to lead to a business closure or a production cutback than a price increase for a large volume chemical. This decreases the likelihood of large quantity impacts or closures. The RIA addresses the benefits, costs, and economic impact of the regulation. Because benefits could only be addressed qualitatively, the RIA is not able to provide guidance as to which regulatory option optimizes net benefits. However, the RIA does summarize the types of benefits associated with the reduction of HAP's, VOC's, and particulate matter formed from VOC's. The consideration given to benefits, cost, and impacts estimates is discussed further in section VII.A.2 of this notice. V. Emissions and Impacts Estimation Methodology A. Overview This section of the preamble explains the methodology used for estimating emissions and control impacts for existing sources. Emissions and control impacts estimates for new sources are derived from the estimates for existing sources. This section provides a broad overview of the methodology; details are presented in the BID, Volume 1A, chapter 4.0. The objective of estimating emissions and control impacts was to compare the characteristics of alternative standard levels. The following types of impacts were estimated: emission reductions, control costs, energy impacts, secondary air pollution impacts such as NO sub x and CO emissions, water pollution, and solid waste generation. Although site-specific data on every chemical manufacturing process were not available, estimates of emissions and control impacts could be derived using a model emission point approach. The model emission point approach could be used because the emission mechanisms and applicable control technologies are well understood for the kinds of emission points regulated in the HON. Furthermore, these characteristics are similar across SOCMI chemical manufacturing processes. The impacts analysis involved three steps: (1) Development of a data base characterizing the SOCMI, (2) development and assignment of model emission points for each kind of emission point, and (3) calculation of emissions and control impacts. The characterization of the SOCMI primarily involved identifying the specific routes, reactants, and process technologies used to produce a chemical and the corresponding SOCMI chemical manufacturing processes. In addition, information on existing State and Federal regulations was compiled for each kind of emission point to determine the baseline control requirements applicable to SOCMI chemical manufacturing processes. Model emission points were developed to represent each kind of emission point in the SOCMI. The models were developed to emphasize those characteristics that most influence emissions, control costs, energy needs, and secondary environmental impacts. These models were applied to individual chemical manufacturing processes in the SOCMI data base using decision rules based on the level of information in the data base and the specificity of a given model. The data base and model emission points used to estimate the impacts of the HON are based on published literature and information that EPA has collected during other rulemaking efforts including NSPS for air oxidation processes, distillation operations, reactor processes, volatile organic liquid storage, and equipment leaks; and NESHAP for vinyl chloride and benzene. Some information on wastewater collection and treatment operations is based on the document ''Industrial Wastewater Volatile Organic Compound Emissions-Background Information for BACT/LAER Determinations'' (EPA 450/3-90-004). Some additional information was obtained on wastewater operations and transfer loading operations by {pg 62621} requesting it from the industry under authority of section 114 of the Act. Surveys were not conducted on the other kinds of emission points because it was judged that they would not add materially to the analysis. Baseline emissions, those emissions that would occur in the absence of the HON, were estimated using calculation algorithms based on known, previously published, well established methodologies. The baseline emissions estimates were influenced by the production capacities of the chemical manufacturing processes in the SOCMI data base and the control requirements in existing Federal and State regulations. It was assumed that all chemical manufacturing processes would be in compliance with any applicable existing Federal or State air pollution regulations. The impacts of the alternative standard levels were estimated using calculation algorithms previously developed for commonly used control technologies such as incinerators, flares, condensers, tank improvements, and steam strippers. The impacts estimates are based on average, representative, or typical emissions and control requirements for each kind of emission point. Thus, the estimates do not reflect the impacts that would be observed at any particular chemical manufacturing process. However, they do provide a reasonable estimate of nationwide emission reductions and control costs. In addition, these estimates are representative of the range of impacts that the SOCMI might incur under alternative standards. More information concerning characterization of the SOCMI, model assignment, and estimation of impacts can be found in the BID, Volume 1A. Additional information concerning control technologies and costs can be found in the BID, Volume 1B. More detailed information concerning model development can be found in the BID, Volume 1C. B. Control Technologies for Impacts Estimation Before estimating the impacts of the proposed standard, EPA considered several different control technologies including, among others, combustion devices, product recovery devices, and pollution prevention opportunities. The control technologies selected for inclusion in the analysis were chosen because they are the most stringent control technologies that are universally applicable to emission points in the SOCMI and they can achieve emission reductions at least as stringent as the MACT floor. While the selected controls were used as the basis of the control impacts estimates, the proposed standards are written using formats that would allow use of other control technologies if the equivalent emissions reduction is achieved. Listed below are the control technologies selected as the basis for the impacts calculations for each kind of emission point. 1. Process Vents Since combustion devices are typically the most efficient control systems and the only types of control systems that can be applied universally to all process vents, they are used as the basis for the impacts analysis. Emission reduction of 98 percent is achievable using combustion devices. For halogenated streams, impacts were calculated based upon the use of a thermal incinerator and acid gas scrubbing. For nonhalogenated streams, impacts were calculated based upon the use of a thermal incinerator or a flare, whichever is less expensive in a given situation. 2. Storage To estimate impacts for storage vessels, it was assumed that storage vessels will be controlled with internal floating roofs or refrigerated condenser systems depending on chemical properties of the stored liquid. Emission reduction of 95 percent is achievable through the use of tank improvements (e.g., installation of internal floating roof with appropriate seals and fittings) or condensers. For halogenated streams, impacts were calculated based upon the use of refrigerated condensers. For nonhalogenated streams, impacts were calculated based upon tank improvements unless the chemical was incompatible with floating roof materials. In this case, use of a condenser was assumed. 3. Wastewater Operations Steam strippers have been evaluated as the basis for estimating impacts for wastewater treatment because steam strippers are efficient treatment systems for the removal of volatile HAP's from wastewater streams, and are also the most widely applicable control technology for wastewater streams. This technology achieves emission reductions of 0 to 99 percent, based on the chemical characteristics (e.g., strippability) of the wastewater stream. However, 95 to 99 percent reduction can be achieved for the majority of compounds regulated by the HON. Impacts were calculated based upon the use of a steam stripper followed by an air emissions control device. The use of enclosed collection and transport systems to suppress emissions up to treatment in the steam stripper was also assumed. 4. Transfer Operations As with process vents, combustion was selected as the most stringent and universally applicable control technology for control of emissions from transfer operations. Emission reduction of 98 percent is achievable based upon the use of capture systems and combustion technologies. For halogenated streams, impacts were calculated based upon the use of a thermal incinerator and acid gas scrubbing. For nonhalogenated streams, impacts were calculated based upon the use of a thermal incinerator or a flare, whichever is less expensive in a given situation. 5. Equipment Leaks Estimates of impacts for controlling emissions from equipment leaks were based on the use of LDAR at various action levels specified in the negotiated rule for equipment leaks. Emission reductions will vary from source to source as the mix of equipment components differs. Overall, sources are expected to achieve approximately 88 percent emission reduction. C. National Emissions and Control Cost Calculations The calculation of emissions and control impacts for four of the five kinds of emission points was performed for individual chemical manufacturing processes. However, for transfer loading operations, emissions and control impacts were estimated at the source level based on total chemical loading throughput at the rack since it is industry practice to have source-wide transfer racks rather than a dedicated rack for each chemical manufacturing process. Some chemical manufacturing processes were not as well characterized as others. In these cases, information was extrapolated to derive estimates of emissions, control costs, and other impacts as described in chapter 4.0 of the BID, Volume 1A. National impacts for existing sources were determined by aggregating the impacts across all the chemical manufacturing processes (or facilities in the case of transfer racks). As previously described in section IV, estimates of emissions and control impacts for new sources are derived from the results for existing sources, and were calculated to be 16 percent of the total national impacts in the fifth year.{pg 62623} VI. Rationale for Provisions in Subpart F This section describes the rationale for the selection and definition of the SOCMI source category as well as the proposed policies regarding area sources and pilot plants. The rationale for the selection and definition of the seven non-SOCMI processes will be discussed in section VIII, ''Rationale for Provisions in subpart H.'' A. Selection of Source Categories 1. Selection and Definition of the Synthetic Organic Chemical Manufacturing Industry The initial source category list (57 FR 31576, July 16, 1992), required by section 112(c) of the Act, identifies source categories for which NESHAP are to be established. This list includes all major source categories of HAP's known to EPA at this time, and all area source categories for which a finding of adverse effects warranting regulation has been made. The source category list identifies the SOCMI as a source category because it contains major sources emitting at least 10 tons of any one HAP or more than 25 tons of any combination of HAP's annually. The SOCMI is a segment of the chemical manufacturing industry that includes the production of many high-volume organic chemicals. The products of SOCMI are derived from approximately 10 petrochemical feedstocks. Of the hundreds of organic chemicals that are produced by the SOCMI, some are final products and some are the feedstocks for production of other non-SOCMI chemicals or synthetic products such as plastics, fibers, surfactants, pharmaceuticals, synthetic rubber, dyes, and pesticides. Production of such non-SOCMI end products is not considered to be part of SOCMI production and, as a result, the proposed standards would not apply to downstream synthetic products industries, such as rubber production or polymers production, that use chemicals produced by SOCMI processes. For this rule, the EPA has defined the source category as consisting of chemical manufacturing processes that produce one or more of the 396 chemicals listed in Sec. 63.105 of subpart G, or one or more of the chemicals listed in Sec. 63.184 of subpart H. The production of these chemicals is believed to involve emissions of organic HAP's. These chemicals were identified from the literature describing SOCMI production processes, reactants, and products. A chemical was listed if organic HAP's could be used as reactants or produced in the manufacture of the listed chemical. However, EPA recognizes that chemicals on the list can be produced by processes that do not use an organic HAP as a reactant. In such cases, even if the chemical is on the list, those processes producing the chemical that do not use an organic HAP as a reactant or produce a HAP are not considered to be included in the source category. In previous rules for SOCMI, EPA considered by-products, co-products, and intermediates to be products of a process. In implementation of these existing rules, there has been confusion over the meaning of the terms ''product'' and ''to produce'' and the correct way to decide whether a source ''produces'' a listed chemical and is subject to the standard. This confusion arises because of the complexity and diversity of SOCMI and the highly integrated nature of the chemical industry. Since the operations that would be regulated in the HON can also be part of an integrated group of operations dedicated to the production of non- SOCMI products such as pesticides or polymers, the industry is concerned that companies would have difficulty determining which standards apply to which process. Regulatory agencies have also found it difficult to determine applicability. Most of the organic chemical manufacturing processes that would be subject to the proposed standards for SOCMI are classified in the four-digit SIC codes 2865, Cyclic Organic Crudes and Intermediates and Organic Dyes and Pigments, and 2869, Industrial Organic Chemicals Not Elsewhere Classified. However, not all processes classified in these two SIC codes would be subject to the proposed rule. Because of this confusion, a different approach to defining applicability of this rule was developed. For the HON, applicability will be based on the primary product that is produced by a process or, where there is no primary product, on the intended purpose of the process. By- products, co-products, and isolated intermediates would not be considered in determining applicability since these were considered in development of the list of chemical products. The proposed standard would apply to air oxidation, reactor, and distillation processes that make as a product any of the listed 396 SOCMI chemicals. For the purposes of this rule, EPA does not consider wastes to be products. Also, impurities or trace contaminants that are coincidentally processed and are not isolated are not considered to be a product. This decisionmaking process, shown in Figures 1a and 1b, is based on the concept that applicability should be determined based on the primary product or purpose of the process, and is determined only once for each process. The primary product would be determined by the product that represents the largest percentage of the total mass produced by the process. {SEE ILLUSTRATION(S) IN ORIGINAL DOCUMENT} Figure 1a identifies a series of logic tests that determine if the chemical manufacturing process would be subject to this proposed rule. Figure 1b addresses situations where a process produces two or more chemicals that are not predominant. It is expected that in the vast majority of cases, the applicability can be determined using the decisionmaking process in Figure 1a. A comparison of this new approach with the approach used in previous rules for SOCMI sources found that both approaches identify the same set of processes as being subject to this rule. A chemical manufacturing process is the group of equipment associated with air oxidation processes, reactor processes, and distillation operations that convert raw materials into one or more products. The chemical manufacturing process may include storage tanks, process equipment, transfer operations, and waste treatment predominantly used in the production of products. Examples of chemical manufacturing processes that would be subject to the proposed standard are: 1. A process that produces ethylbenzene as the product; 2. A process that produces phenol or acetone as the product from cumene; 3. A chemical manufacturing process that produces methylmethacrylate by purification of an impure feedstock received from another plant site (a distillation operation on a polymer unit would not be considered a chemical manufacturing process); 4. A chemical manufacturing process that produces methanol as the intended product; or 5. A chemical manufacturing process that produces chloroform as the intended product. Examples of processes that would not be considered subject to the proposed standard are: 1. A chemical manufacturing process that produces divinylbenzene as the predominant product and creates a benzene- containing waste that is sent to a benzene production process (the process producing benzene obviously would be subject to the proposed standard); or 2. A polymer process that produces polyethylene terephthalate that also generates an impure methanol stream (the polyethylene terephthalate process will be covered by the MACT standard for polymers and resins production). The EPA selected the proposed approach for defining applicability because it is consistent with previous standards for this industry and the supporting information for the proposed standard. Specifically, the use of a set of specific chemicals to define the category is consistent with standards established for SOCMI under Section 111 of the Act. The definition of ''product'' being proposed for this standard reflects the assumptions used in the identification of the list of 396 SOCMI chemicals and the need for consistency with the benzene waste NESHAP, (40 CFR Part 61, Subpart FF) and future Section 112 standards for wastes. The definition of product was modified from the definition used in the NSPS for SOCMI air oxidation, reactor, and distillation processes to improve implementation of this rule. In addition, the definition was modified to recognize that wastes are sold and that waste materials may be added to other processes for purposes of waste minimization and treatment. These modifications were made to encourage waste minimization and to ensure consistency in what is considered a waste among Section 112 standards. The proposed standard also includes provisions that specifically exclude certain industrial activities from the standard. These provisions were added to prevent any ambiguity in the applicability of the standard. For example, petroleum refinery processes are not part of the SOCMI source category and are therefore not subject to this standard. Many refinery processes make multiple-chemical mixtures for use as fuels. These processes would not be covered by the HON even if one of the 396 chemicals is present in the mixture because EPA plans to regulate refinery processes under a separate MACT standard. For the same reason, refinery processes used to produce feedstocks that are supplied to SOCMI chemical manufacturing processes are not within the definition of SOCMI and are not subject to the HON. However, a SOCMI chemical manufacturing process that is located at a refinery and produces one or more of the 396 chemicals as a single chemical product (rather than a mixture) would be considered a SOCMI process and would be subject to the HON. Ethylene processes are also not considered to be part of the SOCMI source category. Ethylene processes, like refinery operations, generate mixed streams that provide the raw materials for subsequent chemical manufacturing processes. The chemical manufacturing processes that produce butadiene and benzene from these ethylene streams would be subject to the proposed rule. Ethylene processes will be evaluated under a separate MACT standard. Solvent reclamation units operated at hazardous waste TSDF facilities requiring a permit under subtitle C that are separate entities and not part of a SOCMI chemical manufacturing process are not covered by the proposed HON. Instead, these facilities will be considered for regulation under separate section 112 standards. Similarly, emission points that are typically associated with SOCMI processes but are not covered by the HON will be considered for regulation in separate standards. For example, separate MACT standards are planned for industrial cooling towers and boilers. 2. Exclusion of Area Sources A SOCMI chemical manufacturing process would be subject to the proposed standard only if it is part of a major source. As noted earlier in this notice, a major source is any stationary source or group of stationary sources located within a contiguous area and under common control that emits or has the potential to emit, considering controls, more than 10 tons per year of any HAP or more than 25 tons per year of total HAP. In implementation of other provisions of the Act, the EPA has defined ''potential to emit'' as the maximum capacity of a stationary source to emit a pollutant under its physical and operational design. Any physical or operational limitation on the capacity to emit includes any air pollution control equipment and restrictions on operations or on the amount of material processed, if such limitation is federally enforceable (see 40 CFR 52.21(4)). For the purpose of the proposed rule, the EPA considers ''potential to emit'' to be based on the same criteria. All operations at the site that emit or have the potential to emit would be considered when determining whether the source is major. Based on the information available on the SOCMI and emission estimates developed for this standard, the EPA does not believe that it would be feasible for any of the identified plant sites to be area sources. Consequently, the EPA has no information that can be used to determine whether area sources in the SOCMI source category would present a threat of adverse effects to human health or to the environment. While EPA is aware that some owners of SOCMI facilities believe there are area sources in the industry, EPA has no information regarding the basis for this belief. The EPA is requesting comment on whether there are any area sources in the SOCMI and if there are, whether the standard should apply to them. Information is requested on the nature of these sources; their chemical manufacturing processes; their potential {pg 62627} health effects; as well as estimates of the number, location, and emissions. 3. Research and Development Facilities The proposed standard would not apply to research and development facilities, such as laboratories and pilot plants, regardless of whether the facilities are located on the same site as a commercial chemical manufacturing process. Research and development facilities cover a wide range of operations and sizes from bench- top operations to small scale operating units. These facilities are operated under a very wide and changing variety of piping configurations, chemicals, chemical concentrations, and equipment to generate information that can be used to improve existing operations or to develop new products or design criteria for new production plants. Due to their very nature, there are frequent changes in the operations of research and development facilities. Although EPA has extensive experience with the SOCMI, EPA has limited information regarding operations of research and development facilities and the appropriate controls for these facilities. In particular, EPA is presently uncertain how to structure a standard for research and development facilities to avoid imposing extremely burdensome recordkeeping and reporting requirements. The EPA concluded, therefore, that it would be appropriate to establish a separate source category covering research and development facilities to ensure equitable treatment. Standards for such facilities may be developed at a later date. B. Selection of Emission Points The proposed standard applies to all emission points in organic HAP emitting SOCMI chemical manufacturing processes that are part of a major source. The bulk of HAP's from SOCMI processes can be characterized as being emitted from five kinds of emission points: Process vents from reactor processes, air oxidation processes, and distillation operations; storage vessels that store reactants or products; transfer racks used to load products into tank trucks or railcars; wastewater streams; and equipment leaks. 1. Process Vents Process vents are typically associated with product recovery systems in a chemical manufacturing process. Process vent emissions result primarily from the venting of VOC and HAP-containing inert gases and evacuation of equipment for vacuum processing. For this proposed standard, EPA has divided process vents into continuous and batch processes. The proposed provisions for vents would only apply to continuous process vents because the data upon which the standard is based, including the cost and control device performance, assume a continuous or near- continuous mode of operation. The EPA is considering developing a separate standard for process vents associated with batch processes. The process vent provisions apply to the point at which emissions are vented to the atmosphere. Emissions may be vented directly to the atmosphere, or one or more process vents streams may be directed through a recovery device before emissions are released to the atmosphere. A Group 1 process vent is a vent with a flow rate of 0.005 scm/min or greater, an organic HAP concentration of 50 ppmv or greater, and a TRE index value less than or equal to 1. Vents that do not meet these applicability criteria are Group 2 vents and are not required to apply additional controls. If a recovery device is present, the outlet of the final recovery device is where the vent stream tests are performed to determine whether it is a Group 1 or a Group 2 vent. Additional rationale for this method of defining process vents and where process vent emissions are measured is presented in the proposal preamble to the SOCMI reactors NSPS (55 FR 26953, June 29, 1990). Process vent streams that contain 0.005 weight-percent organic HAP or less are not regulated under the HON because it would be impractical to impose requirements for such small streams. 2. Storage Vessels The proposed provisions for storage vessels apply to each individual vessel storing one or more liquids that are organic HAP's because the point of emissions to the atmosphere is typically the individual vessel, and control technologies such as tank improvements apply to each individual storage vessel rather than to a group of vessels. Vessels that store liquids containing organic HAP's as impurities are not subject to the HON. Such liquids contain only trace levels of organic HAP's and no significant HAP emission reductions would be achieved by controlling them. An impurity is produced coincidentally with another chemical substance and is processed, used, or distributed with it. The storage vessel provisions would not apply to storage vessels permanently attached to motor vehicles such as trucks or railcars. The storage vessel provisions also do not apply to pressurized vessels designed to operate in excess of 204.9 kPa (29.7 psia) because such vessels operate under pressure and have no measurable emissions to the atmosphere. 3. Transfer Operations The provisions of Subpart G are applicable to each rack, leg, or arm of the rack at which organic HAP liquids are loaded. This selection was made because there may be multiple points of release for the emissions from transfer operations. In addition, HAP emissions may pass through the loading arm back into the rack and be discharged to the atmosphere through a different arm. Emissions of HAP's may also be released directly to the atmosphere from the vehicle being loaded. To ensure that vapors are collected and transferred to a control device, the standard requires that organic HAP be loaded only into DOT certified or vapor tight tank trucks and railcars. The transfer of liquids containing organic HAP's as impurities is not regulated by the HON. Such liquids contain only trace levels of organic HAP's and no significant HAP emission reductions would be achieved by controlling them. For purposes of the HON transfer operations provisions, an impurity is produced coincidentally with another chemical substance and is processed, used, or distributed with it. 4. Wastewater Collection and Treatment Operations In the manufacture of many chemicals, wastewater streams containing organic HAP compounds are generated. These organic compounds can volatilize and be emitted to the atmosphere if the wastewater is managed in an open system or vented to the atmosphere. There are three categories of HAP- containing wastewaters produced at SOCMI facilities: (1) Process wastewaters; (2) maintenance wastewaters; and (3) water contaminated through leaks from heat exchangers and condensers. Because each category of wastewater is generated in a different way, control of emissions also differs. Recognizing that different approaches are needed to reduce the burden of control and ensure good practices for control of emissions from wastewater, EPA has developed separate provisions for each of the three categories. Wastewaters generated during maintenance turnaround and routine maintenance activities would have to be properly managed to minimize air emissions. As part of the startup, shutdown, and malfunction plan for the source, the owner or operator would {pg 62628} have to specify procedures that will be followed during maintenance turnarounds to ensure that wastewaters are collected, treated, and managed in a manner that minimizes emissions to the atmosphere. If the procedures in the plan are followed during the maintenance turnaround, the owner or operator only needs to document that the procedures specified in the plan were followed. The startup, shutdown, and malfunction plan would also have to include a description of the procedures that would be followed to properly manage process fluids drained from equipment during routine maintenance activities. The proposed provisions for control of emissions from contaminated cooling water are based on a leak detection and repair program to minimize leakage of process fluids into cooling water. The intent of these provisions is to ensure proper operation and maintenance of the process. The leak detection and repair provisions require monitoring of cooling systems for significant increases in HAP content in the cooling water. If concentration increases above the action level are detected, then the leaking equipment would have to be repaired or by-passed within a specific time period. The EPA is requesting comments and data on what is an appropriate action level and what time period should be allowed for repairs. The proposed provisions present a range of possible values for action levels and time periods for repairs. The proposed provisions for process wastewater apply to wastewaters which, during manufacturing or processing, come into direct contact with or result from the production of process fluids. Applicability of the proposed provisions is determined at the point of generation of the wastewater. (The point of generation is the location where the wastewater leaves the process equipment and enters waste management units.) This point was selected because: (1) At this point, the production process ends and the wastewater collection system begins; (2) This is the point where the concentration of organics is highest, therefore allowing judgment to be made regarding the need for control or monitoring of downstream treatment processes; and (3) After this point, there is a potential for emissions from wastewater streams to occur. 5. Equipment Leaks The term equipment leaks refers to the loss of process fluid through the sealing mechanism separating the process from the atmosphere. Equipment that can leak process fluid includes the valves, pumps, connectors, compressors, agitators, pressure relief devices, sampling connection system, open-ended lines or valves, product accumulator vessels, and instrumentation systems that are associated with all operations of the chemical production process. Based on the negotiated agreement, equipment that only contacts or contains process materials that are less than 5 percent HAP or are operated in HAP service for 300 hrs/yr or less is not subject to subpart H. VII. Rationale for Provisions in Subpart G A. Selection of Emission Control Requirements The Act specifies that EPA, in determining the MACT level of control for sources regulated under section 112, must select emission control requirements that are at least as stringent as, or more stringent than, the emission control level identified as the floor. As a result, EPA began the process of selecting control requirements for the HON by determining the floor for the sources that would be subject to the HON. Once the floor was established for both new and existing sources, EPA considered additional control for each kind of emission point and the source as a whole taking into consideration the criteria enumerated in section 112(d) of the Act: Cost of achieving such reductions, any non-air quality health and environmental impacts, and energy requirements. This section of the preamble describes the process EPA used to determine the floors for new and existing sources, the criteria EPA used to evaluate additional control requirements and the outcome of EPA's floor analysis and control selection process for each kind of emission point in the source. Section V of this preamble gives an overview of the data base used in this process, and memoranda in the docket provide a detailed description of the methodology and data used to derive the floors for new and existing sources. 1. Overview of the Process and Factors Considered For SOCMI, what distinguishes a well-controlled facility is not only the type of control equipment used, but also the number of emission points that are controlled. Facilities differ in the number, combination, and design of their chemical manufacturing processes; the production capacities of their processes; the particular chemicals manufactured; and the control equipment used. Consequently, although SOCMI consists of similar kinds of emission points and the same controls are applicable, actual emissions and characteristics of SOCMI facilities vary widely from plant site to plant site. Due to this diversity, no ''typical'' source could be identified that would be representative across the source category. This diversity affected the approach used to define the floor for existing and new sources. Specifically, this diversity precluded the use of mass emission rate as a measure of performance since a mass rate based on an ''average'' source could require no control at some sources and be unachievable at other sources. As with previous rules for the SOCMI category, the EPA used the weight percent reduction achieved by the control device as the most appropriate measure of best performing technology (55 FR 26963). These characteristics of the controlled emission points and the control efficiencies for each kind of emission point were then combined to develop the source-wide floor. This process can be expected to result in a floor determination that is at least as stringent as that which would have been generated with actual source-wide data. The information EPA used in determining the source-wide floor consisted of the estimates of the number and characteristics of the model emission points, the emission control requirements currently in place for each point based on information available to EPA, and the expected control efficiencies for the control technology. As discussed earlier in this notice, EPA used data on the control requirements in existing State and Federal regulations to identify those emission points that must be controlled in the absence of this rule and to identify the required controls. (The regulations were used as a surrogate for actual data on the control levels achieved in practice. In this analysis, EPA assumed all facilities would be in compliance with all applicable regulations.) In the analysis of existing regulations, EPA found that where State and Federal rules require controls on emission points, they typically require use of the most effective control technologies (or performance levels) that are generally applicable. These control technologies are the same as the controls required in previous NSPS standards for the SOCMI and are widely accepted by the industry and regulatory agencies as technologies appropriate as bases for emission standards. These control technologies have been designated as the ''reference {pg 62629} control technologies'' for the purpose of the rule being proposed today. The reference control technologies (and their performance levels) specified in the proposed rule reflect information and knowledge of SOCMI that EPA has been developing since 1976. Through development of standards under section 111 and section 112 for SOCMI, EPA has developed extensive knowledge of the range of demonstrated control technologies applicable to SOCMI and the expected performance of these technologies. The selected technologies identified as the basis for the reference control requirements are: -98 percent combustion control for process vents and transfer operations. For halogenated streams, 98 percent control is achieved using a thermal incinerator, boiler, or process heater, plus acid gas scrubbing. For nonhalogenated streams, 98 percent control is achieved using either a thermal incinerator, a flare, or a boiler or process heater. -95 to 98 percent volatile organic HAP control of wastewater (for highly volatile chemicals) or control to a target concentration using a controlled steam stripper or other treatment technology. The actual control depends on individual chemical properties. -Approximately 95 percent control of storage tanks through tank modifications or application of a vapor recovery system and control device. The actual efficiency depends on the individual chemical properties. The above control technologies and work practices were selected as the technological basis for the HON since EPA is not aware of any demonstrated control technologies and operations that would perform with higher efficiencies and also be universally applicable to SOCMI. In many cases, application of the reference control technology is already required by either an existing State or Federal regulation. To determine the source-wide floor for existing sources, EPA next examined the supporting information to identify the characteristics of the top 12 percent of the source category that applied the reference control technology to the smallest emission points. This analysis was done for each kind of emission point. The characteristics used to identify groups of emission points were physical parameters such as flow rate, HAP concentration, and vapor pressure. All the identified groups of emission points controlled by the reference control technology were then combined to define the weighted average percent reduction achieved by these best performing 12 percent of the sources. The results of this analysis are described in the next section as part of the discussion of specific considerations for each kind of emission point. A similar method was used to determine the source- wide floor for new sources. For each kind of emission point, the characteristics of the smallest emission point controlled by the reference control technology were identified as the means for determining the best controlled similar source. Again, the source- wide floor was determined by the combination of control levels for all emission points. Once the floors were established, EPA considered whether to establish a standard that requires an emission reduction that is more stringent than the floor. In selecting the standard, EPA considered the magnitude of the reduction in HAP emissions, the cost and economic impacts, energy impacts, non-air quality health impacts, and other environmental impacts. The objective in this consideration is to achieve the maximum degree of emission reduction that does not result in unreasonable economic or other impacts. As with the floors, EPA determined the standard for the source by combining selected control levels for each kind of emission point. The next section presents the additional control levels considered for each kind of emission point and the basis for the selected level. 2. Alternative Control Levels and Selection of Requirements In the selection of the proposed standard, the EPA considered the merits of alternative control levels for individual emission points as well as the overall impacts of the group of decisions in light of the statutory criteria. In the first step of the process, regulatory alternatives were developed for each kind of emission point. These alternatives differed only in the number of emission points that would be controlled by the reference control technology. Regulatory alternatives were developed using information for the chemical processes that could be characterized sufficiently to permit assignment of model emission points. Approximately 97 percent of the nationwide chemical production capacity is associated with these well characterized processes. The emission reduction and control cost estimates for the regulatory alternatives are summarized in Tables 5 and 6. Table 5.- Control Alternatives for Existing Sources Subject to Subpart G sup a Kinds of emission points sup b Process vents Control Option 1 Emission reduction mg/yr 232,000 Percent emission reduction 93 Cost $1,000/yr 51,000 Avg. $/mg 220 Inc. $/mg Control Option *2 Emission reduction mg/yr 234,000 Percent emission reduction 93 Cost $1,000/yr 54,000 Avg. $/mg 230 Inc. $/mg 1,800 Control Option 3 Emission reduction mg/yr 235,000 Percent emission reduction 94 Cost $1,000/yr 58,000 Avg. $/mg 250 Inc. $/mg 2,500 Control Option 4 Emission reduction mg/yr 236,000 Percent emission reduction 94 Cost $1,000/yr 62,000 Avg. $/mg 260 Inc. $/mg 3,900 Control Option 5 Emission reduction mg/yr 238,000 Percent emission reduction 95 Cost $1,000/yr 93,000 Avg. $/mg 390 Inc. $/mg 22,000 Kinds of emission points sup b Wastewater Control Option 0 Emission reduction mg/yr 0 Percent emission reduction 0 Cost $1,000/yr 0 Avg. $/mg 0 Inc. $/mg Control Option *1 Emission reduction mg/yr 82,100 Percent emission reduction 84 Cost $1,000/yr 24,000 Avg. $/mg 290 Inc. $/mg 290 Control Option 2 Emission reduction mg/yr 82,800 Percent emission reduction 85 Cost $1,000/yr 26,000 Avg. $/mg 310 Inc. $/mg 2,600 Control Option 3 Emission reduction mg/yr 85,700 Percent emission reduction 88 Cost $1,000/yr 38,000 Avg. $/mg 440 Inc. $/mg 4,200 Control Option 4 Emission reduction mg/yr 88,900 Percent emission reduction 91 Cost $1,000/yr 104,000 Avg. $/mg 1,200 Inc. $/mg 21,000 Kinds of emission points sup b Transfer Control Option *1 Emission reduction mg/yr 360 Percent emission reduction 65 Cost $1,000/yr 3,100 Avg. $/mg 8,700 Inc. $/mg Control Option 2 Emission reduction mg/yr 420 Percent emission reduction 77 Cost $1,000/yr 6,500 Avg. $/mg 15,000 Inc. $/mg 54,000 Kinds of emission points sup b Storage: Control Option *1 Emission reduction mg/yr 0 Percent emission reduction 0 Cost $1,000/yr 0 Avg. $/mg 0 Inc. $/mg Kinds of emission points sup b Small sup c Control Option 2 Emission reduction mg/yr 360 Percent emission reduction 95 Cost $1,000/yr 19,000 Avg. $/mg 53,000 Inc. $/mg 53,000 Kinds of emission points sup b Storage: Control Option *1 Emission reduction mg/yr 330 Percent emission reduction 70 Cost $1,000/yr 2,100 Avg. $/mg 6,500 Inc. $/mg Kinds of emission points sup b Medium sup d Control Option 2 Emission reduction mg/yr 410 Percent emission reduction 88 Cost $1,000/yr 5,700 Avg. $/mg 14,000 Inc. $/mg 43,000 Kinds of emission points sup b Storage: Control Option 1 Emission reduction mg/yr 1,700 Percent emission reduction 17 Cost $1,000/yr 4,000 Avg. $/mg 2,400 Inc. $/mg Kinds of emission points sup b Large sup e Control Option *2 Emission reduction mg/yr 4,800 Percent emission reduction 48 Cost $1,000/yr 7,300 Avg. $/mg 1,500 Inc. $/mg 1,100 Control Option 3 Emission reduction mg/yr 8,600 Percent emission reduction 87 Cost $1,000/yr 19,000 Avg. $/mg 2,100 Inc. $/mg 2,900 Kinds of emission points sup b Total sup f Floor sup g Control Option Emission reduction mg/yr 234,000 Percent emission reduction 65 Cost $1,000/yr 60,000 Avg. $/mg 260 Inc. $/mg Proposed option Control Option Emission reduction mg/yr 322,000 Percent emission reduction 89 Cost $1,000/yr 91,000 Avg. $/mg 280 Inc. $/mg 350 Total control Control Option Emission reduction mg/yr 337,000 Percent emission reduction 94 Cost $1,000/yr 247,000 Avg. $/mg 730 Inc. $/mg 10,000 sup a The impacts in this table are based on well characterized chemical manufacturing processes and were estimated using the model emission point approach described in Section V of this notice. sup b Only the impacts for emission points subject to Subpart G are described. Equipment leaks are also part of a SOCMI source but are subject to Subpart H. sup c Small denotes storage vessels with capacity greater than or equal to 38 m fn 3 (10,000 gal), but less than 75 m fn 3 (20,000 gal). sup d Medium denotes storage vessels with capacity greater than or equal to 75 m fn 3 (20,000 gal), but less than 151 m fn 3 (40,000 gal). sup e Large denotes storage vessels with capacity greater than or equal to 151 m fn 3 (40,000 gal). sup f These totals do not include control impacts for equipment leaks. Floor tables are option 1 for each emission point. Proposed option totals are the option for each emission point. Total control tables are the last option for each emission point. sup g The first option for each kind of emission point represents the floor. Table 6.- Control Alternatives for New Sources Subject to Subpart G sup a,b Kinds of Emission Points sup c Process Vents Control option *1 Emission Reduction Mg/yr 45,000 Percent Emission Reduction 95 Cost $1,000/yr 13,000 Avg. $/Mg 290 Inc. $/Mg Control option 2 Emission Reduction Mg/yr 45,000 Percent Emission Reduction 95 Cost $1,000/yr 18,000 Avg. $/Mg 390 Inc. $/Mg 47,000 Kinds of Emission Points sup c Wastewater Control option 1 Emission Reduction Mg/yr 12,900 Percent Emission Reduction 70 Cost $1,000/yr 5,100 Avg. $/Mg 400 Inc. $/Mg Control option *2 Emission Reduction Mg/yr 15,900 Percent Emission Reduction 86 Cost $1,000/yr 6,900 Avg. $/Mg 430 Inc. $/Mg 600 Control option 3 Emission Reduction Mg/yr 16,900 Percent Emission Reduction 91 Cost $1,000/yr 19,800 Avg. $/Mg 1,200 Inc. $/Mg 13,000 Kinds of Emission Points sup c Transfer Control option *1 Emission Reduction Mg/yr 68 Percent Emission Reduction 65 Cost $1,000/yr 590 Avg. $/Mg 8,700 Inc. $/Mg Control option 2 Emission Reduction Mg/yr 80 Percent Emission Reduction 77 Cost $1,000/yr 1,200 Avg. $/Mg 15,000 Inc. $/Mg 54,000 Kinds of Emission Points sup c Storage: Control option *1 Emission Reduction Mg/yr 61 Percent Emission Reduction 85 Cost $1,000/yr 1,500 Avg. $/Mg 24,000 Inc. $/Mg Kinds of Emission Points sup c Small sup d Control option 2 Emission Reduction Mg/yr 68 Percent Emission Reduction 94 Cost $1,000/yr 3,600 Avg. $/Mg 53,000 Inc. $/Mg 304,000 Kinds of Emission Points sup c Storage: Control option *1 Emission Reduction Mg/yr 62 Percent Emission Reduction 70 Cost $1,000/yr 400 Avg. $/Mg 6,400 Inc. $/Mg Kinds of Emission Points sup c Medium sup e Control option 2 Emission Reduction Mg/yr 78 Percent Emission Reduction 88 Cost $1,000/yr 1,000 Avg. $/Mg 14,000 Inc. $/Mg 42,000 Kinds of Emission Points sup c Storage: Control option 1 Emission Reduction Mg/yr 0 Percent Emission Reduction 0 Cost $1,000/yr 0 Avg. $/Mg 0 Inc. $/Mg Kinds of Emission Points sup c Large sup f Control option *2 Emission Reduction Mg/yr 1,100 Percent Emission Reduction 81 Cost $1,000/yr 800 Avg. $/Mg 730 Inc. $/Mg 730 Control option 3 Emission Reduction Mg/yr 1,100 Percent Emission Reduction 81 Cost $1,000/yr 1,000 Avg. $/Mg 950 Inc. $/Mg 117,000 Total sup g Floor sup h Control option Emission Reduction Mg/yr 58,000 Percent Emission Reduction 86 Cost $1,000/yr 21,000 Avg. $/Mg 350 Inc. $/Mg Proposed Option Control option Emission Reduction Mg/yr 62,000 Percent Emission Reduction 92 Cost $1,000/yr 23,000 Avg. $/Mg 370 Inc. $/Mg 630 Total Control Control option Emission Reduction Mg/yr 63,000 Percent Emission Reduction 93 Cost $1,000/yr 45,000 Avg. $/Mg 710 Inc. $/Mg 21,000 sup a The impacts in this table are based on well characterized chemical manufacturing processes and were estimated using the model emission point approach described in Section V of this notice. sup b Estimated control impacts for fifth year after promulgation of the HON based on an assumed industry growth of 3.5 percent each year. sup c Only the impacts for emission points subject to Subpart G are described. Equipment leaks are also part of a SOCMI source but are subject to Subpart H. sup d Small denotes storage vessels with capacity greater than or equal to 38 m fn 3 (10,000 gal), but less than 75 m fn 3 (20,000 gal). sup e Medium denotes storage vessels with capacity greater than or equal to 75 m fn 3 (20,000 gal), but less than 151 m fn 3 (40,000 gal). sup f Large denotes storage vessels with capacity greater than or equal to 151 m fn 3 (40,000 gal). sup g These totals do not include control impacts for equipment leaks. Floor tables are option 1 for each emission point. Proposed option totals are the option for each emission point. Total control tables are the last option for each emission point. sup h The first option for each kind of emission point represents the floor. Table 5 provides the control costs and emission reductions associated with alternative control levels considered for existing sources in the source category. Table 6 presents the same information for the alternative control levels considered for new sources in the source category. The estimates presented in Tables 5 and 6 differ from the estimates summarized in section IV of this notice because those earlier estimates include an extrapolation to account for processes that could not be modeled. In selection of the proposed standard, the EPA considered: (1) Magnitude of the emission reduction; (2) cost of the emission reduction; (3) economic impacts and feasibility; (4) consistency with previous decisions; (5) other non-air quality health and environmental impacts; and (6) energy requirements. Based on consideration of these factors, the EPA selected a standard that would ensure a significant reduction in HAP emissions from the SOCMI. The EPA's analysis estimates that the selected standard would reduce HAP emissions from the four kinds of emission points by 322,000 Mg/yr (355,000 tons/yr) from existing sources and 62,000 Mg/yr (68,000 tons/yr) from new sources in the fifth year of the standard. This represents an emission reduction of about 89 percent for existing sources (92 percent for new sources) in comparison to the emissions that would have occurred without the standard. For the group of well characterized sources, the total nationwide annual cost associated with this emission reduction is estimated to be about $114 million/yr-$91 million/yr of this cost is from control of existing sources and $23 million/yr is from control of new sources. (As stated above, these estimates represent processes sufficiently well characterized to assign models. These estimates differ from the national numbers presented in section IV which include an extrapolation to account for processes that could not be modeled.) The nationwide annual cost of this rule is estimated to be $182 million. Of this total cost, about $48 millon/yr results from the costs of monitoring, recordkeeping, and reporting provisions. a. Process vents: Existing sources. The determination of the best 12 percent of the process vents was based on consideration of the various parameters that affect emission rates (flow rate, HAP concentration, net heating value, and corrosion properties). These parameters are highly variable from one process to another depending on the constituents of the vent stream. Therefore, it is not appropriate to define best performing in terms of any single parameter or even any specific combination of them since the combination would be different from one process to another. A surrogate measure for these parameters that influence whether or not a vent stream is controlled and the specific control {pg 62631} technology applied is cost effectiveness. Cost- effectiveness values can be used to reflect all possible combinations of these parameters and the cost of controlling streams with the reference control technology (98 percent efficient combustion control). Use of the single criterion of cost effectiveness would result in a more understandable and enforceable rule. The procedure used to define the groups of process vents was to rank all process vents in the data base from highest to lowest cost-effectiveness of control and to determine the point where less than 12 percent of the vents were controlled. This analysis showed that the dividing line between the groups occurred at a cost effectiveness of $1,500/Mg ($1,360/ton). Of the process vents with cost effectiveness values of less than $1,500/Mg, 44 percent were controlled with the 98 percent efficient combustion devices. Thus, all process vents with cost effectiveness of $1,500/Mg and lower have 98 percent control efficiency in the determination of the source-wide floor. In selection of the proposed standard for existing sources, EPA believes that a level of emission reduction from process vents more stringent than the level associated with the floor is achievable. The EPA is proposing to require the control level in Option 2, which has an incremental cost-effectiveness of $1,800/Mg ($1,630/ton) and would require control of vents that have TRE cost-effectiveness values of up to $2,000/Mg ($1,800/ton). Although EPA is proposing this specific control level, EPA is requesting comment on Options 1 through 4. These options would require control of vents with TRE cost-effectiveness values of up to $1,500/Mg ($1,360/ton) in Option 1 (the floor component); vents with TRE cost- effectiveness values of up to $3,000/Mg ($2,720/ton) in Option 3; and vents with TRE cost-effectiveness values of up to $5,000/Mg ($4,540/ton) in Option 4. The EPA anticipates that in consideration of the final decision EPA will give more weight to Options 1 through 3. The Option 2 control level was selected in the proposed rule based on considerations specific to the SOCMI category and would not necessarily be appropriate for other source categories. Therefore, it should not be viewed as a precedent for decisions on other standards. The EPA recognizes that there are differing views regarding the appropriate criteria for determining the control level. Consequently, EPA is requesting comments on the appropriateness of particular control choices within the range of options described above and asks that commenters provide supporting rationale for their preferences. The EPA selected the proposed control level considering the emission reduction achieved by the alternative control options and considering the criteria enumerated in section 112(d) of the Act: The cost of achieving such emission reduction; any non-air quality health and environmental impacts; and energy requirements. As a matter of general policy in decisions to select control levels above the floor, EPA believes that the cost- effectiveness of controls and a comparison of benefits, both quantifiable and nonquantifiable, and costs are primary considerations. In any such comparison of costs and benefits, the uncertainties associated with the benefit and cost estimates should be characterized. In this proposed rulemaking, however, EPA's ability to do such a comparison was severely limited by the lack of sufficient data on the characteristics of the emission sources and the complexity of the SOCMI category. A preliminary assessment of the benefits of the potential emission reductions was prepared using the available data. However, because of the data limitations cited above, EPA does not believe that the analysis provides an adequate basis for decisionmaking. This analysis has been placed in Docket A- 90-19. The EPA invites comment on this analysis, its usefulness and defensibility, and requests the submittal of data and methods that could improve the analysis. The EPA will review any such comments and data and will, to the extent practicable, evaluate their implications for the risk assessment prior to the final rule. Despite the difficulty of estimating quantitative benefits for this rulemaking due to the complexity of the source category, data limitations, and timing of the rule, EPA remains committed to evaluating benefits as well as costs associated with decisions to go above the floor in future MACT standards. The EPA also intends to review cost-effectiveness of decisions as part of these future rulemakings. Due to this lack of a benefits assessment, EPA used benchmarks from previous air toxics regulatory decisions and past air toxics benefits studies as a surrogate measure for benefits considerations in the decision on the proposed control level. The information considered consisted of the Benzene NESHAP decisions (49 FR 23498, 54 FR 38044, and 55 FR 8292) and the Vinyl Chloride NESHAP (41 FR 46560). For example, one decision that reflected consideration of balancing emission reductions against costs was the benzene storage standard (55 FR 38044). In that decision, the cost- effectiveness of the selected option was estimated to range from $128/Mg to $909/Mg ($116/ton to $824/ton) in 1982 dollars ($160/Mg to $1,100/Mg $145/ton to $1,000/ton in 1989 dollars). Although the cost-effectiveness measures from past decisions were considered in selection of the proposed control level, EPA also considered the differences between this proposed rule and the earlier decisions on benzene and vinyl chloride. This rule applies to emissions of 112 HAP's while the earlier rules only addressed cancer health effects of a single HAP (i.e., benzene or vinyl chloride). Because of the noncancer health effects associated with many of the 112 HAP's, there are questions about the degree to which past decisions may serve as a guide to this proposal. The EPA is not able at this time to quantify the noncancer health effects so that they can be combined with the cancer health effects for the HON. Additional factors were considered in the selection of the proposed control level. These factors include the magnitude of the emission reduction, the cost of this reduction, consistency with past decisions, location of facilities near population centers, and other non-air quality environmental benefits from the control. The consideration of these factors is summarized below. Existing source process vents are the single largest contributor of HAP emissions from the SOCMI source category. The additional emission reduction beyond that achieved at the floor that would result from the proposed control level is about 1,600 Mg/yr (1,800 tons/yr), which represents 0.7 percent of the emission reduction achieved at the floor. This additional control has a cost effectiveness of $1,800/Mg ($1,630/ton) of HAP and with a TRE equivalent to a cost- effectiveness of $2,000/Mg ($1,800/ton). The cost of this additional control (along with the cost of the other proposed control requirements) is estimated to result in less than a 3 percent price increase for SOCMI chemicals. The EPA requests comments on whether the costs of the selected control requirements are reasonable and asks that the commenters provide supportive rationale for their judgments on the costs. The EPA is requesting comment on a range of control options because of the previously discussed uncertainties in the measure of the benefits of control and because of the uncertainties in the cost that would actually be experienced. Because of the conservative assumptions used to develop the control {pg 62632} costs and the TRE format of the vent provisions, it is unlikely that any source would actually incur costs at the TRE level (the criterion for defining vents that must apply control) that is equivalent to $2,000/Mg ($1,800/ton). The cost analysis used in developing the supporting information for this standard was based on use of a dedicated control device for each vent. Actual costs would be lower where one device could be used to control emissions from several vents. The cost of control would also be lower in those cases where it is feasible to use a boiler or process heater instead of an incinerator. Finally, the provisions are structured using a TRE index approach. Since the standard is in a TRE format, many facilities may be able to use product recovery to change vent stream characteristics so that they are below the relevant applicability criteria. This flexibility is expected to allow some sources to lower their cost of compliance through changes in equipment or operations. As a result, the impacts analysis may overstate the costs of complying with the proposed control requirement. The EPA solicits comments concerning whether the control costs for process vents are overestimated and the extent of the overestimate. b. Process vents: New sources. The analysis of the data base showed that the maximum degree of emission reduction being achieved by the best-controlled vent occurs at a cost effectiveness value of $11,000/Mg ($9,980/ton). In the determination of the process vent component of the source-wide floor for a new source, vents with a cost effectiveness value for control of $11,000/Mg ($9,980/ton) would be controlled. For the standard for new sources, EPA considered selecting a level of emission reduction more stringent than the level associated with the source-wide floor. However, a standard more stringent than the floor component is not being proposed because the costs were considered high given the very small additional emission reduction available. The additional control would achieve an additional emission reduction of about 100 Mg/yr at a cost of about $5 million/yr. Therefore, the control level associated with the source-wide floor was considered to represent the maximum reduction achievable considering cost and other impacts. The proposed standard for new sources reflects the floor level of control for new vents. c. Storage vessels: Existing sources. In the analysis of the data base to determine the storage vessel components of the source-wide floor, EPA divided the population of model vessels into three size ranges. The size ranges were: 38 m sup 3 to 75 m sup 3 (10,000 to 20,000 gal) (small); 75 to 151 m sup 3 (20,000 to 40,000 gal) (medium); and greater than or equal to 151 m sup 3 (40,000 gal) (large). The first two of the ranges include the two smallest model vessel sizes in the data base and the third range is the combination of the remaining four model vessel size ranges. The larger size range vessels were aggregated into one group because no differences in control based on vessel capacity were expected for the vessels exceeding 40,000 gallons capacity. Only one State regulation distinguishes among vessels with capacities greater than 151 m sup 3 (40,000 gal) and none of the greater than 151 m sup 3 (40,000 gal) storage vessels in the data base are affected by those requirements. Consequently, separate analysis of each of the larger capacity range model vessels would provide the same results as a combined analysis. The parameter used in the analysis to determine the storage vessel components of the source-wide floor was the vapor pressure of the liquid being stored. Vapor pressure is one of three major factors that most influence emissions from storage vessels and potential emission reductions. Furthermore, vapor pressure is commonly a prime determining factor in whether or not a vessel is controlled. For each segment of the storage vessel population, the procedure used to define components for the source-wide floor was to rank storage vessels from lowest to highest vapor pressure. Next, the vapor pressure at which at least 12 percent of the vessels is controlled by the reference control technology was determined. The analysis showed that at liquid vapor pressures of 13.1 kPa (1.9 psia) and higher more than 12 percent of the medium and large vessels are controlled with the reference technology. Of the vessels storing liquids with vapor pressures of 13.1 kPa (1.9 psia) and higher, about 30 percent of the medium vessels and 45 percent of the large vessels were controlled with the reference technology. Therefore, the storage vessel components of the source-wide floor are control of vessels 75 m sup 3 (20,000 gal) and higher that store liquids with vapor pressures of 13.1 kPa (1.9 psia) and higher. For small vessels, the analysis of the data base showed that only 6 percent of the vessels were expected to be controlled with the reference technology. Therefore, since less than 12 percent of the small vessels are controlled with the reference control technology the small vessel component of the source-wide floor is no control. For each of the three populations of storage vessels, EPA considered several alternative levels of emission limitation. The alternatives differed in the vapor pressures of the liquids that would require control and ranged from the vapor pressure associated with the component of the source-wide floor (i.e., 13.1 kPa 1.9 psia ) to vapor pressures of 0.07 kPa (0.01 psia). After considering the alternatives and the associated impacts, EPA proposes to require the following storage vessels be equipped with the reference control technology: (1) Vessels with a capacity greater than or equal to 151 m sup 3 (40,000 gal) storing HAP's with vapor pressures of 5.2 kPa (0.75 psia) or greater and (2) vessels with a capacity greater than or equal to 75 m sup 3 (20,000 gal) storing HAP's with vapor pressures of 13.1 kPa (1.9 psia) and greater. No control requirements are being proposed for storage vessels with capacities less than 75 m sup 3 (20,000 gal). In the selection of these proposed control requirements, EPA believes, based on the available data, that an emission reduction more stringent than the level associated with the floor component for large vessels is achievable considering the statutory criteria. The EPA also believes that no emission levels more stringent than the level associated with the floor components for medium and small vessels are achievable considering the statutory criteria. The statutory criteria considered in selection of the proposed control requirements were the magnitude of the emission reductions, the cost and economic impacts, energy impacts, non-air quality health impacts, and other environmental impacts. The specific considerations of the statutory factors in each of these decisions are summarized below. The proposed control requirements for large storage vessels containing liquids with HAP vapor pressures of 5.2 kPa (0.75 psia) or higher are estimated to achieve an emission reduction of 4,800 Mg/yr (5,280 tons/yr) of HAP's compared to emissions that would occur without the standard. This represents a 48 percent reduction of emissions from this segment of the SOCMI storage vessel population. The annual cost to achieve this reduction is about $7.3 million and the incremental cost- effectiveness of the control beyond the floor is $1,100/Mg. Although EPA considered proposing to control HAP's with vapor pressures of 10.3 kPa (1.5 psia) for large vessels, it was found that this option was less economically efficient (i.e., higher incremental cost-effectiveness) than the proposed option. The cost- effectiveness of the proposed {pg 62633} option is in the range of the cost- effectiveness values of the Benzene Storage NESHAP. The non-air quality health impacts as well as the energy and other environmental impacts of the alternative control levels were also considered in the selection of the proposed requirements for large storage vessels at existing sources. No non-air quality health impacts were expected from any of the alternatives and the energy and other environmental impacts did not vary significantly among the alternatives. Thus, these considerations did not affect the choice of the proposed requirements. The controls required by the proposed requirements are not expected to create any secondary emissions of carbon monoxide or nitrogen oxides. In addition, no secondary benefits of control of non-HAP VOC's are expected. The energy impacts estimated for the selected control requirements were about 5,400,000 kW-hr/yr for electricity. The energy requirements associated with control of all large vessels storing HAP's of any vapor pressure was about 20 percent greater. The energy required for the control equipment represents a small percentage of the total energy requirement for this industry. Based on previous studies of SOCMI, no unreasonable adverse energy impacts are expected from any of the alternative control levels considered. A more stringent level of emission reduction is not being proposed because the additional reduction that could be achieved through further control of large storage vessels was not considered significant, given the additional cost. The additional emission reduction achieved through control of HAP's with vapor pressures lower than 5.2 kPa (0.75 psia) is about 4,000 Mg/yr (4,300 tons/yr). This control would cost an additional $11.2 million per year. The cost-effectiveness of this additional emission reduction is $2,900/Mg. The cost of achieving this emission reduction was not considered reasonable. In selecting the proposed emission limitations for small and medium storage vessels at existing sources, EPA considered levels of emission reduction more stringent than the level associated with the source- wide floor for these emission points. The alternative control options considered for small vessels were: (1) Vapor pressures of 76.6 kPa (11.1 psia) and higher; (2) vapor pressures of 5.2 kPa (1.9 psia) and higher; and (3) vapor pressures greater than 0.07 kPa (0.01 psia). The alternative control options beyond the floor component considered for medium vessels were: (1) Vapor pressures of 0.7 kPa (0.1 psia) and higher; and (2) vapor pressures greater than 0.07 kPa (0.01 psia). None of the alternative options for either small or medium vessels is being proposed since the costs were considered high given the very small potential emission reductions. Control beyond the floor component for medium storage vessels would reduce emissions by less than 100 Mg/yr (110 tons/yr) at a cost of $3.6 million/yr. With small storage vessels the maximum potential reduction of 360 Mg/yr would cost about $19 million/yr. The average cost- effectiveness of these control requirements varied from $53,000/Mg ($48,000/ton) for small vessels to about $14,000/Mg ($12,700/ton) for medium vessels. Therefore, EPA believes that the control level for the small and medium storage vessels components of the source-wide floor represented the maximum reduction achievable considering cost and other impacts. The EPA is proposing to require control of medium storage vessels storing HAP's with vapor pressures of 5.2 kPa (1.9 psia) and higher. No control requirements are being proposed for small storage vessels. For both small and medium storage vessels, this proposal requires no control beyond the respective components of the source-wide floor. An additional consideration in the selection of the proposed options was the non- air quality health impacts as well as the energy and other environmental impacts of the alternative control levels. No non-air quality health impacts were expected from any of the alternatives and the other environmental impacts did not vary significantly among the alternatives. Thus, these considerations did not affect the choice of the proposed requirements. The co-controls required by the proposed requirements are not expected to create any secondary emissions of carbon monoxide or nitrogen oxides. In addition, no secondary benefits of co-control of non-HAP VOC's are expected. The energy impacts for the required controls for medium storage vessels would be about 2,600,000 kW-hr/yr for electricity. The energy requirements associated with control of medium storage vessels containing liquid HAP's of any vapor pressure is slightly greater, but is not considered significant. The energy required for the control equipment represents a small percentage of the total energy requirement for this industry. Based on previous studies of SOCMI, no unreasonable adverse energy impacts are expected from any of the alternative control levels considered. d. Storage vessels: New sources. The determination of the best controlled storage vessels also used the three size ranges of model vessels. For the reasons previously described, the parameter used in the analysis to determine the storage vessel components of the source-wide floor was the vapor pressure of the liquid being stored. The analysis of the data base showed that the maximum degree of emission limitation being achieved by the best-controlled vessels in each population segment occurs at 13.1 kPa (1.9 psia) for small and medium vessels and at 5.2 kPa (0.75 psia) for large vessels. For each of the three segments of the storage vessel population, EPA considered several alternative levels of emission limitation more stringent than the emission level associated with the floor component. The alternatives were structured similarly to those for existing storage vessels but differed in the vapor pressures of the liquids that would be required to apply controls. The alternative control options examined for storage vessels at new sources were: (1) for large and medium vessels, vapor pressures 0.7 kPa (0.1 psia) and greater and vapor pressures of 0.07 kPa (0.01 psia) and greater; and (2) for small vessels, vapor pressures of 0.07 kPa (0.01 psia) and greater. The EPA is proposing to control large storage vessels that store HAP's with vapor pressures of 0.7 kPa (0.1 psia) and higher. This proposed control will result in a significant emission reduction at a reasonable cost. The proposed control requirement for large storage vessels is estimated to achieve an emission reduction of 1,100 Mg/yr (1,200 tons/yr) of HAP's compared to emissions that would occur without the standard. This represents an 87 percent reduction from this segment of the SOCMI storage vessel population. The annual cost to achieve this reduction is about $0.8 million and the cost- effectiveness of this control is $730/Mg ($670/ton). The non-air quality health impacts were considered in this decision as well as the energy and other environmental impacts of the alternative control levels. As with the other decisions, these factors did not vary significantly among the alternatives. Thus, these considerations did not affect the choice of the proposed requirements. The energy required for control beyond the floor component would be about 200,000 kW-hr/yr for electricity. The EPA is not proposing to require control of vapor pressures lower than 0.7 kPa (0.1 psia) for large storage vessels at new sources because the {pg 62634} additional emission reduction achieved through further control was not significant, given the additional cost. Further control was estimated to result in only 2 Mg/yr additional reduction at an additional cost of $233,000/yr. This cost was judged to be disproportionately high. For small and medium vessels at new sources, none of the alternative control options more stringent than the floor components were selected. After considering the emission reductions, costs, and other impacts of the alternatives, EPA determined the cost to achieve the additional reduction was high given the very small potential emission reductions. Additional control would reduce emissions from medium storage vessels by less than 20 Mg/yr (22 tons/yr) at an additional cost of about $700,000/yr. For the small storage vessels segment of the population, further control would result in less than 10 Mg/yr (11 tons/yr) emission reduction at an added cost of about $2.1 million/yr. The incremental cost effectiveness values for these control requirements are about $304,000/Mg ($276,000/ton) for small vessels and $42,000/Mg ($38,100/ton) for medium vessels. The consideration of the non-air quality health impacts as well as the energy requirements and other environmental impacts was similar to that in the decision for existing storage vessels. Therefore, EPA determined that the control level for the small and medium storage vessels components of the sourcewide floor represented the maximum reduction achievable considering cost and other impacts. e. Transfer operations: Existing sources. In the analysis of the data base to determine the transfer rack components of the source wide floor, EPA divided the population of model racks into two groups based on the vapor pressure of materials loaded. The vapor pressure groupings were transfer racks loading materials with average vapor pressures less than 10.3 kPa (1.5 psia) and racks loading materials with vapor pressures of 10.3 kPa (1.5 psia) and higher. These two groups are the result of several simplifying assumptions used in the analysis and the fact that most State regulations only require control of racks loading chemicals with vapor pressures of 10.3 kPa (1.5 psia) and higher. The procedure used to determine the transfer rack components of the source-wide floor was to rank the racks in these two groups by quantity loaded (throughput) in ascending order. Since emissions from transfer operations are largely determined by throughput and vapor pressure, throughput was expected to provide a good measure of whether or not a rack is controlled. For both groups, the throughput at which more than 12 percent of the racks is controlled by the reference control technology was determined. The analysis showed that for transfer racks that load HAP's with vapor pressures of 10.3 kPa (1.5 psia), and higher, that have throughputs greater than or equal to 0.65 million l/yr (170,000 gal/yr) more than 12 percent of the racks are controlled with the reference technology. None of the transfer racks that load HAP's with vapor pressures less than 10.3 kPa (1.5 psia) are controlled with the reference technology. Therefore, the transfer rack component of the source-wide floor is control of racks that load HAP's with vapor pressures of 10.3 kPa (1.5 psia) and higher and that have throughputs of 0.65 million l/yr (170,000 gal/yr) or greater. Only two levels of emission limitation were evaluated and considered in the selection of the proposed control requirements. These alternatives were the emission level associated with the transfer rack component of the sourcewide floor and the level associated with control of all transfer racks. To simplify the consideration of control alternatives, the emissions and cost information for the two groups of racks were combined. After considering the emission reductions, costs, and other impacts of the alternatives, EPA believes that the level of emission limitation associated with the component of the source-wide floor is achievable. The more stringent level of emission reduction is not being proposed because the additional emission reduction achieved is not significant and the cost of this reduction is high. Control beyond the level associated with the source-wide floor would reduce emissions from transfer racks by less than 70 Mg/yr (77 tons/yr) at a cost of $3.4 million/yr. The average cost effectiveness of this control is $54,000/Mg ($49,000/ton). As with the decisions on the other emission points, no non-air quality health impacts were expected from any of the alternatives and the other environmental impacts did not vary significantly among the alternatives. Thus, these considerations did not affect the choice of the proposed requirements. The controls required by the proposed requirements are not expected to create any significant secondary emissions of carbon monoxide or nitrogen oxides. In addition, no secondary benefits of co-control of non-HAP VOC's are expected. The energy impacts for the required controls are also not believed to be significant. f. Transfer operations: New sources. The analysis of the data base showed that the maximum degree of emission reduction being achieved by the best controlled transfer rack occurs at vapor pressures of 10.3 kPa (1.5 psia) and throughputs greater than or equal to 0.65 million l/yr (0.17 million gal/yr). Thus, for new sources the transfer rack component of the source wide floor is identical to that for existing sources. The decision on the control requirements considered the same two levels of emission limitation that were considered for transfer racks at existing sources. For the same reasons discussed earlier, EPA is not proposing to require a more stringent level of emission reduction. After considering emission reductions and other information, EPA believes that the level of emission limitation associated with the component of the source-wide floor is achievable. g. Process wastewater: Existing sources. The parameters used in the analysis of the data base to identify the floor level of control for wastewater streams were: Concentration of chemicals with high volatility in wastewater (e.g., benzene or vinyl chloride); concentration of chemicals with lower volatilities than benzene, or semivolatile HAP's; and flow rate. These parameters were used because they are among the primary factors that most influence air emissions from wastewater streams. As previously noted, EPA used data on the control requirements in existing State and Federal regulations to identify streams that must be controlled and to identify the required controls. The only controlled wastewater streams in the data base were those estimated to meet the applicability requirements of either the Vinyl Chloride NESHAP, 40 CFR part 61, subpart F, or the Benzene Waste Operations NESHAP, 40 CFR part 61, subpart FF. There are no State regulations requiring control of air emissions from wastewater. Consequently, the procedure used in the analysis of the existing control level for wastewater operations was to determine the proportion of processes and wastewater streams that have concentrations above the control criteria in the Benzene Waste NESHAP or Vinyl Chloride NESHAP (greater than 10 ppm benzene or vinyl chloride) compared to the total number of processes and wastewater streams. The estimate of the number of processes and benzene containing wastewater streams subject to the Benzene Waste NESHAP was adjusted by the expected proportion of {pg 62635} the facilities exceeding the facility-wide threshold for control. This analysis showed that fewer than 5 percent of SOCMI processes and fewer than 3 percent of the wastewater streams are currently controlled to the efficiency of the reference control technology. Therefore, since less than 12 percent of existing wastewater streams are controlled, the component of the source-wide floor for existing wastewater streams (Option 0) is estimated to be no control. The EPA considered alternative levels of emission reduction that varied from controlling those process wastewater streams that are larger and more cost effective to control to controlling all wastewater streams. The EPA is proposing to require control of process wastewater streams with greater than or equal to 10 l pm flow rate and greater than 1000 ppmw total VOHAP concentration (Option 1). The proposed control level was selected considering the emission reductions of the alternative control options and the criteria enumerated in section 112(d) of the Act. The proposed alternative is considered to achieve the maximum emission limitation that is achievable considering cost, non-air quality health and environmental impacts, and energy requirements. Although EPA is proposing this specific option, commenters should be aware that there are a number of technical issues regarding the emission and cost estimation methodologies. (These issues are discussed in section VII.E.3 of this notice.) The EPA anticipates that for wastewater controls there may be changes in the emission and cost estimates that result from the resolution of the technical and scientific issues and any new information received during the public comment period. The final rule will be based on consideration of the revised estimates. The considerations of the currently available information on the alternative control options are summarized below. Wastewater streams at existing sources are believed to be the second largest contributor of HAP emissions from the emission points regulated by the HON. The proposed control requirements are expected to achieve an emission reduction of 82,100 Mg/yr (90,300 tons/yr) of HAP's compared to current emission levels for the well characterized processes. This represents an 84 percent reduction from the uncontrolled emission rate. The annual cost to achieve this reduction is $23.9 million. The cost-effectiveness of going from no control to the proposed control is $290/Mg ($265/ton) for the well characterized processes. The estimated costs of the proposed wastewater control requirements are based on use of steam strippers to control the wastewater stream and do not reflect any potential cost savings from use of the other means of demonstrating compliance that are provided in the wastewater provisions. One of these compliance options allows a source to demonstrate compliance through use of pollution prevention measures and waste treatment to change the wastewater stream characteristics. This compliance option is expected to lower the cost of compliance for some sources. Thus, EPA believes that the cost analysis may overstate the cost of complying with the proposed wastewater provisions. Industry representatives have indicated that they believe the cost estimates to be understated. The EPA requests comments on whether the costs of the proposed control requirements are reasonable and asks that commenters provide supporting information for their estimates. The non-air quality health impacts as well as the energy and other environmental impacts of the alternative control levels were also considered in the selection of the proposed requirements. No non-air quality health impacts were expected from any of the alternatives and the energy and other environmental impacts did not vary significantly among the alternatives. Thus, these considerations did not affect the choice of the proposed requirements. The energy impacts estimated for the selected control requirements were about 4,000,000 kW- hr/yr for electricity and 3,600,000 million Btu/yr for steam. The energy required for the most stringent control level was only slightly greater for electricity and steam. These energy requirements represent a small percentage of the total energy requirement for this industry. Based on previous studies of SOCMI, no unreasonable adverse energy impacts are expected from any of the alternative control levels considered. While EPA could not quantify the non-air quality benefits, EPA considers control of wastewater streams to result in non-air quality benefits. These benefits include generation of less hazardous waste, less groundwater contamination, and less contamination of stormwater. An additional air quality benefit considered in the selection of the proposed alternative is that SOCMI wastewater streams also can contain non- HAP VOC's. These non-HAP VOC's will be controlled when streams are controlled for HAP emissions. This co-control is expected to significantly reduce emissions of non-HAP VOC's, which in specific locations may contribute to efforts to reduce tropospheric ozone. Alternative control Options 2 through 4 were not selected because the additional emission reduction achieved through further control was not significant, given the costs and the uncertainty regarding the characterization of SOCMI wastewater systems. Specifically, control of wastewater streams with flows of 5 l pm or greater and concentrations of 800 ppmw VOHAP (Option 2) was estimated to result in about 700 Mg/yr (770 tons/yr) additional reduction at a cost of about $1.8 million/yr. This control option has a cost-effectiveness value of $2,600/Mg ($2,400/ton). Options 3 and 4 achieve only a small additional emission reduction at cost- effectiveness values of $4,200/Mg to $21,000/Mg ($3,800/ton to $19,000/ton). Given the technical uncertainties that exist regarding the representation of SOCMI wastewater streams and industry practices in design of wastewater collection and treatment systems, it is uncertain whether any of the alternative control options considered would result in additional emission reductions. Therefore, EPA is not proposing any options more stringent than Option 1. h. Process wastewater: New sources. The analysis of the data base also showed the maximum emission reduction being achieved is determined by the control requirements for the Benzene Waste Operations NESHAP and the Vinyl Chloride NESHAP. Under these rules, wastewater streams with 10 ppmw or greater and flow rates greater than 0.02 l pm have to be treated prior to discharge and units in which waste is managed before treatment must be equipped with 95 percent efficient air emission controls. This emission limitation was assumed to be equally applicable to other chemicals that have volatilities in water equal to or greater than benzene and vinyl chloride. These chemicals, which are termed VVHAP, are listed in Table 8 of the proposed Subpart G. As previously noted, there are no State or Federal regulations that would require control of chemicals other than benzene or vinyl chloride. Therefore, there is no control at the floor for chemicals with volatilities lower than that of benzene or vinyl chloride. In the determination of the source-wide floor component for new sources, wastewater streams containing 10 ppmw, or more, of VVHAP would be required to be controlled to the efficiency of the reference technology {pg 62636} and chemicals with lower volatilities would not. The EPA developed alternative control options for process wastewater streams at new sources by first considering whether to require control beyond the floor component for VVHAP containing streams (Option 1) and then considering control options for lower volatility chemicals in wastewater streams. No alternatives were developed for further control of VVHAP since the potential additional emission reduction for VVHAP was minimal. For the less volatile chemicals, EPA examined several alternative options. Table 6 shows the emission reductions and costs associated with the floor control for VVHAP combined with the emission reduction and costs for control of total VOHAP concentrations of either 1,000 ppmw (Option 2) or 5 ppmw (Option 3). After considering the alternatives and the associated impacts, EPA is proposing the control requirements in Option 2. The proposed control requirements for new source wastewater streams would apply to 3 sets of streams: streams with flow rates of 0.02 lpm or greater and concentrations of 10 ppmw or greater of VVHAP (as defined in Table 8 in Sec. 63.132 of subpart G); streams with flow rates of 10 lpm or greater and concentrations of 1000 ppmw or greater of VOHAP (as defined in Table 9 of Sec. 63.138 of subpart G); and all streams with concentrations of 10,000 ppmw or greater of VOHAP. The proposed control level was selected considering the emission reduction achieved by the alternative control options for VOHAP emissions and considering the criteria enumerated in section 112(d) of the Act. The proposed alternative is considered to achieve the maximum emission reduction considering the cost, non-air quality health effects, environmental impacts, and energy requirements. The considerations of these factors and the uncertainties in the estimates are described below. The proposed control requirements are estimated to achieve a reduction of 15,900 Mg/yr (17,500 tons/yr) of VVHAP and VOHAP compared to emissions in absence of this rule. This represents an 86 percent reduction from uncontrolled emission rates for wastewater. The annual cost to achieve this reduction is about $6.9 million. As noted in the discussion of the proposed control for existing wastewater streams, there is uncertainty whether the costs are understated or overstated. Option 2 is estimated to achieve an emission reduction of about 3,000 Mg/yr (3,300 tons/yr) of VOHAP emissions above the floor. This control would have cost of about $1.8 million/yr and a cost-effectiveness value of $600/Mg ($540/ton). In selecting Option 2 for the proposed control requirements, EPA also considered the non-air quality health effects as well as the energy requirements and other environmental impacts. The consideration of these criteria was similar to that in the selection of controls for existing wastewater streams. No non-air quality health impacts were expected from any of the alternatives. In comparing the alternatives, EPA found that the environmental and energy impacts did not vary significantly among the alternatives. The energy and secondary impacts from the control devices were not considered significant. The proposed control requirements are expected to have secondary non-air quality benefits such as reduced potential for groundwater contamination. The selected alternative is estimated to result in control of about 30 percent of the total wastewater volume from new SOCMI sources. The EPA is not proposing a more stringent level of emission limitation because control beyond the selected level is estimated to achieve only a small additional emission reduction. The further control would reduce emissions by an additional 1,000 Mg/yr (1,100 tons/yr) of VOHAP. This control would cost about $20 million per year, an increase of about $13 million per year over the cost of the proposed control level. Because the cost is disproportionately large compared to this additional emission reduction and the great uncertainties regarding the potential emission reduction, EPA is not proposing the more stringent control option. B. Selection of Process Vents Provisions 1. Selection of Format The format chosen for process vent streams is dependent upon the control device selected. For vent streams controlled by control devices other than flares, the format is a combination of a weight-percent reduction and an outlet concentration. A weight-percent reduction format is appropriate for streams with HAP concentration above 1000 ppmv because such a format ensures that the stream will meet the reference control technology requirements. For process vents with concentrations below about 1000 ppmv, a 20 ppmv outlet concentration was selected because 98 percent reduction may not be achievable. Further details on selection of this format are presented in the proposal preamble for the SOCMI reactors NSPS (55 FR 26953, June 29, 1990). The combustion of vent streams containing halogenated organic compounds can produce emissions of halogens and hydrogen halides, some of which are HAP's, such as hydrogen chloride, chlorine, and hydrogen fluoride. To reduce these emissions, the proposed standard requires the use of a scrubber after the combustion device for halogenated process vent streams. The format of the standard for such scrubbers is a percent reduction or outlet concentration of those halogens and hydrogen halides that can be measured using the EPA Method 26 or Method 26A. A percent reduction format ensures that most streams will meet the reference control technology requirements. However, an alternative outlet concentration level is needed for low concentration streams where the specified percent reduction would result in outlet levels too low to measure. For vent streams controlled by a flare, the proposal includes equipment and operating specifications because it is very difficult to measure the emissions from a flare to determine its efficiency. 2. Selection of Group Determination Procedures, Performance Tests, Monitoring Requirements, and Test Methods The standard specifies the group determination procedures, performance tests, monitoring requirements, and test methods necessary to determine whether a process vent stream is required to apply control devices and to demonstrate that the allowed emission levels are achieved when controls are applied. As with the format of the process vent provisions, these requirements are dependent on the control device selected. a. Group determination procedures. Each owner or operator would be required to follow group determination methods and procedures to determine whether the vent is a Group 1 or Group 2 process vent or comply directly with the requirement to reduce organic HAP emissions by 98 weightpercent or to an outlet concentration of 20 ppmv through use of a control device. There are three group determination procedures: (1) Process vent flow rate measurement, (2) process vent HAP concentration measurement, and (3) TRE index value determination. The specific test methods for these three determinations are described under section VII.B.2.c ''Test Methods'' of this notice. Process vents with a flow rate less than 0.005 scm/min are considered Group 2 process vents. Vent streams {pg 62637} with flow rates less than 0.005 scm/min are expected to have a TRE greater than 1.0. The flow rate measurement would allow sources with very low flows a less burdensome way to determine if they are Group 2 (instead of performing a TRE calculation). Process vents with a HAP concentration less than 50 ppmv are considered Group 2 process vents. Vent streams with organic HAP concentrations less than 50 ppmv are expected to have a TRE greater than 1.0. The HAP or TOC concentration measurement using Method 18 or 25A (instead of performing a TRE calculation) would allow sources with very low organic HAP concentrations a less burdensome way to determine they are Group 2. An analysis using the HON TRE equation and model process vent stream characteristics in the HON data base showed that process vents with flow rates or HAP concentrations below these levels would have TRE index values above 1. Process vents with a TRE index value greater than 1.0 are considered Group 2 process vents. The TRE index value can be calculated by using inputs derived from engineering assessment (including process knowledge) or test method measurements. The inputs to the equations are the HAP and TOC concentrations, the net heating value, and the flow rate of the vent stream. If the TRE index value calculated using engineering assessment is greater than 4.0, then the owner or operator would not be required to measure stream characteristics. The TRE of 4.0 was selected by considering the uncertainty of the engineering assessment of the inputs to the TRE equation and process variability. If a TRE index value is calculated to be above 4.0, it was determined that it is very unlikely that either uncertainties in estimating inputs to the equation or normal process variations could cause the actual TRE index value to be below 1.0. Therefore, the vent can be considered Group 2 without making measurements. However, engineering assessment procedures must meet the specifications in the regulation and must be fully documented in order to assure that the TRE calculation is acceptable. If the TRE index value is less than 4.0, testing using the appropriate flow rate method and Method 18 is required to determine flow rate, HAP and TOC concentrations, and net heating value for input to the TRE equation. b. Performance test. Initial performance tests are required for all control devices other than flares and certain boilers and process heaters. Specifically, testing would be required for: (1) Incinerators, (2) scrubbers used with combustion devices to control halogenated vent streams, and (3) some boilers and process heaters smaller than 44 MW (150 million Btu/hr). Performance tests: (1) Ensure that a control device can achieve the required control level and (2) help establish operating parameters that are indicative of proper operation and maintenance. An initial performance test is not required for boilers and process heaters larger than 44 MW (150 million Btu/hr) because they operate at high temperatures and residence times. Analysis shows that when vent streams are introduced into the flame zone of these boilers and process heaters, over 98 percent reduction or an outlet concentration of 20 ppmv is achieved. Therefore, a performance test is not necessary. This is more fully explained in the proposal notice for the SOCMI reactor processes NSPS (55 FR 26966, June 29, 1990). Unlike the proposed SOCMI reactor processes NSPS, the proposed HON would not require a performance test for boilers that mix the vent stream with the primary fuel to a boiler or process heater because available information shows that in such situations boilers and process heaters achieve 98 percent control or better. Because percent reduction and outlet concentration cannot feasibly be measured at flares, the flare must meet the requirements in Sec. 63.11 for operating conditions. c. Test methods. The proposed process vent provisions would require the use of approved test methods to ensure consistent and verifiable results for group determination procedures, initial performance tests, and compliance demonstrations. Performance tests are required to demonstrate compliance for control devices other than flares and certain boilers and process heaters. For group determination, Method 2, 2A, 2C, or 2D of 40 CFR part 60, appendix A is specified for measuring vent stream flow rate (prior to combustion). Also, Method 18 or 25A of 40 CFR part 60, appendix A is specified for measuring total vent stream HAP or TOC concentration to determine whether HAP concentration is below 50 ppmv. Method 18 measures individually- speciated organic compounds. Method 25A measures total organic compound content of the vent stream and does not speciate organic HAP; thus, it includes any methane and ethane in the vent. Method 25A may be used only if a single organic HAP compound is greater than 50 percent of the total organic HAP in the vent stream and that HAP compound is used for calibration. In addition, if Method 25A is used, TOC must be below 25 ppmv (instead of 50 ppmv as in Method 18) for a vent to be considered Group 2. A safety factor of 2 is being applied because the method will measure the HAP to which it is calibrated, but could under-estimate other HAP's in the vent stream. Method 18 is specified for measuring TOC and HAP concentrations for use in the TRE equation because concentrations of individually- speciated organic compounds are needed to calculate the net heating value for input to the TRE equation. In order to determine whether a vent stream is halogenated and to calculate TRE, Method 18 is specified for measuring speciated halogenated organic compounds. The total vent stream concentration of total halogen atoms shall be summed from the individual halogen atoms in each organic HAP compound based on the molecular formula of the compound and the concentrations of the compounds containing halogens. For example, 150 ppmv of ethylene dichloride would contain 300 ppmv of total halogen atoms. Process knowledge that no halogenated compounds are present is acceptable for determination that a vent stream is not halogenated. Method 26 or proposed Method 26A is specified for measuring halogens and hydrogen halides from scrubbers following combustors. Proposed revisions to Method 26 and the proposed addition of Method 26A to 40 CFR part 60, appendix A are discussed in a separate notice in this issue of the Federal Register. In order to allow owners or operators greater flexibility, the proposed provisions also allow the use of any test method or test results validated according to the protocol in Method 301 of 40 CFR part 63, appendix A. The EPA considered allowing Method 25A as an alternative to Method 18 for demonstrating compliance of control devices applied to process vents; however, Method 25A is not included as an alternative for demonstrating compliance with the emissions reduction or control device outlet concentration in the proposed rule. The basis for the decision was that the EPA determined that the results obtained with Method 25A would not consistently demonstrate HAP control efficiency. Process vent streams often contain mixtures of multiple organic HAP's and other organic compounds. The TOC measurements obtained with Method 25A would vary depending on {pg 62638} how the method is calibrated because response factors for individual compounds vary. Furthermore, some compounds such as formaldehyde and halogenated compounds are not well detected by Method 25A. Another concern is that the relative proportion of individual organic compounds may change across the combustor. Therefore, specifying calibration with the principal HAP in the inlet would not necessarily produce reliable results. The EPA requests comments and data on whether the use of Method 25A would provide accurate measurements of TOC control. In particular, EPA requests comment on the two following procedures for using Method 25A: Procedure I: a. Calibrate Method 25A with the primary organic HAP constituent (i.e., greater than 50 percent of the total organic HAP by volume) at the control device inlet; b. Measure greater than 99 percent TOC reduction (to be more conservative, as opposed to 98 percent reduction measured by Method 18) or; c. Measure less than 10 ppmv TOC concentration at control device outlet (to be more conservative, as opposed to 20 ppmv measured by Method 18). Procedure II: a. Calibrate to propane; b. Measure 99 percent reduction or 10 ppmv TOC at outlet. The EPA also requests comments and data on other potential test methods or procedures that would provide accurate TOC reduction measurements. In particular, data comparing vent stream composition at the inlet and outlet of combustion devices are requested. Data comparing results of Method 25A and Method 18 tests for combustion of vent streams composed of multiple organic compound mixtures are requested. d. Monitoring. Control devices used to comply with the proposed standard need to be maintained and operated properly if either a 98 percent reduction or reduction to an outlet concentration of 20 ppmv is to be achieved on a continuing basis. Monitoring of the control device operating parameters can be used to ensure that such proper operation and maintenance are occurring. The proposed standard lists the parameters that can be monitored for the common types of combustion devices: Thermal incinerators; catalytic incinerators; boilers and process heaters; and flares. These parameters were selected because they are good indicators of combustion device performance, and instruments are available at a reasonable cost to continuously monitor these parameters. These parameters are generally the same as those required by the reactor processes, air oxidation, and distillation NSPS. The rationale for their selection is fully explained in the proposal notice for the SOCMI reactor processes NSPS (55 FR 2966- 26968, June 29, 1990). The proposed rule also allows the owner or operator to request to monitor other parameters on a site-specific basis. The proposed standard would require the owner or operator to establish site- specific parameter ranges through the Notification of Compliance Status report and operating permit. Site-specific parameter ranges accommodate site-specific differences in control design and process vent stream characteristics. The EPA requests comment and data on whether an alternative range or minimum value of monitored parameters should be set and, if so, what the values should be. Unlike the proposed SOCMI reactor processes NSPS, the proposed HON would not require monitoring of boilers and process heaters of 44 MW (150 million Btu/hr) or greater or of boilers and process heaters below 44 MW (150 million Btu/hr) that introduce the process vent stream as a primary fuel or mix it with the primary fuel and introduce it through the same burner. This decision was made because the burning characteristics of these units generally ensure a 98 percent reduction in the organic content of the process vent stream. The proposed rule also specifies monitoring requirements for scrubbers installed to remove halogens and hydrogen halides from the combustor outlet. For Group 2 process vent streams that have TRE index values greater than 1.0 but less than or equal to 4.0, monitoring of the final recovery device would be required to ensure that it continues to be operated as it was during the group determination test when the initial TRE index value was calculated. Improper recovery device operation and maintenance could lead to increased organic HAP concentration, potentially reducing the TRE index value below 1.0, and causing the vent to become a Group 1 process vent. Continuous monitoring will ensure continued good performance of recovery devices. The TRE index value monitoring level of 4.0 is being proposed because the variability of the process parameters established during normal operating conditions are unlikely to vary to the extent that a TRE value above 4.0 would be reduced to a TRE level less than 1.0 and thus require control. The proposed standard specifies the parameters that can be monitored for the three common types of recovery devices: absorbers, condensers, and carbon adsorbers. These parameters were selected because they are good indicators of recovery device performance, and instruments are available at a reasonable cost to continuously monitor these process parameters. These monitoring parameters are the same as those already required by the reactor processes, air oxidation, and distillation NSPS. The rationale for their selection is fully explained in the proposal notice for the SOCMI reactor processes NSPS (55 FR 26968-26969, June 29, 1990). The proposed rule also allows the owner or operator to request to monitor parameters on a site-specific basis. The owner or operator would establish a site- specific range for the parameters through the Notification of Compliance Status report and operating permit. C. Selection of Storage Vessel Provisions 1. Selection of Format For Group 1 storage vessels, the storage vessel provisions require control by: (1) Tank improvements (internal or external floating roofs with proper seals and fittings) or (2) a closed vent system and control device. The format for the storage vessel provisions is dependent upon the control device selected. For storage vessels controlled by internal or external floating roofs, the format is a combination of design, equipment, work practice, and operational standards because all of these are necessary to ensure that the vessel will meet the reference control technology requirements. The EPA chose not to propose an emission limit format for storage vessels because that would require equipping each vessel with a capture system; the corresponding costs would be prohibitive. The design requirements for vessels controlled with tank improvements are five different equipment configurations specified in the rule. Additional operational and work practice requirements, which consist of inspection and repair requirements, are necessary to ensure the continued integrity of the control equipment. For vessels controlled by a closed vent system and control device, EPA is proposing a design and equipment format. This format accommodates the wide variation in emissions and flow rates being vented from the vessel, and requires that the closed vent system and control device meet the reference control technology requirements. The closed vent system must be capable of collecting HAP vapors and gases discharged from the storage vessel. {pg 62639} The control device must reduce the HAP emissions discharged into it at an efficiency of at least 95 percent by weight and must be operated to achieve this level of emission reduction. Operational requirements, which consist of, among other things, inspection, repair, and work practice requirements, are necessary to ensure the proper operation and integrity of control equipment meeting a design and equipment standard. 2. Selection of Compliance Provisions The proposed storage vessel provisions require control by tank improvements or a closed vent system and control device; however, the choice of control technologies is limited depending on the material stored. For vessels storing liquids with vapor pressures less than 76.6 kPa, either control option may be selected. However, for vessels storing liquids with vapor pressures greater than or equal to 76.6 kPa, tank improvements do not achieve the expected level of emission reductions. As a result, Group 1 storage vessels containing liquids with a maximum true vapor pressure of organic HAP's greater than or equal to 76.6 kPa must be controlled with a closed vent system and control device. Tank improvements would not be allowed as the reference control technology for these vessels. Compliance provisions in the proposed rule are dependent upon the control device selected. The following discussion separately addresses the provisions for internal floating roofs, external floating roofs, and closed vent systems and control devices. a. Internal floating roof vessels. After a vessel is filled, it is impossible to accurately ascertain the condition of the primary seal. Additionally, most repairs cannot be performed on an internal floating roof that is in service. For these reasons, for storage vessels at new sources, the proposed standards would require the owner or operator to inspect and report the condition of the internal floating roof, seals, and other required equipment before placing the storage vessel in service. Because internal floating roofs and seals can fail, resulting in an increase in emissions, owners and operators are required to inspect each storage vessel at new and existing sources periodically and to repair any failures. If failures are detected during an inspection, the vessel must be repaired or emptied within 45 days. This 45-day limit was selected because survey data indicated that most facilities could empty a storage vessel having equipment in need of repair within 45 days. In addition, two 30-day extensions may be requested from the Administrator. As with the previous NSPS for storage vessels and the benzene NESHAP, two types of inspections are required by the proposed rule: (1) Visual inspection of the internal floating roof and seals through the manholes and roof hatches on the fixed roof (i.e., by an observer or television monitor); and (2) internal inspection of the internal floating roof, seals, and deck fittings of a vessel that has been emptied and degassed. To allow flexibility, the proposed rule includes two different schedules for inspections. The first schedule would require a visual inspection at least once every year and an internal inspection at least once every 10 years. The second schedule, which may be used only if a secondary seal is in place, would require an internal inspection at least once every 5 years but would not require annual visual inspections. The owner or operator may find it necessary on occasion to empty and degas a storage vessel for reasons other than equipment inspections (e.g., to repair a failure detected during an annual visual inspection). In order to further reduce the emissions due to degassing for inspections, the proposed rule allows an internal inspection any time a storage vessel is degassed for any purpose. The date of this inspection would become the beginning of the 5- or 10-year time period before the next required inspection. b. External floating roof vessels. As described in section VII.C.2.a for the internal floating roof vessels, the proposed provisions for the external floating roof storage vessels at new sources would require the owner or operator to inspect and report the condition of the seal system of the external floating roof before placing the storage vessel in service. In order to ensure the continued effectiveness of external floating roof controls, EPA is proposing the following requirements. The owner or operator must: (1) Measure the seal gaps in both the primary and secondary seals within 90 days of introducing HAP's into the vessel; (2) measure the primary seal gap once every 5 years; and (3) measure the secondary seal gap once every year. In addition, the proposed rule requires that the vessel be visually inspected each time the vessel is emptied and degassed because any failures in the seals would have to be repaired before the vessel is refilled. Whenever seal gaps exceed the limits specified in the rule, the owner or operator would be required to repair the seal or empty the vessel within 45 days unless an extension is granted. However, the proposed rule would allow delay of the seal gap measurement if the owner or operator determined that the floating roof was structurally unsound and that it would be unsafe to conduct the inspection. To minimize the potential for an increase in emissions, the proposed rule would require that the inspection be performed within 30 days of the determination that the roof is unsafe or that the vessel be emptied within 45 days of the determination. In addition, two 30-day extensions for emptying the vessel may be requested from the Administrator. c. Closed vent systems and control devices. To enable EPA to determine compliance with the requirements for the closed vent system and control device, the proposed rule requires the owner or operator to submit plans and specifications for the system to the EPA as part of their Implementation Plan. In addition, because closed vent systems and vapor control devices are also subject to failures or improper operation, the proposed standard requires periodic inspection of closed vent systems for leaks. Many failures can be detected by regular inspection of operational parameters. Therefore, the proposed standard requires the owner or operator to monitor those operational parameters that would indicate that the device is operating properly. To ensure the integrity of the closed vent system and control device, the proposed standard also requires that inspections for leaks greater than 500 ppm be performed during filling of the vessel and at least once per year. Leaks must be repaired no later than 15 days after being detected. However, repair of the leaks may be delayed until the next process unit shutdown if the owner or operator demonstrates in a report to the Administrator that the repair is technically infeasible without a process unit shutdown or that emissions of purged material resulting from immediate repair would be greater than the fugitive emissions likely to result from delay of repair. D. Selection of Transfer Loading Operations Provisions 1. Selection of Format For Group 1 racks, the transfer provisions require vapor collection and control. The format for the provisions for vapor collection systems includes equipment design and work practice standards to ensure that the HAP- containing vapors are collected and routed to the control device or vapor balancing system. To ensure vapors are captured by the collection system and {pg 62640} are not lost to the atmosphere through leaks in the vehicle, the provisions include a work practice standard requiring owners and operators of transfer racks subject to control to load organic HAP's only into tank trucks and railcars that are DOT certified or are vapor tight. The format of the proposed provisions for control devices depends upon the control device selected. For streams controlled by control devices other than flares, the format is a combination of a weight-percent reduction and an outlet concentration. A weight-percent reduction is required for streams with HAP concentration above 1000 ppmv because such a format ensures that the stream will meet the reference control technology requirements. As mentioned in Section VII.B, ''Selection of Process Vents Provisions,'' the 20 ppmv level is needed as an alternative format for sources with inlet organic HAP concentrations below about 1,000 ppmv. These streams may not be able to achieve 98 percent control, but can achieve an outlet concentration of 20 ppmv. As with process vents, scrubbers are required to remove halogens and hydrogen halides from combustor outlet streams when streams containing halogenated compounds are combusted. The format for the scrubber compliance is a combination of percent reduction and an outlet concentration limit. For streams controlled by a flare, an equipment standard with stated equipment and operating specifications is being proposed as the format because it is very difficult to measure the emissions from a flare to determine its efficiency. For vapor balancing systems, an equipment and work practices standard is being proposed. Vapor balancing systems are required to return emissions from transfer operations back to the storage vessel from which the liquid being transferred originated. If the vapor balancing system is properly operated and inspected for leaks, there should be virtually no emissions from a vapor balancing system. Thus, the proposed provisions require that the system be inspected annually for leaks. 2. Selection of Performance Tests, Monitoring Requirements, and Test Methods a. Initial performance test. Performance tests: (1) Ensure that a control device applied to a Group 1 rack can achieve the required control level and (2) establish operating conditions under which the device should continue to achieve the required level of control. For these reasons, an initial performance test would be required for all control devices except: (1) Flares; (2) boilers and process heaters with design heat input capacities of 44 MW (150 million Btu/hr) or greater; and (3) boilers and process heaters with design heat input capacities less than 44 MW that introduce the process vent stream as a primary fuel or mix it with the primary fuel and introduce it through the same burner. A discussion of the rationale for excluding flares and boilers and process heaters is presented in the discussion of process vents, section VII.B.2 of this preamble. Because the format for the provisions for vapor balancing systems includes equipment design and work practice standards, a performance test is not required. A vapor balancing system inherently prevents emissions to the atmosphere. The requirement for annual leak detection and repair ensures that fugitive emissions from the system will be minimal. As described in Sec. 63.128 of subpart G, a performance test measuring the percent reduction or outlet concentration would not be required for streams controlled by a flare. However, the flare must comply with Sec. 63.11 which includes a compliance determination according to Method 22 of 40 CFR part 60, appendix A, and design specifications for velocity and heat content. The EPA attempted to specify a performance test duration that would provide a representative test period to measure control device performance and establish monitoring parameter ranges. Emissions and control performance may vary over the filling cycle and over the control device cycle where intermittent vapor processing systems are used. Intermittent vapor processing systems are systems that have a vapor holder to accumulate vapors and treats the vapors during automatically controlled cycles. It was considered that filling duration, filling frequency, and control device cycle duration will be different from facility to facility. The performance test period should be long enough to obtain representative results, but short enough that the test is not unduly burdensome for the source. The proposed performance test duration is three filling periods for continuous vapor processing systems; and three control device cycles for intermittent vapor processing systems. The EPA requests comments or suggestions with supporting data and/or rationale on the proposed performance test duration. b. Test methods. The proposed standard requires the use of approved test methods to ensure consistent and verifiable results for initial performance tests and compliance demonstration. The transfer provisions allow the use of either Method 18 or Method 25A for compliance with the percent reduction and outlet concentration. Method 25A is proposed as a method for determining emissions from transfer racks but not process vents because emission characteristics make this method more appropriate for transfer operations. Transfer operations typically load products of known composition. One, or at most a few, organic HAP's would be routed to the control device at one time. Therefore, Method 25A can be calibrated to the specific organic HAP being controlled and will provide a reliable measurement of control device inlet and outlet emission levels. This is in contrast to the situation described in Section VII.B.2 for process vents containing complex mixtures of several organic compounds. However, EPA requests comments and data on the accuracy and use of these methods for transfer operations and any data comparing the Method 25A results to Method 18 results. As with process vents, Method 26 or proposed Method 26A is specified for measuring halogens and hydrogen halides from scrubbers following combustors. Proposed revisions to Method 26 and the proposed addition of Method 26A to 40 CFR part 60, appendix A are discussed in a separate notice in this issue of the Federal Register. The proposed rule also allows the use of any test method or test results validated according to the protocol in Method 301 of 40 CFR part 63, appendix A to allow owners or operators greater flexibility. c. Vapor tightness testing. The proposed transfer provisions require that organic HAP's be loaded only into vehicles that are DOT certified, or vehicles that have been determined to be vapor tight to ensure that emissions generated during loading are captured. Previous regulations required vapor tightness testing for tank trucks and railcars according to Method 27 of 40 CFR part 60, appendix A. This requirement was reconsidered for the HON in light of updated DOT regulations. In order to decrease the burden to SOCMI facilities and to tank truck and railcar owners, the proposed rule allows either DOT certification in accordance with pressure test requirements of 49 CFR 180 for tank trucks and 49 CFR 173.31 for railcars, or vapor tightness testing according to Method 27. d. Monitoring. The proposed standard lists the parameters that must be {pg 62641} monitored for the common types of control devices (i.e., thermal incinerators, catalytic incinerators, flares, boilers, process heaters, absorbers, condensers, and carbon adsorbers). These parameters were selected because they are good indicators of combustion or recovery device performance, and instruments to monitor these parameters at the frequency required are available at a reasonable cost. The rationale for their selection is fully explained in the proposal notice for the SOCMI reactor processes NSPS (55 FR 26966-26969, June 29, 1990). The proposed rule also allows the owner or operator to request to monitor other parameters on a site-specific basis. The proposed standard would require the owner or operator to establish site- specific parameter ranges through the Notification of Compliance Status report and operating permit. Site-specific parameter ranges accommodate site-specific differences in control design and vent stream characteristics. The EPA requests comment and data on whether an alternative range or minimum value of monitored parameters should be set and, if so, what the values should be. In previous rules, the frequency of monitoring parameters has been linked to the duration of the loading cycle. In order to give flexibility, the proposed HON rule gives owners and operators a choice whether to link the monitoring frequency to duration of loading cycle or length of time the control device is operating. If the owner or operator decides to base the monitoring program on the duration of the loading cycle, measurements would have to be made every 15 minutes for loading cycles of 3 hours or more and measurements would have to be made every 5 minutes for loading cycles less than 3 hours in duration. If the owner or operator decides to base the monitoring program on the length of time the control device is operating, measurements would have to be made every 15 minutes for control device operating periods of 3 hours or more, and measurements would have to be made every 5 minutes for control device operating periods less than 3 hours in duration. The monitoring frequency option chosen must be recorded in the Notification of Compliance Status report. For vapor balancing systems, the proposed provisions require an annual visual inspection and a test using Method 21 of 40 CFR part 60, appendix A. The purpose of these provisions is to ensure that there are no leaks in the vapor balancing system. E. Selection of Wastewater Collection and Treatment Operations Provisions 1. Selection of Format The provisions for controlling air emissions from Group 1 wastewater streams are a combination of equipment, operational, work practice, and emission standards. The rationale for choosing the format of these provisions is discussed below. a. Transport and handling equipment. Several formats were considered in developing the proposed provisions for transport and handling equipment. These formats included a numerical emission standard, and an equipment standard with the work practices necessary to ensure proper operation and maintenance of the equipment. Although considered first, it was determined that a numerical standard would not be feasible because it would be difficult to capture and measure emissions from this equipment for the purpose of evaluating compliance. Due to the number of openings and possible emission points, accurate measurement would require enclosure of the entire airspace around a piece of equipment. This would not be practical for numerous equipment components. The second format considered was an equipment standard. Since the intent of the standard is to capture all emissions from transport and handling equipment, an equipment standard is appropriate. The best method for controlling emissions is to require the installation and proper maintenance of roofs, covers, lids, and enclosures on vessels and tanks. Based on the evaluation of these formats, a combination of equipment standards and work practices was selected for transport and handling equipment. The proposed provisions would require that emissions from wastewater transport and handling equipment be controlled from the point of wastewater generation through the treatment system. Equipment used to control emissions would include covers, lids, roofs, and enclosures designed to eliminate emissions. Proper work practices are needed to ensure that the equipment will control emissions. The proposed work practices include periodic monitoring, inspection, and repair provisions. These provisions were selected at action levels designed to minimize emissions, and are consistent with past NSPS and NESHAP standards. b. Reduction of volatile organic hazardous air pollutant concentration in the wastewater streams. Two formats were considered in developing the proposed standards for reduction of wastewater stream VOHAP concentration: a numerical format, and an equipment design and operation format. Five alternative numerical emission limit formats are proposed to provide sources with a maximum degree of operational flexibility in complying with the provisions. These emission limit formats are: (1) An overall percent reduction for total VOHAP in the wastewater stream (existing sources); (2) percent reductions for individually-speciated HAP's (new and existing sources); (3) an effluent concentration limit for total VOHAP (existing sources); (4) effluent concentration limits for individually-speciated HAP (new sources); and (5) a required mass removal for HAP (new and existing sources). The rationale for providing alternative emission limits based on both a percent reduction and an effluent concentration is given below. Percent reduction. The percent reduction format is based on the organic HAP removal efficiency of a steam stripper; however, any treatment process that can achieve the proposed efficiency can be used to comply with the standard. This format was chosen because it is the best representation of control technology performance. A second alternative limit is based on the percent reduction for individually- speciated HAP's. Some HAP compounds have low volatilities and cannot be removed as easily by steam stripping as other compounds. Wastewater streams composed mostly of compounds with low volatility may not be able to achieve the total HAP percent reduction. Therefore, the organic HAP compounds have been grouped by Henry's Law constants into three strippability groups. (Strippability refers to the predicted removal efficiency of a compound using the design steam stripper specified in the regulation.) Target percentage removals for HAP in each group have been developed. Sources may choose to use this alternative emission limit. Because the objective of the proposed regulation is to control air emissions of HAP's from wastewater streams, and not to control HAP's in the wastewater streams, this approach will result in adequate control of air emissions of organic HAP's within the range of volatilities. Effluent concentration. The effluent concentration limits are also based on the performance of a steam stripper. Effluent concentration limits are provided as alternatives to the percent reduction standard to allow compliance flexibility for facilities required to treat wastewater streams having low organic HAP concentrations. Requiring a {pg 62642} percent reduction standard alone for these wastewater streams would not be reasonable. At very low concentrations, it is technically much more difficult and costly to achieve the same level of percent reduction. Mass removal. Required mass removal is an alternative for combined wastewater streams where Group 1 wastewater streams might be mixed with Group 2 wastewater streams. It is based on the removal performance of a steam stripper for the different strippability groups of compounds. The equations for computing required mass removal are designed to require removal of only the VOHAP concentration contributed by the Group 1 wastewater streams. The mass removal alternative was provided in lieu of concentration limits because concentration limits could be achieved by dilution and, therefore, no emission reduction would occur. Equipment design and operation. Another regulatory format considered for wastewater stream treatment was an equipment design and operation format. The equipment standard consists of the installation of a steam stripper designed and operated at specified parametric levels. The specifications for the steam stripper were developed to provide a standard piece of equipment (with associated operating conditions) that can achieve greater than 95 percent total HAP removal for most wastewater streams, and greater than 99 percent for streams containing primarily high volatility compounds. This equipment format was included to provide an alternative means of compliance that all sources would be able to use, while achieving the desired emission reduction. In addition, the monitoring requirements for the design steam stripper require much less effort on the part of the owner or operator. Thus, this equipment design and operation standard provides an alternative with fewer performance testing requirements. c. Vapor recovery or destruction devices. An emission standard is generally appropriate for vapor recovery and destruction devices used to control vapor streams containing HAP from transport, handling, and treatment equipment. The rationale for an emission limit (percent reduction or outlet concentration format) for this equipment was discussed previously in the sections titled ''Selection of Format'' for process vents and transfer racks. A weight-percent reduction of 95 percent is proposed for all types of devices except flares. The 95 percent efficiency level will allow the use of product recovery devices, and thus encourage product recovery and pollution prevention. Two alternative formats are also proposed for combustion devices other than flares: An outlet volume concentration of 20 ppmv, or a minimum residence time of 0.5 seconds and a minimum temperature of 760 degrees C (1400 degrees F). For flares, an equipment standard with stated equipment and operating specifications is being proposed as the format because it is very difficult to measure emissions from a flare to determine its efficiency. 2. Selection of Performance Tests, Monitoring Requirements, and Test Methods a. Wastewater stream concentration and flow determination. Two important parameters must be quantified initially and whenever process changes are made to determine whether a process wastewater stream is a Group 1 or Group 2 stream. These parameters are the annual wastewater quantity for a stream and the VOHAP concentration of HAP's in the stream. The VOHAP concentration can be quantified as a flow-weighted annual average for total VOHAP or for individually-speciated HAP's. Several methods can be used to determine wastewater quantity. These methods include using knowledge about the capacity of the wastewater-generating process or the waste management unit, and using measurements that are representative of maximum annual wastewater generation rates. Knowledge-based methods are allowed to provide flexibility and to provide less expensive alternatives than actual annual measurement if the appropriate information is available. For quantifying the VOHAP concentration of the wastewater streams, three methods are available: (1) Knowledge of the wastewater streams, (2) bench scale or pilot scale test data, or (3) physical measurements of total VOHAP or individually- speciated VOHAP concentrations. Again, the three methods have been allowed to provide flexibility and to provide less expensive alternatives than actual measurement if the appropriate information is available. If the actual VOHAP concentration of the wastewater stream is measured, the regulation specifies that the procedures in proposed Method 305 (''Measurement of Individual Volatile Organics in Wastewater''), accompanying this package, must be used. This method is a laboratory test method for the measurement of emission potential of individual volatile organics in wastewater. It is based on a combination of Methods 25D and 18 of 40 CFR part 60, appendix A. Method 305 may be used in conjunction with calculations and procedures outlined in Sec. 63.144 of subpart G to determine VOHAP average concentration. To allow owners or operators flexibility, the proposed rule also allows the use of analysis methods which measure organic HAP concentrations in the wastewater streams, in conjunction with Method 301. The concentrations of the individual organic HAP compounds measured in the water may be corrected by multiplying each concentration by the compound-specific fraction measured factor in Table 13 of Sec. 63.144 of subpart G. This procedure will give an estimate of the volatile portion of the HAP which would be measured if the purge-and-trap procedure in proposed Method 305 (or Method 25D) were used. Method 301 provides a protocol for validating the accuracy of any alterative test method chosen by a facility in lieu of the methods specified by the regulation. The proposed rule also specifies procedures for determining the source-wide annual VOHAP loading. Also being proposed today is Method 304, ''Method for the Determination of the Biodegradation Rates of Organic Compounds.'' This method is a laboratory test method for the measurement of biodegradation rates of organic compounds and the calculation of biodegradation rate constants. Method 304 is required to be used in conjunction with calculations and procedures outlined in Sec. 63.145 of subpart G to demonstrate compliance of properly operated biological treatment units. b. Performance tests. Initial performance tests for control of Group 1 wastewater streams are not required by the proposed rule. For treatment processes and control devices, facilities have the choice of using either performance tests or engineering calculations to demonstrate the compliance of those units with the standards. Engineering calculations, supported by the appropriate documentation, have been allowed to provide a less costly alternative to that of actual testing. The proposed rule includes treatment process performance test procedures for the effluent concentration, percent reduction, and required mass removal standards. These test procedures involve measurements of total VOHAP concentration or individually-speciated VOHAP concentrations, depending on the standard for which compliance is being demonstrated. Similar to measurements of VOHAP concentration used to determine applicability, these measurements may be made using: (1) {pg 62643} Proposed Method 305 to quantify total VOHAP or individually-speciated VOHAP concentrations, (2) Method 25D alone to quantify VO as a surrogate for total VOHAP, or (3) any other method for which the results are validated using Method 301. A performance test is not specified for the design steam stripper; installation of the specified equipment, along with monitoring to show attainment of the specified operating parameter levels, demonstrates compliance with the equipment design and operation provisions. A vent stream control device performance test procedure is also specified by the proposed rule. Method 18 may be used to measure the organic content of the vapor stream entering and exiting the control device. Alternatively, any other method may be used if the results are validated using Method 301. All transport and handling systems and closed vent systems used to control emissions from them must be evaluated initially and at annual intervals using Method 21 to determine the presence of HAP-containing VOC fugitive emissions. Method 21 incorporates the use of a portable hydrocarbon detector to measure the concentration of VOC's. Method 21 is used to test compliance in several standards in 40 CFR parts 60 and 61 and represents the best available method for detecting fugitive emissions from these sources. The organic compounds measured by the hydrocarbon detector are not necessarily HAP's. However, if HAP's are contained in the transport, handling, and control equipment being tested, Method 21 is the best procedure available for providing an indication of fugitive HAP emissions. c. Monitoring requirements. The proposed process wastewater provisions include requirements for periodic monitoring and inspections to ensure proper operation and maintenance of the control system and continued compliance. Waste management units. These units are required to be visually inspected semiannually for improper work practices and control equipment failures which may potentially be a source of emissions. The equipment must also be tested annually for the presence of leaks. Wastewater treatment units. Monthly monitoring of effluent streams or influent and effluent streams will be required to ensure compliance with the numerical limits. As an alternative, continuous monitoring of parameters that indicate proper system operation can be used. When continuous monitoring is employed, daily inspection of monitoring records is required. If the wastewater treatment units used to comply with the provisions has any openings such as doors and hatches, initial and semiannual inspections will be required to verify that the system is being properly operated without emissions. If a design steam stripper is used to comply with the requirements, continuous monitoring of the operating parameters listed in the regulation will be required. Continuous monitoring is necessary to ensure proper operation of the stripper, and therefore maximize emission reductions. Closed vent systems and control devices. With the exception of a non- regenerative carbon adsorber, continuous monitoring will be required for all control devices. The monitoring equipment, parameters, and frequency of monitoring for each control device are given. Monitoring is necessary to ensure control devices continue to be properly operated and maintained. The parameters were selected because they are good indicators of control device performance, and instruments are available at a reasonable cost to monitor these parameters. Each control device is required to have specific types of monitoring equipment as itemized in the regulation. Other parameters may be monitored upon approval from the Administrator in accordance with the requirements specified in Sec. 63.143(d) of subpart G. Continuous monitoring requirements were selected because they are known to result in optimum control device performance. A nonregenerative carbon adsorber requires different frequencies because the carbon bed must be replaced at regular intervals. For collection systems transporting vapors to any device, there are two alternatives for ensuring that the collected vapors actually reach the control device without being diverted to the atmosphere. One is use of flow indicators in bypass lines that could divert the vent stream. These would produce continuous records. An alternative is the use of sealed or locked valves to prevent diversion. In this case, monthly inspection of the valves is required to ensure the integrity of the seals or locks, instead of continuous monitoring. 3. Review of Technical Considerations Underlying Selection Process for Process Wastewater Provisions Industry representatives have expressed a number of concerns with EPA's basic approach to estimating and controlling emissions, and estimating control costs. The overall view of industry is that emission estimates are overstated and control costs are understated. Underlying these overall opinions are a number of technical issues on the characterization of the industry, the emission models, and the assumptions in the cost analysis. This section of the notice presents summaries of these technical issues and requests comments and data on these issues. Consequently, there may be changes to the emission estimation and control cost methodologies. The final rule will be based on consideration of the revised estimates. a. Physical/chemical properties of HAP's. One concern, raised by industry representatives, is the range of required removal efficiencies of HAP's within each strippability group in Table 9 of subpart G entitled Organic HAP Strippability Groups and Target Removal Efficiencies. The removal efficiencies for the compounds were predicted based on physical and chemical properties of the chemicals, the design steam stripper conditions, etc. The compounds were grouped with others having similar removal efficiencies and then each group was assigned a target removal efficiency. As a result, three strippability groups were formed. The target removal efficiency for each strippability group is: Strippability Group A, 99 percent; Strippability Group B, 95 percent; and Strippability Group C, 70 percent. The concern is that the ranges represented in each strippability group could hinder a compliance demonstration for specific HAP's that cannot individually attain removal efficiencies at the level assigned to the strippability group. This could result because each strippability group comprises several HAP's and each HAP in each strippability group does not necessarily have the same removal efficiency that is assigned to the strippability group. For example, the removal efficiency that is achievable for a particular HAP in Strippability Group B might be 92 percent, while the target removal efficiency for Strippability Group B is 95 percent. The EPA is considering whether it is more appropriate to develop more strippability groups with smaller ranges of removal efficiencies in each group, or to assign an individual target removal efficiency for each HAP. To make the determination of whether to revise the strippability groups, additional information is needed for the current physical/chemical properties data base. Specifically, the information needed includes: (1) Experimental data and documentation for Henry Law's constants at 25 degrees C and 100 degrees C; (2) documentation (e.g., reaction kinetics) for HAP's that cannot readily exist in wastewater (e.g., due to rapid hydrolysis); and (3) documentation of HAP's that are difficult to remove by steam stripping. This information would be compared against the documentation EPA used to derive the fraction emitted values and strippability factors used in the development of the proposed regulation. The EPA requests that commenters provide data on the items listed above. b. Scenarios and modeling emissions. As part of the technical basis for estimating emissions, EPA developed three scenarios representing SOCMI wastewater collection and treatment systems. Equilibrium and mass transfer equations were used to model the emissions from the waste management units (e.g., individual drain systems, wastewater tanks, biological treatment units, etc.) in each of the scenarios. These scenarios and models are discussed in section V.A. of this preamble, in the BID, and in the document ''Industrial Wastewater Volatile Organic Compound Emissions- Background Information for BACT/LAER Determinations'' (EPA- 450/3-90- 004). Industry representatives questioned whether the scenarios are representative of SOCMI wastewater collection and treatment systems. Specifically, industry representatives pointed out that many facilities have installed traps on drains and seals on other waste management units, therefore controlling some air emissions from the systems. In response to these concerns, CMA developed an alternative scenario based on input from CMA member companies and provided it to EPA; however, the documentation for the scenario is not complete enough to provide a complete evaluation of how well it represents systems used at SOCMI facilities. In order to assess the scenario best representing the industry's wastewater collection and treatment systems, EPA requests information characterizing these systems as they are typically found at SOCMI facilities. If such additional documentation can be provided on the industry as a whole, EPA will consider revising the scenarios used as part of the basis of the proposed rule. Because EPA may revise the scenarios and because industry representatives have expressed concerns about some of the models used for estimating emissions from waste management units, EPA will be reevaluating some models between proposal and promulgation. Revisions to the models will reflect technical issues. The EPA requests results of studies measuring air emissions from waste management units, especially individual drain systems (e.g., drains, manholes, sumps, and junction boxes) as well as wastewater tanks and biological treatment units. c. Point of generation. The proposed rule requires that the owner or operator of a facility determine which process wastewater streams are Group 1 or Group 2. This determination is based on concentration and flow rate. The concentration and flow rate are determined by: (1) Measurements taken at the point of generation; or (2) calculations based on process knowledge or measurements not taken at the point of generation, and adjusted to estimate concentration and flow rate at the point of generation. Industry has commented that: (1) There are difficulties with characterizing the flow rate and concentration at the point of generation of a wastewater stream; and (2) the concentration and flow rate should be determined at the first air- water interface (e.g., a process sump). Industry representatives have stated that many newer chemical process units use complex piping systems that convey wastewater directly to collection tanks or sumps, making it impractical or potentially impossible to measure flow rate and concentration for each individual stream. Furthermore, industry representatives have stated that: (1) Engineering estimates using process knowledge are possible, but are difficult due to the considerable variations in the process, especially batch operations; and (2) material balance calculations would be most appropriate but may not be accurate enough when the number of streams and process variables become large. The basic foundation of the proposed provisions for process wastewater is to identify wastewater streams for control and treatment based on action levels at their point of generation, prior to dilution and air emissions losses. This approach, which is also discussed in Section VI.B.4 of this notice, focuses control efforts on the streams with the highest loadings. If dilution prior to determination of the need to control is allowed by changing the location of the point of generation, some streams that could have been treated cost-effectively would not be treated and some dilute streams that were mixed with more concentrated streams would be controlled less cost-effectively. The EPA believes that sufficient flexibility for compliance is in the proposed wastewater provisions without the need to change the point of generation. For example, the owner or operator may elect to use the alternative process unit compliance option. This compliance option allows mixing of process wastewater streams from the same process unit and does not require a concentration and flow rate determination at the point of generation unless the owner or operator wishes to demonstrate that certain streams are Group 2 streams. Comments are requested on the difficulties that could be experienced from the requirement to determine wastewater stream characteristics at the point of generation. Commenters opposed to determining characteristics at the point of generation should suggest alternative approaches to achieve the same result. d. Wastewater tanks. The proposed provisions require wastewater tanks storing Group 1 wastewater to be equipped with a fixed roof and a control device or a floating roof with specific rim seals and deck fittings. Comments have been received stating that for some wastewater streams, a fixed roof alone would achieve effective control and the control device would not be necessary. The EPA is considering establishing a threshold based on the total partial pressure of the HAP's in process wastewater stored in effected tanks or covered basins. The owner or operator of a wastewater tank storing Group 1 wastewater with a total HAP partial pressure greater than or equal to the threshold would be required to equip the tank with an internal or external floating roof or a closed vent system with a 95 percent efficient control device. The owner or operator of a wastewater tank storing Group 1 wastewater with a total HAP partial pressure less than the threshold would be required to use only a fixed cover. The threshold would not apply to wastewater tanks used as treatment processes. One option being considered for the partial pressure threshold is the pressure for the proposed provisions for storage vessels. e. Biological treatment technologies. While biological treatment units and other technologies may be used to comply with the HON, they must achieve a comparable control efficiency as the reference control technology, which has been proposed to be a design steam stripper. Concerns have been raised that EPA's analysis of the wastewater component of the floor did not reflect industry practice. Industry has stated that biological treatment units should be given more serious consideration as reference control technology. Two concerns have been raised by industry representatives: 1. Do well-operated and maintained biological treatment units in {pg 62645} conjunction with trapped individual drain systems define the wastewater component of the source-wide floor? 2. Is an appropriate reference control technology for biodegradable HAP's a biological treatment unit instead of the design steam stripper? As previously described, EPA's analysis of the floor level of control was based on existing air emission standards for each kind of emission point. Although EPA was aware that some degree of control of air emissions resulted from compliance with other regulations under CWA, OSHA, and RCRA, EPA did not have the information needed to evaluate the resulting emissions control efficiency that was being achieved. Specifically, information was lacking on the extent to which facilities used vapor suppression systems in the individual drain systems and waste management units. Late in the development of the wastewater provisions, the chemical industry reported that traps and seals are commonly used on components in the individual drain systems (e.g., which include drains, junction boxes, and manholes) and that many treatment components such as ponds and tanks are covered as well. If current industry practice for biodegradable HAP's is to suppress air emissions down to the biological treatment unit, then the existing analysis of the source-wide floor may include an underestimate of the control efficiency being achieved for wastewater. Consequently, EPA plans to evaluate the performance achieved by individual drain systems and biological treatment systems at existing facilities and then to reassess the source-wide floor. To do this analysis, a number of technical issues need to be resolved. Specific issues that must be resolved include: (1) Appropriate biokinetic data; (2) appropriate models to predict rates of volatilization; and (3) the best 12 percent of design and operating practices representative of the industry. The EPA expects to meet with industry representatives and other interested parties during the period between proposal and promulgation to obtain the necessary data and to resolve these and other technical issues. These issues are discussed briefly elsewhere in this section of the notice. The degree of control achieved with biological treatment systems depends on the biodegradability of the compounds and the system design. In some cases, high removal efficiencies have been reported, and industry sources have claimed that control performance for all degradable organics is generally quite good with overall removals exceeding 80 to 85 percent of the volatiles. Information on performance and characteristics of biological treatment units (e.g., retention time, aeration rate, aeration gas, mixed liquor suspended solids) will be needed from as many SOCMI sources as possible. The EPA will evaluate control options based on biological treatment and emission containment in individual drain systems. In order to evaluate biological treatment unit control options, additional information on individual drain systems and other information is required. If the results of EPA's evaluation indicate that biological treatment and emission containment in individual drain systems should be selected as a reference control technology, a supplemental proposal would be published in the Federal Register. Comments are requested on this issue and should address the specific concerns and data needs discussed in this section. Specific data EPA will need to evaluate biological treatment as a reference control technology are: 1. HAP-specific efficiency of biological treatment units in terms of mass destruction versus volatilization; 2. Design and operating parameters to attain a high degree of destruction; 3. Data on biodegradation rates for all types of systems; 4. Collection system design, specifically, the number and type of treatment units and their configuration; 5. Control devices used on treatment units (e.g., equalization basins and primary clarifiers) that occur prior to the biological treatment unit, and their efficiency for the control of air emissions; and 6. Identification of HAP's that do not readily biodegrade (e.g., carbon tetrachloride and vinyl chloride). It should be noted that for non-biodegradable HAP's, the design steam stripper would remain the reference control technology. The EPA is also evaluating biological treatment unit models and the biorate constant data base not only to reassess floor and reference control technology decisions, but also to ensure the most accurate results possible for compliance demonstration. Industry representatives have stated that kinetic constants obtained from different experiments for an individual chemical should not be used to predict biological treatment unit performance. However, not all the literature and data reporting experimental results are extensive enough to cover the range of constants necessary as inputs for the models that predict biological treatment performance. To reassess the data base used in the models, EPA requests additional data on biokinetic rate constants and will use the most scientifically defensible constants in the data base. The EPA will also compare the EPA-developed model, WATER7, for predicting biological treatment unit performance to other models such as PAVE, BASTE, and TOXCHEM, to determine if the results are similar. For models having results similar to WATER7 results, EPA may allow their use in compliance demonstration. A summary of the models EPA will evaluate is located in Table 5.1 entitled ''Computer-Based Fate and Transport Models'' (Docket A-90-23, section II-B). f. Test methods. The proposed wastewater provisions require use of Method 304, Determination of Biodegradation Rates of Organic Compounds, and modeling with WATER7, or another approved model, to predict HAP removal achieved in a biological treatment unit. The EPA is considering allowing WATER7 or other approved models to be used without Method 304. With this approach, site-specific input would not be used to derive the biodegradation rate or to establish parameter ranges. If models are used without Method 304 inputs, the model parameters would be required to match the biological treatment unit's operating parameters, such as effluent concentration. If the two sets of parameters are not consistent, the owner or operator would be required to reestablish parameter ranges and derive the biodegradation rate by running Method 304. Comments concerning safety issues were also raised by industry representatives. These safety issues have been addressed in proposed Method 304 by allowing alternative types of heaters (immersion heaters were originally specified) and by calling for headspace gas monitoring when formation of explosive gases is a possible concern. Another concern expressed is what triggers the need to perform Method 304 to demonstrate compliance. While it is clear that an initial demonstration of the biological treatment system's ability to biodegrade HAP's is necessary, it is not explicitly stated in the proposed provisions whether or when subsequent demonstrations are required. Examples of when Method 304 may be required to be run after the initial demonstration are: (1) Addition of a new process unit to the source or after a change in the characteristics of an existing process; (2) scheduled checks at least once every five years; and (3) whenever a performance test is required (e.g., {pg 62646} changes in established parameters or operation of the biological treatment unit). Proposed Method 305, Measurement of Individual Volatile Organics in Wastewater, is the method developed by EPA to provide a relative measure of the volatile organic emission potential of a waste or wastewater stream. Industry representatives report that because facilities already use other methods developed by EPA for Clean Water Act regulations to comply with their effluent discharge permits, facility owners or operators may prefer to use these other methods to demonstrate compliance with the HON. The proposed wastewater provisions allow any EPA-approved method to be used. However, because Method 305 is the basis of the concentration threshold for Group 1 wastewater in the proposed provisions, any other EPA method used for applicability determination and compliance demonstration will have to be validated in accordance with the procedures in sections 5.1 or 5.3 (as applicable) of Method 301 of 40 CFR part 63, appendix A. The results from the alternate method (if valid according to the procedures listed above) are then adjusted according to the fraction measured (F sub m) values listed in Table 13 of subpart G. Industry representatives have suggested that preapproval of Methods 304 and 305 would streamline the validation approval process, benefiting both industry and the regulatory authorities. Method 304 would be preapproved for specific apparatus configurations; Method 305 would be preapproved for specific compounds or ranges compounds for methods other than Method 305. The EPA agrees with this suggestion and requests validation data. This data would be used to establish a data base for public use on the EPA bulletin board system. The data base would list EPA methods approved as alternates for Method 305 and the compounds and concentrations for which the alternative method is valid (as defined by Method 301). Method 305 has a set of established heating and purging requirements, but does allow the owner or operator latitude in choosing sample stream trapping and analytical methodology as long as the recovery criteria listed in the method are met. The EPA requests information on trapping and analytical methodologies for specific compounds for discussion in this data base. The data base would also list sampling and analytical techniques for compounds which are appropriate for use in the ''back half'' (i.e., the speciation portion) of Method 305. These techniques would be suggested analytical systems for Method 305; the recovery requirements listed in Method 305 would still be required. The EPA requests that data be submitted to Docket Number A-90-19 (see ADDRESSES) and to Director, Technical Support Division, Office of Air Quality Planning and Standards, MD-14, Environmental Protection Agency, Research Triangle Park, North Carolina 27711. F. Selection of Emissions Averaging Provisions This section of the preamble presents the rationale for the proposed emissions averaging provisions (described in Sec. 63.150 of subpart G) and the alternative policies that were considered in developing these provisions. As part of the EPA's general policy of encouraging the use of market-based systems where they can be properly monitored and enforced, the Administrator is proposing to allow sources the option of using emissions averaging to comply with subpart G. Emissions averaging provides sources the flexibility to comply in the least costly manner while still maintaining a regulation that is workable and enforceable. The rationale for the specific provisions of the emissions averaging policy is detailed below. 1. The Scope of Emissions Averaging The Administrator proposes to allow emissions averaging across all the emission points, except equipment leaks, within a single new or existing source, as source is defined for the SOCMI source category. As such, emissions from the following kinds of emission points can be averaged: process vents, wastewater operations, storage vessels, and transfer operations. The Administrator is proposing to allow averaging across these 4 kinds of emission points in order to provide as much flexibility as possible while maintaining an enforceable standard. Equipment leaks are included in SOCMI sources, but they cannot be included in emissions averages because: (1) The negotiated standard for equipment leaks has no fixed performance level; and (2) no method currently exists for determining the magnitude of allowable emissions to assign for leaks. Without a method to determine the magnitude of allowable emissions to assign for equipment leaks, an averaging policy that included equipment leaks would be difficult to enforce. When methods are developed to assign allowable emission levels for particular leak points, EPA will consider revising the HON to allow the inclusion of equipment leaks in emissions averages. As previously described, emissions averaging, and in particular the scope of emissions averaging being proposed today, was selected to provide sources broad flexibility in compliance with the HON. Emissions averaging, as proposed, is designed to result in equal or lesser total emissions from any one source, compared to point-by-point compliance with the HON. Though the proposed emissions averaging policy does not specifically address the issue of toxicity, emissions averaging is not designed to result in more toxic emissions than point-by-point compliance. Owners and operators of SOCMI sources have an incentive to avoid increases in emissions of highly toxic chemicals under emissions averaging because such increases could result in additional controls being required after the subsequent evaluation of the residual risk associated with individual sources and the source category as a whole. Because the issue of calculating residual risk, under section 112(f) of the Act, comes up in the context of the HON, the EPA requests comment on whether residual risk should be calculated on a plant-wide basis, on a source category basis, or on the basis of some other reasonable alternative. The issue of how averaging will influence the potential toxicity of emissions from HON sources, and what EPA is considering in response to this issue, is further discussed in section VII.F.8 of this notice. In the future, EPA may regulate other industrial processes at SOCMI facilities (e.g., polymers and resins manufacturing). The emission points in these other industrial processes will not be part of the SOCMI source as defined in the proposed rule. These points cannot be included in emissions averages with the points comprising the sources in the SOCMI source category because the EPA interpretation of the floor requirement in the Act prevents averaging across sources. However, the EPA is seeking comment on a complementary legal interpretation of sections 112(d) and 112(i) of the Act. This legal interpretation is described in the following section. 2. Complementary Legal Interpretation for Broader Emissions Averaging The EPA's proposed rule allows an averaging approach for HON-covered portions of a plant. The EPA is also soliciting comment on a broader averaging alternative, as set forth below. Significantly, this latter averaging scheme is not being proposed as an alternative to the averaging approach, {pg 62647} described above. Rather, the two approaches are entirely complementary. However, because the legal issues posed by the broader averaging scheme differ appreciably from those implicated by the more narrow averaging proposal, the EPA believes that a separate discussion of the former is warranted. The broad averaging alternative relies upon a legal theory which would permit compliance with the HON MACT standard (or, for that matter, with any other MACT standard to be developed by EPA) by averaging emissions from points located anywhere within an entire contiguous facility, which contains HON-covered processing units. This alternative would differentiate between the term ''source in a category or subcategory,'' used in section 112(d), and the term ''source,'' used in section 112(i). To begin with, it is a maxim of statutory construction that the use of different words by Congress in two different subsections is not accidental and reflects well thought- out Congressional intent. See, Sutherland Stat. Const. Sec. 46.06 (4th Ed.). Specifically, for purposes of establishing emissions standards under subsection 112(d), Congress expected EPA to identify specific production lines, pieces of equipment, etc., located within major stationary sources or area sources, and to treat them as sources only for the purposes of subsections 112(d)(3)(A) and (B), i.e., for the purposes of determining what is the average emission limitation achieved by the best performing twelve percent of existing sources or average emission limitation achieved by the five best performing sources. Since the term source was used in subsection 112(d) for a distinct and unique purpose, the relevant statutory language contains a qualifying set of words- ''in a category or subcategory''-appended to the word ''source.'' Significantly, in subsection 112(i), the relevant statutory language contains the term ''source''-a fact that would allow EPA to conclude, for purposes of subsection 112(i), that a source which must achieve compliance is a ''major source'' which encompasses an entire plant. See, e.g., Chevron, U.S.A. Inc., versus NRDC, 467 U.S. 837(1984). This reading is supported by the fact that the definitions of statutory terms, set forth in subsection 112(a), contain the traditional definitions of such terms as ''major source,'' and ''stationary source.'' Under this construction, the statutory language ''any emissions standard, limitation, or regulation promulgated under this section 112 and applicable to a source'' describes all of the existing MACT standards, established under subsection 112(d), and applicable to a source, which contains within it ''sources in a category or subcategory'' (i.e., production lines, pieces of equipment, etc.) covered by such MACT standards. The EPA also notes that, in its ''early reductions'' rule, it adopted a plant- wide definition of source. See 56 FR 27342 (June 13, 1991)-an early indication that the Agency believed that an entire contiguous facility could be considered to be a source for purposes of qualifying for a 6 year extension of the MACT compliance deadlines. The EPA is mindful of the fact that it has traditionally enjoyed wide latitude in coming up with diverse definitions of the term ''source,'' depending on what regulatory objectives were implicated, and EPA's considerable discretion in this area has been affirmed by the courts. Chevron U.S.A. Inc., versus NRDC, 467 U.S. 837 (1984). The EPA solicits comment on the legal authority to adopt this approach, and also requests comment on implementation considerations. Under the broader averaging approach, EPA would make no changes in the way it has identified categories and subcategories of sources in the HON and devised appropriate emission standards for them, nor in the way it would prescribe future emission standards for other emission categories or subcategories. For compliance purposes, however, if an owner or operator does not wish to control a particular HON-covered emission point to the level that would result in compliance with the relevant HON emission standard, the extra emissions from that point could be offset by emissions reductions greater than what is required by any standard then in effect under Section 112 at one or more other emission points within the entire stationary source. In addition, emissions averaging would be allowed across all the emission points within the entire plant, including the SOCMI source category as well as points not yet covered by any MACT standard. As such, emissions from process vents, wastewater operations, storage vessels and transfer operations could be averaged among themselves or with other emission points within the entire stationary source, in order to provide as much flexibility as possible. However, a distinction would be made between emission points physically similar to emission points covered under the HON or any other MACT and all other emission points. For the former, to participate in the averaging scheme, they must comply with whatever compliance routines have been established under the SOCMI HON or other relevant MACT. For the latter, to participate in averaging, the source must petition EPA to establish an appropriate compliance routine. To be sure, once EPA establishes such a compliance routine for any new type of emission point, all similar emission points located within any sources can participate in the averaging, so long as they use the EPA-approved compliance routine. The EPA solicits comment also on how it would verify the baseline for unregulated emission points, and how it would structure the averaging compliance process to assure the standard as a whole remains enforceable. One possibility to consider would be to use the average emissions from such emission points over some representative time period as their baselines. While equipment leaks are included in SOCMI sources, they cannot at this time be included in emissions averages. Equipment leaks are not included because: (1) The negotiated standard does not require any fixed performance level; and (2) no practical method currently exists to verify performance. Under this alternative, EPA would expeditiously develop methods to assign allowable emission levels to particular leak points or groups of leak points, and would similarly develop methods for calculating emissions from emission points not regulated by the HON. The Administrator seeks comment on means to verify and document the emission performance levels achieved for situations where equipment leaks are included in an average. Commenters should include a description of the compliance verification procedure that would be used and what data needs to be supplied to support the suggested system. Comments are also requested on types of non-SOCMI emission points that might be included in broader averages. Under the broader approach, emission reductions from emission points not covered by the HON would be usable to generate credits, subject to the previously discussed limitations and any limitations added in any other MACT standard applicable to the relevant emission points. Credits could be generated at any time from non-HON emission points until they become subject to a MACT standard. When they become subject to a MACT standard, the future value of reductions at those emission points would be governed by the emission standards required by the new MACT. Stated differently, the offsets available from such points would have to be recomputed using the new {pg 62648} baseline established for them. Each future MACT standard must contain a floor calculation. To the extent that points covered by these later standards have already been controlled for purposes of participating in an averaging scheme, that relates to the compliance with an earlier standard, the floor for the later standard may be more stringent. As future MACT standards are adopted, EPA would attempt to maximize averaging opportunities and therefore minimize compliance delays and costs. The EPA believes that the broad averaging scheme, described above, provides significant policy benefits. To begin with, it would enable sources to achieve the same environmental gains, as under the more narrow averaging proposal, but at significantly reduced costs. Second, establishing MACT standards with a facility- wide averaging program could result in the MACT ''floor'' for future standards being tighter than would have been the case if the emission points outside of the already covered MACT categories would have stayed uncontrolled. Last, but not least, EPA believes that the utilization of the broad averaging scheme may result in the discovery of emission points within facilities that otherwise might have been overlooked by EPA and State regulators. A factor that must be considered in implementing this approach is that it would require a determination of the permissibility of allowing averages to emit a different combination of pollutants than sources that meet MACT without averaging. This concern is already being evaluated for VOC HAP's in the SOCMI source category. The alternative approach might further complicate this issue by adding dissimilar categories (e.g., industrial cooling towers) emitting dissimilar pollutants (e.g., chromium). Since the HON currently addresses only organic HAP's, significant changes would be necessary to accommodate non-VOC HAP controls. To be sure, this concern could be resolved by limiting trading under the broad averaging proposal to only organic HAP's already covered by a MACT standard, but emitted anywhere within the fenceline. As nonorganic HAP's become covered by subsequent MACT standards the EPA will develop averaging regimes covering such pollutants. To facilitate comment on this alternative, the EPA is setting forth below the regulatory changes that would be necessary to implement it. 1. Section 63.100. Applicability and designation of source. Amend to add paragraph (h), as follows: ''(h) subparts F, G, and H allow emissions averaging within the entire stationary source subject to the requirements of this Part, including additional information submittals and detailed credit/debit calculations.'' 2. Section 63.101. Definitions. Amend as follows: ''Emission point means an individual process vent, storage vessel, transfer rack, wastewater stream, or equipment leak, and for purposes of emissions averaging, any individual item of equipment within the stationary source. ''Stationary source means any building, structure, facility, or installation which emits or may emit any air pollutant subject to regulation under the ACT. Building, structure, facility, or installation means all of the pollutant-emitting activities which belong to the same industrial grouping, are located on one or more contiguous or adjacent properties, and are under the control of the same person (or persons under common control) except the activities of any vessel. Pollutant- emitting activities shall be considered as part of the same industrial grouping if they belong to the same Major Group (i.e., which have the same two-digit code) as described in the Standard Industrial Classification Manual, 1987 (National Technical Information Service stock number PB 87-100012).'' 3. Section 63.112. Emission limits. Amend paragraph (c), as follows: ''(c) Compliance with the emission standard in paragraph (a) or (b) shall be demonstrated in one of the following two ways: ''(2) The owner or operator may elect to control different groups of emission points with the stationary source to different levels than specified under Secs. 63.113 through 63.147 as long as the overall emissions reductions from the stationary source equal or exceed the emission reduction required by Sec. 63.112. ''(i) Owners or operators using this emissions averaging compliance approach must calculate their emission debits and credits for those emission points involved in the emission average as specified in Sec. 63.150, develop an Implementation Plan as required in Sec. 63.151, and comply with the general reporting requirements in Sec. 63.152. ''(ii) Emission debits and credits must be calculated separately for new and existing sources. New sources may be included in the same emission average as existing sources. The determination of whether an emission point is part of a new or existing source shall be made according to the provisions of subparts A and F of this part.'' 4. Section 63.150. Emissions Averaging Provisions. a. Amend paragraph (c), as follows: ''(c) The following emission points can be credited in an emissions average to offset use of controls less stringent than the reference technology on Group 1 emission points: ''(8) Any other emissions points within the stationary source except those excluded from use as credits in an emissions average by a standard promulgated under section 112.'' b. Amend paragraph (d), as follows: ''(d) The following emission points cannot be used to generate credits in emission averaging: ''(1) Emission points subject to Secs. 63.113 through 63.149 already controlled on or before (date of promulgation), except those that were controlled as part of the section 112(i)(5) early reduction program, the 33/50 program, or a pollution prevention program as described in paragraph (c) of this section. '' c. Amend paragraph (e) as follows: ''(e) For all points included in an emission average, the owner or operator shall: ''(2) Calculate credits for all emission points that are overcontrolled to compensate for the debits using equations in paragraph (g). Emission points that meet the criteria of paragraph (c) may be included in the credit calculation, whereas those described in paragraph (d) shall not be included. 5. Section 63.151. Initial Notification and Implementation Plan. ''(d) For emission points included in an emission average, the following information shall be provided in the Implementation Plan. ''(8.5) For each emission point included in the average of other than a process vent, storage vessel, transfer rack or wastewater stream, the owner or operator shall document the following information: ''(i) The information regarding emissions averaging required by any other standard proposed under section 112; or ''(ii) Information reasonably similar to information regarding emissions averaging required by any standard proposed under section 112. The Administrator may request any reasonably related supplemental information within 30 days after submission.'' The EPA further solicits comment on the legal and policy implications of allowing averaging between new and existing sources within a plant in reliance upon the legal interpretation outlined above. An alternative approach would be for the EPA to rely upon an interpretation of section 112(d) under which new and existing sources would be considered as comprising separate subcategories within a category. The EPA solicits comment on this interpretation in light of the approach to identifying subcategories that the EPA followed publishing the list of categories and subcategories required under section 112(c) of the Act (57 FR 31576, July 16, 1992). 3. Credits Under the proposed emissions averaging system, a source can get credits for emission reductions achieved after passage of the 1990 amendments to the Act if they result in greater emission reductions than required by the proposed rule for the relevant points. {pg 62649} There are three ways a source might generate credits: (1) Using control equipment that EPA agrees has a higher efficiency than the reference control technology on a Group 1 point; (2) controlling a Group 2 point; and (3) using a pollution prevention measure on a Group 2 point or using a pollution prevention measure that results in lower emissions than use of the reference control technology alone on a Group 1 point. In addition to the three ways of generating credits, the EPA is considering the feasibility and desirability of allowing credits for recycling activities which can be clearly shown to have resulted in quantifiable emission reductions. The EPA envisions that sources would be required to account for all emissions to the atmosphere during the entire recycling process and those emissions would reduce the amount of credits attributable to the recycling activity. At this point, the EPA has not determined which specific recycling activities can be shown to meet these requirements. The Agency intends to investigate this potential further and is therefore soliciting comments on what activities would qualify for this treatment and on procedures for quantifying credits and procedures for ensuring that all atmospheric emissions are accounted for. The EPA is interested in providing credits for recycling if the emission reduction can be quantified and made enforceable. The EPA is willing to include provisions in the final rule for crediting recycling if sufficient information on quantification, methodology, and enforceable mechanisms for such recycling measures is received during the public comment period. a. Reference efficiency ratings. Sources cannot use a type of control equipment or pollution prevention measure to generate credits unless the source can demonstrate the efficiency or level of emission reduction achievable through the measure from its use over time. For innovative control technologies that are different either in use or design from the reference controls, the effectiveness of the technologies must be demonstrated to the EPA or the operating permit authority prior to their use for compliance with the proposed rule. If the technology in question would be used for credit in no more than three applications in a given facility, the operating permitting authority can assign it a control efficiency as part of the permit review. However, if the permitting authority concludes that the technology has broad applicability or the source wishes to use it in more than three applications, the EPA will assign its control efficiency and publish a description of the technology in the Federal Register so that it is available for widespread use in averaging. This process is being proposed to encourage innovation in control technologies by establishing a relatively low approval hurdle for technologies with limited applicability and potential for return to the developer while maintaining a thorough federal review for technologies with the potential for broad applicability, widespread use and high returns. In addition, this process ensures that information about significant advancements in technology will reach a wide and varied audience, thus encouraging further innovation. The reference efficiency assigned by EPA to a new type of control equipment would be based on the level of emission reductions that could be expected from that equipment if it were in use in a variety of situations. The EPA has established this process because subpart G of the HON is a national standard and the reference efficiencies for new controls must be established at a level that can be met nationwide. In general, sources cannot get emissions averaging credit for the use of control equipment above its designated reference efficiency rating. For example, if the EPA considers a certain control to have a 98 percent control efficiency, the source cannot get credit for operating it above that efficiency. There is an exception to this policy for storage vessels controlled with closed vent systems with a 98-percent efficient control device and for process vents on which the source has demonstrated to the EPA that control can achieve 99.9 percent control. In addition, for these sources to get credit for the 99.9 percent control, the source must submit and the EPA must approve a continuous emissions monitoring plan for the relevant process vent and its control. The general policy of reference control efficiency ratings has been established because the reference efficiency levels are set at the minimum level of emissions reduction that is generally achieved by the control device. Many sources may, by simply applying the device, achieve greater emissions reduction than predicted by the technology's reference efficiency rating. To provide credits in this situation without requiring certain demonstrations of a higher efficiency would give the source a windfall and result in a net increase in emissions over the level that would be expected if there were no emissions averaging. This policy is fair to the sources, since the emission credits and the emission debits would usually be based on the same reference efficiency ratings. Emission debits, the amount of emissions that the source must make up for not controlling a Group 1 point, are based on the efficiency of the point's reference control technology. A source is not required to test the emission point to determine the level of control that would be achieved in practice if the reference control technology were to be applied at that point. If such testing were required, then the amount of emission debits for a particular point would, in many cases, be greater. In addition, to grant credits for the small amount of emission difference that might occur above a reference efficiency would lead to significant enforcement problems. It would be very difficult to ensure that, on a continuous basis, a technology achieves an emissions reduction above its reference efficiency rating. It would be even more difficult, if not impossible, for sources to prove to inspectors that they are in fact achieving these higher levels of efficiency. Use of a reference control efficiency for each control technology allows inspectors to ensure compliance merely by checking that the equipment is in place and operating properly. Use of reference efficiency ratings helps ensure that the emissions averaging system will result in the same or better emission reductions as a rule that does not allow emissions averaging. In addition, the use of reference efficiency ratings simplifies the emissions averaging system, thus making it enforceable. The system to provide higher nominal efficiency ratings for the storage vessels controlled with closed vent systems with a control device and for process vent reference control technologies in certain circumstances is being proposed to encourage the use of emissions averaging, the improved operation and maintenance of the reference controls, and expanded use of continuous emissions monitoring. The proposed rule is essentially a performance standard and does not require the installation of any particular controls. In fact, sources are encouraged to meet the performance level of the standard using the least costly method available, including emissions averaging. However, the level of the standard is based on the minimum level of control anticipated with application and proper operation and maintenance of the reference control technologies. The reference efficiency assigned to each of the reference controls also contributes to {pg 62650} the determination of the level of the standard. Some sources may be able to achieve higher levels of emissions reduction than the reference control efficiencies while using the reference control technologies because of the particular characteristics of their emission streams. Some would argue that this higher level of control, over the efficiency assigned to a reference control technology, should be allowed to be used for emissions averaging credits. The EPA agrees that significantly higher emission control levels should be credited, but believes that marginally higher levels should not. As a result, the EPA is proposing to allow higher nominal efficiency ratings only for those applications of process vent reference controls that can be demonstrated to achieve 99.9 percent control and controlled closed vent systems on storage vessels that achieve 98 percent control. In addition, the source would be required to institute an EPA-approved plan for continuous emissions monitoring at the point assigned the higher nominal efficiency for the reference control. The proposed rule does not allow a source to get a higher nominal efficiency rating for the reference controls for transfer operations, wastewater operations, or storage vessels (except controlled closed vent systems) because the technologies are not as well characterized. The EPA is seeking comment on how a source might use a reference control for a transfer operation, wastewater operation, or storage vessel to create control more stringent than the reference efficiency rating, and how monitoring might be used to ensure that the extra control was achieved on a continuous basis. In addition, the EPA is seeking comment on the inclusion of this system for assigning process vent reference controls a higher nominal efficiency rating. b. Pollution prevention credits. Credits can also be generated through use of a pollution prevention measure. For the purposes of the proposed rule, the EPA is referring to any pollution prevention activities described in the Agency's Pollution Prevention Strategy (56 FR 7849) that are applicable to this industry. The following activities are included in the description of pollution prevention: substitution of non-toxic for toxic feedstocks in making a product; alterations to the production process to reduce the volume of materials released to the environment; equipment modifications; housekeeping measures; and in-process, recycling that returns waste materials directly to production as raw materials. Other pollution prevention approaches that are identified in the EPA's Pollution Prevention Strategy and are applicable to this industry would also be acceptable for credit. The EPA solicits comment on the specific pollution prevention approaches identified in the strategy that might be applicable to this industry. The EPA proposes that shutdowns (i.e., permanent closures, not maintenance turnarounds) cannot be used to generate credits, even if they are part of an Early Reduction commitment under section 112(i)(5) of the Act, unless they are part of a change that qualifies as a pollution prevention measure as defined in the EPA's Pollution Prevention Strategy (56 FR 7849). One example would be if a source converts a chemical manufacturing process in an emissions averaging program from a HAP-using process to a non-HAP- using process. The EPA solicits comment on whether pollution prevention credit should be granted for cases in which a source reduces its emissions by switching from production of one chemical to another as well as the cases in which a source reduces its emissions through pollution prevention measures but continues to produce the same chemical. Pollution prevention measures will be allowed to generate credits equal to the difference between the emissions allowed by the rule for the previous process after control and the emissions from the modified production process. This credit would be allowed each month the modified production process is in place. The amount of credit would be adjusted according to the actual production volume of the relevant production process for that month. If a pollution prevention measure is used as a means of compliance with the rule for Group 1 points, then the activity would not generate credits unless the control efficiency of the pollution prevention measure is greater than the efficiency of the reference control technology. For example, if a process change results in a reduction in the mass rate of HAP emissions from a Group 1 process vent, such a change would only generate credits if the emission reduction were greater than 98 percent. However, if the process change resulted in a 98 percent reduction, equal to the reference control technology, the pollution prevention measure could be used to meet the standard but would not generate a credit or debit. See section VII.F.4, ''Credits for Previous Actions'' of this notice for an example regarding credits for pollution prevention. The EPA believes that the above method of valuing credits will offer industry flexibility to cost effectively comply with the HON through emissions averaging and achieve a level of emission reduction that is the same or better than a rule that does not allow emissions averaging. 4. Credits for Previous Actions To utilize emissions averaging, an owner or operator must first determine the baseline level of control on those emission points that will be included in the average. Control equipment is considered part of the source's baseline level of control if it was in place before the passage of the 1990 amendments to the Act (November 15, 1990). Generally, this equipment can be used to meet the control level at an individual emission point, but not to generate emission credits for averages. However, sources can get emission credits for a control action taken before the 1990 amendments to the Act (November 15, 1990) if the action achieves more emission reduction than the standard requires for the relevant point and: (1) It is a pollution prevention measure, taken after 1987, qualifying under the Agency's Pollution Prevention Strategy (56 FR 7849); (2) it is being used to satisfy a 33/50 commitment as described in EPA Publication Number EPA-741-K- 92- 001; or (3) it is an Early Reductions commitment, other than an equipment shutdown, approved under the proposed 40 CFR 63.70 through 63.81 (56 FR 27338). Controls applied as part of an Early Reductions commitment can begin to generate credits only after the relevant point becomes subject to the HON, that is after the expiration of the 6-year extension for the Early Reductions source. The proposed rule does not allow most actions taken before passage of the 1990 amendments to be used to generate emission credits because such reductions would have occurred anyway, for reasons unrelated to the 1990 amendments or the proposed rule. If EPA allowed these actions to generate emission credits, then the source would be able to generate more emission debits and, thus, more total emissions. Emissions averaging is a method for complying with subpart G and should not result in more emissions than the other compliance options. The proposed rule allows credit for controls put in place since the passage of the 1990 amendments for two reasons. First, since the 1990 amendments require the promulgation of emission standards, many sources have begun putting in place controls in anticipation of upcoming regulations. If {pg 62651} these controls could not be credited for averaging, these sources would be at a disadvantage relative to other sources that chose to postpone emission reductions until required by rule. Thus, allowing credit for controls put in place since, and presumably because of, passage of the amendments creates a more equitable emissions averaging system. The second reason for the policy on crediting existing controls has to do with the precedent set by this proposed rule. This proposed rule describes the first application of emission averaging for compliance with standards developed under section 112(d) of the Act. Many industry groups have interpreted, and will continue to interpret this proposed rule as an indication of the types of requirements that will be incorporated into future standards. By proposing the passage of the 1990 amendments as the date for determining the source's baseline level of control, the Administrator is setting a precedent that will encourage sources in other source categories to initiate emission reductions before their standards are developed. Selection of any date specifically associated with the proposed rule, such as the date of proposal or promulgation, would not be interpreted as a clear indication of the Administrator's intentions for future standards. The EPA is proposing to make limited exceptions to the general policy of not allowing credit for reductions achieved before passage of the amendments in three cases, the Early Reductions program, the 33/50 program, and pollution prevention. These exceptions are proposed to set a precedent that encourages future participation in these voluntary emissions reductions programs by sources in other source categories and to reward innovative pollution prevention efforts. The EPA believes these actions are beneficial to the environment and wishes to encourage sources to undertake them. The following two examples illustrate the policy regarding credit for previous actions. In the first case, a Group 2 process vent was controlled with an incinerator before November 15, 1990. The HON would not require that a Group 2 process vent be controlled, thus in this situation, the source is achieving more emission reduction than required on the Group 2 vent. However, the incinerator cannot be used to generate credits because it is not a pollution prevention measure or part of either an Early Reductions or a 33/50 commitment. In the second example, the source used a pollution prevention measure on a wastewater stream in 1988. This stream contains specific pollutants for which the steamstripping reference technology can achieve 99 percent removal. Further, of the 99 percent stripped from the wastewater into a vapor stream, the vapor stream reference technology can achieve 95 percent control. Through the pollution prevention process change, the source reduced the annual amount of wastewater it generates from 50 million liters (13,209,000 gal) to 25 million liters (6,604,500 gal). This process change also reduced potential emissions from the wastewater stream by 50 percent, from 70 to 35 Mg/yr (77 to 38.5 tons/yr). In this case, if the source had not undertaken a pollution prevention measure, then the allowed emissions after 99 percent wastewater stream control followed by 95 percent vapor stream control would be 4.2 Mg/yr (4.6 tons/yr). After taking the pollution prevention measure, emissions were 35 Mg/yr (38.5 tons/yr), which is greater than the 4.2 Mg/yr (4.6 tons/yr) allowed to the point by the rule. Without further controls, the source would not get credit for the pollution prevention measure because it did not reduce emissions below what would have occurred with application of the reference control. However, if the source uses 99 percent wastewater stream control followed by 95 percent vapor stream control in addition to the pollution prevention measure, the residual emissions would be 2.1 Mg/yr (2.3 tons/yr). With the pollution prevention measure and the stream control, the source would receive an annual credit equal to the difference between what the rule would allow if a pollution prevention action had not been undertaken, 4.2 Mg/yr (4.6 tons/yr), and what was actually emitted after pollution prevention and control, 2.1 Mg/yr (2.3 tons/yr). Thus, the credit would be for 2.1 Mg/yr (2.3 tons/yr). 5. Credit Discount Factors Some have argued that if industry receives a cost savings as a result of emissions averaging, the environment should also share that benefit by experiencing greater emissions reductions. Credit discounting is one way to provide such a benefit to the environment. The EPA is seeking comment on whether it is appropriate to require the use of a credit discount factor in calculating emissions averages. A discount factor would reduce the value of credits in the emissions average by a certain percentage before the credits are compared to the debits. In considering a discount factor, the EPA examined the requirements for determining MACT in section 112(d) of the Act. Section 112(d)(2) specifies that MACT standards shall require the maximum degree of reduction in emissions of HAP's, taking into consideration, among other things, the cost of achieving those reductions. By defining the source broadly and including the option for emissions averaging in the proposed rule, it could be argued that the EPA is providing flexibility for source owners and operators that would lower the costs of compliance. Some have suggested that, to carry out the mandate of section 112(d)(2), such cost savings should be shared with the environment by requiring sources using averaging to achieve more emission reductions than they would otherwise. Another view is that discount factors place a tax on market-based incentives that is not placed on traditional command and control compliance methods. This tax discourages emissions averaging thereby increasing the costs of complying with the rule yet not necessarily increasing emission reductions. The increased cost imposed by the inclusion of any discount factor may reduce or completely eliminate the incentive to average thereby increasing the overall cost to society of complying with the rule compared to what it would have been with emissions averaging without decreasing emissions, generating no additional health protection for the public. There may also be a technical reason for using a discount factor in averaging. Emission estimates that would be used for averaging are inherently uncertain. It should be noted that discounting is not required in most standards that do not allow compliance through some form of trading. Some of the same technical uncertainties also exist for those standards. However, for these standards, the significance of technical uncertainties can be considered more limited because, without emissions averaging, emissions estimates are not the basis for trading. With the technical uncertainties of emissions estimation in mind, the EPA is seeking comment on whether the emissions estimation procedures included in the proposed rule for any kind of emission point can be expected to consistently produce biased results, either over or under estimates of emissions. Given the above considerations, the EPA is proposing a discount factor to be selected from a range of values from 0 to 20 percent. This range reflects a reasonable span of values that have been included in previous rules involving emissions trading. It is emphasized that a decision on whether to include a {pg 62652} discount factor, and if so what the specific value will be, will be settled for promulgation in the final rule. Comments are requested on what that value should be. One option under consideration is for the value of the discounting factor to vary with the number and kinds of emission points included in the average. The degree of uncertainty associated with estimating emissions varies among different kinds of emission points. For example, there is more certainty regarding the accuracy of the method used for estimating emissions from vents than there is for the method used to estimate emissions from storage vessels. Therefore, the value of the discounting factor could be higher if certain kinds of points, such as storage vessels, were included in the average. A second option would be to increase the value of the discount factor with increases in the number of points included in the average. The intent of such a provision would be to account for the increased uncertainty associated with including more points, and the corresponding emissions estimates, into the average. A third option would be to have a different discount factor for different points based on the height at which their emissions are released. This system would attempt to account for the dispersion characteristics and actual exposure impacts of different emission points. However, EPA is concerned that the adoption of a variable discounting factor, as in any of these three options, would greatly increase the administrative complexity of emissions averaging, reducing its workability. Another option being considered is to include a two-tiered discounting factor so that pollution prevention measures would be assigned a lower factor than other credit- generating activities. This would serve as an incentive to generate credits through activities that prevent pollution. Congress has indicated that, where possible, EPA should encourage emissions reduction through pollution prevention, which often results in reduced emissions from all emission points in the source, both fugitive and point. The EPA specifically requests comments on the use of a discounting factor for emissions averaging in the HON. Commenters should address what value in the proposed span (0 to 20 percent) should be selected, whether the value should vary (and to what degree) according to the kinds and number of emission points included in the average, and whether a lower value should be assigned to pollution prevention measures. 6. Compliance Period The proposed rule requires that the credits and debits generated in emissions averages balance on an annual basis. In addition, the proposed rule requires that debits do not exceed credits by more than 25-35 percent in any one quarter of the year. These two requirements are used together to establish an emissions averaging system that provides flexibility for changes in production over time without allowing for wide-ranging fluctuations in HAP emissions over time. The proposed rule also provides sources the opportunity to ''bank'' extra credits generated in one compliance period for use in a later compliance period. a. Annual and quarterly compliance requirements. The EPA is proposing an annual compliance period for emission points included in averages by requiring that credits and debits balance on an annual basis. This compliance period was selected to provide sources considerable latitude in selecting points for inclusion in emissions averages. With an annual compliance period, sources can average emission points that may not have the same emission rates during some periods of the year, as long as they are similar on an annual basis. This latitude will also be useful to accommodate averages with points that must undergo temporary maintenance shut- downs at different times over the year. Several other factors were evaluated in making the decision to propose an annual compliance period for emission points in averages. To determine the appropriate compliance period for averaging, EPA examined the ability of control and monitoring equipment to measure emissions or other parameters from the kinds of emission points subject to the HON. Because of short term fluctuations in emissions from some of the points, such as transfer racks, EPA concluded that 30 days was the shortest compliance period that could reasonably be applied to all the kinds of points that can be included in averages. Though the administrator did not choose to require a 30-day compliance period, the proposed rule does require that sources maintain records of their emissions averaging credits and debits on a monthly basis. In selecting a compliance period for averaging, EPA also considered the need to verify compliance and, when appropriate, take enforcement action in a timely fashion. One concern about an annual compliance period is that the EPA's authority to take administrative enforcement actions would be significantly reduced because section 113(d) of the Act limits assessment of administrative penalties to violations which occur no more than 12 months prior to the initiation of the administrative proceeding. Administrative proceedings are far less costly than judicial proceedings for both EPA and the regulated community. The requirement that debits not exceed credits by more than 25-35 percent in any quarter enables EPA to use this administrative enforcement authority by providing a shorter period in which to verify compliance. A fourth factor considered in the selection of a compliance period for averaging was the effect of averaging on the distribution of a source's emissions over time. Averaging is intended to allow sources flexibility in how they create emissions reduction without resulting in a significantly different emissions scenario than would have occurred under point-by-point compliance with the proposed rule. The requirement that debits not exceed credits by more than 25-35 percent in any quarter limits the potential for wide variations in emissions over time, thus ensuring that an annual compliance period will not result in a significantly different emissions scenario than a shorter compliance period. As described above, the requirement that debits do not exceed credits by more than 25-35 percent in any quarter is an important element in the rationale for the annual compliance period. The range for the variability factor included in this requirement, 25-35 percent, was selected based on EPA's assessment of likely differences in emissions across quarters for emission points with similar annual emission values. The EPA is seeking comment on what other factors should be considered in setting this number and data regarding the variability in emissions from individual points over a year. This comment, and the included data, will be used to select a single value for the quarterly variability factor in the final rulemaking. While the proposed rule establishes a ratio of quarterly credits to debits to provide both flexibility and an enforceable short-term check on emissions, the same goals could be accomplished with a different requirement. Industry sources have suggested that EPA structure the quarterly check on emissions as an absolute emissions limit. The limit would be set by the sum of the residual emissions that would be emitted from the points in the average after application of the reference control to each Group 1 point and with existing controls being applied at each Group 2 {pg 62653} point. The owner or operator of the source would set this limit in the operating permit or Implementation Plan based on anticipated operations for upcoming quarters. The EPA is seeking comment on this alternative to the proposed requirement that debits do not exceed credits by more than 25-35 percent in any quarter. Commenters are encouraged to describe why the industry suggestion may be preferable to the proposed requirement and to provide any data that EPA might find useful in comparing the two approaches. b. Banking provisions. One way a source can meet the annual compliance requirements for the proposed rule involves the use of ''banked'' emission credits. The proposed rule allows sources to bank their extra credits if they generate more credits than are necessary to offset the debits from a given compliance period. These banked credits are then available for use in future compliance periods when the source has generated more debits than credits. Section 63.150(e) of subpart G details how banked credits can be generated and used. Banking is allowed by the proposed rule to provide flexibility when the number of credits or debits generated over the year differ from what the source anticipated in its averaging plan. Specifically, the banking provisions were developed to provide flexibility when: (1) There is an unanticipated mismatch in production or utilization rates for the credit and debit generators; (2) The maintenance shutdowns for averaged points are in different compliance periods; and (3) One or more of the credit generators must be shut down unexpectedly. Allowing sources to bank extra credits also encourages earlier emission reductions. Knowing that extra emission credits can be banked for possible use in the future, sources may choose to reduce emissions more than required to generate extra credits at an earlier time. The final reason for allowing banking in the proposed rule is that banking can be expected to reduce compliance costs in certain circumstances. Without banked credits, sources can be expected to establish averages that should generate more credits than needed to offset debits. Sources would build this cushion of credits into their averaging plans to avoid noncompliance if an unexpected change in their operations results in fewer credits or more debits than anticipated. While this cushion of credits has the benefit of creating increased emissions reductions, it can have a cost to the source if extra control is necessary to create the cushion. Having banked credits reduces the need to build a cushion of credits into the average, thus reducing the costs of compliance in those circumstances where extra control would have been needed to create a cushion of credits. Banking can also be expected to reduce compliance costs in those situations where creating fewer debits or more current credits would be especially expensive. The EPA considers policies to reduce the cost of compliance in an effort to fulfill the Act's statutory requirement to consider cost in setting MACT standards. The inclusion of a banking provision is another mechanism by which the EPA is implementing the requirements of section 112(d)(2). With the above listed reasons for allowing banking in mind, EPA is seeking comment on whether or not banking provisions should be included in the HON. Some have argued that, despite the benefits banking might provide, it is inherently inconsistent with a NESHAP. As technology based standards, NESHAP does not establish an absolute cap on emissions, but instead allow emissions to grow and change with changes in processes or production. Including banking as a means of compliance with a NESHAP is, in effect, an extension of the compliance period for the source with banked credits. Thus, the primary effect of allowing banking in the proposed rule is to shift emissions over time. However, others would argue that the effect of banking is to achieve emissions reductions earlier than they would otherwise have been achieved and at less cost to society. Thus, to include banking allows a more efficient way to achieve emissions reductions. Based upon this line of argument, the Administrator has concluded that banking is appropriate for the HON in the context of emissions averaging. However, he welcomes comment on if and how banking should be included in the HON. The only way banked credits can be used is to meet the requirement that credits and debits balance on an annual basis. Banked credits cannot be used to offset debits that would exceed credits by more than 25-35 percent in any one quarter of the year. This policy has been established because, in the Administrator's judgment, the quarterly variability factor provides sufficient operational flexibility on a quarterly basis. In addition, the purpose of the quarterly requirement is to ensure that the levels of over-control and under-control used to comply with emissions averages are roughly equivalent over each quarter of the year. As a result, the quarterly requirement must be based on actual credits and debits from the same time period to be meaningful. The proposed rule includes a range, from 2 to 5 years, for the length of time that banked credits are available for use. This range is being proposed as a means of soliciting comment on the length of time that banked credits should be available for use. The shorter end of the range is being proposed because EPA has some concerns about the difficulty of taking enforcement actions involving banked credits. Enforcement actions involving banked credits would be based on data from multiple years and multiple emission points. Thus, emissions averaging, especially with the possibility of banking, increases the complexity of the evidence in controversy compared to an enforcement action for point-by-point compliance. In addition, because enforcement actions involving banked credits may cover several years' worth of data, the action may be barred by the general 5-year statute of limitations. If banking is permitted for 5 years, the EPA might be in the position of asserting that there are discrepancies in data, false reporting or other violations up to 10 years after the banked credit was originally generated. The longer end of the range being proposed for the availability of banked credits is included because sources might create more extra emissions reductions earlier if banked credits are available for use over a longer period. In addition, limiting the length of time that banked credits are available for use can be expected to create an incentive for sources to use the banked credits earlier than they otherwise might have. Thus, it could be argued that a longer time for the use of banked credits could result in later emissions and a longer period of lowered emissions. The EPA is seeking comment on both the enforcement concerns and incentive value associated with how long banked credits are available for use. In order to preserve the ability of the government to seek penalties for the full 5-year period authorized by the general statute of limitations, 28 U.S.C. Section 2462, the EPA is proposing to require that the underlying documentation which supports the existence of a banked credit be maintained by the source for 5 years after the credit is used. For example, if the credit is generated in Year One and used by the source in Year Two, then the records showing the validity of that credit would have to be kept until the end of Year Seven. This would enable the {pg 62654} Agency to seek those records in an enforcement action commenced in Year Seven, the last year in which it could seek penalties for a violation of the HON in Year Two. It should be noted that if an enforcement action were actually begun in, for example, Year Three, the source would need to keep the records until such time as the enforcement action was concluded. In other words, the 5- year rule proposed here is what authorizes the source to destroy the records in the absence of an enforcement action. The EPA is interested in comments on whether there is a basis for requiring records retention for a period of less than 5 years. 7. Emissions Averaging Enforcement The Implementation Plan or operating permit for each source must reflect which points will be included in an emissions average, and how each of those points will be controlled. The controls will be cited in these documents so that inspectors have a relatively simple way to verify compliance by emissions averaging. To verify compliance, inspectors will ensure that the proper controls are installed in the proper places and will examine the source's records to ensure that each emissions average will balance. Thus, the application and proper operation and maintenance of controls is separately enforceable from the credit/debit balance for emission points included in averages. The EPA has specified the monitoring and recordkeeping requirements necessary to ensure that the credits and debits actually balance in each emissions average and that these balances are enforceable. These balances are considered one enforceable commitment made in either the Implementation Plan or the operating permit. The EPA is requesting comment on the process for ensuring that emissions averages meet the requirements of this rule in those cases where a source must comply prior to the approval of their Title V operating permit. The proposed rule attempts to ensure this compliance by requiring that sources obtain approval of those Implementation Plans that include emissions averaging. Thus, approval of Implementation Plans for emissions averaging is required even if no operating permit application has been filed. 8. The Potential Influence of Averaging on the Toxicity of Emissions The current proposal for emission averaging compares ''debits'' with ''credits'' without consideration of toxicity. Concerns have been raised that such a scheme would allow emissions of a ''more hazardous'' pollutant to be increased (debit) for corresponding decreases in a ''less hazardous'' pollutant (credit). Although the influence of such increases and decreases on the risk to public health posed by the HON sources is unclear, the EPA is currently investigating two approaches which use toxicity data in emission averaging, and requests comments on both. The first approach is to allow increased emissions of a HAP (debit) to be compensated for by decreases in an equal or greater amount of ''a more hazardous pollutant'' (credit). Under this approach, EPA's task is to determine the relative hazard of one pollutant to another. For any one pollutant whose emissions are increasing, pollutants which are determined to be equally or more hazardous may be used in the current emission averaging methodology described in the HON. Pollutants which would be considered to be less hazardous would simply not be allowed to be used in the current emission averaging formula. This approach would be the easiest to implement. An alternative approach provides for more flexibility or greater emission averaging opportunities but increases the complexity involved in integrating hazard into the current emissions averaging methodology. Under this approach, a more hazardous quantity (credit) may be used for emission averaging purposes regardless of whether the pollutant whose emissions are to be decreased is itself ''more hazardous'' than the pollutant with increased emissions. Using this approach, not only must the relative hazard of the pollutants be established but also the magnitude of the difference in hazard between pollutants. As the data and science concerning determination of the magnitude in difference between pollutants is limited, this approach is based on a number of policy judgments. The EPA requests comment on the general issue of whether and how averaging may influence the toxicity and risk of emissions from HON sources. In addition, the EPA specifically requests comment on the two previously described alternative approaches to limit the potential for averaging to increase the toxicity of HON source emissions. G. Selection of Reporting and Recordkeeping Requirements The proposed rule would require sources to submit the following five types of reports: 1. Initial Notification, 2. Implementation Plan (if an operating permit application has not been submitted), 3. Notification of Compliance, Status, 4. Periodic Reports, and 5. Other reports. The purpose and contents of each of these reports are described in this section. The wording of the proposed rule requires all draft reports to be submitted to the ''Administrator''. The term Administrator means either the Administrator of the EPA, an EPA regional office, a State agency, or other authority that has been delegated the authority to implement this rule. In most cases, reports will be sent to State agencies. Addresses will be provided in the General Provisions (subpart A) of 40 CFR part 63 that will be proposed in the future. Records of reported information and other information necessary to document compliance with the regulation are generally required to be kept for 5 years. A few records pertaining to equipment design would be kept for the life of the equipment. 1. Initial Notification The proposed rule would require owners or operators who are subject to subpart G to submit an Initial Notification. This report will establish an early dialog between the source and the regulatory agency, allowing both to plan for compliance activities. The notice is due 120 days after the date of promulgation for existing sources. For new sources, it is due 180 days before commencement of construction or reconstruction, or 45 days after promulgation of subpart G, whichever is later. The notification must list the chemical manufacturing processes at the source that are subject to subpart G, and which provisions may apply (e.g., process vents, transfer operations, storage vessel, and/or wastewater provisions). A detailed identification of emission points is not required. The Initial Notification must include a statement of whether the source can achieve compliance by the specified compliance date. The regulated industry anticipates that, due to the large number of sources and emission points required to comply with the HON, there may be delays in permitting processes, and there may be insufficient engineering services and {pg 62655} control equipment to achieve compliance in the 3-year time period allowed for existing sources. If a particular source anticipates a delay that is beyond its control, it will be important for the owner or operator to discuss the problem with the regulatory authority as early as possible. Pursuant to section 112(d) of the Act, the proposed rule has provisions for 1-year compliance extensions to be granted on a case- by-case basis. Further discussion of compliance issues is included in section VII.H. of this notice. 2. Implementation Plan The Implementation Plan details how the source plans to comply with subpart G. Implementation Plans are only required for sources that have not submitted an operating permit application. An operating permit application would contain all the types of information required in the Implementation Plan, so it would be redundant to require sources to submit both. Existing sources must submit the Implementation Plan for points in averages 18 months prior to the compliance date. For emission points not included in an emissions average, the Implementation Plan is due 12 months prior to the compliance date. For new sources, Implementation Plans would be submitted with the Notification of Compliance Status. It is critical that regulatory authorities have the Implementation Plans well before the compliance date so they can plan their implementation and enforcement activities. The early submission of these plans may also benefit regulated sources by allowing them to receive any feedback on their control plans prior to the actual compliance dates. The Implementation Plan for points included in an emissions average is required 18 months prior to the compliance date to allow time for review and approval of the average. Because of the complexities and site-specific nature of emissions averaging, an approval process is necessary to assure all parties that the specific plan will result in emissions credits outweighing debits. The Implementation Plans for points in averages must be more detailed and thorough than the plans for other emission points. The additional information is necessary for the reviewing authority to make an informed decision about approving the average. The projected credits and debits included in the Implementation Plan may be based on calculations, design analyses, or engineering assessments instead of measured values. This flexibility is provided because, in many cases, control measures will not have been implemented at the time the plan is due, and actual measurements would not be possible. 3. Notification of Compliance Status The Notification of Compliance Status would be submitted 150 days after the source's compliance date. It contains the information necessary to demonstrate that compliance has been achieved, such as the results of performance tests, TRE determinations, and design analyses. Further information on the requirements for performance tests and other methods of compliance determination are provided in section VII.B, C, D, and E of this notice for process vents, storage vessels, transfer operations, and wastewater, respectively. Sources with a large number of emission points are likely to be submitting results of multiple performance tests for each kind of emission point. For each test method used for a particular kind of emission point (e.g., a process vent), one complete test report would be submitted. For additional tests performed for the same kind of emission point using the same method, the results would be submitted, but a complete test report is not required. Results would include values needed to determine compliance (e.g., inlet and outlet concentrations, flow rates, percent reduction) as well as the values of monitored parameters averaged over the period of the test. The submission of one test report will allow the regulatory authority to verify that the source has followed the correct sampling and analytical procedures and has done calculations correctly. Complete test reports for other emission points may be kept at the plant rather than submitted. This reporting system was established to ensure that reviewing authorities have sufficient information to evaluate the monitoring and testing used to demonstrate compliance with the HON while minimizing the reporting burden. Another type of information to be included in the Notification of Compliance Status is the specific range for each monitored parameter for each emission point, and the rationale for why this range indicates proper operation of the control device. (If this range has already been established in the operating permit, it does not need to be repeated in the Notification of Compliance Status). As an example, for a process vent controlled by an incinerator, the notification would include the site-specific minimum firebox temperature that will ensure proper operation of the incinerator, and the data and rationale to support this minimum temperature. Table 6a presents illustrative examples of the kinds of limits, or ranges that might be set for monitored parameters. These ranges are provided only as a guide to possible values, and actual values should be determined based on the design and operating characteristics of the control device as well as process- specific considerations. For a full discussion of this approach and EPA's rationale, see section VII.H.2 of this notice. Table 6 a.-Example Range Limits for Continuously Monitored Parameters Control device Thermal incinerator Parameters to be monitored Firebox temperature sup a Example parameter ranges Average firebox temperature must not be more than 28 degrees C (50 degrees F) below the average value measured during the most recent performance test. Control device Catalytic incinerator Parameters to be monitored Temperature upstream and downstream of the catalyst bed Example parameter ranges Average upstream temperature must not be more than 28 degrees C (50 degrees F) below the average value measured during the most recent performance test. Parameters to be monitored Example parameter ranges Average temperature difference across the catalyst bed must be greater than 80 percent of the average temperature difference measured during the most recent performance test. Control device Boiler or process heater with a design heat input capacity less than 44 megawatts Parameters to be monitored Firebox temperature sup a Example parameter ranges Average firebox temperature must not be more than 28 degrees C (50 degrees F) below the average value measured during the most recent performance test. Control device Scrubber for halogenated vent streams (Note: Controlled by a combustion device other than a flare) Parameters to be monitored pH of scrubber effluent Example parameter ranges Average pH of the scrubber effluent must not be more than 1 pH unit below the average value measured during the most recent performance test. Parameters to be monitored Scrubber liquid and gas flow rates Example parameter ranges Average scrubber liquid/gas ratio must be greater than 95 percent of the average value measured during the most recent performance test. Control device Absorber Parameters to be monitored Exit temperature of the absorbing liquid Example parameter ranges Average exit temperature of the absorbing liquid must not be more than 11 degrees C (20 degrees F) above the average value measured during the most recent performance test. Parameters to be monitored Exit specific gravity Example parameter ranges Average exit specific gravity must be within 0.1 unit above or below the average value measured during the most recent performance test. Control device Condenser Parameters to be monitored Exit (product side) temperature Example parameter ranges Average exit temperature must not be more than 6 degrees C (11 degrees F) above the average value measured during the most recent performance test. Control device Carbon adsorber Parameters to be monitored Total regeneration stream mass flow during carbon bed regeneration cycle(s) Example parameter ranges Total regeneration stream mass flow for a regeneration cycle must not be more than 10 percent below the value measured during the most recent performance test. Parameters to be monitored Temperature of the carbon bed after regeneration and within 15 minutes of completing any cooling cycle(s) Example parameter ranges Temperature of the carbon bed after regeneration must not be more than 10 percent or 5 degrees C more than the value measured during the most recent performance test. Control device All Control Devices (as an alternative to the above) Parameters to be monitored Concentration level or reading indicated by an organic monitoring device at the outlet of the control device Example parameter ranges Average concentration level or reading must not be more than 20 percent more than the average value measured during the most recent performance test. sup a Monitor may be installed in the firebox or in the ductwork immediately downstream of the firebox before any substantial heat exchange is encountered. For emission points included in an emissions average, the notification would also include the measured or calculated values of all parameters needed to calculate emission credits and debits, and the result of the calculation for the first quarter. This information is needed to ensure that the points in the average are being controlled as described in the Implementation Plan and that the average itself is balancing as planned. 4. Periodic Reports Periodic Reports are required to ensure that the standards continue to be met and that control devices are operated and maintained properly. Generally, Periodic Reports would be submitted semiannually. However, quarterly reports must be submitted for the emission points included in an emissions average. This reporting frequency is necessary to allow verification of the credit and debit balance on a quarterly basis. In addition, if monitoring results show that the parameter values for a particular emission point are outside the established range for more than 1 percent of the operating time in a reporting period, or the monitor is out of service for more than 5 percent of the time, the Administrator (or delegated regulatory authority) may request that the owner or operator submit quarterly reports for that emission point. After 1 year, the source can return to semiannual reporting, unless the regulatory authority requests continuation of quarterly reports. The EPA has established this reporting system in order to provide an incentive (less frequent reporting) for good performance. Because of uncertainty about the periods of time over which sources are likely to experience excursions outside the parameter ranges or monitoring system failures, the EPA is seeking comment on the 1 percent and 5 percent criteria triggering more frequent reporting. In particular, data are requested on both the frequency of excursions and monitoring system downtime. Periodic Reports specify periods when the values of monitored parameters are outside the ranges established in the Notification of Compliance Status or operating permit. For continuously monitored parameters, records must be kept of the parameter value recorded once every 15 minutes. If a parameter is monitored more frequently than once every 15 minutes, the 15-minute averages may be kept instead of the individual values. This requirement ensures that there will be enough monitoring values recorded to be representative of the monitoring period without requiring the source to retain additional data on file and readily accessible. For some types of emission points and controls, periodic (e.g., monthly, quarterly, or annual) inspections or measurements are required instead of continuous monitoring. Records that such inspections or measurements were done must be kept; but results are included in Periodic Reports only if a problem is found. This requirement is designed to minimize the recordkeeping and reporting burden of the proposed rule. For emission points included in an emissions average, the results of the quarterly credit and debit calculation are also included in the Periodic Reports, so the reviewing authority can ensure that the quarterly requirements for the average have been met. The role of Periodic Reports for compliance purposes is described in section VII.H. of this notice. 5. Other Reports There are a very limited number of other reports. Where possible, subpart G is structured to allow information to be reported in the semiannual (or quarterly) Periodic Reports. However, in a few cases, it is necessary for the source to provide information to the regulatory authority shortly before or after a specific event. For example, if a process change is made that causes a process vent to change from Group 2 to Group 1, the source must report the change within 90 days. For storage vessels, notification prior to internal tank inspections is required to allow the regulatory authority to have an observer present. For storage and wastewater, if an owner or operator requests an extension of the repair period or a delay of repair, the request needs to be submitted separately from the Periodic Reports because the requests require a quick response from the reviewing authority. Certain notifications and reports required by the Part 63 General Provisions must also be submitted. These are described in section IX.A of this notice, ''Coordination with Other Clean Air Act Requirements.'' 6. Possible Alternative Recordkeeping Requirements The proposed rule requires sources to keep readily accessible records of monitored parameters. For those control devices that must be monitored continuously, records which include at least one monitored value for every 15 minutes of operation are considered sufficient. These monitoring records must be maintained for 5 years. However, there are some existing monitoring systems that might not satisfy these requirements. To comply with the HON, sources would have to replace these existing monitoring systems. As a result, the EPA is seeking {pg 62657} data and comment on these existing monitoring systems. Specifically, industry sources have informed EPA that some existing computer- controlled processes have monitoring systems that only store data that is outside some predetermined range of acceptable values. For example, these systems could be set to record and store all monitored values outside a range such as sup 8 1 percent. If a monitored value did not exceed the specified range, no value would be stored. When the value exceeded the range, a value would be stored. It is then deemed that all data in between the stored values is the same as the last recorded value. This system could also be used to record those periods when a monitored parameter may be outside the parameter ranges established by the source to represent proper operation of a control device. Keeping only these records would dramatically reduce the data storage requirements. Industry sources have also informed EPA that many existing process control computer systems obtain monitoring data much more frequently than every 15 minutes, but are not designed to maintain a record of such data for 5 years. Such systems use this extensive monitoring data to calculate average parameter values for the compliance period for the emission source (e.g., 3 hour average). The individual data points could be kept in an accessible record for a period of several days so that the averaging procedure could be verified, and then could be ''written over'' to conserve computer time and memory storage space. The average for the 3-hour compliance period would be retained in an accessible record for 5 years. At this time, EPA does not have a sufficient understanding of these systems to ensure that they provide sufficient support data to accurately and reliably reflect the source's continued compliance. Therefore, EPA has not included them as a recordkeeping option in the proposed rule. Instead, EPA is seeking comment on whether and how these systems should be allowed for compliance with the recordkeeping requirements in the HON. Specifically, EPA is interested in: what criteria are used to determine the values that are stored by these existing monitoring systems; how the validity of the data is verified; the frequency of calibration for this type of system; how operators ensure the accuracy of the results from these existing systems; and what types of processes or controls are currently being monitored with these systems. In addition, EPA is seeking comment on how the requirements allowing the use of these systems to comply with the HON might be structured. Finally, EPA is seeking comment on the concept of determining compliance based on data that do not include values to represent the entire compliance period (i.e., absence of data indicating a violation would constitute evidence of compliance). H. Selection of Compliance Provisions 1. Compliance Schedule The compliance date for existing sources is 3 years after promulgation of the HON. The 3-year compliance time is required by section 112(i)(3) of the Act. The compliance date for new sources is the date of start-up or the date of promulgation, whichever occurs later, as specified by section 112(i)(1) of the Act. During development of the HON, EPA received comment from the regulated community regarding the process that would be used to comply with the rule and certain difficulties that were anticipated. Because the HON will regulate such a large segment of SOCMI operations, the regulated community anticipates that there may be insufficient engineering services and control equipment to achieve compliance in the time allowed. One way in which the proposed rule addresses this issue is to allow for a 1- year compliance extension on a case-by- case basis. Section 112(i)(3)(B) of the Act allows for site-specific 1-year extensions to be granted through the operating permit. However, because of the schedule for States to establish and implement operating permit programs, the 3-year compliance date for the HON may occur prior to the submission of permit applications or the approval of operating permits for some sources. With this potential timing difficulty in mind, the Administrator is proposing that HON sources be allowed to request extensions through a submittal other than the permit application in States where operating permit applications will not have been submitted prior to the due date for the HON Implementation Plan. The proposed rule allows the owner or operator of a HON source to request the 1- year extension from the Administrator with their initial notification or with a separate submittal at any point prior to the submission of an Implementation Plan. The EPA is seeking comment on the significance of the potential difficulties of complying with the HON in the allotted 3 or 4 years. In addition, EPA is seeking comment regarding how these difficulties can be addressed within the confines of the statutory requirements of sections 112(d) and 112(i) of the Act. Specifically, EPA is seeking comment on what types of non-regulatory activities, such as technical assistance, can be provided to assist sources attempting to come into compliance with the HON. 2. Parameter Monitoring and Compliance Certification The proposed Subpart G requires monitoring of control device operating parameters and reporting of periods when parameter values are outside site- specific ranges. Although in previous NSPS and NESHAP, the EPA has specified a pre- determined range of operating parameter values, such values could be considered inadequate given the increased importance of parameter monitoring in determining and certifying compliance due to the new requirements in Section 114 of the Act. For the proposed HON, EPA is requiring sources to establish site-specific ranges. Allowing site-specific ranges for monitored parameters accommodates site-specific variation in emission point characteristics and control device designs. Based on the information available at proposal, it appeared to be difficult to establish ranges or minimum or maximum values that would be applicable in all cases. The proposed system for establishing operating parameter ranges attempts to balance the need for technical certainty and operational feasibility. The ranges may be established by performance testing supplemented by engineering assessments and manufacturer's recommendations. However, the performance test is not required to be conducted over the entire range of permitted parameter values because such a requirement could impose significant technical difficulties and costs on the source. The EPA believes that a performance test conducted for a smaller, yet representative, range of operating conditions can still provide a range for the operating parameters that ensures proper operation of the control device. For emission points and control devices where a performance test is not required (for example, a closed vent system and control device on a storage vessel), the range may be established by engineering assessment. Under the NSPS and NESHAP programs, parameter monitoring has traditionally been used as a tool in determining whether control devices are being maintained and operated properly. However, section 114(a)(3) of the Act and Sec. 70.6(c) of the operating permit rule (57 FR 32251) require the submission of ''compliance {pg 62658} certifications'' from sources subject to the operating permit program. In light of these requirements, the EPA has considered how sources subject to this rule would demonstrate compliance. The EPA has concluded that operating parameter monitoring can be used for this purpose. The EPA considered three alternatives for using continuous parameter monitoring in determining compliance. The first alternative was that each excursion of a parameter outside the established range would constitute a violation of the permitted operating conditions for the control device. The first alternative was not selected because correlation of operating parameters with performance of the control device is not exact and operation outside the parameter range is not necessarily indicative of improper operation and maintenance. The second approach was to require corrective action to be taken within 24 hours of first recording the excursion. The excursion would only be considered a violation if the problem was not corrected within 24 hours. The second alternative also was not selected because of the associated recordkeeping burden and difficulties in verifying compliance. Another disadvantage of the second alternative is it did not provide an incentive for the source to avoid operation outside the parameter ranges. The third approach was to excuse a certain number of excursions per reporting period based on evidence that a certain number of excursions could be expected even with properly operated pollution control devices. For example, the rule could excuse three excursions and, if there were fewer than three excursions in a semiannual period, the source could certify continuous compliance; however, if a fourth excursion occurred, it would be a violation of the permitted operating conditions. The Administrator chose the third approach described above. The proposed rule requires the source to record daily average values for continuously monitored parameters. The daily average is the average of all of the 15-minute values generated by the continuous recorder during the operating day. If the daily average value is outside the established range, it must be reported. The EPA is proposing to allow from 3 to 6 excused excursions (3 to 6 operating days) per semiannual reporting period for each control device. The daily averaging period was selected because the purpose of monitoring data is to ensure proper operation and maintenance of the control device. Because it often takes from 12 to 24 hours to correct a problem, this averaging period was considered to best reflect operation and maintenance practices. This averaging period therefore gives the owner or operator a reasonable period of time to take action. If a shorter averaging period (for example 3 hours) was selected, sources would be likely to have multiple excursions caused by the same operational problem because it would not be possible to correct problems in one 3-hour reporting period. The proposed range of 3 to 6 days of excusable excursions per semiannual reporting period (or 1 to 3 days per quarterly reporting period) equates to roughly 1 to 3 percent of the days in the reporting period. The range of time allowed as excused excursions was selected based on information about the types of events that cause parameter excursions; the duration of typical excursions; and the frequency of the events that create excursions. Examples of events that could cause excursions that would count toward the number of excused excursions are: a thermocouple failure in an incinerator; water contamination in a condenser; off-specification feedstocks; electrical problems; control valve problems such as leaky pneumatic drivers; and extreme environmental conditions. Events that would be considered malfunctions under the Start-up, Shut- down and Malfunction Plan required by the General Provisions (subpart A) are to be handled separately and would not be counted toward the allowed number of excused excursions for purposes of compliance with subpart G of the HON. In addition, the provision for excusable excursions is not meant to allow actions that are specifically disallowed by other sections of the HON or the General Provisions, such as bypass of a control device. Comments on the proposed approach and the other alternative approaches that were considered and any other suggested approaches are requested. Regarding the proposed approach, comment is requested on the number of days or percent of operating time that should be allowed as excused excursions, and whether the number of excused days should decrease over time, after an initial break-in period. In particular, EPA requests that commenters submit data that might be used to better characterize the cost of such requirements, the relationship between operating parameter monitoring results and control device performance, and data that might indicate how many excursions are associated with proper operation and maintenance of various control devices. The EPA also requests comment on the availability of methods for continuous monitoring of operating parameters and whether such methods could be used for compliance determinations and certification. VIII. Rationale for Provisions in Subpart H A. Background Equipment leak emissions refer to the loss of VOC's and VHAP's through the sealing mechanism separating process fluid contained in equipment from the atmosphere. Because of the large number of valves, pumps, and other components within a process unit, total emissions from such equipment can be large. Equipment leaks have been estimated to contribute about one-third of all routine (non- accidental) VOC emissions from the chemical industry. Existing regulations adopted under sections 111 and 112 of the Act (i.e., 40 CFR 60, subparts VV, GGG, and KKK, 40 CFR part 61, subpart V) hereafter referred to as ''existing rules'') and in SIP's have been effective in heightening awareness of the significance of equipment leaks and in stimulating control efforts. These rules basically require that pumps and valves be inspected periodically for leaks with a portable hydrocarbon detector. If a VOC concentration greater than 10,000 ppm, as methane or hexane, is found, the component is identified as a ''leaker'' and maintenance is required to repair the leak. This approach is known as LDAR. When these rules were established, EPA estimated that emissions would be reduced by about 60 to 70 percent and that after control, leak frequencies would be approximately 5 percent. Data gathered over the past several years on equipment leaks at some chemical plants indicate that much lower leak frequencies can be achieved. These data, however, did not identify specific factors that led to lower leak frequencies, nor indicate how low leak frequencies could be obtained at all chemical plants. Consequently, EPA saw a need for a new regulatory approach that would achieve low leak frequencies at all chemical plants. It was recognized that establishing such a regulation for as broad and varied a source category as chemical production units would be difficult. The challenges included determining how to achieve low leak frequency at all plants with a simple set of rules, how to provide more flexibility in achieving low leak rates than that provided by LDAR alone, how to apply standards across the industry using data from only a part of the {pg 62659} industry, and the EPA's need to establish standards consistent with the MACT requirements of the Act. On April 25, 1989, EPA announced its intention to establish a committee to negotiate a new approach for control of volatile organic chemical equipment leaks (54 FR 17944), and conducted an initial informational meeting on May 15, 1989, to determine among interested parties whether negotiation would be desirable. The participants at the initial meeting responded favorably to the concept of negotiation, and on September 12, 1989, EPA established a negotiating committee (54 FR 37725). The Committee met over a period of 1 year, holding nine 2-day meetings and one 1-day meeting, to resolve the various issues related to developing a MACT standard for equipment leaks. The Committee members are listed in Table 7. Table 7.- List of Negotiators, Facilitator, and Observer Negotiators Robert L. Ajax Affiliations Environmental Protection Agency. Negotiators Alfred Bickum Affiliations International Institute of Synthetic Rubber Producers. Negotiators Bruce Bowers Affiliations Standard Chlorine. Negotiators Linda Curran Affiliations Amoco Oil. Negotiators David Doniger, Allen Hershkowitz Affiliations Natural Resources Defense Council. Negotiators David Dunn Affiliations Sterling Chemicals, Incorporated. Negotiators Larry Goodheart, Ellen Siegler Affiliations American Petroleum Institute. Negotiators Jack Kace Affiliations Pharmaceutical Manufacturers Association. Negotiators Thomas Kittleman Affiliations Chemical Manufacturers Association. Negotiators Robert Majewski Affiliations Northeast States for Coordinated Air Use Management. Negotiators Les Montgomery Affiliations Texas Air Control Board. Negotiators Harvel Rogers Affiliations Jefferson County (Kentucky) Air Pollution Control District. Negotiators Gustave Von Bodungen Affiliations Louisiana Department of Environmental Quality. Facilitator: Phillip J. Harter Affiliations Consultant to EPA. Observer: Nicolas Garcia Affiliations Office of Management and Budget. The Committee considered the many factors and uncertainties associated with regulating equipment leaks at a wide variety of chemical plants and developed an acceptably balanced approach, weighing the need to be flexible, the technical uncertainties, the requirement for MACT standards, and the data limitations. At the final negotiating session, the Committee members conceptually resolved all outstanding major issues and decided to reach final agreement through a two-step process. The Committee members first agreed in principle to the regulatory language to be proposed and then concurred on a draft preamble to the regulation describing in detail the scope, application, effect, and rationale. All Committee members have agreed to support the standard as long as EPA proposes and promulgates a regulation and its preamble with the same substance and effect as the regulation and preamble that are the subject of the final agreement. It is important to note that the parties to the negotiation concurred with the regulation when considered as a whole. Inevitably, in any negotiation, this means that some parties may have made concessions in one area in exchange for concessions from other parties in other areas. B. Scope and Applicability 1. Source Categories The negotiators originally were to develop standards for equipment leaks for 13 source categories that would be affected by the EPA's then expected HON. These source categories included both SOCMI and non-SOCMI source categories, and the standards under development would have applied to eight hazardous organic chemicals. Over the course of the negotiations, and in anticipation of the Clean Air Act Amendments of 1990, EPA saw the need to expand this scope to include not only the original HON source categories, but also all SOCMI processes that use as a reactant or produce one of the organic chemicals listed in the Clean Air Act list of 189 HAP's. The EPA identified 396 such processes. The Committee agreed to expand the scope of the negotiations to the larger group of SOCMI processes and to retain the non-SOCMI categories. In addition, EPA determined that petroleum refinery processes would not be covered by these standards, regardless of whether the unit supplies feedstocks that include chemicals listed in Sec. 63.183, and that MACT standards for petroleum processes would be established in a separate rulemaking. These standards also would not be applicable to other petroleum-related facilities, including those engaged in petroleum exploration, production, marketing, or transportation. Refinery processes not covered by this regulation include, but are not limited to, cracking, reforming, coking, and other processes that produce transportation fuels, heating oils, or lubricants. Table 8 presents a more detailed list of examples of refinery processes not included in the scope of this rule. Table 8.- Examples of Refinery Processes Excluded From the Negotiated Regulation Thermal processes -Gas-oil cracking -Thermal cracking -Visbreaking -Coking (fluid) -Coking (delayed) -Coking (flexi) -Other Catalytic cracking -Fluid -Other Catalytic reforming -Semiregenerative -conventional catalyst -bimetallic catalyst -Cyclic -conventional catalyst -bimetallic catalyst -Other -continuous catalytic reforming -conventional catalyst -bimetallic catalyst Catalytic hydrocracking -Distillate upgrading -Residual upgrading -Lube-oil manufacturing -Other Catalytic hydrorefining -Residual desulfurizing -Heavy gas-oil desulfurizing -Cat-cracker and cycle-stock feed pretreatment -Middle distillate -Other Fractionation -Pipe stills -Light ends -Gas recovery units On-site transfer and blending operations of gasoline and other fuels Lube oil and specialties processes Catalytic hydrotreating -Pretreating cat-reformer feeds -Naphtha desulfurizing -Naphtha olefin or aromatics saturation -Straight-run distillate -Other distillate -Lube-oil "polishing" -Other Alkylation -Sulfuric acid -Hydrofluoric acid Refinery polymerization processes Crude units (Atmospheric and vacuum) Refinery isomerization process 2. Relationship Between This Regulation and Future Regulations for Refinery Equipment Leaks The Committee agreed on the following language to describe the relationship between this regulation and future regulations for refinery equipment leaks: The standards incorporated in this agreement were established by the Committee based primarily on data from well-controlled ethylene oxide plants and data from a number of SOCMI plants. The Committee did not review data from petroleum refining processes because they were outside the scope of the negotiations, and the Committee did not consider whether the numerical standards in this agreement are achievable by refinery processes. The Committee recognizes that there are technical differences between SOCMI categories and petroleum processes that may affect the achievability of the numerical standards (including percent leakers). Potential differences include the availability and effectiveness of emission control technologies, plant shutdown practices, line sizes, process temperatures and process pressures, and cost. These differences could work in the direction of making the numerical standards adopted by the Committee more easily achievable, or less easily achievable, for refinery processes. The Committee agrees that the framework used for the regulation developed during the current negotiation should be adopted in the MACT regulation for petroleum refinery equipment leaks unless there are sound technical reasons why a different framework would be more effective. The American Petroleum Institute understands that the numerical standards included in this negotiated rule are more stringent than those that would have been selected if the rule had provided that exceedences of numerical standards, by themselves, constituted violations of the CAA. The American Petroleum Institute recognizes that this will also be true of the numerical standards that will be included in a MACT regulation for refinery equipment leaks utilizing the framework of the currently- negotiated regulation. If EPA develops MACT refinery equipment leak regulations, it will conduct a separate rulemaking. Technical differences between SOCMI and refinery processes that may affect the achievability of the standards will be considered by EPA in the refinery equipment leak rulemaking. 3. Equipment The negotiated rule would apply to those pieces of equipment which are regulated in the existing sections 111 and 112 equipment leaks rules, including all valves, pumps, compressors, pressure relief devices, open-ended valves or lines, connectors, closed vent systems and control devices, sampling connection systems, and product accumulator vessels. In the existing rules, connectors are referred to as ''flanges and other connectors'' or simply as flanges. To avoid potential misinterpretations of the negotiated rule and to eliminate redundancy within the phrase, the term connector is used in the negotiated standard to designate the same types of fittings which were previously termed flanges and other connectors; i.e., all flanged, screwed, or other joined fittings used to connect two pipelines or a pipeline and a piece of equipment. The negotiated rule would not apply to flanges between sections of a vessel (e.g., body flanges on a distillation column reactor or heat exchanger, etc.), head gaskets on vessels, or access hatches (e.g., manholes). These types of seals are not included in the definition of connector. The negotiated rule also contains provisions for agitators and instrumentation systems. The rationale for the inclusion of agitators and the separate treatment of components in instrumentation systems in the negotiated rule is discussed later under the basis for the negotiated rule. The negotiated rule would apply to both existing and new process units. The negotiated rule categorizes the regulated processes into five groups and uses a staggered implementation scheme, requiring some process units to comply in 1/2 year, while others would comply as late as 1 1/2 years after final promulgation of the rule. This staggered implementation was provided to alleviate the impact of applying the rules simultaneously to all sources. An affected ''process unit'' means equipment that uses a VHAP as a reactant or produces a VHAP or its derivatives as intermediate or final product(s), including all equipment associated with the unit process operation, storage and transfer of feed material to the unit process operation and final or intermediate product from the unit process operation, and operations treating process wastewater (e.g., strippers, decanters) from the unit process operation. The proposed standards would apply to chemical manufacturing processes operated to produce one or more of the chemicals listed in Sec. 63.184. Examples of SOCMI production processes that would be subject are a unit process operation that produces ethylbenzene from benzene, a unit that produces phenol and acetone from cumene, and a unit that produces butadiene by separation from an impure mixed C-4 stream received from another plant site. Examples of operations that would not be considered subject to this standard are waste solvent reclamation or a SOCMI process using any of the chemicals listed in Sec. 63.183 only as a solvent. The standards would also apply to equipment handling specific chemicals for the non-SOCMI source categories listed in Section 63.160(c). These source categories and chemicals are: styrene-butadiene rubber production (styrene, BD); polybutadiene production (BD); chlorine production (carbon tetrachloride); pesticide production (carbon tetrachloride, methylene chloride, and ethylene dichloride); chlorinated hydrocarbon use in production of chlorinated paraffins, Hypalon sup , OBPA/1,3- diisocyanate, polycarbonate, polysulfide rubber, and symmetrical tetrachloropyridine (carbon tetrachloride, methylene chloride, tetrachloroethylene, chloroform, and ethylene dichloride); pharmaceutical production (carbon tetrachloride, methylene chloride); and miscellaneous BD use (BD). The lines within a unit process operation containing process fluids are considered to be part of the process unit and thus, subject to regulatory requirements, while lines and equipment not containing process fluids are not subject to these requirements. Utilities, and other nonprocess lines, such as heating and cooling systems, are not considered to be part of a process unit. For example, any direct heating and cooling systems, which generally service many processes at a plant and do not combine their materials with those in the processes they service, are also not subject to these requirements. A plant site may consist of one or more process units. Process units covered by the negotiated rule are listed specifically in Secs. 63.160 (b) and (c) of the negotiated rule. C. Background Information on Equipment Leaks This section presents an overview of findings from previous equipment leak studies and a summary of information that led to the EPA's decision to develop a new regulatory approach for equipment leaks through negotiation. This synopsis is not intended to reflect the Committee discussions. Rather, it is intended to provide basic information for readers unfamiliar with the existing standards, the underlying studies, and recent trends. More detailed information on the basis for the existing requirements and underlying studies is available in the dockets and Federal Register notices for the existing rules. 1. Overview of Background Information One of the first published studies of equipment leak emissions was conducted in the 1950's in several petroleum refineries in the Los Angeles County Air Pollution Control District. The results of this study showed that a large quantity of hydrocarbons could be lost to the atmosphere from various sources such as valves, pump and compressor seals, flanges, and pressure relief devices. In the late 1970's and early 1980's, EPA conducted several studies to evaluate and quantify emissions from equipment leaks in petroleum refinery operations and in chemical process units. In these studies, EPA collected data and evaluated leak frequency, mass emissions, and {pg 62661} methods and effectiveness of leak prevention. These studies showed that equipment leaks were a significant source of emissions and that the majority of emissions at that time were associated with equipment leaks measured at concentrations greater than 10,000 ppm. In the EPA Refinery Assessment Study, data were gathered on equipment screening values (using a portable VOC instrument) and mass emissions. These data permitted the development of average emission factors and screening value/emission rate correlations. The study also provided other important results. Analyses to identify equipment or process variables (e.g., equipment manufacturer, age, or line size, process pressure, stream volatility) that affect leak frequency led to the separation of equipment component emissions by stream phase: gas/vapor, light liquid, and heavy liquid. These classifications have been used to design regulations based on leak potential. In 1980, EPA conducted a study of equipment leak emissions from 24 individual SOCMI process units. In this study, the 24-Unit Study, EPA investigated leak frequency (at a concentration of 10,000 ppm) in equipment at 24 individual chemical process units. A study of the effects of maintenance on emissions was performed concurrently at six of the units screened in the 24-Unit Study. The data from these studies were used to develop average emission factors and screening value/leak rate correlations; to evaluate leak frequency as a function of process parameters and equipment design; to evaluate the effect of instrument response factors on leak frequency; and to estimate the effect of leak occurrence and recurrence rates on mass emissions. These studies also demonstrated that a program consisting of inspection of equipment and maintenance of the leaking equipment was an effective means of reducing emissions. The control measures identified at that time were estimated to achieve overall a 60 to 70-percent reduction in emissions from equipment leaks. The data and conclusions from the studies on petroleum refinery units and chemical process units served as a basis for the EPA's equipment leak regulations and guidelines that were issued in the early 1980's. Information obtained in these studies also showed that leak frequencies varied widely among process units and source types and that factors that affect leak frequencies were not well understood. Specifically, studies such as the EPA's 24- Unit Study showed that the frequency of leaks greater than 10,000 ppm ranged from 0 to approximately 30 percent among 15 different chemical processes. Similar chemical processes also showed large differences in leak frequencies; for example, the leak frequencies for gas valves varied from 0 to 18 percent for three ethylene dichloride units. In subsequent studies, EPA evaluated leak frequency as a function of process parameters and equipment design and identified no single factor that determined leak frequency. Leak frequencies are believed to be a function of component design, specifications, construction, material, and age; quality and frequency of maintenance; operating and training practices used by the company; diligence; process fluids and operating conditions; and other unidentified factors. Current regulations adopted under sections 111 and 112 of the Act (e.g., 40 CFR part 60, subpart VV and 40 CFR part 61, subpart V) require: (1) An LDAR program for valves in gas/vapor and light liquid service and pumps in light liquid service; (2) equipment for compressors, sampling systems, and open-ended lines; and (3) no detectable emissions (500 ppm as determined by Method 21) for pressure relief devices in gas/vapor service during normal operation. These rules do not require sources to achieve particular performance levels or to install particular equipment designs (e.g., DMS on pumps). As such, residual emissions will vary among process units and the best controlled units will emit less than typically or poorer controlled units. 2. Recent Studies In the last 3 years, EPA has received information that shows significantly lower leak frequencies at 10,000 ppm for some chemical processes than was representative for the industry as a whole in the early 1980's. This information, not necessarily known by the industry in general, was one consideration in the EPA's decision to develop a new regulatory approach for equipment leaks through the regulatory negotiation process. The information considered by EPA is briefly summarized in the paragraphs below. In addition, where sufficient information is available, these data are included in Table 9 which shows leak frequency at several leak definitions by component type. In general, these data were collected following the procedures specified in Method 21, and the results represent a single monitoring survey conducted over a limited time period. Table 9.- Summary of Equipment Screening Data sup a Unit ID EO 1 Valves 250 0.6 % 500 0.4 Leakers 1,000 0.3 Pumps 500 % 1,000 Leakers 2,000 Leakers 5,000 Connectors sup b 250 0 % 500 0 Leakers 1,000 0 Unit ID EO 2 Valves 250 2.8 % 500 2.2 Leakers 1,000 1.1 Pumps 500 0 % 1,000 0 Leakers 2,000 0 Leakers 5,000 0 Connectors sup b 250 0.9 % 500 0.9 Leakers 1,000 0.9 Unit ID EO 3 Valves 250 1.4 % 500 1.1 Leakers 1,000 1.1 Pumps 500 23 % 1,000 18 Leakers 2,000 14 Leakers 5,000 9 Connectors sup b 250 3.2 % 500 3.2 Leakers 1,000 3.2 Unit ID EO 4 Valves 250 2 % 500 1.3 Leakers 1,000 0.9 Pumps 500 36 % 1,000 27 Leakers 2,000 9 Leakers 5,000 9 Connectors sup b 250 2.8 % 500 1.5 Leakers 1,000 0.7 Unit ID EO 5 Valves 250 0.4 % 500 0.2 Leakers 1,000 0.2 Pumps 500 0 % 1,000 0 Leakers 2,000 0 Leakers 5,000 0 Connectors sup b 250 % 500 Leakers 1,000 Unit ID EO 6 Valves 250 1.6 % 500 1.4 Leakers 1,000 1.3 Pumps 500 % 1,000 Leakers 2,000 Leakers 5,000 Connectors sup b 250 4.1 % 500 4.1 Leakers 1,000 4.1 Unit ID EO 7 sup c Valves 250 1.2 % 500 1.2 Leakers 1,000 1.2 Pumps 500 0 % 1,000 0 Leakers 2,000 0 Leakers 5,000 0 Connectors sup b 250 0 % 500 0 Leakers 1,000 0 Unit ID EO 8 Valves 250 2.9 % 500 2.5 Leakers 1,000 1.5 Pumps 500 6 % 1,000 0 Leakers 2,000 0 Leakers 5,000 0 Connectors sup b 250 2.5 % 500 2.5 Leakers 1,000 2.5 Unit ID EO 9 Valves 250 5.3 % 500 4.5 Leakers 1,000 4.1 Pumps 500 25 % 1,000 25 Leakers 2,000 25 Leakers 5,000 25 Connectors sup b 250 7.6 % 500 5.9 Leakers 1,000 4.9 Unit ID BD 1 Valves 250 10.7 % 500 8.8 Leakers 1,000 7.5 Pumps 500 % 1,000 Leakers 2,000 Leakers 5,000 Connectors sup b 250 0 % 500 0 Leakers 1,000 0 Unit ID BD 3 sup c Valves 250 5.6 % 500 4.8 Leakers 1,000 4.2 Pumps 500 24 % 1,000 21 Leakers 2,000 15 Leakers 5,000 13 Connectors sup b 250 0.7 % 500 0.7 Leakers 1,000 0.7 Unit ID BD 4 Valves 250 7.3 % 500 6.3 Leakers 1,000 5.4 Pumps 500 2.9 % 1,000 0 Leakers 2,000 0 Leakers 5,000 0 Connectors sup b 250 10 % 500 8.6 Leakers 1,000 7.1 Unit ID BD 5 Valves 250 11 % 500 7.7 Leakers 1,000 5.1 Pumps 500 sup d50 % 1,000 Leakers 2,000 Leakers 5,000 Connectors sup b 250 3.4 % 500 1.5 Leakers 1,000 1.1 Unit ID BD 6 Valves 250 7.5 % 500 5.4 Leakers 1,000 3.3 Pumps 500 sup d33 % 1,000 Leakers 2,000 Leakers 5,000 Connectors sup b 250 % 500 Leakers 1,000 Unit ID BD 7 Valves 250 2.2 % 500 2.2 Leakers 1,000 2.2 Pumps 500 % 1,000 Leakers 2,000 Leakers 5,000 Connectors sup b 250 0 % 500 0 Leakers 1,000 0 Unit ID BD 8 Valves 250 21.9 % 500 18.4 Leakers 1,000 15.1 Pumps 500 58 % 1,000 50 Leakers 2,000 41.7 Leakers 5,000 41.7 Connectors sup b 250 7.4 % 500 6.7 Leakers 1,000 5.9 Unit ID BD 10 Valves 250 25 % 500 19.8 Leakers 1,000 17.1 Pumps 500 0 % 1,000 0 Leakers 2,000 0 Leakers 5,000 0 Connectors sup b 250 5.1 % 500 2.5 Leakers 1,000 1.9 Unit ID BD 11 Valves 250 11.3 % 500 10.1 Leakers 1,000 8 Pumps 500 16.7 % 1,000 16.7 Leakers 2,000 16.7 Leakers 5,000 16.7 Connectors sup b 250 1.4 % 500 1.4 Leakers 1,000 0.7 Unit ID BD 12 Valves 250 5.9 % 500 4.5 Leakers 1,000 3.7 Pumps 500 10 % 1,000 10 Leakers 2,000 10 Leakers 5,000 10 Connectors sup b 250 0.6 % 500 0.6 Leakers 1,000 0.6 Unit ID Acrolein Valves 250 0.1 % 500 0.1 Leakers 1,000 0.1 Pumps 500 4 % 1,000 4 Leakers 2,000 4 Leakers 5,000 4 Connectors sup b 250 0.1 % 500 0.1 Leakers 1,000 0.1 Unit ID Average SOCMI at 10,000 ppm Valves 250 11.5 % 500 Leakers 1,000 Pumps 500 % 1,000 Leakers 2,000 Leakers 5,000 9 Connectors sup b 250 % 500 Leakers 1,000 2.1 sup a Leak frequencies are not shown for phosgene units since all observations are "0." sup b Data includes flanges, threaded fittings, unions, and any other pipe-to-pipe connections other than welds. The records are not sufficient to permit differentiating among types of connectors. sup c Certain ranges of data for these facilities were discarded due to excessive rounding and truncation errors. sup d Type of pump seal is unknown. Units subject to the Benzene Equipment Leak National Emission Standards for hazardous air pollutants. During the EPA's reconsideration of the benzene equipment leak NESHAP (September 14, 1989, 54 FR 38044), EPA examined compliance reports from 1987 and 1988 for a randomly selected sample of 25 plants subject to the standard, which requires a 10,000 ppm LDAR program. This review showed that plants had more than 1.5 percent valves or more than 12.5 percent pumps exceeding 10,000 ppm. The average frequency of sources exceeding 10,000 ppm was lower than the average frequencies of 3 to 5 percent for valves and approximately 10 percent for pumps which had been expected after application of the standard. Acrolein process units. In 1989, EPA received screening data and process information on two acrolein process units. These data were collected voluntarily by the company to provide better emission estimates than could be obtained using the SOCMI average factors. The monitoring data represented eight types of equipment (gas valves, light liquid valves, light liquid pumps, flanges and other connectors, pressure relief valves, sample points, open-ended lines and compressors). These monitoring data were collected following the procedures specified in ''Protocols for Generating Unit-Specific Emission Estimates for Equipment Leaks of VOC and VHAP'', (Protocols) EPA- 450/3-88-010. In general, particular care was taken in measurements of equipment with screening values of less than 200 ppm. In addition, one of the units had approximately 60 percent sealed bellows valves and the other unit had no sealed bellows valves. The screening data indicated average leak frequencies of 0.26 to 0.09 percent for valves and of 0.13 to 0 percent for flanges and other connectors at leak definitions of 100, 1,000, and 10,000 ppm. These data showed that both conventional and sealed bellow valve designs can achieve very good performance with few leaking valves. The observed leak frequencies of all components were substantially below the levels found by EPA in studies conducted in the early 1980's. Amine unit-West Virginia. In 1987, EPA received monitoring data for two process units handling amines. These units were selected for testing because of the service conditions (i.e., high temperature and high pressure) and high vapor pressures of the process materials. One unit (Unit 1) was characterized as handling relatively nontoxic and innocuous substances and the other unit (Unit 2) was characterized as handling a more toxic compound, one with a threshold limit value of 10 ppm. These data were collected voluntarily by the company to calculate a unit specific emission estimate using the EPA's leak/no leak emission factors. Monitoring data were provided for pumps, valves, and flanges and other connectors. The monitoring data for Unit 1 showed 33 percent of the pumps, 16 percent of the light liquid valves, 27 percent of the gas valves, and 0.4 percent of the flanges and other connectors exceeding 10,000 ppm. Unit 2 had 5.6 percent of the pumps, 0.8 percent of the light liquid valves, 1 percent of the gas valves, and 0.1 percent of the flanges and other connectors exceeding 10,000 ppm. Ethylbenzene/styrene unit-Texas. Limited information was received on screening values measured at this facility's EB/S process unit, which is subject to the benzene equipment leak NESHAP. Screening data were provided for a total of approximately 2,500 valves and pumps. The data were provided as time aggregated summaries of screening values for a 1-year period from 1987 to 1988. These data indicated that 99 percent of the time the valves and 90 percent of the time the pumps had screening values less than 500 ppm. Information submitted for compliance with the benzene equipment leak NESHAP showed on an annual average basis 0.04 percent of the valves and 3.6 percent of the pumps exceeded 10,000 ppm. No description was provided of the procedures used in the data collection, other than that Method 21 was used. Chemical Manufacturers Association studies of Butadiene, Ethylene Oxide, and Phosgene process units. In addition to the preceding information, comprehensive screening data were also provided to EPA on 33 BD, EO, and phosgene production process units. The monitoring data represented eight types of equipment (gas valves, light liquid valves, light liquid pumps, flanges and other connectors, pressure relief valves, sample points, open-ended lines and compressors) at essentially all BD, EO, and phosgene producers operating in the United States. These data were collected voluntarily by the CMA committees to provide EPA with better industry-wide estimates of emissions from equipment leaks at these process units than could be obtained using EPA average factors for SOCMI. These monitoring data were collected following the procedures specified in the Protocols. These studies provide the most comprehensive equipment screening data available that document the full range of leak frequencies for large numbers of process units used for the production of specific chemicals. The EPA also obtained information on the plant work practices, operations, and the equipment design specifications. Conditions in the equipment surveyed spanned a wide range of temperatures and pressures 10 degrees C to 288 degrees C and 136 to 3,550 kPa (50 degrees F to 550 degrees F and 5 to 500 pounds per square inch gauge) , as well as a range of line diameters (0.5 inch to greater than 6 inches). The screening data indicated that most process units at these facilities are characterized by equipment leak frequencies far below levels found by EPA in studies conducted in the 1970's and 1980's. In addition, the data base indicated much lower leak frequencies at the phosgene process units than at the EO and BD units, and the EO units in general had lower leak frequencies than {pg 62663} the BD units. Table 9 presents the plant-specific leak frequency data by component type, at various leak definitions for the three processes. The data shown in Table 9 are the screening data that were judged to be consistent with proper calibration and use of the test equipment. The screening values for the ten BD and nine EO process units were also adjusted for the instrument response factor to represent more accurately the screening concentration of equipment in BD or EO service. The data presented in Table 9 represent process units in various stages of control and operations. Some plants operated using formal LDAR programs and others did not; some units were screened just before a shutdown for maintenance and others were screened just after startup following a scheduled maintenance shutdown. The above information, when compared with the earlier studies and analyses indicated to EPA that the performance capability of available technology is better than had been expected from a 10,000 ppm LDAR program alone. Furthermore, analyses of these data showed, like earlier studies, that no single factor or technology was responsible for the better performance. Rather, the recent data confirmed that effective leak control involves a combination of factors. For example, statistical analysis of the recent EO and BD data failed to identify specific process parameters or equipment designs that alone would significantly affect leak frequencies. D. Development of Framework and Selection of Maximum Achievable Control Technology Selection of MACT for equipment leaks, unlike many other source types, requires a balancing of a number of interrelated factors and is not based on identification of a specific best control technique or approach. The following discussion explains the Committee's assessment and consideration of the available data in the development of the framework as well as the selection of MACT. 1. Information Considered In deliberations over the framework, the Committee considered screening data from the CMA surveys of EO, BD, and phosgene production units as well as data and information from other process units. The additional information considered included data provided on an EB/S process unit and on three facilities in Texas subject to a 500 ppm leak permit condition. The information available for the EO, BD, and phosgene units, and for the EB/S unit was described in the preceding section. The information provided on the three facilities in Texas is described below. a. Texas Air Control Board data. Additional data from three plant sites were provided by the Committee member representing the TACB. These plants are subject to a Texas regulation that requires LDAR with directed maintenance at a 500 ppm leak definition. This requirement has been applied as of the permit program for new and modified sources. Data were provided for cumene, phenol/acetone, and butyraldehyde/butanol process units which had been operating a 500 ppm LDAR program for 1 to 6 quarters. These units had average leak frequencies at 500 ppm for valves ranging from 0 to 3.6 percent. The average leak frequency at 500 ppm reported for pumps varied from 0 to 20 percent in the first period and from 3.6 to 5 percent in the second period. Average leak frequencies for flanges varied from 0 to 2 percent. These data show for several chemicals that a 500 ppm LDAR program is effective and will significantly decrease leak frequency. b. Plant visit. In addition to considering the available screening data, several members of the Committee visited three integrated chemical plants, one of which had 20 operating units including a phosgene unit, a toluene diisocyanate unit, and a hydrogen fluoride unit. One conclusion drawn from these visits was that the acute toxicity of phosgene and toluene diisocyanate, combined with other features of these units, has resulted in the routine use of a number of effective leak control practices in these process units. Measures taken to prevent and minimize leaks in phosgene and isocyanate units include design to minimize the number of components, stringent quality assurance/control programs, bench testing of equipment for leaks, and careful management and extensive continuous monitoring of the unit to detect leaks and problems. In addition, phosgene and isocyanate units are shut down for immediate repair whenever any leak is detected. Immediate shutdown is possible because phosgene and isocyanate process units are small and are not operated as of an integrated process. In units where less acutely toxic compounds are used or produced and in facilities where operations are large scale and highly integrated, such measures were not observed and the units are not (and cannot be) immediately shut down for repair whenever a leak is detected. Rather, shutdown of facilities with large integrated processes can require several days to cease production, followed by up to several weeks to drain and decontaminate the process equipment prior to actually initiating repair. These units are typically operated for a fixed time period before they are shut down for maintenance and repair. Representative operating times between scheduled shutdowns range from 3 months to several years. The Committee also concluded that equipment design and an on-going quality control/assurance program has a significant role in achieving low leak frequencies. This conclusion was supported by remarks made to the Committee by a chemical industry representative who described his company's recent experience with two EO production units. Following completion of a screening study, the company undertook a program to reduce the leak frequency and emissions from the units. Steps taken consisted of component inspection, complete or partial replacement of over 700 valves and 2,000 gaskets, modification of design to reduce leaks, and regular monitoring and inspection for leaks. The company reported that there was an overall reduction of about 80 percent in the number of leaking components following completion of this project. 2. Committee Analysis and Consideration of the Data The Committee considered the data and information described in the preceding section and concluded that the EO, BD, and phosgene data should serve as the principal basis for establishing the performance of MACT for SOCMI processes. The EO, BD, and phosgene data, however, were considered in light of the broader data; especially, the Texas 500 ppm LDAR program and the site visits. On this basis, the Committee agreed that phosgene units are not similar in design, operation, and maintenance to most SOCMI processes. Some of the maintenance and operating practices for phosgene units control programs went beyond MACT for other SOCMI processes. One example of this is the practice of immediate shutdown. Immediate shutdown is economically and technically infeasible for other processes because shut down of the one unit would require shut down of all the integrated operations at the site and it would be impractical to clear large volumes of chemicals in short time frames. In some cases, emissions from clearing process material from or all of the units could greatly exceed the emissions that would result from the leaking equipment. Therefore, while the {pg 62664} phosgene units' performance did provide an indication of the potential for leak reduction, the Committee concluded that the EO and BD units are more representative of the range of processes in SOCMI than are phosgene or toluene diisocyanate units. The Committee did recognize, as indicated by the framework and the base performance levels, that low leak frequencies are generally achievable at chemical plants and that the rule should encourage attaining low emissions. From the data and observations, the Committee also concluded that low emission performance results from combinations of monitoring (or surveillance), repair of leaking equipment, use of quality assured and quality controlled equipment suitable for the process operating conditions, and a quality controlled maintenance and repair program. The Committee concluded that such combinations are MACT for SOCMI processes. The information available did not identify any single factor, such particular equipment design or specific work practices or maintenance programs that by itself would guarantee low emission performance in all cases. Furthermore, the Committee did not identify the results at any one specific process unit or performance level as representing MACT performance because there is no apparent break point that represents a limit of what is achievable with a combination of technologies that are available. Thus, the Committee identified the best performing EO process units, the EB/S unit, the Texas 500 ppm units, and possibly the best performing BD process units (for pumps) as reflecting the performance of MACT for production processes involving HAP's. 3. Regulatory Approach Because available information indicated that best performance for equipment leaks could not be defined or reflected in a single numerical standard, type of technology, or group of work practices, the Committee focused on developing a regulatory framework that, when combined with performance levels, would reflect MACT. Key principles that evolved during the negotiations and that led to the regulatory framework are: (1) Incentives should be included in the standard to reward good performance and prescribed corrective actions should be included to ensure poor performance is improved in a timely manner. (2) Flexibility in achieving specified results is important. This is necessary to allow consideration of process to process variations in operating conditions and equipment specifications and to provide incentives for identifying the most effective combination of equipment and practices. (3) Identifying, designing, and implementing systems to meet the specified results requires varying amounts of lead time. Effective control requires continual analysis and adjustment, and may involve several technologies which cannot be identified or applied at one time or in one step. (4) More frequent monitoring and maintenance or a prescribed program should be employed, at a minimum, for those process units that do not achieve the specified results through a program of their own design or through quarterly inspection and maintenance alone. (5) There is insufficient information available to set an enforceable not-to- be-exceeded standard. The existing information is too limited in scope to predict with certainty what level of performance could be achieved by processes for which no data were available. The majority of the Committee deliberations concerned development of a regulatory framework and performance levels that would incorporate the above key principles, while reflecting the performance of MACT. These deliberations primarily focused on developing provisions for valves, pumps, and flanges because these emission sources offered the greatest potential for emission reductions. The Committee used the existing LDAR program as a starting point for development of the framework for these sources. Early-on, the Committee recognized the need for a phased-in approach for valves and pumps that would provide time for plants to develop and implement effective programs and to focus on larger emission sources first. The phased-in approach is necessary because effective control requires continual analysis and adjustment of programs as experience is gained. The Committee defined MACT for pumps and valves in terms of three phases of requirements of increasing stringency. Each phase, as it occurs, reflects the performance of MACT for units in that phase. The Committee also deliberated on several methods of linking monitoring frequency with performance and of providing incentives to use low emission equipment or to reduce the amount of equipment in VHAP service. The final decision was to link reduced monitoring frequency with lower leak frequencies, and to give credit for removal of equipment or use of low leak design equipment. The performance format of the standards would also encourage the use of low-leak technologies. This approach is consistent with the data which indicate that there are combinations of technology, maintenance practices, and quality assurance programs that can achieve very low leak frequencies while using less frequent monitoring intervals. The Committee also debated whether to require shut down of process units for repair of leaking equipment whenever performance deteriorates below some specified performance level. The Committee ultimately concluded that process units should not be shut down to meet a performance level. It was generally agreed that clearing of process materials could cause greater emissions than delaying repairs until the next scheduled shutdown. The Committee also agreed to retain the provisions of the existing equipment standards for pressure relief devices, sample points, open-ended lines, product accumulator vessels, and compressors. The existing standards for these items of equipment require installation of equipment or control devices and essentially eliminate emissions from equipment leaks. The Committee considered these requirements to be MACT. For the valve, pump, and connector standards, the Committee primarily used the data from the better-performing EO units, the best-performing BD units (for pumps only), the Texas 500 ppm units, and the EB/S unit to establish the performance levels for the monitoring frequency. A major consideration in the selection of performance levels was the understanding that the levels are interrelated with the framework and phasing and are not necessarily the levels that would have been chosen for a not-to-be-exceeded standard. In the case of the valve standard, the selection of the base performance levels was also influenced by the inclusion of provisions that provide an alternative to monthly monitoring. Referred to as the QIP, this program ensures that plants subject to this QIP replace poorer performing valves (during normal replacement) with superior performing technologies until less than 2 percent leaking valves is achieved and implement a quality assurance/control program to ensure that all elements of MACT are utilized. Also, the QIP was included in the framework to allow those process units that do use all elements of MACT but do not achieve the specified numerical performance levels to remain on a quarterly monitoring schedule instead of increasing monitoring frequency.{pg 62665} The basis for the selection of the base performance levels and other elements of MACT for valves, pumps, and connectors is discussed below. The basis for the other provisions of the valve, pump, and connector standards as well as the other specific requirements of the regulation is given in Section VIII.F of this preamble. a. Valves. For valves, the Committee initially considered both equipment standards and LDAR programs and ultimately agreed upon a three-phase LDAR program. Initially, the Committee considered requiring the use of equipment such as sealed bellows valves because the equipment is often believed to have zero emissions. This option was not selected for several reasons. First, these valves are not suitable for use in all operating conditions and services encountered in SOCMI processes. Specifically, these valves cannot be used in corrosive streams, with gritty materials, or in extremely high pressure or temperature conditions. Second, sealed bellows valves can leak internally, and will eventually fail. When failure occurs, there is potential for massive leaks. Due to this possibility for leaks, monitoring of these valves is necessary. Considering this, the Committee thought that the leak detection, repair, and base performance level standards framework described above would best reflect MACT. This approach allows and would give credit for any low emission performing valve regardless of its design. The Committee considered several LDAR programs which tied monitoring frequency to the performance level achieved and created a framework that consists of several combinations of work practices and performance levels. In determining the maximum monitoring frequencies for Phase III, the Committee considered requiring weekly monitoring and monthly monitoring. The Committee agreed that a monthly program would be generally feasible and would reduce emissions, while being sufficiently costly to provide an incentive for owners of such sources to identify more cost- effective approaches. For plants that do not achieve the performance level required for quarterly monitoring of valves, the standard requires monthly monitoring or participation in the QIP program. In addition, the Committee also decided to allow less frequent monitoring to provide an incentive to achieve better performance than that required for the base quarterly monitoring program. Annual and semi- annual frequencies were selected for the incentive programs. In selecting the base performance level and leak definition, the Committee considered the performance of the better performing EO units and the performance of the units subject to the Texas 500 ppm standard. Using these data, the Committee considered leak definitions in the range of 50 to 2,000 ppm and performance levels from 1 to 5 percent leaking valves. These data and the fact that, in practice, low leak frequencies were achieved by a 500 ppm LDAR program alone supported the practicality and achievability of a performance limit for a quarterly LDAR program of 2 percent leaking valves at a 500 ppm leak definition. Leak definitions lower than 500 ppm were not selected due to some Committee members' concerns regarding the practical ability to measure and repair smaller leaks. In contrast, other members of the Committee favored leak definitions lower than 500 ppm. Ultimately, a leak definition of 500 ppm was selected in light of data demonstrating the practicality and current implementation of a 500 ppm LDAR program. The Committee also agreed that units with less than 1 percent leaking valves could monitor the valves semiannually and those with less than 0.5 percent leaking valves could monitor annually. These performance levels were selected considering the Committee had previously agreed to exclude nonrepairable valves (up to a maximum of 1 percent of the total number of valves in VHAP Service) from the calculation of percent leaking valves (see Section VIII.F.4 of this notice for discussion of the basis for the nonrepairable provision). These performance levels have been demonstrated by the better performing EO process units, the EB/S unit, and by the plants subject to the Texas 500 ppm regulation. The Committee considered these levels to represent the performance of MACT. b. Pumps. In deliberations on the standard, the Committee considered the data summarized in Table 9, additional analyses of the EO/BD data, and the comments of pump seal manufacturers. The EO/BD data showed pump leak frequencies and overall performance to be highly variable with leak frequencies ranging from 0 to 50 percent at a 1,000 ppm leak definition. Additional analyses were done to compare the performance of DMS and SMS pumps. These comparisons showed that the performance of both DMS pumps and SMS pumps varied widely among the process units. In general, DMS pumps screened at less than 500 to 1,000 ppm. In some cases, the SMS pumps screened at less than 100 ppm and could be viewed as having comparable performance to that of DMS pumps. For the majority of SMS pumps screening at less than 10,000 ppm, the screening values were fairly uniformly distributed from 100 to 9,100 ppm. Thus, the data did not provide a clear indication of the performance level that can be achieved by SMS without incurring high maintenance costs. From its consideration of the available data, the Committee generally concluded that pumps equipped with DMS, as a group, have lower leak frequencies than those equipped with SMS. Consequently, the Committee deliberated at great length on whether MACT is reflected by an equipment standard for DMS on all pumps or by a performance standard. The Committee decided not to establish an equipment standard because DMS are not suitable for use in all cases and because a DMS equipment standard would be very costly and would preclude effective lower cost options. Dual mechanical seals cannot be used with materials where leakage of the barrier fluid would affect product purity (such as with medical products), with polymerizing monomers, or on reciprocating pumps. Also, the pump casing of some existing packed seal and SMS pumps does not allow installation of a DMS assembly. In such cases, a DMS equipment standard would require replacement of the entire pump. The cost of a DMS equipment standard was considered in a general sense; however, no agreement was reached on the cost of installation of DMS systems or on the cost of a DMS equipment standard, including the cost of pump replacements. The opinion that a DMS equipment standard would be very costly was based on general knowledge and the experience of some Committee members. However, to give all Committee members some perspective on the potential costs, EPA provided an estimate of the annualized cost for retrofitting an existing pump with a DMS system. This estimate showed the cost would be approximately $2,200 per year (1990 dollars) for each pump in VHAP service. In previous rulemakings, EPA has estimated that a requirement of DMS on pumps has an incremental cost effectiveness over a 10,000 ppm LDAR program of $5,600/Mg ($5,080/ton) of total VOC or pure VHAP (1978 dollars), which is comparable to approximately $15,000/Mg ($13,600/ton) in 1990 dollars. Some Committee members expressed an additional concern that a DMS equipment standard would have a {pg 62666} significant adverse economic effect on small producers. To establish a performance standard that would reflect MACT for pumps, the Committee considered the performance capabilities of SMS (and packed seals) compared to the performance, applicability, and cost of DMS. Discussions among the Committee members, chemical industry representatives, and pump seal manufacturers led to the conclusion that it is not possible to identify precisely best performance levels achievable by SMS and that, although advances in technology could reasonably be expected over the next 5 years, the limits on these advances could not be established. Because of technical limitations on the use of DMS on pumps in food/medical service and the use of mechanical seals on pumps handling polymerizing monomers, the Committee agreed to establish separate performance standards for pumps in those two services. For pumps in general chemical service, the Committee agreed to establish 1,000 ppm as a performance target reflecting MACT and to establish 2,000 ppm as the concentration at which repair is required. The Committee agreed to require repair of pumps with an instrument reading greater than 2,000 ppm as a result of some Committee members concerns that repair at lower concentrations could result in significant and costly maintenance, with little to no emission reduction. The Committee also concluded that a 10 percent leak frequency (or three leaking pumps) is the appropriate point to impose a QIP that includes a mandatory replacement provision, if the leak frequency remains greater than 10 percent. On balance, it was believed that this combination of requirements would provide the time and incentive needed to achieve best performance from SMS, would allow use of SMS where they can achieve low emissions, and would allow use of both SMS and DMS pumps, as appropriate, to achieve the standard. For pumps in food/medical service, the Committee selected 2,000 ppm and 10 percent leak frequency as the performance level. The leak definition was set at 2,000 ppm to account for the limited applicability of DMS pumps and the uncertainty that SMS seals could achieve lower performance levels without excessive replacements in this service. The Committee also agreed that processes with pumps in food/medical service that exceed 10 percent leaking pumps would not be subject to the mandatory replacement provisions in the QIP. For pumps handling polymerizing monomers, the Committee selected 5,000 ppm and 10 percent leak frequency for the performance level. As mechanical seals cannot be used on pumps in polymerizing monomer service, industry representatives generally maintained that pumps in this service could not achieve a 2,000 ppm leak performance level. The Committee ultimately agreed to the 5,000 ppm leak definition based on expert judgment that 5,000 ppm would reflect best performance and on the general lack of data for pumps in this service. It was also agreed that facilities that exceed 10 percent leak frequency would not be subject to the mandatory replacement provision of the QIP. c. Connectors. Provisions for a LDAR program for connectors were developed after the Committee generally agreed that connectors could be a significant source of emissions at a well-controlled plant and that emissions could be reduced. In the development of these provisions, the Committee considered the data summarized in Table 9 and the contribution of connector emissions to total emissions for several EO/BD process units. These data showed a range of connector leak frequencies at different leak definitions (e.g., 3 percent at 10,000 ppm to less than 2 percent at 250 ppm) and showed that connectors could be a significant source of the total emissions. Some Committee members believed the relatively high leak rates observed at some process units were a result of infrequent or no inspections and maintenance. The Committee agreed that connector leaks should be controlled, therefore, a LDAR program was established to ensure that low leak frequencies are attained. In development of these provisions, the Committee agreed that LDAR can reduce connector leak frequencies and that assuring MACT performance requires less frequent monitoring than is necessary for pumps or valves. Less frequent monitoring is needed because connectors have no movings and once repaired they should remain leak free for extended periods. Information provided by industry members on the Committee indicated that a number of actions can be taken to reduce or eliminate leaks. In most cases, it was expected that tightening the flange bolts on flanged connectors would eliminate the leak. It was also expected that in other cases it may be necessary to replace the gasket or to correct faulty alignment of surfaces. These latter cases are expected to be relatively infrequent. The Committee, therefore, decided that annual monitoring was reasonable. The Committee also concluded that process units that demonstrate sustained performance at the level of the standard should be allowed to monitor less frequently than annually. This concept is similar to the skip-period provisions for valves in the existing equipment leak regulations. This approach is consistent with quality control principles and the Committee concluded that this approach will ensure MACT performance. The Committee selected the performance level and leak definition considering the data from the EO/BD units, the Texas 500 ppm units, and the general conclusions drawn from the Committee's discussions. These data and the fact that, in practice, low leak frequencies were achievable persuaded the Committee that a leak definition of 500 ppm and a base performance level of 0.5 percent leaking connectors would reflect MACT. Process units that have 0.5 percent, or greater, leaking connectors are required to implement an annual LDAR program for connectors. Process units that have less than 0.5 percent leaking connectors are allowed to monitor all connectors in a biennial or quadrennial program. E. Selection of Format of Standards Under section 112 of the Act, national emission standards must, whenever possible, take the format of a numerical emission standard. Typically, an emission standard is written in terms of an allowable emission rate (mass per unit of time), performance level (e.g., 90 percent control), or an allowable concentration. These types of standards require the direct measurement of emissions to determine compliance. For some source types, emission standards cannot be prescribed because it is not feasible to measure emissions. Section 112(h)(2) recognizes this situation by defining two conditions under which it is not feasible to establish an emission standard. These conditions are: (1) If the pollutants cannot be emitted through a conveyance designed and constructed to emit or capture the pollutant; or (2) if the application of measurement methodology is not practicable due to technological and economic limitations. If an emission standard cannot be established, EPA may instead establish a design, equipment, work practice, or operational standard or combination thereof. For equipment leak sources, such as pumps and valves, EPA has previously determined that it is not feasible to prescribe or enforce emission standards. Except for those items of equipment for which standards can be set at a specific concentration, the only method of measuring emissions is total enclosure {pg 62667} of individual items of equipment, collection of emissions for a specified time period, and measurement of the emissions. This procedure, known as bagging, is a time-consuming and prohibitively expensive technique considering the great number of individual items of equipment in a typical process unit. Moreover, this procedure would not be useful for routine monitoring and identification of leaking equipment for repair. The proposed standards incorporate several formats: equipment, design, base performance levels, work practices, and operational practices. Different formats are required for different types of equipment because of the nature of the equipment, available control techniques, and applicability of the measurement method. In the next section, the rationale for selecting particular format is explained for each type of equipment. For each source type, the feasibility of prescribing or enforcing an emission standard is discussed. F. Selection of Emission Limits and Work Practice Requirements 1. Applicability The Committee considered and ultimately established several additional applicability criteria that further defined the scope of the negotiated standard. This section presents the Committee's consideration of minimum VHAP concentration, time in VHAP service, and pilot plants and research facilities. a. Stream volatile hazardous air pollutant concentration. The Committee discussed at some length what minimum concentration should be used for determining applicability of the negotiated standard for specific streams within an affected process. This minimum concentration was of particular concern to industry members of the Committee because there are trace quantities of VHAP in many process streams. The number of equipment components potentially subject to the standard and the associated costs increase substantially as the concentration is decreased; and the emission reduction becomes small. Other Committee members were concerned that emissions from equipment handling streams containing low VHAP concentrations (e.g., less than 10 percent) become relatively more significant as higher concentration streams are controlled. Based on these considerations, the Committee agreed the standard would apply to equipment containing or contacting process materials that are 5 percent VHAP or greater. In SOCMI processes, this concentration was viewed as being a point of relatively low emissions and diminishing returns. b. Time-in-VHAP service. In certain chemical plants, particularly batch processes which produce a number of different products, there is equipment that is used in VHAP service only occasionally. In such cases, implementation of the standard could be difficult and would achieve very little emission reduction. Pumps and compressors used only during startup or shutdown of a process unit are one example of such equipment. Other examples include equipment used in batch pharmaceutical processes or batch steps in continuous processes. For these situations, the Committee concluded that equipment that is operated 300 hrs/yr, or less, in VHAP service should be exempt. c. Research facilities. Exemption of research facilities (including pilot plants) from the negotiated standard was discussed on several occasions by the Committee. Some members thought that frequent changes in operations at these facilities would make compliance with the standard difficult or would incur unnecessary costs. Other Committee members were of the opinion that the negotiated standard should apply equally well to pilot plants and research facilities as production units, and that chemical industry pilot plants can be quite large and have substantial emissions. An additional concern expressed by some members was that any exemption should be based on objective criteria, not the intended use of the material or the purpose of the facility. The Committee did not resolve this issue, and concluded that the final Clean Air Act Amendments should dictate the disposition of research facilities, as several versions of the pending legislation provided for limited exemptions. The final amendments, however, are not definite on how to deal with research facilities and leave it to the EPA's discretion to '' establish a separate category covering research or laboratory facilities, as necessary (emphasis added), to assure the equitable treatment of such facilities.'' (Section 112(c)(7).) The issue of regulating research facilities extends beyond equipment leaks, because research facilities can also have other emission points such as process vents or storage tanks. During preparation of this preamble, after enactment of the Clean Air Act Amendments, a consensus of Committee members was reached that it was more appropriate to address research facilities in the overall HON and to deal with it outside the negotiation, since no consensus was reached during the negotiation and the issue is broader than the negotiated rule. Consequently, those portions of today's notice addressing the regulation of research facilities are not of the negotiated agreement and may be commented on by the negotiators. 2. Compliance Dates In developing the compliance schedule and framework of the standards, the Committee anticipated the Clean Air Act Amendments requirement that compliance dates for each category apply as expeditiously as practicable, but not later than 3 years after the effective date of the standards. The Committee recognized in the selection of compliance dates that the equipment leak regulation is primarily a work practice standard. Unlike standards that require installation of control equipment, the equipment leak regulation does not require major installation of equipment or large capital expenditures. Also the time needed to identify all components in VHAP service, establish a recordkeeping system, obtain monitoring equipment, and initiate an inspection and maintenance program is relatively short. Based on these considerations, and recognizing that advance notice would be provided when the rule is proposed in 1991, the Committee concluded that 6 months after promulgation was a reasonable lead time for compliance with Phase I. An additional concern to the Committee in establishing the compliance schedule was the potential for the negotiated rule to overwhelm the resources of both industry and regulatory agencies because of the large number of affected facilities (about 1,000 process units) expected. The Committee considered several methods of distributing the peak demand for equipment and staff, and ultimately agreed to a staggered implementation schedule that spreads the compliance over a 1-year period. The affected source categories were divided into five groups roughly based on the carcinogenicity and weight of evidence for the chemicals associated with the category. Sources handling chemicals with the highest health hazard were assigned to Group I and those with the lowest were assigned to Group V. In addition, the seven non-SOCMI source categories as well as benzene, toluene, and xylene production were included in the first group of source categories. The Committee agreed that this would allow industry, contractors, vendors, and regulatory agencies the time necessary {pg 62668} to manage successfully the increased demand on resources and personnel that is anticipated to result from the negotiated rule. The provisions of the negotiated rule would become applicable to those processes listed in first group 6 months after promulgation of the HON, and to one additional group of production processes each successive quarter. The negotiated rule would, therefore, become applicable to the last group 18 months after promulgation of the negotiated rule. Apart from staggered applicability dates, all other provisions of applicability, including the provisions for a source to request an extension, remain as provided for in the General Provisions for NESHAP. The negotiated rule also allows an owner or operator to elect to comply with the applicability date of an earlier group. The Committee agreed that some owners or operators at plants having process units from two or more groups might prefer to comply with the provisions of the negotiated rule all at once, or in fewer phases, in order to avoid confusion or misunderstandings about what rules or procedures were applicable to which individual items of equipment. 3. Phases of Valve and Pump Standards Both the valve and pump standards were structured in three phases to allow time to improve performance and to achieve progress toward lower emission levels. The Committee designed Phase I to allow existing facilities unfamiliar with the existing rules the time necessary to develop and implement a 10,000 ppm LDAR program, as well as to assess the necessary changes in operations, maintenance, and training of employees. Phase I is considered necessary because the majority of existing facilities do not presently implement LDAR programs and it was recognized that establishing programs can be a lengthy process. The Committee agreed Phase I should begin on the applicability date for an existing unit and last for 1 year. A 1-year interval will allow 4 quarters of monitoring which should provide sufficient time to establish a program and to identify and repair valves and pumps with the highest emissions. Phase II was established to achieve further emission reductions by reducing the leak definition to 500 ppm for valves and 5,000 ppm for pumps. This phase is intended to expand the focus from the higher screening value leaks to lower values and to continue the identification of problem equipment. The Committee agreed that Phase II would begin 1 year after the applicability date for an existing process unit and would continue for 1 and 1/2 years before Phase III is implemented. In Phase III for valves, the Committee added a base performance level to the work practice requirements of Phase II. The base performance level was used to determine the monitoring frequency or the applicability of the QIP. The base performance level was added to ensure a more certain emission performance than achieved through the work practice requirement alone and to create an economic incentive to improve performance. For Phase III of the pump standard, the Committee added a base performance level to the work practice requirements and reduced the concentration defined as a leak. The Committee established separate leak definitions for three types of pumps to ensure that each category achieves MACT. For new process units, the Committee agreed that for both pumps and valves Phase II should be implemented at startup and Phase III should be implemented beginning 1 year after startup. It was expected that new units will be designed using better equipment design or practices and considering the requirements of the standard. The Committee provided a 1-year period in Phase II to allow time needed for adjustments to operations and to correct any problems encountered during startup of the process unit. 4. Valve Standard As discussed earlier, the Committee developed the valve standard using the LDAR program of the existing rule as a starting point. The standard recommended by the Committee consists of three phases of requirements of increasing stringency and these requirements are a combination of LDAR programs and performance levels. Figure 2 presents the conceptual approach for the valve standard. The rationale for the framework of the standard and the selection of the base performance levels was explained in Section VIII.D. This section explains the rationale for other provisions in the proposed valve standard. {SEE ILLUSTRATION(S) IN ORIGINAL DOCUMENT} The Committee used LDAR programs as the starting point because LDAR is an effective means of detecting and eliminating emissions from seal failure. For emission sources such as valves, emission standards do not provide appropriate control procedures. As previously discussed, an equipment standard was not selected because equipment, such as sealed bellows valves, cannot be uniformly applied to all SOCMI processes. Moreover, proper use of equipment such as sealed bellows valves, also requires a LDAR program because these valves can fail and can have high emission rates upon failure. a. Valve repair intervals. Repair of leaks soon after detection is a key feature in the proposed rule and is a factor in the effectiveness of LDAR programs. The Committee agreed to retain, in the negotiated standard, the repair intervals (i.e., first attempt at repair in 5 days and repair within 15 days) specified in the existing rules (e.g., 40 CFR part 61, subpart V). The repair intervals in the existing standards are intended to provide effective emission reduction while allowing the time necessary for scheduling of more complex repairs. As in the existing equipment leak standards, the first attempt at repair is required as soon as practicable after detection of the leak and no later than 5 days after discovery. Most valve repairs, such as tightening the bonnet bolts, can be performed quickly, and 5 days should provide sufficient time to schedule simple field repairs that do not require isolation of the valve from the process. Attempting to repair a leak within 5 days will help identify those valves that cannot be repaired with simple field repair or without shutdown of the process unit. Valves that are not repairable by simple field procedures may require removal from the process for repair. The 15-day repair interval provides sufficient time to identify methods for isolating leaking valves for repair when isolation is necessary and extensive repairs are required. Shorter repair intervals were not selected because they could cause problems in performing effective repairs on valves that must be isolated from the process. The Committee also recognized that some valves cannot be repaired without shutting down the process and that process shutdown is costly and may result in greater emissions than delaying repair until the next scheduled shutdown. For these reasons, the existing equipment leak regulations allow, under certain conditions, delay of repair of these ''nonrepairables'' until the next facility shutdown. These rules, in general, require that these valves be repaired during that shutdown. The Committee reviewed the delay of repair provisions in the existing rules and concluded that these provisions are appropriate and should be included in the negotiated standard. b. Nonrepairables. There was considerable discussion by the Committee of whether those valves that cannot be repaired without a process unit shutdown should be considered in determining the required monitoring frequency. One view was, if unrepairable components were included, the potential for increased monitoring frequency would provide a strong incentive for companies to take every possible step to prevent leaks or to repair leaking valves. However, it was also recognized that nonrepairable valves may not always be preventable and counting these could result in higher leak percentages in process units or facilities that have infrequent shutdowns. Moreover, increased monitoring frequency, if triggered by nonrepairable components, is of little direct benefit since the nonrepairable valves will remain unrepaired in spite of more frequent inspection. In light of these considerations, the Committee sought to provide a means to provide motivation for plants to limit the number of nonrepairable valves, while at the same time avoiding imposition of unproductive costs or inadvertently increasing emissions. The agreed upon approach achieves such a balance by excluding nonrepairable valves up to a total of 1 percent of the valves in VHAP service from the calculation of percent leaking valves for sampling periods after the leaking valve was first identified. c. Averaging periods for calculation of percent leaking. In establishing the base performance levels in Phase III, the Committee considered the variability inherent in measurements of equipment leaks and the variability in leak rates over time. There was considerable debate as to whether measurements of equipment leaks are sufficiently precise to warrant performance criteria or how to consider this variability in the framework of the rule. The concept of a 2-period rolling average was introduced as a means of ensuring that random fluctuations, of themselves, do not force a facility into more frequent monitoring. In addition to the benefits of averaging, the rolling average of 2 consecutive monitoring periods provides the opportunity for the plant to take action, such as increased surveillance on a subset of valves, based on higher than normal leak rates in a single period. It also allows the owner or operator the necessary flexibility to implement supplemental quality assurance programs to ensure that performance remains below base levels. The 2-period rolling average is designed to encourage such plant designed quality assurance programs and it does not penalize more frequent inspection and maintenance. The Committee selected this approach for semiannual and more frequent monitoring schedules. The Committee did not choose to apply the 2-period average approach to annual monitoring programs because of the low performance level required for annual monitoring and concerns that effective control programs could be unintentionally penalized due to a single high monitoring result. Specifically, a Committee member described the effect of extreme weather that caused an unplanned shutdown and resulted in an abnormally high number of leaking valves after startup. In this case, the plant would have to revert to more frequent monitoring if only two monitoring periods were considered even though the leak frequency was normally near zero. Such a penalty would be inconsistent with the intent to encourage effective, plant designed quality assurance programs that achieve continuing low leak performance. In light of this, the Committee agreed to base the annual monitoring criterion on the average of 3 out of 4 consecutive monitoring periods. This averaging procedure is intended to provide relief for facilities with histories of good performance from the effects of unusual and infrequent perturbations. d. Credits for removing valves. In developing the negotiated standard, the Committee recognized that minimizing the number of components in a process is one of the most effective means for reducing leaks. Therefore, the Committee gave careful consideration to formats for, or means of, providing an incentive to minimize the number of components and the amount of credit provided. For new process units, no mechanisms could be identified that would provide such an incentive for all SOCMI processes. New unit designs differ widely from process to process and it was not possible to define a norm or baseline from which component reductions could be determined for new units. In contrast, equipment count baselines can be established for existing process units at the time these units become subject to the standards (for new units at startup) and can be used as the basis for calculating credits for future reductions in the number of valves. The Committee elected to provide these credits in the calculation of percent leaking valves and the associated determination of monitoring {pg 62671} frequency. The Committee also decided that use of the valve credit provisions should be at the option of the owner or operator of the process unit because of the burden associated with the recordkeeping needed to document the reduction. The Committee considered credits for components removed ranging from full credit to one-half credit for each component removed. The full credit option was not selected because some Committee members thought that it could be used to show a specific performance level without actually reducing the number of leaking valves. Based on a desire to provide credit for reducing the number of components, while at the same time ensuring that removal credits do not reduce the incentive for low- leak performance in the remaining components, the Committee selected a 67-percent credit for net reductions in the number of valves; i.e., two out of every three valves removed from a process unit may continue to be counted in the total number of valves when calculating leak frequency. e. Unsafe- to-monitor and difficult-to-monitor valves. The Committee agreed that the ''unsafe-to-monitor'' and the ''difficult-to-monitor'' provisions in the existing equipment leak standards are appropriate and should be included in the negotiated standard. Valves that are ''unsafe-to-monitor'' are defined as valves that could expose personnel to imminent hazards from temperature, pressure, or explosive process conditions. Examples of such locations include valves at the top of or located nearby high pressure reactors. Valves that are unsafe-to-monitor cannot be eliminated from new or existing units. The rule, therefore, would exempt valves in unsafe locations from routine monitoring requirements, but would require monitoring as frequently as practicable during safe-to-monitor periods. At existing process units, valves may be located where they can be reached only through extraordinary means. The Committee concluded that the criteria for difficult-to-monitor valves are appropriate and that routine monitoring should not be required for valves that require elevating personnel more than 1.8 meters (6 feet) above any permanent support surface. This means that valves that cannot be safely monitored from a step ladder or portable scaffolding could be classified as inaccessible and exempt from routine monitoring. As in the existing rules, the difficult-to- monitor exemption would be available only for existing units. The Committee agreed that routine monitoring of valves that have the potential to leak should be considered in the design of new units. f. 250 valve exemption from Phase III monthly monitoring. The Committee agreed to require a quarterly LDAR program in Phases I and II and in Phase III to base the monitoring frequency on the performance level achieved. In Phase III, sources that exceed the base performance level have the option of monthly monitoring or participation in a QIP. For small plants, however, there was concern that either monthly monitoring or QIP would present an unreasonable burden. Because small plants may have limited technical and financial resources and the emissions from small plants with a quarterly LDAR program may be relatively low, the Committee established 250 valves per plant site as the level below which quarterly monitoring would be the most frequent monitoring interval required. The Committee defined a small plant as a site with 250 valves, or fewer, based on an industry estimate of representative number of valves at small facilities and the estimated emissions from such facilities. 5. Pump Standard As discussed earlier in ''Development of Framework and Selection of MACT,'' the Committee developed the pump standard using the provisions of the existing standards as a starting point. The recommended standard consists of three phases of requirements of increasing stringency and these are a combination of LDAR programs and performance levels. The Committee also retained the existing provisions for DMS systems and for sealless pumps (e.g., canned pumps). Figure 3 presents the conceptual approach for the pump standard. The rationale for the framework and the selection of the base performance levels was explained earlier in this notice. This section explains the rationale for other provisions in the proposed pump standard. {SEE ILLUSTRATION(S) IN ORIGINAL DOCUMENT} As with valves, the Committee used LDAR programs as the starting point because it is an effective means of detecting and controlling emissions and because these sources are not amenable to application of an emission standard. As discussed previously, the Committee did not establish a standard requiring the use of DMS systems or sealless pump designs because this equipment cannot be used in all SOCMI processes. Moreover, this equipment can fail and leaks can occur. Thus, proper operation and maintenance of DMS systems and sealless pumps requires some monitoring. For these reasons, the Committee developed a standard that requires use of LDAR programs (including QIP's, if necessary) for conventional pumps, or alternatively allows the use of DMS systems or sealless pump designs with periodic monitoring. a. Pump repair intervals. For the reasons discussed in the valve standard, the Committee retained the 5-day first attempt and the 15-day repair intervals of the existing rules in the pump standard. The Committee agreed that, with a minor clarification, these repair intervals were appropriate. The Committee added language to clarify the types of simple field repair that are intended to represent first attempt at repair. This provision was added to reduce uncertainty regarding compliance and to consider the increased maintenance that will be required with leak definitions lower than 10,000 ppm. The Committee also recognized that some pumps are not spared in chemical process units and cannot be repaired without shutting down the process. The delay of repair provisions for pumps in the existing standards were reviewed and the Committee concluded that it was appropriate to include these in the proposed standard. b. Averaging period for calculation of percent leaking. During development of the Phase III base performance level, concerns were expressed by Committee members regarding the inherent variability of percent leaking calculations given the small number of pumps typically present in process units and the inherent variability of the leak measurements. While there were no data available to quantify the month-to- month variability of pump leak frequencies, it was expected from general experience that leak frequencies would be highly variable. It was also judged that leak frequencies based on short time periods of data would be poor predictors of the type of performance that the QIP was intended to remedy. The Committee, therefore, concluded that pump leak frequencies should be averaged over a period of several months to provide an indication of trends in performance. A rolling 6-month average was selected by the Committee as an appropriate measure for indicating upward trends and the need to take additional measures to reduce pump seal failures. c. Calculation of percent leaking pumps. The calculation procedure specified in the pump standard is designed to address additional concerns about the variability of pump leak frequencies and to encourage the use of low leak design pumps and pump seals. Because of the small number of pumps in typical process units, the Committee provided the owner or operator the option of calculating percent leaking on a process unit or plant site basis. This option was provided to allow the necessary flexibility to consider further problems associated with small population statistics and site specific concerns. The Committee also decided to require the owner or operator to designate the basis for calculating percent leaking pumps in the first monitoring period and to use that basis for all subsequent calculations. This was required: (1) To preclude sources from using the most favorable basis for the period and circumventing the intent of the pump standard, and (2) to simplify enforcement. The Committee also agreed that pumps equipped with DMS systems and sealless pumps (e.g., canned or magnetic pumps) could be included in the calculation of percent leaking pumps. For the purpose of this calculation, these pumps are assumed not to leak. The calculation procedure, thus, gives credit for use of inherently low leak designs and is intended to provide an incentive to install these designs. The negotiated standard would exclude pumps that leak shortly after startup of a process unit or startup of an individual pump after maintenance or repair activities on that pump from the calculation of percent leaking for that monitoring period. This exclusion was added because industry members of the Committee reported that preinstallation testing does not necessarily ensure that a pump will not leak when put into service. The industry members also thought that these leaks are not indicative of a long term problem. The Committee concluded that such pumps should be subject to the repair requirement, but that this problem is not indicative of the type of poor performance that QIP is intended to remedy. d. Requirements for low leak design pumps. The Committee retained, with slight modifications, the provisions in the existing standards for pumps with DMS systems and for sealless pumps (e.g., 40 CFR 61.242-2 (d) and (e)). For pumps with DMS systems, the Committee agreed that it was only necessary to prohibit use of light liquid VHAP as barrier fluids. The Committee allowed the use of heavy liquid VHAP as barrier fluids because the operation and monitoring requirements for DMS systems were considered sufficient to minimize any emissions from leaks from the DMS system. The Committee also reevaluated the requirement for annual demonstration of no detectable leaks from sealless pumps in the existing standards (e.g., 40 CFR 61.242-2(e)). The Committee considered the causes of failures of these designs and concluded that weekly visual inspection would achieve equivalent performance and assure that these pumps have no emissions. Consequently, under the negotiated pump standard these sealless design pumps would be subject to a weekly visual inspection work practice standard and would not be subject to an annual performance test requirement to demonstrate no emissions. 6. Quality Improvement Program for Valves and Pumps As discussed earlier in the ''Development of Framework and Selection of MACT'' section of this notice, the Committee recognized that there is uncertainty in the achievability of the proposed performance levels by all SOCMI processes, given that these levels are based on information from a small number of chemicals. Consequently, the Committee developed provisions that would require process units not achieving the base performance levels for valves to implement more frequent monitoring or a QIP designed to ultimately achieve the performance levels and units not achieving the base performance levels for pumps to implement a QIP. A process unit not able to achieve the performance levels is not out of compliance if it undertakes these additional measures. The provisions for quality improvement were included in the framework to allow those owners or operators of process units that do use all elements of MACT but do not achieve the base performance levels the flexibility to develop process unit-specific, cost-effective methods for improving emission performance. The QIP also provides an innovative mechanism for focusing efforts on reducing emissions while avoiding lengthy enforcement actions. Only a few process units are expected to use the valve QIP. It is expected that {pg 62674} in the majority of cases, a systematic program of monitoring individual components and repair of those that leak will achieve the performance levels of the standards. It is recognized, however, that there may be situations where the performance levels cannot be achieved due to specific process and operating conditions. The valve and pump QIP's were designed to provide a mechanism for improving performance in such situations. These QIP's prescribe programs that will require a commitment to quality improvement and will involve many aspects of plant operations (e.g., engineering and maintenance). This section presents the basis for the specific provisions in the valve and pump QIP. a. Valves. The Committee developed two alternative QIP's for valves and agreed to restrict the availability of both of these alternatives. The QIP may be a program that either demonstrates progress in reducing the percent leaking valves or implements a technology review and improvement program. The decision to use either of these alternative QIP's must be made during the first year of Phase III for both new and existing units. Under either alternative QIP, after the process unit has achieved less than 2 percent leaking valves, the owner or operator may elect to continue the QIP. If the owner or operator discontinues the QIP, however, the QIP may not be used in the future if the unit again exceeds 2 percent leaking valves. In such cases, the owner or operator must implement monthly monitoring until the unit has less than 2 percent leaking valves. The availability of this QIP was restricted due to concerns of some Committee members that the QIP might be used to delay improving performance. These Committee members were concerned that, unrestricted, the QIP could provide for extended study and would never result in any improvements in performance. The Committee also agreed to allow facilities to remain indefinitely in the QIP after achieving less than 2 percent leaking valves. This option was provided since it would assure continued good performance and was consistent with the objectives of the QIP. Demonstration of progress quality improvement program. This alternative QIP would allow an owner or operator of a source to design and implement a site specific program to achieve steady progress in lowering the percent leaking valves. Any combination of measures such as increased maintenance frequency or replacement of components may be used, provided it achieves the required reductions in percent leaking valves. Under this QIP, the owner or operator would be required to continue quarterly monitoring and to demonstrate an average 10-percent reduction in the percent leaking valves each quarter, based on a rolling average of 2 quarters of monitoring data. If an owner or operator fails for 2 consecutive averaging periods to demonstrate at least an overall average reduction of 10 percent per quarter, the owner or operator must either implement a technology review and improvement QIP or monthly monitoring. If this QIP is continued after a process unit has less than 2 percent leaking valves, the owner or operator must continue quarterly monitoring, but does not have to continue to demonstrate the 10- percent reduction per quarter. The Committee developed this QIP to provide an ad hoc performance limit for those owners or operators interested in a less formal improvement program than that specified in the technology review and improvement QIP. It was envisioned that this alternative QIP would be useful primarily for those process units with leak frequencies only slightly greater than 2 percent. The Committee judged that an average reduction in percent leaking valves of 10 percent per quarter, calculated as a rolling average of 2 quarters of monitoring data, would provide a reasonable basis for determining whether progress was being made in approaching the 2 percent performance level. Moreover, it was expected that this rate of reduction would achieve the performance levels of the base monitoring program at a time comparable to that expected under the technology review and improvement QIP. To ensure that continual progress is achieved, the Committee required that units failing to show progress, as measured by the reduction in percent leaking valves, either implement monthly monitoring or a technology review and improvement QIP. With either of these more structured alternatives, it was expected that adherence to the program would reduce emissions. Technology review and improvement quality improvement program. This QIP was designed to provide a generic process for identifying appropriate solutions to systematic problems. The Committee developed these provisions considering the fundamentals of problem solving and quality improvement principles. The QIP is structured as three classes of requirements: data collection and analysis, performance trials, and quality assurance and improvement. The principal basis for the provisions in the technology review and improvement QIP was general knowledge of factors affecting equipment leaks and recent experience in leak control at some chemical production units. Key principles used in developing this QIP are: (1) The details of programs (e.g., specific designs or equipment) are best determined by the source for the conditions in the process unit and (2) good performance is the cumulative result of analysis and attention to quality improvement over a period of several years. Specific information that the Committee considered included the reported recent experience of one EO producer in reducing and eliminating leaks from two process units. The experience of CMA's phosgene panel and the process that was used over the past 10 or more years to achieve their present performance were also considered. How the Committee weighed the various factors in arriving at the details of this QIP is described in the following paragraphs. The Committee included a requirement for extensive data collection because these data are essential to understanding the causes of problems and to development of a corrective program. The QIP would require that specific information be recorded on each valve's design, materials of construction, packing, type, service conditions, and screening values. Engineering evaluations would also be required to determine the causes of failure of all valves removed from the process unit due to leaks. These records are required to ensure that the data necessary for evaluation of causes of leaks is available. The technology review and improvement QIP would require analysis of the data to determine if there are specific problem services, operating conditions, valve types, designs, and materials of construction. A second, and equally important, objective of the analysis would be to identify valve designs and operating practices that will operate with less than 2 percent leaking valves under conditions comparable to those in the process problem areas. Such valve designs or technologies are referred to in this QIP as superior performing valve technologies. Because process conditions and causes of valve leakage do vary among process units and the combinations of equipment and procedures that are ''superior'' may differ, superior performing valve technology is not specified exclusively in terms of specific valve types or designs. Rather, superior performance is {pg 62675} defined as including any combination of valve type and process, operating practices, and maintenance practices that achieves the base performance level of the standard. This flexibility was provided because it was recognized that best performing equipment has to be determined in context of the application. In some cases, superior performance may be achievable by use of a different packing material in the valve; in other cases, use of a different valve design or design changes to the process configuration may be required. Such specific decisions are best determined by the source owner or operator. These superior performing valve technologies may be identified through analysis of the process unit (or plant) data, or through review of available literature and the experience of other facilities. Both the process unit (or plant) data and outside experience must be considered in identification of candidate superior performing valve technologies for performance trials. This was specified to ensure that a comprehensive assessment was conducted. For similar reasons, the Committee agreed that the analysis and identification of better performing equipment may be conducted through an inter- or intra-company program and may encompass a single process unit, a company, or group of process units. The Committee decided that the first data analysis must be completed no later than 18 months after the start of Phase III. This schedule was established to balance both the time necessary for understanding the problem(s) and the need to achieve a reasonable rate of progress. The data analysis must be reevaluated each year the process unit is in the QIP and has 2 or more percent leaking valves. This was included to ensure continued improvement. Performance trials would be required where the data analysis for the process unit does not identify any superior performing valve designs or technologies that can be applied to the process unit. Performance trials of candidate equipment will ensure that possible solutions are evaluated and that application of new equipment or operating procedures makes good engineering sense. These trials must be conducted as of an experimental program that identifies all the designs and technologies to be evaluated, the stages of the evaluation, the range of planned test conditions, estimated duration of each stage, and documentation of the conclusions for each test. The purpose of the performance trials provision is to determine if candidate equipment or operation and maintenance procedures can be used in the conditions in the process unit. Because these performance trials are primarily intended to be tests of feasibility of application to the process conditions, the number of valves in the trial program was set at a realistic number for experimental evaluation. The performance trials are not intended to provide an estimate of the leak frequency. The QIP would require performance trials for 1 percent of the valves up to a maximum of 20 valves in a single process unit or 50 valves in a plant site study. The first performance trials shall be completed by 24 months after the start of Phase III, or 6 months after conclusion of the first data analysis. At that time, the owner or operator shall have identified valve designs or technologies that combined with appropriate process, operation, and maintenance practices operate with low emissions. The owner or operator shall continue performance trials until a superior performing technology is identified, the process unit has less than 2 percent leaking valves, or there are no additional technically feasible candidate technologies remaining. As additional information and experience are gained in the QIP, the list of identified low emission performing equipment shall be updated. The owner or operator would be required to prepare a quality assurance plan that is based on the results of the data analysis, engineering evaluations of valves, and the performance trials. The plan must include procedures that will ensure that each replacement valve is a quality-assured valve. These valves must be valve technologies that have been identified as superior emission performance technology for that category of valves, unless no superior performers could be identified. Where this occurs, the valves must be one of the lowest emission technologies identified for the specific application. The quality assurance plan must specify minimum design standards for each category of valves established by the owner or operator, include a written procedure for bench testing of valves for leaks, provide an audit procedure for quality control of purchased equipment, and include procedures to ensure quality of any rebuilt equipment. The quality assurance plan for replacement valves must be implemented as long as the process unit remains in the QIP. These provisions were included to guarantee there would be continued improvement in the emission performance of the process unit. This quality assurance plan is to be established by the start of the third year of Phase III for most plant sites and by the start of the fourth year of Phase III for plant sites that meet the exemption for small facilities. The Committee exempted plant sites that have fewer than 400 valves and that are owned by a company with fewer than 100 employees from the requirement to conduct performance trials. This exemption from performance trials was provided to consider the limited technical resources available for conducting research at smaller companies and to minimize any adverse cost impacts of the standard on small companies. The Committee selected the exemption criteria based on typical valve counts in a small process unit and an industry estimate of the typical number of employees for a small company. These facilities were also allowed an additional year to develop a quality assurance plan and to begin using quality assured valves that will operate with low emissions in the process unit. The Committee provided this additional time to allow these smaller facilities the time needed to obtain the necessary information from vendors, literature, and other sources. The Committee considered, but rejected, provisions that would have required mandatory replacement of some proportion of the valves in the process unit each year. Such a provision was not established for valves for several reasons. Leaking valves can only be repaired on-line a limited number of times before they can no longer be adjusted to achieve lower emissions and must be replaced. Typically, on an industry wide basis, 7 to 10 percent of the valves are reportedly replaced each year for reasons other than leaking. If valves that are replaced (for any reason) are replaced with quality assured valves appropriate to the service conditions, an improvement in emission performance can be achieved at nominal additional cost. The Committee judged that the normal replacement of 7 to 10 percent of the valves each year was a reasonable rate of improvement considering that a mandatory replacement program would cause very high expenditures for negligible benefit. b. Pumps. The Committee developed a technology review and improvement QIP as a means for process units (or plants) to ultimately achieve the performance level of the standard. Only this type of QIP was employed because the Committee believed that the only practical solution for pumps was an engineering analysis to determine the causes of systematic problems (e.g., design, application, etc.). Like the technology review and improvement QIP for valves, this QIP was designed to provide a generic approach to evaluate {pg 62676} the problem, identify solutions, and improve equipment and operations. The Committee also recognized that sufficient time must be provided to identify both the causes of leaks and cost-effective solutions. Thus, the pump QIP specifies similar requirements on the same schedule as the valve QIP. The pump QIP differs from the valve QIP in several respects due to differences between the two standards and between the equipment. This section presents the rationale for the differences between the provisions for the two QIP and the additional factors that were considered in developing the pump QIP. Unlike the valve provisions, the pump QIP would not be an option for selection by the owner or operator and there would be no restrictions on when a process unit (or plant) enters a QIP. An owner or operator must comply with the requirements of the QIP whenever the greater of either 10 percent of the pumps or three pumps in a process unit (or at a plant) leak. A process unit must remain in the QIP as long as the percent leaking pumps is greater than the base performance level. Theoretically, a process unit (or plant) could be subject to the QIP provisions more than one time. In cases where a process unit (or plant) again becomes subject to the QIP, the QIP must be resumed starting at the performance trials stage. Resumption at the performance trials was specified, rather than the stage at which the unit exited the QIP, to ensure that the QIP results in a reasonable rate of progress toward the base performance level. Furthermore, some Committee members thought adequate data and analysis would be available from the earlier QIP. It was also assumed that owners and operators of process units (or plants) with pump leak frequencies near the base performance level would be prepared to conduct performance trials should the leak frequency increase. Several Committee members also thought that resumption at performance trials was appropriate for process units (or plants) that reenter the pump QIP after completion of performance trials. These additional trials were judged necessary since technology (e.g., pump seal materials, design, etc.) may have changed since the previous evaluations and additional features may need investigation. As with the technology review and improvement QIP for valves, the Committee required that a comprehensive data base be developed on pump type, design, materials of construction, service characteristics, and screening values. The engineering evaluation requirement differs in that inspections are required for pumps removed from the process due to leaks and for pump seal designs associated with high failure rates. These inspections are required to ensure that evaluations are of equipment with apparently fundamental design or application problems and are not of equipment at the end of the design life. Both pumps and pump seals were included to ensure that all possible sources of leakage (e.g., split casings on large pumps) are included. As in the valve QIP, the data are analyzed to identify problem areas and services as well as to identify superior performing pump technologies. Again, these technologies are not considered to solely include particular pump or seal designs. Superior emission performance technology may include material or design changes to existing pumps, pump seals, seal support systems, use of multiple mechanical seals, or replacement by a canned pump or a magnetically driven pump. Any combination of equipment and operating practices that achieves the specified level of performance is considered to be superior emission performance technology. The pump performance trial provision sets a lower limit than the valve QIP does on the number of performance trials; for pumps the number of tests is 1 percent of the pumps up to a maximum of 2 for a single process unit and 5 for a plant. The smaller number of trials is specified because pump populations are much smaller than valve populations. Furthermore, with pumps there will be far fewer changes in operation, maintenance, and design parameters that can be evaluated. Again, because the purpose of the trials is to test the feasibility of using the technology and not to predict the leak frequency, a large number of trials is not necessary. The quality assurance plan for pumps would have the same basic objectives and requirements as does the valve program. The minor differences between the pump and valve requirements arise because bench testing of pumps is not a useful predictor of on-line performance and was not required. For pumps in general chemical service, the Committee decided to require replacement of 20 percent per year of the pumps or pump seals with equipment, operations, and maintenance practices identified as superior emission performance technology. As discussed earlier, this superior emission performance technology is not considered to be exclusively represented by specific designs such as canned pumps or specific types of pump seals such as DMS. Rather, it is envisioned as consisting of the combination of pump and seal design combined with appropriate process, operation, and maintenance practices that will achieve the performance level of the proposed standard. The mandatory replacement provision was included for general chemical service pumps to ensure that process units (or plants) (ultimately) achieve demonstrated performance levels. The Committee intended for this provision to serve as an incentive to improve performance. Mandatory replacement provisions were not applied to pumps in food/medical service or to pumps handling polymerizing monomers since in both cases no demonstrated performance level or technology could be identified. No data were available to the Committee that showed the performance level that could be achieved by these services. As explained earlier in discussion of the proposed pump standard, the Committee also concluded that DMS may not be appropriate for use on pumps in these services. Consequently, the Committee agreed that replacement equipment must be quality assured and be one of the lowest emission performance technologies identified. 7. Standard for Connectors in Gas/Vapor and Light Liquid Service As discussed earlier, the Committee developed a standard for connectors that consists of a LDAR program in which the monitoring frequency is based on the percent leaking connectors and a leak is defined as an instrument reading of 500 ppm or higher. This standard would not be phased in and would apply as soon as the rule is effective for a process unit. The Committee judged that phase in of the standard was unnecessary since once connectors are leak tight they should remain leak tight. Figure 4 presents the conceptual approach to the connector standard. In addition to the basic LDAR program, the Committee also established several provisions to address situations unique to connectors. This section explains the rationale for these additional provisions. {SEE ILLUSTRATION(S) IN ORIGINAL DOCUMENT} As with other equipment, the Committee judged that it is not technically or economically feasible to establish an emission standard for connectors. A work practice standard, consisting of a LDAR program and a base performance level, was developed because it provided a practical means of detecting and controlling emissions. a. Connector repair intervals. For the reasons discussed in the valve standard, the Committee applied the 5-day first attempt at repair and the 15-day repair interval of the existing rules to connectors. These repair intervals were considered appropriate for instances where leaks can be reduced by tightening bolts or where the connector can be isolated from the process. The Committee also recognized that most connectors cannot be isolated from the process and cannot be repaired without shutting down the process. In such cases, the Committee agreed that delay of repair until the next process unit shutdown is appropriate and the delay of repair provisions in the existing rules should be allowed for connectors. In some situations, repair of a leaking connector would expose personnel to imminent danger. Therefore, a special provision was added to the connector standard that allows the owner or operator to designate unsafe-to-repair connectors and to delay repair of these connectors until the next process unit shutdown. Connectors which are unsafe-to-repair are those that would expose repair personnel to imminent hazards from temperature, pressure, or explosive process conditions. An example of such a hazard would be connectors on high pressure lines where tightening a bolt could cause failure and result in a major or catastrophic release. b. Nonrepairable connectors. The Committee recognized that nonrepairable connectors may not be completely preventable and that increased monitoring frequency, if triggered by nonrepairable components, would be of little benefit. As with valves, the Committee sought to strike a balance between providing an incentive to limit the number of nonrepairable connectors and avoiding imposition of unproductive costs. The agreed approach allows excluding up to 2 percent of the connectors from the calculation of percent leaking if disturbed connectors are monitored for leaks within 3 months. The Committee allowed more connectors than valves to be excluded from the calculation (i.e., 2 versus 1 percent) because it is more likely that connectors cannot be removed from the process and repair would have to be delayed until the next process unit shutdown. The Committee also allowed owners or operators the option of foregoing followup monitoring of disturbed connectors in exchange for not excluding nonrepairable connectors in the calculation of percent leaking connectors. This option provides flexibility and encourages owners or operators to develop their own program to ensure low leak frequencies for connectors. Exercising this option would mean that a process unit would have to have less than 0.5 percent leaking connectors to monitor less frequently than annually. The Committee also agreed to allow an owner or operator to switch from one option to the other provided the new option begins with annual monitoring. This optional provision was developed because of some Committee members' concerns that the administrative burdens would exceed any benefits provided. c. Calculation of percent leaking. For connectors, the Committee decided that the percent leaking should be calculated for each monitoring period and should not be based on rolling averages of several monitoring periods. This decision was based on the judgment that properly designed, constructed, and installed connectors should operate for an extended time before failure and exhibit a low frequency of random failures. The results of a single monitoring period were, therefore, expected to provide a reliable estimate of the actual leak frequency. d. Credits for removing connectors. For the reasons discussed previously for valves, the Committee elected to provide partial credit in the calculation of percent leaking connectors for connectors permanently removed from the process unit. If a connector is removed from a process unit by welding the connector or by welding the pipe together, the integrity of the weld must be verified to be eligible for removed connector credits. This was required to ensure that credits are only provided for successful welds around the entire circumference. Like the valve credit, this credit is included to provide an incentive for owners of facilities to minimize the number of connectors. e. Unsafe-to-monitor and inaccessible connectors. The Committee added provisions to the connector standard to consider situations where connectors may be unsafe to monitor or inaccessible. Like valves, unsafe-to-monitor connectors are connectors that could expose personnel to imminent hazards from temperature, pressure, or explosive conditions. These connectors must be monitored as frequently as practicable during safe-to-monitor periods. The Committee also exempted inaccessible connectors and glass or glass-lined connectors from the requirement for routine monitoring. Inaccessible connectors are defined as buried; insulated in a manner that prevents access by the monitor probe; obstructed by equipment or piping; or not accessible from a 7.6-meter (25-foot) portable scaffold on the ground and greater than 1.8 meters (6 feet) above a support surface. These different categories of inaccessible connectors were identified as presenting situations where monitoring would be extremely difficult, dangerous, or physically impossible. Specifically, buried and obstructed connectors were exempted because it is not possible to monitor these connectors. Insulated connectors were exempted because removal of the insulation from cold systems could cause leaks due to thermal stress and would be very costly. Removal of insulation from hot systems would create a safety hazard in addition to being costly. Connectors that cannot be reached from a 7.6-meter (25-foot) portable scaffold or from a step ladder would also present a safety hazard to monitoring personnel. Glass and glass-lined connectors were exempted because the potential for on-line repair by tightening bolts is limited due to the possibility of breakage and attendant accidental releases. For both inaccessible and glass or glass-lined connectors, however, if leaks are detected by visual, audible, or other means, the leaking connector shall be repaired no later than 15 days after the leak is detected. f. Provisions for screwed connectors. The connector standard would include a provision that allows an owner or operator the option of exempting existing screwed connectors of 2 inches or smaller nominal diameter from routine monitoring provided these connectors are monitored once after the standard takes effect and after the seal is broken or disturbed. Any leaking connectors shall be repaired. Some Committee members felt that it would be very costly to monitor large numbers of small connectors in some process units and the emission reduction achieved would be small. The Committee agreed to include this provision for several reasons. The general opinion of engineers on the Committee was that screwed connectors once properly threaded and leak tight should stay leak tight, since there are no moving parts and the seal does not rely on gaskets or other materials that can deteriorate. Thus, an initial check and {pg 62679} one-time followup monitoring after opening the connector were judged to provide sufficient assurance that the connectors are leak tight. However, because the available data were insufficient to demonstrate this, new screwed connectors of any size and existing screwed connectors greater than 2 inches nominal diameter were not allowed this option and are treated the same as any other connector. The standard would also require that data on the leak frequency of screwed connectors be reported. The EPA intends to review this information and, based on the findings, will consider whether to amend the standard to require routine monitoring for existing screwed connectors of 2 inches, or smaller, nominal diameter, or to apply the initial check and one- time follow-up to all connectors. 8. Provisions for Other Equipment a. Compressors. As with other equipment, an emission standard was not developed for compressors because application of available measurement methods is not technically or economically feasible. The requirements in the existing standards were judged to be MACT and were used as the basis for the negotiated standard for compressors. The standard requires the use of mechanical seals equipped with a barrier fluid system and controlled degassing vents or enclosure of the compressor seal area and venting of emissions through a closed vent system to a control device. These systems can provide control efficiencies approaching 100 percent. b. Pressure relief devices in gas/vapor service. The negotiated standard, like the existing standards, is based on the use of rupture disk systems or control devices, which are considered to effectively eliminate emissions, when properly installed, maintained, and operated. The existing equipment leak standards establish a ''no detectable emission limit'' of 500 ppm for control techniques that eliminate emissions. The Committee considered redefining the emission limit for pressure relief devices, but agreed that there were insufficient data to redefine the limit below 500 ppm. The general opinion was that a 500 ppm standard would ensure essentially zero emissions and would ensure that an effective control system would not be found to be in violation due to residual VHAP in the vent exit. Therefore, the negotiated rule would require that pressure relief devices be operated with an instrument reading of less than 500 ppm above background, except during pressure relief. The 500 ppm above background limit would not apply to discharges through the relief device during pressure relief, because the function of relief devices is to discharge process fluid, thereby reducing dangerously high pressures within the equipment. Relief devices must operate with an instrument reading of less than 500 ppm above background within 5 days after such a discharge. A definition of ''set pressure'' was added to the standard to clarify that set pressure is where a properly operating pressure relief device begins to open to relieve atypical process system operating pressure. As an alternative to rupture disks and other techniques that will achieve less than 500 ppm above background, owners or operators may vent pressure relief devices to closed vent systems connected to a control device. c. Sampling connection systems. The Committee agreed the closed-purge sampling, closed-loop sampling, and closed vent vacuum systems in the existing rules represent MACT for sampling connection systems. Closed-purge sampling systems eliminate emissions due to purging by either returning the purge material directly to the process or by collecting the purge in a collection system which is not open to the atmosphere for recycle or disposal. Closed-loop sampling systems also eliminate emissions due to purging by returning process fluid to the process through an enclosed system that is not directly vented to the atmosphere. Closed vent vacuum systems capture and transport the purged process fluid to a control device. An emission limit was not specified because measuring mass emissions from sampling systems would require enclosing each system (i.e., bagging), a measurement method which is time-consuming, costly, and impractical. A concentration limit is also not feasible because although the VOC and VHAP control efficiency on a closed- purge or closed-loop sampling system is approximately 100 percent, some VOC and VHAP could be emitted during its transfer to a closed collection device or during its ultimate disposal. Because emission standards cannot feasibly be prescribed for sampling connection systems, several alternative formats were considered and equipment standards found to be most appropriate. The proposed standards for sampling connection systems in the negotiated rule would require the use of closed-purge or closed-loop sampling equipment or a closed vent system. In situ sampling systems would be exempted from these regulations. d. Open-ended valves or lines. Emissions from open- ended valves or lines can be eliminated, except when the line is used for draining, venting, or sampling operations, by enclosing the open end by a cap, plug, or a second valve. The control efficiency associated with these techniques is approximately 100 percent and is considered to be MACT. As with other items of equipment, establishing an emission limit is neither effective nor technologically feasible. Therefore, the Committee based the standard on the combination of equipment and operational requirements in the existing rules. The negotiated rule would require the use of a cap, plug, blind flange, or a second valve or other equipment to close the open-ended valve or line. To ensure the proper operation of the equipment, open-ended lines are also covered by operational standards. If a second valve is used to close the open end, the proposed standards would require the upstream valve to be closed first. After the upstream valve is completely closed, the downstream valve would be closed. This operational requirement is necessary in order to prevent trapping process fluid between the two valves, which could result in a situation equivalent to the uncontrolled open-ended line. e. Product accumulator vessels. The technique for controlling product accumulator vessels is to connect the vessel to a closed vent system and control device. Anything in compliance with the requirements for closed vent systems and control devices is acceptable. f. Control devices. Control devices would be used to dispose of VHAP captured in closed vent systems from barrier fluid degassing systems and enclosed pump and compressor seal areas. In all cases, these control devices would receive streams with low and intermittent flow rates. These control devices would in some cases be designed to dispose of organic streams from other sources in the plant, so that the VHAP streams may contribute a very small percentage of the total loading on the control device. Because it would be technologically and economically impractical to measure very low-flow streams and differentiate these streams from others, an emission standard was not proposed for these control devices. Design requirements for control devices were considered to ensure that appropriate emission reductions would be achieved from control devices used in conjunction with closed vent systems. Enclosed combustion sources, flares, and vapor recovery systems were considered as control devices for the closed vent system. The proposed {pg 62680} standard requires that these control devices reduce organic emissions by 95 percent or meet minimum design requirements. The minimum design requirements specified for enclosed combustion devices is to provide a minimum residence time of 0.5 seconds at a minimum temperature of 760 degrees C (1400 degrees F). Flares used as control devices to comply with the negotiated standard shall comply with the requirements of 40 CFR 60.18. Vapor recovery systems would also be allowed as control devices for VHAP from closed vent systems. A control efficiency of at least 95 percent was chosen as the design requirement because it is the highest control efficiency that has been demonstrated consistently for vapor recovery systems such as carbon adsorption or condensation units. g. Agitators. Agitators were not evaluated during regulatory development of the existing regulations and are not presently regulated by any standard. Therefore, very limited information is available on this equipment. The Committee determined that agitators should be included in the negotiated standard for several reasons. First, a limited amount of screening data indicate that agitators may be a significant source of emissions. Second, agitators are technologically similar to pumps, and, like pumps, emissions can be controlled using seal technology. However, agitators have longer and larger diameter shafts than pumps and produce greater tangential shaft loadings. The performance of pump seal systems, therefore, cannot be used to estimate agitator seal performance. Considering this and the potential for large leaks, the Committee agreed to require a monthly LDAR program only and to define a leak as a concentration of 10,000 ppm or higher. This program will require replacement of agitator seals with significant leaks and will encourage development of effective bearing and seal systems. h. Instrumentation systems. The Committee created this equipment category to address industry members concerns with application of LDAR programs to equipment used in monitoring process operations. Small diameter tubing and other components are used to convey process samples to analyzers to determine chemical composition and to instruments such as pressure and flow transducers. Instrumentation systems typically contain valves 0.5 inch in nominal diameter or less and connectors 0.75 inch nominal diameter or less and are located in a confined area such that monitoring of individual components is generally infeasible. Because these systems provide critical process operating information, they are subject to frequent surveillance and maintenance to assure the reliability of measurements. Leaking equipment in these systems would be readily detected by changes in temperature, pressure, flow rates, or by observation. Additionally, it is common practice in the industry after maintenance or repair to verify the integrity of these systems by soap bubble testing or pressure checks. Therefore, a routine LDAR program would be redundant and would provide no benefit. The Committee concluded for these reasons that it was appropriate to develop alternative provisions for these systems. The Committee judged that it was appropriate to require repair of leaking instrumentation systems in a timely manner and agreed to treat these systems in a manner similar to equipment in heavy liquid service. In addition, the Committee agreed that monitoring of individual components in instrumentation systems by Method 21 is not necessary if the leak is repaired and the repair is verified. The verification may be by soap testing, a pressure check, or any other visible, audible, olfactory, or other means. i. Miscellaneous. As in the existing equipment leak standards, pumps, valves, connectors, and agitators in heavy liquid service (HAP fluids with vapor pressures less than 0.3 kPa at 20 degrees C), and pressure relief devices in light liquid or heavy liquid service would be excluded from the routine monitoring and inspection requirements. However, if leaks are detected from these sources, the same allowable repair interval that applies to pumps, valves, connectors, and compressors would apply. These sources were excluded from routine monitoring on the basis that they contribute only a very small portion of overall emissions from a process unit and including them in the monitoring and equipment requirements was not considered reasonable. 9. Alternative Standards Under the provisions of section 112(h) of the Act, if the Administrator establishes work practice, equipment, design, or operational standards, then the Administrator must allow use of alternative means of emission reductions if an owner or operator can demonstrate emission reductions equivalent to that achieved by the standards. Generally, alternative means of emission reduction are based on specific circumstances at individual plant sites and must be handled individually. During the course of the Committee's deliberations, however, two situations were identified where development of generic alternative standards was appropriate. These situations were batch processes and enclosed process units. Consequently, the negotiated rule includes alternative standards for batch operations and for process units located in enclosed buildings as well as general provisions for demonstration of equivalency. a. Batch processes. At one of the early meetings of the Committee, the problems associated with application of LDAR programs to batch processes were raised. At continuous processes, equipment can be monitored rapidly, one after another on an established route, without consideration of the process schedule because there is always process fluid in the lines. For several reasons, rapid monitoring of one component after another is not possible at some batch processes. This is particularly true for processes such as those in the pharmaceutical industry where equipment may not be dedicated to a particular process and where different components are used in different batch operations to produce different products. In any particular configuration, process fluid may flow through a series of valves, connectors, and a pump into a reactor. Several hours later, the process fluid would exit the reactor through other equipment for subsequent processing steps and storage. Also, the equipment used will contain process fluids for only a brief period during the batch (e.g., a few minutes per batch). In this case, the monitoring team would have to wait for an extended time between screening equipment on the upstream and downstream side of the reactor. Equipment not used in a