National Emission Standards for Hazardous Air Pollutants for Major Sources: Industrial, Commercial, and Institutional Boilers and Process Heaters

ENVIRONMENTAL PROTECTION AGENCY

40 CFR Part 63

[EPA-HQ-OAR-2002-0058; FRL-9272-8]

RIN 2060-AQ25

National Emission Standards for Hazardous Air Pollutants for Major Sources: Industrial, Commercial, and Institutional Boilers and Process Heaters

AGENCY:

Environmental Protection Agency (EPA).

ACTION:

Final rule.

SUMMARY:

On September 13, 2004, under authority of section 112 of the Clean Air Act, EPA promulgated national emission standards for hazardous air pollutants for new and existing industrial/commercial/institutional boilers and process heaters. On June 19, 2007, the United States Court of Appeals for the District of Columbia Circuit vacated and remanded the standards.

In response to the Court's vacatur and remand, EPA is, in this action, establishing emission standards that will require industrial/commercial/institutional boilers and process heaters located at major sources to meet hazardous air pollutants standards reflecting the application of the maximum achievable control technology. This rule protects air quality and promotes public health by reducing emissions of the hazardous air pollutants listed in section 112(b)(1) of the Clean Air Act.

DATES:

This final rule is effective on May 20, 2011. The incorporation by reference of certain publications listed in this rule is approved by the Director of the Federal Register as of May 20, 2011.

ADDRESSES:

EPA established a single docket under Docket ID No. EPA-HQ-OAR-2002-0058 for this action. All documents in the docket are listed on the

http://www.regulations.gov

Web site. Although listed in the index, some information is not publicly available,

e.g.,

confidential business information or other information whose disclosure is restricted by statute. Certain other material, such as copyrighted material, is not placed on the Internet and will be publicly available only in hard copy form. Publicly available docket materials are available either electronically through

http://www.regulations.gov

or in hard copy at EPA's Docket Center, Public Reading Room, EPA West Building, Room 3334, 1301 Constitution Avenue, NW., Washington, DC 20004. This Docket Facility is open from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding legal holidays. The telephone number for the Public Reading Room is (202) 566-1744, and the telephone number for the Air Docket is (202) 566-1741.

FOR FURTHER INFORMATION CONTACT:

Mr. Brian Shrager, Energy Strategies Group, Sector Policies and Programs Division, (D243-01), Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711; Telephone number: (919) 541-7689; Fax number (919) 541-5450; E-mail address:

shrager.brian@epa.gov.

SUPPLEMENTARY INFORMATION:

The information presented in this preamble is organized as follows:

I. General Information

A. Does this action apply to me?

B. Where can I get a copy of this document?

C. Judicial Review

II. Background Information

A. What is the statutory authority for this final rule?

B. EPA's Response to the Vacatur

C. What is the relationship between this final rule and other combustion rules?

D. What are the health effects of pollutants emitted from industrial/commercial/institutional boilers and process heaters?

E. What are the costs and benefits of this final rule?

III. Summary of this Final Rule

A. What is the source category regulated by this final rule?

B. What is the affected source?

C. What are the pollutants regulated by this final rule?

D. What emission limits and work practice standards must I meet?

E. What are the requirements during periods of startup, shutdown, and malfunction?

F. What are the testing and initial compliance requirements?

G. What are the continuous compliance requirements?

H. What are the notification, recordkeeping and reporting requirements?

I. Submission of Emissions Test Results to EPA

IV. Summary of Significant Changes Since Proposal

A. Applicability

B. Subcategories

C. Emission Limits

D. Work Practices

E. Energy Assessment Requirements

F. Requirements During Startup, Shutdown, and Malfunction

G. Testing and Initial Compliance

H. Continuous Compliance

I. Notification, Recordkeeping and Reporting

J. Technical/Editorial Corrections

K. Other

V. Major Source Public Comments and Responses

A. MACT Floor Analysis

B. Beyond the Floor

C. Rationale for Subcategories

D. Work Practices

E. New Data/Technical Corrections to Old Data

F. Startup, Shutdown, and Malfunction Requirements

G. Health Based Compliance Alternatives

H. Biased Data Collection From Phase II Information Collection Request Testing

I. Issues Related to Carbon Monoxide Emission Limits

J. Cost Issues

K. Non-Hazardous Secondary Materials

VI. Impacts of This Final Rule

A. What are the air impacts?

B. What are the water and solid waste impacts?

C. What are the energy impacts?

D. What are the cost impacts?

E. What are the economic impacts?

F. What are the benefits of this final rule?

G. What are the secondary air impacts?

VII. Relationship of Final Action to Section 112(c)(6) of the Clean Air Act

VIII. Statutory and Executive Order Reviews

A. Executive Orders 12866 and 13563: Regulatory Planning and Review

B. Paperwork Reduction Act

C. Regulatory Flexibility Act as Amended by the Small Business Regulatory Enforcement Fairness Act of 1996, 5 U.S.C. 601

et seq.

D. Unfunded Mandates Reform Act of 1995

E. Executive Order 13132: Federalism

F. Executive Order 13175: Consultation and Coordination With Indian Tribal Governments

G. Executive Order 13045: Protection of Children From Environmental Health Risks and Safety Risks

H. Executive Order 13211: Actions Concerning Regulations That Significantly Affect Energy Supply, Distribution, or Use

I. National Technology Transfer and Advancement Act

J. Executive Order 12898: Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations

K. Congressional Review Act

I. General Information

A. Does this action apply to me?

The regulated categories and entities potentially affected by the final standards include:

Category

NAICS code

1

Examples of potentially regulated entities

Any industry using a boiler or process heater as defined in the final rule

211

Extractors of crude petroleum and natural gas.

321

Manufacturers of lumber and wood products.

322

Pulp and paper mills.

325

Chemical manufacturers.

324

Petroleum refineries, and manufacturers of coal products.

316, 326, 339

Manufacturers of rubber and miscellaneous plastic products.

331

Steel works, blast furnaces.

332

Electroplating, plating, polishing, anodizing, and coloring.

336

Manufacturers of motor vehicle parts and accessories.

221

Electric, gas, and sanitary services.

622

Health services.

611

Educational services.

1

North American Industry Classification System.

This table is not intended to be exhaustive, but rather provides a guide for readers regarding entities likely to be affected by this action. To determine whether your facility, company, business, organization, etc., would be regulated by this action, you should examine the applicability criteria in 40 CFR 63.7485 of subpart DDDDD (National Emission Standards for Hazardous Air Pollutants (NESHAP) for Industrial, Commercial, and Institution Boilers and Process Heaters). If you have any questions regarding the applicability of this action to a particular entity, consult either the air permitting authority for the entity or your EPA regional representative as listed in 40 CFR 63.13 of subpart A (General Provisions).

B. Where can I get a copy of this document?

In addition to being available in the docket, an electronic copy of this action will also be available on the Worldwide Web (WWW) through the Technology Transfer Network (TTN). Following signature, a copy of the action will be posted on the TTN's policy and guidance page for newly proposed or promulgated rules at the following address:

http://www.epa.gov/ttn/oarpg/.

The TTN provides information and technology exchange in various areas of air pollution control.

C. Judicial Review

Under the Clean Air Act (CAA) section 307(b)(1), judicial review of this final rule is available only by filing a petition for review in the U.S. Court of Appeals for the District of Columbia Circuit by May 20, 2011. Under CAA section 307(d)(7)(B), only an objection to this final rule that was raised with reasonable specificity during the period for public comment can be raised during judicial review. This section also provides a mechanism for us to convene a proceeding for reconsideration, “[i]f the person raising an objection can demonstrate to EPA that it was impracticable to raise such objection within [the period for public comment] or if the grounds for such objection arose after the period for public comment (but within the time specified for judicial review) and if such objection is of central relevance to the outcome of this rule.” Any person seeking to make such a demonstration to us should submit a Petition for Reconsideration to the Office of the Administrator, Environmental Protection Agency, Room 3000, Ariel Rios Building, 1200 Pennsylvania Ave., NW., Washington, DC 20004, with a copy to the person listed in the preceding

FOR FURTHER INFORMATION CONTACT

section, and the Associate General Counsel for the Air and Radiation Law Office, Office of General Counsel (Mail Code 2344A), Environmental Protection Agency, 1200 Pennsylvania Ave., NW., Washington, DC 20004. Note, under CAA section 307(b)(2), the requirements established by this final rule may not be challenged separately in any civil or criminal proceedings brought by EPA to enforce these requirements.

II. Background Information

A. What is the statutory authority for this final rule?

Section 112(d) of the CAA requires EPA to set emissions standards for hazardous air pollutants (HAP) emitted by major stationary sources based on the performance of the maximum achievable control technology (MACT). The MACT standards for existing sources must be at least as stringent as the average emissions limitation achieved by the best performing 12 percent of existing sources (for which the Administrator has emissions information) or the best performing 5 sources for source categories with less than 30 sources (CAA section 112(d)(3)(A) and (B)). This level of minimum stringency is called the MACT floor. For new sources, MACT standards must be at least as stringent as the control level achieved in practice by the best controlled similar source (CAA section 112(d)(3)). EPA also must consider more stringent “beyond-the-floor” control options. When considering beyond-the-floor options, EPA must consider not only the maximum degree of reduction in emissions of HAP, but must take into account costs, energy, and nonair environmental impacts when doing so.

With respect to alkylated lead compounds; polycyclic organic matter (POM); hexachlorobenzene; mercury (Hg); polychlorinated biphenyls; 2,3,7,8-tetrachlorodibenzofurans; and 2,3,7,8-tetrachlorodibenzo-p-dioxin, the CAA section 112(c)(6) requires EPA to list categories and subcategories of sources assuring that sources accounting for not less than 90 percent of the aggregate emissions of each such pollutant are subject to standards under subsection 112(d)(2) or (d)(4). Standards established under CAA section 112(d)(2) must reflect the performance of MACT. “Industrial Coal Combustion,” “Industrial Oil Combustion,” “Industrial Wood/Wood Residue Combustion,” “Commercial Coal Combustion,” “Commercial Oil Combustion,” and “Commercial Wood/Wood Residue Combustion” are listed as source categories for regulation pursuant to CAA section 112(c)(6) due to emissions of POM and Hg (63 FR 17838, 17848, April 10, 1998). In the documentation for the 112(c)(6) listing, the commercial fuel combustion categories included institutional fuel combustion (“1990 Emissions Inventory of Section 112(c)(6) Pollutants, Final Report,” April 1998).

CAA section 129(a)(1)(A) requires EPA to establish specific performance standards, including emission limitations, for “solid waste incineration units” generally, and, in particular, for “solid waste incineration units combusting commercial or industrial waste” (section 129(a)(1)(D)). Section 129 defines “solid waste incineration unit” as “a distinct operating unit of any facility which combusts any solid waste material from commercial or industrial establishments or the general public.”

Section 129(g)(1). Section 129 also provides that “solid waste” shall have the meaning established by EPA pursuant to its authority under the Resource Conservation and Recovery Act. Section 129(g)(6).

In

Natural Resources Defense Council

v.

EPA,

489 F. 3d 1250, 1257-61 (D.C. Cir. 2007), the court vacated the Commercial and Industrial Solid Waste Incineration (CISWI) Definitions Rule, 70 FR 55568 (September 22, 2005), which EPA issued pursuant to CAA section 129(a)(1)(D). In that rule, EPA defined the term “commercial or industrial solid waste incineration unit” to mean a combustion unit that combusts “commercial or industrial waste.” The CISWI definitions rule defined “commercial or industrial waste” to mean waste combusted at a unit that does not recover thermal energy from the combustion for a useful purpose. Under these definitions, only those units that combusted commercial or industrial waste and were not designed to, or did not operate to, recover thermal energy from the combustion would be subject to section 129 standards. The District of Columbia Circuit (DC Circuit) rejected the definitions contained in the CISWI Definitions Rule and interpreted the term “solid waste incineration unit” in CAA section 129(g)(1) “to unambiguously include among the incineration units subject to its standards any facility that combusts any commercial or industrial solid waste material at all—subject to the four statutory exceptions identified in [CAA section 129(g)(1).]”

NRDC

v.

EPA,

489 F.3d 1250, 1257-58. A more detailed discussion of this decision, as well as other court decisions relevant to today's action, can be found in the June 4, 2010, preamble to the proposed rule. See 75 FR 32009.

CAA section 129 covers any facility that combusts any solid waste; CAA section 129(g)(6) directs the Agency to the Resource Conservation and Recovery Act (RCRA) in terms of the definition of solid waste. In this

Federal Register

, EPA is issuing a definition of solid waste for purposes of Subtitle D of RCRA. If a unit combusts solid waste, it is subject to CAA section 129 of the Act, unless it falls within one of the four specified exceptions in CAA section 129(g).

The solid waste definitional rulemaking under RCRA is being finalized in a parallel action and is relevant to this proceeding because some industrial, commercial, or institutional boilers and process heaters combust secondary materials as alternative fuels. If industrial, commercial, or institutional boilers or process heaters combust secondary materials that are solid waste under the final definitional rule, those units would be subject to emission standards issued under section 129. The units subject to this final rule include those industrial, commercial, or institutional boilers and process heaters that do not combust solid waste, as well as boilers and process heaters that combust solid waste but qualify for one of the statutory exclusions contained in section 129(g)(1). EPA recognizes that it has imperfect information on the exact nature of the secondary materials which boilers and process heaters combust, including, for example, how much processing of such materials occurs, if any. We used the information currently available to the Agency to determine which units combust solid waste materials and, therefore, are subject to CAA section 129, and which units do not combust solid waste (or qualify for an exclusion from section 129) and, therefore, are subject to CAA section 112.

B. EPA's Response to the Vacatur

A description of EPA's information collection efforts and a description of the development of EPA's proposed response to the

NRDC

v.

EPA

mandate is contained in the preamble to the proposed rule.

See

75 FR 32010-32011. After consideration of public comments on the proposed rule, we have made appropriate revisions to the final rule, and a description of the major changes is provided in this preamble. The changes reflect EPA's consideration of public comments and the consideration of additional information and emissions data provided through the public comment process. The changes also reflect adjustments to the definition of non-hazardous solid waste as set forth in a parallel final action. That final rule contains some revisions to the definition of non-hazardous solid waste proposed by EPA in June 2010. Accordingly, the population of combustion units subject to CAA section 129 (because they combust solid waste) and the population of boilers and process heaters subject to CAA section 112 (because they do not combust solid waste) were established considering the final solid waste definition issued today. We used the updated inventories and all available data, as appropriate, to develop the final standards for boilers and process heaters under CAA section 112 and, in a separate parallel action, the final standards for commercial and industrial solid waste incineration units covered by CAA section 129. We used all of the appropriate information available to the Administrator to calculate the MACT floors, set emission limits, and evaluate the emission impacts of various regulatory options for these final rulemakings.

C. What is the relationship between this final rule and other combustion rules?

This final rule addresses the combustion of non-solid waste materials in boilers and process heaters located at major sources of HAP. If an owner or operator of an affected source subject to these standards were to start combusting a solid waste (as defined by the Administrator under RCRA), the affected source would cease to be subject to this action and would instead be subject to regulation under CAA section 129. A rulemaking under CAA section 129 is being finalized in a parallel action and is relevant to this action because it would apply to boilers and process heaters that combust any solid waste and are located at a major source. In this final boiler rulemaking, EPA is providing specific language to ensure clarity regarding the necessary steps that must be followed for combustion units that begin combusting non-hazardous solid waste materials and become subject to section 129 standards instead of section 112 standards or combustion units that discontinue combustion of non-hazardous solid waste materials and become subject to section 112 standards instead of section 129 standards.

In addition to combustion units that may switch between the section 112 boiler standards and the section 129 incinerator standards, there are certain instances where boilers and process heaters are already regulated under other MACT standards. In such cases, the boilers and process heaters that are already subject to another MACT standard are not subject to the boiler standards.

In 1986, EPA codified new source performance standards (NSPS) for industrial boilers (40 CFR part 60, subparts Db and Dc) and portions of those standards were revised in 1999 and 2006. The NSPS regulates emissions of particulate matter (PM), sulfur dioxide (SO

2

), and nitrogen oxide (NO

X

) from boilers constructed after June 19, 1984. Sources subject to the NSPS will also be subject to the final CAA section 112(d) standards for boilers and process heaters because the section 112(d) standards regulate HAP emissions while the NSPS do not. However, in developing this final rule, we considered the monitoring requirements, testing requirements, and recordkeeping requirements of the NSPS to avoid duplicating requirements.

D. What are the health effects of pollutants emitted from industrial/commercial/institutional boilers and process heaters?

This final rule protects air quality and promotes the public health by reducing emissions of some of the HAP listed in CAA section 112(b)(1). As noted above, emissions data collected during development of the rule show that hydrogen chloride (HCl) emissions represent the predominant HAP emitted by industrial, commercial, and institutional (ICI) boilers, accounting for 69 percent of the total HAP emissions.

1

ICI boilers and process heaters also emit lesser amounts of hydrogen fluoride, accounting for about 21 percent of total HAP emissions, and metals (arsenic, beryllium, cadmium, chromium, cobalt, lead, manganese (Mn), Hg, nickel, and selenium) accounting for about 6 percent of total HAP emissions. Organic HAP (formaldehyde, POM, acetaldehyde, benzene) account for about 4 percent of total HAP emissions. Exposure to these HAP, depending on exposure duration and levels of exposures, can be associated with a variety of adverse health effects. These adverse health effects may include, for example, irritation of the lung, skin, and mucus membranes, effects on the central nervous system, damage to the kidneys, and alimentary effects such as nausea and vomiting. We have classified two of the HAP as human carcinogens (arsenic and chromium VI) and four as probable human carcinogens (cadmium, lead, dioxins/furans, and nickel). We do not know the extent to which the adverse health effects described above occur in the populations surrounding these facilities. However, to the extent the adverse effects do occur, this final rule would reduce emissions and subsequent exposures.

1

See

Memorandum “Methodology for Estimating Impacts from Industrial, Commercial, Institutional Boilers and Process Heaters at Major Sources of Hazardous Air Pollutant Emissions” located in the docket.

E. What are the costs and benefits of this final rule?

EPA estimated the costs and benefits associated with the final rule, and the results are shown in the following table. For more information on the costs and benefits for this rule,

see

the Regulatory Impact Analysis (RIA).

Summary of the Monetized Benefits, Social Costs, and Net Benefits for the Boiler MACT in 2014

[Millions of 2008$]

3% Discount rate

7% Discount rate

Selected

Total Monetized Benefits

2

$22,000 to $54,000

$20,000 to $49,000

Total Social Costs

3

$1,500

$1,500

Net Benefits

$20,500 to $52,500

$18,500 to $47,500

Non-monetized Benefits

112,000 tons of CO, 30,000 tons of HCl, 820 tons of HF, 2,800 pounds of Hg

2,700 tons of other metals, 23 grams of dioxins/furans (TEQ), Health effects from SO

2

exposure, Ecosystem effects, Visibility impairment

Alternative

Total Monetized Benefits

2

$18,000 to $43,000

$16,000 to $39,000

Total Social Costs

3

$1,900

$1,900

Net Benefits

$16,100 to $41,100

$14,100 to $37,100

Non-monetized Benefits

112,000 tons of CO, 22,000 tons of HCl, 620 tons of HF, 2,400 pounds of Hg, 2,600 tons of other metals, 23 grams of dioxins/furans (TEQ), Health effects from SO

2

exposure, Ecosystem effects, Visibility impairment

1

All estimates are for the implementation year (2014), and are rounded to two significant figures. These results include units anticipated to come online and the lowest cost disposal assumption.

2

The total monetized benefits reflect the human health benefits associated with reducing exposure to PM

2.5

through reductions of directly emitted PM

2.5

and PM

2.5

precursors such as SO

2

, as well as reducing exposure to ozone through reductions of VOCs. It is important to note that the monetized benefits include many but not all health effects associated with PM

2.5

exposure. Benefits are shown as a range from Pope

et al.

(2002) to Laden

et al.

(2006). These models assume that all fine particles, regardless of their chemical composition, are equally potent in causing premature mortality because there is no clear scientific evidence that would support the development of differential effects estimates by particle type. These estimates include energy disbenefits valued at $23 million for the selected option and $35 million for the alternative option. Ozone benefits are valued at $3.6 to $15 million for both options.

3

The methodology used to estimate social costs for one year in the multimarket model using surplus changes results in the same social costs for both discount rates.

III. Summary of This Final Rule

This section summarizes the requirements of this action. Section IV below provides a summary of the significant changes to this final rule following proposal.

A. What is the source category regulated by this final rule?

ICI boilers and process heaters located at major sources of HAP are regulated by this final rule. Waste heat boilers and boilers and process heaters that combust solid waste, except for specific exceptions to the definition of a solid waste incineration unit outlined in section 129(g)(1), are not subject to this final rule.

B. What is the affected source?

This final rule affects industrial boilers, institutional boilers, commercial boilers, and process heaters. A process heater is defined as a unit in which the combustion gases do not directly come into contact with process material or gases in the combustion chamber (

e.g.,

indirect fired). A boiler is defined as an enclosed device using controlled flame combustion and having the primary purpose of recovering thermal energy in the form of steam or hot water.

C. What are the pollutants regulated by this final rule?

This final rule regulates HCl (as a surrogate for acid gas HAP), PM (as a surrogate for non-Hg HAP metals), carbon monoxide (CO) (as a surrogate for non-dioxin/furan organic HAP), Hg, and dioxin/furan emissions from boilers and process heaters.

D. What emission limits and work practice standards must I meet?

You must meet the emission limits presented in Table 1 of this preamble. This final rule includes 15 subcategories. Emission limits are established for new and existing sources for each of the subcategories, which are based on unit design.

Metallic HAP (regulated using PM as a surrogate), HCl, and Hg are “fuel-based pollutants” that are a direct result of contaminants in the fuels that are combusted. For those pollutants, if your new or existing unit combusts at least 10 percent solid fuel on an annual basis, your unit is subject to emission limits that are based on data from all of the solid fuel-fired combustor designs. If your new or existing unit combusts at least 10 percent liquid fuel and less than 10 percent solid fuel and your facility is located in the continental United States, your unit is subject to the liquid fuel emission limits for the fuel-based pollutants. If your facility is located outside of North America (referred to as a non-continental unit for the remainder of the preamble and in this final rule) and your new or existing unit combusts at least 10 percent liquid fuel and less than 10 percent solid fuel, your unit is subject to the non-continental liquid fuel emission limits for the fuel-based pollutants. Finally, for the fuel-based pollutants, if your unit combusts gaseous fuel that does not qualify as a “Gas 1” fuel, your unit is subject to the Gas 2 emission limits in Table 1 of this preamble. If your unit is a Gas 1 unit (that is, it combusts only natural gas, refinery gas, or equivalent fuel (other gas that qualifies as Gas 1 fuel)), with limited exceptions for gas curtailments and emergencies, your unit is subject to a work practice standard that requires an annual tune-up in lieu of emission limits.

For the combustion-based pollutants, CO (used as a surrogate for non-dioxin organic HAP) and dioxin/furan, your unit is subject to the emission limits for the design-based subcategories shown in Table 1 of this preamble. If your new or existing boiler or process heater burns at least 10 percent biomass on an annual average heat input

2

basis, the unit is in one of the biomass subcategories. If your new or existing boiler or process heater burns at least 10 percent coal, on an annual average heat input basis, and less than 10 percent biomass, on an annual average heat input basis, the unit is in one of the coal subcategories. If your facility is located in the continental United States and your new or existing boiler or process heater burns at least 10 percent liquid fuel (such as distillate oil, residual oil) and less than 10 percent coal and less than 10 percent biomass, on an annual average heat input basis, your unit is in the liquid subcategory. If your non-continental new or existing boiler or process heater burns at least 10 percent liquid fuel (such as distillate oil, residual oil) and less than 10 percent coal and less than 10 percent biomass, on an annual average heat input basis, your unit is in the non-continental liquid subcategory. Finally, for the combustion-based pollutants, if your unit combusts gaseous fuel that does not qualify as a “Gas 1” fuel, your unit is subject to the Gas 2 emission limits in Table 1. If your unit combusts only natural gas, refinery gas, or equivalent fuel (other gas that qualifies as Gas 1 fuel), with limited exceptions for gas curtailment and emergencies, your unit is subject to a work practice standard that requires an annual tune-up in lieu of emission limits.

2

Heat input means heat derived from combustion of fuel in a boiler or process heater and does not include the heat derived from preheated combustion air, recirculated flue gases or exhaust from other sources (such as stationary gas turbines, internal combustion engines, and kilns).

Table 1—Emission Limits for Boilers and Process Heaters

[Pounds per million British thermal units]

Subcategory

Particulate

matter

(PM)

Hydrogen

chloride

(HCl)

Mercury

(Hg)

Carbon

monoxide

(CO)

(ppm @3% oxygen)

Dioxin/furan

(TEQ)

(ng/dscm)

Existing—Coal Stoker

0.039

0.035

0.0000046

270

0.003

Existing—Coal Fluidized Bed

0.039

0.035

0.0000046

82

0.002

Existing—Pulverized Coal

0.039

0.035

0.0000046

160

0.004

Existing—Biomass Stoker/other

0.039

0.035

0.0000046

490

0.005

Existing—Biomass Fluidized Bed

0.039

0.035

0.0000046

430

0.02

Existing—Biomass Dutch Oven/Suspension Burner

0.039

0.035

0.0000046

470

0.2

Existing—Biomass Fuel Cells

0.039

0.035

0.0000046

690

4

Existing—Biomass Suspension/Grate

0.039

0.035

0.0000046

3,500

0.2

Existing—Liquid

0.0075

0.00033

0.0000035

10

4

Existing—Gas 2 (Other Process Gases)

0.043

0.0017

0.000013

9.0

0.08

Existing—non-continental liquid

0.0075

0.00033

0.00000078

160

4

New—Coal Stoker

0.0011

0.0022

0.0000035

6

0.003

New—Coal Fluidized Bed

0.0011

0.0022

0.0000035

18

0.002

New—Pulverized Coal

0.0011

0.0022

0.0000035

12

0.003

New—Biomass Stoker

0.0011

0.0022

0.0000035

160

0.005

New—Biomass Fluidized Bed

0.0011

0.0022

0.0000035

260

0.02

New—Biomass Dutch Oven/Suspension Burner

0.0011

0.0022

0.0000035

470

0.2

New—Biomass Fuel Cells

0.0011

0.0022

0.0000035

470

0.003

New—Biomass Suspension/Grate

0.0011

0.0022

0.0000035

1,500

0.2

New—Liquid

0.0013

0.00033

0.00000021

3

0.002

New—Gas 2 (Other Process Gases)

0.0067

0.0017

0.0000079

3

0.08

New—non-continental liquid

0.0013

0.00033

0.00000078

51

0.002

The emission limits in Table 1 apply only to new and existing boilers and process heaters that have a designed heat input capacity of 10 million British thermal units per hour (MMBtu/hr) or greater. We also are providing optional output-based standards in this final rule. Pursuant to CAA section 112(h), we are requiring a work practice standard for four particular classes of boilers and process heaters: New and existing units that have a designed heat input capacity of less than 10 MMBtu/hr, and new and existing units in the Gas 1 (natural gas/refinery gas) subcategory and in the metal process furnaces subcategory. The work practice standard for these boilers and process heaters requires the implementation of a tune-up program as described in section III.F of this preamble.

We are also finalizing a beyond-the-floor standard for all existing major source facilities having affected boilers or process heaters that would require the performance of a one-time energy assessment, as described in section III.F of this preamble, by qualified personnel, on the affected boilers and facility to identify any cost-effective energy conservation measures.

E. What are the requirements during periods of startup, shutdown, and malfunction?

Consistent with

Sierra Club

v.

EPA

, EPA has established standards in this final rule that apply at all times. In establishing the standards in this final rule, EPA has taken into account startup and shutdown periods and, for the reasons explained below, has established different standards for those periods.

EPA has revised this final rule to require sources to meet a work practice standard, which requires following the manufacturer's recommended procedures for minimizing periods of startup and shutdown, for all subcategories of new and existing boilers and process heaters (that would otherwise be subject to numeric emission limits) during periods of startup and shutdown. As discussed in Section V.F of this preamble, we considered whether performance testing, and therefore, enforcement of numeric emission limits, would be practicable during periods of startup and shutdown. EPA determined that it is not technically feasible to complete stack testing—in particular, to repeat the multiple required test runs—during periods of startup and shutdown due to physical limitations and the short duration of startup and shutdown periods. Therefore, we have established the separate work practice standard for periods of startup and shutdown.

Periods of startup, normal operations, and shutdown are all predictable and routine aspects of a source's operations. However, by contrast, malfunction is defined as a “sudden, infrequent, and not reasonably preventable failure of air pollution control and monitoring equipment, process equipment or a process to operate in a normal or usual manner * * * ”(40 CFR 63.2). EPA has determined that malfunctions should not be viewed as a distinct operating mode and, therefore, any emissions that occur at such times do not need to be factored into development of CAA section 112(d) standards, which, once promulgated, apply at all times. In

Mossville Environmental Action Now

v.

EPA

, 370 F.3d 1232, 1242 (D.C. Cir. 2004), the court upheld as reasonable standards that had factored in variability of emissions under all operating conditions. However, nothing in section 112(d) or in case law requires that EPA anticipate and account for the innumerable types of potential malfunction events in setting emission standards.

See, Weyerhaeuser

v.

Costle

, 590 F.2d 1011, 1058 (D.C. Cir. 1978) (“In the nature of things, no general limit, individual permit, or even any upset provision can anticipate all upset situations. After a certain point, the transgression of regulatory limits caused by `uncontrollable acts of third parties,' such as strikes, sabotage, operator intoxication or insanity, and a variety of other eventualities, must be a matter for the administrative exercise of case-by-case enforcement discretion, not for specification in advance by regulation.”)

Further, it is reasonable to interpret section 112(d) as not requiring EPA to account for malfunctions in setting emissions standards. For example, we note that Section 112 uses the concept of “best performing” sources in defining MACT, the level of stringency that major source standards must meet. Applying the concept of “best performing” to a source that is malfunctioning presents significant difficulties. The goal of best performing sources is to operate in such a way as to avoid malfunctions of their units.

Moreover, even if malfunctions were considered a distinct operating mode, we believe it would be impracticable to take malfunctions into account in setting CAA section 112(d) standards for boilers and process heaters. As noted above, by definition, malfunctions are sudden and unexpected events and it would be difficult to set a standard that takes into account the myriad different types of malfunctions that can occur across all sources in the category. Moreover, malfunctions can vary in frequency, degree, and duration, further complicating standard setting.

In the event that a source fails to comply with the applicable CAA section 112(d) standards as a result of a malfunction event, EPA would determine an appropriate response based on, among other things, the good faith efforts of the source to minimize emissions during malfunction periods, including preventative and corrective actions, as well as root cause analyses to ascertain and rectify excess emissions. EPA would also consider whether the source's failure to comply with the CAA section 112(d) standard was, in fact, “sudden, infrequent, not reasonably preventable” and was not instead “caused in part by poor maintenance or careless operation.” 40 CFR 63.2 (definition of malfunction).

Finally, EPA recognizes that even equipment that is properly designed and maintained can sometimes fail and that such failure can sometimes cause an exceedance of the relevant emission standard. (

See, e.g.,

State Implementation Plans: Policy Regarding Excessive Emissions During Malfunctions, Startup, and Shutdown (Sept. 20, 1999); Policy on Excess Emissions During Startup, Shutdown, Maintenance, and Malfunctions (Feb. 15, 1983)). EPA is, therefore, adding to this final rule an affirmative defense to civil penalties for exceedances of numerical emission limits that are caused by malfunctions.

See

40 CFR 63.7575 (defining “affirmative defense” to mean, in the context of an enforcement proceeding, a response or defense put forward by a defendant, regarding which the defendant has the burden of proof, and the merits of which are independently and objectively evaluated in a judicial or administrative proceeding.). We also have added other regulatory provisions to specify the elements that are necessary to establish this affirmative defense; the source must prove by a preponderance of the evidence that it has met all of the elements set forth in 63.7501. (

See

40 CFR 22.24). The criteria ensure that the affirmative defense is available only where the event that causes an exceedance of the emission limit meets the narrow definition of malfunction in 40 CFR 63.2 (sudden, infrequent, not reasonably preventable and not caused by poor maintenance and or careless operation). For example, to successfully assert the affirmative defense, the source must prove by a preponderance of the evidence that excess emissions “[w]ere caused by a sudden, infrequent, and unavoidable failure of air pollution control and monitoring equipment,

process equipment, or a process to operate in a normal or usual manner * * *.” The criteria also are designed to ensure that steps are taken to correct the malfunction, to minimize emissions in accordance with section 63.7500(a)(3) and to prevent future malfunctions. For example, the source must prove by a preponderance of the evidence that “[r]epairs were made as expeditiously as possible when the applicable emission limitations were being exceeded * * *” and that “[a]ll possible steps were taken to minimize the impact of the excess emissions on ambient air quality, the environment and human health * * *.” In any judicial or administrative proceeding, the Administrator may challenge the assertion of the affirmative defense and, if the respondent has not met its burden of proving all of the requirements in the affirmative defense, appropriate penalties may be assessed in accordance with Section 113 of the CAA (

see also

40 CFR 22.77).

F. What are the testing and initial compliance requirements?

We are requiring that the owner or operator of a new or existing boiler or process heater must conduct performance tests to demonstrate compliance with all applicable emission limits. Affected units would be required to conduct the following compliance tests where applicable:

(1) Conduct initial and annual stack tests to determine compliance with the PM emission limits using EPA Method 5 or 17.

(2) Conduct initial and annual stack tests to determine compliance with the Hg emission limits using EPA method 29 or ASTM-D6784-02 (Ontario Hydro Method).

(3) Conduct initial and annual stack tests to determine compliance with the HCl emission limits using EPA Method 26A or EPA Method 26 (if no entrained water droplets in the sample).

(4) Use EPA Method 19 to convert measured concentration values to pound per million Btu values.

(5) Conduct initial and annual test to determine compliance with the CO emission limits using EPA Method 10.

(6) Conduct initial test to determine compliance with the dioxin/furan emission limits using EPA Method 23.

As part of the initial compliance demonstration, we are requiring that you monitor specified operating parameters during the initial performance tests that you would conduct to demonstrate compliance with the PM, Hg, HCl, CO, and dioxin/furan emission limits. You must calculate the average hourly parameter values measured during each test run over the three run performance test. The lowest or highest hourly average of the three test run values (depending on the parameter measured) for each applicable parameter would establish the site-specific operating limit. The applicable operating parameters for which operating limits would be required to be established are based on the emissions limits applicable to your unit as well as the types of add-on controls on the unit. The following is a summary of the operating limits that we are requiring to be established for the various types of the following units:

(1) For boilers and process heaters with wet PM scrubbers, you must measure pressure drop and liquid flow rate of the scrubber during the performance test, and calculate the average hourly values during each test run. The lowest hourly average determined during the three test runs establishes your minimum site-specific pressure drop and liquid flow rate operating levels.

(2) If you are complying with an HCl emission limit using a wet acid gas scrubber, you must measure pH and liquid flow rate of the scrubber sorbent during the performance test, and calculate the average hourly values during each test run of the performance test for HCl and determine the lowest hourly average of the pH and liquid flow rate for each test run for the performance test. This establishes your minimum pH and liquid flow rate operating limits.

(3) For boilers and process heaters with sorbent injection, you must measure the sorbent injection rate for each acid gas sorbent used during the performance tests for HCl and for activated carbon for Hg and dioxin/furan and calculate the hourly average for each sorbent injection rate during each test run. The lowest hourly average measured during the performance tests becomes your site-specific minimum sorbent injection rate operating limit. If different acid gas sorbents and/or injection rates are used during the HCl test, the lowest hourly average value for each sorbent becomes your site-specific operating limit. When your unit operates at lower loads, multiply your sorbent injection rate by the load fraction (operating heat input divided by the average heat input during your last compliance test for the appropriate pollutant) to determine the required parameter value.

(4) For boilers and process heaters with fabric filters not subject to PM Continuous Emission Monitoring System (CEMS) or continuous compliance with an opacity limit (

i.e.,

COMS), the fabric filter must be operated such that the bag leak detection system alarm does not sound more than 5 percent of the operating time during any 6-month period unless a CEMS is installed to measure PM.

(5) For boilers and process heaters with electrostatic precipitators (ESP) not subject to PM CEMS or continuous compliance with an opacity limit (

i.e.,

COMS) and you must measure the secondary voltage and secondary current of the ESP collection fields during the Hg and PM performance test. You then calculate the average total secondary electric power value from these parameters for each test run. The lowest average total secondary electric power measured during the three test runs establishes your site-specific minimum operating limit for the ESP.

(6) For boilers and process heaters that choose to demonstrate compliance with the Hg emission limit on the basis of fuel analysis, you are required to measure the Hg content of the inlet fuel that was burned during the Hg performance test. This value is your maximum fuel inlet Hg operating limit.

(7) For boilers and process heaters that choose to demonstrate compliance with the HCl emission limit on the basis of fuel analysis, you are required to measure the chlorine content of the inlet fuel that was burned during the HCl performance test. This value is your maximum fuel inlet chlorine operating limit.

(8) For boilers and process heaters that are subject to a CO emission limit and a dioxin/furan emission limit, you are required to measure the oxygen concentration in the flue gas during the initial CO and dioxin/furan performance test. The lowest hourly average oxygen concentration measured during the most recent performance test is your operating limit, and your unit must operate at or above your operating limit on a 12-hour block average basis.

These operating limits do not apply to owners or operators of boilers or process heaters having a heat input capacity of less than 10 MMBtu/hr or boilers or process heaters of any size which combust natural gas or other clean gas, metal process furnaces, or limited use units, as discussed in section IV.D.3 of this preamble. Instead, owners or operators of such boilers and process heaters shall submit to the delegated authority or EPA, as appropriate, if requested, documentation that a tune-up meeting the requirements of this final rule was conducted. In order to comply with the work practice standard, a tune-up procedure must include the following:

(1) Inspect the burner, and clean or replace any components of the burner as necessary,

(2) Inspect the flame pattern and make any adjustments to the burner necessary to optimize the flame pattern consistent with the manufacturer's specifications,

(3) Inspect the system controlling the air-to-fuel ratio, and ensure that it is correctly calibrated and functioning properly,

(4) Optimize total emissions of CO consistent with the manufacturer's specifications,

(5) Measure the concentration in the effluent stream of CO in parts per million by volume dry (ppmvd), before and after the adjustments are made,

(6) Submit to the delegated authority or EPA an annual report containing the concentrations of CO in the effluent stream in ppmvd, and oxygen in percent dry basis, measured before and after the adjustments of the boiler, a description of any corrective actions taken as a part of the combustion adjustment, and the type and amount of fuel used over the 12 months prior to the annual adjustment.

Further, all owners or operators of major source facilities having boilers and process heaters subject to this final rule are required to submit to the delegated authority or EPA, as appropriate, documentation that an energy assessment was performed, by a qualified energy assessor, and the cost-effective energy conservation measures indentified.

G. What are the continuous compliance requirements?

To demonstrate continuous compliance with the emission limitations, we are requiring the following:

(1) For units combusting coal, biomass, or residual fuel oil (

i.e.,

No 4, 5 or 6 fuel oil) with heat input capacities of less than 250 MMBtu/hr that do not use a wet scrubber, we are requiring that opacity levels be maintained to less than 10 percent (daily average) for existing and new units with applicable emission limits. Or, if the unit is controlled with a fabric filter, instead of continuous monitoring of opacity, the fabric filter must be continuously operated such that the bag leak detection system alarm does not sound more than 5 percent of the operating time during any 6-month period (unless a PM CEMS is used).

(2) For units combusting coal, biomass, or residual oil with heat input capacities of 250 MMBtu/hr or greater, we are requiring that PM CEMS be installed and operated and that PM levels (monthly average) be maintained below the applicable PM limit.

(3) For boilers and process heaters with wet PM scrubbers, we are requiring that you monitor pressure drop and liquid flow rate of the scrubber and maintain the 12-hour block averages at or above the operating limits established during the performance test to demonstrate continuous compliance with the PM emission limits.

(4) For boilers and process heaters with wet acid gas scrubbers, you must monitor the pH and liquid flow rate of the scrubber and maintain the 12-hour block average at or above the operating limits established during the most recent performance test to demonstrate continuous compliance with the HCl emission limits.

(5) For boilers and process heaters with dry scrubbers, we are requiring that you continuously monitor the sorbent injection rate and maintain it at or above the operating limits, which include an adjustment for load, established during the performance tests. When your unit operates at lower loads, multiply your sorbent injection rate by the load fraction (operating load divided by the load during your last compliance test for the appropriate pollutant) to determine the required parameter value.

(6) For boilers and process heaters having heat input capacities of less than 250 MMBtu/hr with an ESP, we are requiring that you monitor the voltage and current of the ESP collection plates and maintain the 12-hour block total secondary electric power averages at or above the operating limits established during the Hg or PM performance test.

(7) For units that choose to comply with either the Hg emission limit or the HCl emission limit based on fuel analysis rather than on performance testing, you must maintain monthly fuel records that demonstrate that you burned no new fuels or fuels from a new supplier such that the Hg content or the chlorine content of the inlet fuel was maintained at or below your maximum fuel Hg content operating limit or your chlorine content operating limit set during the performance tests. If you plan to burn a new fuel, a fuel from a new mixture, or a new supplier's fuel that differs from what was burned during the initial performance tests, then you must recalculate the maximum Hg input and/or the maximum chlorine input anticipated from the new fuels based on supplier data or own fuel analysis, using the methodology specified in Table 6 of this final rule. If the results of recalculating the inputs exceed the average content levels established during the initial test then, you must conduct a new performance test(s) to demonstrate continuous compliance with the applicable emission limit.

(8) For all boilers and process heaters, except those that are exempt from the incinerator standards under section 129 because they are qualifying facilities burning a homogeneous waste stream, you must maintain records of fuel use that demonstrate that your fuel was not solid waste.

(9) For boilers and process heaters with an oxygen monitor installed for this final rule, you must maintain an oxygen concentration level, on a 12-hour block average basis, no less than lowest hourly average oxygen concentration measured during the most recent performance test.

(10) For boilers and process heaters that demonstrate compliance using a performance test. You must maintain an operating load no greater than 110 percent of the operating load established during the performance test.

If an owner or operator would like to use a control device other than the ones specified in this section to comply with this final rule, the owner/operator should follow the requirements in 40 CFR 63.8(f), which presents the procedure for submitting a request to the Administrator to use alternative monitoring.

H. What are the notification, recordkeeping and reporting requirements?

All new and existing sources are required to comply with certain requirements of the General Provisions (40 CFR part 63, subpart A), which are identified in Table 10 of this final rule. The General Provisions include specific requirements for notifications, recordkeeping, and reporting.

Each owner or operator is required to submit a notification of compliance status report, as required by § 63.9(h) of the General Provisions. This final rule requires the owner or operator to include in the notification of compliance status report certifications of compliance with rule requirements.

Semiannual compliance reports, as required by § 63.10(e)(3) of subpart A, are required only for semiannual reporting periods when a deviation from any of the requirements in the rule occurred, or any process changes occurred and compliance certifications were reevaluated.

This final rule requires records to demonstrate compliance with each emission limit and work practice standard. These recordkeeping requirements are specified directly in the General Provisions to 40 CFR part

63, and are identified in Table 10. Owners or operators of sources with units with heat input capacity of less than 10 MMBtu/hr, units combusting natural gas or other clean gas, metal process furnaces, limited use units, and temporary use units must keep records of the dates and the results of each required boiler tune-up.

Records of either continuously monitored parameter data for a control device if a device is used to control the emissions or CEMS data are required.

You are required to keep the following records:

(1) All reports and notifications submitted to comply with this final rule.

(2) Continuous monitoring data as required in this final rule.

(3) Each instance in which you did not meet each emission limit and each operating limit (

i.e.,

deviations from this final rule).

(4) Daily hours of operation by each source.

(5) Total fuel use by each affected source electing to comply with an emission limit based on fuel analysis for each 30-day period along with a description of the fuel, the total fuel usage amounts and units of measure, and information on the supplier and original source of the fuel.

(6) Calculations and supporting information of chlorine fuel input, as required in this final rule, for each affected source with an applicable HCl emission limit.

(7) Calculations and supporting information of Hg fuel input, as required in this final rule, for each affected source with an applicable Hg emission limit.

(8) A signed statement, as required in this final rule, indicating that you burned no new fuel type and no new fuel mixture or that the recalculation of chlorine input demonstrated that the new fuel or new mixture still meets chlorine fuel input levels, for each affected source with an applicable HCl emission limit.

(9) A signed statement, as required in this final rule, indicating that you burned no new fuels and no new fuel mixture or that the recalculation of Hg fuel input demonstrated that the new fuel or new fuel mixture still meets the Hg fuel input levels, for each affected source with an applicable Hg emission limit.

(10) A copy of the results of all performance tests, fuel analysis, opacity observations, performance evaluations, or other compliance demonstrations conducted to demonstrate initial or continuous compliance with this final rule.

(11) A copy of your site-specific monitoring plan developed for this final rule as specified in 63 CFR 63.8(e), if applicable.

We are also requiring that you submit the following reports and notifications:

(1) Notifications required by the General Provisions.

(2) Initial Notification no later than 120 calendar days after you become subject to this subpart, even if you submitted an initial notification for the vacated standards that were promulgated in 2004.

(3) Notification of Intent to conduct performance tests and/or compliance demonstration at least 60 calendar days before the performance test and/or compliance demonstration is scheduled.

(4) Notification of Compliance Status 60 calendar days following completion of the performance test and/or compliance demonstration.

(5) Compliance reports semi-annually.

I. Submission of Emissions Test Results to EPA

EPA must have performance test data and other compliance data to conduct effective reviews of CAA Section 112 and 129 standards, as well as for many other purposes including compliance determinations, emissions factor development, and annual emissions rate determinations. In conducting these required reviews, we have found it ineffective and time consuming not only for us but also for regulatory agencies and source owners and operators to locate, collect, and submit emissions test data because of varied locations for data storage and varied data storage methods. One improvement that has occurred in recent years is the availability of stack test reports in electronic format as a replacement for cumbersome paper copies.

In this action, we are taking a step to improve data accessibility. Owners and operators of ICI boilers located at major source facilities will be required to submit to EPA an electronic copy of reports of certain performance tests required under this final rule. Data will be collected through an electronic emissions test report structure called the Electronic Reporting Tool (ERT) that will be used by the staff as part of the emissions testing project. The ERT was developed with input from stack testing companies who generally collect and compile performance test data electronically and offices within State and local agencies which perform field test assessments. The ERT is currently available, and access to direct data submittal to EPA's electronic emissions database (WebFIRE) is scheduled to become available by December 31, 2011.

The requirement to submit source test data electronically to EPA will not require any additional performance testing and will apply to those performance tests conducted using test methods that are supported by ERT. The ERT contains a specific electronic data entry form for most of the commonly used EPA reference methods. The Web site listed below contains a listing of the pollutants and test methods supported by ERT. In addition, when a facility submits performance test data to WebFIRE, there will be no additional requirements for emissions test data compilation. Moreover, we believe industry will benefit from development of improved emissions factors, fewer follow-up information requests, and better regulation development as discussed below. The information to be reported is already required for the existing test methods and is necessary to evaluate the conformance to the test method.

One major advantage of collecting source test data through the ERT is that it provides a standardized method to compile and store much of the documentation required to be reported by this final rule while clearly stating what testing information we require. Another important benefit of submitting these data to EPA at the time the source test is conducted is that it will substantially reduce the effort involved in data collection activities in the future. Specifically, because EPA would already have adequate source category data to conduct residual risk assessments or technology reviews, there would likely be fewer or less substantial data collection requests (e.g., CAA Section 114 letters). This results in a reduced burden on both affected facilities (in terms of reduced manpower to respond to data collection requests) and EPA (in terms of preparing and distributing data collection requests).

State/local/Tribal agencies may also benefit in that their review may be more streamlined and accurate because the States will not have to re-enter the data to assess the calculations and verify the data entry. Finally, another benefit of submitting these data to WebFIRE electronically is that these data will improve greatly the overall quality of the existing and new emissions factors by supplementing the pool of emissions test data upon which the emissions factor is based and by ensuring that data are more representative of current industry operational procedures. A common complaint we hear from industry and regulators is that emissions factors are outdated or not representative of a particular source category. Receiving and incorporating

data for most performance tests will ensure that emissions factors, when updated, represent accurately the most current operational practices. In summary, receiving test data already collected for other purposes and using them in the emissions factors development program will save industry, State/local/Tribal agencies, and EPA time and money and work to improve the quality of emissions inventories and related regulatory decisions.

As mentioned earlier, the electronic data base that will be used is EPA's WebFIRE, which is a database accessible through EPA's TTN. The WebFIRE database was constructed to store emissions test and other data for use in developing emissions factors. A description of the WebFIRE data base can be found at

http://cfpub.epa.gov/oarweb/index.cfm?action=fire.main.

Source owners and operators will be able to transmit data collected via the ERT through EPA's Central Data Exchange (CDX) network for storage in the WebFIRE data base. Although ERT is not the only electronic interface that can be used to submit source test data to the CDX for entry into WebFIRE, it makes submittal of data very straightforward and easy. A description of the ERT can be found at

http://www.epa.gov/ttn/chief/ert/ert_tool.html.

Source owners and operators must register with the CDX system to obtain a user name and password before being able to submit data to the CDX. The CDX registration page can be found at:

https://cdx.epa.gov/SSL/CDX/regwarning.asp?Referer=registration.

If they have a current CDX account (

e.g.,

they submit reports for EPA's Toxic Release Inventory Program to the CDX), then the existing user name and password can be used to log in to the CDX.

IV. Summary of Significant Changes Since Proposal

A. Applicability

Since proposal, several changes to the applicability of this final rule have been made. First, at proposal, we excluded all units that combust solid waste from the standards, but we have extended the coverage of this final rule to boilers and process heaters that combust solid waste but are exempt, by statute, from section 129 incinerator rules because they are qualifying small power producers or cogeneration units that combust a homogeneous waste stream. This final rule continues to exclude other waste burning units. This is a clarifying change that is consistent with the intent of the proposed rule to establish emissions standards for all boilers and process heaters that are not solid waste incineration units subject to regulation under section 129.

The proposed rule definition of coal was revised to include all types of fossil-based fuels in the coal definition. The final coal definition is: “

Coal

means all solid fuels classifiable as anthracite, bituminous, sub-bituminous, or lignite by the American Society for Testing and Materials in ASTM D388-991, “Standard Specification for Classification of Coals by Rank” (incorporated by reference,

see

§ 63.14(b)), coal refuse, and petroleum coke. For the purposes of this subpart, this definition of “coal” includes synthetic fuels derived from coal for the purpose of creating useful heat, including but not limited to, solvent-refined coal, coal-oil mixtures, and coal-water mixtures. Coal derived gases are excluded from this definition.” Similarly, for biomass, the definition of biomass fuel was revised to include any potential biomass-based fuels. This is also a clarifying change consistent with the intent of the proposed rule as described above. The final definition is: “Biomass or bio-based solid fuel means any solid biomass-based fuel that is not a solid waste. This may include, but is not limited to, the following materials: Wood residue; wood products (

e.g.,

trees, tree stumps, tree limbs, bark, lumber, sawdust, sanderdust, chips, scraps, slabs, millings, and shavings); animal manure, including litter and other bedding materials; vegetative agricultural and silvicultural materials, such as logging residues (slash), nut and grain hulls and chaff (

e.g.,

almond, walnut, peanut, rice, and wheat), bagasse, orchard prunings, corn stalks, coffee bean hulls and grounds. This definition of biomass fuel is not intended to suggest that these materials are or not solid waste.”

The proposed rule included a definition of waste heat boiler that excluded from the definition units with supplemental burners that are designed to supply 50 percent or more of the total rated heat input capacity. The final definition was revised to include all waste heat boilers. The final definition is: “Waste heat boiler means a device that recovers normally unused energy and converts it to usable heat. Waste heat boilers are also referred to as heat recovery steam generators.” Similarly, the waste heat process heater definition was revised to read as follows: “Waste heat process heater means an enclosed device that recovers normally unused energy and converts it to usable heat. Waste heat process heaters are also referred to as recuperative process heaters.” These changes were made in order to exempt the types of units intended at proposal.

The proposed rule exempted blast furnace gas fuel-fired boiler or process heaters, and defined these units as units combusting 90 percent or more of its total heat input from blast furnace gas. We have changed the requirement to 90 percent or more of its total volume of gas in this final rule. This change was made so that the units that were intended to be exempted from this final rule would be exempted. The wording of the proposed exemption did not exempt units that were intended to be exempted because the heating value of blast furnace gas is not as high as that of natural gas.

The proposed rule exempted units that are an affected source in another MACT standard. We amended this language to include any unit that is part of the affected source subject to another MACT standard. We also exempted any unit that is used as a control device to comply with another MACT standard, provided that at least 50 percent of the heat input is provided by the gas stream that is regulated under another MACT standard. This change was made in order to encourage the recovery of energy from high heating value gases that would otherwise be flared.

B. Subcategories

In the proposed rule, for the fuel-dependent HAP (metals, Hg, acid gases), we identified the following five basic unit types as subcategories: (1) Units designed to burn coal, (2) units designed to burn biomass, (3) units designed to burn liquid fuel, (4) units designed to burn natural gas/refinery gas, and (5) units designed to burn other process gases. In this final rule, for fuel-dependent HAP, we combined the subcategories for units designed to combust coal and biomass into a subcategory for units designed to burn solid fuels. We changed the subcategory for units designed to burn natural gas/refinery gas to a subcategory for units that burn natural gas, refinery gas, and other clean gas. We also added subcategories for non-continental liquid units and limited-use units.

As described in the preamble to the proposed rule, within the basic unit types there are different designs and combustion systems that, while having a minor effect on fuel-dependent HAP emissions, have a much larger effect on pollutants whose emissions depend on the combustion conditions in a boiler or process heater. In the case of boilers and process heaters, the combustion-related pollutants are the organic HAP. In the proposed rule, we identified the

following 11 subcategories for organic HAP: (1) Pulverized coal units; (2) stokers designed to burn coal; (3) fluidized bed units designed to burn coal; (4) stokers designed to burn biomass; (5) fluidized bed units designed to burn biomass; (6) suspension burners/dutch ovens designed to burn biomass; (7) fuel cells designed to burn biomass; (8) units designed to burn liquid fuel; (9) units designed to burn natural gas/refinery gas; (10) units designed to burn other gases; and (11) metal process furnaces. In this final rule, we added subcategories for biomass suspension/grate units, non-continental liquid units, and limited-use units.

C. Emission Limits

The proposed rule included numerical emission limits for PM, Hg, HCl, CO, and dioxin/furan, and limits for those same pollutants are included in this final rule. Unlike the proposed rule, we included a compliance alternative in the final rule to allow owners and operators of existing affected sources to demonstrate compliance on an output-basis instead of on a heat input basis. Compliance with the alternate output-based emission limits would require measurement of boiler operating parameters associated with the mass rate of emissions and energy outputs. If you elect to comply with the alternate output-based emission limits, you must use equations provided in the final rule to demonstrate that emissions from the applicable units do not exceed the output-based emission limits specified in the final rule. If you use this compliance alternative using the emission credit approach, you must also establish a benchmark, calculate and document the emission credits generated from energy conservation measures implemented, and develop and submit the implementation plan no later than 180 days before the date that the facility intends to demonstrate compliance.

D. Work Practices

This final rule includes work practice standards for most of the same units for which we proposed work practice standards, including new and existing units in the Gas 1 subcategory, existing units with heat input capacity less than 10 MMBtu/hr, and new and existing metal process furnaces. In addition to those subcategories for which we proposed work practices, this final rule includes work practices for all units during periods of startup and shutdown, new units with heat input capacity less than 10 MMBtu/hr, limited use units, and units combusting other clean gases. Other clean gases are gases, other than natural gas and refinery gas (as defined in this final rule), that meet contaminant level specifications that are provided in the final rule.

E. Energy Assessment Requirements

In this final rule, we have expanded the definition of energy assessment with respect to the requirements of Table 3 of this final rule, by providing a duration for performing the energy assessment and defining the evaluation requirements for each boiler system and energy use system. These requirements are based on the total annual heat input to the affected boilers and process heaters.

This final rule requires an energy assessment for facilities with affected boilers and process heaters using less than 0.3 trillion Btu per year (TBtu/y) heat input to be one day in length maximum. The boiler system and energy use system accounting for at least 50 percent of the energy output from these units must be evaluated to identify energy savings opportunities within the limit of performing a one day energy assessment. An energy assessment for a facility with affected boilers and process heaters using 0.3 to 1 TBtu/year must be three days in length maximum. From these boilers, the boiler system and any energy use system accounting for at least 33 percent of the energy output will be evaluated, within the limit of performing a three day energy assessment. For facilities with affected boilers and process heaters using greater than 1 TBtu/year heat input, the energy assessment must address the boiler system and any energy use system accounting for at least 20 percent of the energy output to identify energy savings opportunities.

The expanded definition for energy assessment clarifies the duration and requirements for each energy assessment for various units based on energy use. We have also added a definition for steam and process heating systems to clarify the components for each boiler system which must be considered during the energy assessment, including elements such as combustion management, thermal energy recovery, energy resource selection, and the steam end-use management of each affected boiler.

Lastly, we have clarified the requirement in Table 3 to evaluate facility energy management practices as part of the energy assessment and a definition of an energy management program was added. The use of the ENERGY STAR Facility Energy Assessment Matrix as part of this review is recommended, but it was removed as a requirement in Table 3. The definition of an energy management program added to the rule is consistent with the ENERGY STAR Guidelines for Energy Management that can be referenced for further guidance. ENERGY STAR provides a variety of tools and resources that support energy management programs. For more information, visit

http://www.energystar.gov.

F. Requirements During Startup, Shutdown, and Malfunction

For startup, shutdown, and malfunction (SSM), the requirements have changed since proposal. For periods of startup and shutdown, EPA is finalizing work practice standards, which require following manufacturers specifications for minimizing periods of startup and shutdown, in lieu of numeric emission limits. For malfunctions, EPA added affirmative defense language to this final rule for exceedances of the numerical emission limits that are caused by malfunctions.

G. Testing and Initial Compliance

The first significant change to the testing and initial compliance requirements is that units greater than 100 MMBtu/hr must comply with the CO limits using a stack test rather than CO CEMS. EPA also added optional output-based limits that promote energy efficient boiler operation. Another significant change is that for units combusting gaseous fuels other than natural gas or refinery gas, in order to qualify for the Gas 1 subcategory work practice standard, the gases that will be combusted must be certified to meet the contaminant levels specified for Hg and hydrogen sulfide (H

2

S) in this final rule. Finally, EPA has changed the dioxin/furan testing requirement to a one-time compliance demonstration due to the low dioxin/furan emissions demonstrated by the vast majority of sources that have tested for dioxin/furan.

H. Continuous Compliance

The only significant change to the continuous compliance requirements is for monitoring of CO. Rather than using CO CEMS, as proposed, units will be required to continuously monitor and record the oxygen level in their flue gas during the initial compliance test and establish an operating limit that requires that the unit operate at an oxygen percentage of at least 90 percent of the operating limit on a 12-hour block average basis. Units will be required to continuously monitor oxygen to ensure continuous compliance.

I. Notification, Recordkeeping, and Reporting

In this final action, we are requiring that owners or operators of boilers that choose to commence or recommence combustion of solid waste must provide 30 days notice of the date upon which the source will commence or recommence combustion of solid waste. The notification must identify the name of the owner or operator of the affected source, the location of the source, the boiler(s) or process heater(s) that will commence burning solid waste, and the date of the notice; the currently applicable subcategory under this subpart; the date on which the unit became subject to the currently applicable emission limits; and the date upon which the unit will commence or recommence combusting solid waste.

For each limited-use unit, owners or operators must monitor and record the operating hours on a monthly basis for the unit. This will ensure that units qualify for the limited-use subcategory.

We also added a requirement that sources keep records of operating load in order to demonstrate continuous compliance with the operating load operating limit.

When malfunctions occur, owners or operators must keep records of the occurrence and duration of each malfunction of the boiler or process heater, or of the associated air pollution control and monitoring equipment, as well as records of actions taken during periods of malfunction to minimize emissions, including corrective actions to restore the malfunctioning boiler or process heater, air pollution control, or monitoring equipment to its normal or usual manner of operation.

Finally, for facilities that elect to use emission credits from energy conservation measures to demonstrate compliance, owners or operators must keep a copy of the Implementation Plan required in this rule and copies of all data and calculations used to establish credits.

J. Technical/Editorial Corrections

In this final action, we are making a number of technical corrections and clarifications to subpart DDDDD. These changes improve the clarity and procedures for implementing the emission limitations to affected sources. We are also clarifying several definitions to help affected sources determine their applicability. We have modified some of the regulatory language that we proposed based on public comments.

In several places throughout the subpart, including the associated tables, we have corrected the cross-references to other sections and paragraphs of the subpart.

We revised 40 CFR 63.7485 to clarify that for the purposes of subpart DDDDD, a major source of HAP is as defined in 40 CFR 63.2, except that for oil and gas facilities a major source of HAP is as defined in 40 CFR 63.761 (40 CFR part 63, subpart HH, National Emission Standards for Hazardous Air Pollutants from Oil and Natural Gas Production Facilities). This change was made because facilities subject to subpart HH contain units that will be subject to subject DDDDD.

The word “specifically” was removed from § 63.7491(i) in order to clarify the exclusion for boilers and process heaters regulated by other HAP regulations.

We revised 40 CFR 63.7505(c) to clarify that performance testing is needed only if a boiler or process heater is subject to an applicable emission limit listed in Table 2.

We made several changes to the initial compliance demonstration requirements. We revised 40 CFR 63.7510(a) to clarify that sources using a second fuel only for start up, shut down, and/or transient flame stability are still considered to be sources using a single fuel. We revised 40 CFR 63.7510(c) to clarify that boilers and process heaters with a heat input capacity below 10 MMBtu per hour are not required to conduct a performance test for CO because they are not subject to a numerical emission limit for CO. In 40 CFR 63.7510(d), we clarified that boilers and process heaters that use a CEMS for PM are exempt from the performance testing and operating limit requirements specified in 40 CFR 63.7510(a) because the CEMS demonstrates continuous compliance. We revised 40 CFR 63.7510(c) and (d) to clarify that compliance for those provisions does not apply to units burning natural gas or refinery gas.

We changed the performance testing requirements in 40 CFR 63.7515(b), (c), and (d) to state that performance testing for a given pollutant may be performed every 3 years, instead of annually, if measured emissions during 2 consecutive annual performance tests are less than 75 percent of the applicable emission limit.

In 40 CFR 63.7515(e), we clarified that boilers and process heaters with a heat input capacity below 10 MMBtu per hour are required to conduct tune-ups biennially, while larger natural gas and other Gas 1 units are required to conduct annual tune-ups.

We revised 40 CFR 63.7515(f) to clarify that monthly fuel analyses are required only for fuel types for which emission limits apply.

We made several changes to 40 CFR 63.7520 to clarify the performance testing requirements. We revised paragraph (c) to clarify that performance tests must be conducted at representative operating load conditions, instead of at the maximum normal operating load. Language was also added to this section and to Table 4 to subpart DDDDD to establish an operating limit for the boiler or process heater and clarified that the operating load must not exceed 110 percent of the load used during the performance test. We revised paragraph (d) to clarify that compliance with operating limits using a continuous parameter monitoring systems are based on the 4-hour block averages of the data collected by the continuous parameter monitoring systems.

In 40 CFR 63.7522, we made several changes to the provisions for using emissions averaging. In paragraph (a), we clarified that average emissions must be “* * * not more than 90 percent of the applicable emission limit.” We also added a sentence to clarify that new boilers and process heaters may not be included in an emissions average used to demonstrate compliance according to that section. Equations 2 and 3 were revised to correct the discount factor from 0.9 to 1.1 because the actual emissions are multiplied by the discount factor. We also revised paragraph (c) to clarify that the deadline to establish emission caps to demonstrate compliance with the emission averaging option is 60 days after the publication of the final rule as referenced in paragraph (g)(2)(i), and revised paragraph (g) to clarify that facilities are required to submit an implementation plan as referenced in § 63.7522(g)(1).

We made several clarifying changes to the monitoring requirements in 40 CFR 63.7525. We revised paragraph (a) to clarify that only boilers or process heaters subject to a CO limit are required to install a continuous oxygen monitoring system. We adopted language from § 63.7525(d)(2) to § 63.7525(a)(6) to clarify what constitutes a deviation. In 40 CFR 63.7525(c)(7), we clarified that owners/operators are required to determine 6-minute and daily block averages excluding data from periods in which the continuous opacity monitoring system is out of control.

The initial compliance provisions in 40 CFR 63.7530(b) were revised to clarify that facilities are exempted from the initial compliance requirements of conducting a fuel analysis if only one

fuel type is used. We revised 40 CFR 63.7530(d) to clarify that units less than 10 MMBtu per hour are required to submit a signed statement with the Notification of Compliance Status report that indicates a tune-up has been conducted.

We revised 40 CFR 63.7540(a)(9)(i) to remove the reference to Procedure 2 in Appendix F to 40 CFR part 60; Procedure 2 specifies the ongoing QA/QC requirements for PM CEMS after certification and is correctly referenced in paragraph (a)(9)(iii) of that section.

We revised the notification requirements in 40 CFR 63.7545 to clarify that notifications should be submitted to the delegated authority, and to clarify that the Notification of Intent to conduct a performance test must be submitted 60 days before the test is scheduled to begin.

The reporting requirements originally in 40 CFR 63.7550(g) and (g)(1) through (g)(3) are more correctly considered notification requirements, so they were moved to § 63.7545(e)(8).

In response to comments asking for clarification, we have added definitions to 40 CFR 63.7575 for “Calendar year,” “Operating day,” “Refinery gas,” and “Valid hourly average.” We have also revised several definitions in that section based on public comments. For example, we revised the definition of “boiler” to describe what is meant by the term “controlled flame combustion” as used in that definition; revised “metal processing furnace” to include homogenizing furnaces; revised the definitions of “dry scrubber,” “electrostatic precipitator,” and “fabric filter,” to indicate that these are all considered dry control systems. The definition of “wet scrubber” was revised to clarify that, “A wet scrubber creates an aqueous stream or slurry as a byproduct of the emissions control process.”

The definition of “Tune-up” was removed from 40 CFR 63.7575 because all of the requirements for a tune-up are provided in the rule language at 40 CFR 63.7540(a)(10), making the definition unnecessary.

Several of the definitions in 40 CFR 64.7575 were revised to clarify the types of equipment to which different standards apply. For example, the definition of “Temporary boiler” was revised to include additional criteria that could be used to identify temporary boilers from permanently installed units. The definition of “Unit designed to burn oil subcategory” was revised to exclude periods of gas curtailment and gas supply emergency from the 48-hour limit on liquid fuel combustion. Likewise, the definition of “Period of natural gas curtailment” was revised to clarify that contractual agreements for curtailed gas usage or fluctuations in price do not constitute periods of gas curtailment under the scope of this regulation. The definition of “Waste heat boiler” was revised to remove the criteria that 50 percent of total rated heat input capacity had to be from waste gases. We also revised the definition of “Natural gas” to include gas derived from naturally occurring mixtures found in geological formations as long as the principal constituent is methane, consistent with the definition provided in 40 CFR part 60 subpart Db. A definition of propane, was also incorporated into the definition of natural gas.

Several changes were made to the tables to subpart DDDDD as a result of the public comments on the proposed rule.

In Tables 1 and 2, the references to “Other gases” were revised to “Gas 2” to clarify that units burning natural gas, refinery gas, or other clean gases are not subject to emission limitations. The emission limits in these two tables were also revised to include averaging times for those pollutants for which measurements are taken with a continuous emission monitor.

In Table 3, the references to “§ 63.11202 and § 63.11203” in the table heading were revised to correctly reference 40 CFR 63.7540. The text in the first and second column of Table 3 was revised to clarify that the requirements apply to both boilers and process heaters. A new row was added to clarify that work practice standards apply to new boilers or process heaters with a rated heat input capacity less than 10 MMBtu per hour. Language was also added to clarify that the energy assessment is a one-time requirement for existing boilers and process heaters. Additionally, new language was added clarifying the evaluation of the facility's energy management program as part of the energy assessment.

In Table 4, operating limits for pH added to Item 1 for wet scrubbers, as specified in 40 CFR 63.7530(b)(3)(i). Item 5 revised to clarify that “Any other control type” only means add-on air-pollution control devices. The operating limits were also revised to clarify which units and control combinations were required to install and operate a bag leak detection system, to install and operate a continuous opacity monitor, or to monitor voltage and amperage of an ESP. These changes removed the appearance that some units would need to do more than one type of monitoring for control of PM. This table was also revised to include a row for an operating limit for unit operating load for those units that demonstrate compliance using a performance test.

Table 5 was revised to include EPA Method 23 as the accepted method for measuring dioxin/furan. A new Table 11 was also added to document the toxic equivalency factors that should be used to demonstrate compliance with the toxic equivalents (TEQ) emission limits.

Table 7 was revised to include dry scrubbers and activated carbon injection used to comply with Hg or dioxin/furan emission limitations, and to include procedures for determining the corresponding operating limit requirements. Procedures were also added for determining the operating limit for unit operating load for units that demonstrate compliance through performance testing. Finally, this table was revised to clarify how the operating limits should be determined for wet scrubbers and for ESPs operated with wet scrubbers.

Table 8 was revised to correct certain cross-references to 40 CFR 63.7530, and to include procedures for demonstrating continuous compliance with the operating limit for unit operating load.

Table 9 was revised to correct cross-references to 40 CFR 63.7550(c) and Table 3 for work practice standards. Language in Item 1.c. revised to more clearly match the language in 40 CFR 63.7530(d) and (e), and Item 1.c. was split into Items 1.c. and 1.d.

K. Other

The definition of a boiler and the definition of a process heater have been revised to include units that combust solid waste but are exempt, by statute, from section 129. This change was necessary in order to provide coverage of units that would otherwise be exempt from any requirements. The revised definitions read as follows:

Boiler

means an enclosed device using controlled flame combustion and having the primary purpose of recovering thermal energy in the form of steam or hot water. Controlled flame combustion refers to a steady-state, or near steady-state, process wherein fuel and/or oxidizer feed rates are controlled. A device combusting solid waste, as defined in 40 CFR 241.3, is not a boiler unless the device is exempt from the definition of a solid waste incineration unit as provided in CAA section 129(g)(1). Waste heat boilers are excluded from this definition.

Process heater

means an enclosed device using controlled flame, and the unit's primary purpose is to transfer heat indirectly to a process material (liquid, gas, or solid) or to a heat transfer material for use in a process unit,

instead of generating steam. Process heaters are devices in which the combustion gases do not directly come into contact with process materials. For purposes of this subpart, a device combusting solid waste, as defined in 40 CFR 241.3, is not a process heater unless the device is exempt from the definition of a solid waste incineration unit as provided in CAA section 129(g)(1). Process heaters do not include units used for comfort heat or space heat, food preparation for on-site consumption, or autoclaves.

As a result of new data received for the floor calculations, revised treatment of low reported CO data to consider measurement error, and a new subcategorization scheme, some of the final CO limits for new sources in Table 1 of this final rule are more stringent than proposed, as are some of the other limits for certain subcategories (e.g., PM and Hg for liquid fuel units, and PM and HCl for solid fuel units when compared to the proposed new source limits for the proposed biomass/bio-based fuel subcategory). Where a final limit is more stringent than proposed, 40 CFR 63.6 of subpart A (General Provisions), requires that new sources that commenced construction between proposal and promulgation be allowed to comply with the proposed limits for 3 years (i.e., up to the existing source compliance date) and then comply with the final limits for new sources listed in Table 1 of this final rule. In this final rule we have added a new Table 12 to outline the emission limits applicable to sources that commenced construction between proposal and promulgation and updated the rule language to provide instructions on which limits apply to them for the 3 year period after this final rule is published. These sources have the option to comply with Table 1 (final) limits from the start, if they choose.

V. Major Source Public Comments and Responses

A. MACT Floor Analysis

1. Pollutant-by-Pollutant Approach

Comment:

Many commenters raised concerns about the way EPA determined the MACT floors using a pollutant-by-pollutant approach. Commenters contended that such a methodology produced limits that are not achievable in combination, and as such, the limits do not comport with the intent of the statute or the recent court decision (

NRDC

v.

EPA

, 2007). Commenters argue that while the Court's 2007 decision in

NRDC

v.

EPA

vacating the first ICI boiler and process heater MACT standard directed EPA to consider individual HAPs, it did not direct EPA to establish a separate floor for each HAP. Commenters further added that the Clean Air Act (CAA) directs EPA to set standards based on the overall performance of “sources” and sections 112(d)(1), (2), and (3) specify that emissions standards be established on the “in practice” performance of a “source” in the category or subcategory. If Congress had intended for EPA to establish MACT floor levels considering the achievable emission limits of individual HAPs, it could have worded 112(d)(3) to refer to the best-performing sources “for each pollutant.” Many commenters added that EPA's discretion in setting standards is limited to distinguishing among classes, types, and sizes of sources. However, Congress limited EPA's authority to parse units and sources with similar design and types but it does not allow EPA to “distinguish” units and sources by individual pollutant as proposed in this rule [

Sierra Club

v.

EPA,

551 F.3d 1019, 1028 (D.C. Cir. 2008)]. By calculating each MACT floor independently of the other pollutants, the combination of HAP limits results in a set of standards that only a hypothetical “best performing” unit could achieve.

Many commenters who criticized the pollutant-by-pollutant approach also filed comments on other rules such as the recent Portland Cement NESHAP and the NSPS and Emission Guidelines for Hospital/Medical Infectious Waste Incinerators (HMIWI). Some commenters expressed concern that EPA used a similar pollutant-by-pollutant approach in the HMIWI rulemaking and that rulemaking is being challenged before the D.C. Circuit. Commenters also submitted a variety of suggestions on calculating a multi-pollutant approach. Some commenters suggested that human health be considered by weighting pollutants according to relative-toxicity and then ranking the units in each subcategory according to their weighted emission totals in order to identify the best performing 12 percent of sources for all pollutants.

Response:

We disagree with the commenters who believe MACT floors cannot be set on a pollutant-by-pollutant basis. Contrary to the commenters' suggestion, section 112(d)(3) does not mandate a total facility approach. A reasonable interpretation of section 112(d)(3) is that MACT floors may be established on a HAP-by-HAP basis, so that there can be different pools of best performers for each HAP. Indeed, as illustrated below, the total facility approach not only is not compelled by the statutory language but can lead to results so arbitrary that the approach may simply not be legally permissible.

Section 112(d)(3) is ambiguous as to whether the MACT floor is to be based on the performance of an entire source or on the performance achieved in controlling particular HAP. Congress specified in section 112(d)(3) the minimum level of emission reduction that could satisfy the requirement to adopt MACT. For new sources, this floor level is to be “the emission control that is achieved in practice by the best controlled similar source.” For existing sources, the floor level is to be “the average emission limitation achieved by the best performing 12 percent of the existing sources” for categories and subcategories with 30 or more sources, or “the average emission limitation achieved by the best performing 5 sources” for categories and subcategories with fewer than 30 sources. Commenters point to the statute's reference to the best performing “sources,” and claim that Congress would have specifically referred to the best performing sources “for each pollutant” if it intended for EPA to establish MACT floors separately for each HAP. EPA disagrees. The language of the Act does not address whether floor levels can be established HAP-by-HAP or by any other means. The reference to “sources” does not lead to the assumption the commenters make that the best performing sources can only be the best-performing sources for the entire suite of regulated HAP. Instead, the language can be reasonably interpreted as referring to the source as a whole or to performance as to a particular HAP. Similarly, the reference in the new source MACT floor provision to “emission control achieved by the best controlled similar source” can mean emission control as to a particular HAP or emission control achieved by a source as a whole.

Industry commenters also stressed that section 112(d) requires that floors be based on actual performance from real facilities, pointing to such language as “existing source”, “best performing”, and “achieved in practice”. EPA agrees that this language refers to sources' actual operation, but again the language says nothing about whether it is referring to performance as to individual HAP or to single facility's performance for all HAP. Industry commenters also said that Congress could have mandated a HAP-by-HAP result by using the phrase “for each HAP” at appropriate points in section 112(d). The fact that Congress did not do so does not compel any inference that Congress was

sub-silentio

mandating a different result

when it left the provision ambiguous on this issue. The argument that MACT floors set HAP-by-HAP are based on the performance of a hypothetical facility, so that the limitations are not based on those achieved in practice, just re-begs the question of whether section 112(d)(3) refers to whole facilities or individual HAP. All of the limitations in the floors in this rule of course reflect sources' actual performance and were achieved in practice. Finally, there are a number of existing units that

meet all

of the final existing source emission limits.

Commenters also point to EPA's subcategorization authority, and claim that because Congress authorized EPA to distinguish among classes, types, and sizes of units, EPA cannot distinguish units by individual pollutant, as they allege EPA did in the proposed rule. However, that statutory language addresses EPA's authority to subcategorize sources within a source category prior to setting standards, which EPA has done for boilers and process heaters. EPA is not distinguishing within each subcategory based on HAP emitted. Rather, it is establishing emissions standards based on the emissions limits achieved by units in each subcategory. Therefore, EPA's subcategorization authority is irrelevant to the question of how EPA establishes MACT floor standards once it has made the decision to distinguish among sources and create subcategories.

EPA's long-standing interpretation of the Act is that the existing and new source MACT floors are to be established on a HAP-by-HAP basis. One reason for this interpretation is that a whole plant approach could yield least common denominator floors—that is floors reflecting mediocre or no control, rather than performance which is the average of what best performers have achieved.

See

61 FR at 173687 (April 19, 1996); 62 FR at 48363-64 (September 15, 1997) (same approach adopted under the very similar language of section 129(a)(2)). Such an approach would allow the performance of sources that are outside of the best-performing 12 percent for certain pollutants to be included in the floor calculations for those same pollutants, and it is even conceivable that the worst performing source for a pollutant could be considered a best performer overall, a result Congress could not have intended. Inclusion of units that are outside of the best performing 12 percent for particular pollutants would lead to emission limits that do not meet the requirements of the statute.

For example, if the best performing 12 percent of facilities for HAP metals were also the worst performing units for organics, the floor for organics or metals would end up not reflecting best performance. In such a situation, EPA would have to make some type of value judgment as to which pollutant reductions are most critical to decide which sources are best controlled.

3

Such value judgments are antithetical to the direction of the statute at the MACT floor-setting stage. Commenters suggested that a multi-pollutant approach could be implemented by weighting pollutants according to relative toxicity and calculating weighted emissions totals to use as a basis for identifying and ranking best performers. This suggested approach would require EPA to essentially prioritize the regulated HAP based on relative risk to human health of each pollutant, where risk is a criterion that has no place in the establishment of MACT floors, which are required by statute to be based on technology.

3

See

Petitioners Brief in

Medical Waste Institute et al.

v.

EPA

, No. 09-1297 (D.C. Cir.) pointing out, in this context, that “the best performers for some pollutants are the worst performers for others” (p. 34) and “[s]ome of the best performer for certain pollutants are among the worst performers for others.”

The central purpose of the amended air toxic provisions was to apply strict technology-based emission controls on HAPs.

See

,

e.g.

, H. Rep. No. 952, 101st Cong. 2d sess. 338. The floor's specific purpose was to assure that consideration of economic and other impacts not be used to “gut the standards. While costs are by no means irrelevant, they should by no means be the determining factors. There needs to be a minimum degree of control in relation to the control technologies that have already been attained by the best existing sources.” A Legislative History of the Clean Air Act

Vol. II

at 2897 (statement of Rep. Collins). An interpretation that the floor level of control must be limited by the performance of devices that only control some of these pollutants effectively “guts the standards” by including worse performers in the averaging process, whereas EPA's interpretation promotes the evident Congressional objective of having the floor reflect the average performance of best performing sources. Since Congress has not spoken to the precise question at issue, and the Agency's interpretation effectuates statutory goals and policies in a reasonable manner, its interpretation must be upheld.

See Chevron

v.

NRDC,

467 U.S. 837 (1984).

4

4

Since industry commenters argued that the statute can only be read to allow floors to be determined on a single source basis, commenters offered no view of why their reading could be viewed as reasonable in light of the statute's goals and objectives. It is not evident how any statutory goal is promoted by an interpretation that allows floors to be determined in a manner likely to result in floors reflecting emissions from worst or mediocre performers.

It is true that legislative history can sometimes be so clear as to give clear meaning to what is otherwise ambiguous statutory text. As just explained, EPA's HAP-by-HAP approach fulfills the evident statutory purpose and is supported by the most pertinent legislative history. A few industry commenters nonetheless indicated that a HAP-by-HAP approach is inconsistent with legislative history to section 112(d), citing to page 169 of the Senate Report. Since this Report was to a version of the bill which did not include a floor provision at all (much less the language at issue here), it is of no relevance.

National Lime II,

233 F. 3d at 638.

Industry commenters also noted that EPA retains the duty to investigate and, if justifiable, to adopt beyond the floor standards, so that potential least common denominator floors resulting from the whole facility approach would not have to “gut the standards.” That EPA may adopt more stringent standards based on what is “achievable” after considering costs and other factors is irrelevant to how EPA is required to set MACT floors. MACT floors must be based on the emission limitation achieved by the best performing 12 percent of existing sources, and, for new sources, on the level achieved by the best controlled similar source, and EPA must make this determination without consideration of cost. At best, standards reflecting a beyond-the-floor level of performance will have to be cost-justified; at worst, standards will remain at levels reflecting mediocre performance. Under either scenario, Congress' purpose in requiring floors is compromised.

EPA notes, however, that if optimized performance for different HAPs is not technologically possible due to mutually inconsistent control technologies (for example, metals performance decreases if organics reduction is optimized), then this would have to be taken into account by EPA in establishing a floor (or floors). The Senate Report indicates that if certain types of otherwise needed controls are mutually exclusive, EPA is to optimize the part of the standard providing the most environmental protection. S. Rep. No. 228, 101st Cong. 1st sess. 168 (although, as noted, the bill accompanying this Report contained no floor provisions). It should be

emphasized, however, that “the fact that no plant has been shown to be able to meet all of the limitations does not demonstrate that all the limitations are not achievable.”

Chemical Manufacturers Association

v.

EPA,

885 F. 2d at 264 (upholding technology-based standards based on best performance for each pollutant by different plants, where at least one plant met each of the limitations but no single plant met all of them).

All available data for boilers and process heaters indicate that there is no technical problem achieving the floor levels contained in this final rule for each HAP simultaneously, using the MACT floor technology. Data demonstrating a technical conflict in meeting all of the limits have not been provided, and, in addition, there are a number of units that meet all of the final existing source emission limits.

2. Minimum Number of Units To Set New Source Floors

Comment:

Many commenters indicated that section 112 requires that data from a minimum of 5 units is required to set MACT floors for existing sources. Commenters noted that EPA's use of less than 5 units for subcategories with greater than 30 units is a legalistic reading of section 112 that could result in such absurd results as using 5 units to set MACT floors for a subcategory with 29 units and data for only 10 units, but using a single unit to set MACT floors for a subcategory with 31 units and data for only 10 units.

Response:

EPA does not agree that section 112(d)(3) mandates a minimum of 5 sources in all instances, notwithstanding the incongruity of having less data to establish floors for larger source categories than is mandated for smaller ones. The literal language of the provision appears to compel this result. Section 112(d)(3) states that for categories and subcategories with at least 30 sources, the MACT floor for existing sources shall be no less stringent than the average emission limitation achieved by the best-performing twelve percent of the sources for which the Administrator has emissions information. The plain language of this provision requires that, for subcategories with at least 30 sources but where the Administrator only has emissions information on a small number of units, the floor can be no less stringent than the average emission limitation achieved by the best-performing twelve percent of those sources.

3. Treatment of Detection Levels

Comment:

When setting the MACT floors, non-detect values are present in many of the datasets from best performing units. Commenters provided input on how these non-detect values should be treated in the MACT floor analysis. Some commenters agreed that it is appropriate to keep the detection levels as reported; while certain commenters suggested that the detection levels should be replaced using a value of half the method detection limit (MDL). Many other commenters stated that data that are below the detection limit should not be used in setting the floors, and these data should be replaced with a higher value including either the MDL, limit of quantitation (LOQ), practical quantitation limit (PQL), or reporting limit (RL) for the purposes of the MACT floor calculations. Other commenters stated all non-detect values should be excluded from the floor analysis, or all values should be treated as 0. Some commenters stated it is necessary to keep the data as reported because changing values would lead to an upward bias. Additional commenters agreed with this basic premise, but suggested that replacing non-detect data with a value of half the MDL is appropriate while still minimizing the bias. They noted that treating measurements below the MDL as occurring at the MDL is statistically incorrect and violates the statute's “shall not be less stringent than” requirement for MACT floors. One commenter also provided a reference for a statistical method based on a log-normal distribution of the data which estimated the “maximum likelihood” of data values; this result is slightly higher than half the MDL. Some commenters stated that it is necessary to substitute the MDL value when performing the MACT floor calculations. With MDL defined as the lowest concentration that can be distinguished from the blank at a defined level of statistical significance, this is an appropriate value. If MDL values are not reported, one commenter suggested an approach for estimating an MDL equivalent value, but recognized that the background laboratory and test report files may not be available to EPA in order to derive these estimates. Most commenters representing industry and industry trade groups argued that either LOQ or PQL values should replace non-detects. The LOQ is defined as the smallest concentration of the analyte which can be measured. These commenters contended that the LOQ leads to a quantifiable amount of the substance with an acceptable level of uncertainty. A few commenters provided calculations showing some of the proposed MACT floors were below the LOQ. Additionally, some of these commenters stated that using LOQ or PQL values also incorporates additional sources of random and inherent sampling error throughout the testing process, which is necessary. These errors occur during sample collection, sample recovery, and sample analysis; MDL values only account for method specific (e.g., instrument) errors. These commenters contended that the three times the MDL approach discussed in the proposal accounts for some measurement errors but does not account for these unavoidable sampling errors. The commenters also noted that an LOQ is calculated as 3.18 times the MDL, and PQL is calculated as 5-10 times the MDL. Many of the commenters in support of using either an LOQ or PQL value ultimately believed a work practice is more appropriate where a MACT floor limit is below either of these two values. They cited 112(h)(1) which allows work practices under 112(h)(2) if “the application of measurement methodology to a particular class of sources is not practicable due to technological and economic limitations”. These commenters stated that the inability of sources to accurately measure a pollutant at the level of the MACT floor qualifies as such a technological limitation that warrants a work practice standard.

Where the proposed MACT floor is below the LOQ or PQL then that source category has a technological measurement limitation. A few commenters suggested RL values should be used when developing the floor limits. They stated that the RL is the lowest level at which the entire analytical system gives reliable signals and includes an acceptable calibration point. They added that use of an acceptable calibration point is critical in showing that numbers are real versus multiplying the MDL by various factors.

Several commenters stated that all non-detect values should be excluded from MACT floor calculations. They believed that excluding all non-detect values would eliminate any potential errors or accuracy issues related to testing for compliance. Due to inconsistencies of the MDL value reported for non-detect data, one commenter suggested treating all such values as zero. This would provide a consistent approach for setting the floor as well as determining compliance. Issues discussed by a multitude of commenters were that a wide range of detection limit values were reported and

that data from Phase I and Phase II information collection requests (ICR) are inconsistent. For all non-detect data, facilities participating in the Phase II ICR were instructed to report a detection limit, but this resulted in a variety of interpretations by the laboratories who reported data. As such, commenters provided examples where detected values were lower than non-detect values, and in some cases measured values were reported lower than typical method detection limits. Many of the commenters stated it is critical that EPA conduct a thorough quality review of the data to determine if non-detect values have been appropriately flagged and to normalize the data on a consistent basis. One commenter presented an example dataset and the potential implications of the treatment of non-detect data for Hg emissions in the biomass subcategory. This commenter noted that a number of the units with Phase I tests would no longer be considered top performers if their data were made consistent with the Phase II criteria. Several commenters provided remarks for EPA's proposed method of three times the MDL as an option for setting limits. A few commenters in support noted that this approach provided a reasonable method to account for data variability as it took into account more than just analytical instrument precision. Many other commenters argued that this method results in limits which are too low, namely that it is still lower than the LOQ value which they are in favor of as a substitute for any reported non-detect data. On the contrary, some other commenters disagreed with this method and claimed that it would lead to results which introduce a high bias in the floor setting process. A few contended that multiplying by 3 would introduce a 300 percent error into the floor, resulting in a floor that is less stringent than required by the Act. Others suggested that the MDL values are antiquated and already too high and thus it is not appropriate to multiply them by three. Also, a few commenters suggested multiplying the MDL by three would not reflect the actual lower emissions achieved by any source and as such is unlawful under section 112(d).

Response:

After consideration of the various comments related to treatment of detection limits in the development of MACT floors, EPA's approach for this final rule is as follows. While commenters suggested using values less than the MDL, such values have not been demonstrated to have been met during the corresponding test run. Therefore, EPA concluded that it is not appropriate, for development of MACT floors, to use any value less than the MDL. EPA also disagrees with comments that emission levels at or near the MDLs are appropriate levels to use for standard setting without consideration of measurement imprecision, because the actual performance of sources may differ significantly from the measured values or the MDL. Accordingly, for the boiler and process heater source category, which includes many sources with emission levels at or near the MDL for the various pollutants, EPA concluded that measurement imprecision was a significant factor that should be included in the development of emission limits. To determine an appropriate methodology, EPA examined the contribution of test method measurement imprecision to the variability of a set of emissions data. One element of variability is associated with method detection capabilities and a second is a function of the measurement value. Measurement imprecision is proportionally highest for values measured below or near a method's detection level and proportionally decreasing for values measured above the method detection level.

The probability procedures applied in calculating the floor or an emissions limit inherently and reasonably account for emissions data variability including measurement imprecision when the database represents multiple tests from multiple emissions units for which all of the data are measured significantly above the method detection level. That is less true when the database includes emissions occurring below method detection capabilities and are reported as the method detection level values.

EPA's guidance to respondents for reporting pollutant emissions used to support the data collection specified the criteria for determining test-specific method detection levels. Those criteria insure that there is only about a 1 percent probability of an error in deciding that the pollutant measured at the method detection level is present when in fact it was absent. Such a probability is also called a false positive or the alpha, Type I, error. Because of sample and emissions matrix effects, laboratory techniques, sample size, and other factors, method detection levels normally vary from test to test for any specific test method and pollutant measurement. The expected measurement imprecision for an emissions value occurring at or near the method detection level is about 40 to 50 percent. Pollutant measurement imprecision decreases to a consistent relative 10 to 15 percent for values measured at a level about three times the method detection level.

5

5

American Society of Mechanical Engineers,

Reference Method Accuracy and Precision (ReMAP): Phase 1, Precision of Manual Stack Emission Measurements,

CRTD Vol. 60, February 2001.

Also in accordance with our guidance, source owners identified emissions data which were measured below the method detection level and reported those values as equal to the method detection level as determined for that test. An effect of reporting data in this manner is that the resulting database is truncated at the lower end of the measurement range (i.e., no values reported below the test-specific method detection level). A floor or emissions limit based on a truncated database or otherwise including values measured near the method detection level may not adequately account for measurement imprecision contribution to the data variability. That is, an emission limit set based on the use of the MDL to represent data below the MDL may be significantly different than the actual levels achieved by the best performing units due to the imprecision of the measurements. This fact, combined with the low levels of emissions measured from many of the best performing units, led EPA to develop a procedure to account for the contribution of measurement imprecision to data variability.

We applied the following procedures to account for the effect of measurement imprecision associated with a database that includes method detection level data. The first step was to define a method detection level that is representative of the data used in establishing the floor or emissions limit and that also minimizes the influence of an outlier test-specific method detection level value. We reviewed each pollutant-specific data set to identify the highest test-specific method detection level reported that was also equal to or less than the average emissions level (

i.e.,

unadjusted for probability confidence level) calculated for the data set. We believe that this approach is representative of the data collected to develop the floor or emissions limit while to some degree minimizing the effect of a test(s) with an inordinately high method detection level (

e.g.,

the sample volume was too small, the laboratory technique was insufficiently sensitive, or the procedure for determining the detection level was other than that specified).

The second step in the process is to calculate three times the representative

method detection level

6

and compare that value to the calculated floor or emissions limit. If three times the representative method detection level were less than the calculated floor or emissions limit calculated from the upper prediction limit (UPL), we would conclude that measurement variability was adequately addressed because the measurement inprecision at that level is a consistent 10 to 15 percent. The calculated floor or emissions limit would need no adjustment. If, on the other hand, the value equal to three times the representative method detection level were greater than the UPL-based emission limit, we would conclude that the calculated floor or emission limit does not account entirely for measurement variability. If indicated, we substituted the value equal to three times the representative method detection level to apply as the adjusted floor or emissions limit. This adjusted value would ensure measurement variability is adequately addressed in the floor or the emissions limit.

6

Ibid.

In response to comments that EPA should have used the PQL, RL, or LOQ values in place of non-detect values, we disagree that use of those values is appropriate for calculating the MACT floors for two reasons. First, these terms are not defined statistically or consistently from method to method but are relatively arbitrary multiples (

e.g.,

3 times, 5 times, or 10 times) of the MDL. In some cases, a RL, LOQ, or PQL is a value determined based on a laboratory-specific procedure and not standardized by the method. We could not apply data arbitrarily adjusted or subject to laboratory-specific variables in establishing the floor. Second, we used a value equal to three times a representative MDL to compare with the floor and to adjust the applicable emissions limit, if necessary. We believe that using a value equal to three times the MDL sufficiently accounts for measurement uncertainty for the purposes of establishing compliance and there is no need to try to define or apply a PQL, LOQ, or RL for this purpose.

4. Instrument Span for CO

Comment:

Many commenters stated that the reported data and limits for CO are within the error range of analyzers and CO CEMS. For Method 10, the calibrated analyzers have an error of ±2 percent of the instrument span, with spans ranging from 50 parts per million (ppm) to 1000 ppm or greater. As such, at a minimum there is a potential error of 1 ppm to 20 ppm (2 percent of 50 ppm and 1000 ppm, respectively) while the liquid and other process gas categories have floor limits set at 1 ppm. Similarly, commenters noted that CO CEMS have an allowable drift of 5 percent of the span, with similar span ranges as Method 10. Commenters questioned the technical feasibility of complying with such low limits given the range in span values and suggested that EPA should review the data and establish more appropriate limits in consideration of measurement precision concerns.

Response:

EPA agrees with the comment that many of the CO measurements are within the error range of analyzers, and EPA has taken steps to mitigate the potential bias of such measurements. The resulting emission limits represent a level of performance that has been demonstrated to be achieved by the average of the best performing 12 percent of sources while considering variability introduced by imprecision of the CO analyzers. As explained below, our assessment indicated that the site-specific estimated measurement errors in some cases may be higher than some of the reported emissions levels. Therefore, for each emission test used in the MACT floor calculations we substituted the site-specific estimated measurement error for reported values below those values in order to ensure the quality of the data used to set the floors.

In response to the comments received, we reviewed the quality of the data relative to information provided for each emissions test. Method 10 is structured such that we can assess measurement data quality relative to the calibration span of the instrument (see

http://www.epa.gov/ttn/emc/promgate/method10r06.pdf

and

http://www.epa.gov/ttn/emc/promgate/method7E.pdf).

For example, the allowable calibration error, system bias, and drift requirements are directly proportional to the site-specific instrument calibration span (

i.e.,

± 2.0 percent of the calibration span value). For instrument calibration span values of 25 ppmv and less, the allowable calibration error, bias, or drift values are each ± 0.5 ppmv.

We can estimate the equivalent of the method detection level for a measurement with an instrumental test method (e.g., EPA Methods 3A, 6C, 7E, and 10) using a square root formula and these allowable data quality criteria. For example, in the case of a calibration span value of 25 ppmv, the square root formula (i.e., square root of the sum of the squares) would indicate a value of 0.9 ppmv. Consistent with the methodology we applied for non-instrumental methods, discussed in the previous comment response where we established limits no less than 3 times the MDL in order to avoid a large degree of measurement imprecision, this estimated measurement error value would translate to a limit of 3.0 ppmv (rounded up from 2.7 ppmv). For tests done with calibration spans of greater than 25 ppmv, the corresponding estimated measurement error would be greater. For example, the estimated measurement error using the square root formula for a calibration span of 100 ppmv would be about 4 ppmv which would translate to a limit of 12 ppmv. For a calibration span of 1000 ppmv, the estimated measurement error would be 35 ppmv or a limit of about 100 ppmv.

5. Achievability of Limits

Comment:

Several commenters were concerned that only small subsets of sources in each subcategory have emissions stack test data. These commenters added that less data means the pool from which the best performing 12 percent of the existing sources are drawn is smaller and, therefore, the actual number of sources used to determine the MACT floor is smaller. The commenters suggested that EPA should collect more data or provide assurances that the limited available data are representative for each subcategory. Commenters suggested that EPA could supplement testing data with “emissions information” such as fuel records, production records and associated emission factors, commercial warranties and guarantees.

Commenters raised concerns that existing units would have difficulty demonstrating compliance with the MACT floor limits. They suggested best performers with advanced air pollution control technologies should not be required to install additional add-on equipment to meet the emission limits. Commenters requested that EPA assess how many existing boilers and process heaters in each subcategory will be able to meet the standards without taking any further control measures. Several commenters contacted manufacturers regarding a retrofit project for their boilers and process heaters and they noted that manufacturers were unwilling to guarantee a retrofit would meet the limits.

Similarly, commenters raised concerns that new units would have even more difficulty demonstrating compliance with the MACT floor limits. These commenters had difficulty identifying a single source whose emissions testing data demonstrated they could achieve all of the MACT

floors for new sources in combination. Several commenters contacted boiler and process heater manufacturers; all were unable to offer commercial emissions guarantees that a new unit would meet the proposed limits. Some commenters raised concerns about the impacts of these stringent new unit floors including: Deterring sources from upgrading to new boilers as efficiency gains provided by a new unit would be offset by extensive controls and threatening fuel diversity.

Some commenters expressed concern that EPA had not properly evaluated whether there are technically feasible means of achieving the MACT floors. The commenters contended that the approach does not identify reasons why best performing sources achieve emissions levels reflected in the test data and they suggested that the intent of the MACT floor standard setting process is to discover effective control techniques so that other performers in the source category could emulate those techniques, reduce their emissions, and achieve similar emission levels. Commenters added that EPA has not adequately considered air pollution control device (APCD) conflicts with one another or compatibility of controls on certain boilers. Additionally, choosing to optimize controls for one pollutant may preclude optimization of controls for another pollutant e.g., minimizing CO in the combustion system is opposed to minimizing NO

X

in most boiler burners.

Response:

As mentioned elsewhere in this preamble, EPA is required to establish MACT floor levels based on emissions limits achieved by sources for which emissions information is available to the Administrator. EPA has revised the proposed MACT floors as well as the proposed subcategories, as explained above. EPA also examined several ways in which it might be able to use other types of emissions information in addition to actual emissions measurements. However, EPA concluded that there was no appropriate method of using different types of information in a manner that could be incorporated into the variability analyses. EPA first assessed the potential for estimating emissions for sources that lacked actual emissions data through the use of emission factors. However, the emission factors lack any degree of variability. Therefore, the use of such data in this rulemaking would have distorted the data variability in many cases, leading to standards that were more stringent than those developed using emissions data only and that likely underestimated actual variability. EPA also considered whether it could otherwise estimate emissions of sources that did not provide emissions data. However, EPA concluded that such estimations were not possible without the development of a technically appropriate approach to evaluate relevant information, and commenters did not provide any such approaches. EPA's approach provides MACT floors that are consistent with the requirements of section 112, because the floors are based on the average emissions performance of the best performers for which the Administrator has emissions information that is appropriate to use in setting the floors.

EPA agrees with commenters who note that many of the data sets are small. However, stakeholders were encouraged to provide additional data, and EPA significantly revised some of the proposed emission limits based on new test data. We received little or no additional data for some subcategories for which data sets were small at proposal. For all data sets, the final emission limits are based on the available data and reflect EPA's assessment of variability. Moreover, after consideration of the comments on the achievability of the emission limits, EPA performed additional analyses and detailed examinations of the data and developed revised limits that are based on what has been demonstrated to be achieved in practice. As described in more detail in the docket memorandum entitled “Revised MACT Floor Analysis (2011) for the Industrial, Commercial, and Institutional Boilers and Process Heaters National Emission Standards for Hazardous Air Pollutants—Major Source,” EPA has made adjustments to treatment of non-detect values, the statistical methodology, and monitoring requirements, and also incorporated new data and data corrections into our analyses. Accordingly, the final emission limits better reflect the performance of the MACT floor units than the proposed limits. EPA notes that for each subcategory, there are existing units that are meeting the MACT floor limits or are expected to meet the limits through application of available control technology.

Finally, in response to comments about low CO limits conflicting with a unit's ability to meet NO

X

requirements, EPA does not have specific information on the NO

X

limits and NO

X

emissions for most of the units that will be subject to the standard. However, the CO limits have been revised as discussed elsewhere in this preamble, and compliance is based on a full load test, while periods of startup and shutdown are subject to a work practice standard. To the extent that units cannot meet the CO floor and maintain NO

X

at the required level, oxidation catalysts can be used to reduce CO without an increase in NO

X

. EPA has included costs for these controls for many units in the cost analysis, although data on NO

X

requirements were not sufficient to allow NO

X

to be part of the analyses. Commenters did not provide any data supporting claims that any of the other emission limits or projected control devices would interfere with a source's ability to meet any of the other emission limits.

6. Comments on Technical Approaches

Comment:

Several commenters offered suggestions for adjusting the treatment of data from common stacks. Commenters suggested that it is improper to count the data twice if two boilers, in the same subcategory, exhaust through a common stack. A test conducted on the common stack does not represent the actual emissions from a single boiler, but rather reflects emissions from the combined simultaneous operation of the two boilers and their associated control device(s). The commenters contended that it is impossible to claim the test result would be exactly the same for each boiler and they added that if a common stack test turns out to be in the lowest 12 percent in a subcategory, counting it twice distorts the average of the best performers and skews the variability calculations. Commenters also noted that it is also not appropriate to divide emissions evenly between each boiler. Instead these commenters suggested that EPA use the data from common stacks only a single time in the MACT floor ranking and UPL calculations.

Response:

EPA's current approach is a reasonable approach for comingled emissions, particularly in light of the limited dataset available for some subcategories, because EPA can not accurately separate the fraction of the emissions that came from the combustion units and process emission points that are comingled in the same stack. Applying the emissions equally to multiple units exhausting through a common stack accurately represents the emissions of those units on average. Further, although the use of a data point twice may dampen variability, the inclusion of an extra unit in the floor has the opposite effect on the overall emission limit by increasing the denominator of the floor calculation. Either method could be used, but the results would not differ significantly. Furthermore, for existing sources, MACT cannot be less stringent than the average emission limitation achieved by

the best performing 12 percent of existing sources (for which emission information is available). If EPA ignored boilers that exhaust through a common stack, it would be ignoring available emissions information that is relevant to setting the MACT floor standards.

Comment:

Some commenters raised concerns that the MACT floor methodology doesn't adequately address the inherent variability with respect to operating conditions and control device performance. Operational variability can include warm-ups, shutdowns, load swings, and variations in fuel quality. They contended that emissions data relied upon in the proposal were produced during reference method performance testing under very limited operating conditions and with a very limited variation in potential fuel quality. Other commenters raised concerns that EPA has not properly acknowledged the impact of fuel quality on emissions. One commenter urged caution to EPA when considering variability to generate compliance margins that are palatable to industry; suggesting that this concept is not incorporated in the statute.

Response:

EPA is mindful of the need to account for sources' variability in assessing sources' performance when developing technology-based standards. EPA reviewed subcategory floor calculations in light of these comments and believes that the two-step MACT floor analysis process adequately addresses: (1) Performance testing variability and (2) fuel analysis variability estimations. EPA revised the MACT floor calculations in light of data submitted during and after the public comment period and also modified the approaches used at proposal for various aspects of the floor calculations.

EPA first took fuel into consideration, to the extent it is reflected in differences in boiler design, when we divided the source category into subcategories. EPA is aware that differences between given types of units, and fuel, can affect technical feasibility of applying emission control techniques, and has addressed this concern in the final rule. For a fuel based pollutant, such as PM, performance testing must be conducted under representative full load operating conditions, which, along with the parameter monitoring requirements, provides an assurance that the standards are being met at all times. For Hg and HCl, we modified the fuel based variability analysis in consideration of comments received on this approach. The first modification to the analysis was the introduction of a solid fuel subcategory, which includes any unit burning at least 10 percent, on an annual heat input basis, of any coal, fossil solid, biomass, or bio-based solid fuel. Given the wide variety in fuel types that compose the floor, the statistical analysis accounts for some of the inter-unit variability for different fuel types identified to be in the floor. The second modification was the development of a fuel variability factor (FVF). The FVF calculations were similar to the calculations used at proposal, but they were simplified to remove the control efficiency calculation and the method for identifying outliers in the data was also adjusted. The revised FVF analysis calculated a ratio for all fuel analysis data points for units in the top 12 percent for existing units and the top performing unit for new units in each subcategory. This ratio compared the reported fuel analysis data, converted to units of lb/MMBtu, to the emission test outlet data, converted to units of lb/MMBtu, during the stack tests. At proposal we conducted an outlier analysis of only the maximum ratios for each unit, but we revised the outlier analysis to consider all of the ratios from top performers within each subcategory. We then defined and identified outliers using the test of 3 times the standard deviation and 3 minus the standard deviation for all of the ratios in the subcategory. After removing outliers, the remaining maximum ratio for each subcategory was identified and multiplied by the 99 percent UPL.

For a discussion of how EPA considered other non-fuel variability operations, such as boiler load, see response to the comments provided under “What did we do with the CO Limits”.

Comment:

Several commenters argued that it is inappropriate to rank units according to the minimum stack test since any boiler can experience a good compliance test if conditions are favorable. Many of these commenters suggested that EPA should instead rank the data on the average of all stack tests. Another commenter suggested that the different emission levels achieved by different sources are just differences in performance and basing the ranking on the average would be more appropriate. This commenter suggested that at a minimum, the data used to rank and the data used as inputs into the MACT floor upper prediction limit calculation should be consistent.

Response:

In this final rule, EPA has reasonably determined that the best-controlled source is the source with the lowest stack test. EPA selected the lowest stack test as a measure of best performer because many units had only a single test available, and the comparison of average performance from two or more tests is not directly comparable to a single test measurement. However, all emission tests of acceptable quality were used to assess variability. As such, all data were considered in the floor analyses. EPA recognizes that each stack test data point represents a true assessment of the emissions for a combustor at a given point in time. However, where units had more than one test available, EPA also considers these other tests to be representative of the unit and relevant to assess run-to-run and test-to-test variability in the MACT floor UPL calculation. EPA did screen and remove certain test data from the MACT floor calculations if that data were not deemed representative of current operating conditions.

7. Statistical Approach

There were several comments made on specific aspects of the statistical variability analysis including suggestions for the appropriate confidence interval, appropriate statistic, and EPA's methods for determining the distribution of the dataset. The specific comments and EPA responses are outlined below.

Comment:

Industry, industry representatives, and environmental advocacy groups had different perspectives on the appropriateness of the proposed 99 percent UPL. Commenters from environmental advocacy groups requested a lower UPL with suggestions ranging between 50 to 95 percent. One commenter stated that EPA over-counts for the potential for future variability by using the 99 percent UPL for the entire data set and it does not adequately account for all variability, such as how unit maintenance and operator training may limit upward variability's effect on emission levels, and requests that EPA explain and justify the selection of the 99 percent UPL as opposed to the 90 or 95 percent UPL. Another commenter stated that most statistical analyses use 90 or 95 percent confidence intervals and prediction intervals. The commenter also claimed that 99 percent is overly conservative and results in twice as much HAP emissions and reduced health benefits compared to a lower UPL. Consequently the commenter stated a lower UPL would better withstand judicial review. One commenter mentioned that there is precedent for setting limits based on the 90th percentile and cited a 2006 analysis where EPA determined the best demonstrated technology, which found Hg reductions based on 90th percentile

and deemed the 90th percentile “reasonable” because of how compliance was to be determined and the high Hg content of the fuel used when the emissions data were collected. These commenters also suggested that EPA did not provide adequate rationale for selecting the 99th percentile instead of the 50th. These commenters noted that civil enforcement of environmental standards is based on a “preponderance of the evidence” which merely requires that a violation be more likely than not.

Commenters from industry and industry representatives advocated for a higher UPL. Commenters requested that EPA increase the UPL to 99.9 percent in order to better encompass unit emissions variability and represent a manageable risk. Industry, like environmental advocacy groups, also requested that EPA take into account operator training and its effect on emissions. The commenters claimed that operators are compelled to set emissions targets lower than limits to create a compliance margin which helps avoid violations and their consequences. Commenters also cited recent consideration of a 99.9 percent UPL in the proposed HMIWI MACT rule. Commenters claimed that since the HMIWI database consisted of a small dataset, it was unlikely full variability was observed and thus EPA had no valid statistical basis for the decisions to use 99 percent in the final HWIMI rule. The commenters suggested similar data limitations in the boiler dataset and argued that the 99.9 percent UPL should be used to allow more of a margin for all operating conditions and sample collection variation due to the limited data for the boiler MACT rule.

Response:

In this final rule, EPA has reasonably determined that 99 percent UPL is appropriate for fuel based HAP, and dioxin/furan, and a 99.9 percent UPL is appropriate for CO. For fuel-based HAP the 99 percent confidence level is consistent with other recent rulemakings. See 75 FR 54975. Many of the subcategories had limited data to establish the MACT floor calculations and EPA determined it was inappropriate to use a confidence level lower than 99 percent to set the standard because doing so would result in limits that the best performers would be expected to exceed, while this final rule requires that units meet the limits at all times. Finally, for the fuel-based pollutants, there are well established control measures currently used on units in the source category (fabric filters for PM and Hg and wet or dry scrubbers for HCl) that serve to mitigate, to some degree, the variability in emissions that can be expected. Given this additional consideration for fuel-based HAP, but recognizing the emission limits must be met at all times yet are based on short term stack test data, EPA selected the 99 percent confidence level. A lower confidence level would result in emission limits that even the best performing sources would be expected to exceed.

For CO, EPA considered several comments from industry and States, which provided both quantitative and qualitative comments on how CO emissions vary with load, fuel mixes and other routine operating conditions. After considering these comments EPA determined that a 99.9 percent confidence level for CO would better account for some of these fluctuations. While a good deal of CO data are available, at least for some of the subcategories, the data show highly variable emissions that can result from situations beyond the control of the operator, such as fuel moisture content after a rain event, elevated moisture in the air, and fuel feed issues or inconsistency in the fuel. The higher confidence level selected for CO is intended to reflect the high degree of variability in the emissions. For dioxin/furan, we also are maintaining the 99 percent UPL. Although much of the uncertainty associated with dioxin/furan testing will be mitigated by the requirement in EPA Method 23 to report non-detect values as zero for compliance purposes, the dioxin emission limits remain quite low and the 99 percent UPL provides a high degree of confidence that the best performing units will be able to meet the standards.

Comment:

Several commenters also addressed concerns with how EPA determined the distribution of the dataset. Many commenters stated that normal distribution theory has been incorrectly applied to positively skewed or log normally distributed emissions data. Based on this, commenters claimed that sample means, and consequently the 99 percent UPL calculation, were incorrectly determined. Commenters suggested that sample means should be computed based on the arithmetic mean of lognormal distribution. One commenter requested that EPA consider using non-normal distributions or non-parametric methods in the analysis. Two commenters noted that the technique used by EPA based on logarithmic transformation underestimates the prediction limit for the mean and requested that EPA use the 2004 Bhaumik and Gibbons procedure for computing the UPL for log-normally distributed data. Three commenters stated that EPA is not following its own guidance document,

Data Quality Assessment: Statistical Methods for Practitioners EPA QA/G-9S,

for determining whether or not a data set is normally distributed and should explain the reasons for not doing so. The commenters then go on to request that EPA follow its guidance documents which recommend use other tests aside from the skewness and kurtosis tests when data are limited or if critical test values are not available.

Response:

EPA appreciates the detailed suggestions for alternative approaches to determine the dataset and it has revised its default selection of data distributions consistent with its guidance document

Data Quality Assessment: Statistical Methods for Practitioners EPA QA/G-9S.

This document indicates that most environmental data are lognormally distributed, so EPA has modified its assumptions when the results of the skewness and kurtosis tests result in a tie, or when there are not enough data to complete the skewness and kurtosis tests. Some of the commenters suggested that more advanced tests are necessary to determine the dataset, such as the Shapiro-Wilkes test. These tests needs a sample size of 50 or more, and would not be appropriate for many of the small sample sizes used to compute the MACT floor UPL.

With respect to the methods used to compute the UPL for a dataset that is determined to be lognormally distributed, EPA also considered the commenters suggested revisions to the calculations in order to avoid skewing the UPL by calculating the UPL of an arithmetic mean instead of the UPL of a geometric mean. To adjust the calculation EPA considered a scale bias correction approach as well as a new UPL equation based on a Bhaumik and Gibbons 2004 paper, which calculates “An Upper Prediction Limit for the Arithmetic Mean of a Lognormal Random Variable”. Given data availability, EPA selected the Bhaumik and Gibbons 2004 approach which addresses commenters concerns with the proposed computations.

Comment:

Several commenters suggested alternatives to the UPL statistics such as upper tolerance limit (UTL), upper limit (UL) and upper confidence limit (UCL). Several commenters stated that EPA's UPL calculation was flawed and did not fully account for variability. Commenters then suggested that if the proposed UPL approach was maintained EPA should adopt the modified UPL equation in the Portland cement NESHAP. Commenters argued that this statistic would

represent floors achieved in practice and account for total variability instead of EPA's proposed UPL statistic based on sample variability. Several commenters claimed the data set was limited and suggested that EPA should use the UTL when data available do not represent the entire population. One commenter claimed that the upper UCL used in the HMIWI MACT rule was not a true prediction limit because it did not adjust the standard deviation for the number of test runs in the future compliance average and it should not be used in the boiler MACT rule.

Response:

EPA considered these comments and reviewed each of the separate statistics. Because statistics is a tool and many statistical approaches could be considered valid, EPA considered the comments and adjusted the approach used to provide a reasonable and technically correct statistical methodology. MACT floors for existing sources must reflect the average emission limitation achieved by the best-performing 12 percent of existing sources. As explained below, only the UCL and UPL adequately get at the notion of average emissions. Use of the UPL is also consistent with other recent rulemakings.

See

75 FR 54975.

In general, confidence intervals are used to quantify one's knowledge of a parameter or some other characteristic of a population based on a random sample from that population. The most frequently used type of confidence interval is the one that contains the population mean. Given this definition, the 99 percent UCL represents the value which we can expect the mean of the population to fall below 99 percent of the time in repeated sampling. Whereas a confidence interval covers a population parameter with a stated confidence, that is, a certain proportion of the time, there is also a way to cover a fixed proportion of the population with a stated confidence. Such an interval is called a tolerance interval. Confidence limits are limits within which we expect a given population parameter, such as the mean, to lie. Statistical tolerance limits are limits within which we expect a stated proportion of the population to lie. Given these definitions, the 99 percent UTL represents the value which we can expect 99 percent of the measurements to fall below 99 percent of the time in repeated sampling. In other words, if we were to obtain another set of emission observations from the five sources, we can be 99 percent confident that 99 percent of these measurements will fall below a specified level. Since you must calculate the sample percentile, and the sample sizes for the boiler MACT floor data are small, the 99th percentile is underestimated. The UTL should only be used where one can calculate a sample percentile,

e.g.,

where there is a sample size of at least 100, and we do not have that many sources represented in any MACT floor.

In contrast to a confidence interval or a tolerance interval, a prediction interval for a future observation is an interval that will, with a specified degree of confidence, contain the next (or some other pre-specified) randomly selected observation from a population. In other words, the prediction interval estimates what future values will be, based upon present or past background samples taken. Given this definition, the UPL represents the value which we can expect the mean of 3 future observations (3-run average) to fall below, based upon the results of the independent sample of size n from the same population. Finally, the upper limit (UL) is roughly equivalent to the percentile of the actual data distribution for the sample. The UL does not have a robust statistical foundation. Basically, the UL formulation assumes that the data: (1) Represent the population rather than a random sample from that population, and (2) are normally distributed. The data used to develop the MACT floors for this rule do not represent the entire population for any subcategory, and most of the data sets are not normally distributed. For these reasons, EPA concluded that it is not appropriate to use the UL in setting the MACT floor limits.

Comment:

Some commenters suggested that EPA's UPL approach fails to accomplish predicting the level of performance achieved by the best performing sources under all operating conditions, not because of a poor statistical framework but because of an inadequate database. These commenters added that as a result, the inputs into the UPL equations are not representative of a distribution of values that reflect all operating conditions.

Response:

Section 112(d) of the Act requires EPA to base MACT floor standards for existing sources on the average emission limitation achieved by the best performing 12 percent of existing sources for which EPA has emissions information. EPA has incorporated new data and data corrections received during the public comment period. EPA also has considered the requests for further subcategorization of the source category in light of limits on the dataset that caution against over-partitioning of the database. The revised analysis is based on all emission stack test data of appropriate quality available to EPA, and the UPL approach provides as complete a picture of variability as possible given the limited data available.

Comment:

Some commenters questioned whether the statistical approach met EPA's legal obligations under Section 112 of the CAA. One commenter stated that in order to withstand judicial review, the UPL should be calculated based on the best 6 percent of sources instead of the best 12 percent in order to establish a floor that would require 94 percent of sources to reduce emissions. One commenter stated that the courts did not endorse the proposed UPL procedure and that its appropriateness should be reviewed. The commenter goes on to say that on a statistical and technical basis, the UPL procedure is antithetical to the instruction in Section 112(d)(3)(A) and contradicts the strong endorsement of the high floor implementation as the best reading of the statutory language.

Response:

While the commenter is correct that the entire MACT floor data pool was used in the calculation of the UPL, EPA notes that statistics is a tool that is used to estimate variability and it is entirely appropriate to consider the variability within the best forming 12 percent of sources in developing emission limits based on the average performance of those sources. As far as the concept that the floors should require 94 percent of the sources to reduce emissions, that is not what is required by the statute. Rather, the statute requires that the MACT floor standards for existing sources be no less stringent than the average emission limitation achieved by the best performing 12 percent of existing sources for which EPA has emissions information. For example, if a category had 100 units and the performance of the best 50 of those units was the same, the emission limits would be based on those 50 units and they all would be projected to meet the limits. While this is a hypothetical scenario, it illustrates that there is no specific percentage of sources that must reduce emissions in order for the MACT floor limits to be consistent with the statutory requirement.

Comment:

One commenter suggested that EPA should incorporate different statistical methods according to the amount and type of data available in each subcategory instead of a one-size-fits-all approach. This commenter also suggested that the approach taken by EPA must be validated by looking at the result it creates and examining whether the end result is reasonable. The commenter suggested applying a simple test to identify whether the resulting

floor requires a substantial majority of each subcategory to make some degree of emission reduction.

Response:

EPA has revised its statistical approach to include a mixed use of confidence levels, as discussed above, as well as a mix of statistical tools to consider the distribution of the datasets and what types of data are used as inputs into the floor analysis. For example, the MACT floor computations for Hg emissions from liquid fuel units were modified to consider data from both fuel analysis and stack test results. EPA appreciates the suggestion for validating the results of the statisti

This text is long and has been trimmed here. Open the source document for the complete record.