[Federal Register Volume 64, Number 189 (Thursday, September 30, 1999)]
[Rules and Regulations]
[Pages 52828-53077]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 99-20430]



[[Page 52827]]

_______________________________________________________________________

Part II





Environmental Protection Agency





_______________________________________________________________________



40 CFR Part 60, et al.



NESHAPS: Final Standards for Hazardous Air Pollutants for Hazardous 
Waste Combustors; Final Rule

Federal Register / Vol. 64, No. 189 / Thursday, September 30, 1999 / 
Rules and Regulations

[[Page 52828]]



ENVIRONMENTAL PROTECTION AGENCY

40 CFR Parts 60, 63, 260, 261, 264, 265, 266, 270, and 271

[FRL-6413-3]
RIN 2050-AEO1


NESHAPS: Final Standards for Hazardous Air Pollutants for 
Hazardous Waste Combustors

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: We are promulgating revised standards for hazardous waste 
incinerators, hazardous waste burning cement kilns, and hazardous waste 
burning lightweight aggregate kilns. These standards are being 
promulgated under joint authority of the Clean Air Act (CAA) and 
Resource Conservation and Recovery Act (RCRA). The standards limit 
emissions of chlorinated dioxins and furans, other toxic organic 
compounds, toxic metals, hydrochloric acid, chlorine gas, and 
particulate matter. These standards reflect the performance of Maximum 
Achievable Control Technologies (MACT) as specified by the Clean Air 
Act. These MACT standards also will result in increased protection to 
human health and the environment over existing RCRA standards.

DATES: This final rule is in effect on September 30, 1999. You are 
required to be in compliance with these promulgated standards 3 years 
following the effective date of the final rule (i.e., September 30, 
2002). You are provided with the possibility of a site-specific one 
year extension for the installation of controls to comply with the 
final standards or for waste minimization reductions. The incorporation 
by reference of certain publications listed in the rule was approved by 
the Director of the Federal Register as of September 30, 1999.

ADDRESSES: The official record (i.e., public docket) for this 
rulemaking is identified as Docket Numbers: F-96-RCSP-FFFFF, F-97-CS2A-
FFFFF, F-97-CS3A-FFFFF, F-97-CS4A-FFFFF, F-97-CS5A-FFFFF, F-97-CS6A-
FFFFF, F-98-RCSF-FFFFF, and F-1999-RC2F-FFFFF. The official record is 
located in the RCRA Information Center (RIC), located at Crystal 
Gateway One, 1235 Jefferson Davis Highway, First Floor, Arlington, 
Virginia. The mailing address for the official record is RCRA 
Information Center, Office of Solid Waste (5305W), U.S. Environmental 
Protection Agency Headquarters, 401 M Street, SW, Washington, DC 20460.
    Public comments and supporting materials are available for viewing 
in the RIC. The RIC is open from 9 a.m. to 4 p.m., Monday through 
Friday, excluding federal holidays. To review docket materials, you 
must make an appointment by calling 703-603-9230 or by sending a 
message via e-mail to: RCRA-D[email protected]. You may copy a 
maximum of 100 pages from any regulatory docket at no charge. 
Additional copies cost 15 cent/page. The index for the official record 
and some supporting materials are available electronically. See the 
``Supplementary Information'' section of this Federal Register notice 
for information on accessing the index and these supporting materials.

FOR FURTHER INFORMATION CONTACT: For general information, you can 
contact the RCRA Hotline at 1-800-424-9346 or TDD 1-800-553-7672 
(hearing impaired). In the Washington metropolitan area, call 703-412-
9810 or TDD 703-412-3323. For additional information on the Hazardous 
Waste Combustion MACT rulemaking and to access available electronic 
documents, please go to our Web page: www.epa.gov/hwcmact. Any 
questions or comments on this rule can also be sent to EPA via our Web 
page.
    For more detailed information on technical requirements of this 
rulemaking, you can contact Mr. David Hockey, 703-308-8846, electronic 
mail: Hockey.D[email protected]. For more detailed information on 
permitting associated with this rulemaking, you can contact Ms. 
Patricia Buzzell, 703-308-8632, electronic mail: 
Buzzell.T[email protected]. For more detailed information on 
compliance issues associated with this rulemaking, you can contact Mr. 
Larry Gonzalez, 703-308-8468, electronic mail: 
Gonzalez.L[email protected]. For more detailed information on the 
assessment of potential costs, benefits and other impacts associated 
with this rulemaking, you can contact Mr. Lyn Luben, 703-308-0508, 
electronic mail: Luben.L[email protected]. For more detailed 
information on risk analyses associated with this rulemaking, you can 
contact Mr. David Layland, 703-308-0482, electronic mail: 
Layland.D[email protected].

SUPPLEMENTARY INFORMATION:
    Official Record. The official record is the paper record maintained 
at the address in ADDRESSES above. All comments that were received 
electronically were converted into paper form and placed in the 
official record, which also includes all comments submitted directly in 
writing. Our responses to comments, whether the comments are written or 
electronic, are located in the response to comments document in the 
official record for this rulemaking.
    Supporting Materials Availability on the Internet. The index for 
the official record and the following supporting materials are 
available on the Internet as:
--Technical Support Documents for HWC MACT Standards:
    --Volume I: Description of Source Categories
    --Volume II: HWC Emissions Database
    --Volume III: Selection of MACT Standards and Technologies
    --Volume IV: Compliance with the MACT Standards
    --Volume V: Emission Estimates and Engineering Costs
--Assessment of the Potential Costs, Benefits and Other Impacts of the 
Hazardous Waste Combustion MACT Standards--Final Rule
--Risk Assessment Support to the Development of Technical Standards for 
Emissions from Combustion Units Burning Hazardous Wastes: Background 
Information Document
--Response to Comments for the HWC MACT Standards Document

    To access the information electronically from the World Wide Web 
(WWW), type: www.epa.gov/hwcmact

Outline

Acronyms Used in the Rule

acfm--Actual cubic feet per minute
BIF--Boilers and industrial furnaces
CAA--Clean Air Act
CEMS--Continuous emissions monitors/monitoring system
CFR--Code of Federal Regulations
DOC--Documentation of Compliance
DRE--Destruction and Removal Efficiency
dscf--Dry standard cubic foot
dscm--Dry standard cubic meter
EPA/USEPA--United States Environmental Protection Agency gr--Grains
HSWA--Hazardous and Solid Waste Amendments
kg--Kilogram
MACT--Maximum Achievable Control Technology
mg--Milligrams
Mg--Megagrams (metric tons)
NOC--Notification of Compliance
NESHAP--National Emission Standards for HAPs
ng--Nanograms
NODA--Notice of Data Availability
NPRM--Notice of Proposed Rulemaking
POHC--Principal Organic Hazardous Constituent

[[Page 52829]]

ppmv--Parts per million by volume
ppmw--Parts per million by weight
RCRA--Resource Conservation and Recovery Act
R & D--Research and Development
SSRA--Site specific risk assessment
TEQ--Toxicity equivalence
g--Micrograms

Outline

Part One: Overview and Background for This Rule
    I. What Is the Purpose of This Rule?
    II. In Brief, What Are the Major Features of Today's Rule?
    A. Which Source Categories Are Affected By This Rule?
    B. How Are Area Sources Affected By This Rule?
    C. What Emission Standards Are Established In This Rule?
    D. What Are the Procedures for Complying with This Rule?
    E. What Subsequent Performance Testing Must Be Performed?
    F. What Is the Time Line for Complying with This Rule?
    G. How Does This Rule Coordinate With the Existing RCRA 
Regulatory Program?
    III. What Is the Basis of Today's Rule?
    IV. What Was the Rulemaking Process for Development of This 
Rule?
Part Two: Which Devices Are Subject to Regulation?
    I. Hazardous Waste Incinerators
    II. Hazardous Waste Burning Cement Kilns
    III. Hazardous Waste Burning Lightweight Aggregate Kilns
Part Three: How Were the National Emission Standards for Hazardous 
Air Pollutants (NESHAP) in This Rule Determined?
    I. What Authority Does EPA Have to Develop a NESHAP?
    II. What Are the Procedures and Criteria for Development of 
NESHAPs?
    A. Why Are NESHAPs Needed?
    B. What Is a MACT Floor?
    C. How Are NESHAPs Developed?
    III. How Are Area Sources and Research, Development, and 
Demonstration Sources Treated in this Rule?
    A. Positive Area Source Finding for Hazardous Waste Combustors
    1. How Are Area Sources Treated in this Rule?
    2. What Is an Area Source?
    3. What Is the Basis for Today's Positive Area Source Finding?
    B. How Are Research, Development, and Demonstration (RD&D) 
Sources Treated in this Rule?
    1. Why Does the CAA Give Special Consideration to Research and 
Development (R&D) Sources?
    2. When Did EPA Notice Its Intent to List R&D Facilities?
    3. What Requirements Apply to Research, Development, and 
Demonstration Hazardous Waste Combustor Sources?
    IV. How Is RCRA's Site-Specific Risk Assessment Decision Process 
Impacted by this Rule?
    A. What Is the RCRA Omnibus Authority?
    B. How Will the SSRA Policy Be Applied and Implemented in Light 
of this Mandate?
    1. Is There a Continuing Need for Site-Specific Risk 
Assessments?
    2. How Will the SSRA Policy Be Implemented?
    C. What Is the Difference Between the RCRA SSRA Policy and the 
CAA Residual Risk Requirement?
Part Four: What Is The Rationale for Today's Final Standards?
    I. Emissions Data and Information Data Base
    A. How Did We Develop the Data Base for this Rule?
    B. How Are Data Quality and Data Handling Issues Addressed?
    1. How Are Data from Sources No Longer Burning Hazardous Waste 
Handled?
    2. How Are Nondetect Data Handled?
    3. How Are Normal Versus Worst-Case Emissions Data Handled?
    4. What Approach Was Used to Fill In Missing or Unavailable 
Data?
    II. How Did We Select the Pollutants Regulated by This Rule?
    A. Which Toxic Metals Are Regulated by This Rule?
    1. Semivolatile and Low Volatile Metals
    2. How Are the Five Other Metal Hazardous Air Pollutants 
Regulated?
    B. How Are Toxic Organic Compounds Regulated By This Rule?
    1. Dioxins/Furans
    2. Carbon Monoxide and Hydrocarbons
    3. Destruction and Removal Efficiency
    C. How Are Hydrochloric Acid and Chlorine Gas Regulated By This 
Rule?
    III. How Are the Standards Formatted In This Rule?
    A. What Are the Units of the Standards?
    B. Why Are the Standards Corrected for Oxygen and Temperature?
    C. How Does the Rule Treat Significant Figures and Rounding?
    IV. How Are Nondioxin/Furan Organic Hazardous Air Pollutants 
Controlled?
    A. What Is the Rationale for DRE as a MACT Standard?
    1. MACT DRE Standard
    2. How Can Previous Successful Demonstrations of DRE Be Used To 
Demonstrate Compliance?
    3. DRE for Sources that Feed Waste at Locations Other Than the 
Flame Zone
    4. Sources that Feed Dioxin Wastes
    B. What Is the Rationale for Carbon Monoxide or Hydrocarbon 
Standards as Surrogate Control of Organic Hazardous Air Pollutants?
    V. What Methodology Is Used to Identify MACT Floors?
    A. What Is the CAA Statutory Requirement to Identify MACT 
Floors?
    B. What Is the Final Rule Floor Methodology?
    1. What Is the General Approach Used in this Final Rule?
    2. What MACT Floor Approach Is Used for Each Standard?
    C. What Other Floor Methodologies Were Considered?
    1. April 19, 1996 Proposal
    2. May 1997 NODA.
    D. How Is Emissions Variability Accounted for in Development of 
Standards?
    1. How Is Within-Test Condition Emissions Variability Addressed?
    2. How Is Waste Imprecision in the Stack Test Method Addressed?
    3. How Is Source-to-Source Emissions Variability Addressed?
    VI. What Are the Standards for Existing and New Incinerators?
    A. To Which Incinerators Do Today's Standards Apply?
    B. What Subcategorization Options Did We Evaluate?
    C. What Are the Standards for New and Existing Incinerators?
    1. What Are the Standards for Incinerators?
    2. What Are the Standards for Dioxins and Furans?
    3. What Are the Standards for Mercury?
    4. What Are the Standards for Particulate Matter?
    5. What Are the Standards for Semivolatile Metals?
    6. What Are the Standards for Low Volatile Metals?
    7. What Are the Standards for Hydrochloric Acid and Chlorine 
Gas?
    8. What Are the Standards for Carbon Monoxide?
    9. What Are the Standards for Hydrocarbon?
    10. What Are the Standards for Destruction and Removal 
Efficiency?
    VII. What Are the Standards for Hazardous Waste Burning Cement 
Kilns?
    A. To Which Cement Kilns Do Today's Standards Apply?
    B. How Did EPA Initially Classify Cement Kilns?
    1. What Is the Basis for a Separate Class Based on Hazardous 
Waste Burning?
    2. What Is the Basis for Differences in Standards for Hazardous 
Waste and Nonhazardous Waste Burning Cement Kilns?
    C. What Further Subcategorization Considerations Are Made?
    D. What Are The Standards for Existing and New Cement Kilns?
    1. What Are the Standards for Cement Kilns?
    2. What Are the Dioxin and Furan Standards?
    3. What Are the Mercury Standards?
    4. What Are the Particulate Matter Standards?
    5. What Are the Semivolatile Metals Standards?
    6. What Are the Low Volatile Metals Standards?
    7. What Are the Hydrochloric Acid and Chlorine Gas Standards?
    8. What Are the Hydrocarbon and Carbon Monoxide Standards for 
Kilns Without By-Pass Sampling Systems?
    9. What Are the Carbon Monoxide and Hydrocarbon Standards for 
Kilns With By-Pass Sampling Systems?
    10. What Are the Destruction and Removal Efficiency Standards?
    VIII. What Are the Standards for Existing and New Hazardous 
Waste Burning Lightweight Aggregate Kilns?
    A. To Which Lightweight Aggregate Kilns Do Today's Standards 
Apply?
    B. What Are the Standards for New and Existing Hazardous Waste 
Burning Lightweight Aggregate Kilns?
    1. What Are the Standards for Lightweight Aggregate Kilns?

[[Page 52830]]

    2. What Are the Dioxin and Furan Standards?
    3. What Are the Mercury Standards?
    4. What Are the Particulate Matter Standards?
    5. What Are the Semivolatile Metals Standards?
    6. What Are the Low Volatile Metals Standards?
    7. What Are the Hydrochloric Acid and Chlorine Gas Standards?
    8. What Are the Hydrocarbon and Carbon Monoxide Standards?
    9. What Are the Standards for Destruction and Removal 
Efficiency?
Part Five: Implementation
    I. How Do I Demonstrate Compliance with Today's Requirements?
    A. What Sources Are Subject to Today's Rules?
    1. What Is an Existing Source?
    2. What Is a New Source?
    B. How Do I Cease Being Subject to Today's Rule?
    C. What Requirements Apply If I Temporarily Cease Burning 
Hazardous Waste?
    1. What Must I Do to Comply with Alternative Compliance 
Requirements?
    2. What Requirements Apply If I Do Not Use Alternative 
Compliance Requirements?
    D. What Are the Requirements for Startup, Shutdown and 
Malfunction Plans?
    E. What Are the Requirements for Automatic Waste Feed Cutoffs?
    F. What Are the Requirements of the Excess Exceedance Report?
    G. What Are the Requirements for Emergency Safety Vent Openings?
    H. What Are the Requirements for Combustion System Leaks?
    I. What Are the Requirements for an Operation and Maintenance 
Plan?
    II. What Are the Compliance Dates for this Rule?
    A. How Are Compliance Dates Determined?
    B. What Is the Compliance Date for Sources Affected on April 19, 
1996?
    C. What Is the Compliance Date for Sources That Become Affected 
After April 19, 1996?
    III. What Are the Requirements for the Notification of Intent to 
Comply?
    IV. What Are the Requirements for Documentation of Compliance?
    A. What Is the Purpose of the Documentation of Compliance?
    B. What Is the Rationale for the DOC?
    C. What Must Be in the DOC?
    V. What Are the Requirements for MACT Performance Testing?
    A. What Are the Compliance Testing Requirements?
    1. What Are the Testing and Notification of Compliance 
Schedules?
    2. What Are the Procedures for Review and Approval of Test Plans 
and Requirements for Notification of Testing?
    3. What Is the Provision for Time Extensions for Subsequent 
Performance Tests?
    4. What Are the Provisions for Waiving Operating Parameter 
Limits During Subsequent Performance Tests?
    B. What Is the Purpose of Comprehensive Performance Testing?
    1. What Is the Rationale for the Five Year Testing Frequency?
    2. What Operations Are Allowed During a Comprehensive 
Performance Test?
    3. What Is the Consequence of Failing a Comprehensive 
Performance Test?
    C. What Is the Rationale for Confirmatory Performance Testing?
    1. Do the Comprehensive Testing Requirements Apply to 
Confirmatory Testing?
    2. What Is the Testing Frequency for Confirmatory Testing?
    3. What Operations Are Allowed During Confirmatory Performance 
Testing?
    4. What Are the Consequences of Failing a Confirmatory 
Performance Test?
    D. What Is the Relationship Between the Risk Burn and 
Comprehensive Performance Test?
    1. Is Coordinated Testing Allowed?
    2. What Is Required for Risk Burn Testing?
    E. What Is a Change in Design, Operation, and Maintenance?
    F. What are the Data In Lieu Allowances?
    VI. What Is the Notification of Compliance?
    A. What Are the Requirements for the Notification of Compliance?
    B. What Is Required in the NOC?
    C. What Are the Consequences of Not Submitting a NOC?
    D. What Are the Consequences of an Incomplete Notification of 
Compliance?
    E. Is There a Finding of Compliance?
    VII. What Are the Monitoring Requirements?
    A. What Is the Compliance Monitoring Hierarchy?
    B. How Are Comprehensive Performance Test Data Used to Establish 
Operating Limits?
    1. What Are the Definitions of Terms Related to Monitoring and 
Averaging Periods?
    2. What Is the Rationale for the Averaging Periods for the 
Operating Parameter Limits?
    3. How Are Performance Test Data Averaged to Calculate Operating 
Parameter Limits?
    4. How Are the Various Types of Operating Parameters Monitored 
or Established?
    5. How Are Rolling Averages Calculated Initially, Upon 
Intermittent Operations, and When the Hazardous Waste Feed Is Cut 
Off?
    6. How Are Nondetect Performance Test Feedstream Data Handled?
    C. Which Continuous Emissions Monitoring Systems Are Required in 
the Rule?
    1. What Are the Requirements and Deferred Actions for 
Particulate Matter CEMS?
    2. What Are the Test Methods, Specifications, and Procedures?
    3. What Is the Status of Total Mercury CEMS?
    4. What Is the Status of the Proposed Performance Specifications 
for Multimetal, Hydrochloric Acid, and Chlorine Gas CEMS?
    5. How Have We Addressed Other Issues: Continuous Samplers as 
CEMS, Averaging Periods for CEMS, and Incentives for Using CEMS?
    D. What Are the Compliance Monitoring Requirements?
    1. What Are the Operating Parameter Limits for Dioxin/Furan?
    2. What Are the Operating Parameter Limits for Mercury?
    3. What Are the Operating Parameter Limits for Semivolatile and 
Low Volatile Metals?
    4. What Are the Monitoring Requirements for Carbon Monoxide and 
Hydrocarbon?
    5. What Are the Operating Parameter Limits for Hydrochloric 
Acid/Chlorine Gas?
    6. What Are the Operating Parameter Limits for Particulate 
Matter?
    7. What Are the Operating Parameter Limits for Destruction and 
Removal Efficiency?
    VIII. Which Methods Should Be Used for Manual Stack Tests and 
Feedstream Sampling and Analysis?
    A. Manual Stack Sampling Test Methods
    B. Sampling and Analysis of Feedstreams
    IX. What Are the Reporting and Recordkeeping Requirements?
    A. What Are the Reporting Requirements?
    B. What Are the Recordkeeping Requirements?
    C. How Can You Receive Approval to Use Data Compression 
Techniques?
    X. What Special Provisions Are Included in Today's Rule?
    A. What Are the Alternative Standards for Cement Kilns and 
Lightweight Aggregate Kilns?
    1. What Are the Alternative Standards When Raw Materials Cause 
an Exceedance of an Emission Standard?
    2. What Special Provisions Exist for an Alternative Mercury 
Standard for Kilns?
    B. Under What Conditions Can the Performance Testing 
Requirements Be Waived?
    1. How Is This Waiver Implemented?
    2. How Are Detection Limits Handled Under This Provision?
    C. What Other Waiver Was Proposed, But Not Adopted?
    D. What Equivalency Determinations Were Considered, But Not 
Adopted?
    E. What are the Special Compliance Provisions and Performance 
Testing Requirements for Cement Kilns with In-line Raw Mills and 
Dual Stacks?
    F. Is Emission Averaging Allowable for Cement Kilns with Dual 
Stacks and In-line Raw Mills?
    1. What Are the Emission Averaging Provisions for Cement Kilns 
with In-line Raw Mills?
    2. What Emission Averaging Is Allowed for Preheater or 
Preheater-Precalciner Kilns with Dual Stacks?
    G. What Are the Special Regulatory Provisions for Cement Kilns 
and Lightweight Aggregate Kilns that Feed Hazardous Waste at a 
Location Other Than the End Where Products Are Normally Discharged 
and Where Fuels Are Normally Fired?
    H. What is the Alternative Particulate Matter Standard for 
Incinerators?

[[Page 52831]]

    1. Why is this Alternative Particulate Matter Standard 
Appropriate under MACT?
    2. How Do I Demonstrate Eligibility for the Alternative 
Standard?
    3. What is the Process for the Alternative Standard Petition?
    XI. What Are the Permitting Requirements for Sources Subject to 
this Rule?
    A. What Is the Approach to Permitting in this Rule?
    1. In General What Was Proposed and What Was Commenters' 
Reaction?
    2. What Permitting Approach Is Adopted in Today's Rule?
    3. What Considerations Were Made for Ease of Implementation?
    B. What Is the Applicability of the Title V and RCRA Permitting 
Requirements?
    1. How Are the Title V Permitting Requirements Applicable?
    2. What Is the Relationship Between the Notification of 
Compliance and the Title V Permit?
    3. Which RCRA Permitting Requirements Are Applicable?
    4. What Is the Relationship of Permit Revisions to RCRA 
Combustion Permitting Procedures?
    5. What is the Relationship to the RCRA Preapplication Meeting 
Requirements?
    C. Is Title V Permitting Applicable to Area Sources?
    D. How will Sources Transfer from RCRA to MACT Compliance and 
Title V Permitting?
    1. In General, How Will this Work?
    2. How Will I Make the Transition to CAA Permits?
    3. When Should RCRA Permits Be Modified?
    4. How Should RCRA Permits Be Modified?
    5. How Should Sources in the Process of Obtaining RCRA Permits 
be Switched Over to Title V?
    E. What is Meant by Certain Definitions?
    1. Prior Approval
    2. 50 Percent Benchmark
    3. Facility Definition
    4. No New Eligibility for Interim Status
    5. What Constitutes Construction Requiring Approval?
    XII. State Authorization
    A. What is the Authority for Today's Rule?
    B. How is the Program Delegated Under the Clean Air Act?
    C. How are States Authorized Under RCRA?
Part Six: Miscellaneous Provisions and Issues
    I. Does the Waiver of the Particulate Matter Standard or the 
Destruction and Removal Efficiency Standard Under the Low Risk Waste 
Exemption of the BIF Rule Apply?
    II. What is the Status of the ``Low Risk Waste'' Exemption?
    III. What Concerns Have Been Considered for Shakedown?
    IV. What Are the Management Requirements Prior to Burning?
    V. Are There Any Conforming Changes to Subpart X?
    VI. What Are the Requirements for Bevill Residues?
    A. Dioxin Testing of Bevill Residues
    B. Applicability of Part 266 Appendix VIII Products of 
Incomplete Combustion List
    VII. Have There Been Any Changes in Reporting Requirements for 
Secondary Lead Smelters?
    VIII. What Are the Operator Training and Certification 
Requirements?
    IX. Why Did the Agency Redesignate Existing Regulations 
Pertaining to the Notification of Intent to Comply and Extension of 
the Compliance Date?
Part Seven: National Assessment of Exposures and Risks
    I. What Changes Were Made to the Risk Methodology?
    A. How Were Facilities Selected for Analysis?
    B. How Were Facility Emissions Estimated?
    C. What Receptor Populations Were Evaluated?
    D. How Were Exposure Factors Determined?
    E. How Were Risks from Mercury Evaluated?
    F. How Were Risks from Dioxins Evaluated?
    G. How Were Risks from Lead Evaluated?
    H. What Analytical Framework Was Used to Assess Human Exposures 
and Risk?
    I. What Analytical Framework Was Used to Assess Ecological Risk?
    II. How Were Human Health Risks Characterized?
    A. What Potential Health Hazards Were Evaluated?
    1. Dioxins
    2. Mercury
    3. Lead
    4. Other Metals
    5. Hydrogen Chloride
    6. Chlorine
    B. What are the Health Risks to Individuals Residing Near HWC 
Facilities?
    1. Dioxins
    2. Mercury
    3. Lead
    4. Other Metals
    5. Inhalation Carcinogens
    6. Other Inhalation Exposures
    C. What are the Potential Health Risks to Highly Exposed 
Individuals?
    1. Dioxins
    2. Metals
    3. Mercury
    D. What is the Incidence of Adverse Health Effects in the 
Population?
    1. Cancer Risk in the General Population
    2. Cancer Risk in the Local Population
    3. Risks from Lead Emissions
    4. Risks from Emissions of Particulate Matter
    III. What is the Potential for Adverse Ecological Effects?
    A. Dioxins
    B. Mercury
Part Eight: Analytical and Regulatory Requirements
    I. Executive Order 12866: Regulatory Planning and Review (58 FR 
51735)
    II. What Activities Have Led to Today's Rule?
    A. What Analyses Were Completed for the Proposal?
    1. Costs
    2. Benefits
    3. Other Regulatory Issues
    4. Small Entity Impacts
    B. What Major Comments Were Received on the Proposal RIA?
    1. Public Comments
    2. Peer Review
    III. Why is Today's Rule Needed?
    IV. What Were the Regulatory Options?
    V. What Are the Potential Costs and Benefits of Today's Rule?
    A. Introduction
    B. Combustion Market Overview
    C. Baseline Specification
    D. Analytical Methodology and Findings--Engineering Compliance 
Cost Analysis
    E. Analytical Methodology and Findings--Social Cost Analysis
    F. Analytical Methodology and Findings--Economic Impact Analysis
    1. Market Exit Estimates
    2. Quantity of Waste Reallocated
    3. Employment Impacts
    4. Combustion Price Increases
    5. Industry Profits
    6. National-Level Joint Economic Impacts
    G. Analytical Methodology and Findings--Benefits Assessment
    1. Human Health and Ecological Benefits
    2. Waste Minimization Benefits
    VI. What Considerations Were Given to Issues Like Equity and 
Children's Health?
    A. Executive Order 12898, ``Federal Actions to Address 
Environmental Justice in Minority Populations and Low-Income 
Populations'' (February 11, 1994)
    B. Executive Order 13045: Protection of Children from 
Environmental Health Risks and Safety Risks (62 FR 19885, April 23, 
1997)
    C. Unfunded Mandates Reform Act of 1995 (URMA) (Pub. Law 104-4)
    VII. Is Today's Rule Cost Effective?
    VIII. How Do the Costs of Today's Rule Compare to the Benefits?
    IX. What Consideration Was Given to Small Businesses?
    A. Regulatory Flexibility Act (RFA) as amended by the Small 
Business Regulatory Enforcement Fairness Act of 1996 (SBREFA), 5 
U.S.C. 601 et seq.
    B. Analytical Methodology
    C. Results--Direct Impacts
    D. Results--Indirect Impacts
    E. Key Assumptions and Limitations
    X. Were Derived Air Quality and Non-Air Impacts Considered?
    XI. The Congressional Review Act (5 U.S.C. 801 et seq., as added 
by the Small Business Regulatory Enforcement Fairness Act of 1996)
    XII. Paperwork Reduction Act (PRA), 5 U.S.C. 3501-3520
    XIII. National Technology Transfer and Advancement Act of 1995 
(Pub L. 104-113, section 12(d)) (15 U.S.C. 272 note)
    XIV. Executive Order 13084: Consultation and Coordination With 
Indian Tribal Governments (63 FR 27655)
Part Nine: Technical Amendments to Previous Regulations
    I. Changes to the June 19, 1998 ``Fast-track'' Rule
    A. Permit Streamlining Section
    B. Comparable Fuels Section

[[Page 52832]]

Part One: Overview and Background for This Rule

I. What Is the Purpose of This Rule?

    In this final rule, we adopt hallmark standards to more rigorously 
control toxic emissions from burning hazardous waste in incinerators, 
cement kilns, and lightweight aggregate kilns. These emission standards 
and continuation of our RCRA risk policy create a national cap for 
emissions that assures that combustion of hazardous waste in these 
devices is properly controlled.
    The standards themselves implement section 112 of the Clean Air Act 
(CAA) and apply to the three major categories of hazardous waste 
burners--incinerators, cement kilns, and lightweight aggregate kilns. 
For purposes of today's rule, we refer to these three categories 
collectively as hazardous waste combustors. Hazardous waste combustors 
burn about 80% of the hazardous waste combusted annually within the 
United States. As a result, we project that today's standards will 
achieve highly significant reductions in the amount of hazardous air 
pollutants being emitted each year by hazardous waste combustors. For 
example, we estimate that 70 percent of the annual dioxin and furan 
emissions from hazardous waste combustors will be eliminated. Mercury 
emissions already controlled to some degree under existing regulations 
will be further reduced by about 55 percent.
    Section 112 of the CAA requires emissions standards for hazardous 
air pollutants to be based on the performance of the Maximum Achievable 
Control Technology (MACT). The emission standards in this final rule 
are commonly referred to as MACT standards because we use the MACT 
concept to determine the levels of emission control under section 
112(d) of the CAA.1 At the same time, these emissions 
standards satisfy our obligation under the main statute regulating 
hazardous waste management, the Resource Conservation Recovery Act 
(RCRA), to ensure that hazardous waste combustion is conducted in a 
manner adequately protective of human health and the environment. Our 
use of both authorities as the legal basis for today's rule and details 
of the MACT standard-setting process are explained more fully in later 
sections of this preamble. Most significantly, by using both 
authorities in a harmonized fashion, we consolidate regulatory control 
of hazardous waste combustion into a single set of regulations, thereby 
eliminating the potential for conflicting or duplicative federal 
requirements.
---------------------------------------------------------------------------

    \1\ The MACT standards reflect the ``maximum degree of reduction 
in emissions of * * * hazardous air pollutants'' that the 
Administrator determines is achievable, taking into account the cost 
of achieving such emission reduction and any nonair quality health 
and environmental impacts and energy requirements. Section 
112(d)(2).
---------------------------------------------------------------------------

    Today's rule also has other important features in terms of our 
legal obligations and public commitments. First, promulgation of these 
standards fulfills our legal obligations under the CAA to control 
emissions of hazardous air pollutants from hazardous-waste burning 
incinerators and Portland cement kilns.2 Second, today's 
rule fulfills our 1993 and 1994 public commitments to upgrade emission 
standards for hazardous waste combustors. These commitments are the 
centerpiece of our Hazardous Waste Minimization and Combustion 
Strategy.3 Finally, today's rulemaking satisfies key terms 
of a litigation settlement agreement entered into in 1993 with a number 
of groups that had challenged our previous rule addressing emissions 
from hazardous waste boilers and industrial furnaces.4
---------------------------------------------------------------------------

    \2\ In a 1992 Federal Register notice, we published the inital 
list of categories of major and area sources of hazardous air 
pollutants including hazardous waste incinerators and Portland 
cement plants. See 57 FR 31576 (July 16, 1992). Today's rule meets 
our obligation to issue MACT standards for hazardous waste 
incinerators. Today's rule also partially meets our obligation to 
issue MACT standards for Portland cement plants. To complete the 
obligation, we have finalized, in a separate rulemaking, MACT 
standards for the portland cement industry source category. Those 
standards apply to all cement kilns except those kilns that burn 
hazardous waste. See 64 FR 31898 (June 14, 1999). Those standards 
also apply to other HAP emitting sources at a cement plant (such as 
clinker coolers, raw mills, finish mills, and materials handling 
operations) regardless of whether the plant has hazardous waste 
burning cement kilns.
    \3\ EPA Document Number 530-R-94-044, Office of Solid Waste and 
Emergency Response, November 1994.
    \4\ ``Burning of Hazardous Waste in Boilers and Industrial 
Furnaces'' (56 FR 7134, February 21, 1991). These groups include the 
Natural Resources Defense Council, Sierra Club, Environmental 
Technology Council, National Solid Waste Management Association, and 
a number of local citizens' groups.
---------------------------------------------------------------------------

II. In Brief, What Are the Major Features of Today's Rule?

    The major features of today's final rule are summarized below.
A. Which Source Categories Are Affected by This Rule?
    This rule establishes MACT standards for three source categories, 
namely: Hazardous waste burning incinerators, hazardous waste burning 
cement kilns, and hazardous waste burning lightweight aggregate kilns. 
As mentioned earlier, we refer to these three source categories 
collectively as hazardous waste combustors.
B. How Are Area Sources Affected by This Rule?
    This rule establishes that MACT standards apply to both major 
sources--sources that emit or have the potential to emit 10 tons or 
greater per year of any single hazardous air pollutant or 25 tons per 
year or greater of hazardous air pollutants in the aggregate--and area 
sources, all others. Area sources may be regulated under MACT standards 
if we find that the category of area sources ``presents a threat of 
adverse effects to human health or the environment * * * warranting 
regulation (under the MACT standards).'' We choose to regulate area 
sources in today's rule and, as a result, all hazardous waste burning 
incinerators, cement kilns, and lightweight aggregate kilns will be 
regulated under standards reflecting MACT.
C. What Emission Standards Are Established in This Rule?
    This rule establishes emission standards for: Chlorinated dioxins 
and furans; mercury; particulate matter (as a surrogate for antimony, 
cobalt, manganese, nickel, and selenium); semivolatile metals (lead and 
cadmium); low volatile metals (arsenic, beryllium, and chromium); 
hydrogen chloride and chlorine gas (combined). This rule also 
establishes standards for carbon monoxide, hydrocarbons, and 
destruction and removal efficiency as surrogates in lieu of individual 
standards for nondioxin/furan organic hazardous air pollutants.
D. What Are the Procedures for Complying With This Rule?
    This rule establishes standards that apply at all times (including 
during startup, shutdown, or malfunction), except if hazardous waste is 
not being burned or is not in the combustion chamber. When not burning 
hazardous waste (and when hazardous waste does not remain in the 
combustion chamber), you may either follow the hazardous waste burning 
standards in this rule or emission standards we promulgate, if any, for 
other relevant nonhazardous waste source categories.
    Initial compliance is documented by stack performance testing. To 
document continued compliance with the carbon monoxide or hydrocarbon 
standards, you must use continuous emissions monitoring systems. For 
the remaining standards, you must document continued compliance by 
monitoring limits on specified operating parameters. These operating 
parameter

[[Page 52833]]

limits 5 are calculated based on performance test conditions 
using specified procedures intended to ensure that the operating 
conditions (and by correlation the actual emissions) do not exceed 
performance test levels at any time. You must also install an automatic 
waste feed cutoff system that immediately stops the flow of hazardous 
waste feed to the combustor if a continuous emissions monitoring system 
records a value exceeding the standard or if an operating parameter 
limit is exceeded (considering the averaging period for the standard or 
operating parameter). The standards and operating parameter limits 
apply when hazardous waste is being fed or remains in the combustion 
chamber irrespective of whether you institute the corrective measures 
prescribed in the startup, shutdown, and malfunction plan.
---------------------------------------------------------------------------

    \5\ The term ``operating parameter limit'' and ``operating 
limit'' have the same meaning and are used interchangeably in the 
preamble and rule language.
---------------------------------------------------------------------------

E. What Subsequent Performance Testing Must Be Performed?
    You must conduct comprehensive performance testing every five 
years. This testing regime is referred to as ``subsequent performance 
testing.'' You must revise the operating parameter limits as necessary 
based on the levels achieved during the subsequent performance test. In 
addition, you must conduct confirmatory performance testing of dioxins/
furans emissions under normal operating conditions midway between 
subsequent performance tests.
F. What Is the Time Line for Complying With This Rule?
    The compliance date of the standards promulgated in today's rule is 
three years after the date of publication of the rule in the Federal 
Register, or September 30, 2002 (See CAA section 112(i)(3)(A) 
indicating that the Environmental Protection Agency (EPA) may establish 
a compliance date no later than three years from the date of 
promulgation.) A one-year extension of the compliance date may be 
requested if you cannot complete system retrofits by the compliance 
date despite a good faith effort to do so.6 CAA section 
112(i)(3)(B).
Continuous emissions monitoring systems and other continuous monitoring 
systems for the specified operating parameters must be fully 
operational by the compliance date. You must demonstrate compliance by 
conducting a performance test no later than 6 months after the 
compliance date (i.e., three and one-half years from the date of 
publication of today's rule in the Federal Register).
---------------------------------------------------------------------------

    \6\ In June 1998, we promulgated a rule to allow hazardous waste 
combustors also to request a one-year extension to the MACT 
compliance date in cases where additional time will be needed to 
install pollution prevention and waste minimization measures to 
significantly reduce the amount or toxicity of hazardous waste 
entering combustion feedstreams. See 63 FR at 43501 (June 19, 1998). 
This provision is recodified in today's rule as 40 CFR 63.1213.
---------------------------------------------------------------------------

    To ensure timely compliance with the standards, by the compliance 
date you must place in the operating record a Documentation of 
Compliance identifying limits on the specified operating parameters you 
believe are necessary and sufficient to comply with the emission 
standards. These operating parameter limits (and the carbon monoxide or 
hydrocarbon standards monitored with continuous monitoring systems) are 
enforceable until you submit to the Administrator a Notification of 
Compliance within 90 days of completion of the performance test.
    The Notification of Compliance must document: (1) Compliance with 
the emission standards during the performance test; (2) the revised 
operating parameter limits calculated from the performance test; and 
(3) conformance of the carbon monoxide or hydrocarbon continuous 
emissions monitoring systems and the other continuous monitoring 
systems with performance specifications. You must comply with the 
revised operating parameter limits upon submittal of the Notification 
of Compliance.
G. How Does This Rule Coordinate With the Existing RCRA Regulatory 
Program?
    You must have a RCRA permit for stack air emissions (or RCRA 
interim status) until you demonstrate compliance with the MACT 
standards. You do so by conducting a comprehensive performance test and 
submitting a Notification of Compliance to the Administrator, as 
explained above.7 Hazardous waste combustors with RCRA 
permits remain subject to RCRA stack air emission permit conditions 
until the RCRA permit is modified to delete those conditions. (As 
discussed later in more detail, we recommend requesting modification of 
the RCRA permit at the time you submit the Notification of Compliance.) 
Only those provisions of the RCRA permit that are less stringent than 
the MACT requirements specified in the Notification of Compliance will 
be approved for deletion.8 Hazardous waste combustors still 
in interim status without a full RCRA permit are no longer subject to 
the RCRA stack air emissions standards for hazardous waste combustors 
in Subpart O of Part 265 and subpart H of part 266 once compliance with 
the MACT standards has been demonstrated and a Notification of 
Compliance has been submitted to the Administrator.
---------------------------------------------------------------------------

    \7\ Hazardous waste combustors, of course, also continue to be 
subject to applicable RCRA requirements for all other aspects of 
their hazardous waste management activities that are separate from 
the requirements being deferred to the CAA by this rule.
    \8\ RCRA permit requirements that may be less stringent than 
applicable MACT standards are nonetheless enforceable until the RCRA 
permit is modified.
---------------------------------------------------------------------------

    You must satisfy both sets of requirements during the relatively 
short period when both RCRA and MACT stack air emissions standards and 
associated requirements in the RCRA permit or in RCRA interim status 
regulations are effective.
    You also may have existing site-specific permit conditions. On a 
case-by-case basis during RCRA permit issuance or renewal, we determine 
whether further regulatory control of emissions is needed to protect 
human health and the environment, notwithstanding compliance with 
existing regulatory standards. Additional conditions may be included in 
the permit in addition to those derived from the RCRA emission 
standards as necessary to ensure that facility operations are 
protective of human health and the environment. Any of these risk-based 
permit provisions more stringent than today's MACT standards (or that 
address other emission hazards) will remain in the RCRA permit.
    After the MACT compliance date, hazardous waste combustors must 
continue to comply with the RCRA permit issuance process to address 
nonMACT provisions (e.g., general facility standards) and potentially 
conduct a risk review under Sec. 270.32(b)(2) to determine if 
additional requirements pertaining to stack or other emissions are 
warranted to ensure protection of human health and the environment.

III. What Is the Basis of Today's Rule?

    As stated previously, this rule issues final National Emissions 
Standards for Hazardous Air Pollutants (NESHAPS) under authority of 
section 112 of the Clean Air Act for three source categories of 
combustors: Hazardous waste burning incinerators, hazardous waste 
burning cement kilns, and hazardous waste burning lightweight aggregate 
kilns. The main purposes of the CAA are to protect and enhance the 
quality of our Nation's

[[Page 52834]]

air resources, and to promote the public health and welfare and the 
productive capacity of the population. CAA section 101(b)(1). To this 
end, sections 112(a) and (d) of the CAA direct EPA to set standards for 
stationary sources emitting (or having the potential to emit) ten tons 
or greater of any one hazardous air pollutant or 25 tons or greater of 
total hazardous air pollutants annually. Such sources are referred to 
as ``major sources.''
    Today's rule establishes MACT emission standards for the following 
hazardous air pollutants emitted by hazardous waste burning 
incinerators, hazardous waste burning cement kilns, and hazardous waste 
burning lightweight aggregate kilns: Chlorinated dioxins and furans, 
mercury, two semivolatile metals (lead and cadmium), three low 
volatility metals (arsenic, beryllium, and chromium), and hydrochloric 
acid/chlorine gas. This rule also establishes MACT control for the 
other hazardous air pollutants identified in CAA section 112(b)(1) 
through the adoption of standards using surrogates. For example, we 
adopt a standard for particulate matter as a surrogate to control five 
metals that do not have specific emission standards established in 
today's rule. These five metals are antimony, cobalt, manganese, 
nickel, and selenium. Also, we adopt standards for carbon monoxide, 
hydrocarbons, and destruction and removal efficiency to control the 
other organic hazardous air pollutants listed in section 112(b)(1) that 
do not have specific emission standards established in this rule.
    Today's standards meet our commitment under the Hazardous Waste 
Minimization and Combustion Strategy, first announced in May 1993, to 
upgrade the emission standards for hazardous waste burning facilities. 
EPA's Strategy has eight goals: (1) Ensure public outreach and EPA-
State coordination; (2) pursue aggressive use of waste minimization 
measures; (3) continue to ensure that combustion and alternative and 
innovative technologies are safe and effective; (4) develop and impose 
more rigorous controls on combustion facilities; (5) continue 
aggressive compliance and enforcement efforts; (6) enhance public 
involvement opportunities in the permitting process for combustion 
facilities; (7) give higher priority to permitting those facilities 
where a final permit decision would result in the greatest 
environmental benefit or the greatest reduction in risk; and (8) 
advance scientific understanding on combustion issues and risk 
assessment and ensure that permits are issued in a manner that provides 
proper protection of human health and the environment.
    We have made significant progress in implementing the Strategy. 
Today's rule meets the Strategy goal of developing and implementing 
rigorous state-of-the-art safety controls on hazardous waste combustors 
by using the best available technologies and the most current 
science.9 We also developed a software tool (i.e., the Waste 
Minimization Prioritization Tool) that allows users to access relative 
persistent, bioaccumulative and toxic hazard scores for any of 2,900 
chemicals that may be present in RCRA waste streams. We also committed 
to the reduction of the generation of the most persistent, 
bioaccumulative and toxic chemicals by 50 percent by 2005. To 
facilitate this reduction we are developing a list of the persistent, 
bioaccumulative and toxic chemicals of greatest concern and a plan for 
working with the regulated community to reduce these chemicals. In 
addition, we promulgated new requirements to enhance public involvement 
in the permitting process 10 and performed risk evaluations 
during the permitting process for high priority facilities. We also 
made allowances for one-year extensions to the MACT compliance period 
as incentives designed to promote the installation of cost-effective 
pollution prevention technologies to replace or supplement emission 
control technologies for meeting MACT standards.
---------------------------------------------------------------------------

    \9\ The three source categories covered by today's final rule 
burn more than 80 percent of the total amount of hazardous waste 
being combusted each year. The remaining 15-20 percent is burned in 
industrial boilers and other types of industrial furnaces, which 
will be addressed in a future NESHAPS rulemaking for hazardous waste 
burning sources.
    \10\ See 60 FR 63417 (December 11, 1995).
---------------------------------------------------------------------------

    Finally, with regard to the regulatory framework that will result 
from today's rule, we are eliminating the existing RCRA stack emissions 
national standards for hazardous waste incinerators, cement kilns, and 
lightweight aggregate kilns. That is, after submittal of the 
Notification of Compliance established by today's rule (and, where 
applicable, RCRA permit modifications at individual facilities), RCRA 
national stack emission standards will no longer apply to these 
hazardous waste combustors. We originally issued air emission standards 
under the authority of section 3004(a) of RCRA, which calls for EPA to 
promulgate standards ``as may be necessary to protect human health and 
the environment.'' In light of today's new MACT standards, we have 
determined that RCRA emissions standards for these sources would only 
be duplicative and so are no longer necessary to protect human health 
and the environment. Under the authority of section 3004(a), it is 
appropriate to eliminate such duplicative standards.
    Emission standards for hazardous waste burning incinerators and 
other sources burning hazardous wastes as fuel must be protective of 
human health and the environment under RCRA. We conducted a 
multipathway risk assessment to assess the ecological and human health 
risks that are projected to occur under the MACT standards. We have 
concluded that the MACT standards are generally protective of human 
health and the environment and that separate RCRA emission standards 
are not needed. Please see a full discussion of the national assessment 
of exposures and risk in Part VIII of this preamble.
    Additionally, RCRA section 1006(b) directs EPA to integrate the 
provisions of RCRA for purposes of administration and enforcement and 
to avoid duplication, to the maximum extent practicable, with the 
appropriate provisions of the Clean Air Act and other federal statutes. 
This integration must be done in a way that is consistent with the 
goals and policies of these statutes. Therefore, section 1006(b) 
provides further authority for EPA to eliminate the existing RCRA stack 
emissions standards to avoid duplication with the new MACT standards. 
Nevertheless, under the authority of RCRA's ``omnibus'' clause (section 
3005(c)(3); see 40 CFR 270.32(b)(2)), RCRA permit writers may still 
impose additional terms and conditions on a site-specific basis as may 
be necessary to protect human health and the environment.

IV. What Was the Rulemaking Process for Development of This Rule?

    We proposed MACT standards for hazardous waste burning 
incinerators, hazardous waste burning cement kilns, and hazardous waste 
burning lightweight aggregate kilns on April 19, 1996. (61 FR 17358) In 
addition, we published five notices of data availability (NODAs):
    1. August 23, 1996 (61 FR 43501), inviting comment on information 
pertaining to a peer review of three aspects of the proposed rule and 
information pertaining to the since-promulgated ``Comparable Fuels'' 
rule (see 63 FR 43501 (June 19, 1998));
    2. January 7, 1997 (62 FR 960), inviting comment on an updated 
hazardous waste combustor data base containing the emissions and 
ancillary

[[Page 52835]]

data that the Agency used to develop the final rule;
    3. March 21, 1997 (62 FR 13775), inviting comment on our approach 
to demonstrate the technical feasibility of monitoring particulate 
matter emissions from hazardous waste combustors using continuous 
emissions monitoring systems;
    4. May 2, 1997 (62 FR 24212), inviting comment on several topics 
including the status of establishing MACT standards for hazardous waste 
combustors using a revised emissions data base and the status of 
various implementation issues, including compliance dates, compliance 
requirements, performance testing, and notification and reporting 
requirements; and
    5. December 30, 1997 (62 FR 67788), inviting comment on several 
status reports pertaining to particulate matter continuous emissions 
monitoring systems.
    Finally, we have had many formal and informal meetings with 
stakeholders, representing an on-going dialogue on various aspects of 
the rulemaking.
    We carefully considered information and comments submitted by 
stakeholders on these rulemaking actions and during meetings. We 
address their comments in our Response to Comments documents, which can 
be found in the public docket supporting this rulemaking. In addition, 
we addressed certain significant comments at appropriate places in this 
preamble.

Part Two: Which Devices Are Subject to Regulation?

I. Hazardous Waste Incinerators

    Hazardous waste incinerators are enclosed, controlled flame 
combustion devices, as defined in 40 CFR 260.10. These devices may be 
fixed or transportable. Major incinerator designs used in the United 
States are rotary kilns, fluidized beds, liquid injection and fixed 
hearth, while newer designs and technologies are also coming into 
operation. Detailed descriptions of the designs, types of facilities 
and typical air pollution control devices were presented in the April 
1996 NPRM and in the technical background document prepared to support 
the NPRM. (See 61 FR 17361, April 19, 1996.) In 1997, there were 149 
hazardous waste incinerator facilities operating 189 individual units 
in the U.S. Of these 149 facilities, 20 facilities (26 units) were 
commercial hazardous waste incinerators, while the remaining 129 
facilities (163 units) were on-site hazardous waste incinerators.

II. Hazardous Waste Burning Cement Kilns

    Cement kilns are horizontally inclined rotating cylinders, lined 
with refractory-brick, and internally fired. Cement kilns are designed 
to calcine, or drive carbon dioxide out of, a blend of raw materials 
such as limestone, shale, clay, or sand to produce Portland cement. 
When combined with sand, gravel, water, and other materials, Portland 
cement forms concrete, a material used widely in many building and 
construction applications.
    Generally, there are two different processes used to produce 
Portland cement: a wet process and a dry process. In the wet process, 
raw materials are ground, wetted, and fed into the kiln as a slurry. In 
the dry process, raw materials are ground and fed dry into the kiln. 
Wet process kilns are typically longer in length than dry process kilns 
to facilitate water evaporation from the slurried raw material. Dry 
kilns use less energy (heat) and also can use preheaters or 
precalciners to begin the calcining process before the raw materials 
are fed into the kiln.
    A number of cement kilns burn hazardous waste-derived fuels to 
replace some or all of normal fossil fuels such as coal. Most kilns 
burn liquid waste; however, cement kilns also may burn bulk solids and 
small containers containing viscous or solid hazardous waste fuels. 
Containers are introduced either at the upper, raw material end of the 
kiln or at the midpoint of the kiln.
    All existing hazardous waste burning cement kilns use particulate 
matter control devices. These cement plants either use fabric filters 
(baghouses) or electrostatic precipitators to control particulate 
matter.
    In 1997, there were 18 Portland cement plants operating 38 
hazardous waste burning kilns. Of these 38 kilns, 27 kilns use the wet 
process to manufacture cement and 11 kilns use the dry process. Of the 
dry process kilns, one kiln uses a preheater and another kiln used a 
preheater and precalciner. Detailed descriptions of the design types of 
facilities and typical air pollution control devices are presented in 
the technical background document.\11\
---------------------------------------------------------------------------

    \11\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume I: Description of Source Categories,'' July 1999.
---------------------------------------------------------------------------

    In developing standards, the Agency considered the appropriateness 
of distinguishing among the different types of cement kilns burning 
hazardous waste. We determined that distinguishing subcategories of 
hazardous waste burning cement kilns was not needed to develop uniform, 
achievable MACT standards. (See Part Four, Section VII of the preamble 
for a discussion of subcategory considerations.)

III. Hazardous Waste Burning Lightweight Aggregate Kilns

    The term ``lightweight aggregate'' refers to a wide variety of raw 
materials (such as clay, shale, or slate) that, after thermal 
processing, can be combined with cement to form concrete products. 
Lightweight aggregate concrete is produced either for structural 
purposes or for thermal insulation purposes. A lightweight aggregate 
plant is typically composed of a quarry, a raw material preparation 
area, a kiln, a cooler, and a product storage area. The material is 
taken from the quarry to the raw material preparation area and from 
there is fed into the rotary kiln.
    A rotary kiln consists of a long steel cylinder, lined internally 
with refractory bricks, which is capable of rotating about its axis and 
is inclined horizontally. The prepared raw material is fed into the 
kiln at the higher end, while firing takes place at the lower end. As 
the raw material is heated, it melts into a semiplastic state and 
begins to generate gases that serve as the bloating or expanding agent. 
As temperatures reach their maximum, the semiplastic raw material 
becomes viscous and entraps the expanding gases. This bloating action 
produces small, unconnected gas cells, which remain in the material 
after it cools and solidifies. The product exits the kiln and enters a 
section of the process where it is cooled with cold air and then 
conveyed to the discharge. Kiln operating parameters such as flame 
temperature, excess air, feed size, material flow, and speed of 
rotation vary from plant to plant and are determined by the 
characteristics of the raw material.
    In 1997, there were five lightweight aggregate kiln facilities in 
the United States operating 10 hazardous waste-fired kilns. Detailed 
descriptions of the lightweight aggregate process and air pollution 
control techniques are presented in the technical support document.\12\
---------------------------------------------------------------------------

    \12\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume I: Description of Source Categories,'' July 1999.

---------------------------------------------------------------------------

[[Page 52836]]

Part Three: How Were the National Emission Standards for Hazardous 
Air Pollutants (NESHAP) in This Rule Determined?

I. What Authority Does EPA Have To Develop a NESHAP?

    The 1990 Amendments to the Clean Air Act (CAA) significantly 
revised the requirements for controlling emissions of hazardous air 
pollutants. EPA is required to develop a list of categories of major 
and area sources of the hazardous air pollutants identified in section 
112 and to develop, over specified time periods, technology-based 
performance standards for sources of these hazardous air pollutants. 
See CAA sections 112(c) and 112(d). These source categories and 
subcategories are to be listed pursuant to section 112(c)(1). We 
published an initial list of 174 categories of such major and area 
sources in the Federal Register on July 16, 1992 (57 FR 31576), which 
was later amended at 61 FR 28197 (June 4, 1996) \13\ and 63 FR 7155 
(February 12, 1998). That list includes the Hazardous Waste 
Incineration, Portland Cement Manufacturing, and Clay Products 
Manufacturing source categories.
---------------------------------------------------------------------------

    \13\ A subsequent Notice was published on July 18, 1996 (61 FR 
37542) which corrected typographical errors in the June 4, 1996 
Notice.
---------------------------------------------------------------------------

    Promulgation of technology-based standards for these listed source 
categories is not necessarily the final step in the process. CAA 
section 112(f) requires the Agency to report to Congress on the 
estimated risk remaining after imposition of technology-based standards 
and make recommendations as to additional legislation needed to address 
such risk. If Congress does not act on any recommendation presented in 
this report, we are required to impose additional controls if such 
controls are needed to protect public health with an ample margin of 
safety or (taking into account costs, energy, safety, and other 
relevant factors) to prevent adverse environmental effects. In 
addition, if the technology-based standards for carcinogens do not 
reduce the lifetime excess cancer risk for the most exposed individual 
to less than one in a million (1 x 10-6), then we must 
promulgate additional standards.
    We prepared the Draft Residual Risk Report to Congress and 
announced its release on April 22, 1998 (63 FR 19914-19916). In that 
report, we did not propose any legislative recommendation to Congress. 
In section 4.2.4 of the report, we state that: ``The legislative 
strategy embodied in the 1990 CAA Amendments adequately maintains the 
goal of protecting the public health and the environment and provides a 
complete strategy for dealing with a variety of risk problems. The 
strategy recognizes that not all problems are national problems or have 
a single solution. National emission standards will be promulgated to 
decrease the emissions of as many hazardous air pollutants as possible 
from major sources.''

II. What Are the Procedures and Criteria for Development of NESHAPs?

A. Why Are NESHAPs Needed?
    NESHAPs are developed to control hazardous air pollutant emissions 
from both new and existing sources. The statute requires a NESHAP to 
reflect the maximum degree of reduction of hazardous air pollutant 
emissions that is achievable taking into consideration the cost of 
achieving the emission reduction, any nonair quality health and 
environmental impacts, and energy requirements. NESHAPs are often 
referred to as maximum achievable control technology (or MACT) 
standards.
    We are required to develop MACT emission standards based on 
performance of the best control technologies for categories or sub-
categories of major sources of hazardous air pollutants. We also can 
establish lower thresholds for determining which sources are major 
where appropriate. In addition, we may require sources emitting 
particularly dangerous hazardous air pollutants such as particular 
dioxins and furans to control those pollutants under the MACT standards 
for major sources.
    In addition, we regulate area sources by technology-based standards 
if we find that these sources (individually or in the aggregate) 
present a threat of adverse effects to human health or the environment 
warranting regulation. After such a determination, we have a further 
choice whether to require technology-based standards based on MACT or 
on generally achievable control technology.
B. What Is a MACT Floor?
    The CAA directs EPA to establish minimum emission standards, 
usually referred to as MACT floors. For existing sources in a category 
or subcategory with 30 or more sources, the MACT floor cannot be less 
stringent than the ``average emission limitation achieved by the best 
performing 12 percent of the existing sources. * * *'' For existing 
sources in a category or subcategory with less than 30 sources, the 
MACT floor cannot be less stringent than the ``average emission 
limitation achieved by the best performing 5 sources. * * *'' For new 
sources, the MACT floor cannot be ``less stringent than the emission 
control that is achieved by the best controlled similar source. * * *''
    We must consider in a NESHAP rulemaking whether to develop 
standards that are more stringent than the floor, which are referred to 
as ``beyond-the-floor'' standards. To do so, we must consider statutory 
criteria, such as the cost of achieving emission reduction, cost 
effectiveness, energy requirements, and nonair environmental 
implications.
    Section 112(d)(2) specifies that emission reductions may be 
accomplished through the application of measures, processes, methods, 
systems, or techniques, including, but not limited to: (1) Reducing the 
volume of, or eliminating emissions of, such pollutants through process 
changes, substitution of materials, or other modifications; (2) 
enclosing systems or processes to eliminate emissions; (3) collecting, 
capturing, or treating such pollutants when released from a process, 
stack, storage, or fugitive emissions point; (4) design, equipment, 
work practice, or operational standards (including requirements for 
operator training or certification); or (5) any combination of the 
above. See section 112(d)(2).
    Application of techniques (1) and (2) are consistent with the 
definitions of pollution prevention under the Pollution Prevention Act 
and the definition of waste minimization under RCRA. In addition, these 
definitions are in harmony with our Hazardous Waste Minimization and 
Combustion Strategy. These terms have particular applicability in the 
discussion of pollution prevention/waste minimization incentives, which 
were finalized at 63 FR 33782 (June 19, 1998) and which are summarized 
in the permitting and compliance sections of this final rule.
C. How Are NESHAPs Developed?
    To develop a NESHAP, we compile available information and in some 
cases collect additional information about the industry, including 
information on emission source quantities, types and characteristics of 
hazardous air pollutants, pollution control technologies, data from 
emissions tests (e.g., compliance tests, trial burn tests) at 
controlled and uncontrolled facilities, and information on the costs 
and other energy and environmental impacts of emission control 
techniques. We use this information in analyzing and developing 
possible regulatory

[[Page 52837]]

approaches. Of course, we are not always able to assemble the same 
amount of information per industry and typically base the NESHAP on 
information practically available.
    NESHAPs are normally structured in terms of numerical emission 
limits. However, alternative approaches are sometimes necessary and 
appropriate. Section 112(h) authorizes the Administrator to promulgate 
a design, equipment, work practice, or operational standard, or a 
standard that is a combination of these alternatives.

III. How Are Area Sources and Research, Development, and Demonstration 
Sources Treated in This Rule?

A. Positive Area Source Finding for Hazardous Waste Combustors
1. How Are Area Sources Treated in This Rule?
    In today's final rule, we make a positive area source finding 
pursuant to CAA section 112(c)(3) for hazardous waste burning 
incinerators, hazardous waste burning cement kilns, and hazardous waste 
burning lightweight aggregate kilns. This rule subjects both major and 
area sources in these three source categories to the same standards--
the section 112(d) MACT standards. We make this positive area source 
determination because emissions from area sources subject to today's 
rule present a threat of adverse effects to human health and the 
environment. These threats warrant regulation under the section 112 
MACT standards.
2. What Is an Area Source?
    Area sources are sources emitting (or having the potential to emit) 
less than 10 tons per year of an individual hazardous air pollutant, 
and less than 25 tons per year of hazardous air pollutants in the 
aggregate. These sources may be regulated under MACT standards if we 
find that the sources ``presen[t] a threat of adverse effects to human 
health or the environment (by such sources individually or in the 
aggregate) warranting regulation under this section.'' Section 
112(c)(3).
    As part of our analysis, we estimate that all hazardous waste 
burning lightweight aggregate kilns are major sources, principally due 
to their hydrochloric acid emissions. We also estimate that 
approximately 80 percent of hazardous waste burning cement kilns are 
major sources, again due to hydrochloric acid emissions. Only 
approximately 30 percent of hazardous waste burning incinerators appear 
to be major sources, considering only the stack emissions from the 
incinerator. However, major and area source status is determined by the 
entire facility's hazardous air pollutant emissions, so that many on-
site hazardous waste incinerators are major sources because they are 
but one contributing source of emissions among others (sometimes many 
others at large manufacturing complexes) at the same facility.
3. What Is the Basis for Today's Positive Area Source Finding?
    The consequences of us not making a positive area source finding in 
this rule would result in an undesirable bifurcated regulation. First, 
the CAA provides independent authority to regulate certain hazardous 
air pollutant emissions under MACT standards, even if the emissions are 
from area sources. These are the hazardous air pollutants enumerated in 
section 112(c)(6), and include 2,3,7,8 dichlorobenzo-p-dioxins and 
furans, mercury, and some specific polycyclic organic hazardous air 
pollutants--hazardous air pollutants regulated under this rule. See 62 
FR at 24213-24214. Thus, all sources covered by today's rule would have 
to control these hazardous air pollutants to MACT levels, even if we 
were not to make a positive area source determination. Second, because 
all hazardous air pollutants are fully regulated under RCRA, area 
source hazardous waste combustors would have not only a full RCRA 
permit, but also (as just explained) a CAA title V permit for the 
section 112(c)(6) hazardous air pollutants. One purpose of this rule is 
to avoid the administrative burden to sources resulting from this type 
of dual permitting, and these burdensome consequences of not making a 
positive area source finding have influenced our decision that area 
source hazardous waste combustors ``warrant regulation'' under section 
112(d)(2).
    a. Health and Environmental Factors. Our positive area source 
finding is based on the threats presented by emissions of hazardous air 
pollutants from area sources. We find that these threats warrant 
regulation under the MACT standards given the evident Congressional 
intent for uniform regulation of hazardous waste combustion sources, as 
well as the common emission characteristics of these sources and 
amenability to the same emission control mechanisms.
    As discussed in both the April 1996 proposal and May 1997 NODA, all 
hazardous waste combustion sources, including those that may be area 
sources, have the potential to pose a threat of adverse effects to 
human health or the environment, although some commenters disagree with 
this point. These sources emit some of the most toxic, bioaccumulative 
and persistent hazardous air pollutants--among them dioxins, furans, 
mercury, and organic hazardous air pollutants. As discussed in these 
Federal Register notices and elsewhere in today's final rule, potential 
hazardous waste combustor area sources can be significant contributors 
to national emissions of these hazardous air pollutants. (See 62 FR 
17365 and 62 FR 24213.)
    Our positive area source finding also is based on the threat posed 
by products of incomplete combustion. The risks posed by these 
hazardous air pollutants cannot be directly quantified on a national 
basis, because each unit emits different products of incomplete 
combustion in different concentrations. However, among the products of 
incomplete combustion emitted from these sources are potential 
carcinogens.\14\ The potential threat posed by emissions of these 
hazardous air pollutants is manifest and, for several reasons, we do 
not believe that control of these products of incomplete combustion 
should be left to the RCRA omnibus permitting process. First, we are 
minimizing the administrative burden on sources from duplicative 
permitting in this rule by minimizing the extent of RCRA permitting and 
hence minimizing our reliance on the omnibus process. Second, we are 
dealing with hazardous air pollutant emissions from these sources on a 
national rather than a case-by-case basis. We conclude that the control 
of products of incomplete combustion from all hazardous waste 
combustors through state-of-the art organic pollution control is the 
best way to do so from an implementation standpoint. Finally, a basic 
premise of the CAA is that there are so many uncertainties and 
difficulties in developing effective risk-based regulation of hazardous 
air pollutants that the first step should be technology-based standards 
based on Maximum Available Control Technology. See generally S. Rep. 
No. 228, 101st Cong. 1st Sess. 128-32 (1990). The positive area source 
finding and consequent MACT controls is consistent with this primary 
legislative objective.
---------------------------------------------------------------------------

    \14\ E.g., benzene, methylene chloride, hexachlorobenzene, 
carbon tetrachloride, vinal chloride, benzo(a)pyrene, and 
chlorinated dioxins and furans. Energy and Environmental Research 
Corp., surrogate Evaluation for Thermal Treatment Systems, Draft 
Report, October 1994. Also see: USEPA, ``Final technical Support 
Document for HWC MACT Standards, Volume III: Section of MACT 
Standards and Technologies,'' July 1999.
---------------------------------------------------------------------------

    The quantitative risk assessment for the final rule did not find 
risk from

[[Page 52838]]

mercury emissions from hazardous waste burning area source cement kilns 
to be above levels we generally consider acceptable. However, the 
uncertainties underlying the analysis are such that only qualitative 
judgments can be made. We do not believe our analysis can be relied 
upon to make a definitive quantitative finding about the precise 
magnitude of the risk. See Part Five, Section XIII for a discussion of 
uncertainty. Background exposures, which can be quite variable, were 
not considered in the quantitative assessment and are likely to 
increase the risk from incremental exposures to mercury from area 
source cement kilns. Commenters, on the other hand, believed that 
cement kilns did not pose significant risk and questioned our risk 
estimates made in the April 1996 NPRM and May 1997 NODA. However, 
taking into account the uncertainty of our mercury analysis and the 
likelihood of background exposures, a potential for risk from mercury 
may exist. Furthermore, the information available concerning the 
adverse human health effects of mercury, along with the magnitude of 
the emissions of mercury from area source cement kilns, also indicate 
that a threat of adverse effects is presumptive and that a positive 
area source finding is warranted.
    b. Other Reasons Warranting Regulation under Section 112. Other 
special factors indicate that MACT standards are warranted for these 
sources.
    The first reason is Congress's, our, and the public's strong 
preference for similar, if not identical, regulation of all hazardous 
waste combustors. Area sources are currently regulated uniformly under 
RCRA, with no distinction being made between smaller and larger 
emitters. This same desire for uniformity is reflected in the CAA. CAA 
section 112(n)(7) directs the Agency, in its regulation of HWCs under 
RCRA, to ``take into account any regulations of such emissions which 
are promulgated under such subtitle (i.e., RCRA) and shall, to the 
maximum extent practicable and consistent with the provisions of this 
section, ensure that the requirements of such subtitle and this section 
are consistent.'' Congress also dealt with these sources as a single 
class by excluding hazardous waste combustion units regulated by RCRA 
permits from regulation as municipal waste combustors under CAA section 
129(g)(1). Thus, a strong framework in both statutes indicates that air 
emissions from all hazardous waste combustors should be regulated under 
a uniform approach. Failure to adopt such a uniform approach would 
therefore be inconsistent with Congressional intent as expressed in 
both the language and the structure of RCRA and the CAA. Although many 
disagree, several commenters support the approach to apply uniform 
regulations for all hazardous waste combustors and assert that it is 
therefore appropriate and necessary to make the positive area source 
finding.
    Second, a significant number of hazardous waste combustors could 
plausibly qualify as area sources by the compliance date through 
emissions reductions of one or more less dangerous hazardous air 
pollutants, such as total chlorine. We conclude it would be 
inappropriate to exclude from CAA 112(d) regulation and title V 
permitting a significant portion of the sources contributing to 
hazardous air pollutant emissions, particularly nondioxin products of 
incomplete combustion should this occur.
    Third, the MACT controls identified for major sources are 
reasonable and appropriate for potential area sources. The emissions 
control equipment (and where applicable, feedrate control) defined as 
floor or beyond-the-floor control for each source category is 
appropriate and can be installed and operated at potential area 
sources. There is nothing unique about the types and concentrations of 
emissions of hazardous air pollutants from any class of hazardous waste 
combustors that would make MACT controls inappropriate for that 
particular class of hazardous waste combustors, but not the others. 
Commenters also raised the issue of applying generally available 
control technologies (GACT), in lieu of MACT, to area sources. 
Consideration of GACT lead us to the conclusion that GACT would likely 
involve the same types and levels of control as we identified for MACT. 
We believe GACT would be the same as MACT because the standards of this 
rule, based on MACT, are readily achievable, and therefore would also 
be determined to be generally achievable, i.e., GACT.
    Finally, we note that the determination here is unique to these 
RCRA sources, and should not be viewed as precedential for other CAA 
sources. In the language of the statute, there are special reasons that 
these RCRA sources warrant regulation under section 112(d)(2)--and so 
warrant a positive area source finding--that are not present for usual 
CAA sources. These reasons are discussed above--the Congressional 
desire for uniform regulation and our desire (consistent with this 
Congressional objective) to avoid duplicative permitting of these 
sources wherever possible. We repeat, however, that the positive area 
source determination here is not meant as a precedent outside the dual 
RCRA/CAA context.
B. How Are Research, Development, and Demonstration (RD&D) Sources 
Treated in This Rule?
    Today's rule excludes research, development, and demonstration 
sources from the hazardous waste burning incinerator, cement kiln, and 
lightweight aggregate kiln source categories. We discuss below the 
statutory mandate to give special consideration to research and 
development (R&D) sources, an Advanced Notice of Proposed Rulemaking to 
list R&D facilities that we published in 1997, and qualifications for 
exclusion of R&D sources from the hazardous waste combustor source 
categories.
1. Why Does the CAA Give Special Consideration to Research and 
Development (R&D) Sources?
    Section 112(c)(7) of the Clean Air Act requires EPA to ``establish 
a separate category covering research or laboratory facilities, as 
necessary to assure the equitable treatment of such facilities.'' 
Congress included such language in the Act because it was concerned 
that research and laboratory facilities should not arbitrarily be 
included in regulations that cover manufacturing operations. The Act 
defines a research or laboratory facility as ``any stationary source 
whose primary purpose is to conduct research and development into new 
processes and products, where such source is operated under the close 
supervision of technically trained personnel and is not engaged in the 
manufacture of products for commercial sale in commerce, except in a de 
minimis manner.''
    We interpret the Act as requiring the listing of R&D major sources 
as a separate category to ensure equitable treatment of such 
facilities. Language in the Act specifying special treatment of R&D 
facilities (section 112(c)(7)), along with language in the legislative 
history of the Act, suggests that Congress considered it inequitable to 
subject the R&D facilities of an industry to a standard designed for 
the commercial production processes of that industry. The application 
of such a standard may be inappropriate because the wide range of 
operations and sizes of R&D facilities. Further, the frequent changes 
in R&D operations may be significantly different from the typically 
large and continuous production processes.
    We have no information indicating that there are R&D sources, major 
or

[[Page 52839]]

area, that are required to be listed and regulated, other than those 
associated with sources already included in listed source categories 
listed today. Although we are not aware of other R&D sources that need 
to be added to the source category list, such sources may exist, and we 
requested information about them in an Advance Notice of Proposed 
Rulemaking, as discussed in the next section.
2. When Did EPA Notice Its Intent To List R&D Facilities?
    In May 1997 (62 FR 25877), we provided advanced notice that we were 
considering whether to list R&D facilities. We requested public 
comments and information on the best way to list and regulate such 
sources. Comment letters were received from industry, academic 
representatives, and governmental entities. After we compile additional 
data, we will respond to these comments in that separate docket. As a 
result we are not deciding how to address the issue in today's rule. 
The summary of comments and responses will be one part of the basis for 
our future decision whether to list R&D facilities as a source category 
of hazardous air pollutants.
3. What Requirements Apply to Research, Development, and Demonstration 
Hazardous Waste Combustor Sources?
    This rule excludes research, development, and demonstration sources 
from the hazardous waste incinerator, cement kiln, or lightweight 
aggregate kiln source categories and therefore from compliance with 
today's regulations. We are excluding research, development, and 
demonstration sources from those source categories because the emission 
standards and compliance assurance requirements for those source 
categories may not be appropriate. The operations and size of a 
research, development, and demonstration source may be significantly 
different from the typical hazardous waste incinerator that is 
providing ongoing waste treatment service or hazardous waste cement 
kiln or hazardous waste lightweight aggregate kiln that is producing a 
commercial product as well as providing ongoing waste treatment.
    We also are applying the exclusion to demonstration sources because 
demonstration sources are operated more like research and development 
sources than production sources. Thus, the standards and requirements 
finalized today for production sources may not be appropriate for 
demonstration sources. Including demonstration sources in the exclusion 
is consistent with our current regulations for hazardous waste 
management facilities. See Sec. 270.65 providing opportunity for 
special operating permits for research, development, and demonstration 
sources that use an innovative and experimental hazardous waste 
treatment technology or process.
    To ensure that research, development, and demonstration sources are 
distinguished from production sources, we have drawn from the language 
in section 112(c)(7) to define a research, development, and 
demonstration source. Specifically, these are sources engaged in 
laboratory, pilot plant, or prototype demonstration operations: (1) 
Whose primary purpose is to conduct research, development, or short-
term demonstration of an innovative and experimental hazardous waste 
treatment technology or process; and (2) where the operations are under 
the close supervision of technically-trained personnel.15
---------------------------------------------------------------------------

    \15\The statute also qualifies that research and development 
sources do not engage in the manufacture of products for commercial 
sale except in a de minimis manner. Although this qualification is 
appropriate for research and development sources, engaged in short-
term demonstration of an innovative or experimental treatment 
technology or process may produce products for use in commerce. For 
example, a cement kiln engaged in a short-term demonstration of an 
innovative process may nonetheless produce marketable clinker in 
other than de minimis quantities. Consequently, we are not including 
this qualification in the definition of a research, development, and 
demonstration source.
---------------------------------------------------------------------------

    In addition, today's rule limits the exclusion to research, 
development, and demonstration sources that operate for not longer than 
one year after first processing hazardous waste, unless the 
Administrator grants a time extension based on documentation that 
additional time is needed to perform research development, and 
demonstration operations. We believe that this time restriction will 
help distinguish between research, development, and demonstration 
sources and production sources. This time restriction draws from the 
one-year time restriction (unless extended on a case-by-case basis) 
currently applicable to hazardous waste research, development, and 
demonstration sources under Sec. 270.65.
    The exclusion of research, development, and demonstration sources 
applies regardless of whether the sources are located at the same site 
as a production hazardous waste combustor that is subject to the MACT 
standards finalized today. A research, development, and demonstration 
source that is co-located at a site with a production source still 
qualifies for the exclusion. A research, development, and demonstration 
source co-located with a production source is nonetheless expected to 
experience the type and range of operations and be of the size typical 
for other research, development, and demonstration sources.
    Finally, hazardous waste research, development, and demonstration 
sources remain subject to RCRA permit requirements under Sec. 270.65, 
which direct the Administrator to establish permit terms and conditions 
that will assure protection of human health and the environment.
    Although we did not propose this exclusion specifically for 
hazardous waste combustor research, development, and demonstration 
sources, the exclusion is an outgrowth of the May 1997 notice discussed 
above. In that notice we explain that we interpret the CAA as requiring 
the listing of research and development major sources as a separate 
category to ensure equitable treatment of such facilities. A commenter 
on the April 1996 hazardous waste combustor NPRM questioned whether we 
intended to apply the proposed regulations to research and development 
sources. We did not have that intent, and in response are finalizing 
today an exclusion of research, development, and demonstration sources 
from the hazardous waste incinerator, hazardous waste burning cement 
kiln, and hazardous waste burning lightweight aggregate kiln source 
categories.

IV. How Is RCRA's Site-Specific Risk Assessment Decision Process 
Impacted by This Rule?

    RCRA Sections 3004(a) and (q) mandate that standards governing the 
operation of hazardous waste combustion facilities be protective of 
human health and the environment. To meet this mandate, we developed 
national combustion standards under RCRA, taking into account the 
potential risk posed by direct inhalation of the emissions from these 
sources.16 With advancements in the assessment of risk since 
promulgation of the original national standards (i.e., 1981 for 
incinerators and 1991 for boilers and industrial furnaces), we 
recognized in the 1993 Hazardous Waste Minimization and Combustion 
Strategy that additional risk analysis was appropriate. Specifically, 
we noted that the risk posed by indirect exposure (e.g., ingestion of 
contamination in the food chain) to long-term deposition of metals,

[[Page 52840]]

dioxin/furans and other organic compounds onto soils and surface waters 
should be assessed in addition to the risk posed by direct inhalation 
exposure to these contaminants. We also recognized that the national 
assessments performed in support of the original hazardous waste 
combustor standards did not take into account unique and site-specific 
considerations which might influence the risk posed by a particular 
source. Therefore, to ensure the RCRA mandate was met on a facility-
specific level for all hazardous waste combustors, we strongly 
recommended in the Strategy that site-specific risk assessments 
(SSRAs), including evaluations of risk resulting from both direct and 
indirect exposure pathways, be conducted as part of the RCRA permitting 
process. In those situations where the results of a SSRA showed that a 
facility's operations could pose an unacceptable risk (even after 
compliance with the RCRA national regulatory standards), additional 
risk-based, site-specific permit conditions could be imposed pursuant 
to RCRA's omnibus authority (section 3005(c)(3)).
---------------------------------------------------------------------------

    \16\ See No CFR part 264, subpart O for incinerator standards 
and 40 CFR part 266, subpart H for BIF standards.
---------------------------------------------------------------------------

    Today's MACT standards were developed pursuant to section 112(d) of 
the CAA, which does not require a concurrent risk evaluation of those 
standards. To determine if the MACT standards would satisfy the RCRA 
protectiveness mandate in addition to the requirements of the CAA, we 
conducted a national RCRA evaluation of both direct and indirect risk 
as part of this rulemaking. If we found the MACT standards to be 
sufficiently protective so as to meet the RCRA mandate as well, we 
could consider modifying our general recommendation that SSRAs be 
conducted for all hazardous waste combustors, thereby lessening the 
regulatory burden to both permitting authorities and facilities.
    In this section, we discuss: The applicability of both the RCRA 
omnibus authority and the SSRA policy to hazardous waste combustors 
subject to today's rulemaking; the implementation of the SSRA policy; 
the relationship of the SSRA policy to the residual risk requirement of 
section 112(f) of the CAA; and public comments received on these 
topics. A discussion of the national risk characterization methodology 
and results is provided in Part Five, Section XIII of today's notice.
A. What Is the RCRA Omnibus Authority?
    Section 3005(c)(3) of RCRA (codified at 40 CFR 270.32(b)(2)) 
requires that each hazardous waste facility permit contain the terms 
and conditions necessary to protect human health and the environment. 
This provision is commonly referred to as the ``omnibus authority'' or 
``omnibus provision.'' It is the means by which additional site-
specific permit conditions may be incorporated into RCRA permits should 
such conditions be necessary to protect human health and the 
environment.17 SSRAs have come to be used by permitting 
authorities as a quantitative basis for making omnibus determinations 
for hazardous waste combustors.
---------------------------------------------------------------------------

    \17\ The risk-based permit conditions are in addition to those 
conditions required by the RCRA national regulatory standards for 
hazardous waste combustors (e.g., general facility requirements).
---------------------------------------------------------------------------

    In the April 1996 NPRM and May 1997 NODA, we discussed the RCRA 
omnibus provision and its relation to the new MACT standards. 
Commenters question whether the MACT standards supersede the omnibus 
authority with respect to hazardous waste combustor air emissions. 
Other commenters agree in principle with the continued applicability of 
the omnibus authority after promulgation of the MACT standards. These 
commenters recognize that there may be unique conditions at a given 
site that may warrant additional controls to those specified in today's 
notice. For those sources, the commenters acknowledge that permit 
writers must retain the legal authority to place additional operating 
limitations in a source's permit.
    As noted above, the omnibus provision is a RCRA statutory 
requirement and does not have a CAA counterpart. The CAA does not 
override RCRA. Each statute continues to apply to hazardous waste 
combustors unless we determine there is duplication and use the RCRA 
section 1006(b) deferral authority to create a specific regulatory 
exemption.18 Promulgation of the MACT standards, therefore, 
does not duplicate, supersede, or otherwise modify the omnibus 
provision or its applicability to sources subject to today's 
rulemaking. As indicated in the April 1996 NPRM, a RCRA permitting 
authority (such as a state agency) has the responsibility to supplement 
the national MACT standards as necessary, on a site-specific basis, to 
ensure adequate protection under RCRA. We recognize that this could 
result in a situation in which a source may be subject to emission 
standards and operating conditions under two regulatory authorities 
(i.e., CAA and RCRA). Although our intent, consistent with the 
integration provision of RCRA section 1006(b), is to avoid regulatory 
duplication to the maximum extent practicable, we may not eliminate 
RCRA requirements if a source's emissions are not protective of human 
health and the environment when complying with the MACT 
standards.19
---------------------------------------------------------------------------

    \18\ The risk-based permit conditions are in addition to those 
conditions required by the RCRA national regulatory standards for 
hazardous waste combustors (e.g., general facility requirements).
    \19\ RCRA section 1006(b) authorizes deferral of RCRA provisions 
to other EPA-implemented authorities provided, among other things, 
that key RCRA policies and protections are not sacrificed. See 
Chemical Waste Management v. EPA, 976 F. 2d 2, 23, 25 (D.C. Cir. 
1992).
---------------------------------------------------------------------------

B. How Will the SSPA Policy Be Applied and Implemented in Light of This 
Mandate?
1. Is There a Continuing Need for Site-Specific Risk Assessments?
    As stated previously, EPA's Hazardous Waste Minimization and 
Combustion Strategy recommended that SSRAs be conducted as part of the 
RCRA permitting process for hazardous waste combustors where necessary 
to protect human health and the environment. We intended to reevaluate 
this policy once the national hazardous waste combustion standards had 
been updated. We view today's MACT standards as more stringent than 
those earlier standards for incinerators, cement kilns and lightweight 
aggregate kilns. To determine if the MACT standards as proposed in the 
April 1996 NPRM would satisfy the RCRA mandate to protect human health 
and the environment, we conducted a national evaluation of both human 
health and ecological risk. That evaluation, however, did not 
quantitatively assess the proposed standards with respect to mercury 
and nondioxin products of incomplete combustion. This was due to a lack 
of adequate information regarding the behavior of mercury in the 
environment and a lack of sufficient emissions data and parameter 
values (e.g., bioaccumulation values) for nondioxin products of 
incomplete combustion. Since it was not possible to suitably evaluate 
the proposed standards for the potential risk posed by mercury and 
nondioxin products of incomplete combustion, we elected in the April 
1996 NPRM to continue recommending that SSRAs be conducted as part of 
the permitting process until we could conduct a further assessment once 
final MACT standards are promulgated and implemented.
    Although some commenters agree with this approach, a number of 
other commenters question the necessity of a quantitative nondioxin 
product of incomplete combustion assessment to demonstrate RCRA 
protectiveness of the MACT standards. These commenters

[[Page 52841]]

assert that existing site-specific assessments demonstrate that 
emissions of nondioxin products of incomplete combustion are unlikely 
to produce significant adverse human health effects. However, we do not 
agree that sufficient SSRA information exists to conclude that 
emissions from these compounds are unlikely to produce significant 
adverse effects on human health and the environment on a national 
basis. First, only a limited number of completed SSRAs are available 
from which broader conclusions can be drawn. Second, nondioxin products 
of incomplete combustion emissions can vary widely depending on the 
type of combustion unit, hazardous waste feed and air pollution control 
device used. Third, a significant amount of uncertainty exists with 
respect to identifying and quantifying these compounds. Many nondioxin 
products of incomplete combustion cannot be characterized by standard 
analytical methodologies and are unaccounted for by standard emissions 
testing.20 (On a site-specific basis, uncharacterized 
nondioxin products of incomplete combustion are typically addressed by 
evaluating the total organic emissions.) Fourth, nondioxin products of 
incomplete combustion can significantly contribute to the overall risk 
posed by a particular facility. For example, in the Waste Technologies 
Industries incinerator's SSRA, nondioxin organics were estimated to 
contribute approximately 30% of the total cancer risk to the most 
sensitive receptor located in the nearest subarea to the 
facility.21 Fifth, national risk management decisions 
concerning the protectiveness of the MACT standards must be based on 
data that are representative of the hazardous waste combustors subject 
to today's rulemaking. We do not believe that the information afforded 
by the limited number of SSRAs now available is sufficiently complete 
or representative to render a national decision.22
---------------------------------------------------------------------------

    \20\ USEPA, ``Development of a Hazardous Waste Incinerator 
Target Analyte List of Products of Incomplete Combustion'' EPA-600/
R-98-076. 1998.
    \21\ The total cancer risk for this receptor was 1 x 10E-6. The 
results derived for the Waste Technologies Industries incinerator's 
SSRA are a combination of measurements and conservative estimates of 
stack and fugitive emissions, which were developed in tandem with an 
independent external peer review. USEPA, ``Risk Assessment for the 
Waste Technologies Industries Hazardous Waste Incineration Facility 
(East Livepool, Ohio)'' EPA-905-R97-002.
    \22\ Since publication of the April 1996 NPRM, we have expanded 
our national risk evaluation of the other hazardous waste combustor 
emissions (e.g., metals) from 11 facilities to 76 facilities 
assessed for today's final rulemaking. The 76 facilities were 
selected using a stratified random sampling approach that allowed 
for a 90 percent probability of including at least one ``high risk'' 
facility. However, this larger set of facility assessments does not 
include an evaluation nondioxin products of incomplete combustion. 
See Part Five, Section XIII for further discussion.
---------------------------------------------------------------------------

    Some commenters recommend discontinuing conducting SSRAs 
altogether. Other commenters, however, advocate continuing to conduct 
SSRAs, where warranted, as a means of addressing uncertainties inherent 
in the national risk evaluation and of addressing unique, site-specific 
circumstances not considered in the assessment.
    In developing the national risk assessment for the final MAC 
standards, we expanded our original analysis to include a quantitative 
assessment of mercury patterned after the recently published Mercury 
Study Report to Congress.23 We were unable to perform a 
similar assessment of nondioxin products of incomplete combustion 
emissions because of continuing data limitations for these compounds, 
despite efforts to collect additional data since publication of the 
April 1996 NPRM . Thus, we conclude that sufficient data are not 
available to quantitatively assess the potential risk from these 
constituents on a national level as part of today's rulemaking.
---------------------------------------------------------------------------

    \23\ USEPA, ``Mercury Study Report to Congress, Volume III: Fate 
and Transport of Mercury in the Environment,'' EPA 452/R-97-005, 
December 1997.
---------------------------------------------------------------------------

    Given the results of the final national risk assessment for other 
hazardous air pollutants, we generally anticipate that sources 
complying with the MACT standards will not pose an unacceptable risk to 
human health or the environment. However, we cannot make a definitive 
finding in this regard for all hazardous waste combustors subject to 
today's MACT standards for the reasons discussed.
    First, as discussed above, the national risk evaluation did not 
include an assessment of the risk posed by nondioxin products of 
incomplete combustion. As reflected in the Waste Technologies 
Industries SSRA, these compounds can significantly contribute to the 
overall risk posed by a hazardous waste combustor. Without a 
quantitative evaluation of these compounds, we cannot reliably predict 
whether the additional risk contributed by nondioxin products of 
incomplete combustion would or would not result in an unacceptable 
increase in the overall risk posed by hazardous waste combustors 
nationally.
    Second, the quantitative mercury risk analysis conducted for 
today's rulemaking contains significant uncertainties. These 
uncertainties limit the use of the analysis for drawing quantitative 
conclusions regarding the risks associated with the national mercury 
MACT standard. Among others, the uncertainties include an incomplete 
understanding of the fate and transport of mercury in the environment 
and the biological significance of exposures to mercury in fish. (See 
Part Five, Section XIII.) Given these uncertainties, we believe that 
conducting a SSRA, which will assist a permit writer to reduce 
uncertainty on a site-specific basis, may be still warranted in some 
cases.24 As the science regarding mercury fate and transport 
in the environment and exposure improves, and greater certainty is 
achieved in the future, we may be in a better position from which to 
draw national risk management conclusions regarding mercury risk.
---------------------------------------------------------------------------

    \24\ An example of the possible reduction in uncertainty which 
may be derived through the performance of a SSRA includes the degree 
of conversion of mercury to methyl mercury in water bodies. Due to 
the wide range of chemical and physical properties associated with 
surface water bodies, there appears to be a great deal of 
variability concerning mercury methylation. In conducting a SSRA, a 
risk assessor may choose to use a default value to represent the 
percentage of mercury assumed to convert to methyl mercury. 
Conversely, the risk assessor may choose to reduce the uncertainty 
in the analysis by deriving a site-specific value using actual 
surface water data. Chemical and physical properties that may 
influence mercury methylation include, but are not limited to: 
dissolved oxygen content, pH, dissolved organic content, salinity, 
nutrient concentrations, and temperature. See USEPA, ``Human Health 
Risk Assessment Protocol for Hazardous Waste Combustion 
Facilities,'' EPA-530-D-98-001A, External Peer Review Draft, 1998.
---------------------------------------------------------------------------

    Third, we agree with commenters who indicated that, by its very 
nature, the national risk assessment, while comprehensive, cannot 
address unique, site-specific risk considerations \25\ As a result of 
these considerations, a separate analysis or ``risk check'' may be 
necessary to verify that the MACT standards will be adequately 
protective under RCRA for a given hazardous waste combustor.
---------------------------------------------------------------------------

    \25\ Including for example, unusual terrain or dispersion 
features, particularly sensitive ecosystems, unusually high 
contaminant background concentrations, and mercury methylation rates 
in surface water.
---------------------------------------------------------------------------

    Thus, we are recommending that for hazardous waste combustors 
subject to the Phase I final MACT standards, permitting authorities 
should evaluate the need for a SSRA on a case-by-case 
basis.26 SSRAs are not anticipated to be necessary for every 
facility, but should be conducted for facilities where there is some 
reason to believe that operation

[[Page 52842]]

in accordance with the MACT standards alone may not be protective of 
human health and the environment. If a SSRA does demonstrate that 
operation in accordance with the MACT standards may not be protective 
of human health and the environment, permitting authorities may require 
additional conditions as necessary. We consider this an appropriate 
course of action to ensure protection of human health and the 
environment under RCRA, given current limits to our scientific 
knowledge and risk assessment tools.
---------------------------------------------------------------------------

    \26\ We continue to recommend that for those HWCs not subject to 
the Phase I final MACT standards, as SSRA should be conducted as 
part of the RCRA permitting process.
---------------------------------------------------------------------------

2. How Will the SSRA Policy Be Implemented?
    Some commenters suggest that EPA provide regulatory language 
specifically requiring SSRAs. Adequate authority and direction already 
exists to require SSRAs on a case-by-case basis through current 
regulations and guidance (none of which are being reconsidered, revised 
or otherwise reopened in today's rulemaking). The omnibus provision 
(codified in 40 CFR 270.32(b)(2)) directs the RCRA permitting authority 
to include terms and conditions in the RCRA permit as necessary to 
ensure protection of human health and the environment. Under 40 CFR 
270.10(k), the permitting authority may require a permittee or permit 
applicant to submit information where the permitting authority has 
reason to believe that additional permit conditions may be warranted 
under Sec. 270.32(b)(2). Performance of a SSRA is a primary, although 
not exclusive mechanism by which the permitting authority may develop 
the information necessary to make the determination regarding what, if 
any, additional permit conditions are needed for a particular hazardous 
waste combustor. Thus, for hazardous waste combustors, the information 
required to establish permit conditions could include a SSRA, or the 
necessary information required to conduct a SSRA.
    In 1994, we provided guidance concerning the appropriate 
methodologies for conducting hazardous waste combustor 
SSRAs.27 This guidance was updated in 1998 and released for 
publication as an external peer review draft.28 We 
anticipate that use of the updated and more detailed guidance will 
result in a more standardized assessments for hazardous waste 
combustors.
---------------------------------------------------------------------------

    \27\ USEPA. ``Guidance for Performing Screening Level Risk 
Analyses at Combustion Facilities Burning Hazardous Wastes'' Draft, 
April 1994; USEPA. ``Implementation of Exposure Assessment Guidance 
for RCRA Hazardous Waste Combustion Facilities'' Draft, 1994.
    \28\ USEPA. ``Human Health Risk Assessment Protocol for 
Hazardous Waste Combustion Facilities'' EPA-520-D-98-001A, B&C. 
External Peer Review Draft, 1998.
---------------------------------------------------------------------------

    To implement the RCRA SSRA policy, we expect permitting authorities 
to continue evaluating the need for an individual hazardous waste 
combustor risk assessment on a case-by-case basis. We provided a list 
of qualitative guiding factors in the April 1996 NPRM to assist in this 
determination. One commenter is concerned that the subjectivity 
inherent in the list of guiding factors might lead to inconsistencies 
when determining if a SSRA is necessary and suggested that we provide 
additional guidance on how the factors should be used. We continue to 
believe that the factors provided, although qualitative, generally are 
relevant to the risk potential of hazardous waste combustors and 
therefore should be considered when deciding whether or not a SSRA is 
necessary. However, as a practical matter, the complexity of the 
multipathway risk assessment methodology precludes conversion of these 
qualitative factors into more definitive criteria. We will continue to 
compile data from SSRAs to determine if there are any trends which 
would assist in developing more quantitative or objective criteria for 
deciding on the need for a SSRA at any given site. In the interim, 
SSRAs provide the most credible basis for comparisons between risk-
based emission limits and the MACT standards.
    The commenter further suggests that EPA emphasize that the factors 
should be considered collectively due to their complex interplay (e.g., 
exposure is dependent on fate and transport which is dependent on 
facility characteristics, terrain, meteorological conditions, etc.). We 
agree with the commenter. The elements comprising multipathway risk 
assessments are highly integrated. Thus, the considerations used in 
determining if a SSRA is necessary are similarly interconnected and 
should be evaluated collectively.
    The guiding factors as presented in the April 1996 NPRM contained 
several references to the proposed MACT standards. As a result, we 
modified and updated the list to reflect promulgation of the final 
standards and to re-focus the factors to specifically address the types 
of considerations inherent in determining if a SSRA is necessary. The 
revised guiding factors are: (1) Particular site-specific 
considerations such as proximity to receptors, unique dispersion 
patterns, etc.; (2) identities and quantities of nondioxin products of 
incomplete combustion most likely to be emitted and to pose significant 
risk based on known toxicities (confirmation of which should be made 
through emissions testing); (3) presence or absence of other off-site 
sources of pollutants in sufficient proximity so as to significantly 
influence interpretation of a facility-specific risk assessment; (4) 
presence or absence of significant ecological considerations, such as 
high background levels of a particular contaminant or proximity of a 
particularly sensitive ecological area; (5) volume and types of wastes 
being burned, for example wastes containing highly toxic constituents 
both from an acute and chronic perspective; (6) proximity of schools, 
hospitals, nursing homes, day care centers, parks, community activity 
centers that would indicate the presence of potentially sensitive 
receptors; (7) presence or absence of other on-site sources of 
hazardous air pollutants so as to significantly influence 
interpretation of the risk posed by the operation of the source in 
question; and (8) concerns raised by the public. The above list of 
qualitative guiding factors is not intended to be all-inclusive; we 
recognize that there may be other factors equally relevant to the 
decision of whether or not a SSRA is warranted in particular 
situations.
    With respect to existing hazardous waste combustion sources, we do 
not anticipate a large number of SSRAs will need to be performed after 
the compliance date of the MACT standards. SSRAs already have been 
initiated for many of these sources. We strongly encourage facilities 
and permitting authorities to ensure that the majority of those risk 
assessments planned or currently in progress be completed prior to the 
compliance date of the MACT standards. The results of these assessments 
can be used to provide a numerical baseline for emission limits. This 
baseline then can be compared to the MACT limits to determine if site-
specific risk-based limits are appropriate in addition to the MACT 
limits for a particular source.
    Several commenters suggest that completed risk assessments should 
not have to be repeated. We do not anticipate repeating many risk 
assessments. It should be emphasized that changes to comply with the 
MACT standards should not cause an increase in risk for the vast 
majority of the facilities given that the changes, in all probability, 
will be the addition of pollution control equipment or a reduction in 
the hazardous waste being burned. For those few situations in which the 
MACT requirements might result in increased potential risk for a 
particular facility due to unique site-specific considerations, the 
RCRA permit writer, however, may determine

[[Page 52843]]

that a risk check of the projected MACT emission rates is in 
order.29 Should the results of the risk check demonstrate 
that compliance with the MACT requirements does not satisfy the RCRA 
protectiveness mandate, the permitting authority should invoke the 
omnibus provision to impose more stringent, site-specific, risk-based 
permit conditions as necessary to protect human health and the 
environment.
---------------------------------------------------------------------------

    \29\ For example, hazardous waste burning cement kilns that 
previously monitored hydrocarbons in the main stack may elect to 
install a mid-kiln sampling port for carbon monoxide or hydrocarbon 
monitoring to avoid restrictions on hydrocarbon levels in the main 
stack. Thus, their stack hydrocarbon emissions may increase.
---------------------------------------------------------------------------

    With respect to new hazardous waste combustors and existing 
combustors for which a SSRA has never been conducted, we recommend that 
the decision of whether or not a SSRA is necessary be made prior to the 
approval of the MACT comprehensive performance test protocol, thereby 
allowing for the collection of risk emission data at the same time as 
the MACT performance testing, if appropriate (see Part Five, Section 
V). In those instances where it has been determined a SSRA is 
appropriate, the assessment should take into account both the MACT 
standards and any relevant site-specific considerations.
    We emphasize that the incorporation of site-specific, risk-based 
permit conditions into a permit is not anticipated to be necessary for 
the vast majority of hazardous waste combustors. Rather, such 
conditions would be necessary only if compliance with the MACT 
requirements is insufficient to protect human health and the 
environment pursuant to the RCRA mandate and if the resulting risk-
based conditions are more stringent than those required under the CAA. 
Risk-based permit conditions could include, but are not limited to, 
more stringent emission limits, additional operating parameter limits, 
waste characterization and waste tracking requirements.
C. What Is the Difference Between the RCRA SSRA Policy and the CAA 
Residual Risk Requirement?
    Section 112(f) of the CAA requires the Agency to conduct an 
evaluation of the risk remaining for a particular source category after 
compliance with the MACT standards. This evaluation of residual risk 
must occur within eight years of the promulgation of the MACT standards 
for each source category. If it is determined that the residual risk is 
unacceptable, we must impose additional controls on that source 
category to protect public health with an ample margin of safety and to 
prevent adverse environmental effects.
    Our SSRA policy is intended to address the requirements of the RCRA 
protectiveness mandate, which are different from those provided in the 
CAA. For example, the omnibus provision of RCRA requires that the 
protectiveness determination be made on a permit-by-permit or site-
specific basis. The CAA residual risk requirement, conversely, requires 
a determination be made on a source category basis. Further, the time 
frame under which the RCRA omnibus determination is made is more 
immediate; the SSRA is generally conducted prior to final permit 
issuance. The CAA residual risk determination, on the other hand, is 
made at any time within the eight-year time period after promulgation 
of the MACT standards for a source category. Thus, the possibility of a 
future section 112(f) residual risk determination does not relieve RCRA 
permit writers of the present obligation to determine whether the RCRA 
protectiveness requirement is satisfied. Finally, nothing in the RCRA 
national risk evaluation for this rule should be taken as establishing 
a precedent for the nature or scope of any residual risk procedure 
under the CAA.

Part Four: What Is the Rationale for Today's Final Standards?

I. Emissions Data and Information Data Base

A. How Did We Develop the Data Base for This Rule?
    To support the emissions standards in today's rule, we use a 
``fourth generation'' data base that considers and incorporates public 
comments on previous versions of the data base. This final data base 
24 summarizes emissions data and ancillary information on 
hazardous waste combustors that was primarily extracted from 
incinerator trial burn reports and cement and lightweight aggregate 
kiln Certification of Compliance test reports prepared as part of the 
compliance process for the current regulatory standards. Ancillary 
information in the data base includes general facility information 
(e.g., location) process operating data (e.g., waste, fuel, raw 
material compositions, feed rates), and facility equipment design and 
operational information (e.g., air pollution control device 
temperatures).
---------------------------------------------------------------------------

    \24\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume II: HWC Emissions Database,'' July 1999.
---------------------------------------------------------------------------

    The data base supporting the April 1996 proposal was the initial 
data base released for public comment.25 We received a 
substantial number of public comments on this data base including 
identification of data errors and submission of many new trial burn and 
compliance test reports not already in the data base. Subsequently, we 
developed a ``second generation'' data base addressing these comments 
and, on January 7, 1997, published a NODA soliciting public comment on 
the updated data base. Numerous industry stakeholders submitted 
comments on the second generation data base. The data base was revised 
again to accommodate these public comments resulting in a ``third 
generation'' data base. We also published for comment a document 
indicating how specific public comments submitted in response to the 
January NODA were addressed.26 In the May 1997 NODA, we used 
this third generation data base to re-evaluate the MACT standards. 
Since the completion of the third generation data base, we have 
incorporated additional data base comments and new test reports 
resulting in the ``fourth generation'' data base. This final data base 
is used to support all MACT analyses discussed in today's rule. 
Compared to the changes made to develop the third generation data base, 
those changes made in the fourth generation are relatively minor. The 
majority of these changes (e.g., incorporating a few trial burn reports 
and incorporating suggested revisions to the third generation data 
base) were in response to public comments received to May 1997 NODA.
---------------------------------------------------------------------------

    \25\ USEPA, ``Draft Technical Support Document for HWC MACT 
Standards, Volume II: HWC Emissions Database,'' February 1996.
    \26\ See USEPA, ``Draft Report of Revisions to Hazardous Waste 
Combustor Database Based on Public Comments Submitted in Response to 
the January 7, 1997 Notice of Data Availability (NODA),'' May 1997.
---------------------------------------------------------------------------

B. How Are Data Quality and Data Handling Issues Addressed?
    We selected approaches to resolve several data quality and handling 
issues regarding: (1) Data from sources no longer burning hazardous 
waste; (2) assigning values to reported nondetect measurements; (3) 
data generated under normal conditions versus worst-case compliance 
conditions; and (4) use of imputation techniques to fill in missing or 
unavailable data. This section discusses our selected approaches to 
these four issues.

[[Page 52844]]

1. How Are Data From Sources No Longer Burning Hazardous Waste Handled?
    Data and information from sources no longer burning hazardous waste 
are not considered in the MACT standards evaluations promulgated today. 
We note that some facilities have recently announced plans to cease 
burning hazardous waste. Because we cannot continually adjust our data 
base and still finalize this rulemaking, we concluded revisions to the 
data base in early 1998. Announcements or actual facility changes after 
that date simply could not be incorporated.
    Numerous commenters responded to our request for comment on the 
appropriate approach to handle emissions data from sources no longer 
burning hazardous waste. In the April 1996 proposal, we considered all 
available data, including data from sources that had since ceased waste 
burning operations. However, in response to comments to the April 1996 
NPRM, in the May 1997 NODA we excluded data from sources no longer 
burning hazardous waste and reevaluated the MACT floors with the 
revised data base. Of the data included in the fourth generation data 
base, the number of sources that have ceased waste burning operations 
include 18 incineration facilities comprising 18 sources; eight cement 
kiln facilities comprising 12 sources; and one lightweight aggregate 
kiln facility comprising one source.
    Several commenters support the inclusion in the MACT analyses of 
data from sources no longer burning hazardous waste. They believe the 
performance data from these sources are representative of emissions 
control achievable when burning hazardous waste because the data were 
generated under compliance testing conditions. Other commenters suggest 
that data from sources no longer burning hazardous waste should be 
excluded from consideration when conducting MACT floor analyses to 
ensure that the identified MACT floor levels are achievable.
    The approach we adopt today is identical to the one we used for the 
May 1997 NODA. Rather than becoming embroiled in a controversy over 
continued achievability of the MACT standards, we exercise our 
discretion and use a data base consisting of only facilities now 
operating (at least as of the data base finalization date). Ample data 
exist to support setting the MACT standards without using data from 
facilities that no longer burn hazardous waste. To the extent that some 
previous data from facilities not now burning hazardous waste still 
remain in the data base, we ascribe to the view that these data are 
representative of achievable emissions control and can be used.
2. How Are Nondetect Data Handled?
    In today's rule, as in the May 1997 NODA, we evaluated nondetect 
values, extracted from compliance test reports and typically associated 
with feedstream input measurements rather than emissions 
concentrations, as concentrations that are present at one-half the 
detection limit. In the proposal, we assumed that nondetect analyses 
were present at the value of the full detection limit.
    Some commenters support our approach to assume that nondetect 
values are present at one-half the detection limit. The commenter 
states that this approach is consistent with the data analysis 
techniques used in other EPA environmental programs such as in the 
evaluation of groundwater monitoring data. Other commenters oppose 
treating nondetect values at one-half the detection limit, especially 
for dioxins/furans because Method 23 for quantitating stack emissions 
states that nondetect values for congeners be treated as zero when 
calculating total congeners and the toxicity equivalence quotient for 
dioxins/furans. As explained in the NODA, the assumption that nondetect 
measurements are present at one-half the reported detection limit is 
more technically and environmentally conservative and increases our 
confidence that standards and risk findings are appropriate. Further, 
we considered assuming that nondetect values were present at the full 
detection limit, but found that there were no significant differences 
in the MACT data analysis results.27 Therefore, in today's 
rule, we assume nondetect measurements are present at one-half the 
detection limit.
---------------------------------------------------------------------------

    \27\ Using dioxins and furans as an example, for those sources 
using MACT control, this difference is no more than approximately 10 
percent of the standard. USEPA, ``Final Technical Support Document 
for HWC MACT Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

3. How Are Normal Versus Worst-Case Emissions Data Handled?
    The majority of the available emissions data for all of the 
hazardous air pollutants except mercury can be considered worst-case 
because they were generated during RCRA compliance testing. Because 
limits on operating parameters are established based on compliance test 
operations, sources generally operate during compliance testing under 
worst-case conditions to account for variability in operations and 
emissions. However, the data base also contains some normal data for 
these hazardous air pollutants. Normal data include those where 
hazardous waste was burned, but neither spiking of the hazardous waste 
with metals or chlorine nor operation of the combustion unit and 
emission control equipment under detuned conditions occurred.
    In the MACT analyses supporting today's rule, normal data were not 
used to identify or define MACT floor control, with the exception of 
mercury, as discussed below. This approach is identical to the one used 
in the May 1997 NODA. 62 FR 24216.
    Several commenters support the use of normal emissions data in 
defining MACT controls because the effect of ignoring the potentially 
lower emitters from these sources would skew the analysis to higher 
floor results. Other commenters oppose the use of normal data because 
they would not be representative of emissions under compliance test 
conditions--the conditions these same sources will need to operate 
under during MACT performance tests to establish limits on operating 
conditions.28
---------------------------------------------------------------------------

    \28\ These commenters are concerned that, if the standards were 
based on normal emissions data, sources would be inappropriately 
constrained to emissions that are well below what is currently 
normal. This is because of the double ratcheting effect of the 
compliance regime whereby a source must first operate below the 
standard during compliance testing, and then again operate below 
compliance testing levels (and associated operating parameters) to 
maintain day-to-day compliance.
---------------------------------------------------------------------------

    We conclude that it is inappropriate to perform the MACT floor 
analysis for a particular hazardous air pollutant using emissions data 
that are a mixture of normal and worst-case data. The few normal 
emissions data would tend to dominate the identification of best 
performing sources while not necessarily being representative of the 
range of normal emissions. Because the vast majority of our data is 
based on worst-case compliance testing, the definition of floor control 
is based on worst-case data.29 Using worst-case emissions 
data to establish a MACT

[[Page 52845]]

floor also helps account for emissions variability, as discussed in 
Section V.D. below.
---------------------------------------------------------------------------

    \29\ We considered adjusting the emissions data to account for 
spiking to develop a projected normal emissions data base. However, 
we conclude that this is problematic and have not done so. For 
example, it is difficult to project (lower) emissions from 
semivolatile metal-spiked emissions data given that system removal 
efficiency does not correlate linearly with semivolatile metal 
feedrate. In addition, we did not know for certain whether some data 
were spiked. Thus, we would have to use either a truncated data base 
of despiked data or a mixed data base of potentially spiked data and 
despiked data, neither of which would be fully satisfactory.
---------------------------------------------------------------------------

    Sources did not generally spike mercury emissions during RCRA 
compliance testing because they normally feed mercury at levels 
resulting in emissions well below current limits.30 
Consequently, sources are generally complying with generic, 
conservative feedrate limits established under RCRA rather than 
feedrate limits established during compliance testing. Because our data 
base is comprised essentially of normal emissions, we believe this is 
one instance where use of normal data to identify MACT floor is 
appropriate. See discussion in Section V.D. below of how emissions 
variability is addressed for the mercury floors.
---------------------------------------------------------------------------

    \30\ Three of 23 incinerators used to define MACT floor (i.e., 
sources for which mercury feedrate data are available) are known to 
have spiked mercury. No cement kilns used to define MACT floor 
(e.g., excluding sources that have stopped burning hazardous waste) 
are known to have spiked mercury. Only one of ten lightweight 
aggregate kilns used to define MACT floor is known to have spiked 
mercury.
---------------------------------------------------------------------------

4. What Approach Was Used To Fill In Missing or Unavailable Data?
    With respect to today's rule, the term ``imputation'' refers to a 
data handling technique where a value is filled-in for a missing or 
unavailable data point. We only applied this technique to hazardous air 
pollutants that are comprised of more than one pollutant (i.e., 
semivolatile metals, low volatile metals, total chlorine). We used 
imputation techniques in both the proposal and May 1997 NODA; however, 
we decided not to use imputation procedures in the development of 
today's promulgated standards. We used only complete data sets in our 
MACT determinations. Several commenters to the proposal and May 1997 
NODA oppose the use of imputation techniques. Commenters express 
concern that the imputation approach used in the proposal did not 
preserve the statistical characteristics (average and standard 
deviation) of the entire data set. Thus, commenters suggest that 
subsequent MACT analyses were flawed. We reevaluated the data base and 
determined that a sufficient number of data sets are complete without 
the use of an imputation technique.31 A complete discussion 
of various data handling conventions is presented in the technical 
support document.32
---------------------------------------------------------------------------

    \31\ This is especially true because antimony is no longer 
included in the low volatile metal standard.
    \32\ See USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

II. How Did We Select the Pollutants Regulated by This Rule?

    Section 112(b) of the Clean Air Act, as amended, provides a list of 
188 33 hazardous air pollutants for which the Administrator 
must promulgate emission standards for designated major and area 
sources. The list is comprised of metal, organic, and inorganic 
compounds.
---------------------------------------------------------------------------

    \33\ The initial list consisted of 189 HAPs, but we have removed 
caprolactam (CAS number 105602) from the list of hazardous air 
pollutants. See Sec. 63.60.
---------------------------------------------------------------------------

    Hazardous waste combustors emit many of the hazardous air 
pollutants. In particular, hazardous waste combustors can emit high 
levels of dioxins and furans, mercury, lead, chromium, antimony, and 
hydrogen chloride. In addition, hazardous waste combustors can emit a 
wide range of nondioxin/furan organic hazardous air pollutants, 
including benzene, chloroform, and methylene chloride.
    In today's rule, we establish nine emission standards to control 
hazardous air pollutants emitted by hazardous waste combustors. 
Specifically, we establish emission standards for the following 
hazardous air pollutants: Chlorinated dioxins and furans, mercury, two 
semivolatile metals (i.e., lead and cadmium), three low volatility 
metals (i.e., arsenic, beryllium, chromium), and hydrochloric acid/
chlorine gas. In addition, MACT control is provided for other hazardous 
air pollutants via standards for surrogates: (1) A standard for 
particulate matter will control five metal hazardous air pollutants--
antimony, cobalt, manganese, nickel, and selenium; and (2) standards 
for carbon monoxide, hydrocarbons, and destruction and removal 
efficiency will control nondioxin/furan organic hazardous air 
pollutants.
A. Which Toxic Metals Are Regulated by This Rule? 34
---------------------------------------------------------------------------

    \34\ RCRA standards currently control emissions of three toxic 
metals that have not been designated as Clean Air Act hazardous air 
pollutants: Barium, silver, and thallium. These RCRA metals are 
incidentally controlled by today's MACT controls for metal hazardous 
air pollutants in two ways. First, the RCRA metals are semivolatile 
or nonvolatile and will, in part, be controlled by the air pollution 
control systems used to meet the semivolatile metal and low volatile 
metal standards in today's rule. Second, these RCRA metals will be 
controlled by the measures used to meet today's MACT participate 
matter standard. See text that follows.
---------------------------------------------------------------------------

1. Semivolatile and Low Volatile Metals
    The Section 112(b) list of hazardous air pollutants includes 11 
metals: antimony, arsenic, beryllium, cadmium, chromium, cobalt, lead, 
manganese, mercury, nickel, and selenium. To establish an implementable 
approach for controlling these metal hazardous air pollutants, we 
proposed to group the metals by their relative volatility and 
established emission standards for each volatility group. We placed six 
of the eleven metals in volatility groups. The high-volatile group is 
comprised of mercury, the semivolatile group is comprised of lead and 
cadmium, and the low volatile group is comprised of arsenic, beryllium, 
and chromium.35 We refer to these six metals for which we 
have established standards based on volatility group as ``enumerated 
metals.'' We have chosen to control the remaining five metals using 
particulate matter as a surrogate as discussed in the next section.
---------------------------------------------------------------------------

    \35\ Antimony was included in the low volatile group at 
proposal, but we subsequently determined that the MACT particulate 
matter standard serves as an adequate surrogate for this metal. See 
the May 1997 NODA (62 FR at 24216). In making this determination, we 
noted that antimony is an noncarcinogen with relatively low toxicity 
compared with the other five nonmercury metals that were placed in 
volatility groups. To be of particular concern, antimony would have 
to be present in hazardous waste at several orders of magnitude 
higher than shown in the available data.
---------------------------------------------------------------------------

    Grouping metals by volatility is reasonable given that emission 
control strategies are governed primarily by a metal's volatility. For 
example, while semivolatile metals and low volatile metals are in 
particulate form in the emission control train and can be removed as 
particulate matter, mercury species are generally emitted from 
hazardous waste combustors in the vapor phase and cannot be controlled 
by controlling particulate matter unless a sorbent, such as activated 
carbon, is injected into the combustion gas. In addition, low volatile 
metals are easier to control than semivolatile metals because 
semivolatile metals volatilize in the combustion chamber and condense 
on fine particulate matter, which is somewhat more difficult to 
control. Low volatile metals do not volatilize significantly in 
hazardous waste combustors and are emitted as larger, easier to remove, 
particles entrained in the combustion gas.36
---------------------------------------------------------------------------

    \36\ The dynamics associated with the fate of metals in a 
hazardous waste combustor are much more complex than presented here. 
For more information, see USEPA, ``Draft Technical Support Document 
for HWC MACT Standards, Volume VII: Miscellaneous Technical 
Issues,'' February 1996.
---------------------------------------------------------------------------

    Commenters agree with our proposal to group metals by their 
relative volatility. We adopt these groupings for the final rule.
    We note that the final rule does not require a source to control 
its particulate matter below the particulate matter standard to control 
semivolatile and low

[[Page 52846]]

volatile metals. It is true that when we were determining the 
semivolatile and low volatile metal floor standards, we did examine the 
feedrates from only those facilities that were meeting the numerical 
particulate standard. See Part Four, Section V.B.2.c. This is because 
we believe that facilities, in practice, use both feedrate and 
particulate matter air pollution control devices in a complementary 
manner to address metals emissions (except mercury). However, our 
setting of the semivolatile and low volatile metal floor standards does 
not require MACT particulate matter control to be installed, either 
directly or indirectly, as a matter of CAA compliance. We do not think 
it is necessary to require compliance with a particulate matter 
standard as an additional express element of the semivolatile/low 
volatile metal emission standards because the particulate matter 
standard is already required to control the nonenumerated metals, as 
discussed below. However, we could have required compliance with a 
particulate matter standard as part of the semivolatile or low volatile 
metal emission standard because of the practice of using particulate 
matter control as at least part of a facility's strategy to control or 
minimize metal emissions (other than mercury).
2. How Are the Five Other Metal Hazardous Air Pollutants Regulated?
    We did not include five metal hazardous air pollutants (i.e., 
antimony, cobalt, manganese, nickel, selenium) in the volatility groups 
because of: (1) Inadequate emissions data for these metals 
37; (2) relatively low toxicity of antimony, cobalt, and 
manganese; and (3) the ability to achieve control, as explained below, 
by means of surrogates. Instead, we chose the particulate matter 
standard as a surrogate control for antimony, cobalt, manganese, 
nickel, and selenium. We refer to these five metals as ``nonenumerated 
metals'' because standards specific to each metal have not been 
established. We conclude that emissions of these metals is effectively 
controlled by the same air pollution control devices and systems used 
to control particulate matter.
---------------------------------------------------------------------------

    \37\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume II: HWC Emissions Database,'' July 1999.
---------------------------------------------------------------------------

    Some commenters suggest that particulate matter is not a surrogate 
for the five nonenumerated metals. Commenters also note that our own 
study, as well as investigations by commenters, did not show a 
relationship between particulate matter and semivolatile metals and low 
volatile metals when emissions from multiple sources were considered. 
However, we conclude that such a relationship is not expected when 
multiple sources are considered because wide variations in source 
operations can affect: (1) Metals and particulate matter loadings at 
the inlet to the particulate matter control device; (2) metals and 
particulate matter collection efficiency; and (3) metals and 
particulate matter emissions. Factors that can contribute to 
variability in source operations include metal feed rates, ash levels, 
waste types and physical properties (i.e., liquid vs. solid), 
combustion temperatures, and particulate matter device design, 
operation, and maintenance.
    Conversely, emissions of semivolatile metals and low volatile 
metals are directly related to emissions of particulate matter at a 
given source when other operating conditions are held constant (i.e., 
as particulate matter emissions increase, emissions of these metals 
also increase) because semivolatile metals and low volatile metals are 
present as particulate matter at the typical air pollution control 
device temperatures of 200 to 400 deg.F that are required under today's 
rule.38 A strong relationship between particulate matter and 
semivolatile/low volatile metal emissions is evident from our emissions 
data base of trial burn emissions at individual sources where 
particulate matter varies and metals feedrates and other conditions 
that may affect metals emissions were held fairly constant. Other work 
also has clearly demonstrated that improvement in particulate control 
leads to improved metals control.39
---------------------------------------------------------------------------

    \38\ The dioxin/furan emission standard requires that gas 
temperatures at the inlet to electrostatic precipitators and fabric 
filters not exceed 400 deg.F. Wet particulate matter control devices 
reduce gas temperatures to below 400 deg.F by virtue of their design 
and operation. The vapor phase contribution (i.e., nonparticulate 
form that will not be controlled by a particulate matter control 
device) of semivolatile metal and low volatile metal at these 
temperatures is negligible.
    \39\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    We also requested comment on whether particulate matter could be 
used as a surrogate for all semivolatile and low volatile metal 
hazardous air pollutants (i.e., all metal hazardous air pollutants 
except mercury). See the May 1997 NODA. This approach is strongly 
recommended by the cement industry. In that Notice, we concluded that, 
because of varying and high levels of metals concentrations in 
hazardous waste, use of particulate matter control alone may not 
provide MACT control for metal hazardous air pollutants.40 
Our conclusion is the same today. Without metal-specific MACT emission 
standards or MACT feedrate standards, sources could feed high levels of 
one or more metal hazardous air pollutant metals. This practice could 
result in high metal emissions, even though the source's particulate 
matter is controlled to the emission standard (i.e., a large fraction 
of emitted particulate matter could be comprised of metal hazardous air 
pollutants). Thus, the use of particulate matter control alone would 
not constitute MACT control of that metal and would be particularly 
troublesome for the enumerated semivolatile and low volatile metal 
because of their toxicity.41
---------------------------------------------------------------------------

    \40\ However, for sources not burning hazardous waste and 
without a significant potential for extreme variability in metals 
feedrates, particulate matter is an adequate surrogate for metal 
hazardous air pollutants (e.g., for nonhazardous waste burning 
cement kilns).
    \41\ Using particulate matter as a surrogate for metals is, 
however, the approach we used in the final rule for five metals: 
Antimony, cobalt, manganese, nickel, selenium. Technical and 
practical reasons unique to these metals support this approach. 
First, these metals exhibit relatively low toxicity. Second, for 
some of these metals, we did not have emissions data adequate to 
establish specific standards. Therefore, the best strategy for these 
particular metals, at this time, is to rely on particulate matter as 
a surrogate.
---------------------------------------------------------------------------

    Many commenters suggest that particulate matter is an adequate 
surrogate for all metal hazardous air pollutants. They suggest that, 
given current metal feedrates and emission rates, particularly in the 
cement industry, a particulate matter standard is sufficient to ensure 
that metal hazardous air pollutants (other than mercury) are controlled 
to levels that would not pose a risk to human health or the 
environment. While this may be true in some cases as a theoretical 
matter, it may not be in all cases. Data demonstrating this 
conclusively were not available for all cement kilns. Moreover, this 
approach may not ensure MACT control of the potentially problematic 
(i.e., high potential risk) metals for reasons discussed above (i.e., 
higher metal feedrates will result in higher metals emissions even 
though particulate matter capture efficiency remains constant). 
Consequently, we conclude that semi-volatile metals and low volatile 
metals standards are appropriate in addition to the particulate matter 
standard.
    Finally, several commenters suggest that a particulate matter 
standard is not needed to control the five nonenumerated metals because 
the standards for the enumerated semivolatile and low volatile metals 
would serve as surrogates for those

[[Page 52847]]

metals. Their rationale is that because the nonenumerated metals can be 
classified as either semivolatile or nonvolatile 42, they 
would be controlled along with the enumerated semivolatile and low 
volatile metals. However, MACT control would not be assured for the 
five nonenumerated metals even though they would be controlled by the 
same emission control device as the enumerated semivolatile and low 
volatile metals. For example, a source with high particulate matter 
emissions could achieve the semivolatile and low volatile metal 
emission standards (i.e., MACT control) by feeding low levels of 
enumerated semivolatile and low volatile metals. But, if that source 
also fed high levels of nonenumerated metals, MACT control for those 
metals would not be achieved unless the source was subject to a 
particulate matter MACT standard. Consequently, we do not agree that 
the semivolatile and low volatile metal standards alone can serve as 
surrogates for the nonenumerated metals.
---------------------------------------------------------------------------

    \42\ As a factual matter, selenium can be classified as a 
semivolatile metal and the remaining four nonenumerated metals can 
be classified as low volatile metals.
---------------------------------------------------------------------------

    We also proposed to use particulate matter as a supplemental 
control for nondioxin/furan organic hazardous air pollutants that are 
adsorbed onto the particulate matter. Commenters state, however, that 
the Agency had not presented data showing that particulate matter in 
fact contains significant levels of adsorbed nondioxin/furan organic 
hazardous air pollutants. We now concur with commenters that, for 
cement kiln and lightweight aggregate kiln particulate matter, 
particulate matter emissions have not been shown to contain significant 
levels of adsorbed organic compounds. This is likely because cement 
kiln and lightweight aggregate kiln particulate matter is primarily 
inert process dust (i.e., entrained raw material). Although particulate 
matter emissions from incinerators could contain higher levels of 
carbon that may adsorb some organic compounds, this is not likely a 
significant means of control for those organic hazardous air 
pollutants.43
---------------------------------------------------------------------------

    \43\ We recognize that sorbent (e.g., activated carbon) may be 
injected into the combustion system to control mercury or dioxin/
furan. In these cases, particulate matter would be controlled as a 
site-specific compliance parameter for these organics. See the 
discussion in Part Five of this preamble.
---------------------------------------------------------------------------

B. How Are Toxic Organic Compounds Regulated by This Rule?
1. Dioxins/Furans
    We proposed that dioxin/furan emissions be controlled directly with 
a dioxin/furan emission standard based on toxicity equivalents. The 
final rule adopts a TEQ approach for dioxin/furans. In terms of a 
source determining compliance, we expect sources to use accepted TEQ 
references.44
---------------------------------------------------------------------------

    \44\ For example, USEPA, ``Interim Procedure for Estimating 
Risks Associated With Exposures to Mixtures of Chlorinated Dibenzo-
p-Dioxin and -Dibenzofurans (CDDs and CDFs) and 1989 Update'', March 
1989; Van den Berg, M., et al. ``Toxic Equivalency Factors (TEFs) 
for PCBs, PCDDs, PCDFs for Humans and Wildlife'' Environmental 
Health Perspectives, Volume 106, Number 12, December 1998.
---------------------------------------------------------------------------

2. Carbon Monoxide and Hydrocarbons
    We proposed that emissions of nondioxin/furan organic hazardous air 
pollutants be controlled by compliance with continuously monitored 
emission standards for either of two surrogates: carbon monoxide or 
hydrocarbons. Carbon monoxide and hydrocarbons are widely accepted 
indicators of combustion conditions. The current RCRA regulations for 
hazardous waste combustors use emissions limits on carbon monoxide and 
hydrocarbons to control emissions of nondioxin/furan toxic organic 
emissions. See 56 FR 7150 (February 21, 1991) documenting the 
relationship between carbon monoxide, combustion efficiency, and 
emissions of organic compounds. In addition, Clean Air Act emission 
standards for municipal waste combustors and medical waste incinerators 
limit emissions of carbon monoxide to control nondioxin/furan organic 
hazardous air pollutants. Finally, hydrocarbon emissions are an 
indicator of organic hazardous air pollutants because hydrocarbons are 
a direct measure of organic compounds.
    Nonetheless, many commenters state that EPA's own surrogate 
evaluation 45 did not demonstrate a relationship between 
carbon monoxide or hydrocarbons and nondioxin/furan organic hazardous 
air pollutants at the carbon monoxide and hydrocarbon levels evaluated. 
Several commenters note that this should not have been a surprise given 
that the carbon monoxide and hydrocarbon emissions data evaluated were 
generally from hazardous waste combustors operating under good 
combustion conditions (and thus, relatively low carbon monoxide and 
hydrocarbon levels). Under these conditions, emissions of nondioxin/
furan organic hazardous air pollutants were generally low, which made 
the demonstration of a relationship more difficult. These commenters 
note that there may be a correlation between carbon monoxide and 
hydrocarbons and nondioxin/furan organic hazardous air pollutants, but 
it would be evident primarily when actual carbon monoxide and 
hydrocarbon levels are higher than the regulatory levels. We agree, and 
conclude that carbon monoxide and hydrocarbon levels higher than those 
we establish as emission standards are indicative of poor combustion 
conditions and the potential for increased emissions of nondioxin/furan 
organic hazardous air pollutants. Consequently, we have adopted our 
proposed approach for today's final rule.46
---------------------------------------------------------------------------

    \45\ See Energy and Environmental Research Corporation, 
``Surrogate Evaluation of Thermal Treatment Systems,'' Draft Report, 
October 17, 1994.
    \46\ As discussed at proposal, however, this relationship does 
not hold for certain types of cement kilns where carbon monoxide and 
hydrocarbons emissions evolve from raw materials. See discussion in 
Section VII of Part Four.
---------------------------------------------------------------------------

3. Destruction and Removal Efficiency
    We have determined that a destruction and removal efficiency (DRE) 
standard is needed to ensure MACT control of nondioxin/furan organic 
hazardous air pollutants.47 We adopt the implementation 
procedures from the current RCRA requirements for DRE (see 
Secs. 264.342, 264.343, and 266.104) in today's final rule. The 
rationale for adopting destruction and removal efficiency as a MACT 
standard is discussed later in Section IV of the preamble.
---------------------------------------------------------------------------

    \47\ Under this standard, several difficult to combust organic 
compounds would be identified and destroyed or removed by the 
combustor to at least a 99.99% (or 99.9999%, as applicable) 
efficiency.
---------------------------------------------------------------------------

C. How Are Hydrochloric Acid and Chlorine Gas Regulated by This Rule?
    We proposed that hydrochloric acid and chlorine gas emissions be 
controlled by a combined total chlorine MACT standard because: (1) The 
test method used to determine hydrochloric acid and chlorine gas 
emissions may not be able to distinguish between the compounds in all 
situations; 48 and (2) both of these hazardous air 
pollutants can be controlled by limiting feedrate of chlorine in 
hazardous waste and wet scrubbing. We have adopted this approach in 
today's final rule.
---------------------------------------------------------------------------

    \48\ See the proposed rule, 61 FR at 17376.
---------------------------------------------------------------------------

    One commenter questions whether it is appropriate to establish a 
combined standard for hydrochloric acid and chlorine gas because the 
removal efficiency of emission control equipment is substantially 
different for the two pollutants. Although we agree that the efficiency 
of emission control equipment is substantially different for the two 
pollutants, we conclude that the MACT control techniques will readily

[[Page 52848]]

enable sources to achieve the hydrochloric acid/chlorine gas emission 
standard. As discussed in Sections VI, VII, and VIII below, MACT 
control for all hazardous waste combustors is control of the hazardous 
waste chlorine feedrate. This control technique is equally effective 
for hydrochloric acid and chlorine gas and represents MACT control for 
cement kilns. MACT control for incinerators also includes wet 
scrubbing. Although wet scrubbing is more efficient for controlling 
hydrochloric acid, it also provides some control of chlorine gas. MACT 
control for lightweight aggregate kilns also includes wet or dry 
scrubbing. Although dry scrubbing does not control chlorine gas, 
chlorine feedrate control combined with dry scrubbing to remove 
hydrochloric acid will enable lightweight aggregate kilns to achieve 
the emission standard for hydrochloric acid/chlorine gas.

III. How Are the Standards Formatted in This Rule?

A. What Are the Units of the Standards?
    With one exception, the final rule expresses the emission standards 
on a concentration basis as proposed, with all standards expressed as 
mass per dry standard cubic meter (e.g., g/dscm), with 
hydrochloric acid/chlorine gas, carbon monoxide, and hydrocarbon 
standards being expressed at parts per million by volume (ppmv). The 
exception is the particulate matter standard for hazardous waste 
burning cement kilns where the standard is expressed as kilograms of 
particulate matter per Mg of dry feed to the kiln.
    Several commenters suggest that the standards should be expressed 
on a mass emission basis (e.g., mg/hour) because of equity concerns 
across source categories and environmental loading concerns. They are 
concerned that expressing the standards on a concentration basis allows 
large gas flow rate sources such as cement kilns to emit a much greater 
mass of hazardous air pollutants per unit time than smaller sources 
such as some on-site incinerators. Concomitantly, small sources would 
incur a higher cost/lb of pollutant removed, they contend, than a large 
source.49 Further, they reason that the larger sources would 
pose a much greater risk to human health and the environment because 
risk is a function of mass emissions of pollutants per unit of time.
---------------------------------------------------------------------------

    \49\ This result is not evident given that the cost of an 
emission control device is generally directly proportional to the 
gas flow rate, not the mass emission rate of pollutants per unit 
time.
---------------------------------------------------------------------------

    Although we agree with commenters' point about differential 
environmental loadings attributable to small versus large sources with 
a concentration-based standard, we note that the mass-based standard 
urged here is inherently incompatible with technology-based MACT 
standards for several reasons.50 A mass-based standard does 
not ensure MACT control at small sources. Small sources have lower flow 
rates and thus would be allowed to emit hazardous air pollutants at 
high concentrations. They could meet the standard with no or minimal 
control. In addition, this inequity between small and large sources 
would create an incentive to divert hazardous waste from large sources 
to small sources (existing and new), causing an increase in emissions 
nationally.
---------------------------------------------------------------------------

    \50\ Although the particulate matter standard for hazardous 
waste burning cement kilns in today's rule is the New Source 
Performance Standard expressed as on a mass basis (i.e., kg of 
particulate matter per megagram of dry feed to the kiln), this 
standard is not based on a ``mass of particulate matter emissions 
per unit of time'' that commenters suggest. Rather, the cement kiln 
standard can be equated to a concentration basis given that cement 
kilns emit a given quantity of combustion gas per unit of dry feed 
to the kiln. In fact, we proposed the cement kiln particulate matter 
standard on a concentration basis, 0.03 gr/dscf, that was calculated 
from the New Source Performance Standard when applied to a typical 
wet process cement kiln.
---------------------------------------------------------------------------

B. Why Are the Standards Corrected for Oxygen and Temperature?
    As proposed, the final standards are corrected to 7 percent oxygen 
and 20 deg.C because the data we use to establish the standards are 
corrected in this manner and because the current RCRA regulations for 
these sources require this correction. These corrections normalize the 
emissions data to a common base, recognizing the variation among the 
different combustors and modes of operation.
    Several commenters note that the proposed oxygen correction 
equation does not appropriately address hazardous waste combustors that 
use oxygen enrichment systems. They recommend that the Agency 
promulgate the oxygen correction factor equation proposed in 1990 for 
RCRA hazardous waste incinerators. See 55 FR at 17918 (April 27, 1990). 
We concur, and adopt the revised oxygen correction factor equation.
C. How Does the Rule Treat Significant Figures and Rounding?
    As proposed, the final rule establishes standards and limits based 
on two significant figures. One commenter notes that a minimum of three 
significant figures must be used for all intermediate calculations when 
rounding the results to two significant figures. We concur. Sources 
should use standard procedures, such as ASTM procedure E-29-90, to 
round final emission levels to two significant figures.

IV. How Are Nondioxin/Furan Organic Hazardous Air Pollutants 
Controlled?

    Nondioxin/furan organic hazardous air pollutants are controlled by 
a destruction and removal efficiency (DRE) standard and the carbon 
monoxide and hydrocarbon standards. Previous DRE tests demonstrating 
compliance with the 99.99% requirement under current RCRA regulations 
may be used to document compliance with the DRE standard provided that 
operations have not been changed in a way that could reasonably be 
expected to affect ability to meet the standard. However, if waste is 
fed at a point other than the flame zone, then compliance with the 
99.99% DRE standard must be demonstrated during each comprehensive 
performance test, and new operating parameter limits must be 
established to ensure that DRE is maintained. A 99.9999% DRE is 
required for those hazardous waste combustors burning dioxin-listed 
wastes. These requirements are discussed in Section IV.A. below.
    In addition, the rule establishes carbon monoxide and hydrocarbons 
emission standards as surrogates to ensure good combustion and control 
of nondioxin/furan organic hazardous air pollutants. Continuous 
monitoring and compliance with either the carbon monoxide or 
hydrocarbon emissions standard is required. If you choose to 
continuously monitor and comply with the carbon monoxide standard, you 
must also demonstrate during the comprehensive performance test 
compliance with the hydrocarbon emission standard. Additionally, you 
must also set operating limits on key parameters that affect combustion 
conditions to ensure continued compliance with the hydrocarbon emission 
standard. Alternatively, continuous monitoring and compliance with the 
hydrocarbon emissions standard eliminates the need to monitor carbon 
monoxide emissions because hydrocarbon emissions are a more direct 
surrogate of nondioxin/furan organic hazardous air pollutant emissions. 
These requirements are discussed in Section IV.B below.
A. What Is the Rationale for DRE as a MACT Standard?
    All sources must demonstrate the ability to destroy or remove 99.99

[[Page 52849]]

percent of selected principal organic hazardous compounds in the waste 
feed as a MACT standard. This requirement, commonly referred to as 
four-nines DRE, is a current RCRA requirement. We are promulgating the 
DRE requirement as a MACT floor standard to control the emissions of 
nondioxin organic hazardous air pollutants. The rule also requires 
sources to establish limits on specified operating parameters to ensure 
compliance with the DRE standard. See Part Five Section VII(B).
    In the April 1996 NPRM, we proposed that the four-nines DRE test 
requirement be retained under RCRA and be performed as part of a RCRA 
approved trial burn because we did not believe that the DRE test could 
be adequately implemented using the generally self-implementing MACT 
performance test and notification process.51 See 61 FR 
17447.
---------------------------------------------------------------------------

    \51\ Historically, under RCRA regulations, the permittiing 
authority and hazardous waste combustion source found it necessary 
to go through lengthy negotiations to develop a RCRA trial burn plan 
that adequately demonstrates the unit's ability to achieve four-
nines DRE.
---------------------------------------------------------------------------

    In response to the April proposal, however, we received comments 
that suggest the MACT comprehensive performance test and RCRA DRE trial 
burn could and should be combined, and that we should combine all stack 
air emission requirements for hazardous waste combustors into a single 
permit. Commenters are concerned that our proposed approach required 
sources to obtain two permits for air emissions and potentially be 
unnecessarily subject to dual enforcement.
    We investigated approaches that would achieve the goals of a single 
air emission permit and inclusion of DRE in MACT. We determined that 
the 40 CFR part 63 general provisions, applicable to all MACT regulated 
sources unless superseded, includes a process similar to the process to 
develop a RCRA trial burn test plan and allows permitting authorities 
to review and approve MACT performance test plans. See 40 CFR 63.7. 
Additionally, we determined that, because all hazardous waste 
combustors are currently required to achieve four-nines DRE, the DRE 
requirement could be included as a MACT floor standard rather than a 
RCRA requirement. In the May 1997 NODA, we discussed an alternative 
approach that used a modified form of the general provision's 
performance test plan and approval process. The approach would allow 
combination of the DRE test with the comprehensive performance test 
and, therefore, facilitate implementation of DRE as a MACT standard. We 
also discussed modifying the general approach to extend the performance 
test plan review period to one year in advance of the date a source 
plans to perform the comprehensive performance test. This extended 
review period would provide sufficient time for negotiations between 
permitting authorities and sources to develop and approve comprehensive 
performance test plans. These test plans would identify operating 
parameter limits necessary to ensure compliance with all the proposed 
MACT standards, as well as, implement the four-nines DRE test as a MACT 
floor standard. See 62 FR at 24241. Commenters support the process to 
combine the applicable stack emission requirements into a single 
permit. As for making the DRE test a MACT standard, we received no 
negative comments. Many commenters, however, question the need for 
subsequent DRE testing once a unit demonstrates four-nines DRE. See 
discussion and our response in Subsection 2 below.
    We believe that requiring the DRE test as a MACT standard is 
appropriate. As we previously noted, the four-nines DRE is firmly 
grounded statutory and regulatory requirement that has proven to be an 
effective method to determine appropriate process controls necessary 
for the combustion of hazardous waste. Specifically, RCRA requires that 
all hazardous waste incinerators must demonstrate the minimum 
technology requirement of four-nines DRE (RCRA section 3004(o)(1)(B)). 
Additionally, the current RCRA BIF regulations require that all boiler 
and industrial furnaces meet the four-nines DRE standard. Moreover, 
current RCRA regulations require all sources incinerating certain 
dioxin-listed contaminated wastes (F020-023 and F026-27) to achieve 
99.9999% (six-nines) DRE. See Secs. 264.343(a)(2) and 266.104(a)(3).
    The statutory requirement for incinerators to meet four-nines DRE 
can be satisfied if the associated MACT requirements ensure that 
incinerators will continue to meet the four-nines DRE minimum 
technology requirement, i.e., that MACT standards provide at least the 
``minimum'' RCRA section 3004(o)(1) level of control. To determine if 
the RCRA statutory requirements could be satisfied, we investigated 
whether DRE could be replaced with universal standards for key 
operating parameters based on previous DRE demonstrations (i.e., 
standards for carbon monoxide and hydrocarbon emissions). We found 
that, in the vast majority of DRE test conditions, if a unit operated 
with carbon monoxide levels of less than 100 ppmv and hydrocarbon 
emissions of less than 10 ppmv, the unit met or surpassed four-nines 
DRE. In a small number of test conditions, units emitted carbon 
monoxide and hydrocarbons at levels less than 100 and 10 ppmv 
respectively, but failed to meet four-nines DRE. Most failed test 
conditions were either due to questionable test results or faulty test 
design.52 See U.S. EPA, ``Draft Technical Support Document 
for HWC MACT Standards (NODA), Volume II: Evaluation of CO/HC and DRE 
Database,'' April 1997. Even though we could potentially explain the 
reasons these units failed to achieve four-nines DRE, we determined 
that universal carbon monoxide and hydrocarbon emissions limits may not 
ensure that all units achieve four-nines DRE because carbon monoxide 
and hydrocarbon emissions may not be representative of good combustion 
for all operating conditions that facilities may desire to operate. In 
addition, we could not identify a better method than the DRE test to 
limit combustion failures modes.
---------------------------------------------------------------------------

    \52\ In many of the failed test conditions that we investigated, 
the facility fed a low concentration of organic compound on which 
the DRE was being calculated. As has been observed many times, 
organic compounds can be reformed in the post combustion gas stream 
at concentrations sufficient to fail DRE. This is not indicative of 
a failure in the systems ability to destroy the compound, but is 
more likely the result of a poorly designed test. If the facility 
had fed a higher concentration of organic compound in the waste to 
the combustor, the unit would have been more likely to meet four-
nines DRE with no change in the operating conditions used during the 
test. In other cases, poor test design (i.e., firing aqueous organic 
waste into an unfired secondary combustion chamber) is considered to 
be the cause.
---------------------------------------------------------------------------

    Commenters state that the test conditions under which the DRE 
failures occurred involved feeding practices that were not common in 
the hazardous waste combustion industry. They further state that, if it 
could be ensured that hazardous waste ignited, hydrocarbon and carbon 
monoxide limits would be sufficient to ensure four-nines DRE is 
achieved continuously. Therefore, a DRE demonstration would not be 
warranted. Although we might agree in theory, the fact that tests were 
performed under these test conditions indicates that a source desired 
to operate in that fashion. Only the DRE test identified that the 
combustion failure occurred and was not susceptible to control via 
carbon monoxide and hydrocarbon emissions. This and other similar 
failures can lead to increased emissions of products of incomplete 
combustion and organic hazardous air pollutants. Also, as commenters 
acknowledge, carbon monoxide and hydrocarbon emissions were effective 
surrogates to ensure four-nines DRE only when

[[Page 52850]]

hazardous waste ignited. However, as we identified in the May 1997 
NODA, there are a number of hazardous waste combustion sources that 
operate in a manner that does not ensure ignition of hazardous waste.
    As a result of the DRE test investigation, we determined that a 
successful DRE demonstration is an effective, appropriate, and 
necessary method to identify operating parameter limits that ensure 
proper and achievable combustion of hazardous waste and to limit the 
emissions of organic hazardous air pollutants. Additionally, the DRE 
standard is a direct measure to ensure that the RCRA section 3004(o)(1) 
mandate and its protectiveness goals are being met, and also serves to 
maintain a consistent test protocol for sources combusting hazardous 
waste. The DRE demonstration requirement is also reasonable, provides a 
sound means to allow deferral of a RCRA mandate to the CAA, and 
simplifies implementation by having all stack emissions-related testing 
and compliance requirements promulgated under one statute, the CAA. 
Therefore, we retain the DRE demonstration as part of the MACT 
comprehensive performance test unless a DRE test has already been 
performed with no relevant changes.
1. MACT DRE Standard
    In today's rule, all affected sources are required to meet 99.99% 
DRE of selected Principal Organic Hazardous Constituents (POCs) that 
are as or more difficult to destroy than any organic hazardous 
pollutant fed to the unit. With one exception discussed in subsection 3 
below, this demonstration need be made only once during the operational 
life of a source, either before or during the initial comprehensive 
performance test, provided that the design, operation, and maintenance 
features do not change in a manner that could reasonably be expected to 
affect the ability to meet the DRE standard.
    The DRE demonstration involves feeding a known mass of POHC(s) to a 
combustion unit, and then measuring for that POHC(s) in stack 
emissions. If the POHC(s) is emitted at a level that exceeds 0.01% of 
the mass of the individual POHC(s) fed to the unit, the unit fails to 
demonstrate sufficient DRE.
    Operating limits for key combustion parameters are used to ensure 
four-nines DRE is maintained. The operating parameter limits are 
established based on operations during the DRE test. Examples of 
combustion parameters that are used to set operating limits include 
minimum combustion chamber temperature, minimum gas residence time, and 
maximum hazardous waste feedrate by mass. See Sec. 63.1209(j).
    Today's MACT DRE requirement is essentially the same as that 
currently required under RCRA. The main difference is that the vast 
majority of the MACT DRE demonstrations would not have to be repeated 
as often as currently required under RCRA, as discussed in section 3 
below.
2. How Can Previous Successful Demonstrations of DRE Be Used To 
Demonstrate Compliance?
    Except as discussed below, today's rule requires that, at least 
once during the operational life of a source during or before the 
initial comprehensive performance test, the source must demonstrate the 
ability to achieve 99.99% DRE and must set operating parameter limits 
to ensure that DRE is maintained. However, we recognize that many 
sources have already undergone approved DRE testing. Further, many 
facilities do not intend to modify their units design or operations in 
such a way that DRE performance or parameters would be adversely 
affected. Therefore, the Agency is allowing sources to use results from 
previous EPA or State-approved DRE demonstrations to fulfill the MACT 
four-nines DRE requirement, as well as to set the necessary operating 
limits on parameters that ensure continued compliance.
    If a facility wishes to operate under new operating parameter 
limits that could reasonably be expected to affect the ability to meet 
the standard, a new DRE demonstration must be performed before or 
concurrent with the comprehensive performance test. If the DRE 
operating limits conflict with operating parameter limits that are set 
to ensure compliance with other MACT standards, the unit must comply 
with the more stringent limits. Additionally, if a source is modified 
in such a way that its DRE operating limits are no longer applicable or 
valid, the source must perform a new DRE test. Moreover, if a source is 
modified in any way such that DRE performance or parameters are 
affected adversely, the source must perform a new DRE test.
3. DRE for Sources That Feed Waste at Locations Other Than the Flame 
Zone
    Today's rule requires sources that feed hazardous waste in 
locations other than the flame zone to perform periodic DRE tests to 
ensure that four-nines DRE continues to be achieved over the life of 
the unit. As indicated in the May 1997 NODA at 62 FR 25877, the Agency 
is concerned that these types of sources have a greater potential of 
varying DRE performance due to their waste firing practices. That is, 
due to the unique design and operation of the waste firing system, the 
DRE may vary over time, and those variations cannot be identified or 
limited through operating limits set during a single DRE test. For 
these units, we are requiring that DRE be verified during each 
comprehensive performance test and that new operating parameter limits 
be established to ensure continued compliance.
4. Sources That Feed Dioxin Wastes
    In today's rule, we are requiring all sources that feed certain 
dioxin-listed wastes (i.e., F020-F023, F026, F027) to demonstrate the 
ability to achieve 99.9999 percent (six-nines) DRE as a MACT standard. 
This requirement will serve to achieve a number of goals associated 
with today's regulations. First, under RCRA, six-nines DRE is required 
when burning certain dioxin-listed wastes. If we did not promulgate 
this requirement as a MACT standard, sources that feed dioxin-listed 
waste would be required to maintain two permits to manage their air 
emissions. Thus, by including this requirement as a MACT standard, we 
eliminate any unnecessary duplication. That outcome is contrary to our 
goal which is to limit, to the greatest extent possible, the need for 
sources to obtain two permits governing air emissions under different 
statutory authorities. Second, six-nines DRE helps to improve control 
of nondioxin organic hazardous air pollutants as well. Finally, this 
requirement properly reflects floor control for sources that feed 
dioxin-listed wastes. Currently, all sources that feed dioxin listed 
wastes must achieve six-nines DRE. Before making the decision to 
include six-nines DRE as a MACT standard, we considered whether the 
requirements could be eliminated given that we are issuing dioxin/furan 
emission standards with today's rule. We concluded, first, that we had 
not provided sufficient notice and comment to depart from the current 
regulations applicable to these sources. Second, we also decided that 
because we currently require other similar highly toxic bioaccumulative 
and persistent compounds (e.g., PCB wastes) to be fed to units that 
demonstrate six-nines DRE, a departure from that policy for RCRA dioxin 
wastes would be inconsistent. Finally, we are in discussions that may 
cause us to reevaluate our overall approach to dioxin-listed wastes, 
with the potential to impact this rule and the land disposal 
restrictions program. Any changes to our approach will be included in a 
single rulemaking that would be proposed later.

[[Page 52851]]

B. What Is the Rationale for Carbon Monoxide or Hydrocarbon Standards 
as Surrogate Control of Organic Hazardous Air Pollutants?
    Today's rule adopts limits on emissions of carbon monoxide and 
hydrocarbons as surrogates to ensure good combustion and control of 
nondioxin organic hazardous air pollutants. We require continuous 
emissions monitoring and compliance with either the carbon monoxide or 
hydrocarbon emissions standard. Sources can choose which of these two 
standards it wishes to continuously monitor for compliance. If a source 
chooses the carbon monoxide standard, it must also demonstrate during 
the comprehensive performance test compliance with the hydrocarbon 
emission standard. During this test the source also must set operating 
limits on key parameters that affect combustion conditions to ensure 
continued compliance with the hydrocarbon emission standard. These 
parameters relate to good combustion practices and are identical to 
those for which you must establish limits under the DRE standard. See 
Sec. 63.109(a)(7) and 63.1209(j). However, this source need not install 
and use a continuous hydrocarbon monitor to ensure continued compliance 
with the hydrocarbon standard. As discussed previously, the limits 
established for DRE are identical. If a source elects to use the 
hydrocarbon limit for compliance, then it must continuously monitor and 
comply with the hydrocarbon emissions standard. However, this type of 
source need not monitor carbon monoxide emissions or carbon monoxide 
operating parameters because hydrocarbon emissions are a more direct 
surrogate of nondioxin organic hazardous air pollutant emissions.
    The April 1996 NPRM proposed MACT emission standards for both 
carbon monoxide and hydrocarbon as surrogates to control emissions of 
nondioxin organic hazardous air pollutants. We also proposed that 
cement kilns comply with either a carbon monoxide or hydrocarbons 
standard due to raw material considerations.53 See 61 FR at 
17375-6. Our reliance on only carbon monoxide or only hydrocarbon has 
drawbacks, and therefore we proposed that incinerators and lightweight 
aggregate kilns comply with emissions standards for both. Nonetheless, 
we also acknowledged that requiring compliance with both carbon 
monoxide and hydrocarbon standards may be redundant, and requested 
comment on: (1) Giving sources the option of complying with either 
carbon monoxide or hydrocarbon emission standards; or (2) establishing 
a MACT standard for either carbon monoxide or hydrocarbon, but not 
both.
---------------------------------------------------------------------------

    \53\ See discussion regarding cement kilns compliance with the 
carbon monoxide and/or hydrocarbon standards in Part Four, Section 
VII.D.
---------------------------------------------------------------------------

    Comments to our proposed approach question the necessity of two 
related surrogates to control organic hazardous air pollutants. Many 
commenters assert they are capable of controlling hydrocarbon emissions 
effectively, but due to their system's unique design, they could not 
comply continuously with the carbon monoxide emission standard. In 
general, commenters prefer an approach that would afford them maximum 
flexibility in demonstrating compliance with organic control standards, 
i.e., more like option (1) in the NPRM.
    The May 1997 NODA included a refined version of the option that 
commenters prefer that allowed sources to monitor and comply with 
either a carbon monoxide or hydrocarbon emission standard. In response 
to the May 1997 NODA, commenters nearly unanimously support the option 
that allowed facilities to monitor and comply with either the carbon 
monoxide or hydrocarbon standard as surrogates to limit emissions of 
nondioxin organic hazardous air pollutants. However, a few commenters 
suggest that compliance with carbon monoxide or hydrocarbons in 
combination with DRE testing is redundant and unnecessary. However, in 
their comments, they do not address the issue of DRE failures 
associated with low carbon monoxide or hydrocarbon emissions, other 
than to state that if ignition failure was avoided, emissions of carbon 
monoxide or hydrocarbons would be good indicators of combustion 
efficiency and four-nines DRE. This does not address our concerns, 
which reflect cases in which ignition failures did not occur and in 
which destruction and removal efficiencies were not met.
    In the May 1997 NODA, we discussed another option that required 
sources to comply with the hydrocarbon emission standard and establish 
a site-specific carbon monoxide limit higher than 100 ppmv. This option 
was developed because compliance with the hydrocarbon standard assures 
control of nondioxin organic hazardous air pollutants, and a site-
specific carbon monoxide limit aids compliance by providing advanced 
information regarding combustion efficiency. However, we conclude that 
this option may be best applied as a site-specific remedy in situations 
where a source has trouble maintaining compliance with the hydrocarbon 
standard.
    Today's final rule modifies the May 1997 NODA approach slightly. 
Complying with the carbon monoxide standard now requires documentation 
that hydrocarbon emissions during the performance test are lower than 
the standard, and requires operating limits on parameters that affect 
hydrocarbon emissions. We adopt this modification because some data 
show that high hydrocarbon emissions are possible while simultaneously 
low carbon monoxide emissions are found.54
---------------------------------------------------------------------------

    \54\ In a number of instances, RCRA compliance test records 
showed that sources emitting carbon monoxide at less than 100 ppmv 
emitted hydrocarbons in excess of 10 ppmv.
---------------------------------------------------------------------------

    In the BIF rule (56 FR at 7149-50), we found that both monitoring 
and compliance with either carbon monoxide or hydrocarbon limits and 
achieving four-nines DRE is needed to ensure control of products of 
incomplete combustion (including nondioxin organic hazardous air 
pollutants) that are a result of hazardous waste combustion. DRE, 
although sensitive to identifying combustion failure modes, cannot 
independently ensure that emissions of products of incomplete 
combustion or organic hazardous air pollutants are being controlled. 
DRE can only provide the assurance that, if a hazardous waste combustor 
is operating normally, the source has the capability to transform 
hazardous and toxic organic compounds into different compounds through 
oxidation. These other compounds can include carbon dioxide, water, and 
other organic hazardous air pollutants. Because carbon monoxide 
provides immediate information regarding combustion efficiency 
potentially leading to emissions of organic hazardous air pollutants 
and hydrocarbon provides a direct measure of organic emissions, these 
two parameters individually or in combination provide additional 
control that would not be realized with the DRE operating parameter 
limits alone.55 Neither our data nor data supplied by 
commenters show that only monitoring

[[Page 52852]]

carbon monoxide, hydrocarbons, or DRE by itself can adequately ensure 
control of nondioxin organics. Therefore, the approach used in the BIF 
rule still provides the best regulatory model. We conclude in today's 
rule that hydrocarbons and carbon monoxide monitoring are not redundant 
with the DRE testing requirement to control emissions of organic 
hazardous air pollutants and require both standards. For an additional 
discussion regarding the use of hydrocarbons and carbon monoxide to 
control emissions of organic hazardous air pollutants, see USEPA, 
``Technical Support Document for HWC MACT Standards, Volume III: 
Selection of MACT Standards and Technologies,'' July 1999.
---------------------------------------------------------------------------

    \55\ We acknowledge that although hydrocarbon emissions are a 
direct measure of organic emissions, they are measured with a 
continuous emissions monitoring system known as a flame ionization 
detector. Some data suggest hydrocarbon flame ionization detectors 
do not respond with the same sensitivity to the full spectrum of 
organic compounds that may be present in the combustion gas. 
Additionally, combustion gas conditions also may affect the 
sensitivity and accuracy of the monitor. Nonetheless, monitoring 
hydrocarbons with these detectors appears to be the best method 
reasonably available to provide real-time monitoring of organic 
emissions from a hazardous waste combustor.
---------------------------------------------------------------------------

V. What Methodology Is Used To Identify MACT Floors?

    This section discusses: (1) Methods used to identify MACT floor 
controls and emission levels for the final rule; (2) the rationale for 
using hazardous waste feedrate control as part of MACT floor control 
for the metals and total chlorine standards; (3) alternative methods 
for establishing floor levels considered at proposal and in the May 
1997 NODA; and (4) our consideration of emissions variability in 
identifying MACT floor levels.
A. What Is the CAA Statutory Requirement To Identify MACT Floors?
    We identify hazardous waste incinerators, hazardous waste burning 
cement kilns, and hazardous waste burning lightweight aggregate kilns 
as source categories to be regulated under section 112. We must, 
therefore, develop MACT standards for each category to control 
emissions of hazardous air pollutants. Under CAA section 112, we may 
distinguish among classes, types and sizes of sources within a category 
in establishing such standards.
    Section 112 prescribes a minimum baseline or ``floor'' for 
standards. For new sources, the standards for a source category cannot 
be less stringent than the emission control that is achieved in 
practice by the best-controlled similar source. Section 112(d)(3). The 
standards for existing sources may be less stringent than standards for 
new sources, but cannot be less stringent than ``(A) * * * the average 
emissions limitation achieved by the best performing 12 percent of the 
existing sources (for which the Administrator has emissions 
information) * * *, in the category or subcategory for categories and 
subcategories with 30 or more sources, or (B) the average emissions 
limitation achieved by the best performing 5 sources (for which the 
Administrator has or could reasonably obtain emissions information) in 
the category or subcategory for categories and subcategories with fewer 
than 30 sources.'' Id.
    We also must consider a more stringent standard than the floor, 
referred to in today's rule as a ``beyond-the-floor'' standard. For 
each beyond-the-floor analysis, we evaluate the maximum degree in 
reduction of hazardous air pollutants determined to be achievable, 
taking into account the cost of achieving those reductions, nonair 
quality health and environmental impacts, and energy costs. Section 
112(d)(2). The object of a beyond-the-floor standard is to achieve the 
maximum degree of emission reduction without unreasonable economic, 
energy, or secondary environmental impacts.
B. What Is the Final Rule Floor Methodology?
    Today's rule establishes MACT standards for the following hazardous 
air pollutants, hazardous air pollutant groups or hazardous air 
pollutant surrogates: dioxin/furans, mercury, two semivolatile metals 
(lead and cadmium), three low volatile metals (arsenic, beryllium, and 
chromium), particulate matter, total chlorine (hydrochloric acid and 
chlorine gas), carbon monoxide, hydrocarbons, and destruction and 
removal efficiency. This subsection discusses the overall engineering 
evaluation and data analysis methods we used to establish MACT floors 
for these standards. Additional detail on the specific application of 
these methods for each source category and standard is presented in 
Part Four, Sections VI-VIII, of the preamble and in the technical 
support document.56
---------------------------------------------------------------------------

    \56\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

1. What Is the General Approach Used in This Final Rule?
    The starting point in developing standards is to determine a MACT 
floor emission level, the most lenient level at which a standard can be 
set. To identify the floor level, we first identified the control 
techniques used by the best performing sources. We designate these best 
performing sources the ``MACT pool'' and the emission control 
technologies they use we call ``MACT floor controls.''
    After identifying the MACT pool and MACT floor controls, we 
determine the emission level that the MACT floor controls are routinely 
achieving--that is, an achievable emission level taking into account 
normal operating variability (i.e., variability inherent in a properly 
designed and operated control system). This is called the floor 
emission level. To ensure that the floor emission level is being 
achieved by all sources using floor controls (i.e., not just the MACT 
pool sources), we generally consider emissions data from all sources in 
a source category that use well-designed and properly operated MACT 
floor controls. (We call the data set of all sources using floor 
controls the ``expanded MACT pool.'') Floor levels in this rule are 
generally established as the level achieved by the source in the 
expanded MACT pool with the highest emissions average 57 
using well-designed and properly operated MACT floor controls.
---------------------------------------------------------------------------

    \57\ Each source's emissions usually are expressed as an average 
of three or more emission measurements at the same set of operating 
parameters. This is because compliance is based on the average of 
three or more runs.
---------------------------------------------------------------------------

    Several commenters oppose considering emissions data from all 
sources using MACT floor controls (i.e., the expanded MACT pool) 
because they assert the expansion of the MACT pool results in inflated 
floors. If we adopt these commenters' recommendation, then many sources 
using MACT controls would not meet the standard, even though they were 
using MACT floor control. (Indeed, in some cases, other test conditions 
from the very system used to establish the MACT pool would not meet the 
standard, notwithstanding no significant change in the system's design 
and operation.) This result is inappropriate in that all sources using 
properly designed and operated MACT floor controls should achieve the 
floor emission level if the technology is well designed and operated. 
In the absence of data indicating a design or operation problem, we 
assume the floor emission level based on an expanded MACT pool reflects 
an emission level consistently achievable by MACT floor technology. Our 
resulting limits account for the fact that sources and emissions 
controls will experience normal operating variability even when 
properly designed and operated.
    The MACT floor methodology in this rule does not use a single 
uniform data analysis approach consistently across all three source 
categories and standards. Our data analysis methods vary due to: (1) 
Limitations of our emissions data and ancillary information; (2) 
emissions of some hazardous air pollutants being related to the 
feedrate of the hazardous air pollutant (e.g., semivolatile metal 
emissions are affected by semivolatile metal feedrates) while emissions 
of

[[Page 52853]]

other hazardous air pollutants are not (e.g., dioxin/furan emissions 
are related to postcombustion dioxin/furan formation rather than 
dioxin/furan feedrates); (3) the various types of emissions controls 
currently in use which do not lend themselves to one type of MACT 
analysis; and (4) consideration of existing regulations as themselves 
establishing floor levels.
    Finally, as discussed in Section D, the MACT floor levels 
established through our data analysis approaches account for emissions 
variability without the separate addition of a statistically-derived 
emissions variability factor.
2. What MACT Floor Approach Is Used for Each Standard?
    a. Dioxins and Furans. For dioxins and furans, we adopt the MACT 
floor methodology discussed in the May 1997 NODA. Based on engineering 
information and principles, we identify temperature of combustion gas 
at the particulate matter control device of 400 deg.F or less as MACT 
floor control of dioxin/furan. This technology and level of control has 
been selected because postcombustion formation of dioxin/furan is 
suppressed by lowering postcombustion gas temperatures, and formation 
is reasonably minimized at gas temperatures of 400 deg.F or below. 
Sources controlling gas temperatures to 400 deg.F or less at the 
particulate matter control device represent the level achieved by the 
median of the best performing 12 percent of sources where the source 
category has more than 30 sources (or the median of the best performing 
five sources where the source category has fewer than 30 sources).
    The next step is to identify an emissions level that MACT floor 
control achieved on a routine basis. We analyzed the emissions data 
from all sources (within each source category) using MACT floor control 
and establish the floor level equal to the highest test condition 
average.
    As discussed in greater detail in Part Four, Section VI, 
incinerators with waste heat recovery boilers present a unique 
situation for dioxin/furan control. Our data base shows that 
incinerators equipped with waste heat recovery boilers have 
significantly higher dioxin/furan emissions compared to other 
incinerators. In the waste heat recovery boiler, combustion gas is 
exposed to particles on boiler tubes within the temperature window of 
450 deg. F to 650 deg. F, which promotes surface-catalyzed formation of 
dioxin/furan. Therefore, we establish separate dioxin/furan standards 
for incinerators with waste heat boilers and incinerators without waste 
heat boilers.58 The specified floor control for both waste 
heat boilers and nonwaste heat boilers is combustion gas temperature 
control to 400 deg.F or less at the particulate matter control 
device.59 Floor levels for waste heat boiler incinerators 
are much higher, however, because of the dioxin/furan formation during 
the relatively slow temperature quench in the boiler. See the 
incinerator dioxin/furan discussion in Part Four, Section VI, of 
today's rule for more details.
---------------------------------------------------------------------------

    \58\ We concluded that separate standards to control other 
hazardous air pollutants were not needed for waste heat boiler-
equipped incinerators versus other incinerators. That is, whether or 
not the incinerator is equipped with a waste heat recovery boiler is 
only of concern for dioxin/furan emissions, not the other hazardous 
air pollutants.
    \59\ Wet particulate matter control devices (e.g., venturi 
scrubbers) inherently preclude dioxin/furan formation because: (1) 
They do not suspend particulate matter in the combustion gas flow as 
do fabric filters and electrostatic precipitators, and (2) gas 
temperatures are below 400 deg.F in the scrubber. Given this, floor 
control is use of a wet particulate matter control device or control 
of combustion gas temperature to 400 deg.F or below at the inlet to 
a dry particulate matter control device.
---------------------------------------------------------------------------

    b. What MACT Floor Methodology Is Used for Particulate Matter? We 
adopt a final MACT floor methodology for particulate matter based on 
the approaches discussed in the May 1997 NODA. For incinerators, the 
final MACT floor is determined through engineering principles and 
information, coupled with analysis of the emissions data base. For 
cement kilns, we base final MACT on the existing requirements of the 
New Source Performance Standard applicable to Portland cement kilns. 
Finally, for lightweight aggregate kilns, the final floor level is 
derived directly from the emissions data base (i.e., the highest test 
condition average for sources using properly designed and operated 
floor control).
    i. Incinerators. Today's rule identifies MACT floor control as 
either a well-designed, operated, and maintained fabric filter, 
ionizing wet scrubber, or electrostatic precipitator, based on 
engineering information and an evaluation of the particulate matter 
control equipment used by at least the median of the best performing 12 
percent of sources and the emission levels achieved. These types of 
particulate matter control equipment routinely and consistently achieve 
superior particulate matter performance relative to other controls used 
by the incinerator source category and thus represent MACT. Using 
generally accepted engineering information and principles, we then 
identify an emission level that well-designed, operated and maintained 
fabric filters, ionizing wet scrubbers, and electrostatic precipitators 
routinely achieve.
    The floor level is not directly identified from the emissions data 
base as the highest test condition average for sources using a fabric 
filter, ionizing wet scrubber, or electrostatic precipitator. The 
hazardous waste combustor incinerator data base, however, was used as a 
tool to determine if the identified floor level, established on 
generally accepted engineering information and principles, is in 
general agreement with available particulate matter data. This is 
because we do not have adequate data on the features of the control 
devices to accurately distinguish only those devices that are well-
designed, operated, and maintained and thus representative of MACT. 
Several sources in the emissions data base that are equipped with 
fabric filters, ionizing wet scrubbers, or electrostatic precipitators 
have emission levels well above the emission levels of other sources 
equipped with those devices. This strongly suggests that the higher 
levels are not representative of those achieved by well-designed, 
operated, and maintained units, even when normal operating variability 
is considered. We accordingly did not use these data in establishing 
the standard. See Kennecott v. EPA, 780 F.2d 445, 458 (4th Cir. 1985) 
(EPA ``can reject data it reasonably believes to be unreliable 
including performance data that is higher than other plants operating 
the same control technology.'')
    ii. Cement Kilns. As discussed in the May 1997 NODA and in more 
detail in the standards section for cement kilns in Part Four, Section 
VII, we base the MACT floor emission level on use of a fabric filter or 
electrostatic precipitator to achieve the New Source Performance 
Standard for Portland cement kilns. The MACT floor is equivalent to and 
expressed as the current New Source Performance Standard of 0.15 kg/Mg 
dry feed (0.30 lb/ton dry feed). In the NPRM and the May 1997 NODA, we 
proposed to express the particulate matter standard on a concentration 
basis. However, because we are not yet requiring sources to document 
compliance with the particulate matter standard by using a particulate 
matter continuous emissions monitoring system in this final rule, we 
establish and express the floor emission level equivalent to the New 
Source Performance Standard. Commenters' concerns about separate MACT 
pools for particulate matter, semivolatile metals, and low volatile 
metals are discussed in Part Four, Section VII.
    iii. Lightweight Aggregate Kilns. All lightweight aggregate kilns 
burning

[[Page 52854]]

hazardous waste are equipped with fabric filters. We could not 
distinguish only those sources with fabric filters better designed, 
operated, and maintained than others, and thus represent MACT control. 
Because we could not independently use engineering information and 
principles to otherwise distinguish which well-designed, operated, and 
maintained fabric filters are routinely achieving levels below the 
highest test condition average in the emissions data base (i.e., 
considering the high inlet grain loadings for lightweight aggregate 
kilns), we establish the floor level as that highest test condition 
average emission level. Commenters concerns about a high floor level 
and separate MACT pools for particulate matter, semivolatile metals, 
and low volatile metals are discussed in Part Four, Section VIII.
    c. Metals and Total Chlorine. This rule establishes MACT standards 
for mercury; semivolatile metals comprised of combined emissions of 
lead and cadmium; low volatility metals comprised of combined emissions 
of arsenic, beryllium, and chromium; and total chlorine comprised of 
combined emissions of hydrogen chloride and chlorine gas. As shown by 
the following analysis, these hazardous air pollutants are all 
controlled by the best performing sources, at least in part, by 
feedrate control of the metal or chlorine in the hazardous waste. In 
addition to hazardous waste feedrate control, some of the hazardous air 
pollutants also are controlled by air pollution control equipment. Both 
semivolatile metals and low volatile metals are controlled by a 
combination of hazardous waste metal feedrate control and by 
particulate matter control equipment. Total chlorine is controlled by a 
combination of feedrate control and, for hazardous waste incinerators, 
scrubbing equipment designed to remove acid gases.
    i. How Are the Metals and Chlorine Floor Control(s) Identified? We 
follow the language of CAA section 112(d)(3) to identify the control 
techniques used by the best performing sources. The hazardous waste 
incinerator and hazardous waste cement kiln source categories are 
comprised of 186 and 33 sources, respectively. From the statutory 
language, we conclude that for this analysis the control techniques 
used by the best performing 6% of sources represents the average of the 
best performing 12% of the sources in those categories. It follows, 
therefore, that floor control for metals and chlorine is the 
technique(s) used by the best performing 12 incinerators and two cement 
kilns.
    Because the hazardous waste lightweight aggregate kiln source 
category is comprised of only 10 sources, we follow the language of 
section 112(d)(3)(B) to identify the control technique(s) used by the 
three best performing sources, which represents the median of the best 
performing five sources.
    Our floor control analysis indicates that the best performing 12 
incinerators, two cement kilns, and three lightweight aggregate kilns 
all use hazardous waste feedrate control to limit emissions of mercury, 
semivolatile metal, low volatile metal, and total chlorine. For the 
semivolatile and low volatile metals, the best performing sources also 
use particulate matter control as part of the floor control technique. 
In addition, the best performing incinerator sources also control total 
chlorine and mercury with wet scrubbing. Accordingly, we identify floor 
control for semivolatile metal and low volatile metal as hazardous 
waste feedrate control plus particulate matter control, and floor 
control for incinerators for total chlorine and mercury as hazardous 
waste feedrate control plus wet scrubbing.
    ii. What is the Rationale for Using Hazardous Waste Feedrate 
Control as MACT Floor Control Technique? As discussed above, MACT floor 
control for mercury, semivolatile metals, low volatile metals, and 
total chlorine is based on, or at least partially based on, feedrate 
control of metal and chlorine in the hazardous waste. The feedrate of 
metal hazardous air pollutants will affect emissions of those 
pollutants, and the feedrate of chlorine will affect emissions of total 
chlorine (i.e., hydrochloric acid and chlorine gas) because metals and 
chlorine are elements and are not destroyed during combustion. 
Emissions controls, if any, control only a percentage of the metal or 
total chlorine fed. Therefore, as concentrations of metals and total 
chlorine in the inlet to the control device increase, emissions 
increase.
    At proposal, we identified hazardous waste feedrates as part of the 
technology basis for the proposed floor emission 
standards.60 MACT maximum theoretical emission 
concentrations 61 (MTECs) were established individually for 
mercury, semivolatile metals, low volatile metals, and total chlorine 
at a level equal to the highest MTEC of the average of the best 
performing 12% of sources. For some hazardous air pollutants, hazardous 
waste feedrate control of metals and chlorine was identified as the 
sole component of floor control (i.e., where the best performing 
existing sources do not use pollution control equipment to remove the 
hazardous air pollutant). Examples include mercury and total chlorine 
from cement kilns. For other hazardous air pollutants, we identified 
hazardous waste feedrate control of metals and chlorine as a partial 
component of MACT floor control (e.g., floor control for semivolatile 
metals include good particulate matter control in addition to feedrate 
control of semivolatile metals in hazardous waste).
---------------------------------------------------------------------------

    \60\ See 61 FR at 17366.
    \61\ We developed a term, Maximum Theoretical Emissions 
Concentration, to compare metals and chlorine feedrates across 
sources of different sizes. MTEC is defined as the metals or 
chlorine feedrate divided by the gas flow rate, and is expressed in 
g/dscm.
---------------------------------------------------------------------------

    In the May 1997 NODA, we continued to consider hazardous waste 
feedrate control of metals and chlorine as a valid floor control 
technology. However, rather than defining a specific MACT control 
feedrate level (expressed as a MTEC), we instead relied on another 
analysis tool, an emissions breakpoint analysis, to identify sources 
feeding metals and/or chlorine at high (and not MACT) levels. At the 
time, we believed that the breakpoint analysis was a less problematic 
approach to identify sources using MACT floor control than the 
approaches proposed initially.62
---------------------------------------------------------------------------

    \62\ Comments had objected to our proposed approach of defining 
MTECs as too reliant on engineering inspection of the data.
---------------------------------------------------------------------------

    Given commenters' subsequent concerns with the emissions breakpoint 
analysis as well (see discussion in Section C below), we conclude that 
specifying MTECs as MACT control (partially or solely) is necessary to 
properly reflect the feedrate component of MACT control.
    Notwithstanding how the MACT floor MTEC is defined, many commenters 
suggest that our consideration of hazardous waste feedrate as a floor 
control technique is inappropriate in a technology-based rulemaking and 
not permissible under the CAA. Commenters also state that hazardous 
waste feedrate control is not a control technique due to the wide 
variations in metals and chlorine in the hazardous waste generated at a 
single facility location. Further, they believe even greater variations 
occur in metals and chlorine levels in the hazardous waste generated at 
multiple production sites representing different industrial sectors. 
Thus, commenters suggest that basing a floor emission level on data 
from sources that feed hazardous waste with low levels of metals or 
chlorine is tantamount to declaring that wastes with higher levels of 
metals or chlorine are not to be generated. Other

[[Page 52855]]

commenters note, however, that hazardous waste feedrate control must be 
considered as a floor control technique because feedrate control is 
being used as a control means to comply with existing RCRA regulations 
for these combustors. Still other commenters recommend that we 
establish uniform hazardous waste feedrate limits (i.e., base the 
standard on an emission concentration coupled with a hazardous waste 
feedrate limit on metals and chlorine) across all three hazardous waste 
combustor source categories. Please refer to Part Five, Section 
VII.D.3.c.iv of today's preamble and the Comment Response Document for 
detailed responses to these comments.
    We do not accept the argument that control of hazardous waste 
metals and chlorine levels in hazardous waste cannot be part of the 
floor technology. First, control of hazardous air pollutants in 
hazardous waste feedstock(s) can be part of a MACT standard under 
section 112(d)(2)(A), which clearly indicates that material 
substitution can be part of MACT. Second, hazardous waste combustors 
are presently controlling the level of metal hazardous air pollutants 
and chlorine in the hazardous waste combusted because of RCRA 
regulatory requirements. (See Sec. 266.103(c)(1) and (j) where metal 
and chlorine feedrate controls are required, and where monitoring of 
feedrates are required.) Simply because these existing controls are 
risk-based, rather than technology-based, does not mean that they are 
not means of controlling air emissions cognizable under the CAA. Floor 
standards are to be based on ``emission limitation[s]'' achieved by the 
best existing sources. An ``emission limitation'' includes ``a 
requirement established by the * * * Administrator which limits the 
quantity, rate, or concentration of emissions. * * * including any 
requirement relating to the operation * * * of a source. * * *'' CAA 
section 302(k). This is precisely what current regulations require to 
control metal and chlorine levels in hazardous waste feed.
    Commenters also note that contemplated floor levels were lower than 
the feed limits specified in current regulations for boilers and 
industrial furnaces. This is true, but not an impediment to identifying 
achievable MACT floor levels. Actual performance levels can serve as a 
basis for a floor. An analogy would be where a group of facilities 
achieve better capture efficiency from air pollution control devices 
than required by existing rule. That level of performance (if generally 
achievable) can serve as the basis for a floor standard. Accordingly, 
we use hazardous waste feedrate, entirely or partially, to determine 
floor levels and beyond-the-floor levels for mercury, semivolatile 
metals, low volatile metals, and total chlorine.
    iii. How Are Feedrate and Emissions Levels Representative of MACT 
Floor Control Identified? After identifying feedrate control as floor 
control, we use a data analysis method called the ``aggregate feedrate 
approach'' to establish floor control hazardous waste feedrate levels 
and emission levels for mercury, semivolatile metals, low volatile 
metals, and total chlorine. The first step in the aggregate feedrate 
approach is to identify an appropriate level of aggregated mercury, 
semivolatile metals, low volatile metals, and total chlorine feedrate 
control, expressed as a MTEC, being achieved in practice by the best 
performing incinerator, cement kiln and lightweight aggregate kiln 
sources. This aggregate MTEC level is derived only from the sources 
using MACT floor emission controls.
    The aggregate feedrate approach involves four steps: (1) 
Identifying test conditions in the data base where data are available 
to calculate hazardous waste feedrate MTECs for all three metal 
hazardous air pollutant groups and total chlorine; (2) screening out 
test conditions where a source was not using the MACT floor emission 
control device for hazardous air pollutants that are cocontrolled by an 
air pollution control device 63; (3) ranking the individual 
hazardous air pollutant MTECs, from the different source test 
conditions, from lowest to highest and assigning each a numerical rank, 
with a rank of one being the lowest MTEC; and (4) summing, for each 
test condition, the individual ranking for each of the hazardous air 
pollutants to determine a composite ranking. The total sum is used to 
provide an overall assessment of the aggregate level of hazardous air 
pollutants in the hazardous waste for each test condition. The 
hazardous waste feed streams with lower total sums (i.e., hazardous air 
pollutant levels) are ``cleaner'' in aggregate than those with higher 
total sums.64 (See the technical support document for more 
details on this procedure.65)
---------------------------------------------------------------------------

    \63\ For example, to potentially be considered a MACT-controlled 
incinerator with respect to both the emissions control device and 
hazardous waste metals and chlorine feedrate, the incinerator must 
use a wet scrubber for hydrochloric acid and mercury control and 
must use either a fabric filter, ionizing wet scrubber, or 
electrostatic precipitator and achieve the floor particulate matter 
level of 0.015 gr/dscf. Similarly, cement kilns must achieve the 
particulate matter MACT floor (for this analysis only, the New 
Source Performance Standard was converted to an estimated equivalent 
stack gas concentration of 0.03 gr/dscf) and lightweight aggregate 
kilns must meet the particulate matter MACT floor of 0.025 gr/dscf. 
There is no MACT floor hydrochloric acid emissions control device 
for cement kilns and lightweight aggregate kilns.
    \64\ This aggregate hazardous waste MTEC ranking is done 
separately for each of the three combustor source categories.
    \65\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    The aggregate MTEC ranking process results in aggregate feedrate 
data from nine incinerators, 10 cement kilns, and 10 lightweight 
aggregate kilns from which to select an appropriate level of feedrate 
control representative of MACT floor control.66 We 
considered selecting the source with either the highest or lowest 
aggregate MTEC in each source category to represent MACT floor control, 
but did not believe this was appropriate based on concerns about 
representativeness and achievability. We conclude that it is 
reasonable, however, to consider the best 50% of the sources for which 
we have data in each source category as the best performing sources. 
This is because, for incinerators and cement kilns, we have only a few 
sources with complete aggregate MTEC data relative to the size of the 
source category. The best 50% of the sources for these categories 
equates to five sources, given that we have aggregate MTEC data for 
nine incinerators and 10 cement kilns. For lightweight aggregate kilns, 
this equates also to five sources given that we have aggregate MTEC 
data for 10 lightweight aggregate kiln sources.
---------------------------------------------------------------------------

    \66\ Only nine incinerators were ultimately used because (1) We 
have complete metal emissions data on relatively few sources, and 
(2) many sources do not use particulate matter floor control, a 
major means of controlling semivolatile metals and low volatile 
metals.
---------------------------------------------------------------------------

    Additionally, we conclude it is appropriate to identify a feedrate 
MTEC representative of floor control based on the median of the best 
performing five sources. In selecting a representative sample and 
identifying the appropriate MTEC floor control level, we draw guidance 
from section 112(d)(3)(B), in which Congress requires the Agency to use 
the average of the best performing five sources when faced with small 
source categories (i.e., less than 30 sources), and therefore limited 
data, to establish a MACT floor. In addition, this methodology is 
reasonable and appropriate because it allows consideration of a number 
of best performing sources (i.e., five), which is within the range of 
reasonable values we could have selected.
    We considered an approach that selected both the control technique 
and level of control as the average of the best performing 12% of 
incinerator and

[[Page 52856]]

cement kiln sources for which we have aggregate MTEC data. This 
approach resulted in using only the best single source as 
representative of MACT floor control for all existing sources because 
there are only nine incinerators and 10 cement kilns for which we have 
adequate aggregate data. However, the level of feedrate control 
achieved by the single best performing existing source is likely not 
representative of the range of higher feedrate levels achieved by the 
best performing existing sources and, indeed, would inappropriately 
establish as a floor what amounts to a new source standard.
    The final step of the aggregate feedrate approach is to determine 
an emission level that is routinely achieved by sources using MACT 
floor control(s). Similar to the April 1996 NPRM and May 1997 NODA, we 
evaluated all available data for each test condition to determine if a 
hazardous air pollutant is fed at levels at or below the MACT floor 
control MTEC. If so, the test condition is added to the expanded MACT 
pool for that hazardous air pollutant.67 We then define the 
floor emission level for the hazardous air pollutant/hazardous air 
pollutant group as the level achieved by the source with the highest 
emissions average in the MACT expanded pool.
---------------------------------------------------------------------------

    \67\ The expanded MACT pool for each hazardous air pollutant is 
comprised of test conditions from sources equipped with the 
prescribed MACT floor emission control device, if any, and feeding 
hazardous waste at an MTEC not exceeding the MACT floor MTEC for 
that hazardous air pollutant.
---------------------------------------------------------------------------

    The aggregate feedrate approach is a logical and reasonable 
outgrowth of the aggregate hazardous air pollutant approach to 
establish floor emission levels that we discussed in the April 1996 
NPRM. The initial proposal determined MACT floors separately for each 
hazardous air pollutant controlled by a different control technology, 
but we also proposed an alternative whereby floors would be set on the 
basis of a source's performance for all hazardous air pollutants.
    Many commenters prefer the total aggregate hazardous air pollutant 
approach over the individual hazardous air pollutant approach because 
it better ensures that floor levels would be simultaneously achievable. 
However, we reject the total aggregate approach because it tends to 
result in floors that are likely to be artificially high, reflective of 
limited emissions data for all hazardous air pollutants at each 
facility. These floor levels, therefore, would not reflect performances 
of the best performing sources for particular hazardous air pollutants. 
We are assured of simultaneous achievability in our final methodology 
by: (1) Establishing the MACT floor feedrate control levels on an 
aggregate basis for metals and chlorine, as discussed above, rather 
than for each individual hazardous air pollutant; (2) using the 
particulate matter MACT pool to establish floor levels for particulate 
matter, semivolatile metals, and low volatile metals; and (3) ensuring 
that floor controls are not technically incompatible. In fact, our 
resulting floor emission levels are already achieved in practice by 9 
to 40 percent of sources in each of the three source categories, 
clearly indicating simultaneously achievable standards.68
---------------------------------------------------------------------------

    \68\ Our analysis shows that approximately nine percent of 
incinerators, 27 percent of cement kilns, and 40 percent of 
lightweight aggregate kilns currently operating can meet all of the 
floor levels simultaneously. See USEPA, ``Final Technical Support 
Document For HWC MACT Standards, Volume V: Emissions Estimates and 
Engineering Costs,'' July 1999.
---------------------------------------------------------------------------

C. What Other Floor Methodologies Were Considered?
    This is a brief overview of the major features of the MACT floor 
methodologies that we proposed in the April 1996 NPRM or discussed in 
the May 1997 NODA, accompanied by our rationale for not pursuing those 
methodologies in this final rule.
1. April 19, 1996 Proposal
    We proposed the same general approach to identify floor control and 
floor emission levels as used in today's final rule. The proposal 
contained an approach to identify the controls used by the best 
performing sources (i.e., the MACT pool) and then identify an emission 
level that those controls are achieving. To identify the floor emission 
level, we considered emissions from all sources using properly designed 
and operated controls (i.e., the expanded MACT pool) and established a 
preliminary floor level as the highest test condition average for those 
sources.
    There are three major differences between the proposed approach and 
today's final approach, however:
    a. Emissions Variability. At proposal, we added a statistically-
derived emissions variability factor to the highest test condition 
average in the expanded MACT pool. Today we conclude that emissions 
variability is considered inherently in the floor methodology. (See 
discussion in section D below for our rationale for not using a 
statistically-derived variability factor.)
    b. MACT Pool for Particulate Matter, Semivolatile Metals, and Low 
Volatile Metals. At proposal, we identified separate and different MACT 
pools (and associated MACT controls) for particulate matter, 
semivolatile metals, and low volatile metals, even though all three are 
controlled by a particulate matter control device. Commenters said this 
is inappropriate and we concur. Specifying the MACT floor particulate 
matter emission control device individually for these pollutants is 
likely to result in three different definitions of floor control. Thus, 
the same particulate matter control device would need to meet three 
different design specifications. As a practical matter, the more 
stringent specification would prevail. But, this highlights the 
impracticability of evaluating floor emission control for these 
standards individually rather than in the aggregate.
    As discussed in the May 1997 NODA, today's approach uses the same 
initial MACT pool to establish the floor levels for particulate matter, 
semivolatile metals, and low volatile metals. The initial MACT pool is 
comprised of those sources meeting the emission control component of 
MACT control. To establish the semivolatile metal and low volatile 
metal floor levels, the particulate matter MACT pool is then analyzed 
to consider MACT hazardous waste feedrate control first for 
semivolatile metals and then for low volatile metals, using the 
aggregate feedrate approach discussed above.
    c. Definition of MACT Control. At proposal, we defined MACT 
emissions control by specifying the design of the emissions control 
device. Commenters suggested that this was problematic because: (1) Our 
data base had limited data on design of the control device; (2) some of 
our available data were incorrect; and (3) the parameters the Agency 
was using to characterize MACT control did not adequately correlate 
with control efficiency. Given these concerns, our May 1997 NODA 
contained an emissions breakpoint approach to identify those sources 
that appeared to have anomalously higher emissions than other sources 
in the potential MACT pool. Our rationale was that given the 
anomalously high emissions, those sources were not, in fact, using MACT 
control.
    Commenters express serious concerns about the validity of the 
nonstatistical approach used to identify the breakpoint. After 
considering various statistical approaches to identify an emissions 
breakpoint, we conclude that the emissions breakpoint approach is 
problematic.69 For these reasons, we are

[[Page 52857]]

not defining MACT emissions control by design parameters or using an 
emissions breakpoint approach to identify MACT emissions or feedrate 
control. Rather, the MACT floor emission control equipment, where 
applicable, is defined generically (e.g., electrostatic precipitator, 
fabric filter), and the aggregate feedrate approach is used to define 
MACT floor feedrates. We believe the aggregate feedrate approach 
addresses the concerns that commenters raise on the proposed approach 
because it more clearly defines MACT control and relies less on 
engineering judgment.
---------------------------------------------------------------------------

    \69\ To improve the rigor of our breakpoint approach, we 
investigated a modified Rosner ``outlier'' test that: (1) Uses a 
single tailed test to consider only high ``outliers'' (i.e., test 
conditions that anomalously high emissions, not necessarily true 
outliers in the statistical sense); (2) presumes that any potential 
``outliers'' are at the 80th percentile value or higher; and (3) has 
a confidence level of 90 percent. We abandoned this statistical 
approach because: (1) Although modifications to the standard Rosner 
test were supportable, the modified test has not been peer-reviewed; 
(2) although the target confidence level was 90 percent, the true 
significance level of the test, as revised, is inappropriately low--
approximately 80 percent; and (3) the ``outlier'' test does not 
identify MACT-like test conditions because it only identifies 
anomalously high test conditions rather than the best performing 
test conditions.
---------------------------------------------------------------------------

2. May 1997 NODA
    We have incorporated into the final rule several of the procedures 
discussed in the May 1997 NODA. The NODA explained why it is 
inappropriate to add a statistically-derived emissions variability 
factor to the highest test condition average of the expanded MACT pool. 
Despite comments to the contrary, we conclude that emissions 
variability is inherently considered in the floor methodology. See 
discussion in section D below.
    In addition, the NODA discussed using the same initial MACT pool to 
establish the floor levels for particulate matter, semivolatile metals, 
and low volatile metals. We use this same approach in this final rule. 
Commenters generally concurred with that approach.
    As discussed above, we considered using an emissions breakpoint 
technique, but conclude that this approach is problematic and did not 
use the approach for this rule.
D. How Is Emissions Variability Accounted for in Development of 
Standards?
    The methodology we use to establish the final MACT emission 
standards intrinsically accounts for emissions variability without 
adding statistically-derived emissions variability factors. Many 
commenters strongly suggest that statistically-derived emissions 
variability factors must be added to the emission levels we identify 
from the data base as floor emission levels to ensure that the 
standards are routinely achievable.70 Other commenters 
suggest that our floor methodology inherently accounts for emissions 
variability. We discuss below the types of emissions variability and 
why we conclude that emissions variability is inherently accounted for 
by our methodology.
---------------------------------------------------------------------------

    \70\ One commenter recommends specific statistical approaches to 
calculate variability factors and provides examples of how the 
statistical methods should be applied to our emissions data base. 
See comment number CS4A-00041.
---------------------------------------------------------------------------

    We account for three types of emissions variability in establishing 
MACT standards: (1) Within test condition variability among test runs 
(a test condition is comprised of at least three runs that are 
averaged); (2) imprecision in the stack test method; and (3) source-to-
source emissions variability attributable to source-specific factors 
affecting the performance of the same MACT control device. (See, e.g. 
FMC Corp. v. Train, 539 F.2d 973, 985-86 (4th Cir. 1976), holding that 
variability in performance must be considered when ascertaining whether 
a technology-based standard is achievable.) The following sections 
discuss the way in which we account for these types of variability in 
the final rule.
1. How Is Within-Test Condition Emissions Variability Addressed?
    Inherent process variability will cause emissions to vary from run-
to-run within a test condition, even if the stack method is 100 percent 
precise and even though the source is attempting to maintain constant 
operating conditions. This is caused by many factors including: Minor 
changes in the feedrate of feedstreams; combustion perturbations (e.g., 
uncontrollable, minor fluctuations in combustion temperature or fan 
velocity); changes in the collection efficiency of the emission control 
device caused by fluctuations in key parameters (e.g., power input to 
an electrostatic precipitator); and changes in emissions of materials 
(e.g., sulfur dioxide) that may cause test method interferences.
    At proposal, we used a statistical approach to account for 
emissions variability. See 61 FR at 17366. The statistical approach 
identified an emissions variability factor, which was added to the log-
mean of the emission level being achieved based on the available 
``short-term'' compliance test data. We called this emission level the 
``design level.'' The variability factor was calculated to ensure that 
the design level could be achieved 99 percent of the time, assuming 
average within-test condition emissions variability for the source 
using MACT control.
    In the May 1997 NODA, we discussed alternative emission standards 
developed without using a statistically-derived variability factor. 
Adding such a variability factor was determined inappropriate because 
it sometimes resulted in nonsensical results. For example, the 
particulate matter MACT floor level for incinerators under one floor 
methodology would have been higher than the current RCRA standard 
allows, simply due to the impact of an added variability factor. In 
other cases, the floor levels would have been much higher than our 
experience would indicate are routinely being achieved using MACT 
control. We reasoned that these inappropriate and illogical results may 
flow from either the data base used to derive the variability factor 
(e.g., we did not have adequate information to screen out potentially 
outlier runs on a technical basis) or selecting an inappropriate floor-
setting test condition as the design level (e.g., we did not have 
adequate information on design, operation, and maintenance of emissions 
control equipment used by sources in the emissions data base to 
definitively specify MACT control).
    Consequently, we reasoned that adequately accounting for within 
test condition emissions variability is achieved where relatively large 
data sets are available to evaluate for identifying the floor level. 
Large sets of emissions data from MACT sources, which have emissions 
below the floor level, are likely to represent the range of emissions 
variability. For small data sets (e.g., dioxin/furan emissions for 
waste heat recovery boiler equipped incinerators; dioxin/furan 
emissions data for lightweight aggregate kilns), we acknowledged that 
the same logic would not apply. For these small data sets, the floor 
level was set at the highest run for the MACT source with the highest 
test condition average emissions. Many commenters suggest that our 
logic was flawed. Commenters say that, if we desire the floor level to 
be achievable 99 percent of the time (i.e., the basis for the 
statistically-derived variability factor at proposal), the emissions 
data base is far too small to identify the floor level as the highest 
test condition average for sources using MACT control.
    We conclude, however, that the final floor levels identified, using 
the procedures discussed above (i.e., without adding a statistically-
derived emissions variability factor), are levels that can be 
consistently achieved by well designed, operated, and maintained MACT 
sources. We

[[Page 52858]]

conclude this because our emissions data base is comprised of 
compliance test data generated when sources have an incentive to 
operate under worst case conditions (e.g., spiking metals and chlorine 
in the waste feed; detuning the emissions control equipment). Sources 
choose to operate under worst case conditions during compliance testing 
because the current RCRA regulations require that limits on key 
operating parameters not exceed the values occurring during the trial 
burn. Therefore, these sources conduct tests in a manner that will 
establish a wide envelope for their operating parameter limits in order 
to accommodate the expected variability (e.g., variability in types of 
wastes, combustion system parameters, and emission control parameters). 
See 56 FR at 7146 where EPA likewise noted that certain RCRA operating 
permit test conditions are to be ``representative of worst-case 
operating conditions'' to achieve needed operating flexibility. One 
company that operates several hazardous waste incinerators at three 
locations comments that, because of the current RCRA compliance regime, 
which is virtually identical to the compliance procedures of today's 
MACT rule, ``the result is that units must be tested at rates which are 
at least three standard deviations harsher than normal operations and 
normal variability in order to simulate most of the statistical 
likelihood of allowable emission rates.'' 71 The commenter 
also states that because of the consequences of exceeding an operating 
parameter limit under MACT, ``* * * clearly a source will test under 
the worst possible operating conditions in order to minimize future 
(exceedances of the limits).'' Finally, the commenter says that 
``Because of variability and the stiff consequences of exceeding these 
limits, operators do not in fact operate their units anywhere near the 
limits for sustained periods of time, but instead tend to operate 
several standard deviations below them, or at about 33 to 50% of the 
limits.'' 72
---------------------------------------------------------------------------

    \71\ See Comment No. CS4A-00029.A, dated August 16, 1996.
    \72\ To estimate the compliance cost of today's rule, we assumed 
that sources would design their systems to meet an emission level 
that is 70% of the standard, herein after called the ``design 
level.''
---------------------------------------------------------------------------

    We conclude from these comments, which are consistent with 
engineering principles and with many discussions with experts from the 
regulated community, that MACT sources with compliance test emissions 
at or below the selected floor level are achieving those levels 
routinely because these test conditions are worst-case and are defined 
by the source itself to ensure 100 percent compliance with the relevant 
standard.
    We acknowledge, however, that mercury is a special case because our 
mercury emission data may not be representative of worst-case 
conditions. As discussed in Section I.B.3 above, sources did not 
generally spike mercury emissions during RCRA compliance testing 
because they normally feed mercury at levels resulting in emissions 
well below current limits.73 Although our data base for 
mercury is comprised essentially of normal emissions, emissions 
variability is adequately accounted for in setting floor levels. First, 
mercury emissions variability is minimal because the source can readily 
control emissions by controlling the feedrate of mercury.74 
For cement and lightweight aggregate kilns, mercury is controlled 
solely by controlling feedrate. Given that there is no emission control 
device that could have perturbations affecting emission rates, 
emissions variability at a given level of mercury feedrate control is 
relatively minor. Any variability is attributable to variability in 
feedrate levels due to feedstream sampling and analysis imprecision, 
and stack method imprecision (see discussion below).
---------------------------------------------------------------------------

    \73\ Three of 23 incenerators used to define MACT floor (i.e., 
sources for which mercury feedrate data are available) are known to 
have spiked mercury. No cement kilns used to define MACT floor 
(e.g., excluding sources that have stopped burning hazardous waste) 
are known to have spiked mercury. Only one of ten lightweight 
aggregate kilns used to define MACT floor is known to have spiked 
mercury.
    \74\ Although incenerators are generally equipped with wet 
scrubbers that can have a mercury removal efficiency of 15 to 60 
percent, feedrate control is nonetheless the primary means of 
mercury emissions control because of the relatively low removal 
efficiency provided by wet scrubbers.
---------------------------------------------------------------------------

    Second, our emissions data indicate that the mercury floor levels 
are being achieved by a wide margin, which is a strong indication that 
a variability factor is not needed. Only one of the 15 incinerators 
using MACT floor control exceeds the design level for the floor 
emission level.75 In addition, only seven of 45 incinerators 
for which we have mercury emissions data exceed the design level, and 
two of those eight are know to have spiked mercury in the hazardous 
waste feed during compliance testing. Only six of the 45 incinerators 
exceed the floor emission level.
---------------------------------------------------------------------------

    \75\ Commenters note that the mercury levels fed during RCRA 
compliance testing may not represent the normal range of feedrates, 
and thus the compliance test emission levels may not be 
representative of emission levels achieved in practice. Given that 
only one of 15 incinerators using floor control exceeds the design 
level, it appears that the floor emission level is, in fact, being 
achieved in practice. Some of these 15 sources were likely feeding 
mercury at the high end of their normal range, even though others 
may have been feeding mercury at normal or below normal levels. This 
is also the situation of cement kilns where only two of 2 kilns 
using floor control exceed the design level, and for lightweight 
aggregate kilns where only one of nine kilns using floor exceeds the 
design level.
---------------------------------------------------------------------------

    The situation is similar for cement kilns and lightweight aggregate 
kilns. Only two of 22 cement kilns using floor control exceed the 
design level, only five of the 33 kilns in the source category exceed 
the design level, and only one of the 33 kilns exceeds the floor 
emission level. Only one of nine lightweight aggregate kilns using 
floor control exceeds the design level, and only two of the 10 kilns in 
the source category exceed the design level (and one of those kilns is 
known to have spiked mercury in the hazardous waste feed during 
compliance testing). Only one of the 10 kilns exceeds the floor 
emission level, and that kiln spiked mercury.
    We conclude from this analysis that the mercury floor emission 
levels in this rule are readily achieved in practice even though our 
mercury emissions data were not spiked (i.e., they may not represent 
worst-case emissions), and therefore a separate variability factor is 
not needed.
2. How Is Waste Imprecision in the Stack Test Method Addressed?
    Method precision is a measure of how closely emissions data are 
grouped together when measuring the same level of stack emissions 
(e.g., using a paired or quad test train). Method imprecision is 
largely a function of the ability of the sampling crew and analytical 
laboratory to routinely follow best practices. Precision can be 
affected by: (1) Measurement of ancillary parameters including gas flow 
rate, pressure, and temperature; (2) recovery of materials from the 
sampling train; and (3) cleaning, concentrating, and quantitating the 
analyte.
    Several commenters state that we must add a factor to the selected 
floor level to account for method imprecision in addition to a factor 
to account for within-test condition emissions variability. We 
investigated the imprecision for the stack methods used to document 
compliance with today's rule and determined that method imprecision may 
be significant for some hazardous air pollutant/method 
combinations.76 Our results indicate, however, that method 
precision is much better than commenters claim, and that as additional 
data sets become available,

[[Page 52859]]

the statistically-derived precision bars for certain pollutants are 
reasonably expected to be reduced significantly. This is mainly because 
data should become available over a wider range of emission levels thus 
reducing the uncertainty that currently results in large precision bar 
projections for some hazardous air pollutants at emission levels that 
are not close to the currently available paired and quad-train 
emissions data.
---------------------------------------------------------------------------

    \76\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    We conclude that method imprecision, in selecting the floor levels 
for hazardous waste combustors, is adequately addressed for the same 
reasons that we accounted for within-test condition emissions 
variability. Method precision is simply a factor that contributes to 
within-test condition variability. As discussed above, sources consider 
emissions variability when defining their compliance test operating 
conditions to balance emissions standards compliance demonstrations 
with the need to obtain a wide operating envelope of operating 
parameter limits.
3. How Is Source-to-Source Emissions Variability Addressed?
    If the same MACT control device (i.e., same design, operating, and 
maintenance features) were used at several sources within a source 
category, emissions of hazardous air pollutants from the sources could 
vary. This is because factors that affect the performance of the 
control device could vary from source to source. Even though a device 
has the same nominal design, operating, and maintenance features, those 
features could never be duplicated exactly. Thus, emissions could vary 
from source to source.
    We agree that this type of emissions variability must be accounted 
for in the standards to ensure the standards are achieved in practice. 
Source-to-source emissions variability is addressed by identifying the 
floor emission level as the highest test condition average for sources 
in the expanded MACT pool, as discussed above.77
---------------------------------------------------------------------------

    \77\ Because of the need to account for this type of 
variability, we disagree with those commenters recommending that: 
(1) The floor emission level be identified as the average emission 
level achieved by the 12 percent of source with the lowest 
emissions; and (2) it is inappropriate to base the floor emission 
level on sources using floor control but that are not within the 12 
percent of sources with the lowest emissions (i.e., the expanded 
MACT pool should not be used to identify floor emission levels). The 
floor emission level must be achieved in practice by sources using 
the appropriately designed and operated floor control. Thus, 
emission levels being achieved by all sources using the 
appropriately designed and operated floor control (i.e., including 
sources using floor control but having emission levels greater than 
the average of the emissions achieved by the 12 percent of sources 
with the lowest emissions) must be considered when identifying the 
floor emission level.
---------------------------------------------------------------------------

    The test condition average emissions for sources in the expanded 
MACT pool for most standards often vary over several orders of 
magnitude. That variability is attributable partially to the type of 
source-to-source emissions variability addressed here as well as the 
inclusion of sources with varying levels of MACT control in the pool. 
Sources are included in the expanded MACT pool if they have controls 
equivalent to or better than MACT floor controls. We are unable to 
identify true source-to-source emissions variability for sources that 
actually have the same MACT controls because we are unable to specify 
in sufficient detail the design, operating, and maintenance 
characteristics of MACT control. Such information is not readily 
available. Therefore, we define MACT control only in general terms. 
This problem (and others) are addressed in today's rule by selecting 
the MACT floor level based on the highest test condition average in the 
expanded MACT pool, which accounts for source-to-source variability.
    We also conclude that the characteristics of the emissions data 
base coupled with the methodology used to identify the floor emission 
level adequately accounts for emissions variability so that the floor 
level is routinely achieved in practice by sources using floor control. 
As further evidence, we note that a large fraction--50 to 100 percent--
of sources in the data base currently meet the floor levels regardless 
of whether they currently use floor control.78
---------------------------------------------------------------------------

    \78\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

VI. What Are the Standards for Existing and New Incinerators?

A. To Which Incinerators Do Today's Standards Apply?
    The standards promulgated today apply to each existing, 
reconstructed, and newly constructed incinerator (as defined in 40 CFR 
260.10) burning hazardous waste. These standards apply to all major 
source and area source incinerator units and to all units whether they 
are transportable or fixed sources. These standards also apply to 
incinerators now exempt from RCRA stack emission standards under 
Secs. 264.340(b) and (c).\79\ Additionally, these standards apply to 
thermal desorbers that meet the definition of a RCRA incinerator, and 
therefore, are not regulated under subpart X of part 264.
---------------------------------------------------------------------------

    \79\ Sections 264.340(b) and (c) exempt from stack emission 
standards incinerators (a) burning solely ignitable, corrosive or 
reactive wastes under certain conditions, and (b) if the waste 
contains no or insignificant levels of hazardous constituents.
---------------------------------------------------------------------------

B. What Subcategorization Options Did We Evaluate?
    We considered whether it would be appropriate to subcategorize 
incinerators based on several factors discussed below and conclude that 
subcategorization is not necessary. However, for waste heat recovery 
boiler-equipped incinerators, we establish a separate emission standard 
solely for dioxin/furan. We explained our rationale for separate 
dioxin/furan standards for waste heat recovery boilers in the May 1997 
NODA (62 FR 24220). We said that waste heat recovery boilers emit 
significantly higher dioxin/furan emissions than other incinerators, 
probably because the heat recovery boiler precludes rapid temperature 
quench of the combustion gases to below 400 deg.F, therefore warranting 
separate standards for dioxin/furan only (i.e., the waste heat boiler 
does not affect achievability of the other emission standards).
    We considered several options for subcategorizing the hazardous 
waste incinerator source category based on: (1) Size of the unit (e.g., 
small and large incinerators); (2) method of use of the hazardous waste 
incinerator (e.g., commercial hazardous waste incinerator, captive (on-
site) unit); (3) facility design (e.g., rotary kiln, liquid injection, 
fluidized bed, waste heat boiler), and (4) type of waste fed (e.g., 
hazardous waste mixed with radioactive waste, munitions, liquid, solid 
or aqueous wastes). Subcategorization would be appropriate if one or 
more of these factors affected achievability of emission standards that 
were established without subcategorization. In the May 1997 NODA (62 FR 
24219), we stated that subdividing the hazardous waste incinerator 
source category by size or method of use (such as commercial or on-
site) would be inappropriate because it would not result in standards 
that are more achievable. Many of the standards would be the same for 
the subcategories while the remainder would be more stringent. That 
conclusion is not altered by any of the changes in today's final rule. 
Therefore, subcategorization would add complexity without any tangible 
achievability benefits.
    In the same notice, we also requested comment on subcategorization 
and/or a deferral of standards for mixed waste incinerators based on a 
comment from the Department of Energy that this type of incinerator has 
several unique features that warrant subcategorization.

[[Page 52860]]

There are three Department of Energy mixed waste incinerators. Each 
mixed waste incinerator has a different type of operation and different 
air pollution control devices, and two of the sources have high dioxin/
furan and mercury emissions (several times the dioxin/furan standards 
adopted in today's rule). We received several comments on the mixed 
waste incinerator issue. These commenters contend that, because of the 
radioactive component of the wastes, mixed waste incinerators pose 
greater than average risk, and regulating these facilities should not 
be deferred. These commenters also note that the MACT controls are not 
incompatible with mixed waste incinerators and thus these incinerators 
can readily achieve the emission standards. We agree that MACT controls 
are compatible with mixed waste incinerators, with one exception 
discussed below, and do not establish a mixed waste incinerator 
subcategory.
    The standards promulgated today are generally achievable by all 
types and sizes of incinerators when using MACT controls. We recognize, 
however, that each of the possible subcategories considered has some 
unique features. At the same time, upon consideration of each 
individual issue, we conclude that unique features of a particular 
hazardous waste incinerator can be better dealt with on an individual 
basis (through the permit process or through petitions) instead of 
through extensive subcategorization. As an example, we agree with the 
Department of Energy's contentions that feedstream testing for metals 
is problematic for mixed waste incinerators due to radioactivity of the 
waste and because risk from metal emissions is minimal in mixed waste 
incinerators that use HEPA filters to prevent radioactive emissions. 
Section 63.1209(g)(1) of today's rule provides a mechanism for 
petitioning the Administrator for use of an alternative monitoring 
method.80 This petition process appears to be an appropriate 
vehicle for addressing the concerns expressed by the Department of 
Energy about feedstream testing for metals and use of HEPA filters at 
its mixed waste incinerators.
---------------------------------------------------------------------------

    \80\ The petition for an alternative monitoring method should be 
included in the comprehensive performances test plan submitted for 
review and approval.
---------------------------------------------------------------------------

    In summary, our decision not to subcategorize hazardous waste 
incinerators is based on four reasons:
    (1) Size differences among hazardous waste incinerators do not 
necessarily reflect process, equipment or emissions differences among 
the incinerators. Many small size hazardous waste incinerators have 
emissions lower than those promulgated today even though they are not 
regulated to those low levels.
    (2) Types and concentrations of uncontrolled hazardous air 
pollutants are similar for all suggested subcategories of hazardous 
waste incinerators.
    (3) The same type of control devices, such as electrostatic 
precipitators, fabric filters, and scrubbers, are used by all hazardous 
waste incinerators to control emissions of particular hazardous air 
pollutants.
    (4) The standards are achievable by all types and sizes of well 
designed and operated incinerators using MACT controls.
C. What Are the Standards for New and Existing Incinerators?
1. What Are the Standards for Incinerators?
    We discuss in this section the basis for the emissions standards 
for incinerators. The emissions standards are summarized below:

                                   Standards for Existing and New Incinerators
----------------------------------------------------------------------------------------------------------------
                                                               Emissions standard \1\
    Hazardous air pollutant or    ------------------------------------------------------------------------------
hazardous air pollutant surrogate    Existing sources                          New sources
----------------------------------------------------------------------------------------------------------------
Dioxin /Furan....................  0.20 ng TEQ \2\/     0.20 ng TEQ/dscm.
                                    dscm; or 0.40 ng
                                    TEQ/dscm and
                                    temperature at
                                    inlet to the
                                    initial
                                    particulate matter
                                    control device  400 deg.F.
Mercury..........................  130 g/dscm  45 g/dscm.
Particulate Matter...............  34mg/dscm (0.015gr/  34mg/dscm (0.015gr/dscf).
                                    dscf).
Semivolatile Metals..............  240 g/dscm  24 g/dscm.
Low Volatile Metals..............  97 g/dscm.  97 g/dscm.
Hydrochloric Acid/Chlorine Gas...  77 ppmv............  21 ppmv.
Hydrocarbons 3, 4................  10 ppmv (or 100      10 ppmv (or 100 ppmv carbon monoxide).
                                    ppmv carbon
                                    monoxide).
Destruction and Removal            99.99% for each      Same as for existing incinerators.
 Efficiency.                        specific principal
                                    organic hazardous
                                    constituent,
                                    except 99.9999%
                                    for specified
                                    dioxin-listed
                                    wastes.
----------------------------------------------------------------------------------------------------------------
\1\ All emission levels are corrected to 7 percent oxygen.
\2\ Toxicity equivalent quotient, the international method of relating the toxicity of various dioxin/furan
  congeners to the toxicity of 2,3,7,8-TCDD.
\3\ Hourly rolling average. Hydrocarbons reported as propane.
\4\ Incinerators that elect to continuously comply with the carbon monoxide standard must demonstrate compliance
  with the hydrocarbon standard of 10ppmv during the comprehensive performance test.

2. What Are the Standards for Dioxins and Furans?
    We establish a dioxin/furan standard for existing incinerators of 
either 0.20 ng TEQ/dscm, or a combination of dioxin/furan emissions up 
to 0.40 ng TEQ/dscm and temperature at the inlet to the initial dry 
particulate matter control device not to exceed 400 deg.F.81 
Expressing the standard as a temperature limit as well as a dioxin/
furan concentration limit provides better control of dioxin/furan, 
because sources operating at temperatures below 400 deg.F generally 
have lower emissions and is consistent with the current practice of 
many sources. Further, without the lower alternative TEQ limit of 0.20 
ng/dscm, sources that may be operating dry particulate matter control 
devices at temperatures higher than 400 deg.F while achieving dioxin/
furan emissions below 0.20 ng TEQ/dscm would nonetheless be required to 
incur costs to lower gas temperatures. This would not be appropriate 
because lowering gas temperatures in this case would likely

[[Page 52861]]

achieve limited reductions in dioxin/furan emissions (i.e., because 
emissions are already below 0.20 ng TEQ).
---------------------------------------------------------------------------

    \81\ Incinerators that use wet scrubbers as the initial 
particulate matter control device are presumed to meet the 400 deg.F 
temperature requirement. Consequently, as a practical matter, the 
standard for such incinerators is simply 0.4 ng TEQ/dscm.
---------------------------------------------------------------------------

    For new incinerators, the dioxin/furan standard is 0.20 ng TEQ/
dscm. We discuss below the rationale for these standards.
    a. What is the MACT Floor for Existing Sources? We establish the 
same MACT floor control, as was evaluated in the May 1997 NODA, based 
on the revised data base and the refinements to the analytical 
approaches. This floor control is based on quenching of combustion 
gases to 400 deg.F or below at the dry particulate matter control 
device.82 We selected a temperature of 400 deg.F because 
that temperature is below the temperature range for optimum surface-
catalyzed dioxin/furan formation reactions--450 deg.F to 650 deg.F--and 
most sources operate their particulate matter control device below that 
temperature. In addition, temperature is an important control parameter 
because dioxin/furan emissions increase exponentially as combustion gas 
temperatures at the dry particulate matter control device increase 
above 400 deg.F.
---------------------------------------------------------------------------

    \82\ The temperature limit applies at the inlet to a dry 
particulate matter control device that suspends particulate matter 
in the combustion gas stream (e.g., electrostatic precipitator, 
fabric filter) such that surface-catalyzed formation of dioxin/furan 
is enhanced. The temperature limit does not apply to a cyclone 
control device, for example.
---------------------------------------------------------------------------

    We identify a MACT floor level of 0.40 ng TEQ/dscm for incinerators 
other than those equipped with waste heat recovery boilers. As 
discussed in the May 1997 NODA, the floor level of 0.40 ng TEQ/dscm is 
based on the highest nonoutlier test condition for sources equipped 
with dry particulate matter control devices operated at temperatures of 
400 deg.F or below or wet particulate matter control devices. We 
screened out four test conditions from three facilities because they 
have anomalously high dioxin/furan emissions and are not representative 
of MACT control practices.83 Three of these test conditions 
are from sources that had other test conditions with emission averages 
well below 0.40 ng TEQ/dscm, indicating that the same facilities can 
achieve lower emission levels in different operating modes.
---------------------------------------------------------------------------

    \83\ USEPA, ``Technical Support Document for HWC MACT Standards, 
Volume III: Selection of MACT Standards and Technologies,'' July 
1999, Section 3.1.1.
---------------------------------------------------------------------------

    We identify a MACT floor level for waste heat boiler-equipped 
hazardous waste incinerators of 12 ng TEQ/dscm based on the highest 
emitting individual run for sources equipped with dry particulate 
matter control devices operated at temperatures of 400 deg.F or below 
or wet particulate matter control devices. We use the highest run to 
set the floor level rather than the average of the runs for the test 
condition to address emissions variability concerns given that we have 
a very small data set for waste heat boilers. All waste heat boiler-
equipped hazardous waste incinerators meet this floor level, except for 
a new test conducted after the publication of the May 1997 NODA at high 
temperature conditions that resulted in dioxin/furan emission levels of 
47 ng TEQ/dscm. This source is not using MACT control, however, because 
the temperature at the particulate matter control device exceeded 
400 deg.F. Thus, we do not consider emissions from this source in 
identifying the floor level.
    We received numerous and diverse comments on the April 1996 
proposal and the May 1997 NODA. While some commenters consider the 
dioxin/furan standards too high, a large number comment that the 
standards are too stringent. Many comment that the methodology used for 
calculating the dioxin/furan MACT floor level is inappropriate and that 
the cost-effectiveness of the standards is not reasonable. In 
particular, some commenters suggest separating ``fast quench'' and 
``slow quench'' units. We have fully addressed this latter concern 
because we now establish separate dioxin/furan standards for waste heat 
boilers given that they are a fundamentally different type of process 
and that they have higher dioxin/furan emissions because of the slow 
quench across the boiler. We address the other comments elsewhere in 
the preamble and in the comment response document.
    Approximately 65% of all test conditions at all incinerator sources 
are achieving the 0.40 ng TEQ/dscm level, and over 50% of all test 
conditions achieve the 0.20 ng TEQ/dscm level. We estimate that 
approximately 60 percent of incinerators currently meet the TEQ limit 
as well as the temperature limit. Under the statute, compliance costs 
are not to be considered in MACT floor determinations. For purposes of 
compliance with Executive Order 12866 and the Regulatory Flexibility 
Act, we calculated the annualized cost for hazardous waste incinerators 
to achieve the dioxin/furan MACT floor levels. Assuming that no 
hazardous waste incinerator exits the market due to MACT standards, the 
annual cost is estimated to be $3 million, and the standards will 
reduce dioxin/furan emissions nationally by 3.4 g TEQ per year from the 
baseline emissions level of 24.8 g TEQ per year.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We investigated the use of activated carbon injection, along 
with limiting temperatures at the inlet to the initial dry particulate 
matter control device to 400 deg.F,84 to achieve two 
alternative beyond-the-floor emission levels: (1) 0.40 ng TEQ/dscm for 
waste heat boiler-equipped incinerators (i.e., slow quench) to reduce 
their emissions to the floor level for other incinerators; and (2) 0.20 
ng TEQ/dscm for all incinerators. Activated carbon injection technology 
is feasible and proven to reduce dioxin/furan emissions by 99 percent 
or greater.85 It is currently used by one waste heat boiler-
equipped hazardous waste incinerator (Waste Technologies Industries in 
East Liverpool, Ohio) and many municipal waste combustors.86 
The removal efficiency of an activated carbon injection system is 
affected by several factors including carbon injection rate and 
adsorption quality of the carbon. Thus, activated carbon injection 
systems can be used by waste heat boiler-equipped incinerators to 
achieve alternative beyond-the-floor emissions of either 0.40 ng TEQ/
dscm or 0.20 ng TEQ/dscm.
---------------------------------------------------------------------------

    \84\ Limiting the temperature at the dry particulate matter 
control device reduces surface-catalyzed formation of dioxin/furan 
and enhances the adsorption of dioxin/furan on the activated carbon.
    \85\ USEPA, ``Technical Support Document for HWC MACT Standards, 
Volume III: Selection of MACT Standards and Technologies,'' July 
1999.
    \86\ We have established in a separate rulemaking that activated 
carbon injection is MACT floor control for municipal waste 
combustors.
---------------------------------------------------------------------------

    We conclude that a beyond-the-floor emission level of 0.40 ng TEQ/
dscm for waste heat boiler-equipped incinerators is cost-effective but 
a 0.20 ng TEQ/dscm emission level for all incinerators is not cost-
effective. We estimate that 23 waste heat boiler-equipped incinerators 
will need to install activated carbon injection systems at an 
annualized cost of approximately $6.6 million. This will result in a 
sizable reduction of 17.9 g TEQ dioxin/furan emissions per year and 
will provide an 84 percent reduction in emissions from the floor 
emission level (21.4 g TEQ per year) for all hazardous waste 
incinerators. This represents a cost-effectiveness of $370,000 per gram 
TEQ removed.
    When we evaluated the alternative beyond-the-floor emission level 
of 0.20 ng TEQ/dscm for all incinerators, we determined that 80 
hazardous waste incinerators would incur costs to reduce dioxin/furan 
emissions by 19.5 g TEQ from the floor level (21.4 g TEQ) at an 
annualized cost of $16.1 million. The cost-effectiveness would be 
$827,000 per gram of TEQ removed. In addition,

[[Page 52862]]

we determined that the vast majority of these emissions reductions 
would be provided by waste heat boiler-equipped incinerators, and would 
be provided by the beyond-the-floor emission level of 0.40 ng TEQ/dscm 
discussed above. The incremental annualized cost of the 0.20 ng TEQ/
dscm option for incinerators other than waste heat boiler-equipped 
incinerators would be $9.5 million, and would result in an incremental 
reduction of only 1.6 g TEQ per year. This represents a high cost for a 
very small additional emission reduction from the floor, or a cost-
effectiveness of $6.0 million per additional gram of TEQ dioxin/furan 
removed. Accordingly, we conclude that the 0.20 ng TEQ/dscm beyond-the-
floor option is not cost-effective.
    We note that dioxin/furan are some of the most toxic compounds 
known due to their bioaccumulative potential and wide range of adverse 
health effects, including carcinogenesis, at exceedingly low doses. We 
consider beyond-the-floor reduction of dioxin/furan emissions a prime 
environmental and human health consideration. As discussed above, our 
data base indicates that a small subset of incinerators--those equipped 
with waste heat recovery boilers--can emit high levels of dioxin/furan, 
up to 12 ng TEQ/dscm, even when operating the dry particulate matter 
control device at 400 deg.F. We are concerned that such high 
dioxin/furan emission levels are not protective of human health and the 
environment, as mandated by RCRA. If dioxin/furan emissions from waste 
heat boiler-equipped incinerators are not reduced by a beyond-the-floor 
emission standard, omnibus RCRA permit conditions would likely be 
needed in many cases. This would defeat our objective of having only 
one permitting framework for stack air emissions at hazardous waste 
incinerators (except in unusual cases). Thus, the beyond-the-floor 
standard promulgated today for waste heat boiler-equipped incinerators 
is not only cost-effective, but also an efficient approach to meed the 
Agency's RCRA mandate.
    Some commenters suggest that the standard for waste heat boiler-
equipped hazardous waste incinerators, which is based on activated 
carbon injection, be set at levels achieved by activated carbon 
injection at the Waste Technologies Industries facility--an average of 
0.07 ng TEQ/dscm. We determined that this would not be appropriate 
because of concerns that such a low emission level may not be routinely 
achievable. An emission level of 0.07 ng TEQ/dscm represents a 99.4 
percent reduction in emissions from the floor level of 12 ng TEQ/dscm. 
Although activated carbon injection can achieve dioxin/furan emissions 
reductions of 99 percent and higher, we are concerned that removal 
efficiency may decrease at low dioxin/furan emission levels. We noted 
our uncertainty about how much activated carbon injection control 
efficiency may be reduced at low dioxin/furan concentrations in the May 
1997 NODA (62 FR at 24220). Several commenters agree with our concern, 
including Waste Technologies Industries.87 No commenters 
provide data or information to the contrary. Because we have data from 
only one hazardous waste incinerator documenting that an emission level 
of 0.07 ng TEQ can be achieved, we are concerned that an emission level 
that low may not be routinely achievable by all sources.
---------------------------------------------------------------------------

    \87\ Waste Technologies Industries suggested, however, that 
after experience with activated carbon injection systems has been 
attained by several hazardous waste incinerators, the Agency could 
then determine whether an emission level of 0.07 ng TEQ/dscm is 
routinely achievable. See comment number 064 in Docket F-97-CS4A-
FFFFF.
---------------------------------------------------------------------------

    c. What Is the MACT Floor for New Sources? For new sources, the CAA 
requires that the MACT floor be the level of control used by the best 
controlled single source. As discussed above, one source, the Waste 
Technologies Industries (WTI) incinerator in Liverpool, Ohio, uses 
activated carbon injection. Therefore, we identify activated carbon 
injection as MACT floor control for new sources. To establish the MACT 
floor emission level that is being achieved in practice for sources 
using activated carbon injection, data are available from only WTI. WTI 
is achieving an emission level of 0.07 ng TEQ/dscm. As discussed above, 
we are concerned that emission level may not be routinely achievable 
because the removal efficiency of activated carbon injection may be 
reduced at such low emission levels. An emission level of 0.20 ng TEQ/
dscm is routinely achievable, however. We note that activated carbon 
injection is MACT floor control for dioxin/furan at new large municipal 
waste combustors. We established a standard of 13 ng/dscm total mass 
``equal to about 0.1 to 0.3 ng/dscm TEQ'' for these sources (60 FR 
65396 (December 19, 1995)), equivalent to approximately 0.20 ng TEQ/
dscm. We conclude, therefore, that a floor level of 0.20 ng TEQ/dscm is 
achievable for new sources using activated carbon injection and 
accordingly set this as the standard.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? As 
discussed in the May 1997 NODA, a beyond-the-floor standard below 0.20 
ng TEQ/dscm would not be appropriate. Although installation of carbon 
beds would enable new hazardous waste incinerators to achieve lower 
dioxin/furan levels, we do not consider the technology to be cost-
effective. The reduction in dioxin/furan emissions would be very small, 
while the costs of carbon beds would be prohibitively high. In 
addition, due to the very small dioxin/furan reduction, the benefit in 
terms of cancer risks reduced also will be very small. Therefore, we 
conclude that a beyond-the-floor standard for dioxin/furan is not 
appropriate.
3. What Are the Standards for Mercury?
    We establish a mercury standard for existing and new incinerators 
of 130 and 45 g/dscm respectively. We discuss below the 
rationale for these standards.
    a. What Is the MACT Floor for Existing Sources? We are establishing 
the same MACT floor level as proposed, 130 g/dscm although, as 
discussed below, the methodology underlying this standard has changed 
from proposal. At proposal, the floor standard was based on the 
performance of either: (1) Feedrate control of mercury at a maximum 
theoretical emission concentration not exceeding 19 g/dscm; or 
(2) wet scrubbing in combination with feedrate control of mercury at a 
level equivalent to a maximum theoretical emission concentration not 
exceeding 51 g/dscm. In the May 1997 NODA, we reevaluated the 
revised data base and defined MACT control as based on performance of 
wet scrubbing in combination with feedrate control of mercury at a 
level equivalent to a maximum theoretical emission concentration of 50 
g/dscm and discussed a floor level of 40 g/dscm.
    Several commenters object to our revised methodology and are 
concerned that we use low mercury feedrates to define floor control. 
These commenters state that standards should not be based on sources 
feeding very small amounts of a particular metal, but rather on their 
ability to minimize the emissions by removing the hazardous air 
pollutant. As discussed previously, we maintain that hazardous waste 
feedrate is an appropriate MACT control technique. We agree with 
commenters' concerns, however, that previous methodologies to define 
floor feedrate control may have identified sources feeding anomalously 
low levels of a metal (or chlorine). To address this concern, we have 
revised the floor determination methodology for mercury, semivolatile 
metals, low volatile metals and total chlorine. A

[[Page 52863]]

detailed description of this methodology--the aggregate feedrate 
approach--is presented in Part Four, Section V of this preamble. 
Adopting this aggregate feedrate approach, we identify a mercury 
feedrate level that is approximately five times higher than the May 
1997 NODA level and higher than approximately 70% of the test 
conditions in our data base.
    Wet scrubbers also provide control of mercury (particularly mercury 
chlorides). Given that virtually all incinerators are equipped with wet 
scrubbers (for control of particulate matter or acid gases), we 
continue to define floor control as both hazardous waste feedrate 
control of mercury and wet scrubbing. The MACT floor based on the use 
of wet scrubbing and feedrate control of mercury is 130 g/
dscm.\88\
---------------------------------------------------------------------------

    \88\ This is coincidentally the same floor level as proposed, 
notwithstanding the use of a different methodology.
---------------------------------------------------------------------------

    The floor level is being achieved by 80% of the test conditions in 
our data base of 30 hazardous waste incinerators. As already discussed 
above, consideration of costs to achieve MACT floor standards play no 
part in our MACT floor determinations, but we nevertheless estimate 
costs to the hazardous waste incinerator universe for administrative 
purposes. We estimate that 35 hazardous waste incinerators, assuming no 
market exit by any facility, will need to adopt measures to reduce 
mercury emissions at their facilities by 3.46 Mg from the current 
baseline of 4.4 Mg at an estimated annualized cost $12.2 million, 
yielding a cost-effectiveness of $3.6 million per Mg of mercury 
reduced.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? As required by statute, we evaluated more stringent beyond-
the-floor controls for further reduction of mercury emissions from the 
floor level. Activated carbon injection systems can achieve mercury 
emission reductions of over 85 percent and we proposed them as beyond-
the-floor control in the April 1996 NPRM. In the May 1997 NODA, we 
reevaluated the use of activated carbon injection 89 as 
beyond-the-floor control, but cited significant cost-effectiveness 
concerns. We reiterate these concerns here. Our technical support 
document 90 provides details of annualized costs and 
reductions that can be achieved.
---------------------------------------------------------------------------

    \89\ Flue gas temperatures would be limited to 400 deg.F at the 
point of carbon injection to enhance mercury removal.
    \90\ USEPA, ``Technical Support Document for HWC MACT Standards, 
Volume V: Emission Estimates and Engineering Costs,'' July 1999.
---------------------------------------------------------------------------

    In addition, we considered a beyond-the-floor level of 50 
g/dscm based on limiting the feedrate of mercury in the 
hazardous waste (i.e., additional feedrate control beyond floor 
control), and conducted an evaluation of the cost of achieving this 
reduction to determine if this beyond-the-floor level would be 
appropriate. The national incremental annualized compliance cost to 
meet this beyond-the-floor level, rather than comply with the floor 
controls, would be approximately $4.2 million for the entire hazardous 
waste incinerator industry and would provide an incremental reduction 
in mercury emissions nationally beyond the MACT floor controls of 0.7 
Mg/yr, yielding a cost-effectiveness of $10 million per additional Mg 
of mercury reduced. Thus, potential benefits in relation to costs are 
disproportionately low, and we conclude that beyond-the-floor mercury 
controls for hazardous waste incinerators are not warranted. Therefore, 
we are not adopting a mercury beyond-the-floor standard.
    Many commenters object to our beyond-the-floor standards as 
proposed, citing high costs for achieving relatively small mercury 
emission reductions, and compare the cost-effectiveness numbers with 
regulations of other sources (electric utilities, municipal and medical 
waste incinerators). Although comparison between rules for different 
sources is not directly relevant (see, e.g., Portland Cement 
Association v. Ruckelshaus 486 F.2d 375, 389 (D.C. Cir. 1973)), we 
nevertheless agree that the cost of a mercury beyond-the-floor standard 
in relation to benefits is substantial. Some commenters, as well as the 
peer review panel, state that beyond-the-floor levels are not supported 
by a need based on risk. Although the issue of residual risk can be 
deferred under the CAA, an immediate question must be addressed if RCRA 
regulation of air emissions is to be deferred. Our analysis \91\ 
indicates that mercury emissions at the floor level do not pose a 
serious threat to the human health and environment and that these 
standards are adequately protective to satisfy RCRA requirements as a 
matter of national policy, subject, of course, to the possibility of 
omnibus permit conditions for individual facilities in appropriate 
cases.
---------------------------------------------------------------------------

    \91\ USEPA, ``Risk Assessment Support to the Development of 
Technical Standards for Emissions from Combustion Units Burning 
Hazardous Wastes: Background Information Document,'' July 1999.
---------------------------------------------------------------------------

    Some commenters state that the technical performance of activated 
carbon injection for mercury control is not adequately proven. 
Activated carbon injection performance has been adequately demonstrated 
at several hazardous waste incinerators, municipal waste combustors, 
and other devices.\92\ Our peer review panel also states that activated 
carbon injection can achieve 85% reduction of mercury emissions.\93\ 
Some commenters also state that we underestimate the cost and 
complexities of retrofitting incinerators to install activated carbon 
injection systems (e.g., air reheaters would be required in many 
cases). We reevaluated the modifications needed for retrofits of 
activated carbon injection systems and have revised the costs of 
installation.
---------------------------------------------------------------------------

    \92\ USEPA, ``Technical Support Document for HWC MACT Standards, 
Volume III: Selection of Proposed MACT Standards and Technologies,'' 
July 1999.
    \93\ Memo from Mr. Shiva Garg, EPA to Docket No. F-96-RCSP-FFFFF 
entitled ``Peer Review Panel Report in support of proposed rule for 
revised standards for hazardous waste combustors'', dated August 5, 
1996.
---------------------------------------------------------------------------

    c. What Is the MACT Floor for New Sources? Floor control must be 
based on the level of control used by the best controlled single 
source. The best controlled source in our data base uses wet scrubbing 
and hazardous waste feedrate control of mercury at a feedrate 
corresponding to a maximum theoretical emission concentration of 0.072 
g/dscm. We conclude that this feedrate is atypically low, 
however, given that the next lowest mercury feedrates in our data base 
are 63, 79, 110, and 130 g/dscm, expressed as maximum 
theoretical emission concentrations. Accordingly, we select the mercury 
feedrate for the second best controlled source under the aggregate 
feedrate approach to represent the floor control mercury feedrate for 
new sources. That feedrate is 110 g/dscm \94\ expressed as a 
maximum theoretical emission concentration, and corresponds to an 
emission level of 45 g/dscm after considering the expanded 
MACT pool (i.e., the highest emission level from all sources using 
floor control). Therefore, we establish a MACT floor level for mercury 
for new sources of 45 g/dscm.\95\ We note that, at proposal 
and in

[[Page 52864]]

the May 1997 NODA, mercury standards of 50 and 40 g/dscm 
respectively were proposed for new sources. Today's final rule is in 
the same range as those proposed emission levels.
---------------------------------------------------------------------------

    \94\ The test conditions with mercury feedrates of 63 and 79 
g/dscm do not have complete data sets for all metals and 
chlorine. Thus, these conditions cannot be used under the aggregate 
feedrate approach to define the floor level of feedrate control. 
Mercury emissions from those test conditions are used, however, to 
identify a floor emission level that is being achieved.
    \95\ In addition, this floor emission level may be readily 
achievable for new sources using activated carbon injection as floor 
control for dioxiin/furan without the need for feedrate control of 
mercury. Activated carbon injection can achieve mercury emissions 
reductions of 85 percent. Given that the upper bound mercury 
feedrate for ``normal'' wastes (i.e., without mercury spiking) in 
our data base corresponds to a maximum theoretical emission 
concentration of 300 g/dscm, such sources could achieve the 
mercury floor emission level of 45 g/dscm using activated 
carbon injection alone.
---------------------------------------------------------------------------

    d. What Are Our Beyond-the-Floor Considerations for New Sources? We 
evaluated the use of activated carbon injection as beyond-the-floor 
control for new sources to achieve emission levels lower than floor 
levels. In the April 1996 NPRM and May 1997 NODA, we stated that new 
sources could achieve a beyond-the-floor level of 4 g/dscm 
based on use of activated carbon injection. We cited significant cost-
effectiveness concerns at that level, however. We reiterate those 
concerns today.
    Many commenters object to our beyond-the-floor standards as 
proposed, citing high costs for achieving relatively small mercury 
emission reductions. They compare the proposed standards unfavorably 
with other sources' regulations (e.g., electric utilities, municipal 
and medical waste incinerators), where the cost-effectiveness values 
are much lower. As stated earlier, comparison between rules for 
different sources is not directly relevant. Nonetheless, we conclude 
that use of activated carbon injection as a beyond-the-floor control 
for mercury for new sources would not be cost-effective. We also note 
that the floor levels are adequately protective to satisfy RCRA 
requirements.
    We also considered additional feedrate control of mercury as 
beyond-the-floor control. We conclude, however, that significant 
emission reductions using feedrate control may be problematic because 
the detection limit of routine feedstream analysis procedures for 
mercury is such that a beyond-the-floor mercury emission limit could be 
exceeded even though mercury is not present in feedstreams at 
detectable levels. Although sources could potentially perform more 
sophisticated mercury analyses, cost-effectiveness considerations would 
likely come into play and suggest that a beyond-the-floor standard is 
not warranted.
4. What Are the Standards for Particulate Matter?
    We establish standards for existing and new incinerators which 
limit particulate matter emissions to 0.015 grains/dry standard cubic 
foot (gr/dscf) or 34 milligrams per dry standard cubic meter (mg/
dscm).\96\ We chose the particulate matter standard as a surrogate 
control for the metals antimony, cobalt, manganese, nickel, and 
selenium. We refer to these five metals as ``nonenumerated metals'' 
because standards specific to each metal have not been established. We 
discuss below the rationale for adopting these standards.
---------------------------------------------------------------------------

    \96\ Particulate matter is a surrogate for the metal hazardous 
air pollutants for which we are not establishing metal emission 
standards: Antimony, cobalt, manganese, nickel, and selenium.
---------------------------------------------------------------------------

    a. What Is the MACT Floor for Existing Sources? Our data base 
consists of particulate matter emissions from 75 hazardous waste 
incinerators that range from 0.0002 gr/dscf to 1.9 gr/dscf. Particle 
size distribution greatly affects the uncontrolled particulate matter 
emissions from hazardous waste incinerators, which, in turn, is 
affected by incinerator type and design, particulate matter entrainment 
rates, waste ash content, waste sooting potential and waste chlorine 
content. Final emissions from the stacks of hazardous waste 
incinerators are affected by the degree of control provided to 
uncontrolled particulate matter emissions by the air pollution control 
devices. Dry collection devices include fabric filters or electrostatic 
precipitators, while wet collection devices include conventional wet 
scrubbers (venturi type) or the newer patented scrubbers like 
hydrosonic, free jet, or the collision type. Newer hazardous waste 
incinerators now commonly use ionizing wet scrubbers or wet 
electrostatic precipitators or a combination of both dry and wet 
devices.
    The MACT floor setting procedure involves defining MACT level of 
control based on air pollution control devices used by the best 
performing sources. Control devices used by these best performing 
sources can be expected to routinely and consistently achieve superior 
performance. Then, we identify an emissions level that well designed, 
well-operated and well-maintained MACT controls can achieve based on 
demonstrated performance, and engineering information and principles.
    The average of the best performing 12 percent of hazardous waste 
incinerators use either fabric filters, electrostatic precipitators 
(dry or wet), or ionizing wet scrubbers (sometimes in combination with 
venturi, packed bed, or spray tower scrubbers). As explained in Part 
Four, Section V, we define floor control for particulate matter for 
incinerators as the use of a well-designed, operated, and maintained 
fabric filter, electrostatic precipitator, or ionizing wet scrubber. 
Sources using certain wet scrubbing techniques such as high energy 
venturi scrubbers, and novel condensation, free-jet, and collision 
scrubbers can also have very low particulate matter emission levels. We 
do not consider these devices to be MACT control, however, because, in 
general, a fabric filter, electrostatic precipitator, or ionizing wet 
scrubber will provide superior particulate matter control. In some 
cases, sources using medium or low energy wet scrubbers are achieving 
very low particulate matter emissions, but only for liquid waste 
incinerators, which typically have low ash content waste. Thus, this 
control technology demonstrates high effectiveness only under atypical 
conditions, and we do not consider it to be MACT floor control for 
particulate matter.
    We conclude that fabric filters, electrostatic precipitators, and 
ionizing wet scrubbers are routinely achieving an emission level of 
0.015 gr/dscf based upon the following considerations:
    i. Sources in our data base are achieving this emission level. Over 
75 percent of the sources in the expanded MACT pool are achieving an 
emission level of 0.015 gr/dscf. We investigated several sources in our 
data base using floor control but failing to achieve this level, and we 
found that the control devices do not appear to be well-designed, 
operated, and maintained. Some of these sources are not using superior 
fabric filter bags (e.g., Gore-tex, Nomex felt, or tri-lift 
fabrics), some exhibit salt carry-over and entrainment from a poorly 
operated wet scrubber located downstream of the fabric filter, and some 
are poorly maintained in critical aspects (such as fabric cleaning 
cycle or bag replacements). \97\
---------------------------------------------------------------------------

    \97\ USEPA, ``Technical Support Document for HWC, MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    ii. Well-designed, operated, and maintained fabric filters and 
electrostatic precipitators can routinely achieve particulate matter 
levels lower than the floor level of 0.015 gr/dscf. Levels less than 
0.005 gr/dscf were demonstrated on hazardous waste incinerators and 
municipal waste combustors in many cases. Well-designed fabric filters 
have a surface collection area of over 0.5 ft2/acfm and high 
performance filter fabrics such as Nomex and Gore-tex. Well-designed 
electrostatic precipitators have advanced power system controls (with 
intermittent or pulse energization), internal plate and electrode 
geometry to

[[Page 52865]]

allow for high voltage potential, flue gas conditioning by addition of 
water or reagents such as sulfur trioxide or ammonia to condition 
particulate matter for lower resistivity, and optimized gas 
distribution within the electrostatic precipitator. The technical 
support document identifies many hazardous waste incinerators using 
such well designed control equipment.
    iii. The 0.015 gr/dscf level is well within the accepted 
capabilities of today's particulate matter control devices in the 
market place. Vendors typically guarantee emission levels for the 
particulate matter floor control devices at less than 0.015 gr/dscf and 
in some cases, as low as 0.005 gr/dscf.
    iv. The 0.015 gr/dscf level is consistent with standards 
promulgated for other incinerator source categories burning municipal 
solid waste and medical waste, both of which are based on performance 
of fabric filters or electrostatic precipitators as MACT. Comparison of 
hazardous waste incinerator floor level to these standards is 
appropriate because particulate matter characteristics such as particle 
size distribution, loading and particulate matter type are comparable 
within the above three types of waste burning source categories.
    v. Hazardous waste incinerators that meet the 0.015 gr/dscf 
particulate matter level also generally achieve semivolatile metal 
system removal efficiencies of over 99% and low volatile metal system 
removal efficiencies over 99.9%. This indicates superior particulate 
matter collection efficiency because these metals are controlled by 
controlling fine and medium-sized particulate matter.
    vi. Over 50 percent of all test conditions in the data base, 
regardless of the type of air pollution control device used, design of 
the hazardous waste incinerator, or the type of waste burned, currently 
meet the 0.015 gr/dscf level. This includes hazardous waste 
incinerators with high particulate matter entrainment rates (such as 
fluidized bed and rotary kilns) as well as those with wastes that 
generate difficult to capture fine particulate matter, such as certain 
liquid injection facilities.
    vii. Many incinerators conducted several tests to develop the most 
flexible operating envelope for day-to-day operations, keeping in view 
the existing RCRA particulate matter standard of 0.08 gr/dscf. In many 
test conditions, they elected to meet (and be limited to) the 0.015 gr/
dscf level, although they were only required to meet a 0.08 gr/dscf 
standard.
    Many commenters object to the use of engineering information and 
principles in the selection of the MACT floor level. Some consider 
engineering information and principles highly subjective and dependent 
on reviewers' interpretation of the data, while others suggest the use 
of accepted statistical methods for handling the data. We performed 
analyses based on available statistical tools for outlier analysis and 
variability, as discussed previously, but conclude that those 
approaches are not appropriate. We continue to believe that the use of 
engineering information and principles is a valid approach to establish 
the MACT floor (i.e., to determine the level of performance 
consistently achievable by properly designed and operated floor control 
technology).
    Some commenters object to the use of ``well-designed, operated and 
maintained'' MACT controls. They consider the term too vague and want 
specific parameters and features (e.g., air to cloth ratio for fabric 
filters and power input for electrostatic precipitators) identified. We 
understand commenters' concerns but such information is simply not 
readily available. Further, many parameters work in relation with 
several others making it problematic to quantify optimum values 
separate from the other values. The system as a whole needs to be 
optimized for best control efficiency on a case-by-case basis.
    Some commenters object to our justification of particulate matter 
achievability on the basis of vendors' claims. They contend that: (1) 
Vendors' claims lack quality control and are driven by an incentive for 
sales; (2) vendors' claims are based on normal operating conditions, 
not on trial burn type conditions; and (3) MACT floor should not be 
based on theoretical performance of state-of-the-art technology. We 
would agree with the comments if the vendor information were from 
advertising literature, but instead, our analysis was based on 
warranties. The financial consequences of vendors' warranties require 
those warranties to be conservative and based on proven performance 
records, both during normal operations and during trial burn 
conditions. In any case, we are using vendor information as 
corroboration, not to establish a level of performance.
    In the May 1997 NODA (62 FR at 24222), we requested comments on the 
alternative MACT evaluation method based on defining medium and low 
energy venturi-scrubbers burning low ash wastes as an additional MACT 
control, but screening out facilities from the expanded MACT floor 
universe that have poor semivolatile metal system removal efficiency. 
The resulting MACT floor emission level under this approach would be 
0.029 gr/dscf. Many commenters agree with the Agency that this 
technique is unacceptable because it ignores a majority (over 75 
percent) of the available particulate matter data in identifying the 
MACT standard. This result is driven by the fact that corresponding 
semivolatile metal data are not available from those sources. Other 
commenters, however, suggest that venturi scrubbers should be 
designated as MACT particulate matter control. These commenters suggest 
that sources using venturi scrubbers are within the average of the best 
performing 12 percent of sources, and there is no technical basis for 
their exclusion. As stated above, we agree that well-designed and 
operated venturi scrubbers can achieve the MACT floor level of 0.015gr/
dscf under some conditions (as when burning low ash wastes), but their 
performance is generally not comparable to that of a fabric filter, 
electrostatic precipitator, or ionizing wet scrubber. Thus, we conclude 
that sources equipped with venturi scrubbers may not be able to achieve 
the floor emission level in all cases, and the floor level would have 
to be inappropriately increased to accommodate unrestricted use of 
those units.
    Some commenters state that we must demonstrate health or 
environmental benefits if the rule were to require sources to replace 
existing, less efficient air pollution control devices (e.g., venturi 
scrubbers incapable of meeting the standard) with a better performing 
device, particularly because particulate matter is not a hazardous air 
pollutant under the CAA. These comments are not persuasive and are 
misplaced as a matter of law. The MACT floor process was established 
precisely to obviate such issues and to establish a minimum level of 
control based on performance of superior air pollution control 
technologies. Indeed, the chief motivation for adopting the technology-
based standards to control emissions of hazardous air pollutants in the 
first instance was the evident failure of the very type of risk-based 
approach to controlling air toxics as is suggested by the commenters. 
(See, e.g., H. Rep. No. 490, 101st Cong. 2d Sess., at 318-19.) Inherent 
in technology-based standard setting, of course, is the possibility 
that some technologies will have to be replaced if they cannot achieve 
the same level of performance as the best performing technologies. 
Finally, with regard to the commenters' points regarding particulate 
matter not being a hazardous air pollutant, we explain

[[Page 52866]]

above why particulate matter is a valid surrogate for certain hazardous 
air pollutants, and can be used as a means of controlling hazardous air 
pollutant emissions. In addition, the legislative history appears to 
contemplate regulation of particulate matter as part of the MACT 
process. (See S. Rep. No. 228, 101st Cong. 1st Sess., at 
170.98)
---------------------------------------------------------------------------

    \98\ Control of particulate matter also helps assure that the 
standards are sufficiently protective to make RCRA regulation of 
these sources' air emissions unnecessary (except potentially on a 
site-specific basis through the omnibus permitting process). See 
Technical Support Document on Risk Assessment.
---------------------------------------------------------------------------

    We do not consider cost in selecting MACT floor levels. 
Nevertheless, for purposes of administrative compliance with the 
Regulatory Flexibility Act and various Executive Orders, we estimate 
the cost burden on the hazardous waste incinerator universe to achieve 
compliance. Approximately 38 percent of hazardous waste incinerators 
currently meet the floor level of 0.015 gr/dscf. The annualized cost 
for the remaining 115 incinerators to meet the floor level, assuming no 
market exits, is estimated to be $17.4 million. Nonenumerated metals 
and particulate matter emissions will be reduced nationally by 5.1 Mg/
yr and 1345 Mg/yr, respectively, or over 50 percent from current 
baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the NPRM, we proposed a beyond-the-floor emission level of 
69 mg/dscm (0.030 gr/dscf) and solicited comment on an alternative 
beyond-the-floor emission level of 34 mg/dscm (0.015 gr/dscf) based on 
improved particulate matter control. (61 FR at 17383.) In the May 1997 
NODA, we concluded that a beyond-the-floor standard may not be 
warranted due to significant cost-effectiveness considerations. (62 FR 
at 24222.)
    In the final rule, we considered more stringent beyond-the-floor 
controls that would provide additional reductions of particulate matter 
emissions using fabric filters, electrostatic precipitators, and wet 
ionizing scrubbers that are designed, operated, and maintained to have 
improved collection efficiency. We considered a beyond-the-floor level 
of 16 mg/dscm (0.007 gr/dscf), approximately one-half the floor 
emission level, for existing incinerators based on improved particulate 
matter control. We then determined the cost of achieving this reduction 
in particulate matter, with corresponding reductions in the 
nonenumerated metals for which particulate matter is a surrogate, to 
determine if this beyond-the-floor level would be appropriate. The 
national incremental annualized compliance cost for incinerators to 
meet this beyond-the-floor level, rather than comply with the floor 
controls, would be approximately $6.8 million for the entire hazardous 
waste incinerator industry and would provide an incremental reduction 
in nonenumerated metals emissions nationally beyond the MACT floor 
controls of 1.7 Mg/yr. Based on these costs of approximately $4.1 
million per additional Mg of nonenumerated metals emissions removed, we 
conclude that this beyond-the-floor option for incinerators is not 
acceptably cost-effective nor otherwise justified. Therefore, we do not 
adopt this beyond-the-floor standard. Poor cost-effectiveness would be 
particularly unacceptable here considering that these metals also have 
relatively low toxicity. Thus, the particulate matter standard for new 
incinerators is 34 mg/dscm. Therefore, the cost-effectiveness threshold 
we would select would be less than for more toxic pollutants such as 
dioxin, mercury or other metals.
    c. What Is the MACT Floor for New Sources? We proposed a floor 
level of 0.030 gr/dscf for new sources based on the best performing 
source in the data base, which used a fabric filter with an air-to-
cloth ratio of 3.8 acfm/ft\2\. In the May 1997 NODA, we reevaluated the 
particulate matter floor level and indicated that floor control for 
existing sources would also appear to be appropriate for new sources. 
We are finalizing the approach discussed in the May 1997 NODA whereby 
floor control is a well-designed, operated, and maintained fabric 
filter, electrostatic precipitator, or ionizing wet scrubber, and the 
floor emission level is 0.015 gr/dscf.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? We 
considered more stringent beyond-the-floor controls that would provide 
additional reductions of particulate matter emissions using fabric 
filters, electrostatic precipitators, and wet ionizing scrubbers that 
are designed, operated, and maintained to have improved collection 
efficiency. We considered a beyond-the-floor level of 16 mg/dscm (0.007 
gr/dscf), approximately one-half the emissions level for existing 
sources, for new incinerators based on improved particulate matter 
control. For analysis purposes, improved particulate matter control 
assumes the use of higher quality fabric filter bag material. We then 
determined the cost of achieving this reduction in particulate matter, 
with corresponding reductions in the nonenumerated metals for which 
particulate matter is a surrogate, to determine if this beyond-the-
floor level would be appropriate. The incremental annualized compliance 
cost for one new large incinerator to meet this beyond-the-floor level, 
rather than comply with floor controls, would be approximately $39,000 
and would provide an incremental reduction in nonenumerated metals 
emissions of approximately 0.05 Mg/yr.99 For a new small 
incinerator, the incremental annualized compliance cost would be 
approximately $7,500 and would provide an incremental reduction in 
nonenumerated metals emissions of approximately 0.008 Mg/yr. Based on 
these costs of approximately $0.8-1.0 million per additional Mg of 
nonenumerated metals removed, we conclude that a beyond-the-floor 
standard of 16 mg/dscm is not warranted due to the high cost of 
compliance and relatively small nonenumerated metals emission 
reductions. Poor cost-effectiveness would be particularly unacceptable 
here considering that these metals also have relatively low toxicity. 
Thus, the particulate matter standard for new incinerators is 34 mg/
dscm.
---------------------------------------------------------------------------

    \99\ Based on the data available, the average emissions in sum 
of the five nonenumerated metals from incinerators using MACT 
particulate matter control is approximately 229 g/dscm. To 
estimate emission reductions of the nonenumerated metals for 
specific test conditions, we assume a linear relationship between a 
reduction in particulate matter and these metals.
---------------------------------------------------------------------------

5. What Are the Standards for Semivolatile Metals?
    Semivolatile metals are comprised of lead and cadmium. We establish 
standards which limit semivolatile metal emissions to 240 g/
dscm for existing sources and 24 g/dscm for new sources. We 
discuss below the rationale for adopting these standards.
    a. What Is the MACT Floor for Existing Sources? As discussed in 
Part Four, Section V of the preamble, floor control for semivolatile 
metals is hazardous waste feedrate control of semivolatile metals plus 
MACT floor particulate matter control. We use the aggregate feedrate 
approach to define the level of semivolatile metal feedrate control. We 
have aggregate feedrate data for 20 test conditions from nine hazardous 
waste incinerators that are using MACT floor control for particulate 
matter. The semivolatile metal feedrate levels, expressed as maximum 
theoretical emission concentrations, for these sources range from 100 
g/dscm to 1.5 g/dscm while the semivolatile emissions range 
from 1 to 6,000 g/dscm. The MACT-defining maximum theoretical 
emission concentration is

[[Page 52867]]

5,300 g/dscm. Upon expanding the MACT pool, only the highest 
emissions test condition of 6,000 g/dscm was screened out 
because the semivolatile metal maximum theoretical emission 
concentration for this test condition was higher than the MACT-defining 
maximum theoretical emission concentration. The highest emission test 
condition in the remaining expanded MACT pool identifies a MACT floor 
emission level of 240 g/dscm.
    We originally proposed a semivolatile metal floor standard of 270 
g/dscm based on semivolatile metal feedrate control. We 
subsequently refined the emissions data base and reevaluated the floor 
methodology, and discussed in the May 1997 NODA a semivolatile metal 
floor level of 100 g/dscm. Commenters express serious concerns 
with the May 1997 NODA approach in two areas. First, they note that the 
MACT-defining best performing sources have very low emissions, not 
entirely due to the performance of MACT control, but also due to 
atypically low semivolatile metal feedrates. Second, they object to our 
use of a ``breakpoint'' analysis to screen out the outliers from the 
expanded MACT pool (which was already small due to the screening 
process to define the feedrate level representative of MACT control). 
Our final methodology makes adjustments to address these concerns. 
Under the aggregate feedrate approach, sources with atypically low 
feedrates of semivolatile metals would not necessarily drive the floor 
control feedrate level. This is because the aggregate feedrate approach 
identifies as the best performing sources (relative to feedrate 
control) those with low feedrates in the aggregate for all metals and 
chlorine. In addition, the floor methodology no longer uses the 
breakpoint approach to identify sources not using floor control. These 
issues are discussed above in detail in Part Four, Section V, of the 
preamble.
    Although cost-effectiveness of floor emission levels is not a 
factor in defining floor control or emission levels, we have estimated 
compliance costs and emissions reductions at the floor for 
administrative purposes. Approximately 66 percent of sources currently 
meet the semivolatile metal floor level of 240 g/dscm. The 
annualized cost for the remaining 64 incinerators to meet the floor 
level, assuming no market exits, is estimated to be $1.8 million. 
Semivolatile metal emissions will be reduced nationally by 55.9 Mg per 
year from the baseline emissions level of 58.5 Mg per year, a reduction 
of 95.5%.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We considered more stringent semivolatile metal feedrate 
control as a beyond-the-floor control to provide additional reductions 
in emissions. Cost effectiveness considerations would likely come into 
play, however, and suggest that a beyond-the-floor standard is not 
warranted. Therefore, we conclude that a beyond-the-floor standard for 
semivolatile metals for existing sources is not appropriate. We note 
that a beyond-the-floor standard is not needed to meet our RCRA 
protectiveness mandate.
    c. What Is the MACT Floor for New Sources? Floor control for new 
sources is: (1) The level of semivolatile metal feedrate control used 
by the source with the lowest aggregate feedrate for all metals and 
chlorine;100 and (2) use of MACT floor particulate matter 
control for new sources (i.e., a fabric filter, electrostatic 
precipitator, or wet ionizing scrubber achieving a particulate matter 
emission level of 0.015 gr/dscf). Three sources in our data base are 
currently using the floor control selected for all new sources and are 
achieving semivolatile emissions ranging from 2 g/dscm to 24 
g/dscm. To ensure that the floor level is achievable by all 
sources using floor control, we are establishing the floor level for 
semivolatile metals for new sources at 24 g/dscm.
---------------------------------------------------------------------------

    \100\ I.e., a semivolatile metal feedrate equivalent to a 
maximum theoretical emission concentration of 3,500 g/dscm.
---------------------------------------------------------------------------

    d. What Are Our Beyond-the-Floor Considerations for New Sources? We 
considered more stringent beyond-the-floor controls (i.e., a more 
restrictive semivolatile metal feedrate) to provide additional 
reduction in emissions. We determined that cost-effectiveness 
considerations would likely be unacceptable due to the relatively low 
concentrations achieved at the floor. This suggests that a beyond-the-
floor standard is not warranted. We note that a beyond-the-floor 
standard is not needed to meet our RCRA protectiveness mandate.
6. What Are the Standards for Low Volatile Metals?
    Low volatile metals are comprised of arsenic, beryllium, and total 
chromium. We establish standards that limit emissions of these metals 
to 97 g/dscm for both existing and new incinerators. We 
discuss below the rationale for adopting these standards.
    a. What Is the MACT Floor for Existing Sources? We are using the 
same approach for low volatile metals as we did for semivolatile metals 
to define floor control. Floor control for low volatile metals is use 
of particulate matter floor control and control of the feedrate of low 
volatile metals to a level identified by the aggregate feedrate 
approach.
    The low volatile metal feedrates for sources using particulate 
matter floor control range from 300 g/dscm to 1.4 g/dscm when 
expressed as maximum theoretical emission concentrations. Emission 
levels for these sources range from 1 to 803 g/dscm. 
Approximately 60 percent of sources using particulate matter floor 
control have low volatile metal feedrates below the MACT floor 
feedrate--24,000 g/dscm, expressed as a maximum theoretical 
emission concentration.
    Upon expanding the MACT pool, the source using floor control with 
the highest emissions is achieving an emission level of 97 g/
dscm. Accordingly, we are establishing the floor level for low volatile 
metals for existing sources at 97 g/dscm to ensure that the 
floor level is achievable by all sources using floor control.
    We identified a low volatile metal floor level of 210 g/
dscm in the April 1996 proposal. The refined data analysis in the May 
1997 NODA, based on the revised data base, reduced the low volatile 
metal floor level to 55 g/dscm. As with semivolatile metals, 
commenters express serious concerns with the May 1997 NODA approach, 
including selection of the breakpoint ``outlier'' screening approach 
and use of hazardous waste incinerator data with atypically low 
feedrates for low volatile metals. We acknowledge those concerns and 
adjusted our methodology accordingly. See discussions above in Part 
Four, Section V.
    We estimated compliance costs to the hazardous waste incinerator 
universe for administrative purposes. Approximately 63 percent of 
incinerators currently meet the 97 g/dscm floor level. The 
annualized cost for the remaining 69 incinerators to meet the floor 
level, assuming no market exits, is estimated to be $1.9 million, and 
would reduce low volatile metal emissions nationally by 6.9 Mg per year 
from the baseline emissions level of 8 Mg per year.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We considered more stringent beyond-the-floor controls (i.e., 
a more restrictive low volatile metal feedrate) to provide additional 
reduction in emissions. Due to the relatively low concentrations 
achieved at the floor, we determined that cost-effectiveness 
considerations would likely be unacceptable. Therefore, we conclude 
that a beyond-the-floor standard for low volatile metals for existing 
sources is not

[[Page 52868]]

appropriate. We note that a beyond-the-floor standard is not needed to 
meet our RCRA protectiveness mandate.
    c. What Is the MACT Floor for New Sources? We identified a floor 
level of 260 g/dscm for new sources at proposal based on the 
best performing source in the data base. That source uses a venturi 
scrubber with a low volatile metal feedrate equivalent to a maximum 
theoretical emission concentration of 1,000 g/dscm. Our 
reevaluation of the data base in the May 1997 NODA identified a floor 
level of 55 g/dscm based on use of floor control for 
particulate matter and feedrate control of low volatile metals. Other 
than the comments on the two issues of low feedrate and the 
inappropriate use of a breakpoint analysis discussed above, no other 
significant comments challenged this floor level.
    Floor control for new sources is the same as discussed in the May 
1997 NODA (i.e., use of particulate matter floor control and feedrate 
control of low volatile metals), except the floor feedrate level under 
the aggregate feedrate approach used for today's final rule is 13,000 
g/dscm. Upon expanding the MACT pool, the source using floor 
control with the highest emissions is achieving an emission level of 97 
g/dscm.101 Accordingly, we are establishing the 
floor level for low volatile metals for new sources at 97 g/
dscm to ensure that the floor level is achievable by all sources using 
floor control.
---------------------------------------------------------------------------

    \101\ The emission level for new sources achieving a feedrate 
control of 13,000 g/dscm (expressed as a maximum 
theoretical emission concentration) is the same as the emission 
level for existing sources achieving a feedrate control of 24,000 
g/dscm because sources feeding low volatile metals in the 
range of 13,000 to 24,000 g/dscm have emission levels at or 
below 97 g/dscm. Although these sources feel low volatile 
metals at higher levels than the single best feedrate-controlled 
source, their emission control devices apparently are more 
efficient. Thus, they achieved lower emissions than the single best 
feedrate-controlled source.
---------------------------------------------------------------------------

    d. What Are Our Beyond-the-Floor Considerations for New Sources? We 
considered more stringent beyond-the-floor controls (i.e., a more 
restrictive low volatile metal feedrate) to provide additional 
reduction in emissions. Because of the relatively low concentrations 
achieved, we determined that cost-effectiveness considerations would 
likely be unacceptable. Therefore, we conclude that a beyond-the-floor 
standard for low volatile metals for new sources is not appropriate. We 
note that a beyond-the-floor standard is not needed to meet our RCRA 
protectiveness mandate.
7. What Are the Standards for Hydrochloric Acid and Chlorine Gas?
    We establish standards for hydrochloric acid and chlorine gas, 
combined, for existing and new incinerators of 77 and 21 ppmv 
respectively. We discuss below the rationale for adopting these 
standards.
    a. What Is the MACT Floor for Existing Sources? Almost all 
hazardous waste incinerators currently use some type of add-on stack 
gas wet scrubbing system, in combination with control of the feedrate 
of chlorine, to control emissions of hydrochloric acid and chlorine 
gas. A few sources use dry or semi-dry scrubbing, alone or in 
combination with wet scrubbing, while a few rely upon feedrate control 
only. Wet scrubbing consistently provides a system removal efficiency 
of over 99 percent for various scrubber types and configurations. 
Current RCRA regulations require 99% removal efficiency and most 
sources are achieving greater than 99.9 percent removal efficiency. 
Accordingly, floor control is defined as wet scrubbing achieving a 
system removal efficiency of 99 percent or greater combined with 
feedrate control of chlorine.
    The floor feedrate control level for chlorine is 22 g/
dscm, expressed as a maximum theoretical emission concentration, based 
on the aggregate feedrate approach. The source in the expanded MACT 
pool (i.e., all sources using floor control) with the highest emission 
levels of hydrogen chloride and chlorine gas is achieving an emission 
level of 77 ppmv. Thus, MACT floor for existing sources is 77 ppmv.
    At proposal, we also defined floor control as wet scrubbing 
combined with feedrate control of chlorine. We proposed a floor 
emission level of 280 ppmv based on a chlorine feedrate control level 
of 21 g/dscm, expressed as a maximum theoretical emission 
concentration. The best performing sources relative to emission levels 
all use wet scrubbing and feed chlorine at that feedrate or lower. We 
identified a floor level of 280 ppmv based on all sources in our data 
base using floor control and after applying a statistically-derived 
emissions variability factor. In the May 1997 NODA, we again defined 
floor control as wet (or dry) scrubbing with feedrate control of 
chlorine. We discussed a floor emission level of 75 ppmv based on the 
revised data base and break-point floor methodology. Rather than using 
a break-point analysis in the final rule, we use a floor methodology 
that identifies floor control as an aggregate chlorine feedrate 
combined with scrubbing that achieves a removal efficiency of at least 
99 percent.
    We estimated compliance costs to the hazardous waste incinerator 
universe for administrative purposes. Approximately 70 percent of 
incinerators currently meet the hydrochloric acid and chlorine gas 
floor level of 77 ppmv. The annualized cost for the remaining 57 
incinerators to meet that level, assuming no market exits, is estimated 
to be $4.75 million and would reduce emissions of hydrochloric acid and 
chlorine gas nationally by 2,670 Mg per year from the baseline 
emissions level of 3410 Mg per year, a reduction of 78%.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We considered more stringent beyond-the-floor controls to 
provide additional reduction in emissions. Due to the relatively low 
concentrations achieved at the floor, we determined that cost-
effectiveness considerations would likely be unacceptable. Therefore, 
we conclude that a beyond-the-floor standard for hydrochloric acid and 
chlorine gas for existing sources is not appropriate. We note that a 
beyond-the-floor standard is not needed to meet our RCRA protectiveness 
mandate.
    c. What Is the MACT Floor for New Sources? We identified a floor 
level of 280 ppmv at proposal based on the best performing source in 
the data base. That source uses wet scrubbing and a chlorine feedrate 
of 17 g/dscm, expressed as a maximum theoretical emission 
concentration. Our reevaluation of the revised data base in the May 
1997 NODA defined a floor level of 75 ppmv. Based on the aggregate 
feedrate approach used for today's final rule, we are establishing a 
floor level of 21 ppmv, based on a chlorine feedrate of 4.7 g/
dscm expressed as a maximum theoretical emission concentration.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? We 
considered more stringent beyond-the-floor controls to provide 
additional reduction in emissions. Due to the relatively low 
concentrations achieved at the floor, we determined that cost-
effectiveness considerations would likely be unacceptable. Therefore, 
we conclude that a beyond-the-floor standard for hydrochloric acid and 
chlorine gas for new sources is not appropriate. We note that a beyond-
the-floor standard is not needed to meet our RCRA protectiveness 
mandate.
8. What Are the Standards for Carbon Monoxide?
    We use carbon monoxide as a surrogate for organic hazardous air 
pollutants. Low carbon monoxide

[[Page 52869]]

concentrations in stack gas are an indicator of good control of organic 
hazardous air pollutants and are achieved by operating under good 
combustion practices.
    We establish carbon monoxide standards of 100 ppmv for both 
existing and new sources based on the rationale discussed below. 
Sources have the option to comply with either the carbon monoxide or 
the hydrocarbon emission standard. Sources that elect to comply with 
the carbon monoxide standard must also document compliance with the 
hydrocarbon standard during the performance test to ensure control of 
organic hazardous air pollutants. See discussion in Part Four, Section 
IV.B.
    a. What Is the MACT Floor for Existing Sources? As proposed, floor 
control for existing sources is operating under good combustion 
practices (e.g., providing adequate excess oxygen; providing adequate 
fuel (waste) and air mixing; maintaining high temperatures and adequate 
combustion gas residence time at those temperatures).102 
Given that there are many interdependent parameters that affect 
combustion efficiency and thus carbon monoxide emissions, we were not 
able to quantify ``good combustion practices.''
---------------------------------------------------------------------------

    \102\ USEPA, ``Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    We are identifying a floor level of 100 ppmv on an hourly rolling 
average, as proposed, because it is being achieved by sources using 
good combustion practices. More than 80 percent of test conditions in 
our data base have carbon monoxide levels below 100 ppmv, and more than 
60 percent have levels below 20 ppmv. Of approximately 20 test 
conditions with carbon monoxide levels exceeding 100 ppmv, we know the 
characteristics of many of these sources are not representative of good 
combustion practices (e.g., use of rotary kilns without afterburners; 
liquid injection incinerators with rapid combustion gas quenching). In 
addition, we currently limit carbon monoxide concentrations for 
hazardous waste burning boilers and industrial furnaces to 100 ppmv to 
ensure good combustion conditions and control of organic toxic 
compounds. Finally, we have established carbon monoxide limits in the 
range of 50 to 150 ppmv on other waste incineration sources (i.e., 
municipal waste combustors, medical waste incinerators) to ensure good 
combustion conditions. We are not aware of reasons why it may be more 
difficult for a hazardous waste incinerator to achieve carbon monoxide 
levels of 100 ppmv.
    We estimated compliance costs to the hazardous waste incinerator 
universe for administrative purposes. Because carbon monoxide emissions 
from these sources are already regulated under RCRA, approximately 97 
percent of incinerators currently meet the floor level of 100 ppmv. The 
annualized cost for the remaining six incinerators to meet the floor 
level, assuming no market exits, is estimated to be $0.9 million and 
would reduce carbon monoxide emissions nationally by 45 Mg per year 
from the baseline emissions level of 9170 Mg per year.103 
Although we cannot quantify a corresponding reduction of organic 
hazardous air pollutant emissions, we estimate these reductions would 
be significant based on the carbon monoxide reductions.
---------------------------------------------------------------------------

    \103\ As discussed previously in the text, you have the option 
of complying with the hydrocarbon emission standard rather than the 
carbon monoxide standard. This is because carbon monoxide is a 
conservative indicator of the potential for emissions of organic 
compounds while hydrocarbon concentrations in stack gas are a direct 
measure of emissions of organic compounds.
---------------------------------------------------------------------------

    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We considered more stringent beyond-the-floor controls (i.e., 
better combustion practices resulting in lower carbon monoxide levels) 
to provide additional reduction in emissions. Although it is difficult 
to quantify the reduction in emissions of organic hazardous air 
pollutants that would be associated with a lower carbon monoxide limit, 
we concluded that cost-effectiveness considerations would likely come 
into play, and suggest that a beyond-the-floor standard is not 
warranted. Therefore, we conclude that a beyond-the-floor standard for 
carbon monoxide for existing sources is not appropriate. We note that, 
although control of carbon monoxide (or hydrocarbon) is not an absolute 
guarantee that nondioxin/furan products of incomplete combustion will 
not be emitted at levels of concern, this problem (where it may exist) 
can be addressed through the RCRA omnibus permitting process.
    c. What Is the MACT Floor for New Sources? At proposal and in the 
May 1997 NODA, we stated that operating under good combustion practices 
defines MACT floor control for new (and existing) 
sources,104 and the preponderance of data indicate that a 
floor level of 100 ppmv over an hourly rolling average is readily 
achievable. For reasons set forth in the proposal, and absent data to 
the contrary, we conclude that this floor level is appropriate.
---------------------------------------------------------------------------

    \104\ Because we cannot quantify good combustion practices, 
floor control for the single best controlled source is the same as 
for existing sources (i.e., that combination of design, operation, 
and maintenance that achieves good combustion as evidenced by carbon 
monoxide levels of 100 ppmv or less on an hourly rolling average).
---------------------------------------------------------------------------

    d. What Are Our Beyond-the-Floor Considerations for New Sources? We 
considered more stringent beyond-the-floor controls (i.e., better 
combustion practices resulting in lower carbon monoxide levels) to 
provide additional reduction in emissions. For the reasons discussed 
above in the context of beyond-the-floor controls for existing sources, 
however, we conclude that a beyond-the-floor standard for carbon 
monoxide for new sources is not appropriate.
9. What Are the Standards for Hydrocarbon?
    Hydrocarbon concentrations in stack gas are a direct surrogate for 
emissions of organic hazardous pollutants. We establish hydrocarbon 
standards of 10 ppmv for both existing and new sources based on the 
rationale discussed below. Sources have the option to comply with 
either the carbon monoxide or the hydrocarbon emission standard. 
Sources that elect to comply with the carbon monoxide standard, 
however, must nonetheless document compliance with the hydrocarbon 
standard during the comprehensive performance test.
    a. What Is the MACT Floor for Existing Sources? We proposed a 
hydrocarbon emission standard of 12 ppmv 105 based on good 
combustion practices, but revised it in the May 1997 NODA to 10 ppmv 
based on refinements of analysis and the corrected data base.
---------------------------------------------------------------------------

    \105\ Based on an hourly rolling average, reported as propane, 
corrected to 7 percent oxygen, dry basis.
---------------------------------------------------------------------------

    As proposed, floor control for existing sources is operating under 
good combustion practices (e.g., providing adequate excess oxygen; 
providing adequate fuel (waste) and air mixing; maintaining high 
temperatures and adequate combustion gas residence time at those 
temperatures). Given that there are many interdependent parameters that 
affect combustion efficiency and thus hydrocarbon emissions, we are not 
able to quantify good combustion practices.
    We are identifying a floor level for the final rule of 10 ppmv on 
an hourly rolling average because it is being achieved using good 
combustion practices. More than 85 percent of test conditions in our 
data base have hydrocarbon levels below 10 ppmv, and nearly 75 percent 
have levels below 5 ppmv. Although 13 test conditions in our data base 
representing 7 sources have hydrocarbon levels higher than 10 ppmv, we 
conclude that these sources

[[Page 52870]]

are not operating under good combustion practices. For example, one 
source is a rotary kiln without an afterburner. Another source is a 
fluidized bed type incinerator that operates at lower than typical 
combustion temperatures without an afterburner while another source is 
operating at high carbon monoxide levels, indicative of poor combustion 
efficiency.106
---------------------------------------------------------------------------

    \106\ USEPA, ``Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    Some commenters on the May 1997 NODA object to the 10 ppmv level 
and suggest adopting a level of 20 ppmv based on the BIF rule 
(Sec. 266.104(c)), and an earlier hazardous waste incinerator proposal 
(55 FR 17862 (April 27, 1990)). These commenters cite sufficient 
protectiveness at the 20 ppmv level. We conclude that this comment is 
not on point because the MACT standards are technology rather than 
risk-based. The MACT standards must reflect the level of control that 
is not less stringent than the level of control achieved by the best 
performing sources. Because hazardous waste incinerators are readily 
achieving a hydrocarbon level of 10 ppmv using good combustion 
practices, that floor level is appropriate.
    Some commenters also object to the requirement to use heated flame 
ionization hydrocarbon detectors 107 in hazardous waste 
incinerators that use wet scrubbers. The commenters state that these 
sources have a very high moisture content in the flue gas that hinders 
proper functioning of the specified hydrocarbon detectors. We agree 
that hydrocarbon monitors may be hindered in these situations. For this 
and other reasons (e.g., some sources can have high carbon monoxide but 
low hydrocarbon levels), the final rule gives sources the option of: 
(1) Continuous hydrocarbon monitoring; or (2) continuous carbon 
monoxide monitoring and demonstration of compliance with the 
hydrocarbon standard only during the performance test.
---------------------------------------------------------------------------

    \107\ See Performance Specification 8A, appendix B, part 60, 
``Specifications and test procedures for carbon monoxide and oxygen 
continuous monitoring systems in stationary sources.''
---------------------------------------------------------------------------

    We estimated compliance costs to the hazardous waste incinerator 
universe for administrative purposes. Approximately 97 percent of 
incinerators currently meet the hydrocarbon floor level of 10 ppmv. The 
annualized cost for the remaining six incinerators to meet the floor 
level, assuming no market exits, is estimated to be $0.35 million, and 
would reduce hydrocarbon emissions nationally by 28 Mg per year from 
the baseline emissions level of 292 Mg per year. Although the 
corresponding reduction of organic hazardous air pollutant emissions 
cannot be quantified, these reductions are qualitatively assessed as 
significant.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We considered more stringent beyond-the-floor controls (i.e., 
better combustion practices resulting in lower hydrocarbon levels) to 
provide additional reduction in emissions. Although it is difficult to 
quantify the reduction in emissions of organic hazardous air pollutants 
that would be associated with a lower hydrocarbon limit, cost-
effectiveness considerations would likely come into play, however, and 
suggest that a beyond-the-floor standard is not warranted. Therefore, 
we conclude that a beyond-the-floor standard for hydrocarbon emissions 
for existing sources is not appropriate. We note further that, although 
control of hydrocarbon emissions is not an absolute guarantee that 
nondioxin products of incomplete combustion will not be emitted at 
levels of concern, this problem (where it may exist) can be addressed 
through the RCRA omnibus permitting process.
    c. What Is the MACT Floor for New Sources? At proposal and in the 
May 1997 NODA, we stated that operation under good combustion practices 
at new (and existing) hazardous waste incinerators defines the MACT 
control.108 As discussed above, sources using good 
combustion practices are achieving hydrocarbon levels of 10 ppmv or 
below. Comments on this subject were minor and did not identify any 
problems in achieving the 10 ppmv level by new sources. Thus, we 
conclude that a floor level of 10 ppmv on hourly rolling average is 
appropriate for new sources.
---------------------------------------------------------------------------

    \108\ Because we cannot quantify good combustion practices, 
floor control for the single best controlled soruce is the same as 
for existing sources (i.e., that combination of design, operation, 
and maintenance that achieves good combustion as evidenced by 
hydrocarbon levels of 10 ppmv or less on an hourly rolling average).
---------------------------------------------------------------------------

    d. What Are Beyond-the-Floor Considerations for New Sources? We 
considered more stringent beyond-the-floor controls (i.e., better 
combustion practices) to provide additional reduction in emissions. For 
the reasons discussed above in the context of beyond-the-floor controls 
for existing sources, however, we conclude that a beyond-the-floor 
standard for hydrocarbons for new sources is not appropriate.
10. What Are the Standards for Destruction and Removal Efficiency?
    We establish a destruction and removal efficiency (DRE) standard 
for existing and new incinerators to control emissions of organic 
hazardous air pollutants other than dioxins and furans. Dioxins and 
furans are controlled by separate emission standards. See discussion in 
Part Four, Section IV.A. The DRE standard is necessary, as previously 
discussed, to complement the carbon monoxide and hydrocarbon emission 
standards, which also control these hazardous air pollutants.
    The standard requires 99.99 percent DRE for each principal organic 
hazardous constituent (POHC), except that 99.9999 percent DRE is 
required if specified dioxin-listed hazardous wastes are burned. These 
wastes are listed as--F020, F021, F022, F023, F026, and F027--RCRA 
hazardous wastes under Part 261 because they contain high 
concentrations of dioxins.
    a. What Is the MACT Floor for Existing Sources? Existing sources 
are currently subject to DRE standards under Sec. 264.342 and 
Sec. 264.343(a) that require 99.99 percent DRE for each POHC, except 
that 99.9999 percent DRE is required if specified dioxin-listed 
hazardous wastes are burned. Accordingly, these standards represent 
MACT floor. Since all hazardous waste incinerators are currently 
subject to these DRE standards, they represent floor control, i.e., 
greater than 12 percent of existing sources are achieving these 
controls.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? Beyond-the-floor control would be a requirement to achieve a 
higher percentage DRE, for example, 99.9999 percent DRE for POHCs for 
all hazardous wastes. A higher DRE could be achieved by improving the 
design, operation, or maintenance of the combustion system to achieve 
greater combustion efficiency.
    Sources will not incur costs to achieve the 99.99 percent DRE floor 
because it is an existing RCRA standard. A substantial number of 
existing incinerators are not likely to be routinely achieving 99.999 
percent DRE, however, and most are not likely to be achieving 99.9999 
percent DRE. Improvements in combustion efficiency will be required to 
meet these beyond-the-floor DREs. Improved combustion efficiency is 
accomplished through better mixing, higher temperatures, and longer 
residence times. As a practical matter, most combustors are mixing-
limited. Thus, improved mixing is

[[Page 52871]]

necessary for improved DREs. For a less-than-optimum burner, a certain 
amount of improvement may typically be accomplished by minor, 
relatively inexpensive combustor modifications--burner tuning 
operations such as a change in burner angle or an adjustment of swirl--
to enhance mixing on the macro-scale. To achieve higher and higher 
DREs, however, improved mixing on the micro-scale may be necessary 
requiring significant, energy intensive and expensive modifications 
such as burner redesign and higher combustion air pressures. In 
addition, measurement of such DREs may require increased spiking of 
POHCs and more sensitive stack sampling and analysis methods at added 
expense.
    Although we have not quantified the cost-effectiveness of a beyond-
the-floor DRE standard, we do not believe that it would be cost-
effective. For reasons discussed above, we believe that the cost of 
achieving each successive order-of-magnitude improvement in DRE will be 
at least constant, and more likely increasing. Emissions reductions 
diminish substantially, however, with each order of magnitude 
improvement in DRE. For example, if a source were to emit 100 gm/hr of 
organic hazardous air pollutants assuming zero DRE, it would emit 10 
gm/hr at 90 percent DRE, 1 gm/hr at 99 percent DRE, 0.1 gm/hr at 99.9 
percent DRE, 0.01 gm/hr at 99.99 percent DRE, and 0.001 gm/hr at 99.999 
percent DRE. If the cost to achieve each order of magnitude improvement 
in DRE is roughly constant, the cost-effectiveness of DRE decreases 
with each order of magnitude improvement in DRE. Consequently, we 
conclude that this relationship between compliance cost and diminished 
emissions reductions associated with a more stringent DRE standard 
suggests that a beyond-the-floor standard is not warranted.
    c. What Is the MACT Floor for New Sources? The single best 
controlled source, and all other hazardous waste incinerators, are 
subject to the existing RCRA DRE standard under Sec. 264.342 and 
Sec. 264.343(a). Accordingly, we adopt this standard as the MACT floor 
for new sources.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? As 
discussed above, although we have not quantified the cost-effectiveness 
of a more stringent DRE standard, diminishing emissions reductions with 
each order of magnitude improvement in DRE suggests that cost-
effectiveness considerations would likely come into play. We conclude 
that a beyond-the-floor standard is not warranted.

VII. What Are the Standards for Hazardous Waste Burning Cement Kilns?

A. To Which Cement Kilns Do Today's Standards Apply?
    The standards promulgated today apply to each existing, 
reconstructed, and newly constructed Portland cement manufacturing kiln 
that burns hazardous waste. These standards apply to all hazardous 
waste burning cement kilns (both major source and area source cement 
plants). Portland cement kilns that do not engage in hazardous waste 
burning operations are not subject to this NESHAP. However, these 
hazardous waste burning kilns would be subject to the NESHAP for other 
sources of hazardous air pollutants at the facility (e.g., clinker 
cooler stack) that we finalized in June 1999.109
---------------------------------------------------------------------------

    \109\ On June 14, 1999, we promulgated regulations for kiln 
stack emissions for nonhazardous waste burning cement kilns and 
other sources of hazardous air pollutants at all Portland 
manufacturing plants. (See 64 FR 31898.)
---------------------------------------------------------------------------

B. How Did EPA Initially Classify Cement Kilns?
1. What Is the Basis for a Separate Class Based on Hazardous Waste 
Burning?
    Portland cement manufacturing is one of the initial 174 categories 
of major and area sources of hazardous air pollutants listed pursuant 
to section 112(c)(1) for which section 112(d) standards are to be 
established.110 We divided the Portland cement manufacturing 
source category into two different classes based on whether the cement 
kiln combusts hazardous waste. This action was taken for two principal 
reasons: If hazardous wastes are burned in the kiln, emissions of 
hazardous air pollutants can be different for the two types of kilns in 
terms of both types and concentrations of hazardous air pollutants 
emitted, and metals and chlorine emissions are controlled in a 
significantly different manner.
---------------------------------------------------------------------------

    \110\ EPA published an initial list of 174 categories of area 
and major sources in the Federal Register on July 16, 1992. (See 57 
FR at 31576.)
---------------------------------------------------------------------------

    A comparison of metals levels in coal and in hazardous waste fuel 
burned in lieu of coal on a heat input basis reveals that hazardous 
waste frequently contains higher concentrations of hazardous air 
pollutant metals (i.e., mercury, semivolatile metals, low volatile 
metals) than coal. Hazardous waste contains higher levels of 
semivolatile metals than coal by more than an order of magnitude at 
every cement kiln in our data base.111 In addition, coal 
concentrations of mercury and low volatile metals were less than 
hazardous waste by approximately an order of magnitude at every 
facility except one. Thus, a cement kiln feeding a hazardous waste fuel 
is likely to emit more metal hazardous air pollutants than a 
nonhazardous waste burning cement kiln. Given this difference in 
emissions characteristics, we divided the Portland cement manufacturing 
source category into two classes based on whether hazardous waste is 
burned in the cement kiln.
---------------------------------------------------------------------------

    \111\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    Today's rule does not establish hazardous air pollutant emissions 
limits for other hazardous air pollutant-emitting sources at a 
hazardous waste burning cement plant. These other sources of hazardous 
air pollutants may include materials handling operations, conveyor 
system transfer points, raw material dryers, and clinker coolers. 
Emissions from these sources are subject to the requirements 
promulgated in the June 14, 1999 Portland cement manufacturing NESHAP. 
See 64 FR 31898. These standards are applicable to these other sources 
of hazardous air pollutants at all Portland cement plants, both for 
nonhazardous waste burners and hazardous waste burners.
    In addition, this regulation does not establish standards for 
cement kiln dust management facilities (e.g., cement kiln dust piles or 
landfills). We are developing cement kiln dust storage and disposal 
requirements in a separate rulemaking.
2. What Is the Basis for Differences in Standards for Hazardous Waste 
and Nonhazardous Waste Burning Cement Kilns?
    Today's final standards for hazardous waste burning cement kilns 
are identical in some respects to those finalized for nonhazardous 
waste burning cement kilns on June 14, 1999. The standards differ, 
however, in several important aspects. A comparison of the major 
features of the two sets of standards and the basis for major 
differences is discussed below.
    a. How Does the Regulation of Area Sources Differ? As discussed 
earlier, this rule makes a positive area source finding under section 
112(c)(3) of the CAA (i.e., a finding that hazardous air pollutant 
emissions from an area source can pose potential risk to human health 
and the environment) for existing hazardous waste burning cement kilns 
and subjects area sources to the same standards that apply to major 
sources. (See Part Three, Section III.B of today's preamble.) For 
nonhazardous waste burning cement kilns, however, we regulate area 
sources under authority of

[[Page 52872]]

section 112(c)(6) of the CAA, and so apply MACT standards only to the 
section 112(c)(6) hazardous air pollutants emitted from such sources.
    The positive finding for hazardous waste burning cement kilns is 
based on several factors and, in particular, on concern about potential 
health risk from emissions of mercury and nondioxin/furan organic 
hazardous air pollutants which are products of incomplete combustion.
    However, we do not have this same level of concern with hazardous 
air pollutant emissions from nonhazardous waste burning cement kilns 
located at area source cement plants, and so did not make a positive 
area source finding. As discussed above, mercury emissions from 
hazardous waste burning cement kilns are generally higher than those 
from nonhazardous waste burning cement kilns. Also, nondioxin and 
nonfuran organic hazardous air pollutants emitted from hazardous waste 
burning cement kilns have the potential to be greater than those from 
nonhazardous waste burning cement kilns because hazardous waste can 
contain high concentrations of a wide-variety of organic hazardous air 
pollutants. In addition, some hazardous waste burning cement kilns feed 
containers of hazardous waste at locations (e.g., midkiln, raw material 
end of the kiln) other than the normal coal combustion zone. If such 
firing systems are poorly designed, operated, or maintained, emissions 
of nondioxin and furan organic hazardous air pollutants could be 
substantial (and, again, significantly greater than comparable 
emissions from nonhazardous waste Portland cement plants). Finally, 
hazardous air pollutant emissions from nonhazardous waste burning 
cement kilns currently are not regulated uniformly under another 
statute as is the case for hazardous waste burning cement kilns which 
affects which pollutants are controlled at the floor for each class.
    Under the June 1999 final rule, existing and new nonhazardous waste 
burning cement kilns at area source plants are subject to dioxin and 
furan emission standards, and a hydrocarbon 112 standard for 
new nonhazardous waste burning cement kilns that are area sources. 
These standards are promulgated under the authority of section 
112(c)(6). That section requires the Agency to establish MACT standards 
for source categories contributing significantly in the aggregate to 
emissions of identified, particularly hazardous air pollutants. The 
MACT process was also applied to the control of mercury, although the 
result was a standard of no control.
---------------------------------------------------------------------------

    \112\ Hydrocarbon emissions would be limited as a surrogate for 
polycyclic organic matter, a category of organic hazardous air 
pollutants identified in section 112(c)(6).
---------------------------------------------------------------------------

    b. How Do the Emission Standards Differ? The dioxin, furan and 
particulate matter emission standards for nonhazardous waste burning 
cement kilns are identical to today's final standard for hazardous 
waste burning cement kilns. The standards for both classes of kilns are 
floor standards and are identical because hazardous waste burning is 
not likely to affect emissions of either dioxin/furan 113 or 
particulate matter. We also conclude that beyond-the-floor standards 
for these pollutants would not be cost-effective for either class of 
cement kilns.
---------------------------------------------------------------------------

    \113\ Later in the text, however, we discuss how hazardous waste 
burning may potentially affect dioxin and furan emissions and the 
additional requirements for hazardous waste burning cement kilns 
that address this concern.
---------------------------------------------------------------------------

    Under today's rule, hazardous waste burning cement kilns are 
subject to emission standards for mercury, semivolatile metals, low 
volatile metals, and hydrochloric acid/chlorine gas, but we did not 
finalize such standards for nonhazardous waste burning cement kilns. 
Currently, emissions of these hazardous air pollutants from hazardous 
waste burning cement kilns are regulated under RCRA. Therefore, we 
could establish floor levels for each pollutant under the CAA. These 
hazardous air pollutants, however, currently are not controlled for 
nonhazardous waste burning cement kilns and floor levels would be 
uncontrolled levels (i.e., the highest emissions currently 
achieved).114 We considered beyond-the-floor controls and 
emission standards for mercury and hydrochloric acid for nonhazardous 
waste burning cement kilns, but conclude that beyond-the-floor 
standards are not cost-effective, especially considering the lower 
rates of current emissions for nonhazardous waste burning plants.
---------------------------------------------------------------------------

    \114\ Although semivolatile metal and low volatile metal are 
controlled by nonhazardous waste burning cement kilns, along with 
other metallic hazardous air pollutants, by controlling particulate 
matter. These metals are not individually controlled by nonhazardous 
waste burning cement kilns as they are for hazardous waste burning 
cement kilns by virtue of individual metal feedrate limits 
established under existing RCRA regulations.
---------------------------------------------------------------------------

    Finally, under today's rule, hazardous waste burning cement kilns 
are subject to emission limits on carbon monoxide and hydrocarbon and a 
destruction and removal efficiency standard to control nondioxin/furan 
organic hazardous air pollutants. We identified these controls as floor 
controls because carbon monoxide and hydrocarbon emissions are 
controlled for these sources under RCRA regulations, as is destruction 
and removal efficiency.115 For nonhazardous waste burning 
cement kilns, carbon monoxide and hydrocarbon emissions currently are 
not controlled, and the destruction and removal efficiency standard, 
established under RCRA, does not apply. Therefore, carbon monoxide, 
hydrocarbon control and the destruction and removal efficiency standard 
are not floor controls for this second group of cement kilns. We 
considered beyond-the-floor controls for hydrocarbon from nonhazardous 
waste burning cement kilns and determined that beyond-the-floor 
controls for existing sources are not cost-effective. The basis of this 
conclusion is discussed in the proposed rule for nonhazardous waste 
burning cement kilns (see 63 FR at 14202). We proposed and finalized, 
however, a hydrocarbon emission standard for new source nonhazardous 
waste cement kilns based on feeding raw materials without an excessive 
organic content.116 See 63 FR at 14202 and 64 FR 31898.
---------------------------------------------------------------------------

    \115\ For hazardous waste burning cement kilns, existing RCRA 
carbon monoxide and hydrocarbon standards do not apply to the main 
stack of a kiln equipped with a by-pass or other means of measuring 
carbon monoxide or hydrocarbon at mid kiln to ensure good combustion 
of hazardous waste. Therefore, there is no carbon monoxide or 
hydrocarbon floor control for such stacks, and we conclude that 
beyond-the-floor controls would not be cost-effective.
    \116\ Consistent with the nonhazardous waste burnign cement kiln 
proposal, however, we subject the main stack of such new source 
hazardous waste burning cemen tkilns to a hydrocarbon standard.
---------------------------------------------------------------------------

    We did not consider a destruction and removal efficiency standard 
as a beyond-the-floor control for nonhazardous waste burning cement 
kilns because, based historically on a unique RCRA statutory provision, 
the DRE standard is designed to ensure destruction of organic hazardous 
air pollutants in hazardous waste fed to hazardous waste combustors. 
The underlying rationale for such a standard is absent for nonhazardous 
waste burning cement kilns that do not combust hazardous waste and that 
feed materials (e.g., limestone, coal) that contain only incidental 
levels of organic hazardous air pollutants.
    c. How Do the Compliance Procedures Differ? We finalized compliance 
procedures for nonhazardous waste burning cement kilns that are similar 
to those finalized today for hazardous waste burning cement kilns. For 
particulate matter, we are implementing a coordinated program to 
document the feasibility of particulate matter continuous emissions 
monitoring

[[Page 52873]]

systems on both nonhazardous waste and hazardous waste burning cement 
kilns. We plan to establish a continuous emissions monitoring systems-
based emission level through future rulemaking that is achievable by 
sources equipped with MACT control (i.e., an electrostatic precipitator 
or fabric filter designed, operated, and maintained to meet the New 
Source Performance Standard particulate matter standard). In the 
interim, we use the opacity standard as required by the New Source 
Performance Standard for Portland cement plants under Sec. 60.62 to 
ensure compliance with the particulate matter standard for both 
hazardous waste and nonhazardous waste burning cement kilns.
    For dioxin/furan, the key compliance parameter will be identical 
for both hazardous waste and nonhazardous waste burning cement kilns--
control of temperature at the inlet to the particulate matter control 
device. Other factors that could contribute to the formation of dioxins 
and furans, however, are not completely understood. As a result, 
hazardous waste burning cement kilns have additional compliance 
requirements to ensure that hazardous waste is burned under good 
combustion conditions. These additional controls are necessary because 
of the dioxin and furan precursors that can be formed from improper 
combustion of hazardous waste, given the hazardous waste firing systems 
used by some hazardous waste burning cement kilns and the potential for 
hazardous waste to contain high concentrations of many organic 
hazardous air pollutants not found in conventional fuels or cement kiln 
raw materials.
    We also require both hazardous waste and nonhazardous waste burning 
cement kilns to conduct performance testing midway between the five-
year periodic comprehensive performance testing to confirm that dioxin/
furan emissions do not exceed the standard when the source operates 
under normal conditions.
C. What Further Subcategorization Considerations Are Made?
    We also fully considered further subdividing the class of hazardous 
waste burning cement kilns itself. For the reasons discussed below, we 
decided that subcategorization is not needed to determine achievable 
MACT standards for all hazardous waste burning cement kilns.
    We considered, but rejected, subdividing the hazardous waste 
burning cement kiln source category on the basis of raw material feed 
preparation, more specifically wet process versus dry process. In the 
wet process, raw materials are ground, wetted, and fed into the kiln as 
a slurry. Approximately 70 percent of the hazardous waste burning 
cement kilns in operation use a wet process. In the dry process, raw 
materials are ground dry and fed into the kiln dry. Within the dry 
process there are three variations: Long kiln dry process, preheater 
process, and preheater-precalciner process. We decided not to 
subcategorize the hazardous waste burning cement kiln category based on 
raw material feed preparation because: (1) The wet process kilns and 
all variations of the dry process kilns use similar raw materials, 
fossil fuels, and hazardous waste fuels; (2) the types and 
concentrations of uncontrolled hazardous air pollutant emissions are 
similar for both process types;117 (3) the same types of 
particulate matter pollution control equipment, specifically either 
fabric filters or electrostatic precipitators, are used by both process 
types, and the devices achieve the same level of performance when used 
by both process types; and (4) the MACT controls we identify are 
applicable to both process types of cement kilns. For example, MACT 
floor controls for metals and chlorine include good particulate matter 
control and hazardous waste feedrate control, as discussed below, the 
particulate matter standard promulgated today is based on the New 
Source Performance Standard, which applies to all cement kilns 
irrespective of process type. Further, a cement kiln operator has great 
discretion in the types of hazardous waste they accept including the 
content of metals and chlorine in the waste. These basic control 
techniques--particulate matter control and feedrate control of metals 
and chlorine--clearly show that subcategorization based on process type 
is not appropriate.
---------------------------------------------------------------------------

    \117\ Although dry process kilns with a separate by-pass stack 
can have higher metals emissions from that stack compared to the 
main stack of other kilns, today's rule allows such kilns to 
flowrate-average its emissions between the main and by-pass stack. 
The average emissions are similar to the emissions from dry and wet 
kilns that have only one stack. Similarly, kilns with in-line raw 
mills have higher mercury emissions when the raw mill is off. 
Today's rule allows such kilns to time-weight average their 
emissions, however, and the time-weighted emissions for those kilns 
are similar to emissions from other hazardous waste burning cement 
kilns.
---------------------------------------------------------------------------

    Some commenters stated that it is not feasible for wet process 
cement kilns to use fabric filters, especially in cold climates, and 
thus subcategorization based on process type is appropriate. The 
problem, commenters contend, is that the high moisture content of the 
flue gas will clog the fabric if the cement-like particulate is wetted 
and subsequently dried, resulting in reduced performance and early 
replacement of the fabric filter bags. Other commenters disagreed with 
these assertions and stated that fabric filter technology can be 
readily applied to wet process kilns given the exit temperatures of the 
combustion gases and the ease of insulating fabric filter systems to 
minimize cold spots in the baghouse to avoid dew point problems and 
minimize corrosion. These commenters pointed to numerous wet process 
applications currently in use at cement kilns with fabric filter 
systems located in cold climates to support their claims.118 
In light of the number of wet process kilns already using fabric 
filters and their various locations, we conclude that wet process 
cement kilns can be equipped with fabric filter systems and that 
subdividing by process type on this basis is not necessary or 
warranted. A review of the particulate matter emissions data for one 
wet hazardous waste burning cement kiln using a fabric filter shows 
that it is achieving the particulate matter standard. We do not have 
data in our data base from the only other wet hazardous waste burning 
cement kiln using a fabric filter; however, this cement kiln recently 
installed and upgraded to a new fabric filter system.
---------------------------------------------------------------------------

    \118\ We are aware of four wet process cement kiln facilities 
operating with fabric filters: Dragon (Thomaston, ME), Giant 
(Harleyville, SC), Holnam (Dundee, MI), and LaFarge (Paulding, OH). 
Commenters also identified kilns in Canada operating with fabric 
filters.
---------------------------------------------------------------------------

    We also fully considered, but ultimately rejected, subdividing the 
hazardous waste burning cement kiln source category between long kilns 
and short kilns (preheater and preheater-precalciner) technologies, and 
those with in-line kiln raw mills. This subcategorization approach was 
recommended by many individual cement manufacturing member companies 
and a cement manufacturing trade organization. Based on information on 
the types of cement kilns that are currently burning hazardous waste, 
these three subcategories consist of the following four subdivisions: 
(1) Short kilns with separate by-pass and main stacks; (2) short kilns 
with a single stack that handles both by-pass and preheater or 
precalciner emissions; (3) long dry kilns that use kiln gas to dry raw 
meal in the raw mill; and (4) others wet kilns, and long dry kilns not 
using in-line kiln raw mill drying. Currently, each of the first three 
categories consists of only one cement kiln facility while

[[Page 52874]]

the kilns at the remaining 15 facilities are in the fourth category: 
wet kilns or long dry kilns that do not use in-line kiln raw mill 
drying.
    Commenters state that these subcategories should be considered 
because the unique design or operating features of the different types 
of kilns could have a significant impact on emissions of one or more 
hazardous air pollutants that we proposed to regulate. Specifically, 
commenters noted the potential flue gas characteristic differences for 
cement kilns using alkali bypasses on short kilns and in-line kiln raw 
mills. For example, kilns with alkali bypasses are designed to divert a 
portion of the flue gas, approximately 10-30%, to remove the 
problematic alkalis, such as potassium and sodium oxides, that can 
react with other compounds in the cool end of the kiln resulting in 
operation problems. Thus, bypasses allow evacuation of the undesirable 
alkali metals and salts, including semivolatile metals and chlorides, 
entrained in the kiln exit gases before they reach the preheater 
cyclones. As a result, the commenters stated that the emission 
concentration of semivolatile metals in the bypass stack is greater 
than in the main stack, and therefore the difference in emissions 
supports subcategorization.
    We agree, in theory, that the emissions profile for some hazardous 
air pollutants can be different for the three kilns types--short kilns 
with and without separate bypass stacks, long kilns with in-line kiln 
raw mills. To consider this issue further, we analyzed floor control 
and floor emissions levels based only on the data and information from 
the other long wet kilns and long dry kilns not using raw mill drying. 
We then considered whether the remaining three kiln types could apply 
the same MACT controls and achieve the resulting emission standards. We 
conclude that these three types of kilns at issue can use the MACT 
controls and achieve the corresponding emission levels identified in 
today's rule for the wet kilns and long dry kilns not using raw mill 
drying.119 As a result, we conclude that there is no 
practical necessity driving a subcategorization approach even though 
one would be theoretically possible. Further, to ensure that today's 
standards are achievable by all cement kilns, we establish a provision 
that allows cement kilns operating in-line kiln raw mills to average 
their emissions based on a time-weighted average concentration that 
considers the length of time the in-line raw mill is on-line and off 
line. We also adopt a provision that allows short cement kilns with 
dual stacks to average emissions on a flow-weighted basis to 
demonstrate compliance with the emissions standards. (See Part Five, 
Section X--Special Provisions for a discussion of these provisions.)
---------------------------------------------------------------------------

    \119\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    In the case of hydrocarbons and carbon monoxide, we developed final 
standards that reflect the concerns raised by several commenters. We 
determined that this approach best accommodated the unique design and 
operating differences between long wet and long dry process and short 
kilns using either a preheater or a preheater and precalciner.
    Existing hazardous waste preheater and preheater-precalciner cement 
kilns, one of each type is burning hazardous waste, are equipped with 
bypass ducts that divert a portion of the kiln off-gas through a 
separate particulate matter control device to remove problematic alkali 
metals. Long cement kilns do not use bypasses designed to remove alkali 
metals. The significance of this operational difference is that 
hydrocarbon and carbon monoxide levels in the bypass gas of short kilns 
is more representative of the combustion efficiency of burning 
hazardous waste and other fuels in the kiln than the measurements made 
in the main stack. Main stack gas measurements of hydrocarbons and 
carbon monoxide, regardless of process type, also include contributions 
from trace levels of organic matter volatilized from the raw materials, 
which can mask the level of combustion efficiency achieved in the kiln.
    Today's tailored standards require cement kilns to monitor 
hydrocarbons and carbon monoxide at the location best indicative of 
good combustion. For short kilns with bypasses, the final rule requires 
monitoring of hydrocarbons and carbon monoxide in the bypass. Long 
kilns are required to comply with the hydrocarbon and carbon monoxide 
standards in the main stack. However, long kilns that operate a mid-
kiln sampling system, for the purpose of removing a representative 
portion of the kiln off-gas to measure combustion efficiency, can 
comply with the hydrocarbon and carbon monoxide standards at the 
midkiln sampling point.
    In addition, establishing separate hydrocarbon and carbon monoxide 
standards reflects the long and short kiln subcategorization approach 
recommended by some commenters. The standards differ because MACT floor 
control for hydrocarbons and carbon monoxide is based primarily on the 
existing requirements of the Boiler and Industrial Furnace rule. In 
that rule, the unique design and operating features of long and short 
kilns were considered in establishing type specific emission limits for 
hydrocarbons and carbon monoxide. Thus, MACT floor control for long and 
short kilns is different. However, we note these same unique design and 
operating features were not a factor in establishing standards for 
other pollutants, including mercury, semivolatile and low volatile 
metals, and hydrochloric acid/chlorine gas, in the Boiler and 
Industrial Furnace rule.
    For the reasons discussed above, subcategorization would not appear 
to be needed to establish uniform, achievable MACT standards for all 
cement kilns burning hazardous waste. Thus, because the differences 
among kiln types ``does not affect the feasibility and effectiveness of 
air pollution control technology,'' subcategorization is not 
appropriate. S. Rep. No. 228, 101st Cong. 1st sess. 166.
D. What Are The Standards for Existing and New Cement Kilns?
1. What Are the Standards for Cement Kilns?
    In this section, the basis for the emissions standards for cement 
kilns is discussed. The kiln emission limits apply to the kiln stack 
gases, in-line kiln raw mill stack gases if combustion gases pass 
through the in-line raw mill, and kiln alkali bypass stack gases if 
discharged through a separate stack from cement plants that burn 
hazardous waste in the kiln. The emissions standards are summarized 
below:

[[Page 52875]]



               Standards for Existing and New Cement Kilns
------------------------------------------------------------------------
 Hazardous air pollutant or              Emissions standard 1
   hazardous air pollutant   -------------------------------------------
          surrogate             Existing sources         New sources
------------------------------------------------------------------------
Dioxin and furan............  0.20 ng TEQ/dscm; or  0.20 ng TEQ/dscm; or
                               0.40 ng TEQ/dscm      0.40 ng TEQ/dscm
                               and control of flue   and control of flue
                               gas temperature not   gas temperature not
                               to exceed 400 deg.F   to exceed 400 deg.F
                               at the inlet to the   at the inlet to the
                               particulate matter    particulate matter
                               control device.       control device.
Mercury.....................  120 g/dscm.  56 g/dscm.
Particulate matter 2........  0.15 kg/Mg dry feed   0.15 kg/Mg dry feed
                               and 20% opacity.      and 20% opacity.
Semivolatile metals.........  240 g/dscm.  180 g/dscm.
Low volatile metals.........  56 g/dscm..  54 g/dscm.
Hydrochloric acid and         130 ppmv............  86 ppmv.
 chlorine gas.
Hydrocarbons: kilns without   20 ppmv (or 100 ppmv  Greenfield kilns: 20
 by-pass 3, 6.                 carbon monoxide) 3.   ppmv (or 100 ppmv
                                                     carbon monoxide and
                                                     50 ppmv 5
                                                     hydrocarbons).
                              ....................  All others: 20 ppmv
                                                     (or 100 ppmv carbon
                                                     monoxide) 3.
Hydrocarbons: kilns with by-  No main stack         50 ppmv 5.
 pass; main stack 4, 6.        standard.
Hydrocarbons: kilns with by-  10 ppmv (or 100 ppmv  10 ppmv (or 100 ppmv
 pass; by-pass duct and        carbon monoxide).     carbon monoxide).
 stack 3, 4, 6.
Destruction and removal        For existing and new sources, 99.99% for
 efficiency.                        each principal organic hazardous
                                   constituent (POHC) designated. For
                                 sources burning hazardous wastes F020,
                               F021, F022, F023, F026, or F027, 99.9999%
                                       for each POHC designated.
------------------------------------------------------------------------
\1\ All emission levels are corrected to 7% O2, dry basis.
\2\ If there is an alkali by-pass stack associated with the kiln or in-
  line kiln raw mill, the combined particulate matter emissions from the
  kiln or in-line kiln raw mill and the alkali by-pass must be less than
  the particulate matter emissions standard.
\3\ Cement kilns that elect to comply with the carbon monoxide standard
  must demonstrate compliance with the hydrocarbon standard during the
  comprehensive performance test.
\4\ Measurement made in the by-pass sampling system of any kiln (e.g.,
  alkali by-pass of a preheater and/or precalciner kiln; midkiln
  sampling system of a long kiln).
\5\ Applicable only to newly-constructed cement kilns at greenfield
  sites (see discussion in Part Four, Section VII.D.9). 50 ppmv standard
  is a 30-day block average limit. Hydrocarbons reported as propane.
\6\ Hourly rolling average. Hydrocarbons are reported as propane.

2. What Are the Dioxin and Furan Standards?
    In today's rule, we establish a standard for new and existing 
cement kilns that limits dioxin/furan emissions to either 0.20 ng TEQ/
dscm; or 0.40 ng TEQ/dscm and temperature at the inlet to the 
particulate matter control device not to exceed 
400 deg.F.120 Our rationale for these standards is discussed 
below.
---------------------------------------------------------------------------

    \120\ The temperature limit applies at the inlet to a dry 
particulate matter control device that suspends particulate matter 
in the combustion gas stream (e.g., electrostatic precipitator, 
fabric filter) such that surface-catalyzed formation of dioxin/furan 
is enhanced. The temperature limit does not apply to a cyclone 
control device, for example.
---------------------------------------------------------------------------

    a. What Is the MACT Floor for Existing Sources? In the April 1996 
proposal, we identified floor control as either temperature control at 
the inlet to the particulate matter control device of less than 
418 deg.F, or achieving a specific level of dioxin/furan emissions 
based upon levels achievable using proper temperature control. (61 FR 
at 17391.) The proposed floor emission level was 0.20 ng TEQ/dscm, or 
temperature at the inlet to the electrostatic precipitator or fabric 
filter not to exceed 418 deg.F. In the May 1997 NODA, we identified an 
alternative data analysis method to identify floor control and the 
floor emission level. Floor control for dioxin/furan was defined as 
temperature control at the inlet to the electrostatic precipitator or 
fabric filter at 400 deg.F, which was based on further engineering 
evaluation of the emissions data and other available information. That 
analysis resulted in a floor emission level of 0.20 ng TEQ/dscm, or 
0.40 ng TEQ/dscm and temperature at the inlet to the electrostatic 
precipitator or fabric filter not to exceed 400 deg.F. (62 FR at 
24226.) The 0.40 ng TEQ/dscm standard is the level that all cement 
kilns, including data from nonhazardous waste burning cement kilns, are 
achieving when operating at the MACT floor control level or better. We 
considered a data set that included dioxin/furan emissions from 
nonhazardous waste burning cement kilns because these data are 
adequately representative of general dioxin/furan behavior and control 
in either type of kiln. The impacts of hazardous waste constituents 
(HAPs) on the emissions of those HAPs prevent us from expanding our 
database for other HAPs in a similar way.
    We conclude that the floor methodology discussed in the May 1997 
NODA is appropriate and we adopt this approach in today's final rule. 
We identified two technologies for control of dioxin/furan emissions 
from cement kilns in the May 1997 NODA. The first technology achieves 
low dioxin/furan emissions by quenching kiln gas temperatures at the 
exit of the kiln so that gas temperatures at the inlet to the 
particulate matter control device are below the temperature range of 
optimum dioxin/furan formation. For example, we are aware of several 
cement kilns that have recently added flue gas quenching units upstream 
of the particulate matter control device to reduce the inlet 
particulate matter control device temperature resulting in 
significantly reduced dioxin/furan levels.121 The other 
technology is activated carbon injected into the kiln exhaust gas. 
Since activated carbon injection is not currently used by any hazardous 
waste burning cement kilns, this technology was evaluated only as part 
of a beyond-the-floor analysis.
---------------------------------------------------------------------------

    \121\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards. Volume III: Selection of Proposed MACT Standards and 
Technologies'', July 1999. See Section 3.2.1.
---------------------------------------------------------------------------

    As discussed in the May 1997 NODA, specifying a temperature 
limitation of 400 deg.F or lower is appropriate for floor control 
because, from an engineering perspective, it is within the range of

[[Page 52876]]

reasonable values that could have been selected considering that: (1) 
The optimum temperature window for surface-catalyzed dioxin/furan 
formation is approximately 450-750 deg.F; and (2) temperature levels 
below 350 deg.F can cause dew point condensation problems resulting in 
particulate matter control device corrosion, filter cake cementing 
problems, increased dust handling problems, and reduced performance of 
the control device. (62 FR at 24226.)
    Several commenters disagreed with our selection of 400 deg.F as the 
particulate matter control device temperature limitation and stated 
that other higher temperature limitations were equally appropriate as 
MACT floor control. Based on these NODA comments, we considered 
selecting a temperature limitation of 450 deg.F, generally regarded to 
be the lower end of the temperature range of optimum dioxin/furan 
formation. However, available data indicate that dioxin/furan formation 
can be accelerated at kilns operating their particulate matter control 
device at temperatures between 400-450 deg.F. Data from several kilns 
show dioxin/furan emissions as high as 1.76 ng TEQ/dscm when operating 
in the range of 400-450 deg.F. Identifying a higher temperature limit 
such as 450 deg.F is not consistent with other sources achieving much 
lower emissions at 400 deg.F, and thus identifying a higher temperature 
limit would not be MACT floor control.
    Some commenters also state that EPA has failed to demonstrate that 
the best performing 12 percent of existing sources currently use 
temperature control to reduce dioxin/furan emissions, and therefore, 
temperature control is more appropriately considered in subsequent 
beyond-the-floor analyses. However, particulate matter control device 
operating temperatures associated with the emissions data used to 
establish the dioxin/furan standard are based on the maximum operating 
limits set during compliance certification testing required by the 
Boiler and Industrial Furnace rule. See 40 CFR 266.103(c)(1)(viii). As 
such, cement kilns currently must comply with these temperature limits 
on a continuous basis during day-to-day operations, and therefore, 
these temperature limits are properly assessed during an analysis of 
MACT floors.
    Several commenters also oppose consideration of dioxin/furan 
emissions data from nonhazardous waste burning cement kilns in 
establishing the floor standard. Commenters state that pooling the 
available emissions data from hazardous waste burning cement kiln with 
data from nonhazardous waste burning cement kilns to determine the MACT 
floor violates the separate category approach that EPA decided upon for 
the two classes of cement kilns. Notwithstanding our decision to divide 
the Portland cement manufacturing source category based on the kiln's 
hazardous waste burning status, we considered both hazardous waste 
burning cement kiln and nonhazardous waste burning cement kiln data 
together because both data sets are adequately representative of 
general dioxin/furan behavior and control in either type of kiln. This 
similarity is based on our engineering judgement that hazardous waste 
burning does not have an impact on dioxin/furan formation, dioxin/furan 
is formed post-combustion. Though the highest dioxin/furan emissions 
data point from MACT (i.e., operating control device less than 
400 deg.F) hazardous waste and nonhazardous waste burning cement kiln 
sources varies somewhat (0.28 vs 0.37 ng TEQ/dscm respectively), it is 
our judgment that additional emissions data, irrespective of hazardous 
waste burning status, would continue to point to a floor of within the 
range of 0.28 to 0.37 ng TEQ/dscm. This approach ensures that the floor 
levels for hazardous waste burning cement kilns are based on the 
maximum amount of relevant data, thereby ensuring that our judgment on 
what floor level is achievable is as comprehensive as possible.
    We estimate that approximately 70 percent of test condition data 
from hazardous waste burning cement kilns are currently emitting less 
than 0.40 ng TEQ/dscm (irrespective of the inlet temperature to the 
particulate matter control device). In addition, approximately 50 
percent of all test condition data are less than 0.20 ng TEQ/dscm. The 
national annualized compliance cost for cement kilns to reduce dioxin/
furan emissions to comply with the floor standard is $4.8 million for 
the entire hazardous waste burning cement industry and will reduce 
dioxin/furan emissions by 5.4 g TEQ/yr or 40 percent from current 
baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We considered in the April 1996 proposal and May 1997 NODA a 
beyond-the-floor standard of 0.20 ng TEQ/dscm based on activated carbon 
injection at a temperature of less than 400 deg.F. We continue to 
believe that a beyond-the-floor standard 0.20 ng TEQ/dscm based on 
activated carbon injection is the appropriate beyond-the-floor standard 
to evaluate given the risks posed by dioxin/furan emissions.
    Carbon injection is routinely effective at removing 99 percent of 
dioxin/furans for numerous municipal waste combustor and mixed waste 
incinerator applications and one hazardous waste incinerator 
application. However, currently no hazardous waste burning cement kilns 
use activated carbon injection for dioxin/furan removal. For cement 
kilns, we believe that it is conservative to assume only 95 percent is 
achievable given that the floor level is already low at 0.40 ng/dscm. 
As dioxin/furans decrease, activated carbon injection efficiency is 
expected to decrease. In addition, we assumed for cost-effectiveness 
calculations that cement kilns needing activated carbon injection to 
achieve the beyond-the-floor standard would install the activated 
carbon injection system after the normal particulate matter control 
device and add a new, smaller fabric filter to remove the injected 
carbon with the absorbed dioxin/furan and mercury.122 The 
costing approach addresses commenter's concerns that injected carbon 
may interfere with cement kiln dust recycling practices.
---------------------------------------------------------------------------

    \122\ We received many comments on the use of activated carbon 
injection as a beyond-the-floor control techniques at cement kilns. 
Since we do not adopt a beyond-the-floor standard based on activated 
carbon injection in the final rule, these comments and our responses 
to them are only discussed in our document that responds to public 
comments.
---------------------------------------------------------------------------

    The national incremental annualized compliance cost for the 
remaining cement kilns to meet this beyond-the-floor level, rather than 
comply with the floor controls, would be approximately $2.5 million for 
the entire hazardous waste burning cement industry and would provide an 
incremental reduction in dioxin/furan emissions nationally beyond the 
MACT floor controls of 3.7 g TEQ/yr. Based on these costs, 
approximately $0.66 million per g dioxin/furan removed, we determined 
that this dioxin/furan beyond-the-floor option for cement kilns is not 
justified. Therefore, we are not adopting a beyond-the-floor standard 
of 0.2 ng TEQ/dscm.
    We note that one possible explanation of high cost-effectiveness of 
the beyond-the-floor standard may be due to the significant reduction 
in national dioxin/furan emissions achieved over the past several years 
by hazardous waste burning cement kilns due to emissions improving 
modifications. The hazardous waste burning cement kiln national dioxin/
furan emissions estimate for 1997 decreased by nearly

[[Page 52877]]

97% since 1990, from 431 g TEQ/yr to 13.1 g TEQ/yr.123
---------------------------------------------------------------------------

    \123\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume V: Emission Estimates and Engineering Costs'', 
July 1999. See also 63 FR 17338, April 10, 1998.
---------------------------------------------------------------------------

    c. What Is the MACT Floor for New Sources? At proposal, we 
identified floor control for new sources as temperature control at the 
inlet to the particulate matter control device at 409 deg.F. The 
proposed floor emission level was 0.20 ng TEQ/dscm, or temperature at 
the inlet to the particulate matter control device not to exceed 
409 deg.F. In the May 1997 NODA, we identified an alternative data 
analysis method to identify floor control and the floor emission level. 
The May 1997 NODA dioxin/furan floor control for new sources was 
defined as temperature control at the inlet to the electrostatic 
precipitator or fabric filter at 400 deg.F, which was based on an 
engineering evaluation of the emissions data and other available 
information. That analysis resulted in a floor emission level of 0.20 
ng TEQ/dscm, or 0.40 ng TEQ/dscm and temperature at the inlet to the 
electrostatic precipitator or fabric filter not to exceed 400 deg.F. We 
continue to believe that the floor methodology is appropriate for new 
sources and we adopt this approach in this final rule.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
both the April 1996 proposal and May 1997 NODA, we proposed activated 
carbon injection as beyond-the-floor control and a beyond-the-floor 
standard of 0.20 ng TEQ/dscm for new sources. For reasons discussed 
above for existing sources, we conclude that it is also not cost-
effective for new cement kilns to achieve this level. Thus, we do not 
adopt a beyond-the-floor dioxin/furan standard for new cement kilns.
3. What Are the Mercury Standards?
    In today's rule, we establish a standard for existing and new 
cement kilns that limits mercury emissions to 120 and 56 g/
dscm, respectively. The rationale for these standards is discussed 
below.
    a. What Is the MACT Floor for Existing Sources? All cement kilns 
use either electrostatic precipitators or fabric filters for 
particulate matter control. However, since mercury is generally in the 
vapor form in and downstream of the combustion chamber, including the 
air pollution control device, electrostatic precipitators and fabric 
filters do not achieve good mercury control. Mercury emissions from 
cement kilns are currently regulated by the Boiler and Industrial 
Furnace rule, which establishes limits on the maximum feedrate of 
mercury in total feedstreams (e.g., hazardous waste, raw materials, 
coal). Thus, MACT floor control is based on hazardous waste feed 
control.
    In the April 1996 proposal, we identified floor control as 
hazardous waste feedrate control not to exceed a feedrate level of 110 
g/dscm, expressed as a maximum theoretical emission 
concentration, and proposed a floor standard of 130 g/dscm 
based on an analysis of data from all cement kilns with a hazardous 
waste mercury feedrate of this level or lower. (61 FR at 17393.) In May 
1997 NODA, we conducted a breakpoint analysis on low to high ranked 
mercury emissions data from sources floor control and established the 
floor level as the test condition average emission of the breakpoint 
source. The breakpoint analysis was intended to reflect an engineering-
based evaluation of the data so that the few cement kilns spiking 
mercury during compliance testing did not drive the floor standard to 
levels higher than the preponderance of the emissions data. We reasoned 
that sources with emissions higher than the breakpoint source were not 
controlling the hazardous waste feedrate of mercury to levels 
representative of MACT. This analysis resulted in a MACT floor level of 
72 g/dscm. (62 FR at 24227.)
    For today's rule, in response to comments questioning our May 1997 
NODA approach, we use a revised engineering evaluation and data 
analysis method to establish the MACT floor for mercury. As discussed 
in greater detail in the methodology section previously, we use an 
aggregate feedrate approach to establish MACT floors for the three 
metal hazardous air pollutant groups and hydrochloric acid/chlorine 
gas. The aggregate feedrate approach first identifies a MACT floor 
feedrate level for mercury and then establishes the floor emission 
level as the highest emissions level achieved by any cement kilns using 
floor control or better. Using this approach, the resulting mercury 
floor emission level is 120 g/dscm.
    We received comments on several overarching issues including the 
appropriateness of considering feedrate control of mercury in hazardous 
waste as a MACT floor control technique and the specific procedure of 
identifying breakpoints in arrayed emissions data. These issues and our 
response to them are discussed in the floor methodology section in Part 
Four, Section V. In addition, we received comment on a special 
provision that would allow cement kilns (and lightweight aggregate 
kilns) to petition the Administrator for an alternative mercury 
standard for kilns with mercury concentrations in their mineral and 
related process raw materials that causes an exceedance of the emission 
standard. This issue and the alternative standard promulgated in the 
final rule is fully discussed in Part Five, Section X.A.
    We also received comments from the cement manufacturing industry 
indicating that cement kilns with in-line raw mills have unique design 
and operating procedures that necessitate the use of emission averaging 
when demonstrating compliance with the emission standards. These 
commenters stated that the mercury standard is not achievable without a 
procedure for kilns to emissions average. The commenters supported a 
provision allowing cement kilns with in-line raw mills to demonstrate 
compliance with the emission standards on a time-weighted average basis 
to account for different emission characteristics when the raw mill is 
active as opposed to when it is inactive. After fully considering 
comments received, we adopt an emission averaging provision in the 
final rule. This provision is fully discussed in Part Five, Section 
X.E.
    Several commenters expressed concern that the mercury emissions 
data base for cement kilns is comprised of normal data, that is, cement 
kilns did not spike mercury during RCRA compliance testing as they did 
for other metals and chlorine. Thus, commenters stated that an 
emissions variability factor should be added to a floor level derived 
directly from the emissions data to ensure that the floor emission 
level is being achieved in practice. As discussed in Section V.D.1 
above, we conclude that emissions variability is adequately accounted 
for by the MACT floor methodology finalized today.
    We estimate that 85 percent of cement kilns currently meet the 
floor level. The national annualized compliance cost for cement kilns 
to reduce mercury emissions to comply with the floor level is $1.1 
million for the entire hazardous waste burning cement industry and will 
reduce mercury emissions by 0.2 Mg/yr or 15 percent from current 
baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 NPRM, we proposed a beyond-the-floor 
standard of 50 g/dscm based on flue gas temperature reduction 
to 400  deg.F followed by activated carbon injection for mercury 
capture. (61 FR at 17394.) In the May 1997 NODA, we considered a 
beyond-the-floor standard of 30 g/dscm based on activated 
carbon

[[Page 52878]]

injection; however, an evaluation was not conducted to determine if 
such a level would be cost-effective. (62 FR at 24227.)
    In developing the final rule, we identified three techniques for 
control of mercury as a basis to evaluate a beyond-the-floor standard: 
(1) Activated carbon injection; (2) limiting the feed of mercury in the 
hazardous waste; and (3) limiting the feed of mercury in the raw 
materials. The results of each analysis are discussed below.
    i. Activated Carbon Injection. To investigate activated carbon 
injection, we applied a carbon injection capture efficiency of 80 
percent to the floor emission level of 120 g/dscm. Our basis 
for selecting a capture efficiency of 80 percent 124 is 
discussed in the support document.125 The resulting beyond-
the-floor emission level is 25 g/dscm.
---------------------------------------------------------------------------

    \124\ We received many comments on the use of activated carbon 
injection as a beyond-the-floor control technique at cement kilns. 
Since we do not adopt a beyond-the-floor standard based on activated 
carbon injection in the final rule, these comments and our responses 
to them are only discussed in our document that responds to public 
comments.
    \125\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies.'' July 1999.
---------------------------------------------------------------------------

    We then determined the cost of achieving this reduction to 
determine if a beyond-the-floor standard of 25 g/dscm would be 
appropriate. The national incremental annualized compliance cost for 
the remaining cement kilns to meet this beyond-the-floor level, rather 
than comply with the floor controls, would be approximately $11.1 
million for the entire hazardous waste burning cement kiln industry and 
would provide an incremental reduction in mercury emissions nationally 
beyond the MACT floor controls of 0.7 Mg/yr. Based on these costs of 
approximately $16 million per additional Mg of mercury removed, we 
conclude that this mercury beyond-the-floor option for cement kilns is 
not acceptably cost-effective nor otherwise justified. Therefore, we do 
not adopt this beyond-the-floor standard.
    ii. Limiting the Feedrate of Mercury in the Hazardous Waste. We 
also considered a beyond-the-floor standard of 50 g/dscm based 
on limiting the feedrate of mercury in the hazardous waste. An emission 
level of 50 g/dscm represents the practicable extent that 
additional feedrate control of mercury in hazardous waste (beyond 
feedrate control needed to achieve the floor emission level) can be 
used and still achieve modest emissions reductions. We investigated the 
cost of achieving this reduction to determine if this beyond-the-floor 
standard would be appropriate. The national incremental annualized 
compliance cost for cement kilns to meet a beyond-the-floor level of 50 
g/dscm, rather than comply with the floor controls, would be 
approximately $4.2 million for the entire hazardous waste burning 
cement kiln industry and would provide an incremental reduction in 
mercury emissions nationally beyond the MACT floor controls of 0.4 Mg/
yr. Based on these costs of approximately $10.9 million per additional 
Mg of mercury removed, we conclude that this mercury beyond-the-floor 
option for cement kilns is not warranted. Therefore, we did not adopt 
this mercury beyond-the-floor standard.
    iii. Limiting the Feedrate of Mercury in Raw Materials. Finally, we 
considered a beyond-the-floor standard based on limiting the feedrate 
of mercury in the raw materials. Cement manufacturing involves the 
heating of raw materials such as limestone, clay, shale, sand, and iron 
ore. Limestone, shale, and clay comprise the vast majority of raw 
material feed to the kiln, and these materials are typically mined at 
quarries nearby the cement kiln. Since feed materials can contain 
significant quantities of hazardous air pollutants, we considered 
establishing a beyond-the-floor standard based on limiting the feedrate 
of mercury in these raw materials. A source can achieve a reduction in 
mercury emissions by substituting a feed material containing lower 
levels of mercury for a primary raw material with higher mercury 
levels. For example, shale is the primary feed material used as a 
source of silica. Under this beyond-the-floor option, a source using a 
high mercury-containing shale could substitute a feed material lower in 
mercury such as a coal ash to achieve lower mercury emissions. This 
beyond-the-floor option appears to be less cost-effective compared to 
either of the options evaluated above, however. This conclusion is 
based on the fact that cement kilns are sited proximate to primary raw 
material supply and transporting large quantities of an alternative 
source of raw material(s) is likely to be cost-prohibitive, thereby 
making a beyond-the-floor standard not cost-effective. Therefore, we do 
not adopt this mercury beyond-the-floor standard.
    Thus, the promulgated mercury standard for existing hazardous waste 
burning cement kilns is the floor level of 120 g/dscm.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal, we identified floor control for new sources as hazardous 
waste mercury feedrate control not to exceed a feedrate level of 28 
g/dscm expressed as a maximum theoretical emission 
concentration. We proposed a floor level of 82 g/dscm. We 
discussed a floor emission level for new cement kilns in the May 1997 
NODA of 72 g/dscm, based on a floor feedrate control level of 
110 g/dscm.
    Today we identify floor control for new cement kilns as feedrate 
control of mercury in the hazardous waste, expressed as a maximum 
theoretical emission concentration, based on the single source with the 
best aggregate feedrate of mercury in hazardous waste. Using the 
aggregate feedrate approach to establish this floor level of control 
and the corresponding floor emission level, we identify a MACT floor 
emission level of 56 g/dscm for new hazardous waste burning 
cement kilns.126
---------------------------------------------------------------------------

    \126\ Given that the emission level is substantially higher than 
the feedrate level expressed as a maximum theoretical emission 
concentration, 56 vs 7 g/dscm, the contributions of mercury 
from raw materials and coal for the floor-setting source must be 
substantial.
---------------------------------------------------------------------------

    d. What Are Our Beyond-the-Floor Considerations for New Sources? At 
proposal, we based beyond-the-floor control for new cement kilns on 
activated carbon injection and proposed a standard of 50 g/
dscm. In the May 1997 NODA we considered a beyond-the-floor standard of 
30 g/dscm based on activated carbon injection as done for 
existing sources.
    We identified two techniques for control of mercury as a basis to 
evaluate a beyond-the-floor standard for new sources: (1) Activated 
carbon injection; and (2) limiting the feedrate of mercury in the 
hazardous waste. The results of each analysis are discussed below.
    i. Activated Carbon Injection. As discussed above, we conclude that 
flue gas temperature reduction to 400 deg.F followed by activated 
carbon injection to remove mercury is an appropriate beyond-the-floor 
control option for improved mercury control at cement kilns. Based on 
the MACT floor emission level of 56 g/dscm and assuming a 
carbon injection capture efficiency of 80 percent, we identified a 
beyond-the-floor emission level of 10 g/dscm. We then 
determined the cost of achieving this reduction to determine if a 
beyond-the-floor standard of 10 g/dscm would be appropriate. 
The incremental annualized compliance cost for one new large cement 
kiln to meet this beyond-the-floor level, rather than comply with floor 
controls, would be approximately $2.3 million and would provide an 
incremental reduction in mercury emissions beyond the MACT floor 
controls of approximately 0.17 Mg/yr. For a new small cement kiln, the

[[Page 52879]]

incremental annualized compliance cost would be approximately $0.9 
million and would provide an incremental reduction in mercury emissions 
beyond the MACT floor controls of approximately 0.04 Mg/yr. Based on 
these costs of approximately $13-22 million per additional Mg of 
mercury removed, we concluded that a beyond-the-floor standard of 10 
g/dscm is not justified due to the high cost of compliance and 
relatively small mercury emissions reductions.
    ii. Limiting the Feedrate of Mercury in Hazardous Waste. We also 
considered a beyond-the-floor standard based on limiting the feedrate 
of mercury in the hazardous waste. Considering that the floor emission 
level for new cement kilns is approximately half of the floor emission 
level for existing kilns (56 versus 120 g/dscm), we conclude 
that a mercury beyond-the-floor standard for cement kilns is not 
warranted. This conclusion is based on the limited incremental 
emissions reductions achieved 127 and because the cost-
effectiveness of beyond-the-floor controls for new cement kilns would 
be even higher than for existing sources, which we found unacceptable 
in paragraph (b) above. Therefore, we do not adopt a mercury beyond-
the-floor standard based on limiting feedrate of mercury in hazardous 
waste.
---------------------------------------------------------------------------

    \127\ Achieving substantial additional mercury emissions 
reductions by further controls on hazardous waste feedrate may be 
problematic because the mercury contribution from raw materials and 
coal represents an even larger proportion of the total mercury fed 
to the kiln.
---------------------------------------------------------------------------

    Thus, the promulgated mercury standard for new hazardous waste 
burning cement kilns is the floor emissions level of 56 g/
dscm.
4. What Are the Particulate Matter Standards?
    We establish standards for both existing and new cement kilns which 
limit particulate matter emissions to 0.15 kg/Mg dry 
feed.128 In addition, opacity cannot exceed 20 percent. We 
chose the particulate matter standard as a surrogate control for the 
metals antimony, cobalt, manganese, nickel, and selenium. We refer to 
these five metals as ``nonenumerated metals'' because standards 
specific to each metal have not been established. The rationale for 
adopting these standards is discussed below.
---------------------------------------------------------------------------

    \128\ Approximately equivalent to a particulate matter 
concentration of 0.03 gr/dscf (69 mg/dscm) as expressed in the April 
1996 NPRM and May 1997 NODA. The calculation is approximate due to 
the different types of cement kilns and their associated flow rates.
---------------------------------------------------------------------------

    a. What Is the MACT Floor for Existing Sources? In the April 1996 
proposal, we discussed particulate matter floor control based upon the 
performance of a fabric filter with an air-to-cloth ratio of 2.3 acfm/
f, 2 resulting in a nominal floor emission level of 0.065 
gr/dscf. However, we believed it more appropriate to establish the 
floor standard based on the cement kiln 1971 New Source Performance 
Standard. (See discussion in 61 FR at 17392.) The 1971 New Source 
Performance Standard is 0.15 kg/Mg dry feed (0.30 lb/ton of dry feed). 
(see 40 CFR 60.60.) Cement kilns currently achieve this standard with 
well-designed and properly operated electrostatic precipitators and 
fabric filters.
    In the May 1997 NODA, we considered two data analysis methods to 
identify the particulate matter floor emission level. The first method 
established and expressed the floor level equivalent to the existing 
New Source Performance Standard promulgated in 1971. We subsequently 
proposed and finalized this approach for nonhazardous waste burning 
cement kilns. See 63 FR at 14198-199 and 64 FR 31898, respectively. The 
second approach discussed expressed the New Source Performance Standard 
as a stack gas concentration limit, as opposed to a production-based 
emission limit format. The May 1997 reevaluation suggested that the 
1971 New Source Performance Standard was approximately equivalent to a 
particulate matter concentration of 0.03 gr/dscf (69 mg/
dscm).129 We indicated a preference for expressing the 
particulate matter standard on a concentration basis because we also 
proposed that sources would comply with the particulate matter standard 
with a particulate matter continuous emissions monitoring system.
---------------------------------------------------------------------------

    \129\ See USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999 for a discussion of the approximate 
equivalency.
---------------------------------------------------------------------------

    However, we now conclude that basing the floor on the 1971 New 
Source Performance Standard is the most appropriate approach. Cement 
kilns achieve the 1971 New Source Performance Standard with well-
designed and properly operated fabric filters and electrostatic 
precipitators. Since approximately 20% of hazardous waste burning 
cement kilns now are subject to the 1971 New Source Performance 
Standard, consideration of this existing federal regulation as a floor 
is appropriate because greater than 12% of existing sources are 
achieving it. The available emissions test data show a wide range of 
particulate matter results--some emissions data are well below while 
other data are at the 1971 New Source Performance Standard 
level.130 Even though the hazardous waste burning cement 
kiln particulate matter data span two orders of 
magnitude,131 we have limited data on design parameters of 
the particulate matter control device and could not identify a cause 
(i.e., differentiate among control equipment) for the wide range in 
particulate matter emissions. We thus believe that the variation 
reflects normal operating variability. Therefore, the MACT floor 
emission level for existing cement kilns is the 1971 New Source 
Performance Standard.
---------------------------------------------------------------------------

    \130\ The variation in the particulate matter data is consistent 
with data from nonhazardous waste burning cement kilns. We neither 
expect nor have any data indicating that waste-burning operations 
increase particulate matter emissions at a cement kiln. An estimated 
30% of existing nonhazardous waste burning cement kilns are subject 
to the requirements of the new Source Performance Standard for 
cement plants. The particulate matter data for these kilns also 
exhibit a wide range in measurements. (63 FR at 14198.)
    \131\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    The New Source Performance Standard at Sec. 60.62 also specifies 
that opacity must be monitored continuously and establishes an opacity 
standard of 20 percent as a measure to ensure compliance with the 
particulate matter standard. We are therefore also adopting this 
opacity standard for today's rule.132 We are adopting it for 
the final rule because: (1) We proposed to base the particulate matter 
standard for hazardous waste burning cement kilns on the New Source 
Performance Standard, and the opacity standard is an integral component 
of that standard; and (2) we proposed to base the MACT particulate 
matter standard for nonhazardous waste burning cement kilns on the New 
Source Performance Standard and explicitly identified both the 
particulate emission and opacity components of the standard. Hazardous 
waste burning cement kiln stakeholders have commented on both the 
nonhazardous waste and hazardous waste cement kiln proposed rules and 
suggest that there is little or no difference in emissions from the two 
classes of kilns and that they should be regulated the same. Although 
we do not agree that emissions of all hazardous pollutants are the same 
for both classes of kilns and should be regulated the same, we agree 
that particulate

[[Page 52880]]

emissions are comprised largely of entrained raw material and are not 
significantly affected by burning hazardous waste. Thus, we concur that 
the standard for particulate matter should be the same for both classes 
of sources and are therefore adopting the New Source Performance 
Standard opacity standard for the final rule.133 In the NPRM 
and the May 1997 NODA, we proposed to express the particulate matter 
standard on a concentration basis rather than express it as the same 
format as the 1971 New Source Performance Standard, which is a 
production-based emission limit format. However, because we are not yet 
requiring sources to document compliance with the particulate matter 
standard by using a particulate matter continuous emissions monitoring 
system in this final rule 134, we establish and express the 
floor emission level equivalent to the 1971 New Source Performance 
Standard. Thus, the particulate matter floor is 0.15 kg/Mg dry feed 
based on the performance of a well-designed and operated fabric filter 
or electrostatic precipitator.
---------------------------------------------------------------------------

    \132\ Given that we adopt the New Source Performance Standard 
for particulate matter and opacity for the MACT standards for 
hazardous waste burning cement kilns, we exempt these sources from 
the New Source Performance Standard to avoid duplicative regulation. 
See Sec. 63.1204(h).
    \133\ We are not adopting the opacity standard component of the 
New Source Performance Standard for hazardous waste burning 
lightweight aggregate kilns, however. This is because that opacity 
standard (see Sec. 60.732) is a measure to ensure compliance with 
the particulate emissions component of that standard, which is 
substantially higher than the MACT standard that we promulgate 
today. Thus, the NSPS opacity standard for lightweight aggregate 
kilns would not be a useful measure of compliance with today's 
particulate matter standard for lightweight aggregate kilns.
    \134\ We anticipate rulemaking on a particulate matter 
continuous emissions monitoring system requirement for hazardous 
waste combustors in the near future. Under this rulemaking, 
combustors would be required to document compliance with national 
emission standards by complying with continuous emissions monitoring 
system-based particulate matter levels that are being achieved by 
sources equipped with MACT controls. See Part Five, Section VII.C. 
for details.
---------------------------------------------------------------------------

    Several commenters expressed concern in their comments to the NPRM 
that the Agency identified separate, different MACT pools and 
associated MACT controls for particulate matter, semivolatile metals, 
and low volatile metals, even though all three are controlled, at least 
in part, by a particulate matter control device. Commenters stated that 
our approach is likely to result in three different design 
specifications. We agree with the need to use the same pool for 
particulate matter, semivolatile metals, and low volatile metals and 
used the same initial MACT pool to establish the floor levels for these 
pollutants. See Part Four, Section V for a detailed discussion of our 
floor methodology.
    We estimate that over 60 percent of cement kilns currently meet the 
floor emission level. The national annualized compliance cost for 
cement kilns to reduce particulate matter emissions to comply with the 
floor level is $6.2 million for the entire hazardous waste burning 
cement industry and will reduce nonenumerated metals and particulate 
matter emissions by 1.1 Mg/yr and 873 Mg/yr, respectively, or over 30 
percent from current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the proposal and May 1997 NODA, we considered a beyond-the-
floor level of 34 mg/dscm (0.015 gr/dscf) based on improved particulate 
matter control. However, after examining the costs of such control and 
the relatively low incremental reductions in air emissions that would 
result, we determined that a beyond-the-floor standard would not likely 
be cost-effective. (61 FR at 17393.)
    Several commenters support a beyond-the-floor option for 
particulate matter because some cement kilns are readily achieving 
particulate matter levels well below the floor emission level based on 
the New Source Performance Standard. Other commenters oppose a beyond-
the-floor option for cement kilns because of the high costs and 
anticipated poor cost-effectiveness. In the final rule, we evaluated a 
beyond-the-floor emission level for existing cement kilns to determine 
if such a level would be appropriate.
    Improved particulate matter control for existing cement kilns would 
require the use of high efficiency electrostatic precipitators and 
fabric filters. These may include fabric filters with low air-to-cloth 
ratios, high performance fabrics, electrostatic precipitators with 
large specific collection areas, and advanced control systems. 
Currently, the majority of hazardous waste burning cement kilns use 
electrostatic precipitators for particulate matter control and usually 
achieve removal efficiencies greater than 99.8%. Cement kilns can meet 
the MACT floor with well designed and properly operated particulate 
matter control equipment that for many kilns may require only minor 
system upgrades from their current systems. A beyond-the-floor 
standard, however, would likely involve more than a minor system 
upgrade, and may require new control equipment or retrofitting a 
baghouse with new higher performance fabric materials. The total 
annualized costs associated with such major system upgrades would be 
significant, while only achieving modest incremental emissions 
reductions in particulate matter and nonenumerated metals.
    In the final rule, we considered a beyond-the-floor level of 34 mg/
dscm, approximately one-half the New Source Performance Standard, for 
existing cement kilns based on improved particulate matter control. For 
analysis purposes, improved particulate matter control entails the use 
of higher quality fabric filter bag material. We then determined the 
cost of achieving this level of particulate matter, with corresponding 
reductions in the nonenumerated metals for which particulate matter is 
a surrogate, to determine if this beyond-the-floor level would be 
appropriate. The national incremental annualized compliance cost for 
cement kilns to meet this beyond-the-floor level, rather than comply 
with the floor controls, would be approximately $7.4 million for the 
entire hazardous waste burning cement kiln industry and would provide 
an incremental reduction in nonenumerated metals emissions nationally 
beyond the MACT floor controls of 0.7 Mg/yr. Based on these costs of 
approximately $10.7 million per additional Mg of nonenumerated metals 
emissions removed, we conclude that this beyond-the-floor option for 
cement kilns is not acceptably cost-effective nor otherwise justified. 
Therefore, we do not adopt this beyond-the-floor standard. The 
promulgated particulate matter standard for existing hazardous waste 
burning cement kilns is the floor emission level of 0.15 kg/Mg dry feed 
and opacity not to exceed 20 percent.
    c. What Is the MACT Floor for New Sources? In the proposal, we 
defined floor control based on the performance of a fabric filter with 
an air-to-cloth ratio of less than 1.8 acfm/ft2. As discussed for 
existing sources, we proposed the floor level based on the existing 
cement kiln New Source Performance Standard. 61 FR at 17400. In the May 
1997 NODA, we again considered basing the floor emission level on the 
New Source Performance Standard and solicited comment on the two 
alternatives to express the standard identical to those discussed above 
for existing cement kilns. (62 FR at 24228.)
    All cement kilns use fabric filters and electrostatic precipitators 
to control particulate matter. As discussed earlier, we have limited 
detailed information on the design and operation characteristics of 
existing control equipment currently used by cement kilns. As a result, 
we are unable to identify a specific design or technology that can 
consistently achieve lower emission levels than the controls used by 
cement kilns achieving the New Source Performance Standard. Cement 
kilns meet the New Source Performance Standard with well-

[[Page 52881]]

designed and properly operated fabric filters and electrostatic 
precipitators. Thus, floor control for new cement kilns is also a well-
designed and properly operated fabric filter and electrostatic 
precipitator. As discussed for existing sources, we conclude that 
expressing the floor based on the New Source Performance Standards is 
appropriate for the final rule. Therefore, the MACT floor level for new 
cement kilns is 0.15 kg/Mg dry feed and opacity not to exceed 20 
percent.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 NPRM and May 1997 NODA, we considered a beyond-the-floor 
standard based on improved particulate matter control to be consistent 
with existing sources. However, we proposed that such a beyond-the-
floor level was not likely cost-effective.
    As discussed for existing sources, we considered a beyond-the-floor 
level of 34 mg/dscm, approximately one-half the New Source Performance 
Standard, for new cement kilns based on improved particulate matter 
control. For analysis purposes, improved particulate matter control 
entails the use of higher quality fabric filter bag material. We then 
determined the cost of achieving this level of particulate matter, with 
corresponding reductions in the nonenumerated metals for which 
particulate matter is a surrogate, to determine if this beyond-the-
floor level would be appropriate. The incremental annualized compliance 
cost for one new large cement kiln to meet this beyond-the-floor level, 
rather than comply with floor controls, would be approximately $309,000 
and would provide an incremental reduction in nonenumerated metals 
emissions of approximately 0.18 Mg/yr.135 For a new small 
cement kiln, the incremental annualized compliance cost would be 
approximately $120,000 and would provide an incremental reduction in 
nonenumerated metals emissions of approximately 0.04 Mg/yr. Based on 
these costs of approximately $1.7-3.0 million per additional Mg of 
nonenumerated metals removed, we conclude that a beyond-the-floor 
standard of 0.015 gr/dscf is not justified due to the high cost of 
compliance and relatively small nonenumerated metals emission 
reductions. Thus, the particulate matter standard for new cement kilns 
is the floor level of 0.15 kg/Mg dry feed and opacity not to exceed 20 
percent.
---------------------------------------------------------------------------

    \135\ Based on the data available, the average emissions in sum 
of the five nonenumerated metals from cement kilns using MACT 
particulate matter control is approximately 80 g/dscm. To 
estimate emission reductions of the nonenumerated metals, we assume 
a linear relationship between a reduction in particulate matter and 
these metals.
---------------------------------------------------------------------------

5. What Are the Semivolatile Metals Standards?
    Today's rule establishes standards for existing and new cement 
kilns that limit semivolatile metals emissions to 240 and 180 
g/dscm, respectively. The rationale for these standards is 
discussed below.
    a. What Is the MACT Floor for Existing Sources? In the April 1996 
proposal, we defined floor control as a fabric filter with an air-to-
cloth ratio less than 2.1 acfm/ft 2 and a hazardous waste 
feedrate level of 84,000 g/dscm, expressed as a maximum 
theoretical emission concentration. The proposed floor emission level 
was 57 g/dscm, based on the level a source with properly 
designed and operated floor technology could achieve. In the proposed 
rule, we also solicited comment on an alternative floor approach 
whereby ``equivalent technology'' to MACT control is identified and 
evaluated. This approach resulted in an emission level of 160 
g/dscm (See 61 FR at 17395.) In the May 1997 NODA, we 
discussed a floor methodology where we used a breakpoint analysis to 
identify sources that were not using floor control with respect either 
to semivolatile metals hazardous waste feedrate or emissions control. 
Under this approach, we ranked semivolatile metals emissions data from 
sources that were using MACT floor particulate matter control, i.e., 
sources achieving the New Source Performance Standard or better. We 
identified the floor level as the test condition average associated 
with the breakpoint source. Thus, sources with atypically high 
emissions because of high semivolatile metals feedrates or poor 
semivolatile metals control even though they appeared to be using floor 
control for particulate matter were screened from the pool of sources 
used to define the floor emission level. Based on this analysis, we 
identified a floor level in the May 1997 NODA of 670 g/dscm. 
(See 62 FR at 24228.)
    As discussed previously in the methodology section, we use a 
revised engineering evaluation and data analysis method to establish 
the MACT floor for semivolatile metals based on the same underlying 
data previously noticed for comment. The aggregate feedrate approach, 
in conjunction with floor control for particulate matter, identified a 
semivolatile metals floor emission level of 650 g/dscm.
    In addition, several commenters stated strongly that the feedrate 
of semivolatile metals in hazardous waste cannot be considered MACT 
floor control in conjunction with particulate matter control. These 
commenters believe that floor control for semivolatile metals is 
control of particulate matter only. We disagree with these commenters 
for reasons we discuss in Part Four, Section V of the preamble, mainly 
that feedrate is currently control for hazardous waste combustors under 
RCRA regulations, and conclude that control of the feedrate of 
semivolatile metals in hazardous waste is floor control, in conjunction 
with particulate matter control.
    We estimate that approximately 60 percent of cement kilns currently 
meet this floor level. The national annualized compliance cost for 
cement kilns to reduce semivolatile metal emissions to comply with the 
floor level is $1.3 million for the entire hazardous waste burning 
cement industry and will reduce semivolatile metal emissions by 19.5 
Mg/yr or 65 percent from current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the proposal, we considered a beyond-the-floor standard for 
semivolatile metals based on improved particulate matter control below 
the New Source Performance Standard. However, we concluded that a 
beyond-the-floor standard would not be cost-effective, given that the 
semivolatile metal floor level of 57 g/dscm alone resulted in 
an estimated 94 percent semivolatile metal reduction in emissions. (see 
61 FR at 17396.) In the May 1997 NODA, we considered a lower 
particulate matter emissions level of 0.015 gr/dscf, based on improved 
particulate matter control, as a beyond-the-floor standard to further 
reduce semivolatile and low volatile metals. Even though we did not 
quantify cost-effectiveness values, we expressed concern that a beyond-
the-floor standard would not likely be cost-effective. (see 62 FR at 
24229.)
    Commenters believed there were several control techniques that 
should be considered, therefore, we identified three potential beyond-
the-floor control techniques in developing the final rule: (1) Limiting 
the feedrate of semivolatile metals in hazardous waste; (2) improved 
particulate matter control; and (3) limiting the feedrate of 
semivolatile metals in raw materials. We conclude that a beyond-the-
floor standard is warranted based on limiting the feedrate of 
semivolatile metals in hazardous waste. The results of each analysis 
are discussed below.
    i. Limiting the Feedrate of Semivolatile Metals in Hazardous Waste. 
Under this approach, we selected a beyond-the-floor emission level of 
240

[[Page 52882]]

g/dscm from among the range of possible levels that reflect 
improved feedrate control. This emission level represents a significant 
increment of emission reduction from the floor of 650 g/dscm, 
it is within the range of levels that are likely to be reasonably 
achievable using feedrate control, and it is consistent with the 
incinerator standard thereby advancing a potential policy objective of 
essentially common standards among combustors of hazardous waste.
    The national incremental annualized compliance cost for the 
remaining cement kilns to meet this beyond-the-floor level, rather than 
comply with the floor controls, would be approximately $2.7 million for 
the entire hazardous waste burning cement kiln industry and would 
provide an incremental reduction, beyond emissions at the MACT floor, 
in semivolatile metal emissions nationally of 5.5 Mg/yr. The cost-
effectiveness of this standard would be approximately $500,000 per 
additional Mg of semivolatile metals removed. Notwithstanding the 
relatively poor cost-effectiveness of this standard on a dollar per Mg 
removed basis, we conclude that additional beyond-the-floor control of 
the feedrate of semivolatile metals in hazardous waste to achieve an 
emission level of 240 g/dscm is warranted because this 
standard would reduce lead and cadmium emissions which are particularly 
toxic hazardous air pollutants. See Health Human Effects discussion in 
USEPA, ``Technical Background Document for HWC MACT Standards: Health 
and Ecological Risk Assessment'', July 1999. Further, approximately 90% 
of the lead and cadmium fed to the cement kiln is from the hazardous 
waste,136 not the raw material (about 9%) or coal (about 
1%). We are willing to accept a more marginal cost-effectiveness to 
ensure that hazardous waste combustion sources are using the best 
controls for pollutants introduced almost exclusively for the burning 
of hazardous waste. We do so to provide a strong incentive for waste 
minimization of lead and cadmium sent for combustion. By providing 
stringent limits, we can help assure that hazardous waste with lead 
does not otherwise move from better controlled units in other 
subcategories to units in this subcategory because of a lesser degree 
of control. Moreover, this beyond-the-floor semivolatile metal standard 
supports our Children's Health Initiative in that lead emissions, which 
are of highest significance to children's health, will be reduced by 
another 20-25 percent from today's baseline. As part of this 
initiative, we are committed to reducing lead emissions wherever and 
whenever possible. Finally, this beyond-the-floor standard is 
consistent with European Union standards for hazardous waste 
incinerators of approximately 200 g/dscm for lead and cadmium 
combined. For all these reasons, we accept the cost-effectiveness of 
this level of feedrate control and adopt a beyond-the-floor standard of 
240 g/dscm for existing cement kilns.
---------------------------------------------------------------------------

    \136\ USEPA, ``Final Technical Support document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies'', July 1999.
---------------------------------------------------------------------------

    Additionally, we received comments shortly before promulgation from 
the cement kiln industry that expressed their achievability and 
economic concerns with a beyond-the-floor standard in the range of 240 
g/dscm based on limiting the feedrate of semivolatile metals 
in the hazardous waste. We considered their comments in adopting the 
240 g/dscm beyond-the-floor standard and included a copy of 
their November 18, 1998 presentation to the Office of Management and 
Budget in the docket along with our responses to their concerns, many 
of which are addressed above.
    ii. Improved Particulate Matter Control. We also evaluated improved 
particulate matter control as a beyond-the-floor control option for 
improved semivolatile metals control. Cadmium and lead are volatile at 
the high temperatures within the cement kiln itself, but typically 
condense onto the fine particulate at control device temperatures, 
where they are collected. As a result, control of semivolatile metals 
emissions is closely associated with particulate matter control. 
Examples of improved particulate matter control include the use of more 
expensive fabric filter bags, optimizing the design and operation 
features of the existing control equipment, and the addition to or the 
replacement of control equipment with a new fabric filter.
    We evaluated the costs to achieve a beyond-the-floor emission level 
of 240 g/dscm based on improved particulate matter control. 
The national incremental annualized compliance cost for cement kilns to 
meet this beyond-the-floor level, rather the floor level, would be 
approximately $4.1 million for the entire hazardous waste burning 
cement kiln industry and would provide an incremental reduction in 
semivolatile metal emissions beyond the MACT floor controls of 5.5 Mg/
yr. Because this beyond-the-floor control option would have a cost-
effectiveness of approximately $800,00 per additional Mg of 
semivolatile metal removed, contrasted to a cost-effectiveness of 
approximately $500,000 using hazardous waste feedrate control and 
remove an identical amount of semivolatile metals, we conclude that 
basing the beyond-the-floor standard on improved particulate matter 
control is not warranted.
    iii. Limiting the Feedrate of Semivolatile Metals in Raw Materials. 
A source can achieve a reduction in semivolatile metal emissions by 
substituting a feed material containing lower levels of lead and/or 
cadmium for a primary raw material with higher levels of these metals. 
We expect this beyond-the-floor option to be less cost-effective 
compared to either of the options evaluated above. Cement kilns are 
sited proximate to primary raw material supply and transporting large 
quantities of an alternative source of raw material(s) is likely to be 
cost-prohibitive. Therefore, we are not adopting a semivolatile metal 
beyond-the-floor standard based on limiting the feedrate of 
semivolatile metals in raw materials.137
---------------------------------------------------------------------------

    \137\ We, however, reject the proposition in comments that we 
are without legal authority to regulate HAPs in raw materials 
processed in cement kilns based on legislative history to the 1990 
amendments. This legislative history is not reflected in the 
statutory text, which unambiguously gives us that authority.
---------------------------------------------------------------------------

    Thus, the promulgated semivolatile metals standard for existing 
hazardous waste burning cement kilns is a beyond-the-floor standard of 
240 g/dscm based on limiting the feedrate of semivolatile 
metals in the hazardous waste.
    c. What Is the MACT Floor for New Sources? In the proposal, we 
defined floor control as a fabric filter with an air-to-cloth ratio 
less than 2.1 acfm/ft 2 and a hazardous waste feedrate level 
of 36,000 g/dscm, expressed as a maximum theoretical emission 
concentration. The proposed floor emission level for new cement kilns 
was 55 g/dscm. (See 61 FR at 17400.) In the May 1997 NODA, we 
concluded that the floor control and emission level for existing 
sources for semivolatile metals also would be appropriate for new 
sources. Floor control was based on a combination of good particulate 
matter control and limiting hazardous waste feedrate of semivolatile 
metals. We used a breakpoint analysis of the semivolatile metal 
emissions data to exclude sources achieving substantially poorer 
semivolatile metal control than the majority of sources because of 
atypically high semivolatile metals feedrates or poor emission control. 
We established the floor level at the test condition average of the 
breakpoint source: 670 g/dscm. (See 62 FR at 24229.)
    As discussed above for existing sources, we developed the final 
rule

[[Page 52883]]

using the aggregate feedrate approach to identify MACT floors for the 
metals. See Methodology Section for detailed discussion of aggregate 
feedrate approach. Using this approach, we establish the semivolatile 
metal floor emission level for new sources at 180 g/dscm.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 NPRM and May 1997 NODA, we considered a semivolatile 
metal beyond-the-floor emission level for new sources, but determined 
that it would not be cost-effective.
    For the final rule, we do not consider a beyond-the-floor level for 
new cement kilns because the MACT floor for new cement kilns is already 
lower than the beyond-the-floor emission standard for existing sources. 
As a result, a beyond-the-floor standard for new cement kilns is not 
warranted due to the likely significant costs of control and the 
minimal incremental emissions reductions. In addition, our policy goal 
of state of the art control of lead is achieved at the floor standard 
for new sources. We, therefore, adopt a semivolatile metal floor 
standard of 180 g/dscm for new hazardous waste burning cement 
kilns.
6. What Are the Low Volatile Metals Standards?
    We establish standards for existing and new cement kilns in today's 
rule that limit low volatile metal emissions to 56 and 54 g/
dscm, respectively. The rationale for these standards is discussed 
below.
    a. What Is the MACT Floor tor Existing Sources? In the April 1996 
NPRM, we defined floor control as either: (1) A fabric filter with an 
air-to-cloth ratio less than 2.3 acfm/ft \2\ and a hazardous waste 
feedrate level of 140,000 g/dscm, expressed as a maximum 
theoretical emission concentration; or (2) an electrostatic 
precipitator with a specific collection area of 350 ft \2\/kacfm and 
the same hazardous waste feedrate level of 140,000 g/dscm. The 
proposed floor level was 130 g/dscm. (See 61 FR at 17396.) In 
the May 1997 NODA, we used a breakpoint analysis to identify sources 
that were not using floor control with respect either to low volatile 
metals hazardous waste feedrate or emissions control. Under this 
approach, we ranked low volatile metals emissions data from sources 
that were achieving the particulate matter floor of 69 mg/dscm or 
better. We identified the floor level as the test condition average 
associated with the breakpoint source. Thus, sources with atypically 
high emissions because of high low volatile metals feedrates or poor 
low volatile metals control, even though they were using floor control 
for particulate matter, were screened from the pool of sources used to 
define the floor emission level. The May 1977 NODA MACT floor level was 
63 g/dscm. (See 62 FR at 24229.)
    We received limited comments in response to the NPRM and May 1997 
NODA concerning the low volatile metals floor standard. We received 
comments, however, on several overarching issues including the 
appropriateness of considering feedrate control of metals including low 
volatile metals in hazardous waste as a MACT floor control technique 
and the specific procedure of identifying breakpoints in arrayed 
emissions data. These issues and our responses to them are discussed in 
the floor methodology section in Part Four, Section V.
    Today we use a revised engineering evaluation and data analysis 
method to establish the MACT floor for low volatile metals on the same 
underlying data previously noticed for comment. As explained earlier, 
the aggregate feedrate approach, in conjunction with floor control for 
particulate matter, replaces the breakpoint analysis for metals and 
results in a low volatile metal floor emission level of 56 g/
dscm.
    We estimate that over 76 percent of cement kilns in our data base 
meet the floor level. The national annualized compliance cost for 
cement kilns to reduce low volatile metal emissions to comply with the 
floor level is $0.8 million for the entire hazardous waste burning 
cement industry, and will reduce low volatile metal emissions by 0.2 
Mg/yr or approximately 25 percent from current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the proposal, we considered a beyond-the-floor standard for 
low volatile metals based on improved particulate matter control. 
However, we concluded that a beyond-the-floor standard would not likely 
be cost-effective based on the limited emissions reductions of low 
volatility metals. In the May 1997 NODA, we considered a lower 
particulate matter emissions level, based on improved particulate 
matter control, as a beyond-the-floor standard with corresponding 
beyond-the-floor reductions in low volatile and semivolatile metals. 
Even though we did not quantify cost-effectiveness values, we expressed 
concern that a beyond-the-floor standard would not likely be cost-
effective. (62 FR at 24229.)
    For today's final rule, we identified three potential beyond-the-
floor techniques for control of low volatile metals: (1) Improved 
particulate matter control; (2) limiting the feedrate of low volatile 
metals in the hazardous waste; and (3) limiting the feedrate of low 
volatile metals in the raw materials. We discuss the results of our 
analysis of each option below.
    Improved Particulate Matter Control. Our judgment is that a beyond-
the-floor standard based on improved particulate matter control would 
be less cost-effective than a beyond-the-floor standard based on 
limiting the feedrate of low volatile metals in the hazardous waste. 
First, our data show that all cement kilns are already achieving 
greater than a 99% system removal efficiency for low volatile metals, 
with most attaining 99.99% removal. Thus, equipment retrofit costs for 
improved control would be significant and result in only a small 
increment in reduction of emissions. Our beyond-the-floor analysis for 
semivolatile metals supports this conclusion. There, the semivolatile 
metals analysis showed that the beyond-the-floor option based on 
limiting the feedrate of semivolatile metals was approximately 30% more 
cost-effective than a beyond-the-floor option based on improved 
particulate matter control. We believe the low volatile metals would 
require similar particulate matter control device retrofits at cement 
kilns as for semivolatile metals. However, the total emissions 
reduction achieved would be less because hazardous waste burning cement 
kilns emit less low volatile metals than semivolatile metals. We do not 
have any of the serious concerns present for semivolatile metals that 
suggest we should accept a more marginal cost-effectiveness. Thus, we 
conclude that a beyond-the-floor standard for low volatile metals based 
on improved particulate matter control is not warranted.
    Limiting the Feedrate of Low Volatile Metals in the Hazardous 
Waste. We also considered a beyond-the-floor standard of 40 g/
dscm for low volatile metals based on additional feedrate control of 
low volatile metals in the hazardous waste. This would reduce the floor 
emission level by approximately 30 percent. Our investigation shows 
that this beyond-the-floor option would achieve an incremental 
reduction in low volatile metals of only 0.1 Mg/yr. Given that this 
beyond-the-floor level would not achieve appreciable emissions 
reductions, we conclude that cost-effectiveness considerations would 
likely come into play suggesting that this beyond-the-floor standard is 
not warranted.

[[Page 52884]]

    Limiting the Feedrate of Low Volatile Metals in the Raw Materials. 
Sources can achieve a reduction in low volatile metal emissions by 
substituting a feed material containing lower levels of arsenic, 
beryllium, and/or chromium for a primary raw material with higher 
levels of these metals. We believe that this beyond-the-floor option 
would be even less cost-effective than either of the options evaluated 
above, however. Cement kilns are sited proximate to primary raw 
material supply and transporting large quantities of an alternative 
source of raw material(s) is likely to be cost-prohibitive. Therefore, 
we do not adopt a low volatile metal beyond-the-floor standard based on 
limiting the feedrate of low volatile metals in raw materials.
    For the reasons discussed above, we do not adopt a beyond-the-floor 
level for low volatile metals and establish the emission standard for 
existing hazardous waste burning cement kilns at 56 g/dscm.
    c. What Is the MACT Floor for New Sources? In the proposal, we 
defined floor control as a fabric filter with an air-to-cloth ratio 
less than 2.3 acfm/ft2 and a hazardous waste feedrate 
control level of 25,000 g/dscm, expressed as a maximum 
theoretical emission concentration. The proposed floor for new cement 
kilns was 44 g/dscm. (61 FR at 17400.) In the May 1997 NODA, 
we concluded that the floor control and emission level for existing 
sources for low volatile metals would also be appropriate for new 
sources. Floor control was based on a combination of good particulate 
matter control and limiting hazardous waste feedrate of low volatile 
metals. We used a breakpoint analysis of the low volatile metal 
emissions data to exclude sources achieving substantially poorer low 
volatile metal control than the majority of sources. We established the 
floor level at the test condition average of the breakpoint source. The 
NODA floor was 63 g/dscm. (62 FR at 24230.)
    As discussed above for existing sources, in developing the final 
rule we use the aggregate feedrate approach to identify MACT floors for 
the metals and hydrochloric acid/chlorine gas in combination with MACT 
floor control for particulate matter. Based on the low volatile metal 
feedrate in hazardous waste from the single best performing cement kiln 
using floor control for particulate matter, the MACT floor for new 
hazardous waste burning cement kilns is 54 g/dscm.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the proposal and May 1997 NODA, we considered a low volatile metal 
beyond-the-floor level for new sources, but determined it would not be 
cost effective. For reasons similar to those discussed for existing 
sources, we do not believe that a beyond-the-floor standard is 
warranted for new cement kilns due to the high expected compliance cost 
and relatively low reductions in emissions of low volatile metals. 
Therefore, we adopt a low volatile metals standard of 54 g/
dscm for new hazardous waste burning cement kilns.
7. What Are the Hydrochloric Acid and Chlorine Gas Standards?
    In today's rule, we establish standards for existing and new cement 
kilns that limit hydrochloric acid and chlorine gas emissions to 130 
and 86 ppmv, respectively. The rationale for these standards is 
discussed below.
    a. What Is the MACT Floor for Existing Sources? In the proposal, we 
identified floor control for hydrochloric acid/chlorine gas as feedrate 
control of chlorine in the hazardous waste and proposed a floor 
standard of 630 ppmv. (61 FR at 17396.) In the May 1997 NODA, we used a 
data analysis method similar to that at proposal and discussed a floor 
emission level of 120 ppmv. (62 FR at 24230.)
    Some commenters to the May 1997 NODA expressed concern that cement 
kilns may not be able to meet the hydrochloric acid/chlorine gas 
standard while making low alkali cement. Commenters noted that chlorine 
is sometimes added specifically to volatilize potassium and sodium 
compounds that must be removed to produce low alkali cement. One 
commenter manufacturing a low alkali cement submitted data showing a 
large range in hydrochloric acid/chlorine gas emissions while operating 
under varying conditions and production requirements. This commenter 
stated that they may not be able to meet the NODA hydrochloric acid/
chlorine gas standard of 120 ppmv while making low alkali cement. We 
conclude, however, that the data they submitted do not adequately 
support this ultimate conclusion. The commenter's emissions data range 
from 6 ppmv to 83 ppmv while operating under RCRA compliance testing 
conditions. These emission levels are well below the final standard of 
130 ppmv, and the expected operational range in this rule is 70% of the 
standard. We conclude that the hydrochloric acid/chlorine gas standard 
of 130 ppmv finalized today is readily achievable by all cement kilns 
irrespective of the type of cement manufactured.
    For today's rule, we use a revised engineering evaluation and data 
analysis method to establish the MACT floor for hydrochloric acid and 
chlorine gas on the same underlying data previously noticed for 
comment. Using the aggregate feedrate approach discussed previously, we 
establish a hydrochloric acid/chlorine gas floor emission level of 130 
ppmv.
    We estimate that approximately 88 percent of cement kilns in our 
data base currently meet the floor level. The national annualized 
compliance cost for cement kilns to reduce hydrochloric acid/chlorine 
gas emissions to comply with the floor level is $1.4 million for the 
entire hazardous waste burning cement industry and will reduce 
hydrochloric acid/chlorine gas emissions by 383 Mg/yr or 12 percent 
from current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the proposal, we defined beyond-the-floor control as wet 
scrubbing with a 99 percent removal efficiency, but determined that a 
beyond-the-floor standard would not be cost-effective. (61 FR at 
17397.) In the May 1997 NODA, we identified a more stringent floor 
standard and therefore reasoned that a beyond-the-floor standard based 
on wet scrubbing would likely also not be cost-effective. (62 FR at 
24230.)
    For today's rule, we identified three potential beyond-the-floor 
techniques for control of hydrochloric acid/chlorine gas emissions: (1) 
Scrubbing; (2) limiting the feedrate of chlorine in the hazardous 
waste; and (3) limiting the feedrate of chlorine in the raw materials. 
We discuss our analysis of each option below.
    Scrubbing. We continue to believe that a beyond-the-floor standard 
based on dry or wet scrubbing is not likely to be cost-effective. 
Cement kilns achieve control of hydrochloric acid/chlorine gas 
emissions from alkaline raw materials in the kiln. Control 
effectiveness varies among kilns based on the alkalinity of the raw 
materials. Thus, the cement manufacturing process serves essentially as 
a dry scrubber. We conclude, therefore, that the addition of a dry 
scrubber will only marginally improve hydrochloric acid/chlorine gas 
removal and is not warranted as beyond-the-floor control.
    It is also our judgment that a beyond-the-floor standard based on 
wet scrubbing is not warranted. The total estimated engineering 
retrofit costs would be approximately equivalent to those identified at 
proposal for this option. However, emissions reductions would be less 
given that the final MACT floor level is more stringent than the

[[Page 52885]]

level proposed. Therefore, the cost-effectiveness of a beyond-the-floor 
standard would be less attractive than the number we rejected at 
proposal. As a result, we must reaffirm that conclusion here.
    Limiting the Feedrate of Chlorine in the Hazardous Waste. We also 
considered a beyond-the-floor standard for hydrochloric acid/chlorine 
gas based on additional feedrate control of chlorine in the hazardous 
waste. We are concerned, however, that cement kilns making low alkali 
cement may not be able to achieve a beyond-the-floor standard by 
controlling feedrate of chlorine in the hazardous waste. As noted 
above, chlorine is sometimes added specifically to volatilize potassium 
and sodium compounds that must be removed from the clinker to produce 
low alkali cement. Based on limited data submitted by a cement facility 
manufacturing low alkali cement, achievability of a beyond-the-floor 
standard of 70 ppmv, representing a 45% reduction from the floor level, 
may not be feasible for this source using feedrate control and others 
by inference. Therefore, we conclude that a beyond-the-floor standard 
based on chlorine feedrate control in the hazardous waste is not 
appropriate.
    Limiting the Feedrate of Chlorine in the Raw Materials. A source 
can achieve a reduction in hydrochloric acid/chlorine gas emissions by 
substituting a feed material containing lower levels of chlorine for a 
primary raw material with higher levels of chlorine. This beyond-the-
floor option is less cost-effective compared to the scrubbing options 
evaluated above because cement kilns are sited proximate to the primary 
raw material supply and transporting large quantities of an alternative 
source of raw material(s) is not technically achievable. Therefore, we 
do not adopt a hydrochloric acid/chlorine gas beyond-the-floor standard 
based on limiting the feedrate of chlorine in raw materials.
    In summary, we establish the hydrochloric acid/chlorine gas 
standard for existing hazardous waste burning cement kilns at the floor 
level of 130 ppmv.
    c. What Is the MACT Floor for New Sources? At proposal, we defined 
floor control for new sources as hazardous waste feedrate control for 
chlorine and the proposed floor level was 630 ppmv. (See 61 FR at 
17401.) In the May 1997 NODA, we concluded that the floor control and 
emission level for existing sources for hydrochloric acid/chlorine gas 
would also be appropriate for new sources. Floor control was based on 
limiting hazardous waste feedrates of chlorine. After screening out 
some data with anomalous system removal efficiencies compared to the 
majority of sources, we established the floor level at the test 
condition average of the breakpoint source. We identified a floor level 
for new kilns of 120 ppmv. (See 62 FR at 24230.)
    As discussed above for existing sources, in developing the final 
rule, we use the aggregate feedrate approach to identify MACT floors 
for hydrochloric acid/chlorine gas. The resulting MACT emissions floor 
for new hazardous waste burning cement kilns is 86 ppmv.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the proposal, we considered a beyond-the-floor standard for new cement 
kilns of 67 ppmv based on wet scrubbing and concluded that it would not 
be cost-effective. In the May 1997 NODA, we also concluded that a 
beyond-the-floor standard based on wet scrubbing would likewise not be 
cost-effective. Considering the level of the floor standard for new 
kilns, we do not believe that a more stringent beyond-the-floor 
standard is warranted for the final rule, especially considering our 
concerns for cement kilns manufacturing low alkali cements.
    In summary, we adopt the floor level of 86 ppmv as the standard for 
hydrochloric acid/chlorine gas for new sources.
8. What Are the Hydrocarbon and Carbon Monoxide Standards for Kilns 
Without By-Pass Sampling Systems? 138
---------------------------------------------------------------------------

    \138\ See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume I: Description of 
Source Categories,'' July 1999, for further explanation of by-pass 
and midkiln sampling systems. Hydrocarbon and carbon monoxide 
standards for kilns equipped with by-pass sampling systems are 
discussed in Section VI.D.9 f the text.
---------------------------------------------------------------------------

    See Sec. 63.1205(a)(5) and (b)(5).
    In today's rule, we establish hydrocarbon and carbon monoxide 
standards for new and existing cement kilns without by-pass sampling 
systems as surrogates to control emissions of nondioxin organic 
hazardous air pollutants. The standards for existing sources limit 
hydrocarbon or carbon monoxide concentrations to 20 ppmv \139\ or 100 
ppmv, \140\ respectively. The standards for new sources limit: (1) 
Hydrocarbons to 20 ppmv; or (2) carbon monoxide to 100. New, greenfield 
\141\ kilns that elect to comply with the 100 ppmv carbon monoxide 
standard, however, must also comply with a 50 ppmv \142\ hydrocarbon 
standard. New and existing sources that elect to comply with the 100 
ppmv carbon monoxide standard, including new greenfield kilns that 
elect to comply with the carbon monoxide standard and 50 ppmv 
hydrocarbon standard, must also demonstrate compliance with the 20 ppmv 
hydrocarbon standard during the comprehensive performance test.\143\ 
(See Part Four, Section IV.B of the preamble for the rationale for this 
requirement.) We discuss the rationale for these standards below.
---------------------------------------------------------------------------

    \139\ Hourly rolling average, reported as propane, dry basis, 
and corrected to 7% oxygen.
    \140\ Hourly rolling average, dry basis, corrected to 7% oxygen.
    \141\ A greenfield cement kiln is a kiln that commenced 
construction or reconstruction after April 19, 1996 at a site where 
no cement kiln previously existed, irrespective of the class of kiln 
(i.e., nonhazardous waste or hazardous waste burning). A newly 
constructed or reconstructed cement kiln at an existing site would 
not be classified as a greenfield cement kiln, and would be subject 
to the same carbon monoxide and hydrocarbon standards as an existing 
cement kiln.
    \142\ Thirty day block average, reported as propane, dry basis, 
and corrected to 7 percent oxygen.
    \143\ As discussed in Part 5, Section X.F, sources that feed 
hazardous waste at a location other than the end where products are 
normally discharged and where fuels are normally fired must comply 
with the 20 ppmv hydrocarbon standard i.e., these sources do not 
have the option to comply with the carbon monoxide standard).
---------------------------------------------------------------------------

    a. What Is the MACT Floor for Existing Sources? As discussed in 
Part Four, Section II.B.2, we proposed limits on hydrocarbon emissions 
for kilns without by-pass sampling systems as a surrogate to control 
nondioxin organic hazardous air pollutants. In the April 1996 proposal 
(61 FR at 17397), we identified a hydrocarbon floor emission level of 
20 ppmv for cement kilns not equipped with by-pass sampling systems, 
and proposed that floor control be based on the current federally-
enforceable RCRA boiler and industrial furnace standards, control of 
organics in raw materials coupled with operating under good combustion 
practices to minimize fuel-related hydrocarbon. In the May 1997 NODA, 
we also indicated that this approach was appropriate.
    Some commenters stated that a carbon monoxide limit of 100 ppmv was 
necessary for these cement kilns to better control organic hazardous 
air pollutants. Commenters also wrote that, alone, neither carbon 
monoxide nor hydrocarbons is an acceptable surrogate for organic 
hazardous air pollutant emissions. Additionally, commenters suggested 
that by requiring both carbon monoxide and hydrocarbon limits, we would 
further reduce emissions of organic hazardous air pollutants.
    We conclude that continuous compliance with both a carbon monoxide 
and hydrocarbon standard is unwarranted for the following reasons. 
First, stack gas carbon monoxide levels are not a universally reliable 
indicator

[[Page 52886]]

of combustion intensity and efficiency for kilns without by-pass 
sampling systems. This is due to carbon monoxide generation by 
disassociation of carbon dioxide to carbon monoxide at the high 
sintering zone temperatures and evolution of carbon monoxide from the 
trace organic constituents in raw material feedstock.\144\ (See 56 FR 
at 7150, 7153-55). Thus, carbon monoxide can be a too conservative 
surrogate for this type of kiln for potential emissions of hazardous 
air pollutants from combustion of hazardous waste. There are other 
sources of carbon monoxide unrelated to combustion of hazardous 
waste.\145\
---------------------------------------------------------------------------

    \144\ Raw materials enter the upper end of the kiln and move 
counter-current to the combustion gas. Thus, as the raw materials 
are heated in the kiln, organic compounds can evolve from trace 
levels of organics in the raw materials. These organic compounds can 
be measured as hydrocarbons and, when only partially oxidized, 
carbon monoxide. This process is not related to combustion of 
hazardous waste or other fuels in the combustion zone at the other 
end of the kiln.
    \145\ Of course, if a source elects to comply with the carbon 
monoxide standard, then we are more assured of good combustion 
conditions in the combustion zone, and thus good control of organic 
hazardous air pollutants that could be potentially emitted from 
feeding hazardous waste in the combustion zone.
---------------------------------------------------------------------------

    Second, requiring continuous compliance with both a carbon monoxide 
and hydrocarbon emission limitation in the stack can be redundant for 
control of organic emissions from combustion of hazardous waste 
because: (1) Hydrocarbon alone is a direct and reliable surrogate for 
organic hazardous air pollutants; and (2) in most cases carbon monoxide 
is a conservative indicator of good combustion conditions and thus good 
control of organic hazardous air pollutants. As discussed in the 
following paragraphs, however, we have concluded that a source must 
demonstrate compliance with the hydrocarbon standard during the 
comprehensive performance test if it elects to continuously comply with 
the carbon monoxide standard to ensure that carbon monoxide is an 
adequate continuously monitored indicator of combustion efficiency. See 
Part Four, Section IV of the preamble for a discussion of the merits of 
using limits on stack gas concentrations of carbon monoxide and 
hydrocarbon to control organic emissions.
    One commenter suggested cement kilns be given the option to comply 
with a carbon monoxide limit of 100 ppmv instead of the 20 ppmv 
hydrocarbon limit. The commenter emphasized that this option is 
currently allowed under the RCRA boiler and industrial furnace 
regulations, and that it would be conservative because hydrocarbon 
levels would always be below 20 ppmv when carbon monoxide levels are 
below 100 ppmv. As discussed below, we agree that cement kilns should 
be given the option to comply with either standard, but do not agree 
that compliance with the carbon monoxide standard ensures compliance 
with the hydrocarbon standard.
    We have determined that it is necessary to require a source that 
elects to continuously comply with the carbon monoxide standard to also 
demonstrate compliance with the 20 ppmv hydrocarbon standard during the 
comprehensive performance test. We concluded that this requirement is 
necessary because we have limited data that shows a source can produce 
high hydrocarbon emissions while simultaneously producing low carbon 
monoxide emissions. This requirement to demonstrate compliance with the 
hydrocarbon standard during the performance test is sufficient to 
ensure that carbon monoxide alone is an appropriate continuously 
monitored indicator of combustion efficiency. See Part 4, Section IV.B, 
for a more detailed discussion. Consistent with this principle, 
incinerators and lightweight aggregate kilns are also required to 
demonstrate compliance with hydrocarbon standard during the 
comprehensive performance test if they elect to comply with the carbon 
monoxide standard.
    In today's final rule, we are identifying a carbon monoxide level 
of 100 ppmv and a hydrocarbon level of 20 ppmv as floor control for 
existing sources because they are currently enforceable Federal 
standards for hazardous waste burning cement kilns. See Sec. 266.104(b) 
and (c). As current rules allow, sources would have the option of 
complying with either limit. However, sources that elect to comply with 
the carbon monoxide standard must also demonstrate compliance with the 
hydrocarbon standard during the comprehensive performance test.
    Given that these are current RCRA rules, all cement kilns without 
by-pass sampling systems can currently achieve these emission levels. 
Thus, we estimate no emissions reductions (or new costs) for compliance 
with these floor levels.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 proposal, we identified beyond-the-floor 
control levels for carbon monoxide and hydrocarbon in the main stack of 
50 ppmv and 6 ppmv, respectively. (See 61 FR at 17397.) These beyond-
the-floor levels were based on the use of a combustion gas afterburner. 
We indicated in the proposal, however, that the beyond-the-floor 
control was not practical since no kilns currently achieved these 
emission levels, and because of the high costs to retrofit a kiln with 
an afterburner.
    One commenter wrote that we rejected the 50 ppmv and 6 ppmv beyond-
the-floor carbon monoxide and hydrocarbon standards, respectively, 
without providing any justification. In order to confirm the reasoning 
discussed above, we have now estimated that the annualized cost for an 
afterburner for cement kilns will range from $3-8 million dollars per 
facility.\146\ As proposed, and as we reiterated in the May 1997 NODA a 
beyond-the-floor standard based on an afterburner would be not be cost-
effective due to the high retrofit costs and minimal incremental 
emissions reductions, and we do not adopt a beyond-the-floor standard 
for existing cement kilns.
---------------------------------------------------------------------------

    \146\ See `Final Technical Support Document for Hazardous Waste 
Combustor MACT Standards, Volume V: Emission Estimates and 
Engineering Costs'', February, 1999.
---------------------------------------------------------------------------

    In summary, we adopt the floor emission levels as standards for 
carbon monoxide, 100 ppmv, and hydrocarbons, 20 ppmv.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal (see 61 FR at 17401) and the May 1997 NODA, we identified a 
new source hydrocarbon floor emission level of 20 ppmv for new cement 
kilns not equipped with by-pass sampling systems based on the current 
Federally-enforceable BIF standards. The hydrocarbon limit is based on 
control of organics in raw materials coupled with good combustion 
practices.
    In developing the final rule, we considered the comment discussed 
above that the rule should allow compliance with either a carbon 
monoxide standard of 100 ppmv or a hydrocarbon standard of 20 ppmv. 
Given that this option is available under the current BIF rule for new 
and existing sources, we now conclude that it represents MACT floor for 
new sources, except as discussed below.
    As discussed previously, we have also proposed MACT standards for 
nonhazardous waste burning cement kilns. See 63 FR 14182, March 24, 
1998. In that proposal, we determined that some existing sources have 
used the combination of feed material selection, site location, and 
feed material blending to optimize operations. We then concluded that 
site selection based on availability of acceptable raw material 
hydrocarbon content is a feasible approach to control hydrocarbon 
emissions at new sources. See 63 FR at 14202-03. We proposed a new 
source

[[Page 52887]]

floor hydrocarbon emission level of 50 ppmv at nonhazardous waste 
burning Portland cement kilns because it is being consistently achieved 
during thirty-day block averaging periods when high hydrocarbon content 
raw materials are avoided. We have since promulgated a standard of 50 
ppmv for hydrocarbons for new nonhazardous waste burning cement kilns. 
64 FR 31898.
    We now conclude for the same reasons that site selection is floor 
control for new source, greenfield hazardous waste burning cement kilns 
\147\ and that the floor hydrocarbon emission level is 50 ppmv.\148\ 
Sources must document compliance with this standard for each thirty-day 
block period of operation. We reconcile this hydrocarbon floor level of 
50 ppmv with the floor levels discussed above of 20 ppmv hydrocarbons 
or 100 ppmv carbon monoxide by establishing the floor as follows. For 
new source greenfield kilns, the floor is either: (1) 20 ppmv 
hydrocarbons; or (2) 100 ppmv carbon monoxide and 50 ppmv hydrocarbons. 
For other new sources not located at greenfield sites, the floor is 
either 20 ppmv hydrocarbons or 100 ppmv carbon monoxide, which is 
identical to the standards for existing sources.
---------------------------------------------------------------------------

    \147\ At least one hazardous waste burning cement kiln in our 
data base used raw material substitution to control hydrocarbon 
emissions.
    \148\ We concluded that this new source hydrocarbon standard of 
50 ppms should not apply to new sources that are not located at 
greenfield sites since these kilns are not capable of using site-
selection to control hydrocarbon emissions.
---------------------------------------------------------------------------

    The combined 20 ppmv hydrocarbon and 100 ppmv carbon monoxide 
standards control organic hazardous air pollutant emissions that 
originate from the incomplete combustion of hazardous waste. The 50 
ppmv hydrocarbon standard for new greenfield kilns controls organic 
hazardous air pollutant emissions that originate from the raw material. 
We conclude that the 50 ppmv hydrocarbon standard is necessary to deter 
new kilns from siting at locations that have on-site raw material that 
is high in organic content, since siting a cement kiln at such a 
location could result in elevated hydrocarbon emissions.
    We considered whether new greenfield kilns would be required to 
monitor hydrocarbons continuously, or just document compliance with the 
50 ppmv limit during the comprehensive performance test. We determined 
that hydrocarbons must be continuously monitored because compliance 
with the 100 ppmv carbon monoxide limit may not always ensure 
compliance with the 50 ppmv hydrocarbon limit. This is because 
hydrocarbons could potentially evolve from raw materials in the upper 
drying zone end of the kiln under conditions that inhibit sufficient 
oxidation of the hydrocarbons to form carbon monoxide.
    As with existing sources, we are requiring new sources that elect 
to continuously comply with the carbon monoxide standard, and new 
greenfield sources that elect to comply with the carbon monoxide and 50 
ppmv hydrocarbon standard, to also demonstrate compliance with the 20 
ppmv hydrocarbon standard during the comprehensive performance test. 
Consistent with this principle, incinerators and lightweight aggregate 
kilns are also required to demonstrate compliance with the hydrocarbon 
standard during the comprehensive performance test if they elect to 
comply with the carbon monoxide standard.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 proposal, we identified beyond-the-floor emission levels 
for carbon monoxide and hydrocarbon of 50 ppmv and 6 ppmv, 
respectively, for new sources. (See 61 FR at 17401.) These beyond-the-
floor levels were based on the use of a combustion gas afterburner. We 
indicated in the proposal, however, that beyond-the-floor control was 
not practical since none of the kilns in our data base are achieving 
these emission levels, and because of the high costs to retrofit kilns 
with an afterburner. We reiterated in the May 1997 NODA that a beyond-
the-floor standard based on use of an afterburner would not be cost-
effective.
    One commenter supported these beyond-the-floor standards for new 
sources, but did not explain why these were considered to be 
appropriate standards. As discussed above for existing sources, we 
continue to believe that a beyond-the-floor standard based on use of an 
afterburner would not be cost-effective.
    In summary, we adopt the floor levels as standards for new sources. 
For new source greenfield kilns, the standard monitored continuously is 
either: (1) 20 ppmv hydrocarbons; or (2) 100 ppmv carbon monoxide and 
50 ppmv hydrocarbons. For other new source kilns, the standard is 
either 20 ppmv hydrocarbons or 100 ppmv carbon monoxide monitored 
continuously. New sources that elect to comply with the carbon monoxide 
standard, and new greenfield sources that elect to comply with the 
carbon monoxide and 50 ppmv hydrocarbon standard, must also demonstrate 
compliance with the 20 ppmv hydrocarbon standard, but only during the 
comprehensive performance test.
9. What Are the Carbon Monoxide and Hydrocarbon Standards for Kilns 
With By-Pass Sampling Systems? 149
---------------------------------------------------------------------------

    \149\ This also includes cement kilns which have midkiln 
sampling systems. See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume I: Description of 
Source Categories,'' July 1999, for further explanation of by-pass 
and midkiln sampling systems.
---------------------------------------------------------------------------

    See Sec. 63.1204(a)(5) and (b)(5).
    We establish carbon monoxide and hydrocarbon standards for existing 
and new cement kilns with by-pass sampling systems as surrogates to 
control emissions of nondioxin organic hazardous air 
pollutants.150 Existing kilns are required to comply with 
either a carbon monoxide standard of 100 ppmv or a hydrocarbon standard 
of 10 ppmv on an hourly rolling average basis. Both standards apply to 
combustion gas sampled in the by-pass or a midkiln sampling port that 
samples representative kiln gas. Sources that elect to comply with the 
carbon monoxide standard, however, must also document compliance with 
the hydrocarbon standard during the comprehensive performance 
test.151 See Part Four, Section IV.B of the preamble for the 
rationale for this requirement.
---------------------------------------------------------------------------

    \150\ As discussed in Part 5, Section X.F, cement kilns equipped 
with bypass sampling systems that feed hazardous waste at a location 
other than the end where products are normally discharged and at a 
location downstream of the bypass sampling location (relative to the 
combustion gas flow direction) must comply with the 20 ppmv main 
stack hydrocarbon standard discussed in the previous section in lieu 
of the bypass gas hydrocarbon standard.
    \151\As discussed in Part 5, Section X.F, cement kilns that feed 
hazardous waste at a location other than the end where products are 
normally discharged and where fuels are normally fired must comply 
wit the 10 ppmv hydrocarbon standard (i.e., these sources do not 
have the option to comply with the carbon monoxide standard).
---------------------------------------------------------------------------

    New kilns are subject to the same by-pass gas carbon monoxide and 
hydrocarbon standards as existing sources. But, new, greenfield 
152 kilns must also comply with a 50 ppmv hydrocarbon 
standard continuously monitored in the main stack. Sources must 
document compliance with this standard for each thirty-day block period 
of operation.
---------------------------------------------------------------------------

    \152\ A greenfield cement kiln is a kiln that commenced 
construction or reconstruction after April 19, 1996 at a site where 
no cement kiln previously existed, irrespective of the class of kiln 
(i.e., nonhazardous waste or hazardous waste burning). A newly 
constructed or reconstructed cement kiln at an existing site would 
not be classified as a greenfield cement kiln, and would be subject 
to the same carbon monoxide and hydrocarbon standards as an existing 
cement kiln.
---------------------------------------------------------------------------

    We discuss the rationale for adopting these standards below.

[[Page 52888]]

    a. What Is the MACT Floor for Existing Sources? In the April 1996 
proposal, we identified floor carbon monoxide and hydrocarbon emission 
standards for by-pass gas of 100 ppmv and 6.7 ppmv, respectively. Floor 
control was good combustion practices. (See 61 FR at 17397.) In the May 
1997 NODA, we used an alternative data analysis method to identify a 
hydrocarbon floor level of 10 ppmv.153 See 62 FR at 24230. 
Our decision to use engineering information and principles to set the 
proposed floor standard was based, in part, on the limited hydrocarbon 
data in our data base. In addition, we reasoned that the hydrocarbon 
levels being achieved in an incinerator, (i.e., 10 ppmv) are also being 
achieved in a cement kiln's by-pass duct.154
---------------------------------------------------------------------------

    \153\ The proposed hydrocarbon standard of 6.7 ppmv was based on 
a statistical and breakpoint analysis. Today's final rule, 
consistent with May 1997 NODA, instead uses engineering information 
and principles to identify the floor hydrocarbon level of 10 ppmv.
    \154\ See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume III: Selection of 
MACT Standards and Technologies,'' February, 1999.
---------------------------------------------------------------------------

    Some commenters stated that we did not have sufficient hydrocarbon 
emissions data from cement kilns equipped with by-pass sampling systems 
to justify a by-pass duct hydrocarbon standard. We disagree and 
conclude that we have adequate data because the MACT data base includes 
seven cement kilns that monitored hydrocarbons at the bypass sampling 
location. These sources are achieving hydrocarbon levels of 10 ppmv or 
less.155 The fact that these sources achieve hydrocarbon 
levels below 10 ppmv supports our use of engineering information and 
principles to set the floor limit at 10 ppmv.156
---------------------------------------------------------------------------

    \155\ Four of these kilns have ceased hazardous waste 
operations, and one of the kilns collected that data during time 
periods other than Certification of Compliance testing.
    \156\ We note that we could have elected to establish this 10 
ppmv hydrocarbon standard as a beyond-the-floor standard rather than 
a floor standard.
---------------------------------------------------------------------------

    Many commenters questioned whether cement kilns with by-pass 
sampling systems should comply with both a hydrocarbon and carbon 
monoxide standard. Those in favor of requiring cement kilns to comply 
with both standards wrote that neither carbon monoxide nor hydrocarbons 
are sufficient surrogates for organic hazardous air pollutant 
emissions. Commenters also noted that by requiring both a carbon 
monoxide and hydrocarbon limit, we would achieve appropriate organic 
hazardous air pollutant emission reductions. Other commenters wrote 
that continuous compliance with both a hydrocarbon and a carbon 
monoxide standard would be redundant and unnecessarily costly. We agree 
with the latter view, in that requiring continuous compliance with both 
standards for bypass gas is redundant for control of organic emissions 
from combustion of hazardous waste because, as previously discussed: 
(1) Hydrocarbon alone is a direct and reliable surrogate for organic 
hazardous air pollutants; and (2) in most cases, carbon monoxide is a 
conservative indicator of good combustion conditions and thus good 
control of organic hazardous air pollutants. However, as discussed 
earlier, we have concluded that a source must demonstrate compliance 
with the hydrocarbon standard during the comprehensive performance test 
if it elects to continuously comply with the carbon monoxide standard 
to ensure that carbon monoxide is an adequate continuously monitored 
indicator of combustion efficiency. See discussion in Part Four, 
Section IV.B of the preamble for more discussion on this issue.
    One commenter stated that due to some by-pass gas quenching 
methods, and the need to correct for moisture and oxygen, it may not be 
possible to accurately measure hydrocarbons to the level of the 
proposed standard, i.e., 6.7 ppmv. We disagree with this reasoning 
because, as explained in the technical support document, cement kiln 
by-pass hydrocarbon levels should be reasonably achievable and 
measurable by decreasing the span and increasing the calibration 
frequency of the hydrocarbon monitor.157 We also note that a 
cement kiln has the option to petition the Administrator for 
alternative monitoring approaches under Sec. 63.8(f) if the source has 
valid reasons why a total hydrocarbon monitor cannot be used to 
document compliance.
---------------------------------------------------------------------------

    \157\ See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume III: Selection of 
MACT Standards and Technologies,'' February, 1999.
---------------------------------------------------------------------------

    We conclude that floor control can achieve by-pass gas emission 
levels of 100 ppmv for carbon monoxide and 10 ppmv for hydrocarbons. As 
discussed in Part Four, Section IV.B, a source may comply with either 
standard. If the source elects to comply with the carbon monoxide 
standard, however, it must also demonstrate compliance with the 
hydrocarbon standard during comprehensive performance testing.
    We estimate that all cement kilns with by-pass sampling systems can 
currently achieve the carbon monoxide floor of 100 ppmv. We also 
estimate that approximately 97 percent of cement kilns with by-pass 
sampling systems meet the hydrocarbon floor level of 10 ppmv. The 
national annualized compliance cost for cement kilns to comply with the 
floor level is $37K and hydrocarbon emissions will be reduced by 11 Mg/
yr, two percent from current baseline emissions .
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 proposal, we identified a beyond-the-floor 
control level for carbon monoxide and hydrocarbons in the main stack of 
50 ppmv and 6 ppmv, respectively, based on the use of a combustion gas 
afterburner. (See 61 FR at 17399.) We indicated in the proposal that 
this beyond-the-floor level was not practical, however, since none of 
the kilns currently achieve these emission levels and because of the 
high costs of retrofitting kilns with an afterburner. We estimate that 
the annualized cost for each cement kiln to operate afterburners range 
from three to eight million dollars.158 We continue to 
believe that it is not cost-effective based on the high retrofit costs 
and minimal incremental emissions reductions to adopt these beyond-the-
floor standards.
---------------------------------------------------------------------------

    \158\ See ``Final Technical Support Document for Hazardous Waste 
Combustor MACT Standards, Volume V: Emission Estimates and 
Engineering Costs'', February, 1999.
---------------------------------------------------------------------------

    In the April 1996 NPRM, we also considered limiting main stack 
hydrocarbon emissions to a beyond-the-floor level of 20 ppmv based on 
the use of a low-organic raw material.159 This was in 
addition to floor controls limiting carbon monoxide and/or hydrocarbon 
levels in the by-pass. See 61 FR at 17398. We considered this beyond-
the-floor option to address concerns that: (1) organics desorbed from 
raw materials may contain hazardous air pollutants, even absent any 
influence from burning hazardous waste; and, (2) it is reasonable to 
hypothesize that the chlorine released from burning hazardous waste can 
react with the organics desorbed from the raw material to form 
generally more toxic chlorinated hazardous air pollutants. Many 
commenters supported this approach. For the reasons discussed below, 
however, we conclude it is not appropriate to adopt this beyond-the-

[[Page 52889]]

floor hydrocarbon standard for existing sources.
---------------------------------------------------------------------------

    \159\ The definition of floor control for existing cement kilns 
equipped with by-pass sampling systems does not include the use of 
low organic raw material. Although we have limited data indicating 
that some kilns used low organic raw material to control hydrocarbon 
emissions, there are enough facilities using this method of control 
to establish it as a floor control for existing sources.
---------------------------------------------------------------------------

    Also, many commenters stated that we should establish a main stack 
hydrocarbon standard because, as stated above, hazardous waste 
combustion byproducts from cement kilns, particularly chlorine, can 
react with organic compounds desorbed from raw materials to form 
hazardous air pollutants. Commenters believe that an additional main 
stack hydrocarbon emission standard would limit the emissions of 
chlorinated organic hazardous air pollutants that are generated due to 
the interaction of the hazardous waste combustion byproducts and the 
organics desorbed from the raw material.
    We disagree that a main stack hydrocarbon emission limit is an 
appropriate beyond-the-floor control for existing sources. First, we do 
not believe it is cost-effective to require an existing kiln to 
substitute its raw material with an off-site raw 
material.160 Cement kilns are sited proximate to the primary 
raw material supply and transporting large quantities of an alternative 
source of raw material(s) is likely to be very costly. Second, 
establishing a main stack hydrocarbon limit for existing sources is 
likely to be counter-productive in controlling organic hazardous air 
pollutants. It may compel the operator to avoid the unacceptable costs 
of importing low organic raw material by increasing back-end kiln 
temperatures to oxidize organics desorbed from raw material, thus 
lowering hydrocarbon levels. This increase in temperature may result in 
increased dioxin formation and is counter to our dioxin control 
strategy. Third, it is debatable whether there is a strong relationship 
between chlorine feedrates and chlorinated organic hazardous air 
pollutant emissions, as is suggested by commenters.161 
Finally, we anticipate that any potential risks associated with the 
possible formation of these chlorinated hazardous air pollutants at 
high hydrocarbon emission levels can be adequately addressed in a site-
specific risk assessment conducted as part of the RCRA permitting 
process. This increased potential for emissions of chlorinated 
hazardous air pollutants is not likely to warrant evaluation via a 
site-specific risk assessment under RCRA, however, unless main stack 
hydrocarbon levels are substantially higher than the 20 ppmv limit 
currently applicable under RCRA for cement kilns not equipped with by-
pass systems.
---------------------------------------------------------------------------

    \160\ We did not quantify actual costs associated with raw 
material substitution due to the lack of information.
    \161\ It is true that some studies have shown a relationship 
between chlorine levels in the flue gas and the generation of 
chlorobenzene in cement kiln emissions: the more chlorine, the more 
chlorobenzene is generated. Some full-scale tests, however, have 
shown that there is no observable or consistent trend when comparing 
``baseline'' (i.e., nonhazardous waste operation) organic hazardous 
air pollutant emissions with organic hazardous air pollutant 
emissions associated with hazardous waste operations, as well as 
comparing hazardous waste conditions with varying levels of 
chlorine. See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume III: Selection of 
MACT Standards and Technologies,'' July 1999, for further 
discussion.
---------------------------------------------------------------------------

    In summary, we adopt the floor levels as standards for carbon 
monoxide, 100 ppmv, and hydrocarbons, 10 ppmv. As discussed above, a 
source may comply with either standard. If the source elects to comply 
with the carbon monoxide standard, however, it must also demonstrate 
compliance with the hydrocarbon standard during comprehensive 
performance testing.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal, we identified new source floor standards for carbon monoxide 
and hydrocarbon emissions in the by-pass of 100 ppmv and 6.7 ppmv, 
respectively. We identified good combustion practices as floor control. 
(See 61 FR at 17401.) In the May 1997 NODA, we used an alternative data 
analyses method, in part, to identify an alternative new source 
hydrocarbon floor level. (See 62 FR at 24230.) As a result of this 
analysis and the use of engineering information and principles, we 
identified a floor hydrocarbon emission level of 10 ppmv in the by-pass 
for new cement kilns. We continue to believe that the new source 
hydrocarbon floor methodology discussed in the May 1997 NODA, and the 
new source carbon monoxide floor methodology discussed in the April 
1996 proposal, are appropriate. Therefore, we adopt these floor 
emission levels for by-pass gas in today's final rule.
    We also establish a 50 ppmv hydrocarbon floor level for the main 
stack of new greenfield kilns. As discussed above (Part Four, Section 
VII.8.c), we concluded during development of the final rule that some 
cement kilns are currently controlling their feed material selection, 
site location, and feed material blending to optimize operations. 
Because these controls can be used to control hydrocarbon content of 
the raw material and, thus, hydrocarbon emissions in the main stack, 
they represent floor control for main stack hydrocarbons for new 
sources.162 We established a floor hydrocarbon emission 
level of 50 ppmv because it is being consistently achieved during 
thirty-day block averaging periods when high hydrocarbon content raw 
materials are avoided.
---------------------------------------------------------------------------

    \162\ At least one hazardous waste burning cement kiln in our 
data base used raw material substitution to control hydrocarbon 
emissions.
---------------------------------------------------------------------------

    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 proposal, we identified main stack beyond-the-floor 
emission levels for carbon monoxide and hydrocarbon of 50 ppmv and 6 
ppmv, respectively, for new sources. (See 61 FR at 17401.) These 
beyond-the-floor levels were based on the use of a combustion gas 
afterburner. We indicated in the proposal, however, that beyond-the-
floor control was not practical since none of the kilns in our data 
base are achieving these emission levels, and because of the high costs 
to retrofit kilns with an afterburner. We reiterated in the May 1997 
NODA, that a beyond-the-floor standard based on use of an afterburner 
would not be cost-effective.
    One commenter wrote that we rejected these beyond-the-floor carbon 
monoxide and hydrocarbon standards without providing any justification. 
Another commenter supported these beyond-the-floor standards for new 
sources. As discussed above (in greater detail) for existing sources, 
we continue to believe that a beyond-the-floor standard based on use of 
an afterburner would not be cost-effective.
    In the April 1996 proposal, we considered limiting main stack 
hydrocarbon emissions at new sources equipped with by-pass sampling 
systems to a beyond-the-floor level of 20 ppmv.163 This 
addressed concerns that: (1) Organics desorbed from raw materials 
contain hazardous air pollutants, even absent any influence from 
burning hazardous waste; and (2) it is reasonable to hypothesize that 
the chlorine released from burning hazardous waste can react with the 
organics desorbed from the raw material to form generally more toxic 
chlorinated hazardous air pollutants. Although not explicitly stated, 
beyond-the-floor control would have been control of feed material 
selection, site location, and feed material blending to control the 
hydrocarbon content of the raw material and, thus, hydrocarbon 
emissions in the main stack. As discussed above, however, we adopt 
today a main stack hydrocarbon floor standard of 50 ppmv for newly 
constructed greenfield cement kilns equipped with by-pass systems. We 
are not adopting a main stack beyond-the floor hydrocarbon standard of 
20 ppmv for these kilns because we

[[Page 52890]]

are concerned that it may not be readily achievable using beyond-the-
floor control.
---------------------------------------------------------------------------

    \163\ This was in addition to limiting hydrocarbon and/or carbon 
monoxide at the by-pass sampling location.
---------------------------------------------------------------------------

    In summary, we establish the following standards for new sources 
based on floor control: (1) By-pass gas emission standards for carbon 
monoxide and hydrocarbons of 100 ppmv and 10 ppmv, respectively; 
164 and (2) a main stack hydrocarbon standard of 50 ppmv at 
greenfield sites.
---------------------------------------------------------------------------

    \164\ A source may comply with either bypass gas standard. If 
the source elects to comply with the carbon monoxide standard, 
however, it must also demonstrate compliance with the hydrocarbon 
standard during comprehensive performance testing.
---------------------------------------------------------------------------

10. What Are the Destruction and Removal Efficiency Standards?
    We establish a destruction and removal efficiency (DRE) standard 
for existing and new cement kilns to control emissions of organic 
hazardous air pollutants other than dioxins and furans. Dioxins and 
furans are controlled by separate emission standards. See discussion in 
Part Four, Section IV.A. The DRE standard is necessary, as previously 
discussed, to complement the carbon monoxide and hydrocarbon emission 
standards, which also control these hazardous air pollutants.
    The standard requires 99.99 percent DRE for each principal organic 
hazardous constituent (POHC), except that 99.9999 percent DRE is 
required if specified dioxin-listed hazardous wastes are burned. These 
wastes are listed as--F020, F021, F022, F023, F026, and F027--RCRA 
hazardous wastes under part 261 because they contain high 
concentrations of dioxins.
    a. What Is the MACT Floor for Existing Sources? Existing sources 
are currently subject to DRE standards under Sec. 266.104(a) that 
require 99.99 percent DRE for each POHC, except that 99.9999 percent 
DRE is required if specified dioxin-listed hazardous wastes are burned. 
Accordingly, these standards represent MACT floor. Since all hazardous 
waste cement kilns are currently subject to these DRE standards, they 
represent floor control, i.e., greater than 12 percent of existing 
sources are achieving these controls.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? Beyond-the-floor control would be a requirement to achieve a 
higher percentage DRE, for example, 99.9999 percent DRE for POHCs for 
all hazardous wastes. A higher DRE could be achieved by improving the 
design, operation, or maintenance of the combustion system to achieve 
greater combustion efficiency.
    Sources will not incur costs to achieve the 99.99% DRE floor 
because it is an existing RCRA standard . A substantial number of 
existing hazardous waste combustors are not likely to be routinely 
achieving 99.999% DRE, however, and most are not likely to be achieving 
99.9999% DRE. Improvements in combustion efficiency will be required to 
meet these beyond-the-floor DREs. Improved combustion efficiency is 
accomplished through better mixing, higher temperatures, and longer 
residence times. As a practical matter, most combustors are mixing-
limited. Thus, improved mixing is necessary for improved DREs. For a 
less-than-optimum burner, a certain amount of improvement may typically 
be accomplished by minor, relatively inexpensive combustor 
modifications--burner tuning operations such as a change in burner 
angle or an adjustment of swirl--to enhance mixing on the macro-scale. 
To achieve higher and higher DREs, however, improved mixing on the 
micro-scale may be necessary requiring significant, energy intensive 
and expensive modifications such as burner redesign and higher 
combustion air pressures. In addition, measurement of such DREs may 
require increased spiking of POHCs and more sensitive stack sampling 
and analysis methods at added expense.
    Although we have not quantified the cost-effectiveness of a beyond-
the-floor DRE standard, we do not believe that it would be cost-
effective. For reasons discussed above, we believe that the cost of 
achieving each successive order-of-magnitude improvement in DRE will be 
at least constant, and more likely increasing. Emissions reductions 
diminish substantially, however, with each order of magnitude 
improvement in DRE. For example, if a source were to emit 100 gm/hr of 
organic hazardous air pollutants assuming zero DRE, it would emit 10 
gm/hr at 90 percent DRE, 1 gm/hr at 99 percent DRE, 0.1 gm/hr at 99.9 
percent DRE, 0.01 gm/hr at 99.99 percent DRE, and 0.001 gm/hr at 99.999 
percent DRE. If the cost to achieve each order of magnitude improvement 
in DRE is roughly constant, the cost-effectiveness of DRE decreases 
with each order of magnitude improvement in DRE. Consequently, we 
conclude that this relationship between compliance cost and diminished 
emissions reductions associated with a more stringent DRE standard 
suggests that a beyond-the-floor standard is not warranted.
    c. What Is the MACT Floor for New Sources? The single best 
controlled source, and all other hazardous waste cement kilns, are 
subject to the existing RCRA DRE standard under Sec. 266.104(a). 
Accordingly, we adopt this standard as the MACT floor for new sources.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? As 
discussed above, although we have not quantified the cost-effectiveness 
of a more stringent DRE standard, diminishing emissions reductions with 
each order of magnitude improvement in DRE suggests that cost-
effectiveness considerations would likely come into play. We conclude 
that a beyond-the-floor standard is not warranted.

VIII. What Are the Standards for Existing and New Hazardous Waste 
Burning Lightweight Aggregate Kilns?

A. To Which Lightweight Aggregate Kilns Do Today's Standards Apply?
    The standards promulgated today apply to each existing, 
reconstructed, and newly constructed lightweight aggregate plant where 
hazardous waste is burned in the kiln. These standards apply to major 
source and area source lightweight aggregate facilities. Lightweight 
aggregate kilns that do not engage in hazardous waste burning 
operations are not subject to this NESHAP; however, these kilns will be 
subject to future MACT standards for the Clay Products source category.
B. What Are the Standards for New and Existing Hazardous Waste Burning 
Lightweight Aggregate Kilns?
1. What Are the Standards for Lightweight Aggregate Kilns?
    In this section, the basis for the emissions standards for 
hazardous waste burning lightweight aggregate kilns is discussed. The 
kiln emission limits apply to the kiln stack gases from lightweight 
aggregate plants that burn hazardous waste. The emissions standards are 
summarized below:

[[Page 52891]]



       Standards for Existing and New Lightweight Aggregate Kilns
------------------------------------------------------------------------
 Hazardous air pollutant or             Emissions standard \1\
   hazardous air pollutant   -------------------------------------------
          surrogate             Existing sources         New sources
------------------------------------------------------------------------
Dioxin/furan................  0.20 ng TEQ/dscm; or  0.20 ng TEQ/dscm; or
                               0.40 ng TEQ/dscm      0.40 ng TEQ/dscm
                               and rapid quench of   and rapid quench of
                               the flue gas at the   the flue gas at the
                               exit of the kiln to   exit of the kiln to
                               less than 400 deg.F.  less than 400
                                                     deg.F.
Mercury.....................  47 g/dscm..  43 g/dscm.
Particulate matter..........  57 mg/dscm (0.025 gr/ 57 mg/dscm (0.025 gr/
                               dscf).                dscf).
Semivolatile metals \2\.....  250 g/dscm.  43 g/dscm.
Low volatile metals \3\.....  110 g/dscm.  110 g/dscm.
Hydrochloric acid/chlorine    230 ppmv............  41 ppmv.
 gas.
Hydrocarbons 2,3............  20 ppmv (or 100 ppmv  20 ppmv (or 100 ppmv
                               carbon monoxide).     carbon monoxide).
Destruction and removal        For existing and new sources, 99.99% for
 efficiency.                        each principal organic hazardous
                                   constituent (POHC) designated. For
                                 sources burning hazardous wastes F020,
                               F021, F022, F023, F026, or F027, 99.9999%
                                       for each POHC designated.
------------------------------------------------------------------------
\1\ All emission levels are corrected to 7% O2, dry basis.
\2\ Hourly rolling average. Hydrocarbons are reported as propane.
\3\ Lightweight aggregate kilns that elect to continuously comply with
  the carbon monoxide standard must demonstrate compliance with the
  hydrocarbon standard of 20 ppmv during the comprehensive performance
  test.

2. What Are the Dioxin and Furan Standards?
    In today's rule, we establish a standard for new and existing 
lightweight aggregate kilns that limits dioxin/furan emissions to 
either 0.20 ng TEQ/dscm; or 0.40 ng TEQ/dscm and rapid quench of the 
flue gas at the exit of the kiln to less than 400 deg.F. Our rationale 
for adopting these standards is discussed below.
    a. What Is the MACT Floor for Existing Sources? In the April 1996 
proposal, we had dioxin/furan emissions data from only one lightweight 
aggregate kiln and pooled that data with the dioxin/furan data for 
hazardous waste burning cement kilns to identify the MACT floor 
emission level. We stated that it is appropriate to combine the two 
data sets because they are adequately representative of general dioxin/
furan behavior and control in either type of kiln. Consequently, floor 
control and the floor emission level for lightweight aggregate kilns 
were the same as for cement kilns. We proposed a floor emission level 
of 0.20 ng TEQ/dscm, or temperature at the inlet to the fabric filter 
not to exceed 418 deg.F. (61 FR at 17403.)
    Several commenters opposed our proposed approach of pooling the 
lightweight aggregate kiln data with the cement kiln dioxin/furan data 
for the MACT floor analysis. In order to respond to commenter concerns, 
we obtained additional dioxin/furan emissions data from lightweight 
aggregate kiln sources. In a MACT reevaluation discussed in the May 
1997 NODA, we presented an alternative data analysis method to identify 
floor control and the floor emission level. In that NODA, dioxin/furan 
floor control was defined as temperature control not to exceed 
400 deg.F at the inlet to the fabric filter. That analysis resulted in 
a floor emission level of 0.20 ng TEQ/dscm, or 4.1 ng TEQ/dscm and 
temperature at the inlet to the fabric filter not to exceed 400 deg.F. 
(62 FR at 24231.) An emission level of 4.1 ng TEQ/dscm represents the 
highest single run from the test condition with the highest run 
average. We concluded that 4.1 ng TEQ/dscm was a reasonable floor 
level, from an engineering perspective, given our limited dioxin/furan 
data base for lightweight aggregate kilns. (We noted that if this were 
a large data set, we would have identified the floor emission level 
simply as the highest test condition average.) Due to variability among 
the runs of the test condition with the highest condition average and 
because a floor level of 4.1 ng TEQ/dscm is 40 percent higher than the 
highest test condition average of 2.9 ng TEQ/dscm lightweight aggregate 
kilns using floor control will be able to meet routinely a floor 
emission level of 4.1 ng TEQ/dscm.
    We maintain that the floor methodology discussed in the May 1997 
NODA is appropriate and we adopt this approach in today's rule. In that 
NODA we identified two technologies for control of dioxin/furan 
emissions from lightweight aggregate kilns. The first technology 
controls dioxin/furans by quenching kiln gas temperatures at the exit 
of the kiln so that gas temperatures at the inlet to the particulate 
matter control device are below the temperature range of optimum 
dioxin/furan formation. The other technology is activated carbon 
injected into the kiln exhaust gas. Because activated carbon injection 
is not currently used by any hazardous waste burning lightweight 
aggregate kilns, this technology was evaluated only as part of a 
beyond-the-floor analysis.
    One commenter opposes our approach specifying a MACT floor control 
temperature limitation of 400 deg.F at the particulate matter control 
device. Instead, the commenter supports a temperature limitation of 
417 deg.F, which is the highest temperature associated with any dioxin/
furan test condition in our data base. Although only two of the three 
test conditions for which we have dioxin/furan emissions data operated 
the fabric filter at 400 deg.F or lower (the third operated at 
417 deg.F), we do have other fabric filter operating temperatures from 
kilns performing RCRA compliance testing for other hazardous air 
pollutants that document fabric filter operations at 400 deg.F or 
lower. From these data, we conclude that lightweight aggregate kilns 
can operate the fabric filter at temperatures of 400 deg.F or lower. 
Thus, identifying floor control at a temperature limitation of 
400 deg.F ensures that all lightweight aggregate kilns will be 
operating consistent with sound operational practices for controlling 
dioxin/furan emissions.
    As discussed in the May 1997 NODA, specifying a temperature 
limitation of 400 deg.F or lower is appropriate for floor control 
because, from an engineering perspective, it is within the range of 
reasonable values that could have been selected considering that: (1) 
The optimum temperature window for surface-catalyzed dioxin/furan 
formation is approximately 450-750 deg.F; and (2) temperature levels 
below 350 deg.F can cause dew point condensation problems resulting in 
particulate matter control device corrosion. Further, lightweight 
aggregate kilns can operate at air pollution control device 
temperatures between 350 to 400 deg.F. In

[[Page 52892]]

fact, all lightweight aggregate kilns use (or have available) fabric 
filter ``tempering'' air dilution and water quench for cooling kiln 
exit gases prior to the fabric filter (some kilns also augment this 
with uninsulated duct radiation cooling). Thus, the capability of 
operating fabric filters at temperatures lower than 400 deg.F currently 
exists and is practical. See the technical support document for further 
discussion.165
---------------------------------------------------------------------------

    \165\ USEPA, ``Final Technical Support document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    In summary, today's floor emission level for dioxin/furan emissions 
for existing lightweight kilns is 0.20 ng TEQ/dscm or 4.1 ng TEQ/dscm 
and control of temperature at the inlet to the fabric filter not to 
exceed 400 deg.F. We estimate that all lightweight aggregate kiln 
sources currently are meeting the floor level.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? We considered in the April 1996 proposal a beyond-the-floor 
standard of 0.20 ng TEQ/dscm based on injection of activated carbon at 
a flue gas temperature of less than 400 deg.F. (61 FR at 17403.) In the 
May 1997 NODA, we considered a beyond-the-floor standard of 0.20 ng 
TEQ/dscm standard based on rapidly quenching combustion gases at the 
exit of the kiln to 400 deg.F, and insulating the duct-work between the 
kiln exit and the fabric filter to maintain gas temperatures high 
enough to avoid dew point problems. (62 FR at 24232.)
    One commenter, however, disagrees that there is adequate evidence 
(test data) supporting rapid quench of kiln exit gases to less than 
400 deg.F can achieve a level of 0.20 ng TEQ/dscm. Based on these NODA 
comments and upon closer analysis of all available data, we find that a 
level of 0.20 ng TEQ/dscm has not been clearly demonstrated for 
lightweight aggregate kilns with rapid quench less than 400 deg.F prior 
to the particulate matter control device. The data show that some 
lightweight aggregate kilns can achieve a level of 0.20 TEQ ng/dscm 
with rapid quench. In addition, one commenter, who operates two 
lightweight aggregate kilns with heat exchangers that cool the flue gas 
to a temperature of approximately 400 deg.F at the fabric filter, 
stated that they achieve dioxin/furan emissions slightly below 0.20 ng 
TEQ/dscm. However, because of the small dioxin/furan data base we are 
concerned that these limited data may not show the full range of 
emissions. Due to the similarity of dioxin/furan control among cement 
kilns and lightweight aggregate kilns, we looked to the cement kiln 
data to complement our limited lightweight aggregate kiln dataset. As 
discussed earlier, cement kilns are able to control dioxin/furans to 
0.40 ng TEQ/dscm with temperature control. Since we do not expect a 
lightweight aggregate kiln to achieve lower dioxin/furan emissions than 
a cement kiln with rapid quench, we agree with these commenters and 
conclude that lightweight aggregate kilns can control dioxin/furans to 
0.40 ng TEQ/dscm with rapid quench of kiln exit gases to less than 
400 deg.F.
    Thus, for the final rule, we considered two beyond-the-floor 
levels: (1) Either 0.20 ng TEQ/dscm; or 0.40 ng TEQ/dscm and rapid 
quench of the kiln exhaust gas to a temperature less than 400 deg.F; 
and (2) a level of 0.20 ng TEQ/dscm based on activated carbon 
injection.
    The first option is a beyond-the-floor standard of either 0.20 ng 
TEQ/dscm, or 0.40 ng TEQ/dscm and rapid quench of the kiln exhaust gas 
to less than 400 deg.F. The national incremental annualized compliance 
cost for lightweight aggregate kilns to meet this beyond-the-floor 
level rather than comply with the floor controls would be approximately 
$50,000 for the entire hazardous waste burning lightweight aggregate 
kiln industry, and would provide an incremental reduction in dioxin/
furan emissions beyond the MACT floor controls of nearly 2 g TEQ/yr.
    Based on these costs of approximately $25 thousand per additional g 
of dioxin/furan removed and on the significant reduction in dioxin/
furan emissions achieved, we have determined that this dioxin/furan 
beyond-the-floor option for lightweight aggregate kilns is justified, 
especially given our special concern about dioxin/furans. Dioxin/furans 
are some of the most toxic compounds known due to their bioaccumulation 
potential and wide range of health effects, including carcinogenesis, 
at exceedingly low doses. Exposure via indirect pathways is a chief 
reason that Congress singled out dioxin/furans for priority MACT 
control in section 112(c)(6) of the CAA. See S. Rep. No. 128, 101st 
Cong. 1st Sess. at 154-155.
    We also evaluated, but rejected, activated carbon injection as a 
beyond-the-floor option. Carbon injection is routinely effective at 
removing 99 percent of dioxin/furans at numerous municipal waste 
combustor and medical waste combustor applications and one hazardous 
waste incinerator application. However, no hazardous waste burning 
lightweight aggregate kiln currently uses activated carbon injection 
for dioxin/furan removal. We believe that it is conservative to assume 
that only 95 percent is achievable given potential uncertainties in its 
application to lightweight aggregate kilns. In addition, we assumed for 
cost-effectiveness calculations that lightweight aggregate kilns 
needing activated carbon injection would install the activated carbon 
injection system after the existing fabric filter device and add a new 
smaller fabric filter to remove the injected carbon with the absorbed 
dioxin/furans and mercury. This costing approach addresses commenter's 
concerns that injected carbon may interfere with current dust recycling 
practices.
    The national incremental annualized compliance cost for lightweight 
aggregate kilns to meet a beyond-the-floor level based on activated 
carbon injection rather than comply with the floor controls would be 
approximately $1.2 million for the entire hazardous waste burning 
lightweight aggregate kiln industry. This would provide an incremental 
reduction in dioxin/furan emissions beyond the MACT floor controls of 
2.2 g TEQ/yr, or 90 percent. Based on these costs of approximately 
$0.53 million per additional g of dioxin/furan removed and the small 
incremental dioxin/furan emissions reduction beyond the dioxin/furan 
beyond-the-floor option discussed above (2.0 g TEQ/yr versus 2.2 g TEQ/ 
yr), we have determined that this second beyond-the-floor option for 
lightweight aggregate kilns is not justified. Therefore, we are not 
promulgating a beyond-the-floor standard of 0.20 ng TEQ/dscm for 
lightweight aggregate kilns based on activated carbon injection.
    Thus, the promulgated dioxin/furan standard for existing 
lightweight aggregate kilns is a beyond-the-floor standard of 0.20 ng 
TEQ/dscm; or 0.40 ng TEQ/dscm and rapid quench to a temperature not to 
exceed 400 deg.F based on rapid quench of flue gas at the exit of the 
kiln.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal, the floor analysis for new lightweight aggregate kilns was 
the same as for existing kilns, and the proposed standard was the same. 
The proposed floor emission level was 0.20 ng TEQ/dscm, or temperature 
at the inlet to the particulate matter control device not to exceed 
418 deg.F. (61 FR at 17408.) In the May 1997 NODA, we used an 
alternative data analysis method to identify floor control and the 
floor emission level. As done for existing sources, floor control for 
new sources was defined as temperature control at the inlet to the 
particulate matter control device to less than 400 deg.F. That

[[Page 52893]]

analysis resulted in a floor emission level of 0.20 ng TEQ/dscm, or 4.1 
ng TEQ/dscm and temperature at the inlet to the fabric filter not to 
exceed 400 deg.F. Our engineering evaluation indicated that the best 
controlled source is one that is controlling temperature control at the 
inlet to the fabric filter at 400 deg.F. (62 FR at 24232.) We continue 
to believe that the floor methodology discussed in the May 1997 NODA is 
appropriate for new sources and we adopt this approach in the final 
rule. The floor level for new lightweight aggregate kilns is 0.20 ng 
TEQ/dscm, or 4.1 ng TEQ/dscm and temperature at the inlet to the 
particulate matter control device not to exceed 400 deg.F.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 proposal, we proposed activated carbon injection as 
beyond-the-floor control and a beyond-the-floor standard of 0.20 ng 
TEQ/dscm. (61 FR at 17408.) In the May 1997 NODA, we identified a 
beyond-the-floor standard of 0.20 ng TEQ/dscm based on rapid quench of 
kiln gas to less than 400 deg.F combined with duct insulation or 
activated carbon injection operated at less than 400 deg.F. (62 FR at 
24232.) These beyond-the-floor considerations are identical to those 
discussed above for existing sources.
    The beyond-the-floor standard identified for existing sources 
continues to be appropriate for new sources for the same reasons. Thus, 
the promulgated dioxin/furan standard for new lightweight aggregate 
kilns is the same as the standard for existing standards, i.e., 0.20 ng 
TEQ/dscm or 0.40 ng TEQ/dscm and rapid quench of the kiln exhaust gas 
to less than 400 deg.F.
3. What Are the Mercury Standards?
    In the final rule, we establish a standard for existing and new 
lightweight aggregate kilns that limits mercury emissions to 47 and 33 
g/dscm, respectively. The rationale for adopting these 
standards is discussed below.
    a. What Is the MACT Floor for Existing Sources? All lightweight 
aggregate kilns use fabric filters, and one source uses a venturi 
scrubber in addition to a fabric filter. However, since mercury is 
generally in the vapor form in and downstream of the combustion 
chamber, including in the air pollution control device, fabric filters 
alone do not achieve significant mercury control. Mercury emissions 
from lightweight aggregate kilns are currently controlled under 
existing regulations through limits on the maximum feedrate of mercury 
in total feedstreams (e.g., hazardous waste, raw materials). Thus, MACT 
floor control is based on limiting the feedrate of mercury in hazardous 
waste.
    In the April 1996 proposal, we identified floor control as 
hazardous waste feedrate control not to exceed a feedrate level of 17 
g/dscm, expressed as a maximum theoretical emissions 
concentration, and proposed a floor emission level of 72 g/
dscm based on an analysis of data from all lightweight aggregate kilns 
with a hazardous waste feedrate of mercury of this level or lower. (61 
FR at 17404.) In the May 1997 NODA, we conducted a breakpoint analysis 
on ranked mercury emissions data and established the floor emission 
level equal to the test condition average of the breakpoint source. (62 
FR at 24232.) The breakpoint analysis was intended to reflect an 
engineering-based evaluation of the data whereby the few lightweight 
aggregate kilns spiking extra mercury during testing procedures did not 
drive the floor emission level to levels higher than the preponderance 
of the emission data. We reasoned that sources with emissions higher 
than the breakpoint source were not controlling the hazardous waste 
feedrate of mercury to levels representative of MACT. The May 1997 NODA 
analysis resulted in a MACT floor level of 47 g/dscm.
    One commenter states that the use of mercury stack gas measurements 
from RCRA compliance test reports is inappropriate for setting the MACT 
floor since they are based on feeding normal wastes. With the exception 
of one source, no mercury spiking was done during the RCRA compliance 
testing because lightweight aggregate kilns complied with Tier I levels 
allowable in the Boiler and Industrial Furnace rule. The commenter 
notes that the Tier I allowable levels are above, by orders of 
magnitude, the total mercury fed into lightweight aggregate kilns. 
Thus, to set the mercury MACT floor, the commenter states that we need 
to consider the potential range of mercury levels in the hazardous 
waste and raw materials, which may not represented by the RCRA 
compliance stack gas measurements.
    We recognize that stack gas tests generating mercury emissions data 
were conducted with normal unspiked waste streams containing normal 
levels of mercury in hazardous waste. However, we concluded that it is 
appropriate in this particular circumstance to use unspiked data to 
define a MACT floor. See discussion in Part Four, Section V.D.1. It 
would hardly reflect MACT to base the floor emission level on a 
feedrate of mercury greater than that which actually occurs in 
hazardous waste fuels burned in these units. Furthermore, the final 
rule standard is projected to be achievable by lightweight aggregate 
kilns for the vast majority of the wastes they are currently handling. 
The standard would allow lightweight aggregate kilns to burn wastes 
with about 0.5 ppmw mercury, without use of add-on mercury control 
techniques such as carbon injection. Data provided by a commenter 
indicates that approximately 90% of the waste streams lightweight 
aggregate kilns currently burn do not contain mercury levels at 2 ppmw. 
Further, the commenter indicates that these wastes are typically less 
than 0.02 ppmw mercury when more refined and costly analysis techniques 
are used. Thus, the standard is consistent with the current practice of 
lightweight aggregate kilns burning low-mercury waste.
    We received comments from the lightweight aggregate kiln industry 
expressing concern with the stringency of the mercury standard. These 
commenters oppose a mercury standard of 47 g/dscm, in part, 
because of the difficulty and increased cost of demonstrating 
compliance with day-to-day mercury feedrate limits. One potential 
problem pertains to raw material mercury detection limits. The 
commenter states that mercury is generally not measured in the raw 
material at detectable levels at their facilities. The commenter points 
out that if a kiln assumes mercury is present in the raw material at 
the detection limit, the resulting calculated uncontrolled mercury 
emission concentration could exceed, or be a significant percentage of, 
the mercury emission standard. This may prevent a kiln from complying 
with the mercury emission standard even though MACT control is used. 
Further, the commenter anticipates that more frequent analysis, 
additional laboratory equipment and staff, and improved testing and 
analysis procedures will be required to show compliance with a standard 
of 47 g/dscm. The commenter states that the costs of 
compliance will increase significantly at each facility to address this 
nondetect issue.
    Four provisions in the final rule offer flexibility in complying 
with the mercury standard. For example, one provision allows sources to 
petition for an alternative mercury standard that only requires 
compliance with a hazardous waste mercury feedrate limitation, provided 
that mercury not been present historically in the raw material at 
detectable levels. This approach ensures that kilns using MACT controls 
can achieve the mercury standard. The details of this provision are 
discussed in Part Five, Section

[[Page 52894]]

X.A.2. Another provision allows kilns a waiver of performance testing 
requirements when the source feeds low levels of mercury. Under this 
provision, a kiln qualifies for a waiver of the performance testing 
requirements for mercury if all mercury from all feedstreams fed to the 
combustion unit does not exceed the mercury emission standard. For 
kilns using this waiver, we allow kilns to assume mercury in the raw 
material is present at one-half the detection limit whenever the raw 
materials feedstream analysis determines that mercury is not present at 
detectable levels. The details of this provision are presented in Part 
Five, Section X.B. For a discussion of the other two methods that can 
be used to comply with the mercury emission standard, see Part Five, 
Section VII.B.6.
    For today's rule we use a revised engineering evaluation and data 
analysis method to establish the MACT floor emission level for mercury. 
The approach used to establish MACT floors for the three metal 
hazardous air pollutant groups and hydrochloric acid/chlorine gas is 
the aggregate feedrate approach. Using this approach, the resulting 
mercury floor emission level is 47 g/dscm.
    We estimate that approximately 75 percent of lightweight aggregate 
kiln sources currently are meeting the floor emission level. The 
national annualized compliance cost for lightweight aggregate kilns to 
reduce mercury emissions to comply with the floor emission level is 
$0.7 million for the entire hazardous waste burning lightweight 
aggregate kiln industry, and will reduce mercury emissions by 
approximately 0.03 Mg/yr or 47 percent from current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 NPRM, we considered a beyond-the-floor 
standard based on flue gas temperature reduction to 400 deg.F or less 
followed by activated carbon injection, but determined that a beyond-
the-floor level would not be cost-effective and therefore warranted. 
(61 FR at 17404.) In the May 1997 NODA, we considered a beyond-the-
floor standard of 15 g/dscm based on an activated carbon 
injection. However, we indicated in the NODA that a beyond-the-floor 
standard would not likely be justified given the high cost of treatment 
and the relatively small amount of mercury removed from air emissions. 
(62 FR at 24232.)
    In developing the final rule, we identified three techniques for 
control of mercury as a basis to evaluate a beyond-the-floor standard: 
(1) Activated carbon injection; (2) limiting the feed of mercury in the 
hazardous waste; and (3) limiting the feed of mercury in the raw 
materials. The results of each analysis are discussed below.
    Activated Carbon Injection. To investigate this beyond-the-floor 
control option, we applied a carbon injection capture efficiency of 80 
percent to the floor emission level of 47 g/dscm. The 
resulting beyond-the-floor emission level is 10 g/dscm.
    The national incremental annualized compliance cost for lightweight 
aggregate kilns to meet this beyond-the-floor level rather than comply 
with the floor controls would be approximately $0.6 million for the 
entire hazardous waste burning lightweight aggregate kiln industry and 
would provide an incremental reduction in mercury emissions beyond the 
MACT floor controls of 0.02 Mg/yr. Based on these costs of 
approximately $34 million per additional Mg of mercury removed and the 
small emissions reductions that would be realized, we conclude that 
this mercury beyond-the-floor option for hazardous waste burning 
lightweight aggregate kilns is not acceptably cost-effective nor 
otherwise justified. Therefore, we do not adopt this beyond-the-floor 
standard.
    Limiting the Feedrate of Mercury in Hazardous Waste. We also 
considered, but rejected, a beyond-the-floor emission level based on 
limiting the feed of mercury in the hazardous waste. This mercury 
beyond-the-floor option for lightweight aggregate kilns is not 
warranted because data submitted by commenters indicate that 
approximately 90% of the hazardous waste burned by lightweight 
aggregate kilns contains mercury at levels below method detection 
limits. We conclude from these data that there are little additional 
mercury reductions possible by reducing the feed of mercury in the 
hazardous waste. Therefore, we are not adopting a beyond-the-floor 
emission level because it will not be cost-effective due to the 
relatively small amount of mercury removed from air emissions and 
likely problems with method detection limitations.
    Limiting the Feedrate of Mercury in Raw Materials. A source can 
achieve a reduction in mercury emissions by substituting a feed 
material containing lower levels of mercury for a primary raw material 
higher mercury levels. This beyond-the-floor option appears to be less 
cost effective compared to either of the options evaluated above. 
Because lightweight aggregate kilns are sited proximate to primary raw 
material supply and transporting large quantities of an alternative 
source of raw material(s) is expected to be cost prohibitive. 
Therefore, we do not adopt this mercury beyond-the-floor standard.
    Thus, the promulgated mercury standard for existing hazardous waste 
burning lightweight aggregate kilns is the floor emission level of 47 
g/dscm.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal, we identified floor control for new sources as hazardous 
waste feedrate control of mercury not to exceed a feedrate level of 17 
g/dscm expressed as a maximum theoretical emissions 
concentration. We proposed a floor emission level of 72 g/
dscm. (61 FR at 17408.) In May 1997 NODA, we conducted a breakpoint 
analysis on ranked mercury emissions data from sources utilizing the 
MACT floor technology and established the floor emission level as the 
test condition average of the breakpoint source. The breakpoint 
analysis was intended to reflect an engineering-based evaluation of the 
data so that the one lightweight aggregate kiln spiking extra mercury 
during testing procedures did not drive the floor emission level to 
levels higher than the preponderance of the emissions data. This 
analysis resulted in a MACT floor level of 47 g/dscm. (62 FR 
at 24233.)
    For the final rule, we identify floor control for new lightweight 
aggregate kilns as feed control of mercury in the hazardous waste, 
based on the single source with the best aggregate feedrate of mercury 
in hazardous waste. Using the aggregate feedrate approach to establish 
this floor level of control and corresponding floor emission level, we 
identify a MACT floor emission level of 33 g/dscm for new 
lightweight aggregate kilns.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
both the proposal and the NODA, we considered a beyond-the-floor 
standard for new sources based on activated carbon injection, but 
determined that it would not be cost-effective to adopt the beyond-the-
floor standard given the high cost of treatment and the relatively 
small amount of mercury removed from air emissions. (61 FR at 17408 and 
62 FR at 24233.)
    In the final rule, we identified three techniques for control of 
mercury as a basis to evaluate a beyond-the-floor standard: (1) 
Activated carbon injection; and (2) limiting the feed of mercury in the 
hazardous waste. The results of each analysis are discussed below.
    Activated Carbon Injection. As discussed above, we conclude that 
flue gas temperature reduction to 400  deg.F followed by activated 
carbon injection to remove mercury is an appropriate beyond-the-floor 
control option for improved mercury control at

[[Page 52895]]

lightweight aggregate kilns. The control of flue gas temperature is 
necessary to ensure good collection efficiency. Based on the MACT floor 
emission level of 33 g/dscm and assuming a carbon injection 
capture efficiency of 80 percent, we identified a beyond-the-floor 
emission level of 7 g/dscm. As discussed above for existing 
sources, we do not believe that a beyond-the-floor standard of 7 
g/dscm is warranted for new lightweight aggregate kilns due to 
the high cost of treatment and relatively small amount of mercury 
removed from air emissions. The incremental annualized compliance cost 
for one new lightweight aggregate kiln to meet this beyond-the-floor 
level, rather than comply with floor controls, would be approximately 
$0.46 million and would provide an incremental reduction in mercury 
emissions beyond the MACT floor controls of approximately 0.008 Mg/yr. 
Based on these costs of approximately $58 million per additional Mg of 
mercury removed, a beyond-the-floor standard of 7 g/dscm is 
not warranted due to the high cost of compliance and relatively small 
mercury emissions reductions. Notwithstanding our goal of reducing the 
loading to the environment by bioaccumulative pollutants such as 
mercury whenever possible, these costs are not justified.
    Limiting the Feedrate of Mercury in Hazardous Waste. As discussed 
above for existing sources, we conclude that a beyond-the-floor based 
on limiting the feed of mercury in the hazardous waste is not 
justified. Considering that the floor emission level for new 
lightweight aggregate kilns is approximately one third lower than the 
floor emission level for existing kilns (33 versus 47 g/dscm), 
we again conclude that a mercury beyond-the-floor standard is not 
warranted because emission reductions of mercury would be less than 
existing sources at comparable costs. Thus, the cost-effectiveness is 
higher for new kilns than for existing kilns. Further, achieving 
substantial additional mercury reductions by further controls on 
hazardous waste feedrate may be problematic because the mercury 
contribution from raw materials and coal represents an even larger 
proportion of the total mercury fed to the kiln. Therefore, we do not 
adopt a mercury beyond-the-floor standard based on limiting feed of 
mercury in hazardous waste for new sources.
    Thus, the promulgated mercury standard for new hazardous waste 
burning lightweight aggregate kilns is the floor emission level of 33 
g/dscm.
4. What Are the Particulate Matter Standards?
    We establish standards for both existing and new lightweight 
aggregate kilns that limit particulate matter emissions to 57 mg/dscm. 
The particulate matter standard is a surrogate control for the metals 
antimony, cobalt, manganese, nickel, and selenium. We refer to these 
five metals as ``nonenumerated metals'' because standards specific to 
each metal have not been established. The rationale for adopting these 
standards is discussed below.
    a. What Is the MACT Floor for Existing Sources? In the April 1996 
NPRM, we defined floor control based upon the performance of a fabric 
filter with an air-to-cloth ratio of 2.8 acfm/ft2. The MACT 
floor was 110 mg/dscm (0.049 gr/dscf). (61 FR at 17403.) In the May 
1997 NODA, we defined the technology basis as a fabric filter for a 
MACT floor, but did not characterize the design and operation 
characteristics of the particulate matter control equipment, air-to-
cloth ratio of a fabric filter, because we had limited information on 
these parameters. (62 FR at 24233.) Instead, for each particulate 
matter test condition, we evaluated the corresponding semivolatile 
metal system removal efficiency and screened out sources with 
relatively poor system removal efficiencies as a means to identify and 
eliminate from consideration those sources not using MACT floor 
control. Our reevaluation of the lightweight aggregate kiln particulate 
matter data resulted in a MACT floor of 50 mg/dscm (0.022 gr/dscf).
    Some commenters state that a floor emission level of 50 mg/dscm 
(0.022 gr/dscf) is too high and a particulate matter standard of 23 mg/
dscm (0.010 gr/dscf) is more appropriate because it is consistent with 
the level of performance achieved by incinerators using fabric filters. 
Even though we agree that well designed and properly operated fabric 
filters in use at all lightweight aggregate kilns can achieve low 
levels, we are concerned that an emission level of 23 mg/dscm would not 
be appropriate given the high inlet grain loading inherent with the 
lightweight aggregate manufacturing process, typically much higher than 
the particulate loading to incinerators.
    Commenters also express concern that the Agency identified 
separate, different MACT pools and associated MACT controls for 
particulate matter, semivolatile metals, and low volatile metals, even 
though all three are controlled, at least in part, by the particulate 
matter control device. These commenters stated that our approach is 
likely to result in three different design specifications. We agree 
with these commenters and, in the final rule, the same initial MACT 
pool is used to establish the floor levels for particulate matter, 
semivolatile metals, and low volatile metals. See discussion in Part 
Four, Section V.
    For the final rule, we conclude that the general floor methodology 
discussed in the May 1997 NODA is appropriate. MACT control for 
particulate matter is based on the performance of fabric filters. Since 
we lack data to fully characterize control equipment from all sources 
and we lack information on the relationship between the design 
parameters and the system performance, we evaluated both low and 
semivolatile metal system removal efficiencies associated with the 
source's particulate matter emissions to identify those sources not 
using MACT floor control. Our data show that all lightweight aggregate 
kilns are achieving greater than 99 percent system removal efficiency 
for both low and semivolatile metals, with some attaining 99.99 percent 
removal. Since we found no sources with system removal efficiencies 
indicative of poor performance, we conclude that all lightweight 
aggregate kilns are using MACT controls and the floor emission limit is 
identified as 57 mg/dscm (0.025 gr/dscf).
    The performance level of 57 mg/dscm is generally consistent with 
that expected from well designed and operated fabric filters, and that 
achieved by other similar types of combustion sources operating with 
high inlet grain loadings. We have particulate matter data from all 
lightweight aggregate kiln sources, and multiple test conditions, 
conducted at 3 year intervals, are available for many of the sources. 
We conclude that the number of test conditions available adequately 
covers the range of variability of well operated and designed fabric 
filters.166
---------------------------------------------------------------------------

    \166\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    We considered, but rejected, basing the particulate matter floor 
for lightweight aggregate kilns on the New Source Performance Standard. 
The New Source Performance Standard limits particulate matter emissions 
to 92 mg/dscm (0.040 gr/dscf), uncorrected for oxygen. (See 40 CFR 
60.730, Standards of Performance for Calciners and Dryers in Mineral 
Industries.) We rejected the New Source Performance Standard as the 
basis for the floor emission level

[[Page 52896]]

because our MACT analysis of data from existing sources indicates that 
a particulate matter floor level lower than the New Source Performance 
Standard is currently being achieved by existing hazardous waste 
burning lightweight aggregate kilns. Further, all available emission 
data for hazardous waste burning lightweight aggregate kilns are well 
below the New Source Performance Standard particulate matter standard. 
Thus, the particulate matter floor emission level is 57 mg/dscm based 
on an analysis of existing emissions data.
    We estimate that, based on a design level of 70 percent of the 
standard, over 90 percent of lightweight aggregate kiln sources 
currently are meeting the floor level. The national annualized 
compliance cost for lightweight aggregate kilns to reduce particulate 
matter emissions to comply with the floor emission level is $18,000 for 
the entire hazardous waste burning lightweight aggregate kiln industry, 
and our floor will reduce nonenumerated metals and particulate matter 
emissions by 0.01 Mg/yr and 2.7 Mg/yr, respectively, or 7 percent from 
current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the NPRM, we proposed a beyond-the-floor emission level of 
69 mg/dscm (0.030 gr/dscf) and solicited comment on an alternative 
beyond-the-floor emission level of 34 mg/dscm (0.015 gr/dscf) based on 
improved particulate matter control. (61 FR at 17403.) In the May 1997 
NODA, we concluded that a beyond-the-floor standard may not be 
warranted given a reduced particulate matter floor level compared to 
the proposed floor emission level. (62 FR at 24233.)
    In the final rule, we considered a beyond-the-floor level of 34 mg/
dscm for existing lightweight aggregate kilns based on improved 
particulate matter control. For analysis purposes, improved particulate 
matter control entails the use of higher quality fabric filter bag 
material. We then determined the cost of achieving this level of 
particulate matter, with corresponding reductions in the nonenumerated 
metals for which particulate matter is a surrogate, to determine if 
this beyond-the-floor level would be appropriate. The national 
incremental annualized compliance cost for lightweight aggregate kilns 
to meet this beyond-the-floor level, rather than comply with the floor 
controls, would be approximately $110,000 for the entire hazardous 
waste burning lightweight aggregate kiln industry and would provide an 
incremental reduction in nonenumerated metals emissions nationally 
beyond the MACT floor controls of 0.03 Mg/yr. Based on these costs of 
approximately $3.7 million per additional Mg of nonenumerated metals 
emissions removed, we conclude that this beyond-the-floor option for 
lightweight aggregate kilns is not acceptably cost-effective nor 
otherwise justified. Therefore, we do not adopt this beyond-the-floor 
standard. Thus, the promulgated particulate matter standard for 
existing hazardous waste burning lightweight aggregate kilns is the 
floor emission level of 57 mg/dscm.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal, we defined floor control for new sources based on the level 
of performance of a fabric filter with an air-to-cloth ratio of 1.5 
acfm/ft2. The MACT floor emission level was 120 mg/dscm (0.054 gr/
dscf). (61 FR at 17408.) In the May 1997 NODA, MACT control was defined 
as a well-designed and properly operated fabric filter, and the floor 
emission level for new lightweight aggregate kilns was 50 mg/dscm 
(0.022 gr/dscf). (62 FR at 24233.)
    All lightweight aggregate kilns use fabric filters to control 
particulate matter. As discussed earlier, we have limited information 
on the design and operation characteristics of existing control 
equipment currently used by lightweight aggregate kilns. As a result, 
we are unable to identify a specific technology that can consistently 
achieve lower emission levels than the controls used by lightweight 
aggregate kilns achieving the MACT floor level for existing sources. 
Lightweight aggregate kilns achieve the floor emission level with well-
designed and properly operated fabric filters. Thus, floor control for 
new kilns is likewise a well-designed and properly operated fabric 
filter. Therefore, as discussed for existing sources, the MACT floor 
level for new lightweight aggregate kilns is 57 mg/dscm (0.025 gr/
dscf).
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 NPRM, we proposed a beyond-the-floor standard of 69 mg/
dscm (0.030 gr/dscf) based on improved particulate matter control, 
which was consistent with existing sources. (61 FR at 17408.) In the 
May 1997 NODA, we concluded, as we did for existing sources, that a 
beyond-the-floor level for particulate matter may not be warranted due 
to the high costs of control and relatively small amount of particulate 
matter removed from air emissions. (62 FR at 24233.)
    As discussed for existing sources, we considered a beyond-the-floor 
level of 34 mg/dscm for new lightweight aggregate kilns based on 
improved particulate matter control. For analysis purposes, improved 
particulate matter control entails the use of higher quality fabric 
filter bag material. We then determined the cost of achieving this 
level of particulate matter, with corresponding reductions in the 
nonenumerated metals for which particulate matter is a surrogate, to 
determine if this beyond-the-floor level would be appropriate. The 
incremental annualized compliance cost for one new lightweight 
aggregate kiln to meet this beyond-the-floor level, rather than comply 
with floor controls, would be approximately $38 thousand and would 
provide an incremental reduction in nonenumerated metals emissions of 
approximately 0.012 Mg/yr.167 Based on these costs of 
approximately $3.1 million per additional Mg of nonenumerated metals 
removed, we conclude that a beyond-the-floor standard of 34 mg/dscm is 
not justified due to the high cost of compliance and relatively small 
nonenumerated metals emission reductions. Further, a standard of 57 mg/
dscm would adequately control the unregulated hazardous air pollutant 
metals for which it is being used as a surrogate. Thus, the particulate 
matter standard for new lightweight aggregate kilns is the floor level 
of 57 mg/dscm.
---------------------------------------------------------------------------

    \167\ Based on the data available, the average emissions in sum 
of the five nonenumerated metal from lightweight aggregate kilns 
using MACT particulate matter control is approximately 83 
g/dscm. To estimate emission reductions of the 
nonenumerated metals, we assume a linear relationship between a 
reduction in particulate matter and these metals.
---------------------------------------------------------------------------

5. What Are the Semivolatile Metals Standards?
    In the final rule, we establish a standard for existing and new 
lightweight aggregate kilns that limits semivolatile metal emissions to 
250 and 43 g/dscm, respectively. The rationale for adopting 
these standards is discussed below.
    a. What Is the MACT Floor for Existing Sources? All lightweight 
aggregate kilns use a combination of particulate matter control, i.e., 
a fabric filter, and hazardous waste feedrate to control emissions of 
semivolatile metals. Current RCRA regulations establish limits on the 
maximum feedrate of lead and cadmium in all feedstreams. Thus, 
hazardous waste feedrate control is part of MACT floor control.
    In the April 1996 proposal, we defined floor control as either (1) 
a fabric filter with an air-to-cloth ratio of 1.5 acfm/ft 2 
and a hazardous waste feedrate level of 270,000 g/dscm,

[[Page 52897]]

expressed as a maximum theoretical emissions concentration; or (2) a 
combination of a fabric filter and venturi scrubber with an air-to-
cloth ratio of 4.2 acfm/ft 2 and a hazardous waste feedrate 
level of 54,000 g/dscm. The proposed floor emission level was 
12 g/dscm. (61 FR at 17405.) In the May 1997 NODA, we 
discussed a floor methodology where we used a breakpoint analysis to 
identify sources that were not using floor control with respect either 
to semivolatile metals hazardous waste feedrate or emissions control. 
Under this approach, we ranked semivolatile metal emissions data from 
sources that were achieving the particulate matter floor level of 50 
mg/dscm or better. We identified the floor level as the test condition 
average associated with the breakpoint source. Thus, sources with 
atypically high emissions because of high semivolatile feedrate levels 
or poor semivolatile metals control were screened from the pool of 
sources used to define the floor emission level. Based on this 
analysis, we identified a floor emission level of 76 g/dscm. 
(62 FR at 24234.)
    We received few public comments in response to the proposal and May 
1997 NODA concerning the lightweight aggregate kiln semivolatile metals 
floor emission level. We did receive comments on the application of 
techniques to identify breakpoints in the arrayed emissions data. This 
issue and our response to it are discussed in the floor methodology 
section in Part Four, Section V. We also received comments that our 
semivolatile metals analysis in the proposal and May 1997 NODA included 
several data base inaccuracies that, when corrected, would result in a 
higher floor level. We agree with the commenters and we revised the 
data base as necessary for the final rule analysis.
    In the final rule, in general response to these comments, we use a 
revised engineering evaluation and data analysis method to establish 
the floor emission level for semivolatile metals. We use the aggregate 
feedrate approach in conjunction with floor control for particulate 
matter of 57 mg/dscm to identify a semivolatile metal floor emission 
level of 1,700 g/dscm. We estimate that all lightweight 
aggregate kiln sources currently are meeting the floor level.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 NPRM, we considered a beyond-the-floor 
emission level for semivolatile metals based on improved particulate 
matter control. We concluded that a beyond-the-floor emission level 
would not be cost-effective given that the proposed semivolatile metal 
floor level of 12 g/dscm alone would result in an estimated 97 
percent reduction in semivolatile metal emissions. (61 FR at 17405.) In 
the May 1997 NODA, we considered a beyond-the-floor emission level 
based on improved particulate matter control, but indicated that such a 
standard was not likely to be cost-effective due to the high costs of 
control. (62 FR at 24234.)
    In developing the final rule, we identified three techniques for 
control of semivolatile metals as a basis to evaluate a beyond-the-
floor standard: (1) Limiting the feed of semivolatile metals in the 
hazardous waste; (2) improved particulate matter control; and (3) 
limiting the feed of semivolatile metals in the raw materials. The 
results of each analysis are discussed below.
    Limiting the Feedrate of Semivolatile Metals in Hazardous Waste. 
Under this option, as with cement kilns, we selected for evaluation a 
beyond-the-floor emission level of 240 g/dscm to evaluate from 
among the range of possible levels that reflect improved feedrate 
control of semivolatile metals in hazardous waste. This emission level 
represents a significant increment of emission reduction from the floor 
level of 1700 g/dscm, it is within the range of levels that 
are likely to be reasonably achievable using feedrate control, and it 
is generally consistent with the incinerator and cement kiln standards, 
thereby advancing a policy objective of essentially common standards 
among combustors of hazardous waste.
    In performing an analysis of the 240 g/dscm beyond-the-
floor limit, we found that additional reductions beyond 250 g/
dscm represent a significant reduction in cost-effectiveness of 
incremental beyond-the-floor levels. A beyond-the-floor standard of 250 
g/dscm achieves the same goals as a beyond-the-floor standard 
of 240 g/dscm in a more cost-effective manner. The national 
incremental annualized compliance cost for the lightweight aggregate 
kilns to meet this 250 g/dscm beyond-the-floor level, rather 
than comply with the floor controls, would be approximately $88,000 and 
would provide an incremental reduction beyond emissions at the MACT 
floor in semivolatile metal emissions of an additional 0.17 Mg/yr. The 
cost-effectiveness of this emission level is approximately $530,000 per 
additional Mg of semivolatile metal removed.
    We conclude that additional control of the feedrate of semivolatile 
metals in hazardous waste to achieve an emission level of 250 
g/dscm is warranted because this standard would reduce lead 
and cadmium emissions, which are particularly toxic hazardous air 
pollutants. In addition, Solite Corporation, which operates the 
majority of the hazardous waste burning lightweight aggregate kilns, 
stated in their public comments that a standard of 213 g/dscm 
is achievable and adequately reflects the variability of lead and 
cadmium in raw material for their kilns. Further, the vast majority of 
the lead and cadmium fed to the lightweight aggregate kiln is from the 
hazardous waste,168 not from the raw material or coal. We 
are willing to accept a more marginal cost-effectiveness for sources 
voluntarily burning hazardous waste in lieu of other fuels to ensure 
that sources are using best controls.
---------------------------------------------------------------------------

    \168\ USEPA, ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies'', July 1999.
---------------------------------------------------------------------------

    Moreover, this beyond-the-floor semivolatile metal standard better 
supports our Children's Health Initiative in that lead emissions, which 
are of highest significance to children's health, will be reduced by 
another 60 percent from today's baseline. We are committed to reducing 
lead emissions wherever and whenever possible. Finally, we note that 
this beyond-the-floor standard is also consistent with European Union 
standards for hazardous waste incinerators of approximately 200 
g/dscm for lead and cadmium combined. Therefore, we are 
adopting today a beyond-the-floor standard of 250 g/dscm for 
existing lightweight aggregate kilns.
    Improved Particulate Matter Control. We also evaluated improved 
particulate matter control as another beyond-the-floor control option 
for improved semivolatile metals control. We investigated a beyond-the-
floor standard of 250 g/dscm, an emission level consistent 
with the preferred option based on limiting the feedrate of 
semivolatile metals in hazardous waste. The national incremental 
annualized compliance cost for lightweight aggregate kilns to meet this 
beyond-the-floor level, rather than comply with the floor controls, 
would be approximately $88,000 thousand for all lightweight aggregate 
kilns and would provide an incremental reduction in semivolatile metal 
emissions beyond the MACT floor controls of 0.17 Mg/yr. Based on these 
costs of approximately $530,000 per additional Mg of semivolatile metal 
removed, we determined that this beyond-the-floor option may be 
warranted. However, as discussed below, the cost-effectiveness for this 
beyond-the-floor option is approximately equivalent to the costs

[[Page 52898]]

estimated for a beyond-the-floor option based on limiting the feed of 
semivolatile metals in the hazardous waste. We decided to base the 
beyond-the-floor standard for semivolatile metals on the feedrate 
option to be consistent with the cement kiln approach. Of course light-
weight aggregate kilns are free to choose to improve particulate matter 
control in lieu of feedrate controls as their vehicle to achieve 
compliance with 250 ug/dscm.
    Limiting the Feedrate of Semivolatile Metals in Raw Materials. A 
source can achieve a reduction in semivolatile metals emissions by 
substituting a feed material containing lower levels of lead and/or 
cadmium for a primary raw material higher in lead and/or cadmium 
levels. This beyond-the-floor option appears to be less cost effective 
compared to either of the options evaluated above because lightweight 
aggregate kilns are sited proximate to primary raw material supply. 
Transporting large quantities of an alternative source of raw 
material(s) is expected to be cost prohibitive. Therefore, we do not 
adopt this semivolatile metal beyond-the-floor standard.
    Thus, the promulgated semivolatile metals standard for existing 
hazardous waste burning lightweight aggregate kilns is a beyond-the-
floor standard of 250 g/dscm based on limiting the feedrate of 
semivolatile metals in the hazardous waste.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal, we defined floor control as a fabric filter with an air-to-
cloth ratio of 1.5 acfm/ft2 and a hazardous waste feedrate 
level of 270,000 g/dscm, expressed as a maximum theoretical 
emissions concentration. The proposed floor emission level was 5.2 
g/dscm. (61 FR at 17408.) In the May 1997 NODA, we concluded 
that the floor control and emission level for existing sources for 
semivolatile metals would also be appropriate for new sources. Floor 
control was based on a combination of good particulate matter control 
and limiting hazardous waste feedrates of semivolatile metals to 
control emissions. We used a breakpoint analysis of the semivolatile 
metal emissions data to exclude sources achieving substantially poorer 
semivolatile metal control than the majority of sources. The NODA floor 
emission level was 76 g/dscm for new sources. (62 FR at 
24234.)
    In the final rule, as discussed previously, we use a revised 
engineering evaluation and data analysis method to establish the floor 
emission level for semivolatile metals. We use the aggregate feedrate 
approach in conjunction with floor control for particulate matter of 57 
mg/dscm to identify a semivolatile metal floor emission level of 43 
g/dscm.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 NPRM and May 1997 NODA, we considered a semivolatile 
metal beyond-the-floor emission level for new sources, but determined 
that the standard would not be cost-effective because the floor 
emission levels already achieved significant reductions in semivolatile 
metals emissions. (61 FR at 17408 and 62 FR at 24234.)
    For the final rule, we do not adopt a beyond-the-floor emission 
level because the MACT floor for new sources is already substantially 
lower than the beyond-the-floor emission standard for existing sources. 
As a result, a beyond-the-floor standard for new lightweight aggregate 
kilns is not warranted due to the high costs of control versus the 
minimal emissions reductions that would be achieved. Therefore, we 
adopt the semivolatile metal MACT floor standard of 43 g/dscm 
for new hazardous waste burning lightweight aggregate kilns.
6. What Are the Low Volatile Metals Standards?
    In the final rule, we establish a standard for both existing and 
new lightweight aggregate kilns that limits low volatile metal 
emissions to 110 g/dscm. The rationale for adopting these 
standards is discussed below.
    a. What Is the MACT Floor for Existing Sources? In the April 1996 
proposal, we defined floor control based on the performance of a fabric 
filter with an air-to-cloth ratio of 1.8 acfm/ft2 and a 
hazardous waste feedrate level of 46,000 g/dscm, expressed as 
a maximum theoretical emissions concentration. The proposed floor 
emission level was 340 g/dscm. (61 FR at 17405.) In the May 
1997 NODA, we discussed a floor methodology where we used a breakpoint 
analysis to identify sources that were not using floor control with 
respect either to low volatile metals hazardous waste feedrate or 
emissions control. Under this approach, we ranked low volatile metal 
emissions data from sources that were achieving the particulate matter 
floor level of 50 mg/dscm or better. We identified the floor level as 
the test condition average associated with the breakpoint source. Thus, 
sources with atypically high emissions because of high low volatile 
feedrate levels or poor low volatile metals control were screened from 
the pool of sources used to define the floor emission level. Based on 
this analysis, we identified a floor emission level of 37 g/
dscm. (62 FR at 24234.)
    We received few comments, in response to the April 1996 NPRM and 
May 1997 NODA, concerning the low volatile metals floor emission level. 
We received comments, however, on several overarching issues including 
the appropriateness of considering feedrate control of metals 
(including low volatile metals) in hazardous waste as a MACT floor 
control technique and the specific procedure of identifying breakpoints 
of arrayed emissions data. These issues and our responses to them are 
discussed in the floor methodology section in Part Four, Section V.
    For today's rule, we use a revised engineering evaluation and data 
analysis method to establish the MACT floor level for low volatile 
metals. The aggregate feedrate approach in conjunction with MACT 
particulate matter control to 57 mg/dscm results in a low volatile 
metal floor emission level of 110 g/dscm.
    We estimate that over 80 percent of existing lightweight aggregate 
kiln sources in our data base meet the floor level. The national 
annualized compliance cost for lightweight aggregate kilns to reduce 
low volatile metal emissions to comply with the floor emission level is 
$52,000 for the entire hazardous waste burning lightweight aggregate 
kiln industry, and will reduce low volatile metal emissions by 0.04 Mg/
yr or 40 percent from current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 NPRM and May 1997 NODA, we considered a 
beyond-the-floor standard for low volatile metals based on improved 
particulate matter control. However, we concluded that a beyond-the-
floor standard would not be cost-effective due to the high cost of 
emissions control and relatively small amount of low volatile metals 
removed from air emissions. (61 FR at 17406 and 62 FR at 24235.)
    For today's rule, we identified three potential beyond-the-floor 
techniques for control of low volatile metals: (1) Improved particulate 
matter control; (2) limiting the feed of low volatile metals in the 
hazardous waste; and (3) limiting the feed of low volatile metals in 
the raw materials. The results of each analysis are discussed below.
    Improved Particulate Matter Control. Our judgment is that a beyond-
the-floor standard based on improved particulate matter control would 
be less cost-effective that a beyond-the-floor option based on limiting 
the feedrate of low

[[Page 52899]]

volatile metals in the hazardous waste. Our data show that lightweight 
aggregate kilns are already achieving a 99.9% system removal efficiency 
of low volatile metals and some sources are even attaining 99.99%. 
Thus, pollution control equipment retrofit costs for improved control 
would be significant. Thus, we conclude a beyond-the-floor emission 
level for low volatile metals based on improved particulate matter 
control for lightweight aggregate kilns is not warranted.
    Limiting the Feedrate of Low Volatile Metals in the Hazardous 
Waste. We also considered a beyond-the-floor level of 70 g/
dscm based on additional feedrate control of low volatile metals in the 
hazardous waste. Our investigation shows that this beyond-the-floor 
option would achieve an incremental reduction in low volatile metals of 
only 0.01 Mg/yr. Given that this beyond-the-floor level would not 
achieve appreciable emissions reductions, significant cost-
effectiveness considerations would likely arise, thus suggesting that 
this beyond-the-floor standard is not warranted.
    Limiting the Feedrate of Low Volatile Metals in Raw Materials. A 
source can achieve a reduction in low volatile metal emissions by 
substituting a feed material containing lower levels of these metals 
for a primary raw material higher low volatile metal levels. This 
beyond-the-floor option appears to be less cost-effective compared to 
either of the options evaluated above because lightweight aggregate 
kilns are sited proximate to primary raw material supply. Transporting 
large quantities of an alternative source of raw material(s) is 
expected to be very costly and not cost-effective considering the 
limited emissions reductions that would be achieved. Therefore, we do 
not adopt this low volatile metals beyond-the-floor standard.
    For reasons discussed above, we do not adopt a beyond-the-floor 
level for low volatile metals, and establish the emissions standard for 
existing hazardous waste burning lightweight aggregate kilns at 110 
g/dscm.
    c. What Is the MACT Floor for New Sources? At proposal, we defined 
floor control based on the performance of a fabric filter with an air-
to-cloth ratio of 1.3 acfm/ft2 a hazardous waste feedrate 
level of 37,000 g/dscm, expressed as a maximum theoretical 
emissions concentration. The proposed floor level was 55 g/
dscm. (61 FR at 17408.) In the May 1997 NODA, we concluded that the 
floor control and emission level for existing sources for low volatile 
metals would also be appropriate for new sources. Floor control was 
based on a combination of good particulate matter control and limiting 
hazardous waste feedrate of low volatile metals to control emissions. 
We used a breakpoint analysis of the low volatile metal emissions data 
to exclude sources achieving substantially poorer low volatile metal 
control than the majority of sources. The NODA floor was 37 g/
dscm. (62 FR at 24235.)
    In the final rule, in response to general comments on the May 1997 
NODA, we use a revised engineering evaluation and data analysis method 
to establish the floor emission level for low volatile metals. We use 
the aggregate feedrate approach in conjunction with floor control for 
particulate matter of 57 mg/dscm to identify a low volatile metal floor 
emission level of 110 g/dscm.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 NPRM and May 1997 NODA, we considered a low volatile 
metal beyond-the-floor level, but determined that a beyond-the-floor 
standard would not be cost-effective due to the high cost of treatment 
and relatively small amount of low volatile metals removed from air 
emissions. We received no comments to the contrary.
    For the final rule, as discussed for existing sources, we do not 
adopt a beyond-the-floor level for new sources, and conclude that the 
floor emission level is appropriate. Therefore, we adopt the low 
volatile metal floor level of 110 g/dscm as the emission 
standard for new hazardous waste burning lightweight aggregate kilns.
7. What Are the Hydrochloric Acid and Chlorine Gas Standards?
    In the final rule, we establish a standard for existing and new 
lightweight aggregate kilns that limits hydrochloric acid and chlorine 
gas emissions to 230 and 41 ppmv, respectively. The rationale for 
adopting these standards is discussed below.
    a. What Is the MACT Floor for Existing Sources? In the April 1996 
proposal, we identified floor control for hydrochloric acid/chlorine 
gas as either: (1) Hazardous waste feedrate control of chlorine to 1.5 
g/dscm, expressed as a maximum theoretical emissions concentration; or 
(2) a combination of a venturi scrubber and hazardous waste feedrate 
level of 14 g/dscm, expressed as a maximum theoretical emissions 
concentration. The proposed floor emission level was 2100 ppmv. (61 FR 
at 17406.) In the May 1997 NODA, we used the same data analysis method 
as proposed, except that a computed emissions variability factor was no 
longer added. The floor emission level was 1300 ppmv. (62 FR at 24235.)
    We received few comments concerning the hydrochloric acid/chlorine 
gas floor methodology and emission level. One commenter supports the 
use of a variability factor in calculating the floor emission level. 
Generally, the final emission standards, including hydrochloric acid/
chlorine gas, already accounts for emissions variability without adding 
a statistically-derived emissions variability factor. This issue and 
our response to it are discussed in detail in the floor methodology 
section in Part Four, Section V.
    For today's rule, we use a revised engineering evaluation and data 
analysis method to establish the MACT floor level for hydrochloric acid 
and chlorine gas. The aggregate feedrate approach results in a floor 
emission level of 1500 ppmv.
    We estimate that approximately 31 percent of lightweight aggregate 
kilns in our data base currently meet the floor emission level. The 
national annualized compliance cost for sources to reduce hydrochloric 
acid and chlorine gas emissions to comply with the floor level is 
$350,000 for the entire hazardous waste burning lightweight aggregate 
kiln industry, and will reduce hydrochloric acid and chlorine gas 
emissions by 182 Mg/yr or 10 percent from current baseline emissions.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 proposal, we defined beyond-the-floor 
control as wet or dry lime scrubbing with a control efficiency of 90 
percent. We proposed a beyond-the-floor standard of 450 ppmv, which 
included a statistical variability factor. (61 FR at 17406.) In the May 
1997 NODA, the beyond-the-floor standard was 130 ppmv based on wet or 
dry scrubbing with a control efficiency of 90 percent. (62 FR at 
24235.)
    We identified three potential beyond-the-floor techniques for 
control of hydrochloric acid and chlorine gas emissions: (1) Dry lime 
scrubbing; (2) limiting the feed of chlorine in the hazardous waste; 
and (3) limiting the feed of chlorine in the raw materials. The result 
of each analysis is discussed below.
    Dry Lime Scrubbing. Based on a joint emissions testing program with 
Solite Corporation in 1997, dry lime scrubbing at a stoichiometric lime 
ratio of 3:1 achieved greater than 85 percent removal of hydrochloric 
acid and chlorine gas. For the final rule, we considered a beyond-the-
floor emission level of 230 ppmv based on a 85 percent removal 
efficiency from the floor level of 1500 ppmv.

[[Page 52900]]

    The national incremental annualized compliance cost for all 
lightweight aggregate kilns to meet this beyond-the-floor level is 
approximately $1.5 million. This would provide an incremental reduction 
in hydrochloric acid/chlorine gas emissions beyond the MACT floor 
controls of an additional 1320 Mg/yr, or 80 percent. Based on these 
costs of approximately $1,100 per additional Mg hydrochloric acid/
chlorine gas removed, this hydrochloric acid/chlorine gas beyond-the-
floor option for lightweight aggregate kilns is justified. Therefore, 
we are adopting a beyond-the-floor standard of 230 ppmv for existing 
lightweight aggregate kilns.
    One commenter disagreed with our proposal to base the beyond-the-
floor standard on dry lime scrubbing achieving 90% removal. The 
commenter states that dry lime scrubbing cannot cost-effectively 
achieve 90 percent control of hydrochloric acid and chlorine gas 
emissions. To achieve a 90 percent capture efficiency at a 
stoichiometric ratio of 3:1, the commenter maintains that a source 
would need to install special equipment and make operational 
modifications that are less cost-effective than simple dry lime 
scrubbing at a lower removal efficiency. The commenter identifies this 
lower level of control at 80 percent based on the joint emissions 
testing program.169 The commenter does agree, however, that 
dry lime scrubbing can achieve 90 percent capture without the 
installation of special equipment by operating at a stoichiometric lime 
ratio greater than 3:1. One significant consequence of operating at 
higher stoichiometric lime ratios, the commenter states, is the adverse 
impact to the collected particulate matter. Currently, the collected 
particulate matter is recycled into the lightweight aggregate product. 
At higher stoichiometric lime ratios, unreacted lime and collected 
chloride and sulfur salts would prevent this recycling practice and 
would require the disposal of all the collected particulate matter at 
significant and unjustified costs.
---------------------------------------------------------------------------

    \169\ See ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    We agree with the commenter that data from the joint emissions 
testing program does not support a 90 percent capture efficiency by 
simple dry lime scrubbing at a stoichiometric lime ratio of 3:1. We 
disagree with the commenter that the data support an efficiency no 
greater than 80 percent. In the testing program, we evaluated the 
capture efficiency of lime during four runs at a stoichiometric lime 
ratio of approximately 3:1. The results show that hydrochloric acid was 
removed at rates ranging from 86 to 91 percent with one exception. For 
that one run, the removal was calculated as 81 percent. For reasons 
detailed in the Comment Response Document and in the technical support 
document,170 we conclude that the data from this run should 
not be considered because the calculated stoichiometric lime ratio is 
suspect. When we remove this data point from consideration, the 
available information clearly indicates that dry lime scrubbing at a 
stoichiometric ratio of 3:1 can achieve greater than 85 percent 
removal. Therefore, in the final rule, we base the beyond-the-floor 
standard of 230 ppmv on 85 percent removal.
---------------------------------------------------------------------------

    \170\ See ``Final Technical Support Document for HWC MACT 
Standards, Volume III: Selection of MACT Standards and 
Technologies,'' July 1999.
---------------------------------------------------------------------------

    Limiting the Feedrate of Chlorine in the Hazardous Waste. We also 
considered a beyond-the-floor standard for hydrochloric acid/chlorine 
gas based on additional feedrate control of chlorine in the hazardous 
waste. This option achieves lower emission reductions and is less cost-
effective than the dry lime scrubbing option discussed above. 
Therefore, we are not adopting a hydrochloric acid/chlorine gas beyond-
the-floor standard based on limiting the feed of chlorine in the 
hazardous waste.
    Limiting the Feedrate of Chlorine in the Raw Materials. A source 
can achieve a reduction in hydrochloric acid/chlorine gas emissions by 
substituting a feed material containing lower levels of chlorine for a 
primary raw material higher chlorine levels. This beyond-the-floor 
option appears to be less cost effective compared to either of the 
options evaluated above because lightweight aggregate kilns are sited 
proximate to primary raw material supply. Transporting large quantities 
of an alternative source of raw material(s) is expected to be very 
costly and not cost-effective considering the limited emissions 
reductions that would be achieved. Therefore, we do not adopt this 
hydrochloric acid/chlorine gas beyond-the-floor standard.
    In summary, we establish the hydrochloric acid/chlorine gas 
standard for existing lightweight aggregate kilns at 230 ppmv based on 
scrubbing.
    c. What Is the MACT Floor for New Sources? In the April 1996 
proposal, we defined MACT floor control for new sources as a venturi 
scrubber with a hazardous waste feedrate level of 14 g/dscm, expressed 
as a maximum theoretical emissions concentration. We proposed a floor 
emission level of 62 ppmv. (61 FR at 17409.) In the May 1997 NODA, we 
concluded that the floor control and emission level for existing 
sources for hydrochloric acid/chlorine gas would also be appropriate 
for new sources. Floor control was based on limiting hazardous waste 
feedrates of chlorine to control hydrochloric acid/chlorine gas 
emissions. We screened out some data with anomalous system removal 
efficiencies compared to the majority of sources. The floor emission 
level for new lightweight aggregate kilns was 43 ppmv. (62 FR at 
24235.)
    In the final rule, we use a similar engineering evaluation and data 
analysis method as discussed in the May 1997 NODA to establish the 
floor emission level for hydrochloric acid/chlorine gas. We identified 
MACT floor control as wet scrubbing since the best controlled source is 
using this control technology. One lightweight aggregate facility uses 
venturi-type wet scrubbers for the control of hydrochloric acid/
chlorine gas. We evaluated the chlorine system removal efficiencies 
achieved by wet scrubbing at this facility. Our data show that this 
facility is consistently achieving greater than 99 percent control of 
hydrochloric acid/chlorine gas. Because we have no data with system 
removal efficiencies indicative of poor performance, we conclude that 
all data from this facility are reflective of MACT control (wet 
scrubbers), and, therefore, the floor emission limit for new sources is 
set equal to the highest test condition average of these data. Thus, 
the MACT floor emission limit for new lightweight aggregate kilns is 
identified as 41 ppmv.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 proposal and May 1997 NODA, we did not propose a beyond-
the-floor standard for new sources because the floor emission level was 
based on wet scrubbing, which is the best available control technology 
for hydrochloric acid/chlorine gas. (61 FR at 17409 and 62 FR at 
24235.) We continue to believe that a beyond-the-floor emission level 
for new sources is not warranted due to the high costs of treatment and 
the small additional amount of chlorine that would be removed. 
Therefore, the MACT standard for new lightweight aggregate kilns is 
identified as 41 ppmv.
8. What Are the Hydrocarbon and Carbon Monoxide Standards?
    In the final rule, we establish hydrocarbon and carbon monoxide 
standards as surrogates to control emissions of nondioxin organic 
hazardous air pollutants for existing and

[[Page 52901]]

new lightweight aggregate kilns. The standards limit hydrocarbon and 
carbon monoxide concentrations to 20 ppmv 171 or 100 ppmv, 
172 respectively. Existing and new lightweight aggregate 
kilns can elect to comply with either the hydrocarbon limit or the 
carbon monoxide limit on a continuous basis. Lightweight aggregate 
kilns that choose to comply with the carbon monoxide limit on a 
continuous basis must also demonstrate compliance with the hydrocarbon 
standard during the comprehensive performance test. However, continuous 
hydrocarbon monitoring following the performance test is not 
required.173 We discuss the rationale for establishing these 
standards below.
---------------------------------------------------------------------------

    \171\ Hourly rolling average, reported as propane, dry basis and 
corrected to 7 percent oxygen.
    \172\Hourly rolling average, dry basis, corrected to 7 percent 
oxygen.
    \173\As discussed in Part 5, Section X.F, lightweight aggregate 
kilns that feed hazardous waste at a location other than the end 
where products are normally discharged and where fuels are normally 
fired must comply with the 20 ppmv hydrocarbon standards (i.e., 
these sources do not have the option to comply with the carbon 
monoxide standard).
---------------------------------------------------------------------------

    a. What Is the MACT Floor for Existing Sources? As discussed in 
Part Four, Section II.A.2, we proposed limits on hydrocarbon and carbon 
monoxide emissions as surrogates to control nondioxin organic hazardous 
air pollutants. In the April 1996 NPRM, we identified floor control as 
combustion of hazardous waste under good combustion practices to 
minimize the generation of fuel-related hydrocarbons. We proposed a 
hydrocarbon emission level of 14 ppmv and a carbon monoxide level of 
100 ppmv. The hydrocarbon level was based on an analysis of the 
available emissions data, while the basis of the carbon monoxide level 
was existing federal regulations (see Sec. 266.104(b)). (61 FR at 
17407.) In the May 1997 NODA, we solicited comment a hydrocarbon 
emission level of 10 ppmv. The hydrocarbon floor level was changed to 
10 ppmv from 14 ppmv because of a change in the lightweight aggregate 
kiln universe of facilities. The lightweight aggregate kiln with the 
highest hydrocarbon emissions stopped burning hazardous waste. With the 
exclusion of the hydrocarbon data from this one source, the remaining 
lightweight aggregate kilns appeared to be able to meet a hydrocarbon 
standard on the order of 6 ppmv. However, since we were unable to 
identify an engineering reason why lightweight aggregate kilns using 
good combustion practices should be able to achieve lower hydrocarbon 
emissions than incinerators, we indicated that it may be more 
appropriate to establish the hydrocarbon standard at 10 ppmv, which was 
equal to the incinerator emission level discussed in that NODA. In the 
NODA, we also continued to indicate our preference for a carbon 
monoxide emission level of 100 ppmv. (62 FR at 24235.)
    One commenter states that some lightweight aggregate kilns may not 
be able to meet a 10 ppmv hydrocarbon standard due to organics in raw 
materials. Notwithstanding our data base of short-term data indicating 
the achievability of a hydrocarbon standard of 10 ppmv, the commenter 
states that this standard may be unachievable over the long-term 
because trace levels of organic matter in the raw materials vary 
significantly. Hydrocarbon emissions could increase as the source uses 
raw materials from different on-site quarry locations. Thus, the 
commenter supports a hydrocarbon emission level consistent with cement 
kilns (i.e., 20 ppmv), and opposes a floor emission level that is 
comparable to incinerators for which low temperature organics 
desorption from raw materials is not a complicating issue.
    Our limited hydrocarbon data, as discussed above, indicates that a 
hydrocarbon level of 10 ppmv is achievable for lightweight aggregate 
kilns.174 However, we agree that over long-term operations, 
lightweight aggregate kilns may encounter variations in the level of 
trace organics in raw materials, similar to cement kilns, that may 
preclude some kilns from achieving a hydrocarbon limit of 10 ppmv. 
Thus, we conclude that a hydrocarbon emission level of 20 ppmv, the 
same floor level for cement kilns, is also appropriate for lightweight 
aggregate kilns. A hydrocarbon standard of 20 ppmv also is based on 
existing federally-enforceable RCRA regulations, to which lightweight 
aggregate kilns are currently subject. (See Sec. 266.104(c).)
---------------------------------------------------------------------------

    \174\ Our data base for hydrocarbons consists of short-term 
emissions data.
---------------------------------------------------------------------------

    Some commenters also support a requirement for both a carbon 
monoxide and hydrocarbon limit for lightweight aggregate kilns. These 
commenters state that requiring both hydrocarbon and carbon monoxide 
limits would further reduce emissions of organic hazardous air 
pollutants. One commenter notes that 83 percent of existing lightweight 
aggregate kilns are currently achieving both a hydrocarbon level of 20 
ppmv and a carbon monoxide standard of 100 ppmv.
    We carefully considered the merits and drawbacks to requiring both 
a hydrocarbon and carbon monoxide standard. First, stack gas carbon 
monoxide levels may not be a universally reliable indicator of 
combustion intensity and efficiency for some lightweight aggregate 
kilns due, first, to carbon monoxide generation by disassociation of 
carbon dioxide to carbon monoxide at high temperatures and, second, to 
evolution of carbon monoxide from the trace organic constituents in raw 
material feedstock.175 One commenter supports our view by 
citing normal variability in carbon monoxide levels at their kiln with 
no apparent relationship to combustion conditions, such as temperature, 
residence time, excess oxygen levels. Thus, carbon monoxide can be 
overly conservative surrogate for some kilns.176
---------------------------------------------------------------------------

    \175\ Raw materials enter the upper end of the kiln and move 
counter-current to the combustion gas. Thus, as the raw materials 
are convectively heated in the upper end kiln above the flame zone, 
organic compounds can evolve from trace levels of organics in the 
raw materials. These organic compounds can be measured as 
hydrocarbons, and when only partially oxidized, carbon monoxide. 
This process is not related to combustion of hazardous waste or 
other fuels in the combustion zone at the other end of the kiln.
    \176\ Of course, if a source elects to comply with the carbon 
monoxide standard, then we are sure that it is achieving good 
combustion conditions and good control of organic hazardous air 
pollutants that could be potentially emitted from hazardous waste 
fed into the combustion zone.
---------------------------------------------------------------------------

    Second, requiring both continuous monitoring of carbon monoxide and 
hydrocarbon in the stack is at least somewhat redundant for control of 
organic emissions from combustion of hazardous waste because: (1) 
Hydrocarbons alone are a direct and reliable surrogate for measuring 
the destruction of organic hazardous air pollutants; and (2) carbon 
monoxide is generally a conservative indicator of good combustion 
conditions and thus good control of organic hazardous air pollutants. 
See Part Four, Section IV.B of the preamble for a discussion of our 
approach to using carbon monoxide or hydrocarbons to control organic 
emissions.
    We identify a carbon monoxide level of 100 ppmv and a hydrocarbon 
level of 20 ppmv as floor control for existing sources because they are 
existing federally enforceable standards for hazardous waste burning 
lightweight aggregate kilns. See Sec. 266.104(b) and (c). As current 
rules allow, sources would have the option of complying with either 
limit. Given that these are current rules, all lightweight aggregate 
kilns can currently achieve these emission levels. Thus, we estimate no 
emissions reductions or costs for these floor levels.
    Lightweight aggregate kilns that choose to continuously monitor and

[[Page 52902]]

comply with the carbon monoxide standard must demonstrate during the 
performance test that they are also in compliance with the hydrocarbon 
emission standard. In addition, kilns that monitor carbon monoxide 
alone must also set operating limits on key parameters that affect 
combustion conditions to ensure continued compliance with the 
hydrocarbon emission standard. We developed this modification because 
of some limited data that show a source can produce high hydrocarbon 
emissions while simultaneously producing low carbon monoxide emissions. 
We conclude from this information that it is necessary to confirm the 
carbon monoxide-hydrocarbon emissions relationship for every source 
that selects to monitor carbon monoxide emissions alone. See discussion 
in Part Four, Section IV.B.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? In the April 1996 proposal, we identified beyond-the-floor 
control levels for carbon monoxide and hydrocarbon in the main stack of 
50 ppmv and 6 ppmv, respectively. (61 FR at 17407.) These beyond-the-
floor levels were based on the use of a combustion gas afterburner. We 
indicated in the proposal, however, that this type of beyond-the-floor 
control would be cost prohibitive. Our preliminary estimates suggested 
that going beyond-the-floor for carbon monoxide and hydrocarbons would 
more than double the national costs of complying with the proposed 
standards. We continue to believe that a beyond-the-floor standard for 
carbon monoxide and hydrocarbons based on an afterburner is not 
justified and do not adopt a beyond-the-floor standard for existing 
lightweight aggregate kilns.
    In summary, we adopt the floor emission levels for hydrocarbons, 20 
ppmv, or carbon monoxide, 100 ppmv, as standards in the final rule.
    c. What Is the MACT Floor for New Sources? In the April 1996 NPRM, 
we identified MACT floor control as operating the kiln under good 
combustion practices. Because we were unable to quantify good 
combustion practices, floor control for the single best controlled 
source was the same as for existing sources. We proposed, therefore, a 
floor emission level of 14 ppmv for hydrocarbons and a 100 ppmv limit 
for carbon monoxide. (61 FR at 17409.) In the May 1997 NODA, we 
continued to identify MACT floor control as good combustion practices 
and we took comment on the same emission levels as existing sources: 20 
ppmv for hydrocarbons and 100 ppmv for carbon monoxide. (62 FR at 
24235.)
    In developing the final rule, we considered the comment that the 
rule should allow compliance with either a carbon monoxide standard of 
100 ppmv or a hydrocarbon standard of 20 ppmv. Given that this option 
is available under the existing regulations for new and existing 
sources, we conclude that this represents MACT floor for new sources. 
These emission levels are achieved by operating the kiln under good 
combustion practices to minimize fuel-related hydrocarbons and carbon 
monoxide emissions. As current rules allow, sources would have the 
option of complying with either limit. See Sec. 266.104(b) and (c).
    We also considered site selection based on availability of 
acceptable raw material hydrocarbon content as an approach to establish 
a hydrocarbon emission level at new lightweight aggregate kilns. This 
approach is similar to that done for new hazardous waste burning cement 
kilns at greenfield sites (see discussion above). For cement kilns, we 
finalize a new source floor hydrocarbon emission standard at a level 
consistent with the proposed standard for nonhazardous waste burning 
cement kilns. Because we are planning to issue MACT emission standards 
for nonhazardous waste lightweight aggregate kiln sources, we will 
revisit establishing a hydrocarbon standard at new lightweight 
aggregate kilns at that time so that a hydrocarbon standard, if 
determined appropriate, is consistent for these sources. We are 
deferring this decision to a later date to ensure that hazardous waste 
sources are regulated no less stringently than nonhazardous waste 
lightweight aggregate kilns.
    In summary, we are identifying a carbon monoxide level of 100 ppmv 
and a hydrocarbon level of 20 ppmv as floor control for new sources 
because they are existing federally enforceable standards for hazardous 
waste burning lightweight aggregate kilns. As discussed for existing 
sources above, lightweight aggregate kilns that choose to continuously 
monitor and comply with the carbon monoxide standard must demonstrate 
during the performance test that they are also in compliance with the 
hydrocarbon emission standard.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? In 
the April 1996 proposal, we identified beyond-the-floor emission levels 
for hydrocarbons and carbon monoxide of 6 ppmv and 50 ppmv, 
respectively for new sources. These beyond-the-floor levels were based 
on the use of a combustion gas afterburner. (61 FR at 17409.) We 
indicated in the proposal, however, that beyond-the-floor control was 
not justified due to the significant costs to retrofit kilns with 
afterburner controls. We estimated that going beyond-the-floor for 
hydrocarbons and carbon monoxide would more than double the national 
costs of complying with the proposed standards. We concluded that 
beyond-the-floor standards were not warranted. In the May 1996 NODA, we 
again indicated that a beyond-the-floor standard based on use of an 
afterburner would not be cost-effective and, therefore, justified. As 
discussed above for existing sources, we conclude that a beyond-the-
floor standard for carbon monoxide and hydrocarbons based on use of an 
afterburner would not be justified and do not adopt a beyond-the-floor 
standard for new lightweight aggregate kilns. (62 FR 24235.)
    In summary, we adopt the floor emission levels for hydrocarbons, 20 
ppmv, or carbon monoxide, 100 ppmv, as standards in the final rule.
9. What Are the Standards for Destruction and Removal Efficiency?
    We establish a destruction and removal efficiency (DRE) standard 
for existing and new lightweight aggregate kilns to control emissions 
of organic hazardous air pollutants other than dioxins and furans. 
Dioxins and furans are controlled by separate emission standards. See 
discussion in Part Four, Section IV.A. The DRE standard is necessary, 
as previously discussed, to complement the carbon monoxide and 
hydrocarbon emission standards, which also control these hazardous air 
pollutants.
    The standard requires 99.99 percent DRE for each principal organic 
hazardous constituent (POHC), except that 99.9999 percent DRE is 
required if specified dioxin-listed hazardous wastes are burned. These 
wastes--F020, F021, F022, F023, F026, and F027--are listed as RCRA 
hazardous wastes under part 261 because they contain high 
concentrations of dioxins.
    a. What Is the MACT Floor for Existing Sources? Existing sources 
are currently subject to DRE standards under Sec. 266.104(a) that 
require 99.99 percent DRE for each POHC, except that 99.9999 percent 
DRE is required if specified dioxin-listed hazardous wastes are burned. 
Accordingly, these standards represent MACT floor. Since all hazardous 
waste lightweight aggregate kilns must currently achieve these DRE 
standards, they represent floor control.
    b. What Are Our Beyond-the-Floor Considerations for Existing 
Sources? Beyond-the-floor control would be a requirement to achieve a 
higher

[[Page 52903]]

percentage DRE, for example, 99.9999 percent DRE for POHCs for all 
hazardous wastes. A higher DRE could be achieved by improving the 
design, operation, or maintenance of the combustion system to achieve 
greater combustion efficiency.
    Even though the 99.99 percent DRE floor is an existing RCRA 
standard, a substantial number of existing hazardous waste combustors 
are not likely to be routinely achieving 99.999 percent DRE, however, 
and most are not likely to be achieving 99.9999 percent DRE. 
Improvements in combustion efficiency will be required to meet these 
beyond-the-floor DREs. Improved combustion efficiency is accomplished 
through better mixing, higher temperatures, and longer residence times. 
As a practical matter, most combustors are mixing-limited and may not 
easily achieve 99.9999 percent DRE. For a less-than-optimum burner, a 
certain amount of improvement may typically be accomplished by minor, 
relatively inexpensive combustor modifications--burner tuning 
operations such as a change in burner angle or an adjustment of swirl--
to enhance mixing on the macro-scale. To achieve higher DREs, however, 
improved mixing on the micro-scale may be necessary. This involves 
significant, energy intensive and expensive modifications such as 
burner redesign and higher combustion air pressures. In addition, 
measurement of such DREs may require increased spiking of POHCs and 
more sensitive stack sampling and analysis methods at added expense.
    Although we have not quantified the cost-effectiveness of a beyond-
the-floor DRE standard, it would not appear to be cost-effective. For 
reasons discussed above, the cost of achieving each successive order-
of-magnitude improvement in DRE will be at least constant, and more 
likely increasing. Emissions reductions diminish substantially, 
however, with each order of magnitude improvement in DRE. For example, 
if a source were to emit 100 gm/hr of organic hazardous air pollutants 
assuming zero DRE, it would emit 10 gm/hr at 90 percent DRE, 1 gm/hr at 
99 percent DRE, 0.1 gm/hr at 99.9 percent DRE, 0.01 gm/hr at 99.99 
percent DRE, and 0.001 gm/hr at 99.999 percent DRE. If the cost to 
achieve each order of magnitude improvement in DRE is roughly constant, 
the cost-effectiveness of DRE decreases with each order of magnitude 
improvement in DRE. Consequently, we conclude that this relationship 
between compliance cost and diminished emissions reductions suggests 
that a beyond-the-floor standard is not warranted in light of the 
resulting, poor cost-effectiveness.
    c. What Is the MACT Floor for New Sources? The single best 
controlled source, and all other hazardous waste lightweight aggregate 
kilns, are subject to the existing RCRA DRE standard under 
Sec. 266.104(a). Accordingly, we adopt this standard of 99.99% DRE for 
most wastes and 99.9999% DRE for dioxin listed wastes as the MACT floor 
for new sources.
    d. What Are Our Beyond-the-Floor Considerations for New Sources? As 
discussed above, although we have not quantified the cost-effectiveness 
of a more stringent DRE standard, diminishing emissions reductions with 
each order of magnitude improvement in DRE suggests that cost-
effectiveness considerations would likely come into play. We conclude 
that a beyond-the-floor standard is not warranted.

Part Five: Implementation

I. How Do I Demonstrate Compliance with Today's Requirements?

    If you operate a hazardous waste burning incinerator, cement kiln, 
or lightweight aggregate kiln, you are required to comply with the 
standards and requirements in today's rule at all times, with one 
exception. If you are not feeding hazardous waste to the combustion 
device and if hazardous waste does not remain in the combustion 
chamber, these rules do not apply under certain conditions discussed 
below. You must comply with all of the notification requirements, 
emission standards, and compliance and monitoring provisions of today's 
rule by the compliance date, which is three years after September 30, 
1999. As referenced later, the effective date of today's rule is 
September 30, 1999. The compliance and general requirements of this 
rule are discussed in detail in the follow sections. Also, we have 
included the following time line that will assist you in determining 
when many of the notifications and procedures, discussed in the later 
sections of this part, are required to be submitted or accomplished.
A. What Sources Are Subject to Today's Rules?
    Sources affected by today's rule are defined as all incinerators, 
cement kilns and lightweight aggregate kilns burning hazardous waste 
on, or following September 30, 1999. This definition is essentially the 
same as we proposed in the April 1996 NPRM. Comments, regarding this 
definition, suggested that there was confusion as to when and under 
what conditions you would be subject to today's hazardous waste MACT 
regulations. In this rule, we specify that once you are subject to 
today's regulations, you remain subject to these regulations until you 
comply with the requirements for sources that permanently suspend 
hazardous waste burning operations, as discussed later.
    However, just because you are subject to today's regulations does 
not mean that you must comply with the emission standards or operating 
limits at all times. In later sections of today's rule, we identify 
those limited periods and situations in which compliance with today's 
emission standards and operating limits may not be required.
1. What Is an Existing Source?
    Today's rule clarifies that existing sources are sources that were 
constructed or under construction on the publication date for our 
NPRM---April 19, 1996. This is consistent with the current regulatory 
definition of existing sources, but is different from the definition in 
our April 1996 
NPRM. In the April 1996 NPRM, we defined existing sources as those 
burning hazardous waste on the proposal date (April 19, 1996) and 
defined new sources as sources that begin burning hazardous waste after 
the proposal date. Commenters note that the proposed definition of new 
sources is not consistent with current regulations found in 40 CFR part 
63 or the Clean Air Act. Commenters also believe that our definition 
does not consider the intent of Congress, i.e., to require only those 
sources that incur significant costs during upgrade or modification to 
meet the most stringent new source emission standards. Commenters note 
that a large number of sources that are currently not burning hazardous 
waste could modify their combustion units to burn hazardous waste at a 
cost that would not surpass the reconstruction threshold and therefore 
they should not be required to meet the new source emission standards. 
Commenters suggest we use the statutory definition of an existing 
source found at section 112(a)(4) of the CAA and codified at 40 CFR 
63.2. We agree with commenters and therefore adopt the definition of an 
existing source found at 40 CFR 63.2.
2. What Is a New Source?
    Today's rule clarifies that new sources are those that commence 
construction or meet the definition of a reconstructed source following 
the proposal date of April 19, 1996. In the proposal, we define new 
sources as those that newly begin to burn hazardous waste after the 
proposal date. However, as noted earlier, commenters object to the

[[Page 52904]]

proposed definition because of conflicts with the statutory language of 
the CAA and the current definition found in MACT regulations. In the 
CAA regulations, we define new sources as those that are newly 
constructed or reconstructed after a rule is proposed. Here again, we 
agree with commenters and adopt the current regulatory definition of 
new sources. We also adopt the CAA definition of reconstruction. This 
definition also is generally consistent with the RCRA definition of 
reconstruction and should avoid any confusion regarding what standards 
apply to reconstructed sources.
B. How Do I Cease Being Subject to Today's Rule?
    Once you become an affected source as defined in Sec. 63.2, you 
remain an affected source until you: (1) Cease hazardous waste burning 
operations, (i.e., hazardous waste is not in the combustion chamber); 
(2) notify the Administrator, and other appropriate regulatory 
authorities, that you have ceased hazardous waste burning operations; 
and (3) begin complying with other applicable MACT standards and 
regulations, if any, including notifications, monitoring and 
performance tests requirements.
    If you permanently stop burning hazardous waste, the RCRA 
regulations require you to initiate closure procedures within three 
months of the date you received your last shipment of hazardous waste, 
unless you have obtained an extension from the Administrator. The 
requirement to initiate closure pertains to your RCRA status and should 
not be a barrier to operational changes that affect your regulatory 
status under today's MACT requirements. This approach is a departure 
from the requirements proposed in the April 1996 NPRM, but is 
consistent with the approach we identified in the May 1997 NODA.
    Once you permanently stop burning hazardous waste, you may only 
begin burning hazardous waste under the procedures outlined for new or 
existing sources that become affected sources following September 30, 
1999. See later discussion.
C. What Requirements Apply If I Temporarily Cease Burning Hazardous 
Waste?
    Under today's rule, if you temporarily cease burning hazardous 
waste for any reason, you remain subject to today's requirements as an 
affected source. However, even as an affected source, you may not have 
to comply with the emission standards or operating limits of today's 
rule when hazardous waste is not in the combustion chamber. Today's 
standards, associated operating parameter limits, and monitoring 
requirements are applicable at all times unless hazardous waste is not 
in the combustion chamber and either: (1) You elect to comply with 
other MACT standards that would be applicable if you were not burning 
hazardous waste (e.g. the nonhazardous waste burning Portland Cement 
Kiln MACT, the nonhazardous waste burning lightweight aggregate kiln 
MACT (Clay Products Manufacturing), or the Industrial Incinerator 
MACT); or (2) you are in a startup, shutdown, or malfunction mode of 
operation. We note that until these alternative MACT standards are 
promulgated, you need to comply only with other existing applicable air 
requirements if any. This approach is consistent with the current RCRA 
regulatory approach for hazardous waste combustion sources, but differs 
from our April 1996 proposed approach.
    In our April 1996 NPRM, we proposed that sources always be subject 
to all of the proposed regulatory requirements, regardless of whether 
hazardous waste was in the combustion chamber. Commenters question the 
legitimacy of this requirement because the requirement was: (1) more 
stringent than current requirements; (2) not based on CAA statutory 
authority; and (3) contrary to current allowances under current MACT 
general provisions.
    In response, we agree with commenters on issues (1) and (3) above. 
However, we disagree with commenters on issue number (2). The CAA does 
not allow sources to be subject to multiple MACT standards 
simultaneously. Because current CAA regulations also allow sources to 
modify their operations such that they can become subject to different 
MACT rules so long as they provide notification to the Administrator, 
our proposed approach appears to further complicate a situation that it 
was intended to resolve. One of the main reasons we proposed to subject 
hazardous waste burning sources to the final standards at all times was 
to eliminate the ability of sources to arbitrarily switch between 
regulation as a hazardous waste burning source and regulation as a 
nonhazardous waste burning source. We were concerned about the 
compliance implications associated with numerous notifications to the 
permitting authority to govern operations that may only occur for a 
short period of time. However, our concern appears unfounded because 
the MACT general provisions currently allow sources to change their 
regulatory status following notification, and we cannot achieve this 
goal without restructuring the entire MACT program. Therefore, 
consistent with the current program, we adopt an approach that allows a 
source to comply with alternative compliance requirements, while 
remaining subject to today's rule. This regulatory approach eliminates 
the reporting requirements and compliance determinations we intended to 
avoid with our proposed approach, while preserving the essence of the 
current RCRA approach, which applies more stringent emissions standards 
when hazardous waste is in the combustor.
1. What Must I Do to Comply with Alternative Compliance Requirements?
    If you wish to comply with alternative compliance requirements, you 
must: (1) Comply with all of the applicable notification requirements 
of the alternative regulation; (2) comply with all the monitoring, 
record keeping and testing requirements of the alternative regulation; 
(3) modify your Notice Of Compliance (or Documentation of Compliance) 
to include the alternative mode(s) of operation; and (4) note in your 
operating record the beginning and end of each period when complying 
with the alternative regulation.
    If you intend to comply with an alternative regulation for longer 
than three months, then you also must comply with the RCRA requirements 
to initiate RCRA closure. You may be able to obtain an extension of the 
date you are required to begin RCRA closure by submitting a request to 
the Administrator.
2. What Requirements Apply If I Do Not Use Alternative Compliance 
Requirements?
    If you elect not to use the alternative requirements for compliance 
during periods when you are not feeding hazardous waste, you must 
comply with all of the operating limits, monitoring requirements, and 
emission standards of this rule at all times.177 However, if 
you are a kiln operator, you also may be able to obtain and comply with 
the raw material variance discussed later.
---------------------------------------------------------------------------

    \177\ The operating requirements do not apply during startup, 
shutdown, or malfunction provided that hazardous waste is not in the 
combustion chamber. See the discussion below in the text.
---------------------------------------------------------------------------

D. What Are the Requirements for Startup, Shutdown and Malfunction 
Plans?
    Sources affected by today's rule are subject to the provisions of 
40 CFR 63.6 with regard to startup, shutdown and malfunction plans. 
However, the plan applies only when hazardous waste is

[[Page 52905]]

not in the combustion chamber. If you exceed an operating requirement 
during startup, shutdown, or malfunction when hazardous waste is in the 
combustion chamber, your exceedance is not excused by following your 
plan. If you exceed an operating requirement during startup, shutdown, 
or malfunction when hazardous waste is not in the combustion chamber, 
you must follow your startup, shutdown, and malfunction plan to come 
back into compliance as quickly as possibly, unless you have elected to 
comply with the requirements of alternative section 112 or 129 
regulations that would apply if you did not burn hazardous waste. 
Failure to comply with the operating requirements to follow your 
startup, shutdown, and malfunction plan during the applicable periods 
is representative of a violation and may subject you to appropriate 
enforcement action.
    In the April 1996 NPRM (see 63 FR at 17449), we proposed that 
startup, shutdown, and malfunction plans would not be applicable to 
sources affected by the proposed rule because affected sources must be 
in compliance with the standards at all times hazardous waste is in the 
combustion chamber. We reasoned that hazardous waste could not be fired 
unless you were in compliance with the emission standards and operating 
requirements, and stated that the information contained in the plan and 
the purpose of the plan was not intended to apply to sources affected 
by this rule.
    In response, commenters state that startup, shutdown, and 
malfunction plans are appropriate for hazardous waste burning sources 
because malfunctioning operations are going to occur, and these plans 
are designed to reestablish compliant or steady state operations as 
quickly as possible. Furthermore, commenters maintain that because 
sources must prepare and follow facility-specific plans to address 
situations that could lead to increased emissions, rather than just 
note such an occurrence in the operating record, the public and we are 
better assured that the noncompliant operations are being remedied 
rather than awaiting for an after-the-fact enforcement action. 
Commenters also note that hazardous waste burning sources are no 
different than other MACT sources who are required to use such plans.
    After considering comments, we agree with commenters that startup, 
shutdown, and malfunction plans are valuable compliance tools and 
should be applicable to hazardous waste burning sources. However, we 
are concerned that some sources may attempt to use startup, shutdown, 
and malfunction plans to circumvent enforcement actions by claiming 
they were never out of compliance if they followed their plan. 
Therefore, we restrict the applicability of startup, shutdown, and 
malfunction plans to periods when hazardous waste is not in the 
combustion chamber. This restriction addresses the concern that 
operations under startup, shutdown, and malfunction could lead to 
increased emissions of hazardous air pollutants.
    We considered whether to specifically prohibit sources from feeding 
hazardous waste during periods of startup and shutdown. However, we 
decided not to adopt this requirement because of a potential regulatory 
problem. The requirement could have inadvertently subjected sources 
that experience unscheduled shutdowns to enforcement action if 
hazardous waste remained in the combustion chamber during the shutdown 
process even if operating requirements were not exceeded. Additionally, 
we decided that the prohibition was unnecessary because performance 
test protocols restrict the operations of all sources when determining 
operating parameter limits. The following factors are pertinent in this 
regard: (1) Sources are required to be in compliance with their 
operating parameter limits at all times hazardous waste is in the 
combustion chamber; (2) operating parameter limits are determined 
through a performance test which must be performed under steady-state 
conditions (see Sec. 63.1207(g)(1)(iii)); and (3) periods of startup 
and shutdown are not steady state conditions and therefore operating 
parameter limits determined through performance testing would not be 
indicative of those periods. Accordingly, burning hazardous waste 
during startup or shutdown would significantly increase the potential 
for a source to exceed an operating parameter limit, and we expect that 
sources would be unwilling to take that chance as a practical matter.
E. What Are the Requirements for Automatic Waste Feed Cutoffs?
    As proposed, you must operate an automatic waste feed cutoff system 
that immediately and automatically cuts off hazardous waste feed to the 
combustion device when:
    (1) Any of the following are exceeded: Operating parameter limits 
specified in Sec. 63.1209; an emission standard monitored by a 
continuous emissions monitoring system; and the allowable combustion 
chamber pressure; (2) The span value of any continuous monitoring 
system, except a continuous emissions monitoring system, is met or 
exceeded; (3) A continuous monitoring system monitoring an operating 
parameter limit under Sec. 63.1209 or emission level malfunctions; or 
(4) Any component of the automatic waste feed cutoff system fails.
    These requirements are provided at Sec. 63.1206(c)(3). The system 
must be fully functional on the compliance date and interlocked with 
the operating parameter limits you specify in the Document of 
Compliance (as discussed later) as well as the other parameters listed 
above.
    Also as proposed, after an automatic waste feed cutoff, you must 
continue to route combustion gases through the air pollution control 
system and maintain minimum combustion chamber temperature as long as 
hazardous waste remains in the combustion chamber. These requirements 
minimize emissions of regulated pollutants, including organic hazardous 
air pollutants, that could result from a perturbation caused by the 
waste feed cutoff. Additionally, you must continue to calculate all 
rolling averages and cannot restart feeding hazardous waste until all 
operating limits are within allowable levels.
    Additionally, as currently required for BIFs, we proposed that the 
automatic waste feed cutoff system and associated alarms must be tested 
at least once every seven days. This must be done when hazardous waste 
is burned to verify operability, unless you document in the operating 
record that weekly inspections will unduly restrict or upset operations 
and that less frequent inspections will be adequate. At a minimum, you 
must conduct operational testing at least once every 30 days.
    Commenters express the following concerns with the proposed 
automatic waste feed cutoff requirements: (1) Violations of the 
automatic waste feed cutoff linked operating parameters should not 
constitute a violation of the associated emission standard; (2) 
apparent redundancy exists between the proposed MACT requirements with 
the current RCRA requirements; (3) the proposed automatic waste feed 
cutoff requirements are inappropriate for all sources; and (4) 
uncertainty exists about how ``instantaneous'' is defined with regard 
to the nature of the automatic waste feed cutoff requirement.
    We address issue (1) later in this section. With respect to issue 
(2), our permitting approach (i.e., a single CAA title V permit to 
control all stack emissions) minimizes the potential redundancy of two 
permitting programs.
    In response to issue (3), we acknowledge that not all sources may 
be capable of setting operating limits or

[[Page 52906]]

continuously monitoring all of the prescribed operating parameters due 
to unique design characteristics inherent to individual units. However, 
you may take advantage of the provisions found in Sec. 63.8(f) which 
allow you to request the use of alternative monitoring techniques. See 
also Sec. 63.1209(g)(1).
    For issue (4), commenters express concern that requiring an 
immediate, instantaneous, and abrupt cutoff of the entire waste feed 
can cause perturbations in the combustion system that could result in 
exceedances of additional operating limits. We agree with commenters 
that a ramping down of the waste feedrate could preclude this problem 
in many cases and in the final rule allow a one-minute ramp down for 
pumpable wastes. To ensure that your ramp down procedures are bona fide 
and not simply a one-minute delay ending in an abrupt cutoff, you must 
document your ramp down procedures in the operating and maintenance 
plan. The procedures must specify that the ramp down begins immediately 
upon initiation of automatic waste feed cutoff and provides for a 
gradual ramp down of the hazardous waste feed. Note that if an emission 
standard or operating limit is exceeded during the ramp down, you 
nonetheless have failed to comply with the emission standards or 
operating requirements. The ramp down is not applicable, however, if 
the automatic waste feed cutoff is triggered by an exceedance of any of 
the following operating limits: minimum combustion chamber temperature; 
maximum hazardous waste feedrate; or any hazardous waste firing system 
operating limits that may be established for your combustor on a site-
specific basis. This is because these operating conditions are 
fundamental to proper combustion of hazardous waste and an exceedance 
could quickly result in an exceedance of an emission standard. We 
restrict the ramp down to pumpable wastes because: (1) Solids are often 
fed in batches where ramp down is not relevant (i.e., ramp down is only 
relevant to continuously fed wastes); and (2) incinerators burning 
solids also generally burn pumpable wastes and ramping down on 
pumpables only should preclude the combustion perturbations that could 
occur if all wastes were abruptly cutoff.
    Finally, with respect to issue number (1), if you exceed an 
operating parameter limit while hazardous waste is in the combustion 
chamber, then you have failed to ensure compliance with the associated 
emission standard. Accordingly, appropriate enforcement action on the 
exceedance can be initiated to address the exceedance. This enforcement 
process is consistent with current RCRA enforcement procedures 
regarding exceedances of operating parameter limits. However, as 
commenters note, we acknowledge that an exceedance of an operating 
parameter limit does not necessarily demonstrate that an associated 
emissions standard is exceeded. Nevertheless, in general, an exceedance 
of an operating parameter limit in a permit or otherwise required is an 
actionable event for enforcement purposes.
    Operating parameter limits are developed through performance tests 
that successfully demonstrate compliance with the standards. If a 
source exceeds an operating limit set during the performance test to 
show compliance with the standard, the source can no longer assure 
compliance with the associated standard. Furthermore, these operating 
parameter limits appear in enforceable documents, such as your NOC or 
your title V permit.
F. What Are the Requirements of the Excess Exceedance Report?
    In today's rule, we finalize the requirement to report to the 
Administrator when you incur 10 exceedances of operating parameter 
limits or emissions standards monitored with a continuous emissions 
monitoring system within a 60 day period. See Sec. 63.1206(c)(3)(vi). 
If a source has 10 exceedances within the 60 day period, the 60 day 
period restarts after the notification of the 10th exceedance. This 
provision is intended to identify sources that have excess exceedances 
due to system malfunction or performance irregularities. This 
notification requirement both highlights the source to regulatory 
officials and provides an added impetus to the facility to correct the 
problem(s) that may exist to limit future exceedances. For example, a 
source that must submit an excess exceedance report may be unable to 
operate under its current operating limits, which suggests that the 
source may need to perform a new comprehensive performance test to 
establish more appropriate operating limits.
    We discussed this provision in the April 1996 NPRM. Some commenters 
may have misunderstood our proposal while others felt that 10 
exceedances in sixty days was not a feasible number to set the 
reporting limit. Other commenters state that an industry wide MACT-like 
analysis is necessary to identify an achievable or appropriate number 
of exceedances upon which to set the reporting limit.
    We disagree with such comments. A MACT-like analysis is not called 
for in this case because this requirement is not an emission standard. 
This is a notification procedure that is a compliance tool to identify 
sources that cannot operate routinely in compliance with their 
operating parameter limits and emissions standards monitored with a 
continuous emissions monitoring system. Ideally, all sources should 
operate in compliance with all the standards and operating parameter 
limits at all times. Because, in the past, sources have been able to 
exceed their operating limits without having to notify the Agency, this 
does not mean that we condone, expect, or are unconcerned with such 
activity. In fact, the main reason we require this notification is 
because such activity exists to the current extent and because the 
Regions and States have identified it as a problem. We select 10 
exceedances in sixty days as the value that triggers reporting after 
discussions with Regional and State permit writers. Our discussions 
revealed that many hazardous waste combustion sources are required to 
notify regulatory officials following a single exceedance of an 
operating limit, while others don't have any reporting requirements 
linked to exceedances. Regions and States noted that because there is 
no current regulatory requirement for exceedance notifications, it is 
very difficult to require such notifications on a site-specific basis. 
Following these discussions, we contemplated requiring a notification 
following a single exceedance, but decided that the such a reporting 
limit might unnecessarily burden regulatory officials with reports from 
facilities that have infrequent exceedances. Therefore, our approach of 
10 exceedances in a 60 day period is a reasonably implementable limit 
and is not overly burdensome. Adopting this approach achieves an 
appropriate balance between burden on facilities and regulators and the 
need to identify underlying operational problems that may present 
unacceptable risks to the public and environment.
    To reiterate, this provision applies to any 10 exceedances of 
operating parameter limits or emission standards monitored with a 
continuous emissions monitoring system.
G. What Are the Requirements for Emergency Safety Vent Openings?
    In today's rule, we finalize requirements that govern the operation 
of emergency safety vents. See Sec. 63.1206(c)(4). These requirements: 
clarify the regulatory status of emergency safety vent events; require

[[Page 52907]]

development of an emergency safety vent operating plan that specifies 
procedures to minimize the frequency and duration of emergency safety 
vent openings; and specify procedures to follow when an emergency 
safety vent opening occurs.
    Key requirements regarding emergency safety vent openings include:
    (1) Treatment of combustion gases--As proposed, you must route 
combustion system off-gases through the same emission control system 
used during the comprehensive performance test. Any bypass of the 
pollution control system is considered an exceedance of operating 
limits defined in the Documentation of Compliance (DOC) or Notification 
of Compliance (NOC);
    (2) Emergency safety vent operating plan--As proposed, if you use 
an emergency safety vent in your system design, you must develop and 
submit with the DOC and NOC an emergency safety vent operating plan 
that outlines the procedures you will take to minimize the frequency 
and duration of emergency safety vent openings and details the 
procedure you will follow during and after an emergency safety vent 
opening; and
    (3) Emergency safety vent reporting requirements--As proposed, if 
you operate an emergency safety vent, you must submit a report to the 
appropriate regulatory officials within five days of an emergency 
safety vent opening. In that report, you must detail the cause of the 
emergency safety vent opening and provide information regarding 
corrective measures you will institute to minimize such events in the 
future.
    Commenters on the April 1996 NPRM (61 FR at 17440) state that 
emergency safety vent openings are safety devices designed to prevent 
catastrophic failures, safeguard the unit and operating personnel from 
pressure excursions and protect the air pollution control train from 
high temperatures and pressures. They suggest that restricting these 
operations is contrary to common sense. Furthermore, they state that 
emergency safety vent openings are most often due to local power 
outages and fluctuations in water flows going to the air pollution 
equipment. Commenters believe that emergency safety vent openings 
should not be considered violations and that not every emergency safety 
vent opening should be reportable for a variety of reasons including:
--Emergency safety vent openings have not been shown to be acutely 
hazardous. A study finds that they will not have any short-term impact 
on the health of workers on-site or residents of the nearby off-site 
community.
--Proper use of emergency safety vent systems minimizes the potential 
for impacts on operators and the neighboring public.
--Many emergency safety vents are downstream of the secondary 
combustion chamber and thus have low organic emissions.
--Some facilities have emergency safety vents connected to the air 
pollution control system and should be considered in compliance as long 
as the continuous emissions monitoring systems monitoring data does not 
indicate an exceedance.

    Commenters propose several alternatives:

--Recording emergency safety vent openings (including the time, 
duration and cause of each event) in the operating record, available to 
the Administrator, or any authorized representative, upon request.
--Making emergency safety vent openings a part of startup, shutdown, 
malfunction and abatement plans.
--Reporting openings that occurs more frequently than once in any 90 
day period, whereupon the Administrator may require corrective 
measures.
--Reporting only emergency safety vent openings in excess of 10 in a 60 
day period.
--Conditions relating to an emergency safety vent operation should be a 
part of the site-specific permit.
--Rely on the present RCRA permit process which provides the 
opportunity for permit writers and hazardous waste combustion device 
owner/operators to review emergency safety vent system designs.

    We agree that emergency safety vents are necessary safety devices 
for some incinerator designs that are intended to safeguard employees 
and protect the equipment from the dangers associated with system over-
pressures or explosions. However, simply because emergency safety vents 
are necessary safety devices for some incinerator designs in the event 
of a major malfunction does not mean that their routine use is 
acceptable. We cannot overlook an event when combustion gases are 
emitted into the environment prior to proper treatment by the pollution 
control system. Therefore, an emergency safety vent opening is evidence 
that compliance is not being achieved. Nonetheless, we expect sources 
to continue to use safety vents when the alternative could be a 
catastrophic failure and substantial liability even though opening the 
vent is evidence of failure to comply with the emission standards.
    Today's requirements are based on the fundamental need to ensure 
protection of human health and the environment against unquantified and 
uncontrolled hazardous air pollutant emissions. We do not agree that a 
change in the proposed emergency safety vent reporting requirement is 
warranted. These events are indicative of serious operational problems, 
and each event should be reported and investigated to reduce the 
potential of future similar events. As for including the emergency 
safety vent operating plan in the source-specific startup, shutdown, 
and malfunction plan, we see no reason to discourage that practice 
provided that a combined plan specifically addresses the events 
preceding and following an emergency safety vent opening.
H. What Are the Requirements for Combustion System Leaks?
    You must prevent leaks of gaseous, liquid or solid materials from 
the combustion system when hazardous waste is being fed to or remains 
in the combustion chamber. To demonstrate compliance with this 
requirement you must either: (1) Maintain the combustion system 
pressure lower than ambient pressure at all times; (2) totally enclose 
the system; or (3) gain approval from the Administrator to use an 
alternative approach that provides the same level of control achieved 
by options 1 and 2.
    Currently, these requirements exist for all sources under RCRA 
regulations. Many commenters question whether they were capable of 
meeting this requirement for various technical reasons. We acknowledge 
that certain situations may exist that prevent or limit a source from 
instantaneously monitoring pressure inside the combustion system, but 
in such situations, we can approve alternative techniques (under 
Sec. 63.1209(g)(1)) that allow sources to achieve the objectives of the 
requirements. Because this requirement is identical to the current RCRA 
requirements, and because we have specifically provided alternative 
techniques to demonstrate compliance, modifications to this provision 
are not warranted.
I. What Are the Requirements for an Operation and Maintenance Plan?
    You must prepare and at all times operate according to a operation 
and maintenance plan that describes in detail procedures for operation, 
inspection, maintenance, and corrective measures for all components of 
the combustor, including associated pollution control equipment, that 
could affect emissions of regulated hazardous

[[Page 52908]]

air pollutants. The plan must prescribe how you will operate and 
maintain the combustor in a manner consistent with good air pollution 
control practices for minimizing emissions at least to the levels 
achieved during the comprehensive performance test. You must record the 
plan in the operating record. See Sec. 63.1206(c)(7)(i).
    In addition, if you own or operate a hazardous waste incinerator or 
hazardous waste burning lightweight aggregate kiln equipped with a 
baghouse, your operation and maintenance plan for the baghouse must 
include a prescribed inspection schedule for baghouse components and 
use of a bag leak detection system to identify malfunctions. This 
baghouse operation and maintenance plan must be submitted to the 
Administrator with the initial comprehensive performance test for 
review and approval. See Sec. 63.1206(c)(7)(ii).
    We require an operation and maintenance plan to implement the 
provisions of Sec. 63.6(e). That paragraph requires you to operate and 
maintain your source in a manner consistent with good air pollution 
control practices for minimizing emissions. That paragraph, as all 
Subpart A requirements, applies to all MACT sources unless requirements 
in the subpart for a source category state otherwise. In addition, 
Sec. 63.6(e)(2) states that the Administrator will determine whether 
acceptable operation and maintenance procedures are used by reviewing 
information including operation and maintenance procedures and records. 
Thus, paragraph (e)(2) effectively requires you to develop operation 
and maintenance procedures. Consequently, explicitly requiring you to 
develop an operation and maintenance plan is a logical outgrowth of the 
proposed rule.
    Similarly, although we did not prescribe baghouse inspection 
requirements or require a bag leak detection system at proposal for 
incinerators and lightweight aggregate kilns, this is a logical 
outgrowth of the proposed rule. Section 63.6(e) requires sources to 
operate and maintain emission control equipment in a manner consistent 
with good air pollution control practices for minimizing emissions. 
Inspection of baghouse components is required to provide adequate 
maintenance, and a bag leak detection system is a state-of-the-art 
monitoring system that identifies major baghouse malfunctions. Absent 
use of a particulate matter CEMS or opacity monitor, use of a bag leak 
detection system is an essential monitoring approach to ensure that the 
baghouse continues to operate in a manner consistent with good air 
pollution control practices. Bag leak detection systems are required 
under the MACT standards for secondary lead smelters. See Sec. 63.548. 
We have also proposed to require them as MACT requirements for several 
other source categories including primary lead smelters (see 63 FR 
19200 (April 17, 1998)) and primary copper smelters (see 63 FR 19581 
(April 20, 1998)). In addition, we have published a guidance document 
on the installation and use of bag leak detection systems: USEPA, 
``Fabric Filter Bag Leak Detection,'' September 1997, EPA-454/R-98-015. 
Thus, although not explicitly required at proposal, a requirement to 
use bag leak detection systems is a logical outgrowth of the (proposed) 
requirements of Sec. 63.6(e).
    We are not prescribing a schedule for inspection of baghouse 
components or requiring a bag leak detection system for cement kilns 
because cement kilns must use a continuous opacity monitoring system 
(COMS) to demonstrate compliance with an opacity standard. A COMS is a 
better indicator of baghouse performance than a bag leak detection 
system. We could not use COMS for incinerators and lightweight 
aggregate kilns, however, because we do not have data to identify an 
opacity standard that is achievable by MACT sources (i.e., sources 
using MACT control and achieving the particulate matter standard).
    We are not specifying the type of sensor that must be used other 
than: (1) The system must be certified by the manufacturer to be 
capable of detecting particulate matter emissions at concentrations of 
1.0 milligram per actual cubic meter; and (2) the sensor must provide 
output of relative particulate matter loadings. Several types of 
instruments are available to monitor changes in particulate emission 
rates for the purpose of detecting fabric filter bag leaks or similar 
failures. The principles of operation of these instruments include 
electrical charge transfer and light scattering. The guidance document 
cited above applies to charge transfer monitors that use 
triboelectricity to detect changes in particle mass loading, but other 
types of monitors may be used. Specifically, opacity monitors may be 
used.
    The economic impacts of requiring fabric filter bag leak detection 
systems are minimal. These systems are relatively inexpensive. They 
cost less than $11,000 to purchase and install. Further, we understand 
that most hazardous waste burning lightweight aggregate kilns are 
already equipped with triboelectric sensors. Finally, there are few 
hazardous waste incinerators that are currently equipped with fabric 
filters.

II. What Are the Compliance Dates for this Rule?

A. How Are Compliance Dates Determined?
    In today's rule, as with other MACT rules, we specify the 
compliance date and then provide you additional time to demonstrate 
compliance through performance testing. Generally, you must be in 
compliance with the emission standards on September 30, 2002 unless you 
are granted a site-specific extension of the compliance date of up to 
one year. By September 30, 2002, you must complete modifications to 
your unit and establish preliminary operating limits, which must be 
included in the Documentation of Compliance (DOC) and recorded in the 
operating record. Following the compliance date you have up to 180 days 
to complete the initial comprehensive performance test and an 
additional 90 days to submit the results of the performance test in the 
Notification of Compliance (NOC). In the NOC, you also must certify 
compliance with applicable emission standards and define the operating 
limits that ensure continued compliance with the emission standards.
    In the April 1996 NPRM, we proposed that sources comply with all 
the substantive requirements of the rule on the compliance date. This 
required sources to conduct their performance test as well as submit 
results in the NOC by the compliance date. The compliance date 
discussed in the April 1996 NPRM contained a statutory limitation of 
three years following the effective date of the final rule (i.e., the 
publication date of the final rule) with the possibility of a site-
specific extension of up to one year for the installation of controls 
to comply with the final standards, or to allow for waste minimization 
reductions.
    In the May 1997 NODA, we acknowledged that the April 1996 NPRM 
definition of compliance date and our approach to implementation 
created a number of unforseen difficulties (see 63 FR at 24236). 
Commenters note that the proposed compliance date definition and the 
ramifications of noncompliance create the potential for an 
unnecessarily large number of source shut-downs due to an insufficient 
period to perform all the required tasks. Commenters recommend we 
follow the general provisions applicable to all MACT regulated sources, 
which allow sources to demonstrate compliance through

[[Page 52909]]

performance testing and submission of emission test results up to 270 
days following the compliance date.
    In the May 1997 NODA, we outlined an approach that allowed 
facilities to use the Part 63 general approach, which requires sources 
to complete performance testing within 180 days of the compliance date 
and submit test results 90 days after completing the performance 
test.178 Today, we adopt this approach to foster consistent 
implementation of this rule as a CAA regulation.
---------------------------------------------------------------------------

    \178\ The general provisions of part 63 allow for 180 days after 
the compliance date to conduct a performance test and 60 days to 
submit its results to the appropriate regulatory agency. However, as 
commenters note, dioxin/furan analyses can require 90 days to 
complete. Therefore, the time allowed for submission of test results 
should be extended to 90 days, increasing the total time following 
the compliance date to 270 days. We agree with commenters and 
increase the time allowed for submission of test results from 60 to 
90 days.
---------------------------------------------------------------------------

    Your individual dates for: (1) Compliance; (2) comprehensive 
performance testing; (3) submittal of test results; and (4) submittal 
of your NOC and title V permit requests depend on whether you were an 
existing source on April 19, 1996. Compliance dates for existing and 
new sources are discussed in the following two subsections.
B. What Is the Compliance Date for Sources Affected on April 19, 1996?
    The compliance date for all affected sources constructed, or 
commencing construction or reconstruction before April 19, 1996 is 
September 30, 2002.
C. What Is the Compliance Date for Sources That Become Affected After 
April 19, 1996?
    If you began construction or reconstruction after April 19, 1996, 
your compliance date is the latter of September 30, 1999 or the date 
you commence operations. If today's final emission standards are less 
stringent or as stringent as the standards proposed on April 19, 1996, 
you must be in compliance with the 1996 proposed standards upon 
startup. If today's final standards are more stringent than the 
proposed standards, you must be in compliance with the more stringent 
standards by September 30, 2002.

III. What Are the Requirements for the Notification of Intent to 
Comply?

    For the reader's convenience, we summarize here the Notice of 
Intent to Comply (NIC) requirements finalized in the ``fast-track'' 
rule of June 19, 1998. (See 63 FR at 33782.)
    The NIC requires you to prepare an implementation plan that 
identifies your intent to comply with the final rule and the basic 
means by which you intend to do so. That plan must be released to the 
public in a public forum and formally submitted to the Agency. The 
notice of intent certifies your intentions--either to comply or not to 
comply--and identifies milestone dates that measure your progress 
toward compliance with the final emission standards or your progress 
toward closure, if you choose not to comply. Prior to submitting the 
NIC to the regulatory Agency, you must provide notice of a public 
meeting and conduct an informal public meeting with your community to 
discuss the draft NIC and your plans for achieving compliance with the 
new standards.
    We have redesignated the existing NIC provisions to meld them into 
the appropriate sections of subpart EEE. We have also revised the 
regulatory language to include references to the new provisions 
promulgated today. See Part Six, Section IX of today's preamble.

IV. What Are the Requirements for Documentation of Compliance?

A. What Is the Purpose of the Documentation of Compliance?
    The purpose of the Documentation of Compliance 179 (DOC) 
is for you to certify by the compliance date that: (1) You have made a 
good faith effort to establish limits on the operating parameters 
specified in Sec. 63.1209 that you believe ensure compliance with the 
emissions standards; (2) required continuous monitoring systems are 
operational and meet specifications; and (3) you are in compliance with 
the other operating requirements. See Sec. 63.1211(d). This is 
necessary because all sources must be in compliance by the compliance 
date even though they are not required to demonstrate compliance, 
through performance testing, until 180 days after the compliance date. 
To fulfill the requirements of the DOC, you must place it in the 
operating record by the compliance date, September 30, 2002. (See 
compliance dates in Section II above.) Information that must be in the 
DOC includes all information necessary to determine your compliance 
status (e.g., operating parameter limits; functioning automatic waste 
feed cutoff system). All operating limits identified in the DOC are 
enforceable limits. However, if these limits are determined, after the 
initial comprehensive performance test, to have been inadequate to 
ensure compliance with the MACT standards, you will not be deemed to be 
out of compliance with the MACT emissions standards, if you complied 
with the DOC limits.180
---------------------------------------------------------------------------

    \179\ We renamed the proposed Precertification of Compliance as 
the Documentation of Compliance to avoid any confusion with the RCRA 
requirement of similar name.
    \180\ Once you determine that you failed to demonstrate 
compliance during the performance test, all monitoring data is 
subject to potential case-by-case use as credible evidence to show 
noncompliance following that determination. Therefore, you could 
potentially find yourself in noncompliance for the period which the 
DOC limits were in effect following that determination, but before 
submission of the NOC.
---------------------------------------------------------------------------

B. What Is the Rationale for the DOC?
    In the May 1997 NODA, we discussed the concept of the 
precertification of compliance (Pre-COC). The discussion required 
sources to precertify their compliance status on the compliance date by 
requiring them to submit a notification to the appropriate regulatory 
agency. This notification would detail the operating limits under which 
a source would operate during the period following the compliance date, 
but before submittal of the initial comprehensive performance test 
results in the Notification of Compliance.
    Commenters question this provision since the Pre-COC operating 
limits would be effective only for the 270 days following the 
compliance date. Other commenters support the Pre-COC requirements 
provided the process is focused, straightforward, and limited to the 
minimum operating parameters necessary to document compliance. 
Commenters also stress that the Agency needed to specify the 
requirements of the prenotification, using appropriate sections of 40 
CFR 266.103(b) and Section 63.9 when developing the specific regulatory 
requirements. In addition, commenters suggest that the Agency clarify 
the relationship between the Pre-COC and the title V permit, and 
indicate how or if the Pre-COC operating limits would be placed in the 
title V permit.
    Other commenters state that the rationale underlying the Pre-COC is 
faulty because sources would remain subject to the RCRA permit 
conditions until the NOC is submitted or until the title V permit is 
issued, which was our proposed approach to permitting at that time. 
Therefore, the Agency's concern that sources could be between 
regulatory regimes is not relevant. Commenters also state that Pre-COC 
requirements would be resource intensive and a needless exercise that 
diverted time and attention from preparing to come into compliance with 
MACT standards.
    The DOC requirements and process adopted today provide the Agency 
and public a sound measure of assurance

[[Page 52910]]

that, on the compliance date, combustion sources are operated within 
limits that should ensure compliance with the MACT standards and 
protection to human health and the environment. We agree that operating 
limits in the DOC will be in effect only for a short period of time and 
that affected sources will not be between regulatory regimes at any 
time. Given the relatively short period of time the DOC conditions will 
be in effect, however, we chose for the final rule not to specify 
whether the conditions need to be incorporated into a title V permit 
and do not require the permitting authority to do so. We provide 
flexibility for agencies implementing title V programs to determine the 
appropriate level of detail to include in the permit, thereby allowing 
them to minimize the potential need for permit revisions. In addition, 
we do not require that the DOC be submitted to the permitting 
authority, to avoid burdening the permitting agency with unnecessary 
paper work during the period that they are reviewing site-specific 
performance test plans. In today's rule, we better define the period 
during which the DOC applies by specifying that the DOC is superseded 
by the NOC upon the postmark date for submittal of the NOC. Once you 
mail the NOC, its contents become enforceable unless and until 
superseded by test results submitted within 270 days following 
subsequent performance testing. This approach provides clarity on when 
the NOC supersedes the DOC.
C. What Must Be in the DOC?
    You must complete your site-specific DOC and place it in your 
operating record by the compliance date. The DOC must contain all of 
the information necessary to determine your compliance status during 
periods of operation including all operating parameter limits. You must 
identify the DOC operating limits through the use of available data and 
information. If your unit requires modification or upgrades to achieve 
compliance with the emission standards, you can base this judgment on 
results of shakedown tests and/or manufacturers assertions or 
specifications. If your unit does not require modifications or upgrades 
to meet the emission standards of today's rule, you can develop the 
operating limits through analysis of previous performance tests or 
knowledge of the performance capabilities of your control equipment.
    Your limitations on operating parameters must be based on an 
engineering evaluation prepared under your direction or supervision in 
accordance with a system designed. This evaluation must ensure that 
qualified personnel properly gathered and evaluated the information and 
supporting documentation, and considering at a minimum the design, 
operation, and maintenance characteristics of the combustor and 
emissions control equipment, the types, quantities, and characteristics 
of feedstreams, and available emissions data.
    This requirement should not involve a significant effort because 
your decisions on whether to upgrade and modify your units will be 
based on the current performance of your control equipment and the 
performance capabilities of new equipment you purchase. We expect that, 
by the compliance date, you will have an adequate understanding of your 
unit's capabilities, given the three years to develop this expertise. 
Therefore, by the compliance date, you are expected to identify 
operating limits that are based on technical or engineering judgment 
that should ensure compliance with the emission standards.

V. What Are the Requirements for MACT Performance Testing?

A. What Are the Compliance Testing Requirements?
    Today's final rule requires two types of performance testing to 
demonstrate compliance with the MACT emission standards: Comprehensive 
and confirmatory performance testing. See Sec. 63.1207. The purpose of 
comprehensive performance testing is to demonstrate compliance and 
establish operating parameter limits. You must conduct your initial 
comprehensive performance tests by 180 days (i.e., approximately six 
months) after your compliance date. You must submit results within 90 
days (i.e., approximately 3 months) of completing your comprehensive 
performance test. If you fail a comprehensive performance test, you 
must stop burning hazardous waste until you can demonstrate compliance 
with today's MACT standards. Comprehensive performance testing must be 
repeated at least every five years, but may be required more frequently 
if you change operations or fail a confirmatory performance test.
    The purpose of confirmatory performance tests is to confirm 
compliance with the dioxin/furan emission standard during normal 
operations. You must conduct confirmatory performance tests midway 
between comprehensive performance tests. Confirmatory performance tests 
may be conducted under normal operating conditions. If you fail a 
confirmatory performance test, you must stop burning hazardous waste 
until you demonstrate compliance with the dioxin/furan standard by 
conducting a comprehensive performance test to establish revised 
operating parameter limits.
    The specific requirements and procedures for these two performance 
tests are discussed later in this section. In addition, this section 
discusses the interaction between the RCRA permitting process and the 
MACT performance test.
1. What Are the Testing and Notification of Compliance Schedules?
    Section 63.7 of the CAA regulations contains the general 
requirements for testing and notification of compliance. In today's 
rule, we adopt some Sec. 63.7 requirements without change and adopt 
others with modifications. As summarized earlier, you must commence 
your initial comprehensive performance test within 180 days after your 
compliance date, consistent with the general Sec. 63.7 requirements. 
You must complete testing within 60 days of commencement, unless a time 
extension is granted. This requirement is necessary because testing and 
notification of compliance deadlines are based on the date of 
commencement or completion of testing. Those deadlines could be 
meaningless if a source had unlimited time to complete testing. 
Although we propose to require testing to be completed within 30 days 
of commencement, commenters state that unforeseen events could occur 
(e.g., system breakdown causing extensive repairs; loss of samples from 
breakage of equipment or other causes requiring additional test runs) 
that could extend the testing period beyond normal time frames. We 
concur, and provide for a 60-day test period as well as a case-by-case 
time extension that may be granted by permit officials if warranted 
because of problems beyond our control.
    Additionally, you must submit comprehensive performance test 
results to the Administrator within 90 days of test completion, unless 
a time extension is granted. We are allowing an additional 30 days for 
result submittal beyond the Secs. 63.7(g) and 63.8(e)(5) 60-day 
deadlines because the dioxin/furan analyses required in today's rule 
may take this additional time to complete. We also are including a 
provision for a case-by-case time extension in the final rule because 
commenters express concern that the limited laboratory facilities 
nationwide may be taxed by the need to handle analyses simultaneously 
for many hazardous

[[Page 52911]]

waste combustors. The available analytical services may not be able to 
handle the workload, that could cause some sources to miss the proposed 
90-day deadline. We concur with commenters' concerns and have added a 
provision to allow permit officials to grant a case-by-case time 
extension, if warranted.
    Test results must be submitted as part of the notification of 
compliance (NOC) submitted to the Administrator under Secs. 63.1207(j) 
and 63.1210(d) documenting compliance with the emission standards and 
continuous monitoring system requirements, and identifying applicable 
operating parameter limits. These provisions are similar to 
Secs. 63.7(g) and 63.8(e)(5), except that the NOC must be postmarked by 
the 90th day following the completion of performance testing and the 
continuous monitoring system performance evaluation.
    Overall, the initial NOC must be postmarked within 270 days (i.e., 
approximately nine months) after your compliance date. You must 
initiate subsequent comprehensive performance tests within 60 months 
(i.e., five years) of initiating your initial comprehensive performance 
test. You must submit subsequent NOCs, containing test results, within 
90 days after the completion of subsequent tests.
    The rule allows you to initiate subsequent tests any time up to 30 
days after the deadline for the subsequent performance test. Thus, you 
can modify the combustor or add new emission control equipment at any 
time and conduct new performance testing to document compliance with 
the emission standards. In addition, this testing window allows you to 
plan to commence testing well in advance of the deadline to address 
unforseen events that could delay testing.181 This testing 
window applies to both comprehensive performance tests and confirmatory 
performance tests. For example, if the deadline for your second 
comprehensive performance test is January 10, 2008, you may commence 
the test at any time after completing the initial comprehensive 
performance test but not later than February 10, 2008. The deadline for 
subsequent comprehensive and confirmatory performance tests are based 
on the commencement date of the previous comprehensive performance 
test.
---------------------------------------------------------------------------

    \181\ We note that a case-by-case time extension for 
commencement of subsequent performance testing is also provided 
under Sec. 63.1207(i).
---------------------------------------------------------------------------

2. What Are the Procedures for Review and Approval of Test Plans and 
Requirements for Notification of Testing?
    In the April 1996 NPRM, we proposed in Sec. 63.7(b)(1) to require 
submittal of a ``notification of performance test'' to the 
Administrator 60 days prior to the planned test date. This notification 
included the site-specific test plan itself for review and approval by 
the Administrator (Sec. 63.8(e)(3)). In the May 1997 NODA, to ensure 
coordination of destruction removal efficiency (DRE) and MACT 
performance testing, we considered requiring you to submit the test 
plan one year rather than 60 days prior to the scheduled test date to 
allow the regulatory official additional time to consider DRE testing 
in context with MACT comprehensive performance testing. This one-year 
test review period would only have applied to sources required to 
perform a DRE test.
    In today's final rule, we maintain the requirement for you to 
submit the test plan one year prior to the scheduled test date, but 
apply that requirement to all sources, not just those performing a DRE 
test. After consideration of comments (described below), we determined 
that this one-year period is needed to provide regulatory officials 
sufficient time (i.e., nine months) to review and approve or notify you 
of intent to disapprove the plan. Nine months is needed for the review 
for all sources given the amount of technical information that would be 
included in the test plan, and would also allow time to assess whether 
a source is required to perform a DRE test (see Part IV, Section IV, 
for discussion of DRE testing requirements; see also 
Sec. 63.1206(b)(8)). During this nine-month period, the regulatory 
officials will review your test plan and determine if it is adequate to 
demonstrate compliance with the emission standards and establish 
operating requirements.
    After submittal of the test plan, review and approval or 
notification of intent to deny approval of the test plan will follow 
the requirements of Sec. 63.7(c)(3). That section provides procedures 
for you to provide additional information before final action on the 
plan. It also requires you to comply with the testing schedule even if 
permit officials have not approved your test plan. The only exception 
to this requirement is if you proposed to use alternative test methods 
to those specified in the rule. In that case, you may not conduct the 
performance test until the test plan is approved, and you have 60 days 
after approval to conduct the test.
    Several commenters suggest that it would be difficult for permit 
officials to review and approve test plans within the nine-month window 
given that many test plans may be submitted at about the same time. 
They cite experiences under RCRA trial burn plan approvals where permit 
officials have taken much longer than nine months to approve a plan, 
and have requested that the final rule allow for a longer review 
period. Commenters are concerned with the consequences of being 
required to conduct the performance test even though permit officials 
may not have had time to approve the test plan. They recite various 
concerns that permit officials may at a later date determine that the 
performance test was inadequate and require retesting. Commenters 
suggest that the rule establish the date for the initial comprehensive 
performance test as 60 days following approval of the test plan, 
whenever that may occur, thus extending the deadline for the 
performance test indefinitely from the current requirement of six 
months after the compliance date.
    We maintain that the nine-month review period is appropriate for 
several reasons. First, we are unwilling to build into the regulations 
an indefinite period for review. This would have the potential to delay 
implementation of the MACT emission standards without any clear and 
compelling reason to do so.
    Second, the RCRA experience with protracted approval schedules, 
sometimes over a decade ago, is not applicable or analogous to the MACT 
situation. Under the RCRA regulatory regime, particularly at the early 
stages, there were few incentives for either permit officials or owners 
or operators to expeditiously negotiate acceptable test plans. No 
statutory deadlines existed for a compliance date, and existing 
facilities operated under interim status (a type of grand fathering 
tantamount to a permit). This interim status scheme placed at least 
some controls on hazardous waste combustors during the permit 
application and trial burn test plan review periods. As a result, 
regulatory officials could take significant amounts of time to address 
what was then a new type of approval, that for trial burn testing to 
meet RCRA final permit standards.
    Under MACT, the situation today is quite different. In light of the 
statutory compliance date of 3 years and the existing regulatory 
framework, sources know as of today's final rule that they need to 
respond promptly and effectively to permit officials' concerns about 
the test plan because the performance test must be conducted

[[Page 52912]]

within six months after the compliance date whether or not the test 
plan is approved. And they have at least two years to prepare and 
submit these plans, and to work with regulatory officials even before 
doing so. For their part, permit officials recognize that they have the 
responsibility to review and approve the plan or notify the source of 
their intent to deny approval within the nine-month window given that 
the source must proceed with expensive testing on a fixed deadline 
whether or not the plan is approved. To the extent regulatory officials 
anticipate that many test plans will be submitted at about the same 
time, the agencies have at least two years to figure out ways to 
accommodate this scenario from a resource and a prioritization 
standpoint. If permit officials nevertheless fail to act within the 
nine-month review and approval period, a source could argue that this 
failure is tacit approval of the plan and that later ``second-
guessing'' is not allowable. This should be a very strong incentive for 
regulatory officials to act within the nine months, especially with a 
two-year lead time to avoid this type of situation
    In addition, the RCRA experience is not a particularly good 
harbinger of the future MACT test plan approval, as commenters suggest, 
because most sources will have already completed trial burn testing 
under RCRA. Thus, both the regulatory agencies and the facilities have 
been through one round of test plan submittal, review, and approval for 
their combustion units. Given that MACT testing is very similar to RCRA 
testing, approved RCRA test protocols can likely be modified as 
necessary to accommodate any changes required under the MACT rule. 
Although some of these changes may be significant, we expect that many 
will not be. For example, RCRA trial burn testing always included DRE 
testing. Under the MACT rule, DRE testing will not be required for most 
sources. And for sources where DRE testing is required under MACT, most 
will have already been through a RCRA approval of the DRE test 
protocol, which should substantially simplify the process under MACT.
    The third reason that we maintain the nine-month review and 
approval window is appropriate is that discussions with several states 
leads us to conclude that they are prepared to meet their obligations 
under this provision. This is a highly significant indicator that the 
nine-month review and approval period is a reasonable period of time, 
particularly since all permitting agencies have at least two years to 
plan for submittal of test plans from the existing facilities in their 
jurisdictions.
    In summary, sound reasons exist to expect that today's final rule 
provides sufficient time for the submittal, review, and approval of 
test plans. Furthermore, clear incentives exist for both owners and 
operators and permit officials to work together expeditiously to ensure 
that an approval or notice of intent to disapprove the test plan can be 
provided within the nine-months allotted.
    On a separate issue, we also retain, in today's final rule, the 60-
day time frame and requirements of Sec. 63.7(b)(1) for submittal of the 
notification of performance test. Additionally, the final rule 
continues to provide an opportunity for, but does not require, the 
regulatory agency to review and oversee testing.
3. What Is the Provision for Time Extensions for Subsequent Performance 
Tests?
    The Administrator may grant up to a one year time extension for any 
performance test subsequent to the initial comprehensive performance 
test. This enables you to consolidate MACT performance testing and any 
other emission testing required for issuance or reissuance of Federal/
State permits.182
---------------------------------------------------------------------------

    \182\ In addition, this provision also may assist you when 
unforseen events beyond your control (e.g., power outage, natural 
disaster) prevent you from meeting the testing deadline.
---------------------------------------------------------------------------

    At the time of proposal, we were concerned about how to allow 
coordination of MACT performance tests and RCRA trial burns. As 
discussed elsewhere, the RCRA trial burn is superseded by MACT 
performance testing. However, a one-year time extension may still be 
necessary for you to coordinate performance of a RCRA risk burn. In 
addition, commenters state that there may be additional reasons to 
grant extension requests (e.g. some TSCA-regulated hazardous waste 
combustors may be required to perform stack tests beyond those required 
by MACT). Furthermore, some sources may have to comply with state 
programs requiring RCRA trial burn testing. To address these 
situations, to promote coordinated testing, and to avoid unnecessary 
source costs, the final rule allows up to a one-year time extension for 
the performance test.
    When performance tests and other emission tests are consolidated, 
the deadline dates for subsequent comprehensive performance tests are 
adjusted correspondingly. For example, if the deadline for your 
confirmatory performance test is January 1 and your state-required 
trial burn is scheduled for September 1 of the same year, you can apply 
to adjust the deadline for the confirmatory performance test to 
September 1. If granted, this also would delay by a corresponding time 
period the deadline dates for subsequent comprehensive performance 
tests.
    The procedures for granting or denying a time extension for 
subsequent performance tests are the same as those found in 
Sec. 63.6(i), which allow the Administrator to grant sources up to one 
additional year to comply with standards.183 These are also 
the same procedures apply to a request for a time extension for the 
initial NOC.
---------------------------------------------------------------------------

    \183\ Note, however, that Sec. 63.6(i) applies to an entirely 
different situation: extension of time for initial compliance with 
the standards, not subsequent performance testing.
---------------------------------------------------------------------------

4. What Are the Provisions for Waiving Operating Parameter Limits 
During Subsequent Performance Tests?
    Operating parameter limits are automatically waived during 
subsequent comprehensive performance tests under an approved 
performance test plan. See Sec. 63.1207(h). This waiver applies only 
for the duration of the comprehensive performance test and during 
pretesting for an aggregate period up to 720 hours of operation. You 
are still required to be in compliance with MACT emissions standards at 
all times during these tests, however.
    In the April 1996 NPRM, we proposed to allow the burning of 
hazardous waste only under the operating limits established during the 
previous comprehensive performance test (to ensure compliance with 
emission standards not monitored with a continuous emissions monitoring 
system). Two types of waivers from this requirement would have been 
provided during subsequent comprehensive performance tests: (1) An 
automatic waiver to exceed current operating limits up to 5 percent; 
and (2) a waiver that the Administrator may grant if warranted to allow 
the source to exceed the current operating limits without restriction. 
We proposed an automatic waiver because, without the waiver, the 
operating limits would become more and more stringent with subsequent 
comprehensive performance tests. This is because sources would be 
required to operate within the more stringent conditions to ensure that 
they did not exceed a current operating limit. This would result in a 
shrinking operating envelope over time.
    A number of commenters question the comprehensive performance 
test's 5%

[[Page 52913]]

limit over existing permit conditions. Some commenters state that the 
EPA should not limit a facility's operating envelope from test to test 
based on operating conditions established during the previous test. The 
operator should be free to set any conditions for the comprehensive 
performance test, short of what the regulator deems to pose a short-
term environmental or health threat or inadequate to ensure compliance 
with an emission standard. Commenters also state that the requirement 
that the facility accept the more stringent of the existing 5% limit or 
the test result will inevitably result in the ratcheting down of limits 
over time. Since certain conditions have much greater variation than 5% 
over a limit, sufficient variability must be allowed so the operator 
can run a test under the conditions it wishes to use as the basis for 
worst case operation.
    We agree that a waiver is necessary to avoid ratcheting down the 
operating limits in subsequent tests. Further, in view of the natural 
variability in hazardous waste combustor operations, a 5% waiver may be 
insufficient. Because you are required to comply with the emission 
standards, there does not appear to be any reason to establish national 
restrictions on operations during subsequent performance tests. 
Therefore, the final rule allows a waiver from previously established 
operating parameter limits, as long as you comply with MACT emission 
standards and are operating under an approved comprehensive performance 
test plan. Operating parameter limits will be reset based on the new 
tests. Furthermore, the permitting authority will review and has the 
opportunity to disapprove any proposed test conditions which may result 
in an exceedance of an emission standard.
B. What Is the Purpose of Comprehensive Performance Testing?
    The purposes of the comprehensive performance test are to: (1) 
Demonstrate compliance with the continuous emissions monitoring 
systems-monitored emission standards for carbon monoxide and 
hydrocarbons; (2) conduct manual stack sampling to demonstrate 
compliance with the emission standards for pollutants that are not 
monitored with a continuous emissions monitoring system (e.g., dioxin/
furan, particulate matter, DRE, mercury, semivolatile metal, low 
volatile metal, hydrochloric acid/chlorine gas); (3) establish limits 
on the operating parameters required by Sec. 63.1209 (Monitoring 
Requirements) to ensure compliance is maintained with those emission 
standards for which a continuous emissions monitoring system is not 
used for compliance monitoring; and (4) demonstrate that performance of 
each continuous monitoring system is consistent with applicable 
requirements and the quality assurance plan. In general, the 
comprehensive performance test is similar in purpose to the RCRA trial 
burn and BIF interim status compliance test, but with relatively less 
Agency oversight and a higher degree of self-implementation, as 
discussed below.
    The basic framework for comprehensive performance testing is set 
forth in the existing general requirements of subpart A, part 63. 
Therefore, for convenience of the reader, we will review key elements 
of those regulations and highlight any modifications made specifically 
for hazardous waste combustors.
1. What Is the Rationale for the Five Year Testing Frequency?
    As discussed earlier, you must perform comprehensive performance 
testing every five years. We require periodic comprehensive performance 
testing because we are concerned that long-term stress to the critical 
components of a source (e.g., firing systems, emission control 
equipment) could adversely affect emissions.
    In the April 1996 NPRM, we proposed that large sources (i.e., those 
with a stack gas flow rate greater than 23,127 acfm) and sources that 
accept off-site wastes would be required to perform comprehensive 
performance testing every three years. We also proposed that small, on-
site sources perform comprehensive performance testing every five years 
unless the Administrator determined otherwise on a case-specific basis. 
Commenters suggest that the proposed three year testing frequency is 
too restrictive. They said that test plan approval time, bad weather, 
mechanical failure, and the testing itself combine to make the proposed 
test frequency too tight for tests of this magnitude.
    We agree that, due to the magnitude of the comprehensive 
performance test, a more appropriate testing schedule is required. 
Therefore, we adopt a comprehensive performance testing frequency of 
every five years for small and large sources. In addition, this 
comprehensive performance testing schedule should correspond to the 
renewal of the title V permit. More frequent comprehensive performance 
testing is required, however, if there is a change in design, 
operation, or maintenance that may adversely affect compliance. See 
Sec. 63.1206(b)(6).
2. What Operations Are Allowed During a Comprehensive Performance Test?
    Because day-to-day limits are established for operating parameters 
during the comprehensive performance test, we allow operation during 
the performance test as necessary provided the unit complies with the 
emission standards. Accordingly, you can spike feedstreams with metals 
or chlorine, for example, to ensure that the feedrate limits are 
sufficient to accommodate normal operations while allowing some 
flexibility to feed higher rates. See Part Four, Section I. B. above 
for further discussion of normal operations. We note that this differs 
from Sec. 63.7(e) which requires performance testing under ``normal'' 
operating conditions. See Sec. 63.1207(g).
    Most commenters agree that the comprehensive performance test 
should be conducted under extreme conditions at the edge of the 
operating envelope. Commenters point out that they needed to operate in 
this mode to establish operating parameter limits to cover all possible 
normal operating emissions values. Commenters also state that 
feedstreams may need to be spiked with metals or chlorine to ensure 
limits high enough to allow operational flexibility. We agree that 
these modes of operation are needed to establish operating parameter 
limits that cover all possible normal operating emissions 
values.184 There is precedent for this approach in current 
rules regulating hazardous waste combustors (e.g., the RCRA incinerator 
and BIF rules).
---------------------------------------------------------------------------

    \184\ Allowing sources to operate during MACT comprehensive 
performance testing under the worst-case conditions, as allowed 
during RCRA compliance testing, rather than under normal conditions 
as provided by Sec. 63.7(e) for other MACT sources, ensures that the 
emissions standards do not restrict hazardous waste combustors using 
MACT control to operations resulting in emissions that are lower 
than normal. Therefore, allowing performance testing on a worst-case 
basis provides that the MACT emission standards are achievable in 
practice by sources using MACT control.
---------------------------------------------------------------------------

    In addition, two or more modes of operation may be identified, for 
which separate performance tests must be conducted and separate limits 
on operating conditions must be established. If you identify two modes 
of operation for your source, you must note in the operating record 
which mode you are operating under at all times. For example, two modes 
of operation must be identified for a cement kiln that routes kiln off-
gas through the raw meal mill to help dry the raw meal. When the raw 
meal mill is not operating (perhaps 15% of the time), the kiln gas 
bypasses the raw meal mill. Emissions of particulate matter and other 
hazardous air

[[Page 52914]]

pollutants or surrogates may vary substantially depending on whether 
the kiln gas bypasses the raw meal mill.
    As discussed below for confirmatory testing, when conducting the 
comprehensive performance test, you also must operate under 
representative conditions for specified parameters that may affect 
dioxin/furan emissions. These conditions must ensure that emissions are 
representative of normal operating conditions. Also, when demonstrating 
compliance with the particulate matter, semivolatile metal, and low 
volatile metal emission standards, when using manual stack sampling, 
and when demonstrating compliance with the dioxin/furan and mercury 
emission standards using carbon injection or carbon bed, you must 
operate under representative conditions for the cleaning cycle of the 
particulate matter control device. This is because particulate matter 
emissions increase momentarily during cleaning cycles and can affect 
emissions of these pollutants.
3. What Is the Consequence of Failing a Comprehensive Performance Test?
    If you determine that you failed any emission standard during the 
performance test based on: (1) Continuous emissions monitoring systems 
recordings; (2) results of analysis of samples taken during manual 
stack sampling; or (3) results of the continuous emissions monitoring 
systems performance evaluation, you must immediately stop burning 
hazardous waste. However, if you conduct the comprehensive performance 
test under two or more modes of operation, and you meet the emission 
standards when operating under one or more modes of operation, you are 
allowed to continue burning under the mode of operation for which the 
standards were met.
    If you fail one or more emission standards during all modes of 
operation tested, you may burn hazardous waste only for a total of 720 
hours and only for the purposes of pretesting (i.e., informal testing 
to determine if the combustor can meet the standards operating under 
modified conditions) or comprehensive performance testing under 
modified conditions. The same standards apply for the retest as applied 
for the original test. These conditions apply when you fail the initial 
or subsequent comprehensive performance test.
    A number of commenters suggest that the 720 operating hours allowed 
after a failed performance test should be renewable, as they are under 
existing incinerator and BIF rules. We are persuaded by the commenters' 
rationale and will adopt this practice in today's rule. The final rule 
allows the 720 hours of operation following a failed performance test 
to be renewed as often as the Administrator deems reasonable. We note 
that hazardous waste combustors are currently subject to virtually 
these same requirements under RCRA rules.
    If you fail a comprehensive performance test, you must still submit 
a NOC as required indicating the failure. We want to ensure that the 
regulatory authorities are fully aware of a failure and the need for 
the facility to initiate retesting.
    We do not specifically address other consequences of failing the 
comprehensive performance test in the regulatory language. We will 
instead rely on the regulating agency's enforcement policy to govern 
the type of enforcement response at a facility that exceeds an emission 
standard, fails to ensure compliance with the standards, or fails to 
meet a compliance deadline.
C. What Is the Rationale for Confirmatory Performance Testing?
    Confirmatory performance testing for dioxin/furan is required 
midway between the cycle required for comprehensive performance testing 
to ensure continued compliance with the emission standard. We require 
such testing only for dioxin/furan given: (1) The health risks 
potentially posed by dioxin/furan emissions; (2) the lack of a 
continuous emissions monitoring system for dioxin/furan; (3) the lack 
of a material that directly and unambiguously relates to dioxin/furan 
emissions which could be monitored continuously by means of feedrate 
control (as opposed to, for example, metals feedrates, which directly 
relate to metals emissions); and (4) wear and tear on the equipment, 
including any emission control equipment, which over time could result 
in an increase in dioxin/furan emissions even though the source stays 
in compliance with applicable operating limits.
    Although emissions of dioxins/furans appear to be primarily a 
function of whether particulate matter is retained in post-combustion 
regions of the combustor (e.g., in an electrostatic precipitator or 
fabric filter, or on boiler tubes) in the temperature range that 
enhances dioxin/furan formation, the factors that affect dioxin/furan 
formation are imperfectly understood. Certain materials seem to inhibit 
formation while others seem to enhance formation. Some materials seem 
to be precursors (e.g., PCBs). Changes in the residence time of 
particulate matter in a control device may affect the degree of 
chlorination of dioxins/furans, and thus the toxicity equivalents of 
the dioxins/furans. Given these uncertainties, the health risks posed 
by dioxins/furans, and the relatively low cost of dioxin/furan testing, 
it appears prudent to require confirmatory testing to determine if 
changes in feedstocks or operations that are not limited by the MACT 
rule may have increased dioxin/furan emissions to levels exceeding the 
standard. We also note that confirmatory dioxin/furan testing is 
required for municipal waste combustors (60 FR at 65402 (December 19, 
1995)).
    Confirmatory testing differs from comprehensive testing, however, 
in that you are required to operate under normal, representative 
conditions during confirmatory testing. This will reduce the cost of 
the test, while providing the essential information, because you will 
not have to establish new operating limits based on the confirmatory 
test.
1. Do the Comprehensive Testing Requirements Apply to Confirmatory 
Testing?
    The following comprehensive performance testing requirements 
discussed above also apply to confirmatory testing: Agency oversight, 
notification of performance test, notification of compliance, time 
extensions, and failure to submit a timely notice of compliance. 
However, we modify some of the comprehensive test requirement for 
confirmatory tests, as discussed below.
2. What Is the Testing Frequency for Confirmatory Testing?
    You are required to conduct confirmatory performance testing 30 
months (i.e., 2.5 years) after the previous comprehensive performance 
test. The same two-month testing window, applicable for comprehensive 
tests, also applies to confirmatory tests.
    Several commenters state that the proposed schedule for 
confirmatory tests is too frequent. The April 1996 NPRM would have 
required large and off-site sources to conduct confirmatory performance 
testing 18 months after the previous comprehensive performance test. 
Small, on-site sources would have been required to conduct the testing 
30 months after the previous comprehensive performance test. One 
commenter suggests that the frequency should be at multiples of 12 
months to avoid seasonal weather problems in many locations. Other 
commenters state that EPA's justification for confirmatory tests is not 
supported by evidence

[[Page 52915]]

showing increased emissions due to equipment aging and that the 
performance of combustion practice parameters is already assured 
through continuous monitoring systems.
    We agree that due to the magnitude and expense of the test, a more 
appropriate testing schedule would be every 2.5 years, mid-way between 
the comprehensive performance test cycle. In addition, we agree that 
testing in certain locations at certain times of the year (e.g., 
northern states in the winter) can be undesirable. Although possible, 
it would add to the difficulty and expense of the testing. As 
previously discussed, sources can request a time extension to allow for 
a more appropriate testing season. However, the regulatory date for 
confirmatory testing remains midcycle to the comprehensive performance 
testing.
3. What Operations Are Allowed During Confirmatory Performance Testing?
    As proposed, you are required to operate under normal conditions 
during confirmatory performance testing. Normal operating conditions 
are defined as operations during which: (1) The continuous emissions 
monitoring systems that measure parameters that could relate to dioxin/
furan emissions--carbon monoxide or hydrocarbons--are recording 
emission levels within the range of the average value for each 
continuous emissions monitoring system (the sum of all one-minute 
averages, divided by the number of one minute averages) over the 
previous 12 months to the maximum allowed; (2) each operating parameter 
limit established to maintain compliance with the dioxin/furan emission 
standard (see discussion in Part Five, Section VI.D.1 below and 
Sec. 63.1209(k)) is held within the range of the average values over 
the previous 12 months and the maximum or minimums, as appropriate, 
that are allowed; (3) chlorine feedrates are set at normal or greater; 
and (4) when using carbon injection or carbon bed, the test is 
conducted under representative conditions for the cleaning cycle of the 
particulate matter control device. See Sec. 63.1207(g)(2).
    We define normal operating conditions in this manner because, 
otherwise, sources could elect to limit levels of the regulated dioxin/
furan operating parameters (e.g., hazardous waste feedrate, combustion 
chamber temperature, temperature at the inlet to the dry particulate 
matter control device) to ensure minimum emissions. Thus, without 
specifying what constitutes normal conditions, the confirmatory test 
could be meaningless. On the other hand, the definition of normal 
conditions is broad enough to allow adequate flexibility in operations 
during the test. The confirmatory test confirms that your under day-to-
day operations are meeting the dioxin/furan standard. Thus, the 
confirmatory test differs from the comprehensive performance test in 
which you may choose to extend to the edge of the operating envelope to 
establish operating parameters.
    The April 1996 NPRM would have required normal operating conditions 
for particulate matter continuous emissions monitoring systems. For the 
final rule, particulate matter levels are limited during confirmatory 
testing to ensure normal operations only when your source is equipped 
with carbon injection or carbon bed for dioxin/furan emissions control 
(see dioxin/furan operating limits discussion below).
    The April 1996 NPRM also would have required you to operate under 
representative conditions for types of organic compounds in the waste 
(e.g., aromatics, aliphatics, nitrogen content, halogen/carbon ratio, 
oxygen/carbon ratio) and volatility of wastes when demonstrating 
compliance with the dioxin/furan emission standard. Several commenters 
object to this requirement. We agree that restrictions on these organic 
compounds in the waste are redundant and not necessary to assure good 
combustion. In addition, the requirement would be impracticable because 
in most cases measured data would not be available on these parameters. 
Therefore, the final rule does not require ``representative'' wastes 
with regard to these organic compounds for confirmatory testing.
    It is prudent to require that chlorine be fed at normal levels or 
greater during the dioxin/furan confirmatory performance test. Although 
most studies show poor statistical correlation between dioxin/furan 
emissions and chlorine feedrate, some practical considerations are 
important. Chlorinated dioxin/furan obviously contain chlorine and some 
level of chlorine is necessary for its formation. During the 
confirmatory testing for dioxin/furan, we want you to operate your 
combustor under normal conditions relative to factors that can affect 
emissions of dioxin/furan. Therefore, you must feed chlorine at normal 
or greater levels given the potential for chlorine feedrates to affect 
dioxin/furan emissions. For the confirmatory performance test, normal 
is defined as the average chlorine fed over the previous 12 months. If 
you have established a maximum chlorine value for metals or total 
chlorine compliance in your previous comprehensive performance test, 
then that value can be used in the confirmatory test.
    Several commenters suggest that when defining normal operation, a 
provision should be made to exclude inappropriate data, such as those 
occurring during instrument malfunction, at unit down time, or during 
instrument zero/calibration adjustment. The April 1996 NPRM did not 
allow for any data to be excluded. To define ``normal'' operation, we 
agree it is reasonable to exclude inappropriate data. For the final 
rule, calibration data, malfunction data, and data obtained when not 
burning hazardous waste do not fall into the definition of ``normal'' 
operation.
4. What Are the Consequences of Failing a Confirmatory Performance 
Test?
    If you determine that you failed the dioxin/furan emission standard 
based on results of analysis of samples taken during manual stack 
sampling, you must immediately stop burning hazardous waste. You must 
then modify the design or operation of the unit, conduct a new 
comprehensive performance test to demonstrate compliance with the 
dioxin/furan emission standard (and other standards if the changes 
could adversely affect compliance with those standards), and establish 
new operating parameter limits. Further, prior to submitting a NOC 
based on the new comprehensive performance test, you can burn hazardous 
waste only for a total of 720 hours (renewable based on the discretion 
of the Administrator) and only for purposes of pretesting or 
comprehensive performance testing. These conditions apply when you fail 
the initial or any periodic confirmatory performance test.
    However, if you conduct the comprehensive performance test under 
two or more modes of operation, and meet the dioxin/furan emission 
standards during confirmatory testing when operating under one or more 
modes of operation, you may continue burning under the modes of 
operation for which you meet the standards.
    Other than stopping burning of hazardous waste, we do not 
specifically address the consequences of failing the confirmatory 
performance test in the regulatory language but will instead rely on 
the regulating agency's enforcement policy to govern the type of 
enforcement response at a facility that exceeds an emission standard, 
fails to ensure compliance with the standards, or fails to meet a 
compliance deadline. This approach is consistent with the way

[[Page 52916]]

other MACT standards are implemented.
    Some commenters suggest that the requirement to stop burning waste 
after a failed confirmatory test is overly harsh. They suggest that 
temporarily restricted burning should be allowed, conservative enough 
to insure compliance, while a permanent solution is developed. We 
continue to believe that a source should stop burning hazardous waste 
until it reestablishes operating parameter limits that ensure 
compliance with the dioxin/furan emission standard. We note that 
hazardous waste combustors are currently subject to virtually these 
same requirements under RCRA rules.
D. What Is the Relationship Between the Risk Burn and Comprehensive 
Performance Test?
1. Is Coordinated Testing Allowed?
    Traditionally, a RCRA trial burn serves three primary functions: 
(1) Demonstration of compliance with performance standards such as 
destruction and removal efficiency; (2) determination of operating 
conditions that assure the hazardous waste combustor can meet 
applicable performance standards; and (3) collection of emissions data 
for incorporation into a SSRA that, subsequently, is used to establish 
risk-based permit conditions where necessary.185 Today's 
rulemaking transfers the first two functions of a RCRA trial burn from 
the RCRA program to the CAA program. The responsibility for collecting 
emissions data needed to perform a SSRA is not transferred because 
SSRAs are exclusively a RCRA matter.
---------------------------------------------------------------------------

    \185\ Under 40 CFR 270.10(k), which is the RCRA Part B 
information requirement that supports implementation of the RCRA 
omnibus permitting authority, a regulatory authority may require a 
RCRA permittee or an applicant to submit information to establish 
permit conditions as necessary to protect human health and the 
environment. Under this authority, risk burns and SSRAs may be 
required.
---------------------------------------------------------------------------

    Generally speaking, the type of emissions data needed to conduct a 
SSRA includes concentration and gas flow rate data for dioxin/furans, 
nondioxin/furan organics, metals, hydrogen chloride, and chlorine gas. 
Additionally, particle-size distribution data are normally needed for 
the air modeling component of the SSRA. We have recently published 
guidance on risk burns and the data to be collected. See USEPA, ``Human 
Health Risk Assessment Protocol for Hazardous Waste Combustion 
Facilities'' External Peer Review Draft, EPA-530-D-98-001A, B & C and 
USEPA, ``Guidance on Collection of Emissions Data to Support Site-
Specific Risk Assessments at Hazardous Waste Combustion Facilities,'' 
EPA 530-D-98-002, August 1998.
    A large number of hazardous waste combustors subject to today's 
rule will have completed a RCRA trial burn and SSRA emissions testing 
prior to the date of the MACT comprehensive performance test. There may 
exist, however, some facilities for which this is not the case. For 
these facilities, the Agency proposed, in both the April 1996 NPRM and 
the May 1997 NODA, an option of coordinating SSRA emissions data 
collection with MACT performance testing. Facilities choosing to 
perform coordinated testing would be expected to factor SSRA data 
collection requirements into the MACT performance test plan. Commenters 
support this approach, emphasizing that coordinated testing would 
conserve the resources of both the regulatory authority and regulated 
source. The Agency agrees with the commenters and continues to support 
coordinated testing. There is no need, however, for today's final rule 
to include regulatory language for coordinated testing since it is 
simply matter of submitting and implementing a test plan which 
accomplishes the objectives of both a risk burn and MACT performance 
test.
    Coordinated testing may not be possible for all hazardous waste 
combustors subject to today's MACT standards. Some sources may not be 
able to test under one set of conditions that addresses all data needs 
for both MACT implementation and SSRAs. SSRA emissions testing 
traditionally is performed under worst-case conditions, but may be 
obtained under normal testing conditions when necessary.186 
As noted in the April 1996 NPRM, as well as in this preamble, we 
generally anticipate sources will conduct MACT performance testing 
under conditions that are at the edge of the operating envelope or the 
worst-case to ensure operating flexibility. Regardless of which test 
conditions are used to collect SSRA emissions data, under the 
coordinated testing scenario, those conditions should be consistent 
with the MACT performance test to the extent possible.
---------------------------------------------------------------------------

    \186\ Criteria for determining the circumstances under which 
SSRA emissions data should be collected using normal versus worst-
case testing conditions are provided in EPA's Guidance on Collection 
of Emissions Data to Support Site-Specific Risk Assessments at 
Hazardous Waste Combustion Facilities (EPA 530-D-98-002, August 
1998).
---------------------------------------------------------------------------

    Similarly, a source may experience difficulty integrating MACT 
performance testing with SSRA emissions testing due to conflicting 
goals in establishing enforceable operating parameters, i.e., a 
parameter cannot be maximized for purposes of the SSRA data collection 
while at the same time be properly maximized or minimized for purposes 
of performance testing. It is additionally important to ensure that the 
feed material used during the performance testing is appropriate for 
SSRA emissions testing. When collecting emissions data for a SSRA, 
testing with actual worst-case waste is preferred to ensure that the 
testing material is representative of the toxic, persistence and 
bioaccumulative characteristics of the waste that ultimately will be 
burned. However, even if multiple tests need to be performed to 
accomplish all of the objectives, it is still advantageous to conduct 
these tests in the same general time frame to minimize mobilization and 
sampling costs.
    The timing of the required tests may cause difficulty for some 
sources wishing to use coordinated testing. As we discussed in the May 
1997 NODA, if the timing of the SSRA data collection does not coincide 
with the MACT performance test requirement, the performance test should 
not be unduly delayed. Commenters agree with this approach.
2. What Is Required for Risk Burn Testing?
    We expect that sources for which coordinated testing is not 
possible will need to obtain SSRA emissions data through a separate 
risk burn. Similar to a traditional RCRA trial burn, risk burn testing 
should be conducted pursuant to a test plan that is reviewed and 
approved by the RCRA permitting authority. 40 CFR 270.10(k) provides 
that the permitting authority may require the submittal of information 
to establish permit conditions to ensure a facility's operations will 
be protective of human health and the environment. This regulatory 
requirement provides for the collection of emissions data, as 
appropriate, for incorporation into a SSRA as well as for the 
performance of the SSRA itself. We clarify in amendments to 
Secs. 270.19, 270.22, 270.62 and 270.66 that the Director may apply 
provisions from those sections, on a case-by-case basis, to establish a 
regulatory framework for conducting the risk burn under Sec. 270.10(k) 
and imposing risk-based conditions under Sec. 270.32(b)(2) (omnibus 
provisions). This clarifying language is intended to prevent any 
confusion from other language added to Secs. 270.19, 270.22, 270.62 and 
270.66 today stating that

[[Page 52917]]

these provisions otherwise no longer apply once a source has 
demonstrated compliance with the MACT standards and limitations of 40 
CFR part 63, subpart EEE. (See Part Five, Section XI.B.3 for further 
discussion.) Facilities and regulatory authorities may consult existing 
EPA guidance documents for information regarding the elements of risk 
burn testing.187
---------------------------------------------------------------------------

    \187\ USEPA. ``Human Health Risk Assessment Protocol for 
Hazardous Waste Combustion Facilities'' External Peer Review Draft. 
EPA-530-D-98-001A,B&C. Date.; USEPA, ``Guidance on Collection of 
Emissions Data to Support Site-Specific Risk Assessments at 
Hazardous Waste Combustion Facilities'' EPA 530-D-98-002. August 
1998.
---------------------------------------------------------------------------

E. What Is a Change in Design, Operation, and Maintenance? (See 
Sec. 63.1206(b)(6).)
    The April 1996 NPRM noted that sources may change their design, 
operation, or maintenance practices in a manner that may adversely 
affect their ability to comply with the emission standards. These 
sources would be required to conduct a new comprehensive performance 
test to demonstrate compliance with the affected emission standards and 
would be required to re-establish operating limits on the affected 
parameters specified in Sec. 63.1209. (See 61 at FR 17518.) The 
proposal stated that until a complete and accurate revised NOC is 
submitted to the Administrator, sources would be permitted to burn 
hazardous waste following such changes for time a period not to exceed 
720 hours and only for the purposes of pretesting or comprehensive 
performance testing. The approach in the April 1996 NPRM remains 
appropriate, and we are adopting it in today's final rule with minor 
modifications.
    For changes made after submittal of your NOC that may adversely 
affect compliance with any emission standard, as defined later in this 
section, today's rule requires you to notify the Administrator at least 
60 days prior to the change unless you document circumstances that 
dictate that such prior notice is not reasonably feasible. The 
notification must include a description of the changes and which 
emission standards may be affected. The notification must also include 
a comprehensive performance test schedule and test plan that will 
document compliance with the affected emission standard(s). You must 
conduct a comprehensive performance test to document compliance with 
the affected emission standard(s) and establish operating parameter 
limits as required and submit a revised NOC to the Administrator. You 
also must not burn hazardous waste for more than a total of 720 hours 
after the change and prior to submitting your NOC, and you must burn 
hazardous waste during this time period only for the purposes of 
pretesting or comprehensive performance testing.
    Some commenters are uncomfortable with the proposed regulatory 
language, stating that it was too generic and that the Agency could 
require a comprehensive performance test even after minor changes in 
maintenance practices. One commenter suggests that EPA incorporate a 
list of changes significant enough to affect compliance, similar to 
what is currently done in the RCRA permit modification classification 
scheme in Appendix I of Sec. 270.42.
    We intentionally proposed an approach that provides some degree of 
flexibility to permit authorities. Individual facilities will need to 
consult with these permit authorities who will make the decision on the 
site-specific facts. We do not intend to require a comprehensive 
performance test after minor modifications to system design, or after 
implementing minor changes to operating or maintenance practices. We 
considered incorporating sections of Appendix I of Sec. 270.42 to 
further clarify when comprehensive performance tests would be 
required.188 However, it is impossible to envision all 
scenarios in which changes in design, operation, or maintenance 
practices may or may not trigger the requirement of a complete, or even 
partial, comprehensive performance test. Discussion of specific 
scenarios is more suitable in an Agency guidance document as opposed to 
regulatory provisions, and implemented on a site-specific basis. Thus, 
the April 1996 NPRM set out the regulatory approach as well as can be 
done, and we are adopting it today with minor modifications.
---------------------------------------------------------------------------

    \188\ One approach would be to require performance tests for 
modifications covered by the class 2 and class 3 permit 
modifications associated with combustion source design and operating 
parameter changes.
---------------------------------------------------------------------------

    In the April 1996 NPRM, we did not address what must be done when 
you change design, operation, or maintenance practices during the time 
period between the compliance date and when you submit your NOC. If you 
make a change during this time period, today's rule requires you to 
revise your DOC, which is maintained on-site, to incorporate any 
revised limits necessary to comply with the standards. For purposes of 
this provision, today's rule defines ``change'' as any change in 
reported design, operation, or maintenance practices you previously 
documented to the Administrator in your comprehensive performance test 
plan, NOC, DOC, or startup, shutdown, and malfunction plan.
    Commenters point out that the proposal did not discuss 
recordkeeping requirements necessary for the Administrator to determine 
if you are adequately concluding that changes in design, operation, or 
maintenance practices do not trigger a comprehensive performance test 
requirement 189. As a result, today's rule requires you to 
document in your operating record whenever you make a change (as 
defined above) in design, operation, or maintenance practices, 
regardless of whether the change may adversely affect your ability to 
comply with the emission standards. See Sec. 63.1206(b)(6)(ii). You are 
also required to maintain on site an updated comprehensive performance 
test plan, NOC, and startup, shutdown, and malfunction plan that 
reflect these changes. See Sec. 63.1211(c).
---------------------------------------------------------------------------

    \189\ We cannot determine if a source has accurately concluded 
that a change does not adversely affect its ability to comply with 
the emission standards if we are never aware that changes were made 
to the source.
---------------------------------------------------------------------------

F. What Are the Data In Lieu Allowances?
    You are allowed to submit data from previous emissions tests in 
lieu of performing a MACT performance test to set operating limits. See 
Sec. 63.1207(c)(2). To use previous emissions test data, the data must 
have been collected less than 5 years before the date you intend to 
submit your notification of compliance. The data must also have been 
collected as part of a test that was for the purpose of demonstrating 
compliance with RCRA or CAA requirements. Additionally, you must submit 
your request to use previous test data in your comprehensive 
performance test plan which is submitted 1 year in advance of the MACT 
performance test. Finally, you must schedule your subsequent MACT 
performance test and MACT confirmatory test 5 years and 2.5 years 
respectively following the date the emissions test data your submitting 
was collected.
    We developed this allowance in response to comments that suggested 
we should allow previous RCRA testing to be used in lieu of performing 
a new MACT performance test if the data could be used to demonstrate 
compliance and establish operating limits to ensure compliance with the 
MACT emissions standards. Commenters reasoned, and we agreed, that such 
an allowance was reasonable and necessary for those sources that

[[Page 52918]]

must perform emissions tests to satisfy other state or federal 
requirements. As we developed this allowance, we decided that it is 
necessary to limit the age of the data and specify the date of the 
following performance test because we need to be consistent with the 
MACT performance test requirements with respect to testing frequency. 
We can further justify the time and testing limitations of the data in 
lieu of allowance by acknowledging that we don't want some sources 
gaining an advantage over others by extending the date between 
performance tests. However, we also weighed the fact that some sources 
may be required to perform RCRA testing fairly close to the compliance 
date or promulgation date of today's rule and we didn't want to 
penalize them by forcing them to perform a new performance test before 
five years had elapsed since their previous test. So we settled on an 
approach that allows the use of previous emissions test data and 
effectively sets the same testing frequency as is applied to test data 
collected via a MACT performance test following the compliance date. 
This approach doesn't penalize or favor any source over another and it 
allows each source to take advantage of this provision when it makes 
sense. For instance, a source may be granted approval to use data from 
a RCRA trial burn performed 1 year before today's date, thus not 
requiring the source to perform a comprehensive performance test 270 
days following the compliance date. Instead, the source must schedule 
its next MACT performance test five years after the date the test was 
performed. However, the source must perform a confirmatory test 270 
days following the compliance date because the test schedule for the 
confirmatory test is also linked to the date of the performance test. 
So in this situation the source must determine if its better to run the 
comprehensive performance test on a normal schedule after the 
compliance date or delay the comprehensive test and perform a 
confirmatory test instead.

VI. What Is the Notification of Compliance?

A. What Are the Requirements for the Notification of Compliance?
    You must submit to the Administrator the results of the 
comprehensive performance test in a notification of compliance (NOC) no 
later than three months after the conclusion of the performance test. 
You must submit the initial NOC later than nine months following the 
compliance date.
B. What Is Required in the NOC?
    You must include the following information in the NOC:

--Results of the comprehensive performance test, continuous monitoring 
system performance evaluation, and any other monitoring procedures or 
methods that you conducted;
--Test methods used to determine the emission concentrations and 
feedstream concentrations, as well as a description of any other 
monitoring procedures or methods that you conducted;
--Limits for the operating parameters;
--Procedures used to identify the operating parameter limits specified 
in Sec. 63.1209;
--Other information documenting compliance with the operating 
requirements, including but not limited to automatic waste feed cutoff 
system operability and operator training;
--A description of the air pollution control equipment and the 
associated hazardous air pollutant that each device is designed to 
control; and
--A statement from you or your company's responsible official that the 
facility is in compliance with the standards and requirements of this 
rule.
C. What Are the Consequences of Not Submitting a NOC?
    The normal CAA enforcement procedures apply if you fail to submit a 
timely notification of compliance. We do not adopt our proposed 
approach that would have required you to immediately stop burning 
hazardous waste if you failed to submit a timely NOC.
    We proposed regulatory language stating that failure to submit a 
notification of compliance by the required date would result in the 
source being required to immediately stop burning hazardous waste. This 
proposal was similar to requirements applied to BIFs certifying 
compliance under RCRA. Under the proposal, if you wanted to burn 
hazardous waste in the future, you would be required to comply with the 
standards and permit requirements for new MACT and RCRA sources.
    In the 1997 NODA, however, we proposed to rely on the regulating 
agency's policy regarding enforcement response to govern the type of 
enforcement response at a facility that fails to submit a notification 
of compliance. Based on NODA comments and review of this enforcement 
process, we are not including in the final rule regulatory language 
addressing the consequences of failure to submit a timely or complete 
NOC. Instead, we rely on the regulating agency's policy regarding 
enforcement response to govern the type of enforcement response at a 
facility that fails to meet a compliance deadline. This approach is 
more practical to implementing today's MACT standards and is more 
consistent with the way other MACT standards are implemented.
D. What Are the Consequences of an Incomplete Notification of 
Compliance?
    In response to our April 1996 NPRM, commenters state that we were 
unclear as to the consequences of an incomplete NOC. Furthermore, 
commenters state that it was important that we specify what is needed 
and the consequences if an NOC is incomplete or more information is 
needed. Additionally, commenters recommend that if the NOC contains 
emission information, the certification statement, and a signature, we 
should judge the NOC to be administratively complete and an acceptable 
submission. In addition, commenters suggest that if the regulatory 
official reviewing the NOC determines that additional information is 
required, the source should be given ample time to submit that 
information.
    Our enforcement approach to incomplete submissions, under RCRA or 
the CAA, is generally determined on a site-specific basis. We will not 
attempt to foresee and develop enforcement responses to all the 
possible levels of incompleteness for the NOC. This is beyond the scope 
of our national rulemaking. Furthermore, defining what constitutes an 
incomplete submission requires us to specifically prescribe a complete 
submission, which is not possible for all situations or all source 
designs. Some sources may require more detail than others in defining 
the parameters necessary to determine compliance on a continuous basis. 
Therefore, we instead define the minimum information necessary in the 
submission and allow the implementing agency to determine if more 
information is necessary in a facility's site-specific NOC.
    In response to comments advocating that facilities be given ample 
time to submit additional information required by the regulatory 
official, we prefer to allow the implementing agency to determine the 
time periods that will be granted to submit additional information 
because some information requests may require widely varying degrees of 
time and effort to develop. Many potential problems associated with 
incomplete submissions can be prevented through interaction between

[[Page 52919]]

the source and the regulatory agency during the test plan review and 
approval process. We do not want our rules to act as disincentive to 
those discussions by providing a complete shield, regardless of the 
severity of the omission.
E. Is There a Finding of Compliance?
    We adopt the requirement we proposed for the regulatory agencies to 
make a finding of compliance based on performance test results (see 
Sec. 63.1206(b)(3)). This provision specifies that the regulatory 
agency must determine whether an affected source is in compliance with 
the emissions standards and other requirements of subpart EEE, as 
provided by the general provisions governing findings of compliance in 
Sec. 63.6(f)(3). Thus, the regulatory agency is obligated to make this 
finding upon obtaining all the compliance information required by the 
standards, including the written reports of performance test results, 
monitoring results, and other applicable information. This includes, 
but may not be limited to, the information submitted by the source in 
its NOC.

VII. What Are the Monitoring Requirements?

    In this section, we discuss the following topics: (1) The 
compliance monitoring hierarchy that places a preference on compliance 
with a CEMS; (2) how limits on operating parameters are established 
from comprehensive performance test data; (3) status and use of CEMS 
other than carbon monoxide, hydrocarbon, and oxygen CEMS; and (4) final 
compliance monitoring requirements for each emission standard.
A. What Is the Compliance Monitoring Hierarchy?
    We proposed the following three-tiered compliance monitoring 
hierarchy in descending order of preference to ensure compliance with 
the emission standards: (1) Use of a continuous emission monitoring 
system (CEMS) for a hazardous air pollutant; (2) absent a CEMS for that 
hazardous air pollutant, use of a CEMS for a surrogate of that 
hazardous air pollutant and, when necessary, setting limits on 
operating parameters to account for the limitations of using 
surrogates; and (3) lacking a CEMS for either, requiring periodic 
emissions testing and site-specific limits on operating parameters. 
Accordingly, we proposed to require the use of carbon monoxide, 
hydrocarbon, oxygen, particulate matter, and total mercury CEMS. We 
also proposed performance specifications for multimetal, hydrochloric 
acid, and chlorine gas CEMS to give sources the option of using a CEMS 
for compliance with the semivolatile and low volatile metal emissions 
standards, and the hydrochloric acid/chlorine gas emission standard.
    Commenters question the availability and reliability of CEMS other 
than those for carbon monoxide, hydrocarbon, and oxygen. We concur with 
some of the commenters' concerns and are not requiring use of a total 
mercury CEMS in the final rule or specifying the installation deadline 
and performance specifications for particulate matter CEMS. In 
addition, we have not promulgated performance specifications for these 
CEMS or multimetal, hydrochloric acid, and chlorine gas CEMS. We 
nonetheless continue to encourage sources to evaluate the feasibility 
of using these CEMS to determine the performance specifications, 
correlation acceptance criteria, and detector availability that can be 
achieved. Sources may request approval from permitting officials under 
Sec. 63.8(f) to use CEMS to document compliance with the emission 
standards in lieu of periodic performance testing and compliance with 
limits on operating parameters. See discussion in Section VII.C below 
on these issues.
B. How Are Comprehensive Performance Test Data Used To Establish 
Operating Limits?
    In this section, we discuss: (1) The definitions of terms related 
to monitoring and averaging periods; (2) the rationale for the 
averaging periods for operating parameter limits, (3) how comprehensive 
performance test data are averaged to calculate operating parameter 
limits; (4) how the various types of operating parameters are 
monitored/established; (5) how nondetect performance test feedstream 
data are handled; and (6) how rolling averages are calculated 
initially, upon intermittent operations, and when the hazardous waste 
feed is cut off.
1. What Are the Definitions of Terms Related to Monitoring and 
Averaging Periods?
    In the April 1996 NPRM, we proposed definitions for several terms 
that relate to monitoring and averaging periods. For the reasons 
discussed below, we conclude that the proposed definitions are 
appropriate and are adopting them in today's rule. We also finalize 
definitions for ``average run average'' and ``average highest or lowest 
rolling average'' which were not proposed. We conclude these new 
definitions are necessary to clarify the meaning and intent of 
regulatory provisions associated with the monitoring requirements that 
are discussed in Part 5, Section VII.D. of this preamble.
    We promulgate the following definitions in today's rule (see 
Sec. 63.1201).
    ``Average highest or lowest rolling average'' means the average of 
each run's highest or lowest rolling average run within the test 
condition for the applicable averaging period.
    ``Average run average'' means the average of each run's average of 
all associated one minute values.
    ``Continuous monitor'' means a device that: (1) Continuously 
samples a regulated parameter without interruption; (2) evaluates the 
detector response at least once every 15 seconds; and (3) computes and 
records the average value at least every 60 seconds, except during 
allowable periods of calibration and as defined otherwise by the CEMS 
Performance Specifications in appendix B of part 60.
    ``Feedrate operating limits'' means limits on the feedrate of 
materials (e.g., metals, chlorine) to the combustor that are 
established based on comprehensive performance testing. The limits are 
established and monitored by knowing the concentration of the limited 
material (e.g., chlorine) in each feedstream and the flow rate of each 
feedstream.
    ``Feedstream'' means any material fed into a hazardous waste 
combustor, including, but not limited to, any pumpable or nonpumpable 
solid, liquid, or gas.
    ``Flowrate'' means the rate at which a feedstream is fed into a 
hazardous waste combustor.
    ``Instantaneous monitoring'' means continuously sampling, 
detecting, and recording the regulated parameter without use of an 
averaging period.
    ``One-minute average'' means the average of detector responses 
calculated at least every 60 seconds from responses obtained at least 
each 15 seconds.
    ``Rolling average'' means the average of all one-minute averages 
over the averaging period.
    One commenter opposes the requirement to take instrument readings 
every 15 seconds. This commenter contends that such an approach is 
simply impractical, unnecessary, and imposes a harsh burden upon 
members of the regulated community. Another commenter maintains that 
the CEMS Data Acquisition System should be capable of sampling the 
analyzer outputs at least every 15 seconds. With today's processing 
power and speed, the commenter states that this can easily be achieved. 
We agree with the second commenter and are requiring instrument

[[Page 52920]]

readings at least every 15 seconds because this is currently required 
in the Boilers and Industrial Furnace rulemaking. (See 
Sec. 266.102(e)(6))
    Another commenter states that the Agency's definition of 
``instantaneous monitoring'' of combustion chamber pressure to control 
combustion system leaks is not clear.190 The commenter 
states that, although an instantaneous limit cannot be exceeded at any 
time, continuous monitoring systems are required to detect parameter 
values only once every 15 seconds. We note that the final rule requires 
instantaneous monitoring only for the combustion chamber pressure limit 
to control combustion system leaks. The rule requires an automatic 
waste feed cutoff if the combustion chamber pressure at any time (i.e., 
instantaneously) exceeds ambient pressure (see Sec. 63.1209(p)). The 
definition of a continuous monitoring system is that it must record 
instrument readings at least every 15 seconds. For instantaneous 
monitoring of pressure, the detector must clearly record a response 
more frequently than every 15 seconds.191 It must detect and 
record pressure constantly without interruption and without any 
averaging period.
---------------------------------------------------------------------------

    \190\ ``Combustion system leaks'' is the term used in today's 
rule to refer to leaks that are called fugitive emissions under 
current RCRA regulations. We use the term combustion system leaks to 
refer to those emissions because the term fugitive emissions has 
other meanings under part 63.
    \191\ Typical pressure transducers in use today are capable of 
responding to pressure changes once every fifty milliseconds. See 
USEPA, ``Final Technical Support Document for Hazardous Waste 
Combustor MACT Standards, Volume IV: Compliance with the Hazardous 
Waste Combustor Standard,'' July 1999.
---------------------------------------------------------------------------

2. What Is the Rationale for the Averaging Periods for the Operating 
Parameter Limits?
    The final rule establishes the following averaging periods: (1) No 
averaging period (i.e., instantaneous monitoring) for maximum 
combustion chamber pressure to control combustion system leaks; (2) 12-
hour rolling averages for maximum feedrate of mercury, semivolatile 
metals, low volatile metals, chlorine, and ash (for incinerators); and, 
(3) one-hour averaging periods for all other operating parameters. As 
discussed later in this section, we conclude that the proposed ten-
minute averaging periods are not necessary, on a national basis, to 
better ensure compliance with the emission standards at hazardous waste 
combustors, and have not adopted these averaging periods in this 
rulemaking.
    a. When Is an Instantaneous Limit Used? An instantaneous limit is 
required only for maximum combustion chamber pressure to control 
combustion system leaks. This is because any perturbation above the 
limit may result in uncontrolled emissions exceeding the standards.
    b. When Is an Hourly Rolling Average Limit Used? An hourly rolling 
average limit is required for all parameters that are based on 
operating data from the comprehensive performance test, except 
combustion chamber pressure and feedrate limits. Hourly rolling 
averages are required for these parameters rather than averaging 
periods based on the duration of the performance test because we are 
concerned that there may be a nonlinear relationship between operating 
parameter levels and emission levels of hazardous air pollutants.
    c. Why Has the Agency Decided Not to Adopt Ten-Minute Averaging 
Periods? Dual ten-minute and hourly rolling averages were proposed for 
most parameters for which limits are based on the comprehensive 
performance test. See 61 FR at 17417. We proposed ten-minute rolling 
averages in addition to hourly rolling averages for these parameters 
because short term excursions of the parameter can result in a 
disproportionately large excursion of the hazardous air pollutant being 
controlled.
    Commenters claim that the Agency's concerns with emission 
excursions due to short term perturbations of these operating 
parameters were not supported with data and are therefore unjustified, 
and claim that averaging periods shorter than those required in the 
existing BIF regulations would provide no environmental benefit.
    We acknowledge that the Agency does not have extensive short-term 
emission data that show operating parameter excursions can result in 
disproportionately large excursions of hazardous air pollutants being 
emitted. These short-term data cannot be obtained without the use of 
continuous emission monitors that measure dioxin/furans, metals, and 
chlorine on a real-time basis. Such monitors, for the most part, are 
not currently used for compliance purposes at hazardous waste 
combustors. However, known relationships between operating parameters 
and hazardous air pollutant emissions indicate that a nonlinear 
relationship exists between operating parameter levels and emissions. 
This nonlinear relationship can result in source emissions that exceed 
levels demonstrated in the performance test if the operating parameters 
are not properly controlled. An explanation of these nonlinear 
relationships, including examples that explain why this relationship 
can result in daily emissions that exceed levels demonstrated in the 
performance test, are included in the Final Technical Support 
Document.192 Thus, at least in theory, an environmental 
benefit can result from shorter averaging periods, including ten-minute 
rolling averages and perhaps instantaneous readings in certain 
situations.
---------------------------------------------------------------------------

    \192\ See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume IV: Compliance With 
the Hazardous Waste Combustor Standards, July 1999, Chapters 2 and 
3.
---------------------------------------------------------------------------

    We also acknowledge, however, that the Agency's ability to assess 
this potential benefit in practice for all hazardous waste combustors 
affected by this final rule is limited significantly by the paucity of 
short-term, minute-by-minute, operating parameter data. Without this 
data we cannot effectively evaluate whether operating parameter 
excursions occur to an extent that warrant national ten-minute 
averaging period requirements for all hazardous waste combustors. We 
therefore conclude that averaging period requirements shorter than 
those required by existing BIF regulations are not now appropriate for 
adoption on a national level, and do not adopt ten-minute averaging 
period requirements in this rulemaking.
    We maintain, however, that there may be site-specific circumstances 
that warrant averaging periods shorter than one hour in duration, 
including possibly instantaneous measurements. Regulatory officials may 
determine, on a site-specific basis, that shorter averaging periods are 
necessary to better assure compliance with the emission standards. The 
provisions in Sec. 63.1209(g)(2) authorize the regulatory official to 
make such a determination. Factors that may be considered when 
determining whether shorter averaging periods are appropriate include 
(1) the ability of a source to effectively control operating parameter 
excursions to levels achieved during the performance test; (2) the 
source's previous compliance history regarding operating parameter 
limit exceedances; and (3) the difference between the source's 
performance test emission levels and the relevant emission standard. 
For additional information, see the Final Technical Support Document, 
Volume 4, Chapter 2.
    d. What Is the Basis for 12-Hour Rolling Averages for Feedrates? 
The rule requires 12-hour averages for the feedrate of mercury, 
semivolatile metals, low volatile metals, chlorine, and ash (for 
incinerators) because feedrate and emissions are, for the most part, 
linearly

[[Page 52921]]

related. A 12-hour averaging period for feedrates is appropriate 
because it is the upper end of the range of time required to perform 
three runs of a comprehensive performance test. Thus, a 12-hour 
averaging period will ensure (if all other factors affecting emissions 
are constant) that emissions will not exceed performance test levels 
during any interval of time equivalent to the time required to conduct 
a performance test. A 12-hour averaging period is also achievable and 
appropriate from a compliance perspective because the emission 
standards are based on emissions data obtained over (roughly) these 
sampling periods.193
---------------------------------------------------------------------------

    \193\ See Chemical Waste Management v. EPA, 976 F.2d, 2, 34 
(D.C. Cir. 1992) (It is inherently reasonable to base compliance on 
the same type of data used to establish the requirement).
---------------------------------------------------------------------------

    e. Has the Agency Over-Specified Compliance Requirements? Some 
commenters state that the Agency is over-specifying compliance 
requirements by requiring limits on many operating parameters, 
requiring dual ten-minute and hourly rolling average limits on many 
parameters, and requiring that sources interlock the operating 
parameter limits with the automatic waste feed cutoff system. These 
commenters wrote that this compliance regime may lead to system over-
control and instability, and an unreasonable and unnecessary increase 
in automatic waste feed cutoffs, a result that is contrary to good 
process control principles. They propose that we work with industry to 
develop a process control system and performance specification 
regulatory approach to establish minimum system standards. These would 
include: (1) Minimum process instrument sampling time; (2) maximum 
calculation capability for output signals; (3) minimum standard for 
process control sequences; and (4) minimum requirements for 
incorporating automatic waste feed cutoffs into the control scheme. The 
specifications would be incorporated into guidance, rather than 
regulation. Commenters suggest that the rule should only specify 
general goals, similar to the guidance approach we took for hazardous 
waste incinerators in the 1981 RCRA regulations.194
---------------------------------------------------------------------------

    \194\ The incinerator regulations promulgated in 1981, at the 
outset of the RCRA regulatory program, used such a general guidance 
approach. However, sources have had over 15 years since then to gain 
experience with process control techniques associated with the 
combustion of hazardous waste.
---------------------------------------------------------------------------

    We evaluated these comments carefully, balancing the need to 
provide industry with operational flexibility with the need for 
compliance assurance. As previously discussed, we are not adopting ten-
minute averaging period requirements in this rulemaking, although it 
can be imposed on a site-specific basis under appropriate 
circumstances. This addresses commenter's concerns that relate to the 
complexity of the proposed dual averaging period requirements. We 
acknowledge, however, that today's rule requires that more operating 
parameter limits be interlocked to the automatic waste feed cutoff 
system than is currently required by RCRA regulations. Nonetheless, we 
conclude that the compliance regime of today's final rule is necessary 
to ensure compliance with the emission standards and will not overly 
constrain process control systems for the following reasons.
    Automatic waste feed cutoffs are (by definition) automatic, and the 
control systems used to avoid automatic waste feed cutoffs require 
adequate response time and are primarily site-specific in design. The 
closer a source pushes the edge of the operating envelope, the better 
that control system must perform to ensure that an operating parameter 
limit (and emission standard) is not exceeded. Therefore, a source has 
extensive control over the impact of these requirements.
    Under the compliance regime of today's rule, sources will continue 
to perform comprehensive performance testing under ``worst case'' 
conditions as they currently do under RCRA requirements to establish 
limits on operating parameters that are well beyond normal levels. This 
cushion between normal operating levels and operating parameter limits 
enables the source to take corrective measures well before a limit is 
about to be exceeded, thus avoiding an automatic waste feed cutoff.
    Regulatory officials do not have the extensive resources that would 
be required to develop and implement industry-specific control 
guidelines and we are not confident that this approach would provide 
adequate compliance assurance. Although specifying only emissions 
standards and leaving the compliance method primarily up to the source 
and the permit writer (aided by guidance) would provide flexibility, it 
would place a burden on the permit writers and the source during the 
development and approval of the performance test plan and the finding 
of compliance subsequent to Notification of Compliance. In addition, 
this level of interaction between permitting officials and the source 
is contrary to our policy of structuring the MACT standards to be as 
self-implementing as possible.195 The Agency therefore 
maintains its position that the compliance scheme adopted in today's 
rule, is appropriate.
---------------------------------------------------------------------------

    \195\ The time that would be associated with this type of review 
and negotiation between permit writer and source would be better 
spent on developing, reviewing, and approving the comprehensive 
performance test plan under today's compliance regime.
---------------------------------------------------------------------------

    f. Why Isn't Risk Considered in Determining Averaging Periods? 
Several commenters state that long averaging periods (e.g., monthly 
metal feedrate rolling averages) for the operating parameter limits and 
CEMS-monitored emission standards would be appropriate. These 
commenters believe that long averaging periods would be appropriate 
given that the Agency has performed a risk assessment and concluded 
that the emission standards would be protective over long periods of 
exposure. They state that long averaging periods would ensure that 
emissions are safe and reduce compliance costs.
    Consideration of risk is not an appropriate basis for determining 
averaging periods to ensure compliance with the technology-based MACT 
emission standards.196 As previously stated, we must 
establish averaging periods that ensure compliance with the emission 
standard for time durations equivalent to the emission sampling periods 
used to demonstrate compliance. Longer averaging periods would not 
ensure compliance with the emission standard because many of the 
operating parameters do not relate to emissions linearly.
---------------------------------------------------------------------------

    \196\ We note, however, that within eight years of promulgating 
MACT standards for a source category, we must consider risk in 
determining under section 112(f) whether standards more stringent 
than MACT are necessary to provide an ample margin of safety to 
protect public health and the environment.
---------------------------------------------------------------------------

    In addition, a longer averaging period is not warranted even for 
those operating parameters than may relate linearly to emissions 
because this would allow a source to emit hazardous air pollutants in 
excess of the emission standard for times periods equivalent to the 
stack emission sampling periods used to demonstrate compliance. For 
example, a monthly averaging period for metal feedrates could result in 
a source emitting metals at a level three times the regulatory standard 
continuously for a one week period.197 This would not be 
consistent with the level of control that was achieved by the best 
performing sources in our data base. Modifying the results of the MACT 
process based on risk considerations is thus contrary to Congressional 
intent that MACT

[[Page 52922]]

standards, at a minimum, must represent the level of control being 
achieved by the average of the best performing 12 percent of the 
sources. We therefore conclude that we must limit averaging times at 
least to time durations equivalent to the emission sampling periods 
used to demonstrate compliance.
---------------------------------------------------------------------------

    \197\ For this to occur, the source would have to emit metals 
far below the standard for time periods before and after this one-
week period.
---------------------------------------------------------------------------

    g. Will Relaxing Feedrate Averaging Times Increase Environmental 
Loading? One commenter questions whether relaxing the averaging time 
for the feedrate of metals and chlorine from an hourly rolling average 
under current RCRA regulations to the 12-hour rolling average of 
today's rule would increase total environmental loading of pollutants 
and be counter to the Agency's pollution prevention objectives. 
Contrary to the commenter's concern, we conclude that today's rule will 
decrease environmental loading of hazardous air pollutants because the 
emission standards are generally more stringent than current RCRA 
standards. Today's standards more than offset any difference in 
environmental loading associated with longer averaging times. As 
previously discussed, the averaging periods in today's rule were chosen 
to ensure compliance with the emission standard for intervals of time 
equivalent to the time required to conduct a performance test.
    Although current RCRA standards generally establish hourly rolling 
averages for the feedrate of metals, sources are actually allowed to 
establish up to 24-hour rolling averages for arsenic, beryllium, 
chromium, cadmium, and lead, provided they restrict the feedrate of 
these metals at any time to ten times what would be normally allowed 
under an hourly rolling average basis. For these reasons, the 
commenter's concern is not persuasive.
3. How Are Performance Test Data Averaged To Calculate Operating 
Parameter Limits?
    The rule specifies which of two techniques you must use to average 
data from the comprehensive performance test to calculate limits on 
operating parameters: (1) Calculate the limit as the average of the 
maximum (or minimum, as specified) rolling averages for each run of the 
test; or (2) calculate the limit as the average of the test run 
averages for each run of the test.
    Hourly rolling averages for two parameters--combustion gas flowrate 
(or kiln production rate as a surrogate) and hazardous waste feedrate--
are based on the average of the maximum hourly rolling averages for 
each run. Hourly rolling average and 12-hour rolling average limits for 
all other parameters, however, are based on the average level occurring 
during the comprehensive performance test. We determined that this more 
conservative approach is appropriate for these parameters because they 
can have a greater effect on emissions, and because it is consistent 
with how manual method emissions results are determined.198
---------------------------------------------------------------------------

    \198\ Manual method emission test results for each run 
represents average emissions over the entire run.
---------------------------------------------------------------------------

    These are examples of how the averages work. The hourly rolling 
average hazardous waste feedrate limit for a source is calculated using 
the first technique. If the highest hourly rolling averages for each 
run of the comprehensive performance test were 200 lbs/hour, 210 lbs/
hr, 220 lbs/hr, the hourly rolling average feedrate limit would be 210 
lbs/hr.
    The second approach uses the average of the test run averages for a 
given test condition to calculate the limit. Each test run average is 
calculated by summing all the one-minute readings within the test run 
and dividing that sum by the number of one-minute readings. For 
example, if: (1) The sum of all the one-minute semivolatile metal 
feedrate readings for each run within a test condition is 2,400 lbs/
hour, 2,500 lbs/hour, and 2,600 lbs/hour; and (2) there are 240, 250, 
and 200 one-minute readings in each run, respectively; then (3) the 
average feedrate for each of these three runs is 10 lbs/hour, 10 lbs/
hour, and 13 lbs/hour, respectively. The 12-hour rolling average 
semivolatile metal feed rate limit for this example is the average of 
these three values: 11 lbs/hour. This averaging methodology is not 
equivalent to an approach where the limit is calculated by taking the 
time-weighted average over all three runs within the test condition, 
because, as noted by the example, sampling times may be different for 
each run. The time-weighted average feedrate over all three test runs 
for the previous example is equivalent to 10.9 lbs/hr.199 
Although the two averaging techniques may not result in averages that 
are significantly different, we conclude that basing the limits on the 
average of the test run averages is more appropriate, because this 
approach is identical to how we determine compliance with the emission 
standards.
---------------------------------------------------------------------------

    \199\ This time weighted average is calculated by summing all 
the one-minute feedrate values in the test condition and dividing 
that sum by the number of one minute readings in the test condition.
---------------------------------------------------------------------------

    These averaging techniques are the same as we proposed (see 61 FR 
at 17418).200 A number of commenters object to the more 
conservative second technique of basing the limits on the average 
levels that occur during the test. The commenters claim that this 
approach ensures a source would not comply with the limits 50% of the 
time when operating under the same conditions as the performance test. 
Further, they are concerned that this approach would establish 
operating parameter limits that would ``ratchet'' emissions to levels 
well below the standards, and further ratcheting would occur with each 
subsequent performance test (i.e., because the current operating limits 
could not be exceeded during subsequent performance testing). Some 
commenters prefer the approach of setting the limit as the average of 
the highest (or lowest) rolling average from each run, technique one 
above, which is the same approach used in the BIF rule.
---------------------------------------------------------------------------

    \200\ Except that average hourly rolling average limits are 
calculated as the average of the test run averages rather than 
simply the average over all runs as proposed.
---------------------------------------------------------------------------

    Notwithstanding the conservatism of the promulgated approach 
(technique two above) for many operating parameter limits, we maintain 
that the approach results in achievable limits and is necessary to 
ensure compliance with the emission standards. Comprehensive 
performance tests are designed to demonstrate compliance with the 
emission standards and establish corresponding operating parameter 
limits. Thus, sources will operate under ``worst-case'' conditions 
during the comprehensive performance tests, just as they do currently 
for RCRA trial burns. Given that the source can readily control (during 
the performance test and thereafter) the parameters for which limits 
are established based on the average of the test run averages during 
performance testing (i.e., rather than on the average of the highest 
(or lowest) hourly rolling averages), and that these parameters will be 
at their extreme levels during the performance test, the limits are 
readily achievable.
    There may be situations, however, where a source cannot 
simultaneously demonstrate worst-case operating conditions for all the 
regulated operating parameters. An example of this may be minimum 
combustion chamber temperature and maximum temperature at the inlet to 
the dry particulate matter control device because when the combustion 
chamber temperature is minimized, the inlet temperature to the control 
device may also be minimized. Sources should consult permitting 
officials to resolve

[[Page 52923]]

compliance difficulties associated with conflicting operating 
parameters. Potential solutions to conflicting parameters could be to 
conduct the performance test under two different modes of operation to 
set these conflicting operating parameter limits, or for the 
Administrator to use the discretionary authority provided by 
Sec. 63.1209(g)(2) to set alternative operating parameter limits.
    We address commenters' concern that subsequent performance tests 
would result in a further ratcheting down of operating parameter limits 
by waiving the operating limits during subsequent comprehensive 
performance tests (see Sec. 63.1207(h)). The final rule also waives 
operating limits for pretesting prior to comprehensive performance 
testing for a total operating time not to exceed 720 hours. See 
discussion in Part Five, Section VI for more information on this 
provision.
    Some commenters suggest that we use a statistical analysis to 
determine rolling average limits, such that the limits are calculated 
as the mean plus or minus three standard deviations of all rolling 
averages for all runs. Commenters state that this would ensure that the 
operating parameter limits are achievable. If such an approach were 
adopted, there would be no guarantee that a source is maintaining 
compliance with the emission standards for the time durations of the 
manual stack sampling method used to demonstrate compliance during the 
comprehensive performance test. Such an approach could conceivably 
encourage a source to intentionally vary operating parameter levels 
during the comprehensive performance test to such an extent that the 
statistically-derived rolling average limits would be significantly 
higher than the true average of the test condition. This could also 
result in widely varying statistical correction factors from one source 
to another, which is undesirable for reasons of consistency and 
fairness.
    Such a statistical approach prevents us from establishing the 
minimum emission standards that Congress generally envisioned under 
MACT because we would not be assured that the sources are achieving the 
emission standard. We would also have difficulty estimating 
environmental benefits if this statistical approach were used because 
we would not know what level of emission control each source achieves. 
Again, the methodology promulgated for averaging performance test data 
to calculate operating parameter limits results in limits that are 
achievable and necessary to ensure compliance with the emission 
standards for time durations equivalent to emission sampling periods.
    Several commenters oppose the compliance regime whereby limits on 
operating parameters are established during performance testing. They 
are concerned that this approach encourages sources to operate under 
worst-case conditions during testing. One commenter states that this 
approach effectively punishes sources for demonstrating emissions 
during their performance test that are lower than the standards (i.e., 
by establishing limits on operating parameters that would be well below 
those needed to comply with the standards).
    We understand these concerns, but absent the availability of 
continuous emissions monitoring systems, we are unaware of another 
compliance assurance approach that effectively addresses the (perhaps 
unique) problem posed by hazardous waste combustors. The Agency is 
using this same approach to implement the RCRA regulations for these 
sources. Compliance assurance for hazardous waste combustors cannot be 
maintained using the general provisions of Subpart A in Part 63--
procedures that apply to all MACT sources unless we promulgate 
superseding provisions for a particular source category. Those 
procedures require performance testing under normal operating 
conditions, but operating limits are not established based on 
performance test operations. This approach is appropriate for most 
industrial processes because process constraints and product quality 
typically limit ``normal'' operations to a fairly narrow range that is 
easily defined.
    Hazardous waste combustors may be somewhat unique MACT sources, 
however, in that the characteristics of the hazardous waste feed (e.g., 
metals concentration, heating value) can vary over a wide range and 
have a substantial effect on emissions of hazardous air pollutants. In 
addition, system design, operating, and maintenance features can 
substantially affect pollutant emissions. This is not the same 
situation for many other MACT source categories where feedstream 
characteristics and system design, operation, and maintenance features 
must be confined to a finite range so that the source can continue to 
produce a product. Hazardous waste incinerators do not have such 
inherent controls (i.e., because they provide a waste treatment service 
rather than produce a product), and cement and lightweight aggregate 
kilns can vary substantially hazardous waste characteristics in the 
fuel, as well as system design, operation, and maintenance features and 
still produce marketable product.
    To address commenters' concerns at least in part, however, we have 
included a metals feedrate extrapolation provision in the final rule. 
This will reduce the incentive to spike metals in feedstreams during 
performance testing (and thus reduce the cost of testing, the hazard to 
test crews, and the environmental loading) by explicitly allowing 
sources to request approval to establish metal feedrate limits based on 
extrapolating upward from levels fed during performance testing. See 
discussion in Section VII.D.4 below, and Secs. 63.1209(l)(1) and 
63.1209(n)(2)(ii).
4. How Are the Various Types of Operating Parameters Monitored or 
Established?
    The operating parameters for which you must establish limits can be 
categorized according to how they are monitored or established as 
follows: (1) Operating parameters monitored directly with a continuous 
monitoring system; (2) feedrate limits; and (3) miscellaneous operating 
parameters. (Each of these parameters is discussed in Section VII.D 
below.)
    a. What Operating Parameters Are Monitored Directly with a 
Continuous Monitoring System? Operating parameters that are monitored 
directly with a continuous monitoring system include: Combustion gas 
temperature in the combustion chamber and at the inlet to a dry 
particulate matter control device; baghouse pressure drop; for wet 
scrubbers, pressure drop across a high energy wet scrubber (e.g., 
venturi, calvert), liquid feed pressure, pH, liquid-to-gas ratio, 
blowdown rate (coupled with either a minimum recharge rate or a minimum 
scrubber water tank volume or level), and scrubber water solids 
content; minimum power input to each field of an electrostatic 
precipitator; flue gas flowrate or kiln production rate; hazardous 
waste flowrate; and adsorber carrier stream flowrate. These operating 
parameters are monitored and recorded on a continuous basis during the 
comprehensive performance test and during normal operations. The 
continuous monitoring system also transforms and equates the data to 
its associated averaging period during the performance test so that 
operating parameter limits can be established. The continuous 
monitoring system must operate in conformance with Sec. 63.1209(b).
    b. How Are Feedrate Limits Monitored? Feedrate limits are monitored 
by knowing the concentration of the regulated parameter

[[Page 52924]]

in each feedstream and continuously monitoring the flowrate of each 
feedstream. See Sec. 63.1209(c)(4). You must establish limits on the 
feedrate parameters specified in Sec. 63.1209, including: semivolatile 
metals, low volatile metals, mercury; chlorine, ash (for incinerators), 
activated carbon, dioxin inhibitor, and dry scrubber sorbent. The 
flowrate continuous monitoring system must operate in conformance with 
Sec. 63.1209(b).
    c. How Are the Miscellaneous Operating Parameters Monitored/
Established? Other operating parameters specified in Sec. 63.1209 
include: Specifications for activated carbon, acid gas sorbent, 
catalyst for catalytic oxidizers, and dioxin inhibitor; and maximum age 
of carbon in a carbon bed. Because each of these operating parameters 
may be unique to your source, you are expected to characterize the 
parameter (e.g., using manufacturer specifications) and determine how 
it will be monitored and recorded. This information must be included in 
the comprehensive performance test plan that will be reviewed and 
approved by permitting officials.
5. How Are Rolling Averages Calculated Initially, Upon Intermittent 
Operations, and When the Hazardous Waste Feed Is Cut Off?
    a. How Are Rolling Averages Calculated Initially? You must begin 
complying with the limits on operating parameters specified in the 
Documentation of Compliance on the compliance date.201 See 
Sec. 63.1209(b)(5)(i). Given that the one-hour, and 12-hour rolling 
averages for limits on various parameters must be updated each minute, 
this raises the question of how rolling averages are to be calculated 
upon initial startup of the rolling average requirements. We have 
determined that an operating parameter limit will not become effective 
on the compliance date until you have recorded enough monitoring data 
to calculate the rolling average for the limit. For example, the hourly 
rolling average limit on the temperature at the inlet to an 
electrostatic precipitator does not become effective until you have 
recorded 60 one-minute average temperature values on the compliance 
date. Given that compliance with the standards begins nominally at 
12:01 am on the compliance date, the hourly rolling average temperature 
limit does not become effective as a practical matter until 1:01 am on 
the compliance date. Similarly, the 12-hour rolling average limit on 
the feedrate of mercury does not become effective until you have 
recorded 12 hours of one-minute average feedrate values after the 
compliance date. Thus, the 12-hour rolling average feedrate limits 
become effective as a practical matter at 12:01 pm on the compliance 
date.
---------------------------------------------------------------------------

    \201\ The operating parameters for which you must specify limits 
are provided in Sec. 63.1209. You must include these limits in the 
Documentation of Compliance, and you must record the Documentation 
of Compliance in the operating record.
---------------------------------------------------------------------------

    Although we did not specifically address this issue at proposal, 
commenters raised the question in the context of CEMS. Given that the 
same issue applies to all continuous monitoring systems, we adopt the 
same approach for all continuous monitoring systems, including CEMS. 
See discussion below in Section VII.C.5.b. We adopt the approach 
discussed here because a rolling average limit on an operating 
parameter does not exist until enough one-minute average values have 
been obtained to calculate the rolling average.
    b. How Are Rolling Averages Calculated upon Intermittent 
Operations? We have determined that you are to ignore periods of time 
when one-minute average values for a parameter are not recorded for any 
reason (e.g., source shutdown) when calculating rolling averages. See 
Sec. 63.1209(b)(5)(ii). For example, consider how the hourly rolling 
average for a parameter would be calculated if a source shuts down for 
yearly maintenance for a three week period. The first one-minute 
average value recorded for the parameter for the first minute of 
renewed operations is added to the last 59 one-minute averages before 
the source shutdown for maintenance to calculate the hourly rolling 
average.
    We adopt this approach for all continuous monitoring systems, 
including CEMS (see discussion below in Section VII.C.5.b) because it 
is simple and reasonable. If, alternatively, we were to allow the 
``clock to be restarted'' after an interruption in recording parameter 
values, a source may be tempted to ``clean the slate'' of high values 
by interrupting the recording of the parameter values (e.g., by taking 
the monitor off-line for a span or drift check). Not only would this 
mean that operating limits would not be effective again until an 
averaging period's worth of values were recorded, but it would be 
contrary to our policy of penalizing a source for operating parameter 
limit exceedances by not allowing hazardous waste burning to resume 
until the parameter is within the limit. Not being able to burn 
hazardous waste during the time that the parameter exceeds its limit is 
intended to be an immediate economic incentive to minimize the 
frequency, duration, and intensity of exceedances.
    c. How Are Rolling Averages Calculated when the Hazardous Waste 
Feed Is Cut Off? Even though the hazardous waste feed is cut off, you 
must continue to monitor operating parameters and calculate rolling 
averages for operating limits. See Sec. 63.1209(b)(5)(iii). This is 
because the emission standards and operating parameter limits continue 
to apply even though hazardous waste is not being burned. See, however, 
the discussion in Part Five, Sections I.C and I.D above for exceptions 
(i.e., when a hazardous waste combustor is not burning hazardous waste, 
the emission standards and operating requirements do not apply: (1) 
During startup, shutdown, and malfunctions; or (2) if you document 
compliance with other applicable CAA section 112 or 129 standards).
6. How Are Nondetect Performance Test Feedstream Data Handled?
    You must establish separate feedrate limits for semivolatile metal, 
low volatile metal, mercury, total chlorine, and/or ash for each 
feedstream for which the comprehensive performance test feedstream 
analysis determines that these parameters are not present at detectable 
levels. The feedrate limit must be defined as nondetect at the full 
detection limit achieved during the performance test. See 
Sec. 63.1207(n).
    You will not be deemed to be exceeding this feedrate limit when 
detectable levels of the constituent are measured, provided that: (1) 
Your total system constituent feedrate, considering the detectable 
levels in the feedstream (whether above or below the detection limit 
achieved during the performance test) that is limited to nondetect 
levels, is below your total system constituent feedrate limit; or (2) 
except for ash, your uncontrolled constituent emission rate for all 
feedstreams, calculated in accordance with the procedures outlined in 
the performance test waiver provisions (see Sec. 63.1207(m)) are below 
the applicable emission standards.
    We did not address in the April 1996 NPRM how you must handle 
nondetect compliance test feedstream results when determining feedrate 
limits, nor did commenters suggest an approach. After careful 
consideration, we conclude that the approach presented above is 
reasonable and appropriate.
    The LWAK industry has expressed concern about excessive costs with 
compliance activities that would be needed for the mercury standard. 
They

[[Page 52925]]

claim that the increased costs associated with achieving lower mercury 
detection limits are large, and does not result in significant 
environmental benefits.
    The final rule includes four different methods an LWAK can use to 
comply with the mercury emission standard in order to provide maximum 
flexibility. The basic compliance approach (described below) does not 
require an LWAK to achieve specified minimum mercury detection limits 
for mercury standard compliance purposes.202 Under this 
approach, analytical procedures that achieve given detection limits are 
evaluated on a site-specific basis as part of the waste analysis plan 
review and approval process, which is submitted as part of the 
performance test plan. An LWAK can make the case to the regulatory 
official that the increased costs associated with achieving a very low 
mercury detection limit is not warranted. We therefore do not believe 
that the LWAK industry will incur significant additional analytical 
costs over current practices for daily mercury compliance activities. 
We acknowledge, however, that site-specific circumstances may lead a 
regulatory official to conclude that lower detection limits are 
warranted. To better understand this concept, the following paragraphs 
summarize this basic mercury emission standard compliance scheme and 
discusses why a regulatory official may determine, on a site-specific 
basis, that lower detection limits are needed to better assure 
compliance with the emission standard.
---------------------------------------------------------------------------

    \202\ The other three approaches are (1) performance test waiver 
provisions (see preamble, part 5, section X.B); (2) alternative 
standards when raw materials cause an exceedance of the emission 
standard (see preamble, part 5, section X.A); and, (3) alternative 
mercury standards for kilns that have non-detect levels of mercury 
in the raw material (see preamble, part 5, section X.A). These 
mercury standard compliance alternatives require a source to achieve 
feedstream detection limits that either ensure compliance with an 
emission standard or ensure compliance with a hazardous waste 
feedrate limit that is used in lieu of a numerical emission 
standard. See previous referenced preamble for further discussion.
---------------------------------------------------------------------------

    Under this basic approach, the source conducts a performance test 
and samples the emissions for mercury to demonstrate compliance with 
the emission standard. To ensure compliance with the emission standard 
during day-to-day operations, the source must comply with mercury 
feedrate limits that are based on levels achieved during the 
performance test. A source must establish separate mercury feedrate 
limits for each feed location. As previously discussed in this section, 
for feedstreams where mercury is not present at detectable levels, the 
feedrate limit must be defined as ``nondetect at the full detection 
limit''.
    There is no regulatory requirement for a source to achieve a given 
detection limit under this approach. We acknowledge, however, that 
feedstream detection limits can be high enough such that a mercury 
feedrate limit that is based on nondetect performance test results may 
not completely ensure compliance with the emission standard during day-
to-day operations. For example, the LWAK industry has indicated that a 
hazardous waste mercury detection limit of 2 ppm is reasonably 
achievable at an on-site laboratory. If we assume that mercury is 
present in the hazardous waste at a concentration of 1.99 ppm (just 
below the detection limit), the expected mercury emission concentration 
would be approximately 80 g/dscm, which is above the 
standard.203 (Note also that this does not consider mercury 
emission contributions from the raw material.) This is not to say that 
this LWAK will be exceeding the mercury emission standard during day-
to-day operations. However, their inability to achieve low mercury 
detection limits results in less assurance that the source is 
continuously complying with the emission standard.
---------------------------------------------------------------------------

    \203\ This assumes that all the mercury fed to the unit is 
emitted, and is based on typical LWAK gas emission rates.
---------------------------------------------------------------------------

    The regulatory official should consider such emission standard 
compliance assurance concerns when reviewing the waste analysis plan to 
determine if lower detection limits are appropriate (if, in fact such 
lower detection limits are reasonably achievable). Factors that should 
be considered in this review should include: (1) The costs associated 
with achieving lower detection limits; and (2) the estimated maximum 
mercury concentrations that can occur if the source's feedstreams 
contain mercury just below the detection limit (as described above).
C. Which Continuous Emissions Monitoring Systems Are Required in the 
Rule?
    Although the final rule does not require you to use continuous 
emissions monitoring systems (CEMS) for parameters other than carbon 
monoxide, hydrocarbon, oxygen, and particulate matter 204 we 
have a strong preference for CEMS because they: (1) Are a direct 
measure of the hazardous air pollutant or surrogate for which we have 
established emission standards; (2) lead to a high degree of certainty 
regarding compliance assurance; and (3) allow the public to be better 
informed of what a source's emissions are at any time. Additionally, 
from a facility standpoint, CEMs provide you with real time feedback on 
your combustion operations and give you a greater degree of process 
control. Therefore, we encourage you to use CEMS for other parameters 
such as total mercury, multimetals, hydrochloric acid, and chlorine 
gas. You may use the alternative monitoring provision of Sec. 63.8(f) 
to petition the Administrator (i.e., permitting officials) to use CEMS 
to document compliance with the emission standards in lieu of emissions 
testing and the operating parameter limits specified in Sec. 63.1209. 
You may submit the petition at any time, such as with the comprehensive 
performance test plan. See Section VII.C.5.c below for a discussion of 
the incentives for using CEMS.
---------------------------------------------------------------------------

    \204\ The final rule requires that particulate matter CEMS be 
installed, but defers the effective date of the requirement to 
install, calibrate, maintain, and operate PM CEMS until these 
actions can be completed.
---------------------------------------------------------------------------

    In this section, we discuss the status of development of particular 
CEMS and provide guidance on issues that pertain to case-by-case 
approval of CEMS in lieu of compliance using operating parameter limits 
and periodic emissions testing. Key issues include appropriate CEMS 
performance specifications, reference methods for determining the 
performance of CEMS, averaging periods, and temporary waiver of 
emission standards if necessary to enable sources to correlate 
particulate matter CEMS to the reference method.
1. What Are the Requirements and Deferred Actions for Particulate 
Matter CEMS?
    In the April 1996 NPRM, we proposed the use of particulate matter 
CEMS to document compliance with the particulate matter emission 
standards. Particulate matter CEMS are used for compliance overseas 
205, but are not yet a regulatory compliance tool in the 
U.S. Concurrent with this proposal, we undertook a demonstration of 
particulate matter CEMS at a hazardous waste incinerator to determine 
if these CEMS were feasible in U.S. applications. We selected the test 
incinerator as representative of a worst-case application for a 
particulate matter CEMS at any hazardous waste

[[Page 52926]]

combustor. It was important to document feasibility of the CEMS at a 
worst-case application to minimize time and resources needed to 
determine whether the CEMS were suitable for compliance assurance at 
all hazardous waste combustors.
---------------------------------------------------------------------------

    \205\ The EU guidelines for hazardous waste combustion state 
that particulate matter is a parameter for which compliance must be 
documented continuously. In addition, proposals from vendors that we 
received in response to our February 27, 1996 NODA (see 61 FR 7262) 
indicate that there are many installations elsewhere overseas where 
particulate matter CEMS are used for compliance assurance.
---------------------------------------------------------------------------

    We published preliminary results of our CEMS testing and sought 
comment on our approach to demonstrating particulate matter CEMS in the 
March 1997 NODA. We then revised our approach and sought comment on the 
final report in the December 1997 NODA. The December 1997 NODA also 
clarified several issues that came to light during the demonstration 
test pertaining to the manual reference method, particulate matter 
CEMS, and general quality assurance issues. These clarifications were 
embodied in a new manual method, Method 5-I (Method 5i), a revision to 
the proposed Performance Specification 11 for particulate matter CEMS, 
and a new quality assurance procedure, Procedure 2.
    We believe that our tests adequately demonstrate that particulate 
matter CEMS are a feasible, accurate, and reliable technology that can 
and should be used for compliance assurance. In addition, preliminary 
analyses of the cost of PM CEMS applied to hazardous waste combustors 
suggest that these costs are reasonable. Accordingly, the final rule 
contains a requirement to install PM CEMS. However, we agree with 
comments that indicate a need to develop source-specific performance 
requirements for particulate matter CEMS and to resolve other 
outstanding technical issues. These issues include all questions 
related to implementation of the particulate matter CEMS requirement 
(i.e. relation to all other testing, monitoring, notification, and 
recordkeeping), relation of the particulate matter CEMS requirement to 
the PM emission standard, as well as technical issues involving 
performance, maintenance and correlation of the particulate matter CEMS 
itself. These issues will be addressed in a subsequent rulemaking. 
Therefore, we defer the effective date of this requirement pending 
further testing and additional rulemaking.
    As a result, in today's final rule, we require that particulate 
matter CEMS be installed at all hazardous waste burning incinerators, 
cement kilns, and lightweight aggregate kilns. However, since we have 
not finalized the performance specifications for the use of these 
instruments or resolved some of the technical issues noted above, we 
are deferring the effective date of the requirement to install, 
calibrate, maintain and operate particulate matter CEMS until these 
actions can be completed. The particulate matter CEMS installation 
deadline will be established through future rulemaking, along with 
other pertinent requirements, such as final Performance Specification 
11, Appendix F Procedure 2. Finally, it should be noted that EPA has a 
concurrent rulemaking process underway for nonhazardous waste burning 
cement kilns and plans to adopt the same approach in that rule.
2. What Are the Test Methods, Specifications, and Procedures for 
Particulate Matter CEMS?
    a. What Is Method 5i? We promulgate in the final rule a new manual 
method for measuring particulate matter, Method 5i. See appendix A to 
part 60. We first published this new method in the December 1997 NODA. 
One outgrowth of these particulate matter CEMS demonstration tests is 
that we made significant improvements in making low concentration 
Method 5 particulate measurements. We first discussed these 
improvements in the preliminary report released in the March 1997 NODA, 
and commenters to that NODA ask that these improvements be documented. 
We documented these improvements by creating Method 5i.
    We incorporated the following changes to Method 5 into Method 5i: 
Improved sample collection; minimization of possible contamination; 
Improved sample analysis; and an overall emphasis on elimination of 
systemic errors in measurement. These improvement achieved significant 
improvements in method accuracy and precision at low particulate matter 
concentrations, relative to Method 5.
    We are promulgating Method 5i today, in advance of any particulate 
matter CEMS requirement, for several reasons. We expect this new method 
will be preferred in all cases where low concentration (i.e., below 45 
mg/dscm (0.02 gr/dscf) 206) measurements are 
required for compliance with the standard. Given that all incinerators, 
nearly all lightweight aggregate kilns, and some cement kilns are 
likely to have emissions lower than 45 mg/dscm, we expect that Method 
5i will become the particulate method of choice for most hazardous 
waste combustors. In addition, we expect that Method 5i will be used to 
correlate manual method results to particulate matter CEMS outputs for 
those sources that elect to petition the Administrator to use a CEMS in 
lieu of operating parameter limits for compliance assurance with the 
particulate matter standard.207 This is because, unlike the 
worst-case particulate matter measurements normally used to verify 
compliance with the standard, low (or lower than normal) concentration 
particulate matter data are required to develop a good correlation 
between the CEMS output and the manual, reference method.
---------------------------------------------------------------------------

    \206\ As noted later in the text, the filter and assembly used 
for Method 5i is smaller than the one used for Method 5. This means 
that the Method 5i filter plugs more easily than the one used for 
Method 5. This issue becomes important at particulate matter 
concentrations above 45 mg/dscm, or 0.02 gr/dscf.
    \207\ As alluded to previously, sources may elect to use a CEMS 
to comply with the numerical value of the particulate matter 
emission standard on a six-hour rolling average in lieu of complying 
with operating parameter limits specified by Sec. 63.1209(m).
---------------------------------------------------------------------------

    Many of the issues commenters raise relate to how Method 5i should 
be used to correlate particulate matter CEMS outputs to manual method 
measurements. Even though we are deferring a CEMS requirement, we 
address several key issues here given that sources may elect to 
petition the Administrator under Sec. 63.8(f) to use a CEMS. This 
discussion may provide a better understanding on our thinking on 
particulate matter CEMS issues. In addition, certain comments are 
specific to how Method 5i is performed. These comments and our 
responses are relevant even if you use Method 5i only as a stack 
particulate method and not to correlate a particulate matter CEMS to 
the reference method.
    i. Why Didn't EPA Validate Method 5i Against Method 5? Several 
commenters recommend that we perform a full Method 301 validation to 
confirm that Method 5i is equivalent to Method 5. We determined that a 
full Method 301 validation is not necessary because the differences in 
the two methods do not constitute a major change in the way particulate 
samples are collected from an operational or an analytical standpoint. 
We validated the filter extraction and weighting process--the only 
modification from Method 5 (see ``Particulate Matter CEMS Demonstration 
Test Final Report,'' Appendix A, in the Technical Support Document 
208) `` and documented that Method 5i gives nearly identical 
results as Method 5. Therefore, we disagree with the commenters' 
underlying concern and conclude that Method 5i has been validated.
---------------------------------------------------------------------------

    \208\ See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume IV: Compliance With 
the Hazardous Waste Combustor Standards,'' July 1999.
---------------------------------------------------------------------------

    ii. When Are Paired Trains Required? We have included in Method 5i 
a requirement that paired trains must be

[[Page 52927]]

used to increase method precision. This requirement applies whether you 
use Method 5i to demonstration compliance with the emission standard or 
to correlate a particulate matter CEMS. In addition, if you elect to 
petition the Administrator for approval to use a particulate matter 
CEMS and elect to use Method 5 to correlate the CEMS, you must also 
obtain paired Method 5 data to improve method precision and, thus, the 
correlation.
    During our CEMS testing, we collected particulate matter data using 
two simultaneously-conducted manual method sampling trains. We called 
the results from these simultaneous runs ``paired data.'' We discussed 
the use of paired trains in the December 1997 NODA as being optional 
but requested comment on whether we should require paired trains, state 
a strong preference for them, or be silent on the issue. Many 
commenters believe paired trains should be used at all times so 
precision can be documented. With these comments in mind, and 
consistent with our continued focus on the collection of high quality 
emission measurements, we include a requirement in Method 5i to obtain 
paired data. Method 5i also includes a minimum acceptable relative 
standard deviation between these data pairs. As discussed below, both 
data in the pair are rejected if the data exceed the acceptable 
relative standard deviation.
    To improve the correlation between the manual method and a 
particulate matter CEMS, we also recommend that sources electing to use 
Method 5 also obtain paired Method 5 data. Again, data sets that exceed 
an acceptable relative standard deviation, as discussed below, should 
be rejected. This recommendation will be implemented during the 
Administrator's review of your petition requesting use a particulate 
matter CEMS. If you elect to correlate the CEMS using Method 5, you are 
expected to include in your petition a statement that you will obtain 
paired data and will conform with our recommended relative standard 
deviation for the paired data.
    iii. What Are the Procedures for Identifying Outliers? We have 
established maximum relative standard deviation values for paired data 
for both Method 5i and Method 5. If a data pair exceed the relative 
standard deviation, the pair is identified as an outlier and is not 
considered in the correlation of a particulate matter CEMS with the 
reference method. In addition, Method 5i pairs that exceed the relative 
standard deviation are considered outliers and cannot be used to 
document compliance with the emission standard.
    In the initial phase of our CEMS tests, we established a procedure 
for eliminating imprecise data. This consisted of eliminating a set of 
paired data if the data disagree by more than some previously 
established amount. Two identical methods running at the same time 
should yield the same result; if they do not, the precision of both 
data is suspect. Commenters agree with the need to identify and 
eliminate imprecise data to enhance method precision. This is an 
especially important step when comparing manual particulate matter 
measurements to particulate matter CEMS measurements. As a result, we 
include criteria in Method 5i to ensure data precision.
    When evaluating the particulate matter CEMS Demonstration Test 
data, we screened the data to remove these precision outliers. Data 
outliers at that time were defined as paired data points with a 
relative standard deviation 209 of greater than 30 percent. 
We developed this 30% criterion by analyzing historical Method 5 data. 
Several commenters, including a particulate matter CEMS vendor with 
extensive European experience with correlation programs, recommend that 
we tighten the relative standard deviation criteria. We concur, because 
Method 5i is more precise than Method 5 given the improvements 
discussed above. Therefore, one would logically expect a reasonable 
precision criterion such as the relative standard deviation derived 
from Method 5i data to be less than a similarly reasonable one derived 
from Method 5 data. We investigated the particulate matter CEMS 
Demonstration Test data base as well other available Method 5i data 
(such as the data from a test program recently conducted at another US 
incinerator). We conclude that a 10% relative standard deviation for 
particulate matter emissions greater than or equal to 10 mg/dscm, 
increased linearly to 25% for concentrations down to 1 mg/dscm, is a 
better representation of acceptable, precise Method 5i paired data 
210. Data obtained at concentrations lower than 1 mg/dscm 
have no relative standard deviation limit.
---------------------------------------------------------------------------

    \209\ RSD, or ``relative standard deviation'', is a 
dimensionless number greater than zero defined as the standard 
deviation of the samples, divided by the mean of the samples. In the 
special case where only 2 data represent the sample, the mathematics 
of determining the relative standard deviation simplifies greatly to 
|CA-CB |/(CA + CB), 
where CA and CB are the concentration results 
from the two trains that represent the pair.
    \210\ See Chapter 11, Section 2 of the technical background 
document for details on the statistical procedures used to derive 
these benchmarks: USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume IV: Compliance With 
the Hazardous Waste Combustor Standards,'' July 1999.
---------------------------------------------------------------------------

    The relative standard deviation criterion for Method 5 data used 
for particulate matter CEMS correlations continues to be 30%.
    iv. Why Didn't EPA Issue Method 5i as Guidance Rather than 
Promulgating It as a Method? Most commenters state that Method 5i 
should be guidance rather than a published method and it should not be 
a requirement for performing particulate matter CEMS correlation 
testing or documenting compliance with the emission standard. In 
particular, several commenters in the cement kiln industry express 
concern over the limitations of Method 5i regarding the mass of 
particulate it could collect. This section addresses these concerns.
    We have promulgated Method 5i as a method because it provides 
significant improvement in precision and accuracy of low level 
particulate matter measurements relative to Method 5. Consequently, 
although Method 5i is not a required method, we expect that permitting 
officials will disapprove comprehensive performance test plans that 
recommend using Method 5 for low level particulate levels. Further, we 
expect that petitions to use a particulate matter CEMS that recommend 
performance acceptance criteria (e.g., confidence level, tolerance 
level, correlation coefficient) based on correlating the CEMS with 
Method 5 measurements will be disapproved. This is because we expect 
the CEMS to be able to achieve better acceptance criteria values using 
Method 5i (because it is more accurate and precise than Method 5), and 
expect better relative standard deviation between test pairs (resulting 
in lower cost of correlation testing because fewer data would be 
screened out as outliers).
    Given that we expect and want widespread use of Method 5i, and to 
ensure that its key provisions are followed, it is appropriate to 
promulgate it as a method rather than guidance. If the procedure were 
issued only as guidance, the source or stack tester could choose to 
omit key provisions, thus negating the benefits of the method.
    Relative to the direct reference in Method 5i that the method is 
``most effective for total particulate matter catches of 50 mg or 
less,'' this means the method is most effective at hazardous waste 
combustors with particulate matter emissions below approximately 45 mg/
dscm (0.02 gr/dscf). This applicability statement is not 
intended to be a bright line; total train catches exceeding 50 mg would 
not invalidate

[[Page 52928]]

the method. Rather, we include this guidance to users of the method to 
help them determine whether the method is applicable for their source. 
Note that this statement is found in the applicability section of the 
method, rather than the method description sections that follow. As 
such, the reference is clearly an advisory statement, not a quality 
assurance criterion. Total train catches above 50 mg are acceptable 
with the method and the results from such trains can be used to 
document compliance with the emission standard and for correlating 
CEMS. But, users of Method 5i are advised that problems (such as 
plugging of the filter) may arise when emissions are expected to exceed 
45 mg/dscm. 211
---------------------------------------------------------------------------

    \211\ Stack testers have developed ways to deal with plugging of 
a filter. Many stack testers simply remove the filter before it 
plugs, install a new, clean filter, and continue the sampling 
process where they left off with the old filter. The mass gain is 
then the total mass accumulated on all filters during the run. 
However, using multiple filters for a single run takes more time, 
not only to install the new filter but also to condition and weigh 
multiple filters for a single run. For Method 5i, it would also 
involve more capital cost because the stack tester would need more 
light-weight filter assemblies to perform the same number of runs. 
For these reasons and even though the situation can be acceptably 
managed, it is impractical to have the filter plug. This led to our 
recommendation that Method 5i is best suited for particulate matter 
(i.e., filter) loadings of at most 50 mg, or stack concentrations of 
less than 45 mg/dscm (roughly 0.02 gr/dscf).
---------------------------------------------------------------------------

    v. What Additional Costs Are Associated with Method 5i? Commenters 
raise several issues regarding the additional costs of performing 
Method 5i testing relative to using Method 5. There is an added cost 
for the purchase of new Method 5i filter housings. These new 
lightweight holders are the key addition to the procedure needed to 
improve precision and accuracy and represent a one-time expense that 
emission testing firms or sources that perform testing in-house will 
have to incur to perform Method 5i. We do not view this cost as 
significant and conclude that the use of a light-weight filter housing 
is a reasonable and appropriate feature of the method.
    Other commenters suggest that the requirement for pesticide-grade 
acetone in the version of Method 5i contained in the December 1997 NODA 
unnecessarily raises the cost of performing the method. Instead, they 
ask us to identify a performance level for the acetone instead of a 
grade requirement because it would allow test crews to meet that 
performance in the most economical manner. We agree that prescribing a 
certain type of acetone may unnecessarily increase costs and removed 
the requirement for pesticide-grade acetone. Accordingly, the same 
purity requirements cited in Method 5 for acetone are maintained for 
Method 5i. The prescreening of acetone purity in the laboratory prior 
to field use, consistent with present Method 5 requirements, is also 
maintained in Method 5i.
    Commenters make similar cost-related comments relative to the 
requirement for Teflon beakers. At the request of several 
commenters, we have expanded the requirement for Teflon 
beakers to allow the use of beakers made from other similar light-
weight materials. Because materials other than Teflon can 
be used to fabricate light-weight breakers, changing the requirement 
from a technology basis to a performance basis will reduce costs while 
achieving the performance goals of the method.
    There were no significant comments regarding the added cost of 
paired-train testing.
    vi. What Is the Practical Quantification Limit of the Method 5i 
Filter Sample? We received several comments related to the minimum 
detection limit of Method 5i, including: the minimum sample required, 
guidance on how long to sample, what mass should ideally be collected 
on any filter, and the practical quantification limit.
    Commenters are concerned that while we address the maximum amount 
of particulate matter the method could handle, we are silent on the 
issue of what minimum sample is required. This is important because 
analytical errors, such as weighing of the filters, tend to have the 
same error value associated with it irrespective of the mass loading. 
To address this concern, Method 5i provides guidance on determining the 
minimum mass of the collected sample based on estimated particulate 
matter concentrations.
    Related to the particulate mass collection issue is the issue of 
how long a user of Method 5i needs to sample in order to an adequate 
amount of particulate on the filter. The amount of particulate matter 
collected is directly related to time duration of the sampling period, 
i.e., the longer one samples, the more particulate is collected and 
vice-versa. Therefore, Method 5i provides guidance on selecting a 
suitable sampling time based on the estimated concentration of the gas 
stream.
    Both these issues directly relate to how much particulate matter 
should ideally be collected on any individual filter. Our experience 
indicates a minimum target mass is 10 to 20 mg.
    Finally, we conclude that the targeted practical quantification 
limit for Method 5i is 3.0 mg of sample. Discussion of how this 
quantification limit is determined is highly technical and beyond the 
scope of this preamble. See the technical support document for more 
details.212
---------------------------------------------------------------------------

    \212\ See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume IV: Compliance With 
the Hazardous Waste Combustor Standards,'' July 1999.
---------------------------------------------------------------------------

    vii. How Are Blanks Used with Method 5i? Several commenters 
question the use of acetone blanks or made recommendations for 
additional blanks. We clarify in this section the collection and use of 
sample blank data.
    We recognize that high blank results can adversely effect the 
analytical results, especially at low particulate matter 
concentrations. To avoid the effect high blank results can have on the 
analytical results, today's Method 5i adopts a strategy similar to 
several of the organic compound test procedures (such as Method 23 in 
part 60 and Method 0010 in SW-846) that require collection of blanks 
but do not permit correction to the analytical results. Collection and 
analysis of blanks remains an important component in the sampling and 
analysis process for documenting the quality of the data, however. If a 
test run has high blank results, the data may be suspect. Permitting 
officials will address this issue on a case-by-case basis.
    The importance of minimizing contamination is stressed throughout 
Method 5i for both sample handling and use of high purity sample media. 
If proper handling procedures are observed, we expect that the blank 
values will be less than the method detection limit or within the value 
for constant weight determination (0.5 mg). Therefore, the allowance 
for blank correction that is provided in Method 5 is not permitted in 
Method 5i. The method also recommends several additional types of 
blanks to provide further documentation of the integrity and purity of 
the acetone throughout the duration of the field sampling program.
    b. What Is the Status of Particulate Matter CEMS Performance 
Specification 11 and Quality Assurance/Quality Control Procedure 2? We 
are not finalizing proposed Performance Specification 11 and Quality 
Assurance/Quality Control Procedure 2 because the final rule does not 
require the use of particulate matter CEMS. We considered stakeholder 
comments on these documents, however, and have incorporated many 
comments into the current drafts. We plan to publish these documents 
when we address the particulate matter CEMS requirement. In the 
interim, we will make them available as guidance to sources that are

[[Page 52929]]

considering the option of using a particulate matter CEMS to document 
compliance.
    c. How Have We Resolved Other Particulate Matter CEMS Issues? In 
this section we discuss two additional issues: (1) Why didn't we 
require continuous opacity monitors for compliance with the particulate 
matter standard for incinerators and lightweight aggregate kilns; and 
(2) can high correlation emissions testing runs exceed the particulate 
matter standard?
    i. Why Didn't We Require Continuous Opacity Monitors for Compliance 
Assurance for Incinerators and Lightweight Aggregate Kilns? As 
discussed elsewhere in today's notice, we require cement kilns to use 
continuous opacity monitors (COMS) to comply with a 20 percent opacity 
standard to ensure compliance with the particulate matter emission 
standard. This is the opacity component of the New Source Performance 
Standard for particulate matter for Portland cement plants. See 
Sec. 60.62. Because we are adopting the mass-based portion of the New 
Source Performance Standard for particulate matter as the MACT standard 
(i.e., 0.15 kg/Mg dry feed), the opacity component of the New Source 
Performance Standard is useful for compliance assurance.
    We do not require that incinerators and lightweight aggregate kilns 
use opacity monitors for compliance assurance because we are not able 
to identify an opacity level that is achievable by sources using MACT 
control and that would ensure compliance with the particulate matter 
standards for these source categories. This is the same issue discussed 
above in the context of particulate matter CEMS and is the primary 
reason that we are not requiring use of these CEMS at this time.
    Although we are requiring that cement kilns use COMS for compliance 
assurance, these monitors cannot provide the same level of compliance 
assurance as particulate matter CEMS. Opacity monitors measure a 
characteristic of particulate matter (i.e., opacity) and cannot 
correlate with the manual stack method as well as a particulate matter 
CEMS. COMS are particularly problematic for sources with small stack 
diameters (e.g., incinerators) and low emissions because both of these 
factors contribute to very low opacity readings which results in high 
measurement error as a percentage of the opacity value. Thus, we are 
obtaining additional data to support rulemaking in the near future to 
require use of particulate matter CEMS for compliance assurance.
    Approximately 80 percent of hazardous waste burning cement kilns 
are not currently subject to the New Source Performance Standard and 
many of these sources may not be equipped with COMS that meet 
Performance Specification 1 in appendix B, part 60. Thus, many 
hazardous waste burning cement kilns will be required to install COMS, 
even though we intend to require use of particulate matter CEMS in the 
near future. We do not believe that this requirement will be overly 
burdensome, however, because sources may request approval to install 
particulate matter CEMS rather than COMS. See Sec. 63.8(f). Our testing 
of particulate matter CEMS at a cement kiln will be completed well 
before sources need to make decisions on how best to comply with the 
COMS requirement of the rule. We will develop regulations and guidance 
on performance specifications and correlation criteria for particulate 
matter CEMS as a result of that testing, and sources can use that 
guidance to request approval to use a particulate matter CEMS in lieu 
of a COMS. We expect that most sources will elect to use this approach 
to minimize compliance costs over the long term.
    ii. Can High Correlation Runs Exceed the Particulate Matter 
Standard? The final rule states that the particulate matter and opacity 
standards of parts 60, 61, 63, 264, 265, and 266 (i.e., all applicable 
parts of Title 40) do not apply during particulate matter CEMS 
correlation testing, provided that you comply with certain provisions 
discussed below that ensure that the provision is not abused. This 
provision, as the rest of the rule, is effective immediately. Thus, you 
need not wait for the compliance date to take advantage of this 
particulate matter CEMS correlation test provision.
    We include this provision in the rule because many commenters 
question whether high correlation test runs that exceed the particulate 
matter emission standard constitute noncompliance with the standard. We 
have responded to this concern previously by stating that a single 
manual method test run that exceeds the standard does not constitute 
noncompliance with the standard because compliance is based on the 
average of a minimum of three runs.213 We now acknowledge, 
however, that during high run correlation testing a source may need to 
exceed the emission standard even after averaging emissions across 
runs. Similarly, a source may need to exceed a particulate matter 
operating parameter limit. Given the benefits of compliance assurance 
using a CEMS, we agree with commenters that short-term excursions of 
the particulate matter standard or operating parameter limits for the 
purpose of CEMS correlation testing is warranted. The benefits that a 
CEMS provides for compliance assurance outweighs the short-term 
emissions exceedances that may occur during high end emissions 
correlation testing. Consequently, we have included a conditional 
waiver of the applicability of all Federal particulate matter and 
opacity standards (and associated operating parameter limits).
---------------------------------------------------------------------------

    \213\ One exception is the destruction and removal efficiency 
standard, for which compliance is based on a single test run and not 
the average of three runs.
---------------------------------------------------------------------------

    The waiver of applicability of the particulate matter and opacity 
emission standards and associated operating parameter limits is 
conditioned on the following requirements to ensure that the waiver is 
not abused. Based on information from commenters and expertise gained 
during our testing, the rule requires that you develop and submit to 
permitting officials a particulate matter CEMS correlation test plan 
along with a statement of when and how any excess emissions will occur 
during the correlation tests (i.e., how you will modify operating 
conditions to ensure a wide range of particulate emissions, and thus a 
valid correlation test). If the permitting officials fail to respond to 
the test plan in 30 days, you can proceed with the tests as described 
in the test plan. If the permitting officials comment on the plan, you 
must address those comments and resubmit the plan for approval.
    In addition, runs that exceed any particulate matter or opacity 
emission standard or operating parameter limit are limited to no more 
than a total of 96 hours per correlation test (i.e., including all runs 
of all test conditions). We determined that the 96 hour total duration 
for exceedances for a correlation test is reasonable because it is 
comprised of one day to increase emissions to the desired level and 
reach system equilibrium, two days of testing 214 at the 
equilibrium condition followed by a return to normal equipment settings 
indicative of compliance with emissions standards and operating 
parameter limits, and one

[[Page 52930]]

day to reach equilibrium at normal conditions. Finally, to ensure these 
periods of high emissions are due to the bona fide need described here, 
a manual method test crew must be on-site and making measurements (or 
in the event some unforseen problem develops, prepared to make 
measurements) at least 24 hours after you make equipment or workplace 
modifications to increase particulate matter emissions to levels of the 
high correlation runs.
---------------------------------------------------------------------------

    \214\ The two days assumes sources will conduct a total of 18 
runs, 6 runs in each of the low, medium, and high particulate matter 
emission ranges. To approve use of a particulate matter CEMS, we 
will likely require that a minimum of 15 runs comprise a correlation 
test. If this is the case, some runs will likely be eliminated 
because they fail method or source-specific quality assurance/
quality control procedures.
---------------------------------------------------------------------------

3. What Is the Status of Total Mercury CEMS?
    We are not requiring use of total mercury CEMS in this rulemaking 
because data in hand do not adequately demonstrate nationally that 
these CEMS are reliable compliance assurance tools at all types of 
facilities. Nonetheless, we are committed to the development of CEMS 
that measure total mercury emissions and are continuing to pursue the 
development of these CEMS in our research efforts.
    In the April 1996 NPRM, we proposed that total mercury CEMS be used 
for compliance with the mercury standards. We also said if you elect to 
use a multimetals CEMS that passed proposed acceptability criteria, you 
could use that CEMS instead of a total mercury CEMS to document 
compliance with the mercury standard. Finally, we indicated that if 
neither mercury nor multimetal CEMS were required in the final rule 
(i.e., because they have not been adequately demonstrated), compliance 
assurance would be based on specified operating parameter limits.
    In the March 1997 NODA, we elicited comment on early aspects of our 
approach to demonstrate total mercury CEMS. And, in the December 1997 
NODA, we presented a summary of the demonstration test results and our 
preliminary conclusion that we were unable to adequately demonstrate 
total mercury CEMS at a cement kiln, a site judged to be a reasonable 
worst-case for performance of the total mercury CEMS. As new data are 
not available, we continue to adhere to this conclusion, and comments 
received in response to the December 1997 NODA concur with this 
conclusion. Therefore, we are not requiring total mercury CEMS in this 
rulemaking.
    Nonetheless, the current lack of data to demonstrate total mercury 
CEMS at a cement kiln or otherwise on a generic bases (i.e., for all 
sources within a category) does not mean that the technology, as 
currently developed, cannot be shown to work at particular sources. 
Consequently, the final rule provides you the option of using total 
mercury CEMS in lieu of complying with the operating parameter limits 
of Sec. 63.1209(l). As for particulate matter and other CEMS, the rule 
allows you to petition the Administrator (i.e., permitting officials) 
under Sec. 63.8(f) to use a total mercury CEMS based on documentation 
that it can meet acceptable performance specifications, correlation 
acceptance criteria (i.e., correlation coefficient, tolerance level, 
and confidence level). Although we are not promulgating the proposed 
performance specification for total mercury CEMS (Performance 
Specification 12) given that we were not able to document that a 
mercury CEMS can meet the specification in a (worst-case) cement kiln 
application, the proposed specification may be useful to you as a point 
of departure for a performance specification that you may recommend is 
achievable and reasonable.
4. What Is the Status of the Proposed Performance Specifications for 
Multimetal, Hydrochloric Acid, and Chlorine Gas CEMS?
    We are not promulgating proposed Performance Specifications 10, 13, 
and 14 for multimetal, hydrochloric acid, and chlorine gas CEMS because 
we have not determined that the CEMS can achieve the specifications.
    In the April 1996 NPRM, we proposed performance specifications for 
multimetal, hydrochloric acid, and chlorine gas CEMS to allow sources 
to use these CEMS for compliance with the metals and hydrochloric acid/
chlorine gas standards. Given that we have not demonstrated that these 
CEMS can meet their performance specifications and our experience with 
a mercury CEMS where we were not able to demonstrate that the mercury 
CEMS could meet our proposed performance specification, we are not 
certain that these CEMS can meet the proposed performance 
specifications. Accordingly, it would be inappropriate to promulgate 
them.
    As discussed previously, we encourage sources to investigate the 
use of CEMS and to petition permitting officials under Sec. 63.8(f) to 
obtain approval to use them. The proposed performance specifications 
may be useful to you as a point of departure in your efforts to 
document performance specifications that are achievable and that ensure 
reasonable correlation with reference manual methods.
5. How Have We Addressed Other Issues: Continuous Samplers as CEMS, 
Averaging Periods for CEMS, and Incentives for Using CEMS?
    a. Are Continuous Samplers a CEMS? Several commenters, mostly 
owner/operators of on-site incinerators, suggest that we should adjust 
certain CEMS criteria (e.g., averaging period, response time) to allow 
use of a continuous sampler known as the 3M Method. The 3M Method is a 
continuous metals sampling system. It automatically extracts stack gas 
and accumulates a sample on a filter medium over any desired period--24 
hours, days, or weeks. The sample is manually extracted, analyzed, and 
reported. Various incinerator operators are using or have expressed an 
interest in using this type of approach to demonstrate compliance with 
current RCRA metals emission limits. Many commenters contend that the 
3M Method is a CEMS and that we developed our performance 
specifications for CEMS to exclude techniques like the 3M Method.
    After careful analysis, we conclude that the 3M Method is not a 
CEMS. It does not meet our long-standing definition of a CEMS in parts 
60 or 63. Specifically, it is not a fully automated piece(s) of 
equipment used to extract a sample, condition and analyze the sample, 
and report the results of the analysis in the units of the standard. 
Also, the 3M Method is unable to ``complete a minimum of one cycle of 
operation (sampling, analyzing, and data recording) for each successive 
15-minute period'' as required by Sec. 63.8(c)(4)(ii). As a result, 
making the subtle changes (e.g., to the averaging period, response 
time) to our multimetal CEMS performance specification that commenters 
recommend would not alter the fact that the device does not 
automatically analyze the sample on the frequency required for a CEMS.
    A continuous sampler (coupled with periodic analysis of the sample) 
is inferior to a CEMS for two reasons. First, if the sampling period is 
longer than the time it takes to perform three manual performance 
tests, compliance with the standard cannot be assured. Approaches like 
the 3M Method tend to have reporting periods on the order of days, 
weeks, or even a month. The reporting period is comprised of the time 
required to accumulate the sample and the additional time to analyze 
the sample and report results. Because the stringency of a standard is 
a function of both the numerical value of the standard and the 
averaging period (e.g., at a given numerical limit, the longer the 
averaging period the less stringent the standard), a compliance 
approach having a sampling period greater than the 12 hours we estimate 
it may take to conduct three manual method stack test runs using Method 
29 cannot ensure

[[Page 52931]]

compliance with the standard.215 If the sampling period were 
greater than the time required to conduct three test runs, the 
numerical value of the standard would have to be reduced to ensure an 
equally stringent standard. Unfortunately, we do not know how to derive 
alternative emission limits as a function of the averaging period that 
would be equivalent to the emission standard. We raised this issue at 
proposal, and commenters did not offer a solution.
---------------------------------------------------------------------------

    \215\ A technical support document for the February 1991 
municipal waste combustor rule contains a good description of how 
not only the numerical limit, but the averaging period as well, 
determines the overall stringency of the standard. See Appendices A 
and B found in ``Municipal Waste Combustion: Background Information 
for Promulgated Standards and Guidelines--Summary of Public Comments 
and Responses Appendices A to C'', EPA-450/3-91-004, December 1990.
---------------------------------------------------------------------------

    Second, the results from a continuous sampler are reported after 
the fact, resulting in higher excess emissions than with a CEMS. 
Depending on the sample analysis frequency, it could take days or weeks 
to determine that an exceedance has occurred and that corrective 
measures need to be taken. A CEMS can provide near real-time 
information on emissions such that exceedances can be avoided or 
minimized.
    Absent the generic availability of multimetal CEMS, continuous 
samplers such as the 3M Method may nonetheless be a valuable compliance 
tool. We have acknowledged that relying on operating parameter limits 
may be an imperfect approach for compliance assurance. Sampling and 
analysis of feedstreams to determine metals feedrates can be 
problematic given the complexities of some waste matrices. In addition, 
the operating parameters for the particulate matter control device for 
which limits must be established may not always correlate well with the 
device's control efficiency for metals and thus metals emissions. 
Because of these concerns, we encourage sources to investigate the 
feasibility of multimetal CEMS. But, absent a CEMS, a continuous 
sampler may provide an attractive alternative or complement to some of 
the operating parameter limits under Secs. 63.1209 (l) and (n). You may 
petition permitting officials under Sec. 63.8(f) to use the 3M Method 
(or other sampler) as an alternative method of compliance with the 
emissions standards. Permitting officials will balance the benefits of 
a continuous sampler with the benefits of the operating parameter 
limits on a case-by-case basis.
    b. What Are the Averaging Periods for CEMS and How Are They 
Implemented? We discuss the following issues in this section: (1) 
Duration of the averaging period; (2) frequency of updating the 
averaging period; and (3) how averaging periods are calculated 
initially and under intermittent operations.
    i. What Is the Duration of the Averaging Period? We conclude that a 
six-hour averaging period is most appropriate for particulate matter 
CEMS, and a 12-hour averaging period is most appropriate for total 
mercury, multi metals, hydrogen chloride, and chlorine gas CEMS.
    We proposed that the averaging period for CEMS (i.e., other than 
carbon monoxide, hydrocarbon, and oxygen) be equivalent to the time 
required to conduct three runs of the comprehensive performance test 
using manual stack methods. As discussed above and at proposal, we 
proposed this approach because, to ensure compliance with the standard, 
the CEMS averaging period must be the same as the time required to 
conduct the performance test.216
---------------------------------------------------------------------------

    \216\ Actually, the CEMS averaging period can be no longer than 
the time required to conduct three runs of the performance test to 
ensure compliance with the standard. Although compliance with the 
standard would be ensured if the CEMS averaging period were less 
than the time required to conduct the performance test, this 
approach would be overly stringent because it would ensure 
compliance with an emission level lower than the standard.
---------------------------------------------------------------------------

    Commenters suggest two general approaches to establish averaging 
periods for CEMS: technology-based and risk-based. Commenters 
supporting a technology-based approach favor our proposed approach and 
rationale where the time duration of three emissions tests would be the 
averaging period for CEMS. Commenters favoring a risk-based approach 
state that the averaging period should be years rather than hours 
because the risk posed by emissions at levels of the standard were not 
found to be substantial, assuming years of exposure. We disagree with 
this rationale. CEMS are an option (that sources may request under 
Sec. 63.8(f)) to document compliance with the emission standard. As 
discussed above, if the averaging period for CEMS were longer than the 
duration of the comprehensive performance test, we could not ensure 
that a source maintains compliance with the standards.
    Establishing an averaging period based on the time to conduct three 
manual method stack test runs is somewhat subjective. There is no fixed 
sampling time for manual methods--sampling periods vary depending on 
the amount of time required to ``catch'' enough sample. Thus, we have 
some discretion in selecting an averaging period using this approach. 
Commenters generally favor longer averaging periods as an incentive for 
using CEMS (i.e., because a limit is less stringent if compliance is 
based on a long versus short averaging period). We agree that choosing 
a longer averaging period would provide an incentive for the use of 
CEMS, but conclude that the selected averaging period must be within 
the range (i.e., high end) of times required to perform the three stack 
test runs.
    We derive the averaging period for particulate matter CEMS as 
follows. Most particulate matter manual method tests are one hour in 
duration, but a few stack sampling companies sample for longer periods, 
up to two hours. Therefore, we use the high end of the range of values, 
2 hours, as the basis for calculating the averaging period. We 
recommend a six-hour rolling average considering that it may require 2 
hours to conduct each of three stack tests.
    For mercury, multi-metals, hydrochloric acid, and chlorine gas 
CEMS, we recommend a 12-hour rolling averaging. The data base we used 
to determine the standards shows that the sampling periods for manual 
method tests for these standards ranged from one to four hours. 
Choosing the high end of the range of values, 4 hours, as the basis for 
calculating the averaging period, we conclude that a 12-hour rolling 
average would be appropriate.
    ii. How Frequently Is the Rolling Average Updated? We conclude that 
the rolling average for particulate matter, total mercury, and 
multimetal CEMS should be updated hourly, while the rolling average for 
hydrochloric acid and chlorine gas CEMS should be updated each minute.
    We proposed that all rolling averages would be updated every minute 
and would be based on the average of the one-minute block average CEMS 
observations that occurred over the averaging period. This proposed 
one-minute update is the same that is used for carbon monoxide and 
total hydrocarbon CEMS under the RCRA BIF regulations. (We are 
retaining that update frequency in the final rule for those monitors, 
and recommend it for hydrochloric acid and chlorine gas CEMS.)
    Commenters favor selecting the frequency of updating the rolling 
average taking into account the variability of the CEMS and limitations 
concerning how the correlation data are collected. We agree with this 
approach, as discussed below.
    1. Particulate Matter CEMS. Commenters said that particulate matter 
CEMS correlation tests are approximately one hour in duration and, if 
the rolling average were updated

[[Page 52932]]

each minute, the CEMS would observe more variability in emissions 
within this one hour than the manual method (which is an average of 
those emissions during the hour). For this reason, we conclude it is 
reasonable that particulate matter CEMS data be recorded as a block-
hour and that the rolling average be updated every hour as the average 
of the previous six block-hours. Updating the particulate matter CEMS 
every hour also means the number of compliance opportunities is the 
same irrespective of whether a light-scattering or beta-gage 
particulate matter CEMS is used (i.e., because beta-gage CEMS make 
observations periodically while light-scattering CEMS make observations 
continuously).
    Furthermore, to ensure consistency with existing air rules 
governing CEMS other than opacity, a valid hour should be comprised of 
four or more equally spaced measurements during the hour. See 
Sec. 60.13(h). This means that batch systems, such as beta gages, must 
complete one cycle of operation every 15 minutes, or more frequently if 
possible. See Sec. 63.8(c)(4)(ii). CEMS that produce a continuous 
stream of data, such as light-scattering CEMS, will produce data 
throughout the hour.
    You may not be able to have four valid 15-minute measurement in an 
hour, however, to calculate an hourly block-average. Examples include 
when the source shuts down or the CEMS produces flagged (i.e., 
problematic) data. In addressing this issue, we balanced the need for 
the average of the measurements taken during the hour to be 
representative of emissions during the hour with the need to 
accommodate problems with data availability that will develop. We 
conclude that a particulate matter CEMS needs to sample stack gas and 
produce a valid result from this sample for most of the hour. This 
means that the CEMS needs to be observing stack gas at least half (30 
minutes, or two 15-minute cycles of operation) of the block-hour. 
Emissions from less than one hour might be unrepresentative of 
emissions during the hour, and on balance we conclude that this 
approach is reasonable. If a particulate matter CEMS does not sample 
stack gas and produce a valid result from that sample for at least 30 
minutes of a given hour, the hour is not a valid block-hour. In 
documenting compliance with the data availability recommendation in the 
draft performance specification, invalid block-hours due to 
unavailability of the CEMS that occur when the source is in operation 
count against data availability. If the hour is not valid because the 
source was not operating for more than 30 minutes of the hour, however, 
the invalid block-hour does not count against the data availability 
recommendation.217
---------------------------------------------------------------------------

    \217\ Data availability is defined as the fraction, expressed as 
a percentage, of the number of block-hours the CEMS is operational 
and obtaining valid data during facility operations, divided by the 
number of block-hours the facility was operating.
---------------------------------------------------------------------------

    2. Total Mercury and Multimetal CEMS. As discussed for particulate 
matter CEMS, we also expect manual methods will be required to 
correlate total mercury and multimetal CEMS prior to using them for 
compliance. For the reasons discussed above in the context of 
particulate matter CEMS, we therefore recommend the observations from 
these CEMS be recorded as block-hour averages and that the 12-hour 
rolling average be updated every hour based on the average of the 
previous 12 block-hour averages.
    3. Hydrochloric Acid and Chlorine Gas CEMS. Unlike the particulate 
matter, total mercury, and multimetal CEMS, hydrochloric acid and 
chlorine gas CEMS are likely to be calibrated using Protocol 1 gas 
bottles rather than correlated to manual method stack test results. 
Therefore, the variability of observations measured by the CEMS over 
some averaging period versus the duration of a stack test is not an 
issue. We conclude that it is appropriate to update the 12-hour rolling 
average for these CEMS every minute, as required for carbon monoxide 
and hydrocarbons CEMS.
    iii. How Are Averaging Periods Calculated Initially and under 
Intermittent Operations?
    1. Practical Effective Date of Rolling Averages for CEMS. As 
discussed in Part Five, Sections VII.B.4 above in the context of 
continuous monitoring systems in general, CEMS recordings will not 
become effective for compliance monitoring on the compliance date until 
you have recorded enough observations to calculate the rolling average 
applicable to the CEMS. For example, the six hourly rolling average for 
particulate matter CEMS does not become effective until you have 
recorded six block-hours of observations on the compliance date. Given 
that compliance with the standards begins nominally at 12:01 am on the 
compliance date, the six hour rolling average for particulate matter 
CEMS does not become effective as a practical matter until 6:01 am on 
the compliance date. Similarly, the 12-hour rolling average for a 
multimetal CEMS does not become effective until you have recorded 12 
block-hours of observations after the compliance date. Thus, the 12-
hour rolling average for multimetals CEMS becomes effective as a 
practical matter at 12:01 p.m. on the compliance date.
    We adopt this approach simply because a rolling average does not 
exist until enough observations have been recorded to calculate the 
rolling average.
    2. How Rolling Averages Are Calculated Upon Intermittent 
Operations. We have determined that you are to ignore periods of time 
when CEMS observations are not recorded for any reason (e.g., source 
shutdown) when calculating rolling averages. For example, consider how 
the six hour rolling average for a particulate matter CEMS would be 
calculated if a source shuts down for yearly maintenance for a three 
week period. The first one-hour block average value recorded when the 
source renews operations is added to the last 5 one-hour block averages 
recorded before the source shut down for maintenance to calculate the 
six hour rolling average.
    We adopt this approach for all continuous monitoring systems, 
including CEMS, because it is simple and reasonable. See discussion in 
Part Five, Section B.4 above.
    c. What Are the Incentives for Using CEMS as Alternative 
Monitoring? We strongly support the use of CEMS for compliance with 
standards, even though we are not requiring their use in today's rule 
(except for carbon monoxide, hydrocarbon, and oxygen CEMS) for the 
reasons discussed above. We endorse the principle that, as technology 
advances, current rules should not act as an obstacle to adopting new 
CEMS technologies for compliance. For instance, today's rule does not 
require total mercury CEMS because implementation and demonstration 
obstacles observed during our tests under what we consider worst-case 
conditions (i.e., a cement kiln) could not be resolved in sufficient 
time to require total mercury CEMS at all hazardous waste combustors. 
However, we fully expect total mercury CEMS will improve to the point 
that the technical issues encountered in our tests can be resolved. At 
that point, we do not want the compliance regime of today's rule--
comprised of emissions testing and limits on operating parameters--to 
be so rigid as to preclude the use of CEMS. Commenters are generally 
supportive of this concept, but note that facilities would be reluctant 
to adopt new technologies without adequate incentives. This section 
describes potential incentives: emissions testing would not be 
required; limits on operating parameters would not apply while the CEMS 
is in service; and the feedstream analysis requirements for the

[[Page 52933]]

parameters measured by the CEMS (i.e., metals or chlorine) would not 
apply.
    i. What Incentives Do Commenters Suggest? Several commenters 
suggest that we provide various incentives to encourage development and 
implementation of new and emerging CEMS. Comments by the Coalition for 
Responsible Waste Incineration (CRWI) include a variety of actions to 
encourage voluntary installation of CEMS,218 including: 
Reduce testing for any parameter measured by a CEMS to the correlation 
and maintenance of that CEMS; waive operating parameter limits that are 
linked to the pollutant measured by the CEMS; minimize regulatory 
oversight on waste analysis if compliance is consistently demonstrated 
by a CEMS; increase the emission limit for a source using a CEMS to 
account for the uncertainty of CEMS observations; allow a phase-in 
period when a source can evaluate CEMS performance and develop 
maintenance practices and the CEMS would not be used for compliance; 
allow a phase-in period to establish a reasonable availability 
requirement for that CEMS at a particular location; and allow sources 
to evaluate CEMS on a trial basis to determine if these instruments are 
appropriate for their operations with no penalties if the units do not 
work or have excessive downtime. Many of CRWI's suggestions have merit, 
as discussed below.
---------------------------------------------------------------------------

    \218\ By ``optional use of CEMS'', we mean using CEM not 
required by this rule, i.e., other than those for carbon monoxide, 
oxygen, and hydrocarbon.
---------------------------------------------------------------------------

    ii. How Do We Respond to Commenter's Recommended Incentives?
    1. Waiver of Emissions Testing and Operating Parameter Limits. 
CRWI's first two suggestions (reduced testing and waiver of operating 
parameter limits) are closely linked. The purpose of conducting a 
comprehensive performance test is to document compliance with emission 
standard initially (and periodically thereafter) and establish limits 
on specified operating parameters to ensure that compliance is 
maintained. Because a CEMS ensures compliance continuously, it serves 
the purpose of both the performance test and compliance with operating 
parameter limits. Accordingly, we agree with CRWI that both emissions 
testing and operating parameter limits for the pollutant in question 
would not apply to sources using a CEMS.
    There is one key caveat to this position, however. Because 100% 
availability of any CEMS is unrealistic, we require a means of assuring 
compliance with the emission standards during periods when the CEMS is 
not available. To meet that need, you may elect to install redundant 
CEMS or assure continuous compliance by monitoring and recording 
traditional operating parameter limits during periods when the CEMS is 
not available. Most likely, you will elect to use operating parameters 
as the back-up when the CEMS is unavailable because it would be a less 
expensive approach. You could establish these operating parameter 
limits, though, through CEMS measurements rather than comprehensive 
performance test measures. In fact, it may be prudent for you to 
evaluate relationships between various operating parameters for the 
particulate matter control device 219 and emission levels 
recorded by the CEMS to develop a good predictive model of emissions. 
You could then petition the Administrator (i.e., permitting officials) 
under Sec. 63.8(f) to base compliance during CEMS malfunctions on 
limits on alternative monitoring parameters derived from the predictive 
model.
---------------------------------------------------------------------------

    \219\ You are not restricted to those specified in Sec. 63.1209. 
You may identify parameters for your source that correlate better 
with particulate emissions than those we have specified generically.
---------------------------------------------------------------------------

    2. Waiver of Feedstream Analysis Requirements. If you obtain 
approval to use a CEMS for compliance under the petitioning provisions 
of Sec. 63.8(f), we agree with the commenter's recommendation that you 
should not be subject to the feedstream analysis requirements pertinent 
to the pollutant you are measuring with a CEMS. As examples, if you use 
a total mercury CEMS, you are not subject to a feedrate limit for 
mercury, and if you operate an incinerator and use a particulate matter 
CEMS, you are not subject to a feedrate limit for total ash.
    If you are not subject to a feedrate limit for ash, metals, or 
chorine because you use a CEMS for compliance, you are not subject to 
the feedstream analysis requirements for these materials. As a 
practical matter, however, this waiver may be moot because, as 
discussed above, you will probably elect to comply with operating 
parameter limits during CEMS malfunctions. However, a second, back-up 
CEMS would also be acceptable. Absent a second CEMS, you would need to 
establish feedrate limits for these materials as a back-up compliance 
approach, and you would need to know the feedrate at any time given 
that the CEMS may malfunction at any time. In addition, even when the 
CEMS is operating within the performance specifications approved by the 
permitting officials, you have the responsibility to minimize 
exceedances by, for example, characterizing your feedstreams adequately 
to enable you to take corrective measures if a CEMS-monitored emission 
is approaching the standard. This level of feedstream characterization, 
however, is less than the characterization required to establish and 
comply with feedrate operating limits during CEMS malfunctions or 
absent a CEMS.
    3. Increase the Averaging Period for CEMS-Monitored Pollutants. The 
averaging period for a CEMS-monitored pollutant should not be 
artificially inflated (i.e., increased beyond the time required to 
conduct three manual method test runs) because the standard would be 
less stringent. See previous discussions on this issue.
    4. Increase Emission Limits to Account for CEMS Uncertainty. We do 
not agree with the suggestion that an emission limit needs to be 
increased on a site-specific basis to accommodate CEMS inaccuracy and 
imprecision (i.e., the acceptance criteria in the CEMS performance 
specification that the source recommends and the permitting officials 
approve will necessarily allow some inaccuracy and imprecision). Again, 
we encourage sources to use a CEMS because it is a better indicator of 
compliance than the promulgated compliance regime (i.e., periodic 
emissions testing and operating parameter limits). We established the 
final emission standards with achievability (through the use of the 
prescribed compliance methods) in mind. We have accounted for the 
inaccuracies and imprecisions in the emissions data in the process of 
establishing the standard. See previous discussions in Part Four, 
Section V.D. If the CEMS performance specification acceptance criteria 
(that must be approved by permitting officials under a Sec. 63.8(f) 
petition) were to allow the CEMS measurements to be more inaccurate or 
imprecise than the promulgated compliance regime of performance testing 
coupled with limits on operating parameters, the potential for improved 
compliance assurance with the CEMS would be negated. Consequently, we 
reject the idea that the standards need to be increased on a site-
specific basis as an incentive for sources to use CEMS.
    5. Allow a CEMS Phase-In Period. CRWI's final three incentive 
suggestions deal with the need for a CEMS phase-in period. This phase-
in period would be used to evaluate CEMS performance, including 
identifying acceptable performance specification levels, maintenance 
requirements, and measurement location. CRWI further suggested that the 
Agency not penalize

[[Page 52934]]

a source if the CEMS does not work or has excessive downtime.
    CRWI provided these comments in response to our proposal to require 
compliance using CEMS and that sources document that the CEMS meets a 
prescribed performance specification and correlation acceptance 
criteria. Although we agree that a phase-in period would be 
appropriate, the issue is moot given that we are not requiring the use 
of CEMS.220 Prior to submitting a petition under 
Sec. 63.8(f) to gain approval to use a CEMS, we presume a source will 
identify the performance specification, correlation criteria, and 
availability factors they believe are achievable. (We expect sources to 
use the criteria we have proposed, as revised after considering 
comments and further analysis and provided through guidance, as a point 
of departure.) Thus, each source will have unlimited opportunity to 
phase-in CEMS and subsequently recommend under Sec. 63.8(f) performance 
specifications and correlation acceptance criteria.
---------------------------------------------------------------------------

    \220\ Other than carbon monoxide, hydrocarbon, and oxygen CEMS.
---------------------------------------------------------------------------

    We do not agree as a legal matter that we can state generically 
that CEMS data obtained during the demonstration period are shielded 
from enforcement if the CEMS data are credible and were to indicate 
exceedance of an emission standard. In this situation, we cannot shield 
a source from action by either by a regulatory agency or a citizen 
suit. On balance, given our legal constraints, our policy desire to 
have CEMS used for compliance, and uncertainty about the ultimate 
accuracy of the CEMS data, we can use our enforcement discretion 
whether to use particulate matter CEMS data as credible evidence in the 
event the CEMS indicates an exceedance until the time the CEMS is 
formally adopted as a compliance tool. Sources and regulators may 
decide to draft a formal testing agreement that states that the CEMS 
data obtained prior to the time the CEMS is accepted as a compliance 
tool cannot be used as credible evidence of exceedance of an emission 
standard.
D. What Are the Compliance Monitoring Requirements?
    In this section we discuss the operating parameter limits that 
ensure compliance with each emission standard.
1. What Are the Operating Parameter Limits for Dioxin/Furan?
    You must maintain compliance with the dioxin/furan emission 
standard by establishing and complying with limits on operating 
parameters. See Sec. 63.1209(k). The following table summarizes these 
operating parameter limits. All sources must comply with the operating 
parameter limits applicable to good combustion practices. Other 
operating parameter limits apply if you use the dioxin/furan control 
technique to which they apply.

BILLING CODE 6560-50-P

[[Page 52935]]

[GRAPHIC] [TIFF OMITTED] TR30SE99.000



[[Page 52936]]

[GRAPHIC] [TIFF OMITTED] TR30SE99.001



BILLING CODE 6560-50-C

[[Page 52937]]

    Dioxin/furan emissions from hazardous waste combustors are 
primarily attributable to surface-catalyzed formation reactions 
downstream from the combustion chamber when gas temperatures are in the 
450  deg.F to 650  deg.F window (e.g., in an electrostatic precipitator 
or fabric filter; in extensive ductwork between the exit of a 
lightweight aggregate kiln and the inlet to the fabric filter; as 
combustion gas passes through an incinerator waste heat recovery 
boiler). In addition, dioxin/furan partition in two phases in stack 
emissions: a portion is adsorbed onto particulate matter and a portion 
is emitted as a vapor (gas). Because of these factors, and absent a 
CEMS for dioxin/furan, we are requiring a combination of approaches to 
control dioxin/furan emissions: (1) Temperature control at the inlet to 
a dry particulate matter control device to limit dioxin/furan formation 
in the control device; (2) operation under good combustion conditions 
to minimize dioxin/furan precursors and dioxin/furan formation during 
combustion; and (3) compliance with operating parameter limits on 
dioxin/furan emission control equipment (e.g., carbon injection) that 
you may elect to use.
    We discuss below the operating parameter limits that apply to each 
dioxin/furan control technique.
    a. Combustion Gas Temperature Quench. To minimize dioxin/furan 
formation in a dry particulate matter control device that suspends 
collected particulate matter in the gas flow (e.g., electrostatic 
precipitator, fabric filter), the rule limits the gas temperature at 
the inlet to these control devices 221 to levels occurring 
during the comprehensive performance test. For lightweight aggregate 
kilns, however, you must monitor the gas temperature at the kiln exit 
rather than at the inlet to the particulate matter control device. This 
is because the dioxin/furan emission standard for lightweight aggregate 
kilns specifies rapid quench of combustion gas to 400  deg.F or less at 
the kiln exit. 222
---------------------------------------------------------------------------

    \221\ The temperature at the inlet to a cyclone separator used 
as a prefiltering process for removing larger particles is not 
limited. Cyclones do not suspend collected particulate matter in the 
gas stream. Thus, these devices do not have the same potential to 
enhance dioxin/furan formation as electrostatic precipitators and 
fabric filters.
    \222\ As discussed in Part Four, Section VIII, lightweight 
aggregate kilns can have extensive ducting between the kiln exit and 
the inlet to the fabric filter. If gas temperatures are limited at 
the inlet to the fabric filter, substantial dioxin/furan formation 
could occur in the ducting.
---------------------------------------------------------------------------

    If your combustor is equipped with a wet scrubber as the initial 
particulate matter control device, you are not required to establish 
limits on combustion gas temperature at the scrubber. This is because 
wet scrubbers do not suspend collected particulate matter in the gas 
stream and gas temperatures are well below 400  deg.F in the 
scrubber.223 Thus, scrubbers do not enhance surface-
catalyzed formation reactions.
---------------------------------------------------------------------------

    \223\ For this reason, you are not required to document during 
the comprehensive performance test that gas temperatures in the wet 
scrubber are not greater than 400  deg.F. Also, we note that the 400 
 deg.F temperature limit of the dioxin/furan standard does not apply 
to wet scrubbers, but rather to the inlet to a dry particulate 
matter control device and the kiln exit of a lightweight aggregate 
kiln.
---------------------------------------------------------------------------

    We proposed limits on the gas temperature at the inlet to a dry 
particulate matter control device (see 61 FR at 17424). Temperature 
control at this location is important because surface-catalyzed 
formation reactions can increase by a factor of 10 for every 150  deg.F 
increase in temperature within the window of 350  deg.F to 
approximately 700  deg.F. We received no adverse comments on the 
proposal, and thus, are adopting this compliance requirement in the 
final rule.
    You must establish an hourly rolling average temperature limit 
based on operations during the comprehensive performance test. The 
hourly rolling average limit is established as the average of the test 
run averages. See Part Five, Sections VII.B.1 and B.3 above for a 
discussion on the approach for calculating limits from comprehensive 
performance test data.
    b. Good Combustion Practices. All hazardous waste combustors must 
use good combustion practices to control dioxin/furan emissions by: (1) 
Destroying dioxin/furan that may be present in feedstreams; (2) 
minimizing formation of dioxin/furan during combustion; and (3) 
minimizing dioxin/furan precursor that could enhance post-combustion 
formation reactions. As proposed, you must establish and continuously 
monitor limits on three key operating parameters that affect good 
combustion: (1) Maximum hazardous waste feedrate; (2) minimum 
temperature at the exit of each combustion chamber; and (3) residence 
time in the combustion chamber as indicated by gas flowrate or kiln 
production rate. We have also determined that you must establish 
appropriate monitoring requirements to ensure that the operation of 
each hazardous waste firing system is maintained. We discuss each of 
these parameters below.
    i. Maximum Hazardous Waste Feedrate. You must establish and 
continuously monitor a maximum hazardous waste feedrate limit for 
pumpable and nonpumpable wastes. See 61 FR at 17422. An increase in 
waste feedrate without a corresponding increase in combustion air can 
cause inefficient combustion that may produce (or incompletely destroy) 
dioxin/furan precursors. You must also establish hazardous waste 
feedrate limits for each location where waste is fed.
    One commenter suggests that there is no reason to limit the 
feedrate of each feedstream; a limit on the total hazardous waste 
feedrate to each combustion chamber would be a more appropriate control 
parameter. We concur in part. Limits are not established for each 
feedstream. Rather, limits apply to total and pumpable wastes feedrates 
for each feed location. Limits on pumpable wastes are needed because 
the physical form of the waste can affect the rate of oxygen demand and 
thus combustion efficiency. Pumpable wastes often will expose a greater 
surface area per mass of waste than nonpumpable wastes, thus creating a 
more rapid oxygen demand. If that demand is not satisfied, inefficient 
combustion will occur. We also note that these waste feedrate limit 
requirements are consistent with current RCRA permitting requirements 
for hazardous waste combustors.
    As proposed, you must establish hourly rolling average limits for 
hazardous waste feedrate from comprehensive performance test data as 
the average of the highest hourly rolling averages for each run. See 
Part Five, Section VII.B.3 above for the rationale for this approach 
for calculating limits from comprehensive performance test data.
    ii. Minimum Gas Temperature in the Combustion Zone. You must 
establish and continuously monitor limits on minimum gas temperature in 
the combustion zone of each combustion chamber irrespective of whether 
hazardous waste is fed into the chamber. See 61 FR at 17422. These 
limits are needed because, as combustion zone temperatures decrease, 
combustion efficiency can decrease resulting in increased formation of 
(or incomplete destruction of) dioxin/furan precursors.224
---------------------------------------------------------------------------

    \224\ See USEPA, ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume IV: Compliance with 
the Hazardous Waste Combustor Standards'', February, 1999.
---------------------------------------------------------------------------

    Monitoring combustion zone temperatures can be problematic, 
however, because the actual burning zone temperature cannot be measured 
at many units (e.g., cement kilns). For this reason, the BIF rule 
requires

[[Page 52938]]

measurement of the ``combustion chamber temperature where the 
temperature measurement is as close to the combustion zone as 
possible.'' See Sec. 266.103(c)(1)(vii). In some cases, temperature is 
measured at a location quite removed from the combustion zone due to 
extreme temperatures and the harsh conditions at the combustion zone. 
We discussed this issue at proposal and indicated that we were 
concerned that monitoring at such remote locations may not accurately 
reflect changes in combustion zone temperatures. See 61 FR at 17423.
    We requested comment on possible options to address the issue. 
Under one option, the final rule would have allowed the source to 
identify a parameter that correlates with combustion zone temperature 
and to provide data or information to support the use of that parameter 
in the operating record. Under another option, the final rule would 
have enabled regulatory officials on a case-specific basis to require 
the use of alternate parameters as deemed appropriate, or to determine 
that there is no practicable approach to ensure that minimum combustion 
chamber temperature is maintained (and what the recourse/consequence 
would be).
    Some commenters recommend the status quo as identified by the BIF 
rule requirements for monitoring combustion zone temperature. These 
commenters suggest that more prescriptive requirements would not be 
implementable for cement kilns because use of the temperature 
measurement instrumentation would simply not be practicable under 
combustion zone conditions in a cement kiln. We agree that combustion 
zone temperature monitoring for certain types of sources requires some 
site-specific considerations (as evidenced in our second proposed 
option discussed above), and conclude that more specific language than 
that used in the BIF rule to address this issue would not be 
appropriate. Accordingly, we adopt language similar to the BIF rule in 
today's final rule. You must measure the temperature of each combustion 
chamber at a location that best represents, as practicable, the bulk 
gas temperature in the combustion zone of that chamber. You are 
required to identify the temperature measurement location and method in 
the comprehensive performance test plan, which is subject to Agency 
approval.
    The temperature limit(s) apply to each combustion zone, as 
proposed. See 61 FR at 17423. For incinerators with a primary and 
secondary chamber, you must establish separate limits for the 
combustion zone in each chamber.225 For kilns, you must 
establish separate temperature limits at each location where hazardous 
waste may be fired (e.g., the hot end where clinker is discharged; and 
the upper end of the kiln where raw material is fed). We also proposed 
to include temperature limits for hazardous waste fired at the midkiln. 
One commenter indicates that it is technically infeasible to measure 
temperature directly at the midkiln waste feeding location, however. We 
agree that midkiln gas temperature is difficult to measure due to the 
rotation of the kiln.226 Thus, the final rule allows 
temperature measurement at the kiln back-end as a surrogate.
---------------------------------------------------------------------------

    \225\ The temperature limits apply to a combustion chamber even 
if hazardous waste is not burned in the chamber for two reasons. 
First, an incinerator may rely on an afterburner that is fired with 
a fuel other than hazardous waste to ensure good combustion of 
organic compounds volatilized from hazardous waste in the primary 
chamber. Second, MACT controls apply to total emissions (except 
where the rule makes specific provisions), irrespective of whether 
they derive from burning hazardous waste or other material, or from 
raw materials.
    \226\ See USEPA. ``Final Technical Support Document for 
Hazardous Waste Combustor MACT Standards, Volume IV: Compliance with 
the Hazardous Waste Combustor Standards'', February, 1999, for 
further discussion.
---------------------------------------------------------------------------

    You must establish an hourly rolling average temperature limit 
based on operations during the comprehensive performance test. The 
hourly rolling average limit is established as the average of the test 
run averages. See Part Five, Sections VII.B.1 and B.3 above for a 
discussion on the approach for calculating limits from comprehensive 
performance test data.
    iii. Maximum Flue Gas Rate or Kiln Production Rate. As proposed, 
you must establish and continuously monitor a limit on maximum flue gas 
flowrate or, as a surrogate, kiln production rate. See 61 FR at 17423. 
Flue gas flowrates in excess of those that occur during comprehensive 
performance testing reduce the time that combustion gases are exposed 
to combustion chamber temperatures. Thus, combustion efficiency can 
decrease potentially causing an increase in dioxin/furan precursors 
and, ultimately, dioxin/furan emissions.227
---------------------------------------------------------------------------

    \227\ We note that an increase in gas flowrate can also 
adversely affect the performance of a dioxin/furan emission control 
device (e.g., carbon injection, catalytic oxidizer). Thus, gas 
flowrate is controlled for this reason as well.
---------------------------------------------------------------------------

    For cement kilns and lightweight aggregate kilns, the rule allows 
the use of production rate as a surrogate for flue gas flowrate. This 
is the approach currently used for the BIF rule for these devices, 
given that flue gas flowrate correlates with production rate (e.g., 
feedrate of raw materials or rate of production of clinker or 
aggregate).
    At proposal, however, we expressed concern that production rate may 
not relate well to flue gas flowrate in situations where the moisture 
content of the feed to the combustor changes dramatically. See 61 FR at 
17423. Some commenters concur and also express concern that production 
rate is not a reliable surrogate for flue gas flowrate because changes 
in ambient temperature can cause increased heat rates and changes in 
operating conditions can result in variability in excess air rates. 
Based on an analysis of kiln processes, however, we conclude that these 
issues should not be a concern. With respect to changes in moisture 
content of the feed, kilns tend to have a steady and homogeneous waste 
and raw material processing system. Thus, the feed moisture content 
does not fluctuate widely, and variation in moisture content of the 
stack does not significantly affect gas flowrate.228 Thus, 
production rate should be an adequate surrogate for gas flowrate for 
our purposes here.
---------------------------------------------------------------------------

    \228\ See USEPA, ``Final TSD for hazardous Waste Combustor MACT 
Standards, Volume IV: Compliance with the Hazardous Waste Combustor 
Standards'', February, 1999 for further discussion.
---------------------------------------------------------------------------

    You must establish a maximum gas flowrate or production rate limit 
as the average of the maximum hourly rolling averages for each run of 
the comprehensive performance test. See Part Five, Sections VII.B.3 
above for the rationale for the approach for calculating limits from 
comprehensive performance test data.
    iv. Operation of Each Hazardous Waste Firing System. You must 
recommend in the comprehensive performance test plan that you submit 
for review and approval operating parameters, limits, and monitoring 
approaches to ensure that each hazardous waste firing system continues 
to operate as efficiently as demonstrated during the comprehensive 
performance test.
    It is important to maintain operation of the hazardous waste firing 
system at levels of the performance test to ensure that the same or 
greater surface area of the waste is exposed to combustion conditions 
(e.g., temperature and oxygen). Oxidation takes place more quickly and 
completely as the surface area per unit of mass of the waste increases. 
If the firing system were to degrade over time such that smaller 
surface area is exposed to combustion conditions, inefficient 
combustion could result leading potentially to an increase in dioxin/
furan precursors.

[[Page 52939]]

    At proposal, we discussed establishing operating parameter limits 
only for minimum nozzle pressure and maximum viscosity of wastes fired 
using a liquid waste injection system. In developing the final rule, 
however, we determined that RCRA permit writers currently establish 
operating parameter limits on each waste firing system to ensure 
compliance with the RCRA destruction and removal efficiency (DRE) 
standard. We are continuing the DRE requirement as a MACT standard, and 
as discussed in Section VII.D.7 below, the DRE operating parameter 
limits are identical to those required to maintain good combustion 
practices for compliance with the dioxin/furan standard. This is 
because compliance with the DRE standard is ensured by maintaining good 
combustion practices. Consequently, we include a requirement to 
establish limits on operating parameters for each waste or fuel firing 
system as a measure of good combustion practices for the dioxin/furan 
standard as well to be technically correct and for purposes of 
completeness.229 Because this requirement is identical to an 
existing RCRA requirement, it will not impose an incremental burden.
---------------------------------------------------------------------------

    \229\ Because incomplete combustion of fuels (e.g., oil, coal, 
tires) could contribute to increased dioxin/furan emissions by 
producing dioxin/furan precursors, permitting official may require 
(during review and approval of the comprehensive performance test 
plan) that you establish limits on operating parameters for firing 
systems in addition to those firing hazardous waste.
---------------------------------------------------------------------------

    The rule does not prescribe generic operating parameters and how to 
identify limits because, given the variety of firing systems and waste 
and fuel properties, they are better defined on a site-specific basis. 
Examples of monitoring parameters for a liquid waste firing system 
would be, as proposed, minimum nozzle pressure established as an hourly 
rolling average based on the average of the minimum hourly rolling 
averages for each run, coupled with a limit on maximum waste viscosity. 
The viscosity limit could be monitored periodically based on sampling 
and analysis. Examples of monitoring parameters for a lance firing 
system for sludges could be minimum pressure established as discussed 
above, plus a limit on the solids content of the waste.
    v. Consideration of Restrictions on Batch Size, Feeding Frequency, 
and Minimum Oxygen Concentration. We proposed site-specific limits on 
maximum batch size, batch feeding frequency, and minimum combustion gas 
oxygen concentration as additional compliance requirements to ensure 
good combustion practices. See 61 FR at 17423. After carefully 
considering all comments, and for the reasons discussed below, we 
conclude that the carbon monoxide and hydrocarbon emission standards 
assure use of good combustion practices during batch feed operations. 
This is because the carbon monoxide and hydrocarbon CEMS are reliable 
and continuous indicators of combustion efficiency. In situations where 
batch feed operating requirements may be needed to better assure good 
combustion practices, however, we rely on the permit writer's 
discretionary authority under Sec. 63.1209(g)(2) to impose additional 
operating parameter limits on a site-specific basis.
    Many hazardous waste combustors burn waste fuel in batches, such as 
metal drums or plastic containers. Some containerized waste can 
volatilize rapidly, causing a momentary oxygen-deficient condition that 
can result in an increase in emissions of carbon monoxide, hydrocarbon, 
and dioxin/furan precursors. We proposed to limit batch size, batch 
feeding frequency, and minimum combustion gas oxygen concentration to 
address this concern.
    Commenters suggest that the proposed batch feed requirements (that 
would limit operations to the smallest batch, the longest time 
interval, and the maximum oxygen concentration demonstrated during the 
comprehensive performance test) would result in extremely conservative 
limits that would severely limit a source's ability to batch-feed 
waste. Given these concerns and our reanalysis of the need for these 
limits, we conclude that the carbon monoxide and hydrocarbon emission 
standards will effectively ensure good combustion practices for most 
batch feed operations. Consequently, the final rule does not require 
limits for batch feed operating parameters.
    Carbon monoxide or hydrocarbon monitoring may not be adequate for 
all batch feed operations, however, to ensure good combustion practices 
are maintained. We anticipate that permitting officials will determine 
on a site-specific basis, typically during review of the initial 
comprehensive performance test plan, whether limits on one or more 
batch feed operating parameters need to be established to ensure good 
combustion practices are maintained. This review should consider your 
previous compliance history (e.g., frequency of automatic waste feed 
cutoffs attributable to batch feed operations that resulted in an 
exceedance of an operating limit or standard under RCRA regulations 
prior to the compliance date), together with the design and operating 
features of the combustor. Providing permitting officials the authority 
under Sec. 63.1209(g)(2) to establish batch feed operating parameter 
limits only where warranted precludes the need to impose the limits on 
all sources.
    Permitting officials may also determine that limits on batch feed 
operating parameters are needed for a particular source based on the 
frequency of automatic waste feed cutoffs after the MACT compliance 
date. Permitting officials would consider cutoffs that are attributable 
to batch feed operations and that result in an exceedance of an 
operating parameter limit or the carbon monoxide or hydrocarbon 
emission standard. Given that you must notify permitting officials if 
you have 10 or more automatic waste feed cutoffs in a 60-day period 
that result in an exceedance of an operating parameter limit or CEMS-
monitored emission standard, permitting officials should take the 
opportunity to determine if batch feed operations contributed to the 
frequency of exceedances. If so, permitting officials should use the 
authority under Sec. 63.1209(g)(2) to establish batch feed operating 
parameter limits.
    Although we are not finalizing batch feed operating parameter 
limits, we anticipate that permitting officials will require you 
(during review and approval of the test plan) to simulate worst-case 
batch feed operating conditions during the comprehensive performance 
test when demonstrating compliance with the dioxin/furan and 
destruction and removal efficiency standards. It would be inappropriate 
for you to operate your batch feed system during the comprehensive 
performance test in a manner that is not considered worst-case, 
considering the types and quantities of wastes you may burn, and the 
range of values you may encounter during operations for batch feed-
related operating parameters (e.g., oxygen levels, batch size and/or 
btu content, waste volatility, batch feeding frequency).
    To ensure that the CEMS-monitored carbon monoxide and hydrocarbon 
emission standards ensure good combustion practices for batch feed 
operations, the final rule includes special requirements to ensure that 
``out-of-span'' carbon monoxide and hydrocarbon CEMS readings are 
adequately accounted for. We proposed batch feed operating parameter 
limits in part because of concern that the carbon monoxide and 
hydrocarbon CEMS may not accurately calculate hourly rolling averages 
when you encounter emission concentrations that exceed the span of the 
CEMS. This is an important

[[Page 52940]]

consideration because batch feed operations have the potential to 
generate large carbon monoxide or hydrocarbon spikes--large enough at 
times to exceed the span of the detector. When this occurs, the CEMS in 
effect ``pegs out'' and the analyzer may only record data at the upper 
end of its span, while in fact carbon monoxide/hydrocarbon 
concentrations are much higher. In these situations, the true carbon 
monoxide/hydrocarbon concentration is not being used to calculate the 
hourly rolling average. This has two significant consequences of 
concern to us.230
---------------------------------------------------------------------------

    \230\ As explained in Part Five, Section VII.D.4 of the text, 
this concern is not limited to batch feed operations.
---------------------------------------------------------------------------

    First, you could experience a large carbon monoxide/hydrocarbon 
spike (as a result of feeding a large or highly volatile batch) which 
causes the monitor to ``peg out.'' In this situation, the CEMS would 
record carbon monoxide/hydrocarbon levels that are lower than actual 
levels. This under-reporting of emission levels would result in an 
hourly rolling average that is biased low. You may in fact be exceeding 
the emission standard even though the CEMS indicates you are in 
compliance. Second, if a carbon monoxide/hydrocarbon excursion causes 
an automatic waste feed cutoff, you may be allowed to resume hazardous 
waste burning much sooner than you would be allowed if the CEMS were 
measuring true hourly rolling averages. This is because you must 
continue monitoring operating parameter limits and CEMS-monitored 
emission standards after an automatic waste feed cutoff and you may not 
restart hazardous waste feeding until all limits and CEMS-monitored 
emission standards are within permissible levels.231
---------------------------------------------------------------------------

    \231\ A higher hourly rolling average carbon monoxide level that 
is above the standard requires a longer period of time to drop below 
the standard.
---------------------------------------------------------------------------

    As explained in Part Five, Section VII.D.4 below, we have resolved 
these ``out of span'' concerns by including special provisions in 
today's rule for instances when you encounter hydrocarbon/carbon 
monoxide CEMS measurements that are above the upper span required by 
the performance specifications.232 These special provisions 
require you to assume hydrocarbons and carbon monoxide are being 
emitted at levels of 500 ppmv and 10,000 ppmv, respectively, when any 
one minute average exceeds the upper span level of the 
detector.233 Although we did not propose these special 
provisions, they are a logical outgrowth of the proposed batch feed 
requirements and commenters concerns about those requirements.
---------------------------------------------------------------------------

    \232\ The carbon monoxide CEMS upper span level for the high 
range is 3000 ppmv. The upper span level for hydrocarbon CEMS is 100 
ppmv. (See Performance Specifications 4B and 8A in Appendix B, part 
60, and the appendix to subpart EEE, part 63--Quality Assurance 
Procedures for Continuous Emissions Monitors Used for Hazardous 
Waste Combustors, Section 6.3).
    \233\ You would not be required to assume these one-minute 
values if you use a CEMS that meets the performance specifications 
for a range that is higher than the recorded one-minute average. In 
this case, the CEMS must meet performance specifications for the 
higher range as well as the ranges specified in the performance 
specifications in Appendix B, part 60. See Sec. 63.1209 (a)(3) and 
(a)(4).
---------------------------------------------------------------------------

    For the reasons discussed above, we conclude that national 
requirements for batch feed operating parameter limits are not 
warranted.
    c. Activated Carbon Injection. If your combustor is equipped with 
an activated carbon injection system, you must establish and comply 
with limits on the following operating parameters: Good particulate 
matter control, minimum carbon feedrate, minimum carrier fluid flowrate 
or nozzle pressure drop, and identification of the carbon brand and 
type or the adsorption characteristics of the carbon. These are the 
same compliance parameters that we proposed. See 61 FR at 17424.
    i. Good Particulate Matter Control. You must comply with the 
operating parameter limits for particulate matter control (see 
discussion in Section VII.D.6 below and Sec. 63.1209(m)) because carbon 
injection controls dioxin/furan in conjunction with particulate matter 
control. Dioxin/furan is adsorbed onto carbon that is injected into the 
combustion gas, and the carbon is removed from stack gas by a 
particulate control device.
    Although we proposed to require good particulate matter control as 
a control technique for dioxin/furan irrespective of whether carbon 
injection was used, commenters indicate that we have no data 
demonstrating the relationship between particulate matter and dioxin/
furan emissions. Commenters further indicate that dioxin/furan occur 
predominately in the gas phase, not adsorbed onto particulate. We agree 
with commenters that hazardous waste combustors operating under the 
good combustion practices required by this final rule are not likely to 
have significant carbon particulates in stack gas (i.e., because 
carbonaceous particulates (soot) are indicative of poor combustion 
efficiency). Thus, unless activated carbon injection is used as a 
control technique, dioxin/furan will occur predominately in the gas 
phase. We therefore conclude that requiring good particulate control as 
a control technique for dioxin/furan is not warranted unless a source 
is equipped with activated carbon injection.234
---------------------------------------------------------------------------

    \234\ We discuss below, however, that good particulate matter 
control is also required if a source is equipped with a carbon bed. 
This is to ensure that particulate control upstream of the carbon 
bed is maintained to performance test levels to prevent blinding of 
the bed and loss of removal efficiency.
---------------------------------------------------------------------------

    ii. Minimum Carbon Feedrate. As proposed, you must establish and 
continuously monitor a limit on minimum carbon feedrate to ensure that 
dioxin/furan removal efficiency is maintained. You must establish an 
hourly rolling average feedrate limit based on operations during the 
comprehensive performance test. The hourly rolling average limit is 
established as the average of the test run averages. See Part Five, 
Sections VII.B.1 and B.3 above for a discussion of the approach for 
calculating limits from comprehensive performance test data.
    iii. Minimum Carrier Fluid Flowrate or Nozzle Pressure Drop. A 
carrier fluid, gas or liquid, is necessary to transport and inject the 
carbon into the gas stream. As proposed, you must establish and 
continuously monitor a limit on either minimum carrier fluid flowrate 
or pressure drop across the nozzle to ensure that the flow and 
dispersion of the injected carbon into the flue gas stream is 
maintained.
    We proposed to require you to base the limit on the carbon 
injection manufacturer's specifications. One commenter notes that there 
are no manufacturer specifications for carrier gas flowrate or pressure 
drop. Therefore, the final rule allows you to use engineering 
information and principles to establish the limit for minimum carrier 
fluid flowrate or pressure drop across the injection nozzle. You must 
identify the limit and the rationale for deriving it in the 
comprehensive performance test plan that you submit for review and 
approval.
    iv. Identification of Carbon Brand and Type or Adsorption 
Properties. You must either identify the carbon brand and type used 
during the comprehensive performance test and continue using that 
carbon, or identify the adsorption properties of that carbon and use a 
carbon having equivalent or better properties. This will ensure that 
the carbon's adsorption properties are maintained.235
---------------------------------------------------------------------------

    \235\ Examples of carbon properties include specific surface 
area, pore volume, average pore size, pore size distribution, bulk 
density, porosity, carbon source, impregnation, and activization 
procedure. See USEPA, ``Technical Support Document for HWC MACT 
Standards, Volume IV: Compliance with the HWC MACT Standards,'' July 
1999.
---------------------------------------------------------------------------

    We proposed to require you to use the same brand and type of carbon 
that was

[[Page 52941]]

used during the comprehensive performance test. Commenters object to 
this requirement and suggest that they should have the option of using 
alternative types of carbon that would achieve equivalent or better 
performance than the carbon used during the performance test. We 
concur, and the final rule allows you to document in the comprehensive 
performance test plan key parameters that affect adsorption and the 
limits you have established on those parameters based on the carbon to 
be used during the performance test. You may substitute at any time a 
different brand or type of carbon provided that the replacement has 
equivalent or improved properties and conforms to the key sorbent 
parameters you have identified. You must include in the operating 
record written documentation that the substitute carbon will provide 
the same level of control as the original carbon.
    d. Activated Carbon Bed. If your combustor is equipped with an 
activated carbon bed, you must establish and comply with limits on the 
following operating parameters: good particulate matter control; 
maximum age of each carbon bed segment; identification of carbon brand 
and type or adsorption properties, and maximum temperature at the inlet 
or exit of the bed. These are the same compliance parameters that we 
proposed. See 61 FR at 17424.
    i. Good Particulate Matter Control. You must comply with the 
operating parameter limits for particulate matter control (see 
discussion in Section VII.D.6 below and Sec. 63.1209(m)). If good 
control of particulate matter is not maintained prior to the inlet to 
the carbon bed, particulate matter could contaminate the bed and affect 
dioxin/furan removal efficiency. In addition, if particulate matter 
control is used downstream from the carbon bed, those controls must 
conform to good particulate matter control. This is because this 
``polishing'' particulate matter control device may capture carbon-
containing dioxin/furan that may escape from the carbon bed. Thus, the 
efficiency of this polishing control must be maintained to ensure 
compliance with the dioxin/furan emission standard.
    ii. Maximum Age of Each Bed Segment. As proposed, you must 
establish a maximum age of each bed segment to ensure that removal 
efficiency is maintained. Because activated carbon removes dioxin/furan 
(and mercury) by adsorption, carbon in the bed becomes less effective 
over time as the active sites for adsorption become occupied. Thus, bed 
age is an important operating parameter.
    At proposal, we requested comment on using carbon aging or some 
form of a breakthrough calculation to identify a limit on carbon age. 
See 61 FR at 17424. A breakthrough calculation would give a theoretical 
minimum carbon change-out schedule that you could use to ensure that 
breakthrough (i.e., the dramatic reduction in efficiency of the carbon 
bed due to too many active sites being occupied) does not occur.
    Commenters indicate that carbon effectiveness depends on the carbon 
bed age and pollutant types and concentrations in the gas streams, and 
therefore a carbon change-out schedule should be based on a 
breakthrough calculation rather than carbon age. We agree that a 
breakthrough calculation may be a better measurement of carbon 
effectiveness, but it would be difficult to define generically for all 
situations. A breakthrough calculation could be performed only after 
experimentation determines the relationship between incoming adsorbed 
chemicals and the adsorption rate of the carbon. The adsorption rate of 
carbon could be determined experimentally, but the speciation of 
adsorbed chemicals in a flue gas stream is site-specific and may vary 
greatly at a given site over time.
    We conclude that because carbon age contributes to carbon 
ineffectiveness, it serves as an adequate surrogate and is less 
difficult to implement on a national basis. Therefore, the rule 
requires sources to identify maximum carbon age as the maximum age of 
each bed segment during the comprehensive performance test. Carbon age 
is measured in terms of the cumulative volume of combustion gas flow 
through the carbon since its addition to the bed. Sources may use the 
manufacturer's specifications rather than actual bed age during the 
initial comprehensive performance test to identify the initial limit on 
maximum bed age. If you elect to use manufacturer's specifications for 
the initial limit on bed age, you must also recommend in the 
comprehensive performance test plan submitted for review and approval a 
schedule of dioxin/furan testing prior to the confirmatory performance 
test that will confirm that the manufacturer's specification of bed age 
is sufficient to ensure that you maintain compliance with the emission 
standard.
    If either existing or new sources prefer to use some form of 
breakthrough calculation to establish maximum bed age, you may petition 
permitting officials under Sec. 63.1209(g)(1) 236 to apply 
for an alternative monitoring scheme.
---------------------------------------------------------------------------

    \236\ We have incorporated the alternative monitoring provisions 
of Sec. 63.8(f) in Sec. 63.1209(g)(1) so that alternative monitoring 
provisions for nonCEMS CMS can be implemented by authorized States. 
The alternative monitoring provisions of Sec. 63.1209(g)(1) do not 
apply to CEMS, however. The alternative monitoring provisions of 
Sec. 63.8(f) continue to apply to CEMS because implementation of 
those provisions is not eligible to be delegated to States at this 
time.
---------------------------------------------------------------------------

    iii. Identification of Carbon Brand and Type or Adsorption 
Properties. You must either identify the carbon brand and type used 
during the comprehensive performance test and continue using that 
carbon, or identify the adsorption properties of that carbon and use a 
carbon having equivalent or better properties. This requirement is 
identical to that discussed above for activated carbon injection 
systems.
    iv. Maximum Temperature at the Inlet or Exit of the Bed. You must 
establish and continuously monitor a limit on the maximum temperature 
at the inlet or exit of the carbon bed. This is because a combustion 
gas temperature spike can cause adsorbed dioxin/furan (and mercury) to 
desorb and reenter the gas stream. In addition, the adsorption 
properties of carbon are adversely affected at higher temperatures.
    At proposal, we requested comment on whether it would be necessary 
to control temperature at the inlet to the carbon bed. See 61 FR at 
17425. Some commenters support temperature control noting the concern 
that temperature spikes could cause desorption of dioxin/furan (and 
mercury). We concur, and are requiring you to establish a maximum 
temperature limit at the inlet or exit of the bed. We are allowing you 
the option of measuring temperature at either end of the bed to give 
you greater flexibility in locating the temperature continuous 
monitoring system. Monitoring temperature at either end of the bed 
should be adequate to ensure that bed temperatures are maintained at 
levels not exceeding those during the comprehensive performance test 
(because the temperature remains relatively constant across the bed).
    You must establish an hourly rolling average temperature limit 
based on operations during the comprehensive performance test. The 
hourly rolling average limit is established as the average of the test 
run averages. See Part Five, Sections VII.B.1 and B.3 above for a 
discussion of the approach for calculating limits from comprehensive 
performance test data.
    e. Catalytic Oxidizer. If your combustor is equipped with a 
catalytic oxidizer, you must establish and comply with limits on the 
following operating parameters: minimum gas temperature

[[Page 52942]]

at the inlet of the catalyst; maximum age in use; catalyst replacement 
specifications; and maximum flue gas temperature at the inlet of the 
catalyst. These are the same compliance parameters that we proposed. 
See 61 FR at 17425.
    Catalytic oxidizers used to control stack emissions are similar to 
those used in automotive and industrial applications. The flue gas 
passes over catalytic metals, such as palladium and platinum, supported 
by an alumina washcoat on some metal or ceramic substrate. When the 
flue gas passes through the catalyst, a reaction takes place similar to 
combustion, converting hydrocarbons to carbon monoxide, then carbon 
dioxide. Catalytic oxidizers can also be ``poisoned'' by lead and other 
metals in the same manner as automotive and industrial catalysts.
    i. Minimum Gas Temperature at the Inlet of the Catalyst. You must 
establish and continuously monitor a limit on the minimum flue gas 
temperature at the inlet of the catalyst to ensure that the catalyst is 
above light-off temperature. Light-off temperature is that minimum 
temperature at which the catalyst is hot enough to catalyze the 
reactions of hydrocarbons and carbon monoxide.
    You must establish an hourly rolling average temperature limit 
based on operations during the comprehensive performance test. The 
hourly rolling average limit is established as the average of the test 
run averages.
    ii. Maximum Time In-Use. You must establish a limit on the maximum 
time in-use of the catalyst because a catalyst is poisoned and 
generally degraded over use. You must establish the limit based on the 
manufacturer's specifications.
    iii. Catalytic Metal Loading, Maximum Space-Time, and Substrate 
Construct. When you replace a catalyst, the replacement must be of the 
same design to ensure that destruction efficiency is maintained. 
Consequently, the rule requires that you specify the following catalyst 
properties: Loading of catalytic metals; space-time; and monolith 
substrate construction.
    Catalytic metal loading is important because, without sufficient 
catalytic metal on the catalyst, it does not function properly. Also, 
some catalytic metals are more efficient than others. Therefore, the 
replacement catalyst must have at least the same catalytic metal 
loading for each catalytic metal as the catalyst used during the 
comprehensive performance test.
    Space-time, expressed in inverse seconds (s-1), is 
defined as the maximum rated volumetric flow through the catalyst 
divided by the volume of the catalyst. This is important because it is 
a measure of the gas flow residence time and, hence, the amount of time 
the flue gas is in the catalyst. The longer the gas is in the catalyst, 
the more time the catalyst has to cause hydrocarbons and carbon 
monoxide to react. Replacement catalysts must have the same or lower 
space-time as the one used during the comprehensive performance test.
    Substrate construction is also an important parameter affecting 
destruction efficiency of the catalyst. Three factors are important. 
First, substrates for industrial applications are typically monoliths, 
made of rippled metal plates banded together around the circumference 
of the catalyst. Ceramic monoliths and pellets can also be used. 
Because of the many types of substrates, you must use the same 
materials of construction, monolith or pellets and metal or ceramic, 
used during the comprehensive performance test as replacements. Second, 
monoliths form a honeycomb like structure when viewed from one end. The 
pore density (i.e., number of pores per square inch) is critical 
because the pores must be small enough to ensure intimate contact 
between the flue gas and the catalyst but large enough to allow 
unrestricted flow through the catalyst. Therefore, if you use a 
monolith substrate during the comprehensive performance test, the 
replacement catalyst must have the same pore density. Third, catalysts 
are supported by a washcoat, typically alumina. We require that 
replacement catalysts have the same type and loading of washcoat as was 
on the catalyst used during the comprehensive performance test.
    iv. Maximum Flue Gas Temperature at the Inlet to the Catalyst. You 
must establish and continuously monitor a limit on maximum flue gas 
temperature at the inlet to the catalyst. Inlet temperature is 
important because sustained high flue gas temperature can result in 
sintering of the catalyst, degrading its performance. You must 
establish the limit as an hourly rolling average, based on manufacturer 
specifications.
    In the proposed rule, we would have allowed a waiver from these 
operating parameter limits if you documented to the Administrator that 
establishing limits on other operating parameters would be more 
appropriate to ensure that the dioxin/furan destruction efficiency of 
the oxidizer is maintained after the performance test. See 61 FR at 
17425. We are not finalizing a specific waiver for catalytic oxidizer 
parameters because you are eligible to apply for the same relief under 
the existing alternative monitoring provisions of Sec. 63.1209(g)(1).
    f. Dioxin/Furan Formation Inhibitor. If you feed a dioxin/furan 
formation inhibitor into your combustor as an additive (e.g., sulfur), 
you must: (1) Establish a limit on minimum inhibitor feedrate; and (2) 
identify either the brand and type of inhibitor or the properties of 
the inhibitor.
    i. Minimum Inhibitor Feedrate. As proposed, you must establish and 
continuously monitor a limit on minimum inhibitor feedrate to help 
ensure that dioxin/furan formation reactions continue to be inhibited 
at levels of the comprehensive performance test. See 61 FR at 17425. 
You must establish an hourly rolling average feedrate limit based on 
operations during the comprehensive performance test. The hourly 
rolling average limit is established as the average of the test run 
averages.
    This minimum inhibitor feedrate pertains to additives to 
feedstreams, not naturally occurring inhibitors that may be found in 
fossil fuels, hazardous waste, or raw materials. At proposal, we 
requested comment on whether it would be appropriate to establish 
feedrate limits on the amount of naturally occurring inhibitors based 
on levels fed during the comprehensive performance test. See 61 FR at 
17425. For example, it is conceivable that a source would choose to 
burn high sulfur fuel or waste only during the comprehensive 
performance test and then switch back to low sulfur fuels or waste 
after the test, thus reducing dioxin/furan emissions during the 
comprehensive test to levels that would not be maintained after the 
test. Commenters do not provide information on this matter and we do 
not have enough information on the types or effects of naturally 
occurring substances that may act as inhibitors. Therefore, the final 
rule does not establish limits on naturally occurring inhibitors. 
Permitting officials, however, may choose to address the issue of 
naturally occurring inhibitors when warranted during review of the 
comprehensive performance test plan. (See discretionary authority of 
permitting officials under Sec. 63.1209(g)(2) to impose additional or 
alternative operating parameter limits on a site-specific basis.)
    ii. Identification of Either the Brand and Type of Inhibitor or the 
Properties of the Inhibitor. As proposed, you must either identify the 
inhibitor brand and type used during the comprehensive performance test 
and continue using that inhibitor, or identify the properties of that 
inhibitor that affect its ability to inhibit dioxin/furan formation 
reactions and use an inhibitor having equivalent

[[Page 52943]]

or better properties. This requirement is identical to that discussed 
above for activated carbon systems.
2. What Are the Operating Parameter Limits for Mercury?
    You must maintain compliance with the mercury emission standard by 
establishing and complying with limits on operating parameters. See 
Sec. 63.1209(l). The following table summarizes these operating 
parameter limits. All sources must comply with the limits on mercury 
feedrate. Other operating parameter limits apply if you use the mercury 
control technique to which they apply.

[GRAPHIC] [TIFF OMITTED] TR30SE99.002


    Mercury emissions from hazardous waste combustors are controlled by 
controlling the feedrate of mercury, wet scrubbing to remove soluble 
mercury species (e.g, mercuric chloride), and carbon adsorption. We 
discuss below the operating parameter limits that apply to each control 
technique. We also discuss why we are not limiting the temperature at 
the inlet to the dry particulate matter control device as a control 
parameter for mercury.
    a. Maximum Mercury Feedrate. As proposed, you must establish and 
comply with a maximum total feedrate limit for mercury for all 
feedstreams. See 61 FR at 17428. The amount of mercury fed into the 
combustor directly affects emissions and the removal efficiency of 
emission control equipment. To establish and comply with the feedrate 
limit, you must sample and analyze and continuously monitor the 
flowrate of all feedstreams (including hazardous waste, raw materials, 
and other fuels and additives) except natural gas, process air, and 
feedstreams from vapor recovery systems for mercury content.\237\ As 
proposed, you must establish a maximum 12-hour rolling average feedrate 
limit based on operations during the comprehensive performance test as 
the average of the test run averages.
---------------------------------------------------------------------------

    \237\ See discussion in Section VII.D.3. below in the text for 
rationale for exempting these feedstreams for monitoring for mercury 
content.
---------------------------------------------------------------------------

    Rather than establish mercury feedrate limits as the levels fed 
during the comprehensive performance test, you may request as part of 
your performance test plan to use the mercury feedrates and associated 
emission rates during the performance test to extrapolate to higher 
allowable feedrate limits and emission rates. See Section VII.D.3 below 
for a discussion of the rationale and procedures for obtaining approval 
to extrapolate metal feedrates.
    In addition, you may use the performance test waiver provision 
under Sec. 63.1207(m) to document compliance with the emission 
standard. Under that provision, you must monitor the total mercury 
feedrate from all feedstreams and the gas flowrate and document that 
the maximum theoretical emission concentration does not exceed the 
mercury emission standard. Thus, this is another compliance approach 
where you would not establish feedrate limits on mercury during the 
comprehensive performance test.
    b. Wet Scrubbing. As proposed, if your combustor is equipped with a 
wet scrubber, you must establish and comply with limits on the same 
operating parameters (and in the same manner) that apply to compliance 
assurance with the hydrochloric acid/chlorine gas emission standard for 
wet scrubbers. See Section VII.D.5 below for a discussion of those 
parameters.
    c. Activated Carbon Injection. As proposed, if your combustor is 
equipped with an activated carbon injection system, you must establish 
and comply with limits on the same operating parameters (and in the 
same manner) that apply to compliance assurance with the dioxin/furan 
emission standard for activated carbon injection systems.
    d. Activated Carbon Bed. As proposed, if your combustor is equipped 
with an activated carbon bed, you must establish and comply with limits 
on the same operating parameters (and in the same manner) that apply to 
compliance assurance with the dioxin/furan emission standard for 
activated carbon beds.
    e. Consideration of a Limit on Maximum Inlet Temperature to a Dry 
Particulate Matter Control Device. The final rule does not require you 
to control inlet temperature to a dry particulate

[[Page 52944]]

matter air pollution control device to control mercury emissions. At 
proposal, we expressed concern that high inlet temperatures to a dry 
particulate matter control device could cause low mercury removal 
efficiency because mercury volatility increases with increasing 
temperature. See 61 FR at 17428. Therefore, we proposed to limit inlet 
temperatures to levels during the comprehensive performance test.
    Commenters suggest that a maximum inlet temperature for dry 
particulate matter control devices is not needed because mercury is 
generally highly volatile within the range of inlet temperatures of all 
dry particulate matter control devices. We are persuaded by the 
commenters that inlet temperature to these devices is not critically 
important to mercury control, although temperature can potentially have 
an impact on the volatility of certain mercury species (e.g., oxides). 
We conclude that the other operating parameter limits are sufficient to 
ensure compliance with the mercury emission standard. In particular, we 
note that a limit on maximum inlet temperature to these control devices 
is required for compliance assurance with the dioxin/furan, 
semivolatile metal, and low volatile metal emission standards.
3. What Are the Operating Parameter Limits for Semivolatile and Low 
Volatile Metals?
    You must maintain compliance with the semivolatile metal and low 
volatile metal emission standards by establishing and complying with 
limits on operating parameters. See Sec. 63.1209(n). The following 
table summarizes these operating parameter limits. All sources must 
comply with the limits on feedrates of semivolatile metals, low 
volatile metals, and chlorine. Other operating parameter limits apply 
depending on the type of particulate matter control device you use.

BILLING CODE 6560-50-P

[[Page 52945]]

[GRAPHIC] [TIFF OMITTED] TR30SE99.003



BILLING CODE 6560-50-C

[[Page 52946]]

    Semivolatile and low volatile metal emissions from hazardous waste 
combustors are controlled by controlling the feedrate of the metals and 
particulate matter emissions. In addition, because chlorine feedrate 
can affect the volatility of metals and thus metals levels in the 
combustion gas, and because the temperature at the inlet to the dry 
particulate matter control device can affect whether the metal is in 
the vapor (gas) or solid (particulate) phase, control of these 
parameters is also important to control emissions of these metals. We 
discuss below the operating parameter limits that apply to each control 
technique. We also discuss use of metal surrogates during performance 
testing, provisions for allowing extrapolation of performance test 
feedrate levels to calculate metal feedrate limits, and conditional 
waiver of the limit on low volatile metals in pumpable feedstreams.
    a. Good Particulate Matter Control. As proposed, you must comply 
with the operating parameter limits for particulate matter control (see 
discussion in Section VII.D.6 below and Sec. 63.1209(m)) because 
semivolatile and low volatile metals are primarily in the solid 
(particulate) phase at the gas temperature (i.e., 400 deg.F or lower) 
of the particulate matter control device. Thus, these metals are 
largely removed from flue gas as particulate matter.
    b. Maximum Inlet Temperature to Dry Particulate Matter Control 
Device. As proposed, you must establish and continuously monitor a 
limit on the maximum temperature at the inlet to a dry particulate 
matter control device. Although most semivolatile and low volatile 
metals are in the solid, particulate phase at the temperature at the 
inlet to the dry control device mandated by today's rule (i.e., 
400 deg.F or lower), some species of these metals remain in the vapor 
phase. We are requiring a limit on maximum temperature at the inlet to 
the control device to ensure that the fraction of these metals that are 
volatile (and thus not controlled by the particulate matter control 
device) does not increase during operations after the comprehensive 
performance test.
    As proposed, you must establish an hourly rolling average 
temperature limit based on operations during the comprehensive 
performance test. The hourly rolling average limit is established as 
the average of the test run averages. See Part Five, Sections VII.B.1 
and B.3 above for a discussion of the approach for calculating limits 
from comprehensive performance test data.
    Commenters suggest that this limit may conflict with the maximum 
temperature limit at the inlet to the particulate matter control device 
that is also required for compliance assurance with the dioxin/furan 
emission standard. We do not understand commenters' concern. If for 
some reason the dioxin/furan and metals emissions tests are not 
conducted simultaneously, the governing temperature limit will be the 
lower of the limits established from the separate tests. This provides 
compliance assurance for both standards.
    c. Maximum Semivolatile and Low Volatile Metals Feedrate Limits. 
You must establish limits on the maximum total feedrate of both 
semivolatile metals and low volatile metals from all feedstreams at 
levels fed during the comprehensive performance test. Metals feedrates 
are related to emissions in that, as metals feedrates increase at a 
source, metals emissions increase. See Part Four, Section II.A above 
for discussion on the relationship between metals feedrates and 
emissions. Thus, metals feedrates are an important control technique.
    For low volatile metals, you must also establish a limit on the 
maximum total feedrate of pumpable liquids from all feedstreams. The 
rule requires a separate limit for pumpable feedstreams because metals 
present in pumpable feedstreams may partition between the combustion 
gas and bottom ash (or kiln product) at a higher rate than metals in 
nonpumpable feedstreams (i.e., low volatile metals in pumpable 
feedstreams tend to partition primarily to the combustion gas). The 
rule does not require a separate limit for semivolatile metals in 
pumpable feedstreams because partitioning between the combustion gas 
and bottom ash or product for these metals does not appear to be 
affected by the physical state of the feedstream.238
---------------------------------------------------------------------------

    \238\ See USEPA., ``Technical Support Document for HWC MACT 
Standards, Volume IV: Compliance with the MACT Standards,'' February 
1998.
---------------------------------------------------------------------------

    To establish and comply with the feedrate limits, you must sample 
and analyze and continuously monitor the flowrate of all feedstreams 
(including hazardous waste, raw materials, and other fuels and 
additives) except natural gas, process air, and feedstreams from vapor 
recovery systems for semivolatile and low volatile metals content. As 
proposed, you must establish maximum 12-hour rolling average feedrate 
limits based on operations during the comprehensive performance test as 
the average of the test run averages.
    i. Use of Metal Surrogates. You may use one metal within a 
volatility group as a surrogate during comprehensive performance 
testing for other metals in that volatility group. For example, you may 
use chromium as a surrogate during the performance test for all low 
volatile metals. Similarly, you may use lead as a surrogate for 
cadmium, the other semivolatile metal. This is because the metals 
within a volatility group have generally the same volatility. Thus, 
they will generally be equally difficult to control with an emissions 
control device.
    In addition, you may use either semivolatile metal as a surrogate 
for any low volatile metal because semivolatile metals will be more 
difficult to control than low volatile metals.239 This will 
help alleviate concerns regarding the need to spike each metal during 
comprehensive performance testing. If you want to spike metals, you 
need not spike each metal to comply with today's rule but only one 
metal within a volatility group (or potentially one semivolatile metal 
for both volatility groups).
---------------------------------------------------------------------------

    \239\ This is because a greater portion of semivolatile metals 
volatilize in the combustion chamber and condenses in the flue gas 
on small particulates or as fume. The major portion of low volatile 
metals in flue gas are entrained on larger particulates (rather than 
condensing from volatile species) and are thus easier to remove with 
a particulate control device.
---------------------------------------------------------------------------

    ii. Extrapolation of Performance Test Feedrate Levels to Calculate 
Metal Feedrate Limits.240 You may request under 
Sec. 63.1209(n)(2)(ii) to use the metal feedrates and emission rates 
associated with the comprehensive performance test to extrapolate 
feedrate limits and emission rates at levels higher than demonstrated 
during the performance test. Extrapolation can be advantageous because 
it avoids much of the spiking that sources normally undertake during 
compliance testing and the associated costs, risks to operating and 
testing personnel, and environmental loading from emissions.
---------------------------------------------------------------------------

    \240\ Although this extrapolation discussion is presented in 
context of semivolatile and low volatile metal feedrates, similar 
provisions could be implemented for mercury feedrates.
---------------------------------------------------------------------------

    Under an approved extrapolation approach, you would be required to 
feed metals at no less than normal rates to narrow the amount of 
extrapolation requested. Further, we expect that some spiking would be 
desired to increase confidence in the measured, performance test 
feedrate levels that will be used to project feedrate limits (i.e., the 
errors associated with sampling and analyzing heterogeneous feedstreams 
can be minimized by spiking known quantities). Extrapolation approaches 
that request feedrate limits that are significantly higher than the 
historical range of

[[Page 52947]]

feedrates should not be approved. Extrapolated feedrate limits should 
be limited to levels within the range of the highest historical 
feedrates for the source. We are taking this policy position to avoid 
creating an incentive to burn wastes with higher than historical levels 
of metals. Metals are not destroyed by combustion but rather are 
emitted as a fraction of the amount fed to the combustor. If you want 
to burn wastes with higher than historical levels of metals, you must 
incur the costs and address the hazards to plant personnel and testing 
crews associated with spiking metals into your feedstreams during 
comprehensive performance testing.
    Although we also investigated downward interpolation (i.e., between 
the measured feedrate and emission level and zero), we are concerned 
that downward interpolation may not be conservative. Our data indicates 
that system removal efficiency can decrease as metal feedrate 
decreases. Thus, actual emissions may be higher than emissions 
projected by interpolation for lower feedrates. Consequently, we are 
not allowing downward interpolation.
    We are not specifying an extrapolation methodology to provide as 
much flexibility as possible to consider extrapolation methodologies 
that would best meet individual needs. We have investigated 
extrapolation approaches 241 and discussed in the May 1997 
NODA a statistical extrapolation methodology. Commenters raise 
concerns, however, about defining a single acceptable extrapolation 
method. They note that other methods might be developed in the future 
that prove to be better, especially for a given source. We agree that 
the approach discussed in the NODA may be too inflexible and are not 
promulgating it today.242 Consequently, today's rule does 
not specify a single method but allows you to recommend a method for 
review and approval by permitting officials.
---------------------------------------------------------------------------

    \241\ See USEPA, ``Draft Technical Support Document for HWC MACT 
Standards (NODA), Volume III: Evaluation of Metal Emissions Database 
to Investigate Extrapolation and Interpolation Issues,'' April 1997.
    \242\ We plan to develop guidance on approaches that provide 
greater flexibility.
---------------------------------------------------------------------------

    Your recommended extrapolation methodology must be included in the 
performance test plan. See Sec. 63.1207(f)(1)(x). Permitting officials 
will review the methodology considering in particular whether: (1) 
Performance test metal feedrates are appropriate (i.e., whether 
feedrates are at least at normal levels, whether some level of spiking 
would be appropriate depending on the heterogeneity of the waste, and 
whether the physical form and species of spiked material is 
appropriate); and (2) the requested, extrapolated feedrates are 
warranted considering historical metal feedrate data.
    We received comments both in favor of and in opposition to metals 
extrapolation and interpolation. Those in favor suggest extrapolation 
would simplify the comprehensive performance test procedure, reduce 
costs, and decrease emissions during testing. Those in opposition are 
concerned about: (1) Whether there is a predictable relationship 
between feedrates and emission rates; (2) the possibility of higher 
overall metals loading to the environment over the life of the facility 
(i.e., because higher feedrate limits would be relatively easy to 
obtain); (3) the difficulty in defining a ``normal'' feedrate for 
facilities with variable metal feeds; and (4) whether all conditions 
influencing potential metals emissions, such as combustion temperature 
and metal compound speciation, could be adequately considered.
    Given the pros and cons associated with various extrapolation 
methodologies and policies, we are still concerned that sources would 
be able to: (1) Feed metals at higher rates without a specific 
compliance demonstration of the associated metals emissions; and (2) 
obtain approval to feed metals at higher levels than normal, even 
though all combustion sources should be trying to minimize metals 
feedrates. However, because the alternative is metal spiking (as 
evidenced in facility testing for BIF compliance) and metal spiking is 
a significant concern as well, we find that the balance is better 
struck by allowing, with site-specific review and where warranted 
approval, extrapolation as a means to reduce unnecessary emissions, 
reduce unnecessary costs incurred by facilities, and better protect the 
health of testing personnel during performance tests.
    iii. Conditional Waiver of Limit on Low Volatile Metals in Pumpable 
Feedstreams. Commenters indicate that they may want to base feedrate 
limits only on the worst-case feedstream--pumpable hazardous waste. The 
feedrate limit would be based only on the feedrate of the pumpable 
hazardous waste during the comprehensive performance test, even though 
nonpumpable feedstreams would be contributing some metals to emissions. 
In this situation, commenters suggest that separate feedrate limits for 
total and pumpable feedstreams would not be needed. We agree that if 
you define the total feedstream feedrate limit as the pumpable 
feedstream feedrate during the performance test, dual limits are not 
required. The feedrate of metals in total feedstreams must be monitored 
and shown to be below the pumpable feedstream-based limit. See 
Sec. 63.1209(n)(2)(C).
    iv. Response to other Comments. We discuss below our response to 
several other comments: (1) Recommendation for national uniform 
feedrate limits; (2) concerns that feedstream monitoring is 
problematic; and (3) recommendations that monitoring natural gas and 
vapor recovery system feedstreams is unnecessary.
    A commenter states that nationally uniform feedrate limits are 
needed for metals and chlorine and that any other approach would be 
inconsistent with the CAA. The commenter stated that hazardous waste 
combustion device operators should not be allowed to self-select any 
level of toxic metal feedrate just because they can show compliance 
with the MACT standard. We believe that standards prescribing national 
feedrate limits on metals or chlorine are not necessary to ensure MACT 
control of metals and hydrochloric acid/chlorine gas and may be overly 
restrictive. Emissions of metals and hydrochloric acid/chlorine gas are 
controlled by controlling the feedrate of metals and chlorine, and 
emission control devices. In developing MACT standards for a source 
category, if we can identify emission levels that are being achieved by 
the best performing sources using MACT control, we generally establish 
the MACT standard as an emission level rather than prescribed operating 
limits (e.g., feedrate limits). This approach is preferable because it 
gives the source the option of determining the most cost-effective 
measures to comply with the standard. Some sources may elect to comply 
with the emission standards using primarily feedrate control, while 
others may elect to rely primarily on emission controls. Under either 
approach, the emission levels are equivalent to those being achieved by 
the best performing existing sources. Other factors that we considered 
in determining to express the standards as an emission level rather 
than feedrate limits include: (1) There is not a single, universal 
correlation factor between feedrate and metal emissions to use to 
determine a national feedrate that would be equivalent to the emission 
levels achieved by the best performing sources; (2) emission standards 
communicate better to the public that meaningful controls are being 
applied because the hazardous waste combustor

[[Page 52948]]

emission standards can be compared to standards for other waste 
combustors (e.g., municipal and medical waste combustors) and 
combustion devices; and (3) CEMS, the ultimate compliance assurance 
tool that we encourage sources to use,243 are incompatible 
with standards expressed as feedrate limits.
---------------------------------------------------------------------------

    \243\ As discussed previously in the text, feedrate limits as a 
compliance tool can be problematic for difficult to sample or 
analyze feedstreams. Further, the emissions resulting from a given 
feedrate level may increase (or decrease) over time, providing 
uncertainty about actual emissions.
---------------------------------------------------------------------------

    Another commenter is concerned that feedrate monitoring of highly 
heterogeneous waste streams is problematic and analytical turnaround 
times can be rather long. The commenter suggests that alternatives 
beyond feedstream monitoring (such as predictive emissions monitoring) 
should be allowed. Although we acknowledge that there may be 
difficulties in monitoring the feedrate of metals or chlorine in 
certain waste streams, there generally is no better way to assure 
compliance with these standards other than using CEMS. Predictive 
modeling appears to introduce unnecessarily some greater compliance 
uncertainty than feedstream testing. Thus, we conclude that feedstream 
monitoring is a necessary monitoring tool if a multimetals CEMS is not 
used. (We also note that feedstream monitoring under MACT will not be 
substantially more burdensome or problematic than the requirements now 
in place under RCRA regulations.)
    In addition, another commenter suggests that sources should not 
have to monitor metals and chlorine in natural gas feedstreams because 
it is impractical and levels are low and unvarying. The commenter 
suggests that sources should be allowed to use characterization data 
from natural gas vendors. We agree that the cost and possible hazards 
of monitoring natural gas for metals and chlorine is not warranted 
because our data shows metals are not present at levels of concern. 
Therefore, you are not required to monitor metals and chlorine levels 
in natural gas feedstreams. However, you must document in the 
comprehensive performance test plan the expected levels of these 
constituents and account for the expected levels in documenting 
compliance with feedrate limits (e.g., by assuming worst-case 
concentrations and monitoring the natural gas flowrate). See 
Sec. 63.1209(c)(5).
    Finally, some commenters are concerned that feedstreams from vapor 
recovery systems (e.g., waste fuel tank and container emissions) are 
difficult, costly, and often dangerous to monitor frequently for metals 
and chlorine levels. Particularly because of some of the safety issues 
concerned, the rule does not require continuous monitoring of metals 
and chlorine for feedstreams from vapor recovery systems. However, as 
is the case for natural gas, you must document in the comprehensive 
performance test plan the expected levels of these constituents and 
account for the expected levels in documenting compliance with feedrate 
limits.
    d. Maximum Chlorine Feedrate. As proposed, you must establish a 
limit on the maximum feedrate for total chlorine (both organic and 
inorganic) in all feedstreams based on the level fed during the 
comprehensive performance test. A limit on maximum chlorine feedrate is 
necessary because most metals are more volatile in the chlorinated 
form. Thus, for example, more low volatile metals may report to the 
combustion gas as a vapor than would be otherwise be entrained in the 
combustion gas absent the presence of chlorine. In addition, the vapor 
form of the metal is more difficult to control. Although most 
semivolatile and low volatile metal species are in the particulate 
phase at gas temperatures at the inlet to the particulate matter 
control device, semivolatile metals that condense from the vapor phase 
partition to smaller particulates and are more difficult to control 
than low volatile metals that are emitted in the form of entrained, 
larger particulates.
    To establish and comply with the feedrate limit, you must sample 
and analyze, and continuously monitor the flowrate, of all feedstreams 
(including hazardous waste, raw materials, and other fuels and 
additives) except natural gas, process air, and feedstreams from vapor 
recovery systems for total chlorine content. As proposed, you must 
establish a maximum 12-hour rolling average feedrate limit based on 
operations during the comprehensive performance test as the average of 
the test run averages.
    Commenters suggest that chlorine feedrate limits are not needed for 
sources with semivolatile and low volatile metal feedrates, when 
expressed as maximum theoretical emission concentrations, less than the 
emission standard. We agree. In this situation, you would be eligible 
for the waiver of performance test under Sec. 63.1207(m). The 
requirements of that provision (e.g., monitor and record metals 
feedrates and gas flowrates to ensure that metals feedrate, expressed 
as a maximum theoretical emission concentration, does not exceed the 
emission standard) apply in lieu of the operating parameter limits 
based on performance testing discussed above. We note, however, that 
you would still need to establish a maximum feedrate limit for total 
chlorine as an operating parameter limit for the hydrochloric acid/
chlorine gas emission standard (discussed below), unless you also 
qualified for a waiver of that emission standard under Sec. 63.1207(m).
4. What Are the Monitoring Requirements for Carbon Monoxide and 
Hydrocarbon?
    You must maintain compliance with the carbon monoxide and 
hydrocarbon emission standards using continuous emissions monitoring 
systems (CEMS). In addition, you must use an oxygen CEMS to correct 
continuously the carbon monoxide and hydrocarbon levels recorded by 
their CEMS to 7 percent oxygen.
    As proposed, the averaging period for carbon monoxide and 
hydrocarbon CEMS is a one-hour rolling average updated each minute. 
This is consistent with current RCRA requirements and commenters did 
not recommend an alternative averaging period.
    We also are promulgating performance specifications for carbon 
monoxide, hydrocarbon, and oxygen CEMS. The carbon monoxide and oxygen 
CEMS performance specifications are codified as Performance 
Specification 4B in appendix B, part 60. This performance specification 
is the same as the specification currently used for BIFs in appendix 
IX, part 266. It also is very similar to existing appendix B, part 60 
Performance Specifications 3 (for oxygen) and 4A (for carbon monoxide). 
New specification 4B references many of the provisions of 
Specifications 3 and 4A.
    The hydrocarbon CEMS performance specification is codified as 
Performance Specification 8A in appendix B, part 60. This specification 
is also identical to the specification currently used for BIFs in 
section 2.2 of appendix IX, part 266, with one exception. We deleted 
the quality assurance section and placed it in the appendix to subpart 
EEE of part 63 promulgated today to be consistent with our approach to 
part 60 performance specifications.
    We discuss below several issues pertaining to monitoring with these 
CEMS: (1) The requirement to establish site-specific alternative span 
values in some situations; (2) consequences of exceeding the span value 
of the CEMS; and (3) the need to adjust the oxygen correction factor 
during startup and shutdown.
    a. When Are You Required to Establish Site-Specific Alternative 
Span

[[Page 52949]]

Values? As proposed, if you normally operate at an oxygen correction 
factor of more than 2 (e.g., a cement kiln monitoring carbon monoxide 
in the by-pass duct), you must use a carbon monoxide or hydrocarbon 
CEMS with a span proportionately lower than the values prescribed in 
the performance specifications relative to the oxygen correction factor 
at the CEMS sampling point. See the appendix to Subpart EEE, part 63: 
Quality Assurance Procedures for Continuous Emissions Monitors Used for 
Hazardous Waste Combustors.
    This requirement arose from our experience with implementing the 
BIF rule when we determined that the prescribed span values for the 
carbon monoxide and hydrocarbon CEMS may lead to high error in 
corrected emission values due to the effects of making the oxygen 
correction. For example, a cement kiln may analyze for carbon monoxide 
emissions in the by-pass duct with oxygen correction factors on the 
order of 10. At the low range of the carbon monoxide CEMS span--200 ppm 
as prescribed by Performance Specification 4B--with an acceptable 
calibration drift of three percent, an error of 6 ppm is the result. 
Accounting for the oxygen correction factor of 10, however, drives the 
error in the measurement due to calibration drift up to 60 ppm. This is 
more than half the carbon monoxide emission standard of 100 ppm and is 
not acceptable. At carbon monoxide readings close to the 100 ppm 
standard, true carbon monoxide levels may be well above or well below 
the standard.
    Consider the same example under today's requirement. For an oxygen 
correction factor of 10, the low range span for the carbon monoxide 
CEMS must be 200 divided by 10, or 20 ppm. The allowable calibration 
drift of three percent of the span allows an error of 0.6 ppm at 20 
ppm. Applying an oxygen correction factor of 10 results in an absolute 
calibration drift error of 6ppm at an oxygen-corrected carbon monoxide 
reading of 200.
    b. What Are the Consequences of Exceeding the Span Value for Carbon 
Monoxide and Hydrocarbon CEMS? If you do not elect to use a carbon 
monoxide CEMS with a higher span value of 10,000 ppmv and a hydrocarbon 
CEMS with a higher span value of 500 ppmv, you must configure your CEMS 
so that a one-minute carbon monoxide value reported as 3,000 ppmv or 
greater must be recorded (and used to calculate the hourly rolling 
average) as 10,000 ppmv, and a one-minute hydrocarbon value reported as 
200 ppmv or greater must be recorded as 500 ppmv.
    If you elect to use a carbon monoxide CEMS with a span range of 0-
10,000 ppmv, you must use one or more carbon monoxide CEMS that meet 
the Performance Specification 4B for three ranges: 0-200 ppmv; 1-3,000 
ppmv; and 0-10,000 ppmv. Specification 4B provides requirements for the 
first two ranges. For the (optional) high range of 0-10,000 ppmv, the 
CEMS must also comply with Performance Specification 4B, except that 
the calibration drift must be less than 300 ppmv and calibration error 
must be less than 500 ppmv. These values are based on the allowable 
drift and error, expressed as a percentage of span, that the 
specification requires for the two lower span levels.
    If you elect to use a hydrocarbon CEMS with a span range of 0-500 
ppmv, you must use one or more hydrocarbon CEMS that meet Performance 
Specification 8A for two ranges: 0-100 ppmv, and 0-500 ppmv. 
Specification 8A provides requirements for the first range. For the 
(optional) high range of 0-500 ppmv, the CEMS must also comply with 
Performance Specification 8A, except: (1) The zero and high-level daily 
calibration gas must be between 0 and 100 ppmv and between 250 and 450 
ppmv, respectively; (2) the strip chart recorder, computer, or digital 
recorder must be capable of recording all readings within the CEMS 
measurement range and must have a resolution of 2.5 ppmv; (3) the CEMS 
calibration must not differ by more than 15 ppmv after each 
24 hour period of the seven day test at both zero and high levels; (4) 
the calibration error must be no greater than 25 ppmv; and (5) the zero 
level, mid-level, and high level values used to determine calibration 
error must be in the range of 0-200 ppmv, 150-200 ppmv, and 350-400 
ppmv, respectively. These requirements for the optional high range (0-
500 ppmv) are derived proportionately from the requirements in 
Specification 8A for the lower range (0-100 ppmv).
    The rule provides this requirement because we are concerned that, 
when carbon monoxide and hydrocarbon monitors record a one-minute value 
at the upper span level, the actual level of carbon monoxide or 
hydrocarbons may be much higher (i.e., these CEMS often ``peg-out'' at 
the upper span level). This has two inappropriate consequences. First, 
the source may actually be exceeding the carbon monoxide or hydrocarbon 
standard even though the CEMS indicates that it is not. Second, if the 
carbon monoxide or hydrocarbon hourly rolling average were to exceed 
the standard, triggering an automatic waste feed cutoff, the emission 
level may drop back below the standard much sooner than it otherwise 
would if the actual one-minute average emission levels were recorded 
(i.e., rather than one-minute averages pegged at the upper span value). 
Thus, this diminishes the economic disincentive for incurring automatic 
waste feed cutoffs of not being able to restart the hazardous waste 
feed until carbon monoxide and hydrocarbon levels are below the 
standard.
    We considered applying these ``out-of-span'' requirements when any 
recorded value (i.e., any value recorded by the CEMS on a frequency of 
at least every 15 seconds), rather than one-minute average values, 
exceeded the upper span level. Commenters point out, however, that CEMS 
may experience short-term electronic glitches that cause the monitored 
output to spike for a very short time period. We concur, and conclude 
that we should be concerned only about one-minute average values 
because these short-term electronic glitches (that are not caused by 
emission excursions) could result in an undesirable increase in 
automatic waste feed cutoffs.
    You may prefer to use carbon monoxide or hydrocarbon CEMS that have 
upper span values between 3,000 and 10,000 ppmv and between 100 and 500 
ppmv, respectively. If you believe that you would not have one-minute 
average carbon monoxide or hydrocarbon levels as high as 10,000 ppmv 
and 500 ppmv, respectively, you may determine that it would be less 
expensive to use monitors with lower upper span levels (e.g., you may 
be able to use a single carbon monoxide CEMS to meet performance 
specifications for all three spans--the two lower spans required by 
Specification 4B, and a higher span (but less than 10,000)). You must 
still record, however, any one-minute average carbon monoxide or 
hydrocarbon levels that are at or above the span as 10,000 ppmv and 500 
ppmv, respectively.
    c. How Is the Oxygen Correction Factor Adjusted during Startup and 
Shutdown? You must identify in your Startup Shutdown, and Malfunction 
Plan a projected oxygen correction factor to use during periods of 
startup and shutdown. The projected oxygen correction factor should be 
based on normal operations. See Sec. 63.1206(c)(2)(iii). The rule 
provides this requirement because the oxygen concentration in the 
combustor can exceed 15% during startup and shutdown, causing the 
correction factor to increase exponentially from the normal value. Such 
large correction factors result in corrected carbon

[[Page 52950]]

monoxide and hydrocarbon levels that are inappropriately inflated.
5. What Are the Operating Parameter Limits for Hydrochloric Acid/
Chlorine Gas?
    You must maintain compliance with the hydrochloric acid/chlorine 
gas emission standard by establishing and complying with limits on 
operating parameters. See Sec. 63.1209(o). The following table 
summarizes these operating parameter limits. All sources must comply 
with the maximum chlorine feedrate limit. Other operating parameter 
limits apply depending on the type of hydrochloric acid/chlorine gas 
emission control device you use.

BILLING CODE 6560-50-P

[[Page 52951]]

[GRAPHIC] [TIFF OMITTED] TR30SE99.004



BILLING CODE 6560-50-C

[[Page 52952]]

    Hydrochloric acid/chlorine gas emissions from hazardous waste 
combustors are controlled by controlling the feedrate of total chlorine 
(organic and inorganic) and either wet or dry scrubbers. We discuss 
below the operating parameter limits that apply to each control 
technique.
    a. Maximum Chlorine Feedrate Limit. As proposed, you must establish 
a limit on the maximum feedrate of chlorine, both organic and 
inorganic, from all feedstreams based on levels fed during the 
comprehensive performance test. Chlorine feedrate is an important 
emission control technique because the amount of chlorine fed into a 
combustor directly affects emissions of hydrochloric acid/chlorine gas. 
To establish and comply with the feedrate limit, you must sample and 
analyze, and continuously monitor the flowrate, of all feedstreams 
(including hazardous waste, raw materials, and other fuels and 
additives) except natural gas, process air, and feedstreams from vapor 
recovery systems for chlorine content.244 Also as proposed, 
you must establish a maximum 12-hour rolling average feedrate limit 
based on operations during the comprehensive performance test as the 
average of the test run averages.
---------------------------------------------------------------------------

    \244\ See discussion in Section VII.D.3 above in the text for 
the rationale for exempting these feedstreams for monitoring for 
chlorine content.
---------------------------------------------------------------------------

    One commenter states that a chlorine feedrate is not necessary for 
cement kilns because cement kilns have an inherent incentive to control 
chlorine feedrates: to avoid operational problems such as the formation 
of material rings in the kiln or alkali-chloride condensation on the 
walls. Although we understand that cement kilns must monitor chlorine 
feedrates for operational reasons, several cement kilns in our data 
base emit levels of hydrochloric acid/chlorine gas at levels above 
today's emissions standard. We conclude, therefore, that the 
operational incentive to limit chlorine feedrates is not adequate to 
ensure compliance with the hydrochloric acid/chlorine gas emission 
standard.
    b. Wet Scrubbers. If your combustor is equipped with a wet 
scrubber, you must establish, continuously monitor, and comply with 
limits on the following operating parameters:
    i. Maximum Flue Gas Flowrate or Kiln Production Rate. As proposed, 
you must establish a limit on maximum flue gas flowrate or kiln 
production rate as a surrogate. See 61 FR at 17433. Gas flowrate is a 
key parameter affecting the control efficiency of a wet scrubber (and 
any emissions control device). As gas flowrate increases, control 
efficiency generally decreases unless other operating parameters are 
adjusted to accommodate the increased flowrate. Cement kilns and 
lightweight aggregate kilns may establish a limit on maximum production 
rate (e.g., raw material feedrate or clinker or aggregate production 
rate) in lieu of a maximum gas flowrate given that production rate 
directly relates to flue gas flowrate.
    As proposed, you must establish a maximum gas flowrate or 
production rate limit as the average of the maximum hourly rolling 
averages for each run of the comprehensive performance test.
    We did not receive adverse comment on this compliance parameter.
    ii. Minimum Pressure Drop Across the Scrubber. You must establish a 
limit on minimum pressure drop across the scrubber. If your combustor 
is equipped with a high energy scrubber (e.g., venturi, calvert), you 
must establish an hourly rolling average limits based on operations 
during the comprehensive performance test. The hourly rolling average 
is established as the average of the test run averages.
    If your combustor is equipped with a low energy scrubber (e.g., 
spray tower), you must establish a limit on minimum pressure drop based 
on the manufacturer's specification. You must comply with the limit on 
an hourly rolling average basis.
    Pressure drop across a wet scrubber is an important operating 
parameter because it is an indicator of good mixing of the two fluids, 
the scrubber liquid and the flue gas. A low pressure drop indicates 
poor mixing and, hence, poor efficiency. A high pressure drop indicates 
good removal efficiency.
    One commenter states that wet scrubber pressure drop is not an 
important parameter for packed-bed, low energy wet scrubbers. The 
commenter states that the performance of a packed-bed scrubber is based 
on good liquid-to-gas contacting. Thus, performance is dependent on 
packing design and scrubber fluid flow. In addition, the commenter 
states that scrubber liquid flow rate (and recirculation rate and make-
up water flow rate) are adequate for assuring proper scrubber 
operation. We note that for many types of low energy wet scrubbers, 
pressure drop can be a rough indicator of scrubber liquid and flue gas 
contacting. Thus, although it is not a critical parameter, the minimum 
pressure drop of a low energy scrubber should still be monitored and 
complied with on a continuous basis.
    Because pressure drop for a low energy scrubber (e.g., spray 
towers, packed beds, or tray towers) is not as important as for a high 
energy scrubber to maintain performance, however, the rule requires you 
to establish a limit on the minimum pressure drop for a low energy 
scrubber based on manufacturer specifications, rather than levels 
demonstrated during compliance testing. You must comply with this limit 
on an hourly rolling average basis. The pressure drop for high energy 
wet scrubbers, such as venturi or calvert scrubbers, however, is a key 
operating parameter to ensure the scrubber maintains performance. 
Accordingly, you must base the minimum pressure drop for these devices 
on levels achieved during the comprehensive test, and you must 
establish an hourly rolling average limit.
    iii. Minimum Liquid Feed Pressure. You must establish a limit on 
minimum liquid feed pressure to a low energy scrubber. The limit must 
be based on manufacturer's specifications and you must comply with it 
on an hourly rolling average basis.
    The rule requires a limit on liquid feed pressure because the 
removal efficiency of a low energy wet scrubber can be directly 
affected by the atomization efficiency of the scrubber. A drop in 
liquid feed pressure may be an indicator of poor atomization and poor 
scrubber removal efficiency. We are not requiring a limit on minimum 
liquid feed pressure for high energy scrubbers because liquid flow rate 
rather than feed pressure is the dominant operating parameter for high 
energy scrubbers.
    We acknowledge, however, that not all wet scrubbers rely on 
atomization efficiency to maintain performance. If manufacturer's 
specifications indicate that atomization efficiency is not an important 
parameter that controls the efficiency of your scrubber, you may 
petition permitting officials under Sec. 63.1209(g)(1) to waive this 
operating parameter limit.
    iv. Minimum Liquid pH. You must establish dual ten-minute and 
hourly rolling average limits on minimum pH of the scrubber water based 
on operations during the comprehensive performance test. The hourly 
rolling average is established as the average of the test run averages.
    The pH of the scrubber liquid is an important operating parameter 
because, at low pH, the scrubber solution is more acidic and removal 
efficiency of hydrochloric acid and chlorine gas decreases.
    These requirements, except for the proposed ten-minute averaging 
period, are the same as we proposed. See 61 FR at 17433. We did not 
receive adverse comments.

[[Page 52953]]

    v. Minimum Scrubber Liquid Flowrate or Minimum Liquid/Gas Ratio. 
You must establish an hourly rolling average limits on either minimum 
scrubber liquid flowrate and maximum flue gas flowrate or minimum 
liquid/gas ratio based on operations during the comprehensive 
performance test. The hourly rolling average is established as the 
average of the test run averages.
    Liquid flowrate and flue gas flowrate or liquid/gas ratio are 
important operating parameters because a high liquid-to-gas-flowrate 
ratio is indicative of good removal efficiency.
    We had proposed to limit the liquid-to-gas ratio only. Commenters 
suggest that a limit on liquid-to-gas flow ratio would not be needed if 
the liquid flowrate and flue gas flowrate were limited instead. They 
reason that, because gas flowrate is already limited, limiting liquid 
flowrate as well would ensure that the liquid-to-gas ratio is 
maintained. We agree. During normal operations, the liquid flowrate can 
only be higher than levels during the performance test, and gas 
flowrate can only be lower than during the performance test. Thus, the 
numerator in the liquid flowrate/gas flowrate ratio could only be 
larger, and the denominator could only be smaller. Consequently, the 
liquid flowrate/gas flowrate during normal operations will always be 
higher than during the comprehensive performance test. Consequently, we 
agree that a limit on liquid-to-gas-ratio is not needed if you 
establish a limit on liquid flowrate and flue gas flowrate. 
Establishing limits on these parameters is adequate to ensure that the 
liquid flowrate/gas ratio is maintained.245
---------------------------------------------------------------------------

    \245\ In fact, complying with limits on liquid flowrate and gas 
flowrate, rather than complying with a liquid flowrate/gas flowrate 
ratio, is a more conservative approach to ensure that the 
performance test ratio is maintained (at a minimum). Thus, we prefer 
that you establish a limit on liquid flowrate (in conjunction with 
the limit gas flowrate) in lieu of a limit on the ratio.
---------------------------------------------------------------------------

    c. Dry Scrubbers. A dry scrubber removes hydrochloric acid from the 
flue gas by adsorbing the hydrochloric acid onto sorbent, normally an 
alkaline substance like limestone. As proposed, if your combustor is 
equipped with a dry scrubber, you must establish, continuously monitor, 
and comply with limits on the following operating parameters: Gas 
flowrate or kiln production rate; sorbent feedrate; carrier fluid 
flowrate or nozzle pressure drop; and sorbent specifications. See 61 FR 
at 17434.
    i. Maximum Flue Gas Flowrate or Kiln Production Rate. As proposed, 
you must establish a limit on maximum flue gas flowrate or kiln 
production rate as a surrogate. The limit is established and monitored 
as discussed above for wet scrubbers.
    ii. Minimum Sorbent Feedrate. You must establish an hourly rolling 
average limit on minimum sorbent feedrate based on feedrate levels 
during the comprehensive performance test. The hourly rolling average 
is established as the average of the test run averages.
    Sorbent feedrate is important because, as more sorbent is fed into 
the dry scrubber, removal efficiency of hydrochloric acid and chlorine 
gas increases.246 Conversely, lower sorbent feedrates tend 
to cause removal efficiency to decrease.
---------------------------------------------------------------------------

    \246\ We note that sorbent should be fed to a dry scrubber in 
excess of the stoichiometric requirements for neutralizing the anion 
component in the flue gas. Lower levels of sorbent, even above 
stoichiometric requirements, would limit the removal of acid gasses.
---------------------------------------------------------------------------

    At proposal, we invited comment on whether a ten-minute rolling 
average is appropriate for sorbent feedrate (61 FR at 17434). We were 
concerned that some facilities may not automate their dry scrubbers to 
add sorbent solutions but instead add batches of virgin sorbent 
solution. Thus, we were concerned that a ten-minute rolling average may 
not be practicable in all cases. Some commenters are concerned that a 
ten-minute limit would be difficult to measure, especially in the case 
of batch addition of sorbent. Nonetheless, we have determined upon 
reanalysis that sorbent is not injected into the flue gas in 
``batches.'' Although sorbent may be added in batches to storage or 
mixing vessels, it must be injected into the flue gas continuously to 
provide continuous and effective removal of acid gases. Thus, ten-
minute rolling average limits would be practicable and appropriate for 
sorbent injection feedrates if ten-minute averages were required in 
this final rule.247 However, as discussed in Part Five, 
Section VII.B, we have decided to not require ten-minute averaging 
periods on a national basis. Permitting officials may, however, 
determine that shorter averaging periods are needed to better assure 
compliance with the emission standard.
---------------------------------------------------------------------------

    \247\ We note that flowrate measurement devices are available 
for ten-minute average times (e.g., those based on volumetric screw 
feeders which provide instantaneous measurements).
---------------------------------------------------------------------------

    iii. Minimum Carrier Fluid Flowrate or Nozzle Pressure Drop. A 
carrier fluid, normally air or water, is necessary to transport and 
inject the sorbent into the gas stream. As proposed, you must establish 
and continuously monitor a limit on either minimum carrier gas or water 
flowrate or pressure drop across the nozzle to ensure that the flow and 
dispersion of the injected sorbent into the flue gas stream is 
maintained. You must base the limit on manufacturer's specifications, 
and comply with the limit on a one-hour rolling average basis.
    Without proper carrier flow to the dry scrubber, the sorbent flow 
into the scrubber will decrease causing the efficiency to decrease. 
Nozzle pressure drop is also an indicator of carrier gas flow into the 
scrubber. At higher pressure drops, more sorbent is carried to the dry 
scrubber.
    iv. Identification of Sorbent Brand and Type or Adsorption 
Properties. You must either identify the sorbent brand and type used 
during the comprehensive performance test and continue using that 
sorbent, or identify the adsorption properties of that sorbent and use 
a sorbent having equivalent or better properties. This will ensure that 
the sorbent's adsorption properties are maintained.
    We proposed to require sources to continue to use the same sorbent 
brand and type as they used during the comprehensive performance test 
or obtain a waiver from this requirement from the Administrator. See 61 
FR at 17434. As discussed above in the context of specifying the brand 
of carbon used in carbon injection systems to control dioxin/furan, we 
have determined that sources should have the option of using 
manufacturer's specifications to specify the sorption properties of the 
sorbent used during the comprehensive performance test. You may use 
sorbent of other brands or types provided that it has equivalent or 
better sorption properties. You must include in the operating record 
written documentation that the substitute sorbent will provide the same 
level of control as the original sorbent.
6. What Are the Operating Parameter Limits for Particulate Matter?
    You must maintain compliance with the particulate matter emission 
standard by establishing and complying with limits on operating 
parameters. See Sec. 63.1209(m). The following table summarizes these 
operating parameter limits. All incinerators must comply with the limit 
on maximum ash feedrate. Other operating parameter limits apply 
depending on the type of particulate matter control device you use.

BILLING CODE 6560-50-P

[[Page 52954]]

[GRAPHIC] [TIFF OMITTED] TR30SE99.005



BILLING CODE 6560-50-C

[[Page 52955]]

    Particulate matter emissions from hazardous waste combustors are 
controlled by controlling the feedrate of ash to incinerators and using 
a particulate matter control device. We discuss below the operating 
parameter limits that apply to each control technique.
    a. Maximum Ash Feedrate. As proposed, if you own or operate an 
incinerator, you must establish a limit on the maximum feedrate of ash 
from all feedstreams based on the levels fed during the comprehensive 
performance test. To establish and comply with the feedrate limit, you 
must sample and analyze, and continuously monitor the flowrate of all 
feedstreams (including hazardous waste, and other fuels and additives) 
except natural gas, process air, and feedstreams from vapor recovery 
systems for ash content.248 Also as proposed, you must 
establish a maximum 12-hour rolling average feedrate limit based on 
operations during the comprehensive performance test as the average of 
the test run averages. See 61 FR at 17438.
---------------------------------------------------------------------------

    \248\ See discussion in Section VII.D.3 above in the text for 
the rationale for exempting these feedstreams from monitoring for 
ash content.
---------------------------------------------------------------------------

    Ash feedrate for incinerators is an important particulate matter 
control parameter because ash feedrates can relate directly to 
emissions of particulate matter (i.e., ash contributes to particulate 
matter in flue gas). We are not requiring an ash feedrate limit for 
cement or lightweight aggregate kilns because particulate matter from 
those combustors is dominated by raw materials entrained in the flue 
gas. The contribution to particulate matter of ash from hazardous waste 
or other feedstreams is not significant. We discussed this issue at 
proposal.
    A commenter states that ash feedrate limits are not needed for 
combustors using fabric filters, suggesting that fabric filter pressure 
drop and opacity monitoring are sufficient for compliance assurance. We 
discuss previously in this section (i.e., Part Five, Section VII) our 
concern that neither opacity monitors, nor limits on control device 
operating parameter, nor limits on the feedrates of constituents that 
can contribute directly to emissions of hazardous air pollutants 
comprise an ideal compliance assurance regime. We would prefer the use 
of a particulate matter CEMS for compliance assurance but cannot 
achieve that goal at this time. Absent the use of a CEMS and given the 
limitations of the individual compliance tools currently available, we 
are reluctant to forgo on a national, generic basis requiring limits on 
an operating parameter such as ash feedrate that we know can relate 
directly to particulate emissions. However, you may petition permitting 
officials under Sec. 63.1209(g)(1) for approval to waive the ash 
feedrate limit based on data or information documenting that pressure 
drop across the fabric filter coupled with an opacity monitor would 
provide equivalent or better compliance assurance than a limit on ash 
feedrate.
    b. Wet Scrubbers. As proposed, if your combustor is equipped with a 
wet scrubber, you must establish, continuously monitor, and comply with 
limits on the operating parameters discussed below. High energy wet 
scrubbers (e.g., venturi, calvert) remove particulate matter by 
capturing particles in liquid droplets and separating the droplets from 
the gas stream. Ionizing wet scrubbers use both an electrical charge 
and wet scrubbing to remove particulate matter. Low energy wet 
scrubbers that are not ionizing wet scrubbers (e.g., packed bed, spray 
tower) are only subject to the scrubber water solids content operating 
parameter requirements for particulate matter control because they are 
primarily used to control emissions of acid gases and only provide 
incidental particulate matter control.
    i. Maximum Flue Gas Flowrate or Kiln Production Rate. For high 
energy and ionic wet scrubbers, you must establish a limit on maximum 
flue gas flowrate or kiln production rate as a surrogate. See 61 FR at 
17438. Gas flowrate is a key parameter affecting the control efficiency 
of a wet scrubber (and any emissions control device). As gas flowrate 
increases, control efficiency generally decreases unless other 
operating parameters are adjusted to accommodate the increased 
flowrate. Cement kilns and lightweight aggregate kilns may establish a 
limit on maximum production rate (e.g., raw material feedrate or 
clinker or aggregate production rate) in lieu of a maximum gas flowrate 
given that production rate directly relates to flue gas flowrate.
    As proposed, you must establish a maximum gas flowrate or 
production rate limit as the average of the maximum hourly rolling 
averages for each run of the comprehensive performance test.
    ii. Minimum Pressure Drop Across the Scrubber. For high energy 
scrubbers only, you must establish an hourly rolling average limits on 
minimum pressure drop across the scrubber based on operations during 
the comprehensive performance test. The hourly rolling average is 
established as the average of the test run averages. See the discussion 
in Section VII.D.5.b above for a discussion on the approach for 
calculating limits from comprehensive performance test data.
    iii. Minimum Scrubber Liquid Flowrate or Minimum Liquid/Gas Ratio. 
For high energy wet scrubbers, you must establish an hourly rolling 
average limits on either minimum scrubber liquid flowrate and maximum 
flue gas flowrate or minimum liquid/gas ratio based on operations 
during the comprehensive performance test. The hourly rolling average 
is established as the average of the test run averages. See the 
discussion in Section VII.D.5.b above for a discussion on the approach 
for calculating limits from comprehensive performance test data.
    iv. Maximum Solids Content of Scrubber Water or Minimum Blowdown 
Rate Plus Minimum Scrubber Tank Volume or Level. For all wet scrubbers, 
to maintain the solids content of the scrubber water to levels no 
higher than during the comprehensive performance test, you must 
establish a limit on either: (1) Maximum solids content of the scrubber 
water; or (2) minimum blowdown rate plus minimum scrubber tank volume 
or level. If you elect to establish a limit on maximum solids content 
of the scrubber water, you must comply with the limit either by: (1) 
Continuously monitoring the solids content and establishing 12-hour 
rolling average limits based on solids content during the comprehensive 
performance test; or (2) periodic manual sampling and analysis of 
scrubber water for solids content. Under option 1, the 12-hour rolling 
average is established as the average of the test run averages. Under 
option 2, you must either comply with a default sampling and analysis 
frequency for scrubber water solids content of once per hour or 
recommend an alternative frequency in your comprehensive performance 
test plan that you submit for review and approval.
    Solids content in the scrubber water is an important operating 
parameter because as the solids content increases, particulate 
emissions increase. This is attributable to evaporation of scrubber 
water and release of previously captured particulate back into the flue 
gas. Blowdown is the amount of scrubber liquid removed from the process 
and not recycled back into the wet scrubber. As scrubber liquid is 
removed and not recycled, solids are removed. Thus, blowdown is an 
operating parameter that affects solids content and can be used as a 
surrogate for measuring solids content directly. See 61 FR 17438.
    The proposed rule would have required continuously monitored limits 
on either minimum blowdown or a

[[Page 52956]]

maximum solids content. In response to comments and upon reanalysis of 
the issues, we conclude that we need to make two revisions to these 
requirements. First, we are concerned that it may be problematic to 
continuously monitor the solids content of scrubber water. 
Consequently, we revised the requirements to allow manual sampling and 
analysis on an hourly basis, unless you justify an alternative 
frequency. Second, we are concerned that a limit on blowdown rate 
without an associated limit on either minimum scrubber water tank 
volume or level would not be adequate to provide control of solids 
content. The solids concentration in blowdown tanks could be higher at 
lower water levels. Therefore, water levels need to be at least 
equivalent to the levels during the comprehensive performance test. 
This should not be a significant additional burden. Sources should be 
monitoring the water level in the scrubber water tank as a measure of 
good operating practices. Consequently, we revise the requirement to 
require a minimum tank volume or level in conjunction with a minimum 
blowdown rate for sources that elect to use that compliance option.
    c. Fabric Filter. If your combustor is equipped with a fabric 
filter, you must establish, continuously monitor, and comply with 
limits on the operating parameters discussed below.
    i. Maximum Flue Gas Flowrate or Kiln Production Rate. As proposed, 
you must establish a limit on maximum flue gas flowrate or kiln 
production rate as a surrogate. Gas flowrate is a key parameter 
affecting the control efficiency of a fabric filter (and any emissions 
control device). As gas flowrate increases, control efficiency 
generally decreases unless other operating parameters are adjusted to 
accommodate the increased flowrate. Cement kilns and lightweight 
aggregate kilns may establish a limit on maximum production rate (e.g., 
raw material feedrate or clinker or aggregate production rate) in lieu 
of a maximum gas flowrate given that production rate directly relates 
to flue gas flowrate.
    As proposed, you must establish a maximum gas flowrate or 
production rate limit as the average of the maximum hourly rolling 
averages for each run of the comprehensive performance test.
    ii. Minimum Pressure Drop and Maximum Pressure Drop Across the 
Fabric Filter. You must establish a limit on minimum pressure drop and 
maximum pressure drop across each cell of the fabric filter based on 
manufacturer's specifications.
    Filter failure is typically due to filter holes, bleed-through 
migration of particulate through the filter and cake, and small ``pin 
holes'' in the filter and cake. Because low pressure drop is an 
indicator of one of these types of failure, pressure drop across the 
fabric filter is an indicator of fabric filter failure.
    We had proposed to establish limits on minimum pressure drop based 
on the performance test. Commenters indicate, however, that maintaining 
a pressure drop not less than levels during the performance test will 
not ensure baghouse performance. We concur. The pressure change caused 
by fabric holes may not be measurable, especially at large sources with 
multiple chamber filter housing units that operate in parallel. In 
addition, operating at high pressure drop may not be desirable because 
high pressures can create pin holes.
    Nonetheless, establishing a limit on minimum pressure drop based on 
manufacturer's recommendations, as suggested by a commenter, is a 
reasonable and prudent approach to help ensure fabric filter 
performance. We have since determined that an operating parameter limit 
for maximum pressure drop across each cell of the fabric filter, based 
on manufacturer specifications, is also necessary. As discussed above, 
a high pressure drop in a cell of a fabric filter may cause small 
pinholes to form or may be indicative of bag blinding or plugging, 
which could result in increased particulate emissions. We do not 
consider this additional provision to be burdensome, especially because 
both the maximum and minimum pressure drop limits are based on 
manufacturer specifications on an hourly rolling average. These 
pressure drop monitoring requirements, in combination with COMS for 
cement kilns and bag leak detection systems for incinerators and 
lightweight aggregate kilns, provide a significant measure of assurance 
that control performance is maintained.
    d. Electrostatic Precipitators and Ionizing Wet Scrubbers. As 
proposed, if your combustor is equipped with an electrostatic 
precipitator or ionizing wet scrubber, you must establish, continuously 
monitor, and comply with limits on the operating parameters discussed 
below.
    i. Maximum Flue Gas Flowrate or Kiln Production Rate. You must 
establish a limit on maximum flue gas flowrate or kiln production rate 
as a surrogate. Gas flowrate is a key parameter affecting the control 
efficiency of an emissions control device. As gas flowrate increases, 
control efficiency generally decreases unless other operating 
parameters are adjusted to accommodate the increased flowrate. Cement 
kilns and lightweight aggregate kilns may establish a limit on maximum 
production rate (e.g., raw material feedrate or clinker or aggregate 
production rate) in lieu of a maximum gas flowrate given that 
production rate directly relates to flue gas flowrate.
    As proposed, you must establish a maximum gas flowrate or 
production rate limit as the average of the maximum hourly rolling 
averages for each run of the comprehensive performance test.
    ii. Minimum Secondary Power Input to Each Field. You must establish 
an hourly rolling average limit on minimum secondary power (kVA) input 
to each field of the electrostatic precipitator or ionizing wet 
scrubber based on operations during the comprehensive performance test. 
The hourly rolling average is established as the average of the test 
run averages.
    Electrostatic precipitators capture particulate matter by charging 
the particulate in an electric field and collecting the charged 
particulate on an inversely charged collection plate. Higher voltages 
improve magnetic field strength, resulting in charged particle 
migration to the collection plate. High current leads to an increased 
particle charging rate and increased electric field strength near the 
collection electrode, increasing collection at the plate, as well. 
Therefore, maximizing both voltage and current by specifying minimum 
power input to the electrostatic precipitator is desirable for good 
particulate matter collection in electrostatic precipitators. For these 
reasons, the rule requires you to monitor power input to each field of 
the electrostatic precipitator to ensure that collection efficiency is 
maintained at performance test levels.
    Power input to an ionizing wet scrubber is important because it 
directly affects particulate removal. Ionizing wet scrubbers charge the 
particulate prior to it entering a packed bed wet scrubber. The 
charging aids in the collection of the particulate onto the packing 
surface in the bed. The particulate is then washed off the packing by 
the scrubber liquid. Therefore, power input is a key parameter to 
proper operation of an ionizing wet scrubber.
    One commenter suggests that a minimum limit on electrostatic 
precipitator voltage be used instead of power input because, at low 
particulate matter loadings, operation at maximum power input is 
inefficient. Another commenter suggests that neither a limit on voltage 
or power input is appropriate because a minimum limit would

[[Page 52957]]

actually cause a potential decrease in operational efficiency (required 
power input and voltage are strong functions of gas and particulate 
characteristics, electrostatic precipitator arcing and sparking at high 
voltage and power requirements, etc.). Alternatively, they recommend 
that a limit on the minimum number of energized electrostatic 
precipitator fields be established. We continue to maintain that a 
minimum limit on power input to each field of the electrostatic 
precipitator is generally accepted as an appropriate parameter for 
assuring electrostatic precipitator performance. Consequently, it is an 
appropriate parameter for a generic, national standard. If you believe, 
however, that in your situation limits on alternative operating 
parameters may better assure that control performance is maintained you 
may request approval to use alternative monitoring approaches under 
Sec. 63.1209(1).
    Another commenter suggests that, in addition to a minimum power 
input for an ionizing wet scrubber, a limit should be set on the 
maximum time allowable to be below the minimum voltage. While feasible, 
we conclude that this limit is not necessary on a national basis 
because the one hour rolling average requirement limits the amount of 
time a source can operate below its minimum voltage limit. We 
acknowledge, however, that a permit writer may find it necessary to 
require shorter averaging periods (e.g., ten-minute or instantaneous 
limits) to better control the amount of time a source can operate at 
levels below its limit.
7. What Are the Operating Parameter Limits for Destruction and Removal 
Efficiency?
    You must establish, monitor, and comply with the same operating 
parameter limits to ensure compliance with the destruction and removal 
efficiency (DRE) standard as you establish to ensure good combustion 
practices are maintained for compliance with the dioxin/furan emission 
standard. See Sec. 63.1209(j) and the discussion in Section VII.D.1 
above. This is because compliance with the DRE standard is ensured by 
maintaining combustion efficiency using good combustion practices. 
Thus, the DRE operating parameters are: maximum waste feedrate for 
pumpable and nonpumpable wastes, minimum gas temperature for each 
combustion chamber, maximum gas flowrate or kiln production rate, and 
parameters that you recommend to ensure the operations of each 
hazardous waste firing system are maintained.249
---------------------------------------------------------------------------

    \249\ You are required to establish operating requirements only 
for hazardous waste firing systems because of DRE standard applies 
only to hazardous waste. Permitting officials may determine on a 
site-specific basis under authority of Sec. 63.1209(g)(2), however, 
that combustion of other fuels or wastes may affect your ability to 
maintain DRE for hazardous waste. Accordingly, permitting officials 
may define operating requirements for other (i.e., other than 
hazardous waste) waste or fuel firing systems. Permitting officials 
may also determine under that provision on a site-specific basis 
that operating requirements other than those prescribed for DRE (and 
good combustion practices) may be needed to ensure compliance with 
the DRE standard.
---------------------------------------------------------------------------

VIII. Which Methods Should Be Used for Manual Stack Tests and 
Feedstream Sampling and Analysis?

    This part discusses the manual stack test and the feedstream 
sampling and analysis methods required by today's rule.
A. Manual Stack Sampling Test Methods
    To demonstrate compliance with today's rule, you must use: (1) 
Method 0023A for dioxin and furans; (2) Method 29 for mercury, 
semivolatile metals, and low volatile metals; (3) Method 26A for 
hydrochloric acid and chlorine; and (4) Method 5 or 5i for particulate 
matter. These methods are found at 40 CFR part 60, appendix A, and in 
``Test Methods for Evaluating Solid Waste, Physical/Chemical Methods,'' 
EPA publication.
    In the NPRM, we proposed that BIF manual stack test methods 
currently located in SW-846 be required to demonstrate compliance with 
the proposed standards. Based on public comments from the proposal, in 
the December 1997 NODA we considered simply citing the ``Air Methods'' 
found in appendix A to part 60. Our rationale was that facilities may 
be required to perform two identical tests, one from SW-846 for 
compliance with MACT or RCRA and one from part 60, appendix A, for 
compliance with other air rules using identical test methods simply 
because one method is an SW-846 method and the other an Air Method. See 
62 FR at 67803. To facilitate compliance with all air emissions stack 
tests, we stated that we would list the methods found in 40 CFR part 
60, appendix A, as the stack test methods used to comply with the 
standards. Later in this section we present an exception for dioxin and 
furan testing.
    In today's rule, we adopt the approach of the December 1997 NODA 
and require that the test methods found in 40 CFR part 60, appendix A 
be used to demonstrate compliance with the emission standards of 
today's rule, except for dioxin and furan. Specifically, today's rule 
requires you to use Method 0023A in SW-846 for sampling dioxins and 
furans from stack emissions. As noted by commenters, improvements have 
been made to the dioxin and furan Method 0023A in the Third Update of 
SW-846 that have been previously incorporated into today's regulations. 
See the 40 CFR 63.1208(a), incorporation of SW-846 by reference. 
However, these have not yet been incorporated into 40 CFR part 60, 
Appendix A. To capture these improvements to the method, today's rule 
incorporates by reference SW-846 Method 0023A. We have evaluated both 
methods. Use of the improved Method 0023A will not affect the 
achievability of the dioxin and furan standard.
    In the proposal, we sought comment on the handling of nondetect 
values for congeners analyzed using the dioxin and furan method. We 
also sought comment on whether the final rule should specify minimum 
sampling times. We proposed allowing facilities to assume that 
emissions of dioxins and furans congeners are zero if the analysis 
showed a nondetect for that congener and the sample time for the test 
method run was at least 3 hours. See 61 FR 17378. Dioxin/furan results 
may not be blank corrected. We received several comments this proposed 
approach, which are summarized below.
    One commenter believes that a minimum dioxin/furan sampling time of 
two hours is sufficient. Another commenter believes that a minimum 
sample time as well as a minimum sample volume should be specified. 
Several commenters agree that nondetects should be treated as zero 
(which is consistent with the German standard) and prefer the three 
hour minimum sample period because this would help eliminate intra-
laboratory differences and difficulties with matrix effects in 
attaining low detection limits. One commenter believes that EPA should 
specify the required detection limit for each congener analysis, 
otherwise the provision to assign zeroes to nondetected congeners in 
the TEQ calculation is open to abuse and could result in an 
understatement of the true dioxin/furan emissions. This commenter also 
believes that a source should not be allowed to sample dioxin/furans 
for time periods less than three hours, even if they assume nondetects 
are present at the detection limit.
    Upon carefully considering all the above comments, we conclude that 
the following approach best addresses the nondetect issue. The final 
rule requires all sources to sample dioxin/furans for a minimum of 
three hours for each run,

[[Page 52958]]

and requires all sources to collect a flue gas sample of at least 2.5 
dscm. We conclude both these requirements are necessary to maintain 
consistency from source to source, and to better assure that the 
dioxin/furan emission results are accurate and representative. We 
conclude that these two requirements are achievable and appropriate 
250. These requirements are consistent with the requirements 
included in the proposed Portland Cement Kiln MACT rule (see 64 FR at 
31898). The final rule also allows a source to assume all nondetected 
congeners are not present in the emissions when calculating TEQ values 
for compliance purposes.
---------------------------------------------------------------------------

    \250\ See Final Technical Support Document, Volume IV, Chapter 
3, for further discussion.
---------------------------------------------------------------------------

    We considered whether it would be appropriate to specify required 
minimum detection limits for each congener analysis in order to better 
assure that sources achieved reasonable detection limits, as one 
commenter recommended. Such a requirement would prevent abuse and 
understatements of the true dioxin/furan emissions. We conclude, 
however, that it is not appropriate to finalize minimum detection 
limits in this rulemaking without giving the opportunity to all 
interested parties to review and comment on such an approach.
    However, we are concerned that (1) sources have no incentive to 
achieve low detection limits; and (2) sources may abuse the provision 
that allows nondetected congener results to be treated as if they were 
not present. As explained in the Final Technical Support Document 
referenced in the preceding paragraph, if one assumes that all dioxin/
furan congeners are present at what we consider to be poor detection 
limits using Method 23A, the resultant TEQ can approach the emission 
standard. This outcome is clearly inappropriate from a compliance 
perspective.
    As a result, we highly recommend that this issue be addressed in 
the review process of the performance test workplan. Facilities should 
submit information that describes the target detection limits for all 
congeners, and calculate a dioxin/furan TEQ concentration assuming all 
congeners are present at the detection limit (similar to what is done 
for risk assessments). If this value is close to the emission standard, 
both the source and the regulatory official should determine if it is 
appropriate to either sample for longer time periods or investigate 
whether it is possible to achieve lower detection limits by using 
different analytical procedures that are approved by the Agency.
    Also, EPA has developed analytical standards for certain mono-
through tri-chloro dioxin and furan congeners. We encourage you to test 
for these congeners in addition to the congeners that comprise today's 
standards. This can be done at very little increased cost. If you test 
for these additional congeners, please include the results in your 
Notification of Compliance. We would like this data so we can develop a 
database from which to determine which (if any) of these compounds can 
act as surrogate(s) for the dioxin and furan congeners which comprise 
the total and TEQ. If easily measurable surrogate(s) can be found, we 
can then start the development of a CEMS for these surrogates. A 
complete list of these congeners will be included in the implementation 
document for this rule and updated periodically through guidance.
    One commenter suggests that a source be allowed to conduct one 
extended dioxin/furan sampling event as opposed to three separate runs 
with three separate sampling trains because this would minimize the 
radioactive waste generated for sources that combust mixed waste. We 
conclude this issue should be handled on a site-specific basis, 
although an allowance of such an approach seems reasonable. A source 
can petition the Agency under the provisions of Sec. 63.7(f) for an 
alternative test method for such a site-specific determination.
    The final rule also adopts the approach discussed in the December 
1997 NODA for sampling of mercury, semi-volatile metals, and low-
volatile metals. Therefore, for stack sampling of mercury, semi-
volatile metals, and low-volatile metals, you are required to use 
Method 29 in 40 CFR part 60, appendix A. No adverse comments were 
received concerning this approach in the December 1997 NODA.
    For compliance with the hydrochloric acid and chlorine standards, 
today's rule requires that you use Method 26A in 40 CFR part 60, 
appendix A. Commenters state that we should instead require a method 
involving the Fourier Transform Infrared and Gas Filter Correlation 
Infrared instrumental techniques. Commenters contend that Method 26A is 
biased high at cement kilns because it collects ammonium chloride in 
addition to the hydrochloric acid and chlorine gas emissions it was 
designed to report. Commenters also indicate that the Fourier Transform 
Infrared and Gas Filter Correlation Infrared were validated against 
Method 26A and that these alternative methods do not bias the results 
high due to ammonium chloride 251. The data for today's 
hydrochloric acid standard was derived using the SW-846 equivalent to 
Method 26A (Method 0050) as the reference method. Therefore, today's 
standard accounts for the ammonium chloride collection bias. We reject 
the idea that we should require other methods. If the commenters are 
correct, other methods would not sample the ammonium chloride portion, 
thus making the standard less stringent. You can obtain Administrator 
approval for using Fourier Transform Infrared or Gas Filter Correlation 
Infrared techniques following the provisions found in 40 CFR 63.7 if 
those methods are found to pass a part 63, appendix A, Method 301 
validation at the source.
---------------------------------------------------------------------------

    \251\ After further review and consideration of the GFCIR Method 
(322), we will not be promulgating its use in the Portland Cement 
Kiln NESHAP rulemaking due to problems encountered with the method 
during emission testing at lime manufacturing plants.
---------------------------------------------------------------------------

    Compliance with the particulate matter standards requires the use 
of either Method 5 or Method 5i in 40 CFR part 60, appendix A. See a 
related discussion of Method 5i in Part 5, section VII.C.2.a of the 
preamble to today's rule. Although Method 5i has better precision than 
Method 5, your choice of methods depends on the emissions during the 
performance test. In cases of low levels of particulate matter (i.e., 
for total train catches of less than 50 mg), we prefer that Method 5i 
be used. For higher emissions, Method 5 may be used 252. In 
practice this will likely mean that all incinerators and most 
lightweight aggregate kilns will use Method 5i for compliance, while 
some lightweight aggregate kilns and most cement kilns will use Method 
5.
---------------------------------------------------------------------------

    \252\ We note that this total train catch is not intended to be 
a data acceptance criteria. Thus, total train catches exceeding 50 
mg do not invalidate the method.
---------------------------------------------------------------------------

    Today's rule also allows the use of any applicable SW-846 test 
methods to demonstrate compliance with requirements of this subpart. As 
an example, some commenters noted a preference to perform particulate 
matter and hydrochloric acid tests together using Method 0050. Today's 
rule would allow that practice. Applicable SW-846 test methods are 
incorporated for use into today's rule via reference. See section 
1208(a).
B. Sampling and Analysis of Feedstreams
    Today's rule does not require the use of SW-846 methods for the 
sampling and analysis of feedstreams. Consistent with our approach to 
move toward performance based measurement

[[Page 52959]]

 systems for other than method-defined parameters,253 
today's rule allows the use of any reliable analytical method to 
determine feedstream concentrations of metals, halogens, and other 
constituents. It is your responsibility to ensure that the sampling and 
analysis are unbiased, precise, and representative of the waste. For 
the waste, you must demonstrate that: (1) Each constituent of concern 
is not present above the specification level at the 80% upper 
confidence limit around the mean; and (2) the analysis could have 
detected the presence of the constituent at or below the specification 
level at the 80% upper confidence limit around the mean. You can refer 
to the Guidance for Data Quality Assessment--Practical Methods for Data 
Analysis, EPA QA/G-9, January 1998, EPA/600/R-96/084 for more 
information. Proper selection of an appropriate analytical method and 
analytical conditions (as allowed by the scope of that method) are 
demonstrated by adequate recovery of spiked analytes (or surrogate 
analytes) and reproducible results. Quality control data obtained must 
also reflect consistency with the data quality objectives and intent of 
the analysis. You can read the January 31, 1996, memorandum from Barnes 
Johnson, Director of the Economics, Methods, and Risk Assessment 
Division, to James Berlow, Director of the Hazardous Waste Minimization 
and Management Division for more information on this topic.
---------------------------------------------------------------------------

    \253\ Feedstream sampling and analysis are not method defined 
parameters.
---------------------------------------------------------------------------

IX. What Are the Reporting and Recordkeeping Requirements?

    We discuss in this section reporting and recordkeeping requirements 
and a provision in the rule for allowing data compression to reduce the 
recordkeeping burden.
A. What Are the Reporting Requirements?
    The reporting requirements of the rule include notifications and 
reports that must be submitted to the Administrator as well as 
notifications, requests, petitions, and applications that you must 
submit to the Administrator only if you elect to request approval to 
comply with certain reduced or alternative requirements. These 
reporting requirements are summarized in the following tables. We 
discuss previously in various sections of today's preamble the 
rationale for additional or revised reporting requirements to those 
currently required under subpart A of part 63 for all MACT sources. In 
other cases, the reporting requirements for hazardous waste combustors 
are the same as for other MACT sources (e.g., initial notification 
under existing Sec. 63.9(b). We also show in the tables the 
reference(s) in the regulations for the reporting requirement.

   Summary of Notifications That You Must Submit to the Administrator
------------------------------------------------------------------------
               Reference                           Notification
------------------------------------------------------------------------
63.9(b)................................  Initial notifications that you
                                          are subject to Subpart EEE.
63.1210(b) and (c).....................  Notification of intent to
                                          comply.
63.9(d)................................  Notification that you are
                                          subject to special compliance
                                          requirements.
63.1207(e), 63.9(e) 63.9(g) (1) and (3)  Notification of performance
                                          test and continuous monitoring
                                          system evaluation, including
                                          the performance test plan and
                                          CMS performance evaluation
                                          plan.
163.1210(d), 63.1207(j), 63.9(h),        Notification of compliance,
 63.10(d)(2), 63.10(e)(2).                including results of
                                          performance tests and
                                          continuous monitoring system
                                          performance evaluations.
63.1206(b)(6)..........................  Notification of changes in
                                          design, operation, or
                                          maintenance.
63.9(j)................................  Notification and documentation
                                          of any change in information
                                          already provided under Sec.
                                          63.9.
------------------------------------------------------------------------
\1\ You may also be required on a case-by-case basis to submit a
  feedstream analysis plan under Sec.  63.1209(c)(3).


      Summary of Reports That You Must Submit to the Administrator
------------------------------------------------------------------------
               Reference                              Report
------------------------------------------------------------------------
63.1211(b).............................  Compliance progress report
                                          associated and submitted with
                                          the notification of intent to
                                          comply.
63.10(d)(4)............................  Compliance progress reports, if
                                          required as a condition of an
                                          extension of the compliance
                                          date granted under Sec.
                                          63.6(i).
63.1206(c)(3)(vi)......................  Excessive exceedances reports.
63.1206(c)(4)(iv)......................  Emergency safety vent opening
                                          reports.
63.10(d)(5)(i).........................  Periodic startup, shutdown, and
                                          malfunction reports.
63.10(d)(5)(ii)........................  Immediate startup, shutdown,
                                          and malfunction reports.
63.10(e)(3)............................  Excessive emissions and
                                          continuous monitoring system
                                          performance report and summary
                                          report.
------------------------------------------------------------------------


Summary of Notifications, Requests, Petitions, and Applications That You
    Must Submit to the Administrator Only if You Elect To Comply With
                   Reduced or Alternative Requirements
------------------------------------------------------------------------
                                              Notification, request,
               Reference                     petition, or application
------------------------------------------------------------------------
63.1206(b)(5), 63.1213, 63.6(i),         You may request an extension of
 63.9(c).                                 the compliance date for up to
                                          one year.
63.9(i)................................  You may request an adjustment
                                          to time periods or postmark
                                          deadlines for submittal and
                                          review of required
                                          information.
63.1209(g)(1)..........................  You may request approval of:
                                          (1) alternative monitoring
                                          methods, except for standards
                                          that you must monitor with a
                                          continuous emission monitoring
                                          system (CEMS) and except for
                                          requests to use a CEMS in lieu
                                          of operating parameter limits;
                                          or (2) a waiver of an
                                          operating parameter limit.
63.1209(a)(5), 63.8(f).................  You may request: (1) approval
                                          of alternative monitoring
                                          methods for compliance with
                                          standards that are monitored
                                          with a CEMS; and (2) approval
                                          to use a CEMS in lieu of
                                          operating parameter limits.

[[Page 52960]]

 
63.1204(d)(4)..........................  Notification that you elect to
                                          comply with the emission
                                          averaging requirements for
                                          cement kilns with in-line raw
                                          mills.
63.1204(e)(4)..........................  Notification that you elect to
                                          comply with the emission
                                          averaging requirements for
                                          preheater or preheater/
                                          precalciner kilns with dual
                                          stacks.
63.1206(b)(1)(ii)(A)...................  Notification that you elect to
                                          document compliance with all
                                          applicable requirements and
                                          standards promulgated under
                                          authority of the Clean Air
                                          Act, including Sections 112
                                          and 129, in lieu of the
                                          requirements of Subpart EEE
                                          when not burning hazardous
                                          waste.
63.1206(b)(9)(iii)(B)..................  If you elect to conduct
                                          particulate matter CEMS
                                          correlation testing and wish
                                          to have federal particulate
                                          matter and opacity standards
                                          and associated operating
                                          limits waived during the
                                          testing, you must notify the
                                          Administrator by submitting
                                          the correlation test plan for
                                          review and approval.
63.1206(b)(10).........................  Owners and operators of
                                          lightweight aggregate kilns
                                          may request approval of
                                          alternative emission standards
                                          for mercury, semivolatile
                                          metal, low volatile metal, and
                                          hydrochloric acid/chlorine gas
                                          under certain conditions.
63.1206(b)(11).........................  Owners and operators of cement
                                          kilns may request approval of
                                          alternative emission standards
                                          for mercury, semivolatile
                                          metal, low volatile metal, and
                                          hydrochloric acid/chlorine gas
                                          under certain conditions.
63.1207(c)(2)..........................  You may request to base initial
                                          compliance on data in lieu of
                                          a comprehensive performance
                                          test.
63.1207(i).............................  You may request up to a one-
                                          year time extension for
                                          conducting a performance test
                                          (other than the initial
                                          comprehensive performance
                                          test) to consolidate testing
                                          with other state or federally-
                                          required testing.
63.1209(l)(1)..........................  You may request to extrapolate
                                          mercury feedrate limits.
63.1209(n)(2)(ii)......................  You may request to extrapolate
                                          semivolatile and low volatile
                                          metal feedrate limits.
63.10(e)(3)(ii)........................  You may request to reduce the
                                          frequency of excess emissions
                                          and CMS performance reports.
63.10(f)...............................  You may request to waive
                                          recordkeeping or reporting
                                          requirements.
63.1211(e).............................  You may request to use data
                                          compression techniques to
                                          record data on a less frequent
                                          basis than required by Sec.
                                          63.1209.
------------------------------------------------------------------------

    Some commenters suggest that the rule needs to provide additional 
reporting of information regarding metals fed to cement kilns, 
including quarterly reporting of daily average metal feedrates, maximum 
hourly feedrates, and all testing and analytical information on the 
toxic metal content of cement kiln dust and clinker product. Also, they 
suggest that toxic metals that are Toxics Release Inventory pollutants 
and that are released to the land from cement kiln dust disposal should 
be reported. While these reports might have some value for other 
purposes, we must carefully scrutinize all reporting and recordkeeping 
burdens for a rulemaking and determine whether the reporting and 
recordkeeping requirements are necessary to ensure compliance with the 
standards. (We, as an agency, cannot increase overall our reporting and 
recordkeeping burden.)
    We do not believe that these reports are needed to ensure 
compliance with the standards and therefore are not requiring them. On 
balance, quarterly filing requirements would be too burdensome. A 
source must document compliance with all operating parameter limits and 
emission standards at all times, and its records are subject to 
inspection at any time. There is no additional need to provide 
quarterly reports.
    One commenter suggests that the proposed rule incorrectly focuses 
on maximizing data collection as opposed to ensuring performance, thus 
frustrating the use of better technology and methods. We, of course, 
are also interested in ensuring performance by all reasonable means, 
which for example accounts for our continued focus on continuous 
emission monitors. However, we are not able to sacrifice data 
collection as a means for ensuring compliance as well as a means to 
undergird future rulemakings, assess achievability, and determine site-
specific compliance limits, where necessary.
B. What Are the Recordkeeping Requirements?
    You must keep the records summarized in the table below for at 
least five years from the date of each occurrence, measurement, 
maintenance, corrective action, report, or record. See existing 
Sec. 63.10(b)(1). At a minimum, you must retain the most recent two 
years of data on site. You may retain the remaining three years of data 
off site. You may maintain such files on: microfilm, a computer, 
computer floppy disks, optical disk, magnetic tape, or microfiche.
    We discuss previously in various sections of today's preamble the 
rationale for additional or revised recordkeeping requirements to those 
currently required under subpart A of part 63 for all MACT sources. In 
other cases, the recordkeeping requirements for hazardous waste 
combustors are the same as for other MACT sources (e.g., record of the 
occurrence and duration of each malfunction of the air pollution 
control equipment; see existing Sec. 63.10(b)(2)(ii)). We also show in 
the table the reference(s) in the regulations for the recordkeeping 
requirement.

[[Page 52961]]



Summary of Documents, Data, and Information That You Must Include in the
                            Operating Record
------------------------------------------------------------------------
               Reference                  Document, data, or information
------------------------------------------------------------------------
63.1201(a), 63.10 (b) and (c)..........  General. Information required
                                          to document and maintain
                                          compliance with the
                                          regulations of Subpart EEE,
                                          including data recorded by
                                          continuous monitoring systems
                                          (CMS), and copies of all
                                          notifications, reports, plans,
                                          and other documents submitted
                                          to the Administrator.
63.1211(d).............................  Documentation of compliance.
63.1206 (c)(3)(vii)....................  Documentation and results of
                                          the automatic waste feed
                                          cutoff operability testing.
63.1209 (c)(2).........................  Feedstream analysis plan.
63.1204 (d)(3).........................  Documentation of compliance
                                          with the emission averaging
                                          requirements for cement kilns
                                          with in-line raw mills.
63.1204 (e)(3).........................  Documentation of compliance
                                          with the emission averaging
                                          requirements for preheater or
                                          preheater/precalciner kilns
                                          with dual stacks.
63.1206(b)(1) (ii)(B)..................  If you elect to comply with all
                                          applicable requirements and
                                          standards promulgated under
                                          authority of the Clean Air
                                          Act, including Sections 112
                                          and 129, in lieu of the
                                          requirements of Subpart EEE
                                          when not burning hazardous
                                          waste, you must document in
                                          the operating record that you
                                          are in compliance with those
                                          requirements.
63.1206 (c)(2).........................  Startup, shutdown, and
                                          malfunction plan.
63.1206(c) (3)(v)......................  Corrective measures for any
                                          automatic waste feed cutoff
                                          that results in an exceedance
                                          of an emission standard or
                                          operating parameter limit.
63.1206(c) (4)(ii).....................  Emergency safety vent operating
                                          plan.
63.1206 (c)(4)(iii)....................  Corrective measures for any
                                          emergency safety vent opening.
63.1206 (c)(6).........................  Operator training and
                                          certification program.
63.1209 (k)(6)(iii), 63.1209             Documentation that a substitute
 (k)(7)(ii), 63.1209 (k)(9)(ii),          activated carbon, dioxin/furan
 63.1209 (o)(4)(iii).                     formation reaction inhibitor,
                                          or dry scrubber sorbent will
                                          provide the same level of
                                          control as the original
                                          material.
------------------------------------------------------------------------

    Some commenters are concerned that the specification of media on 
which these files may be maintained unnecessarily limits the options to 
facilities, especially those not equipped with computer or other 
electronic data gathering equipment. We conclude, however, that the 
options listed under Sec. 63.10(b)(1) seem to provide the greatest 
flexibility possible, including the reasonable management of paper 
records through the use of microfilm or microfiche. We encourage the 
use of computer and electronic equipment, however, for logistical 
reasons (retrieval and inspection can be easier) and as a means to 
enhance dissemination to the local community to foster an atmosphere of 
full and open disclosure about facility operations.
C. How Can You Receive Approval to Use Data Compression Techniques?
    You may submit a written request to the Administrator under 
Sec. 63.1211(f) for approval to use data compression techniques to 
record data from CMS, including CEMS, on a frequency less than that 
required by Sec. 63.1209. You must submit the request for review and 
approval as part of the comprehensive performance test plan. For each 
CEMS or operating parameter for which you request to use data 
compression techniques, you must provide: (1) A fluctuation limit that 
defines the maximum permissible deviation of a new data value from a 
previously generated value without requiring you to revert to recording 
each one-minute average; and (2) a data compression limit defined as 
the closest level to an operating parameter limit or emission standard 
at which reduced recording is allowed.
    You must record one-minute average values at least every ten 
minutes. If after exceeding a fluctuation limit you remain below the 
limit for a ten-minute period, you may reinitiate your data compression 
technique provided that you are not exceeding the data compression 
limit.
    The fluctuation limit should represent a significant change in the 
parameter measured, considering the range of normal values. The data 
compression limit should reflect a level at which you are unlikely to 
exceed the specific operating parameter limit or emission standard, 
considering its averaging period, with the addition of a new one-minute 
average.
    We provide the following table of recommended fluctuation and data 
compression limits as guidance. These are the same limits that we 
discussed in the May 1997 NODA.

                               Recommended Fluctuation and Data Compression Limits
----------------------------------------------------------------------------------------------------------------
                                               Fluctuation limit ()                    Data compression limit
----------------------------------------------------------------------------------------------------------------
Continuous Emission Monitoring System:
    Carbon monoxide........................  10 ppm.......................  50 ppm.
    Hydrocarbon............................  2 ppm........................  60% of standard.
Combustion Gas Temperature Quench: Maximum   10 deg.F.....................  Operating parameter limit (OPL)
 inlet temperature for dry particulate                                       minus 30 deg.F.
 matter control device or, for lightweight
 aggregate kilns, temperature at kiln exit.
Good Combustion Practices:
    Maximum gas flowrate or kiln production  10% of OPL...................  60% of OPL.
     rate.
    Maximum hazardous waste feedrate.......  10% of OPL...................  60% of OPL.
    Maximum gas temperature for each         20 deg.F.....................  OPL plus 50 deg.F.
     combustion chamber.
Activated Carbon Injection:
    Minimum carbon injection feedrate......  5% of OPL....................  OPL plus 20%.
    Minimum carrier fluid flowrate or        20% of OPL...................  OPL plus 25%.
     nozzle pressure drop.
Activated Carbon Bed: Maximum gas            10 deg.F.....................  OPL minus 30 deg.F.
 temperature at inlet or exit of the bed.
Catalytic Oxidizer:
    Minimum flue gas temperature at          20 deg.F.....................  OPL plus 40 deg.F.
     entrance.
    Maximum flue gas temperature at          20 deg.F.....................  OPL minus 40 deg.F.
     entrance.
Dioxin Inhibitor: Minimum inhibitor          10% of OPL...................  60% of OPL.
 feedrate.
Feedrate Control:

[[Page 52962]]

 
    Maximum total metals feedrate (all       10% of OPL...................  60% of OPL.
     feedstreams).
    Maximum low volatile metals feedrate,    10% of OPL...................  60% of OPL.
     pumpable feedstreams.
    Maximum total ash feedrate (all          10% of OPL...................  60% of OPL.
     feedstreams).
    Maximum total chlorine feedrate (all     10% of OPL...................  60% of OPL.
     feedstreams).
Wet scrubber:
    Minimum pressure drop across scrubber..  0.5 inches water.............  OPL plus 2 inches water.
    Minimum liquid feed pressure...........  20% of OPL...................  OPL plus 25%.
    Minimum liquid pH......................  0.5 pH unit..................  OPL plus 1 pH unit.
    Maximum solids content in liquid.......  5% of OPL....................  OPL minus 20%.
    Minimum blowdown (liquid flowrate).....  5% of OPL....................  OPL plus 20%.
    Minimum liquid flowrate or liquid        10% of OPL...................  OPL plus 30%.
     flowrate/gas flowrate ratio.
Dry scrubber:
    Minimum sorbent feedrate...............  10% of OPL...................  OPL plus 30%.
    Minimum carrier fluid flowrate or        10% of OPL...................  OPL plus 30%.
     nozzle pressure drop.
Fabric filter: Minimum pressure drop across  1 inch water.................  OPL plus 2 inches water.
 device.
Electrostatic precipitator and ionizing wet  5% of OPL....................  OPL plus 20%.
 scrubber: Minimum power input (kVA:
 current and voltage).
----------------------------------------------------------------------------------------------------------------

    Data compression is the process by which a facility automatically 
evaluates whether a specific data point needs to be recorded. Data 
compression does not represent a change in the continuous monitoring 
requirement in the rule. One-minute averages will continue to be 
generated. With data compression, however, each one-minute average is 
automatically compared with a set of specifications (i.e., fluctuation 
limit and data compression limit) to determine whether it must be 
recorded. New data are recorded when the one-minute average value falls 
outside these specifications.
    We did not propose data compression techniques in the April 1996 
NPRM. In response to the proposed monitoring and recording 
requirements, however, commenters raise concerns about the burden of 
recording one-minute average values for the array of operating 
parameter limits that we proposed. Commenters suggest that allowing 
data compression would significantly reduce the recordkeeping burden 
while maintaining the integrity of the data for compliance monitoring. 
We note that data compression should also benefit regulatory officials 
by allowing them to focus their review on those data that are 
indicative of nonsteady-state operations and that are close to the 
operating parameter limit or, for CEMS, the emission standard.
    In response to these concerns, we presented data compression 
specifications in the May 1997 NODA. Public comments on the NODA are 
uniformly favorable. Therefore, we are including a provision in the 
final rule that allows you to request approval to use data compression 
techniques. The fluctuation and data compression limits presented above 
are offered as guidance to assist you in developing your recommended 
data compression methodology.
    We are not promulgating data compression specifications because the 
dynamics of monitored parameters are not uniform across the regulated 
universe. Thus, establishing national specifications would be 
problematic. Various data compression techniques can be successfully 
implemented for a monitored parameter to obtain compressed data that 
reflect the performance on a site-specific basis. Thus, the rule 
requires you to recommend a data compression approach that addresses 
the specifics of your operations. The fluctuation and data compression 
limits presented above are offered solely as guidance and are not 
required.
    The rule requires that you record a value at least once every ten 
minutes to ensure that a minimum, credible data base is available for 
compliance monitoring. If you operate under steady-state conditions at 
levels well below operating parameter limits and CEMS-monitored 
emission standards, data compression techniques may enable you to 
achieve a potential reduction in data recording up to 90 percent.

X. What Special Provisions Are Included in Today's Rule?

A. What Are the Alternative Standards for Cement Kilns and Lightweight 
Aggregate Kilns?
    In the May 1997 NODA, we discussed alternative standards for cement 
kilns and lightweight aggregate kilns that have metal or chlorine 
concentrations in their mineral and related process raw materials that 
might cause an exceedance of today's standard(s), even though the 
source uses MACT control. (See 62 FR 24238.) After carefully 
considering commenters input, we adopt a process that allows sources to 
petition the Administrator for alternative mercury, semivolatile metal, 
low volatile metal, or hydrochloric acid/chlorine gas standards under 
two different sets of circumstances. One reason for a source to 
consider a petition is when a kiln cannot achieve the standard, while 
using MACT control, because of raw material contributions to their 
hazardous air pollutant emissions. The second reason is limited to 
mercury, and applies when mercury is not present at detectable levels 
in the source's raw material. These alternative standards are discussed 
separately below.
1. What Are the Alternative Standards When Raw Materials Cause an 
Exceedance of an Emission Standard? See sections 1206(b) (10) and (11)
    a. What Approaches Have We Publicly Discussed? We acknowledge that 
a kiln using properly designed and operated MACT control technologies, 
including control of metals levels in hazardous waste feedstocks, may 
not be capable of achieving the emission standards (i.e., the mercury, 
semivolatile metal, low volatile metal, and/or hydrochloric acid/
chlorine gas standards). This can occur when hazardous air pollutants 
(i.e., metals and chlorine) contained in the raw material volatilize or 
are entrained in the flue gas such that their contribution to total 
metal and chlorine emissions cause an exceedance of the emission 
standard.
    Our proposal first acknowledged this possible situation. In the 
April 1996 NPRM, we proposed metal and chlorine standards that were 
based, in part, on specified levels of hazardous waste feedrate control 
as MACT control. To address our concern that kilns may not

[[Page 52963]]

be able to achieve the standards when using MACT control technologies, 
given raw material contributions to emissions, we performed an 
analysis. Our analysis estimated the total emissions of each kiln 
including emissions from raw materials, while also assuming the source 
was using MACT hazardous waste feedrate and particulate matter control. 
Results of this analysis, which were discussed in the proposal, 
indicated that there may be several kilns that would not be able to 
achieve the proposed emission standards while using MACT control, due 
to levels of metals and chlorine in raw material and/or conventional 
fuel. (See 61 FR at 17393-17406.) Commenters requested that we provide 
an equivalency determination to allow sources to comply with a control 
efficiency requirement (e.g., a minimum metal system removal 
efficiency) in lieu of the emission standard. (See response below.)
    In the May 1997 NODA, we discussed revised standards that defined 
MACT control, in part, based on hazardous waste metal and chlorine 
feedrate control--as did the NPRM. (See 62 FR 24225-24235.) However, 
our revised approach did not define specific levels of hazardous waste 
metal and chlorine feedrate control, therefore, making it difficult to 
attribute a kiln's failure to meet emission standards to metals levels 
in raw materials.254 In response to a commenter's request, 
we discussed, in the May 1997 NODA, an alternative approach to address 
raw material contributions. Our approach did not subject a source to 
the MACT standards if the source could document that metal or chlorine 
concentrations in their hazardous waste, and any nonmineral feedstock, 
is within the range of normal industry levels. The purpose of this 
requirement was to ensure that metal and chlorine emissions 
attributable to nonmineral feedstreams were roughly equivalent to those 
from sources achieving the MACT emission standards. The use of an 
industry average, or normal metal and chlorine level, was to serve as a 
surrogate MACT feedrate control level for the alternative standard 
because we did not define a specific level of control as MACT. We also 
requested comment on how best to determine normal hazardous waste metal 
and chlorine levels.
---------------------------------------------------------------------------

    \254\ We could not estimate a cement kiln's total emissions 
(i.e., to determine emission standard achievability) based on the 
assumption that the kiln is feeding metals in the hazardous waste at 
the MACT control feedrate levels.
---------------------------------------------------------------------------

    Today's final rule uses a revised standard setting methodology that 
defines specific levels of hazardous waste metal and chlorine feedrates 
as MACT control.255 As a result, we do not need to define 
normal, or average, metal and chlorine levels for the purposes of this 
alternative standard provision.
---------------------------------------------------------------------------

    \255\ As explained earlier, the emission standards for metals 
and chlorine reflect the performance of MACT control, which includes 
control of metals and chlorine in the hazardous waste feed 
materials. As further explained, sources are not required to adopt 
MACT control. Sources must, however, achieve the level of 
performance which MACT control achieves. Therefore, sources are not 
required to control metals and chlorine hazardous waste feedrates to 
the same levels as MACT control in order to comply with the 
standards for metals and chlorine. Rather, the source can elect to 
achieve the emission standard by any means, which may or may not 
involve hazardous waste feedrate control
---------------------------------------------------------------------------

    b. What Comments Did We Receive on Our Approaches? There were many 
comments supporting and many opposing the concept of allowing 
alternative standards. Several commenters focus on the Agency's legal 
basis for this type of alternative standard. Some, supporting an 
alternative standard, wrote that feedrate control of raw materials at 
mineral processing plants is not a permissible basis for MACT control. 
In support of their position, some directed our attention to the 
language found in the Conference Report to the 1990 CAA 
amendments.256 However, as we noted in the April 1996 NPRM 
and as was mentioned by many commenters 257, the Conference 
Report language is not reflected in the statute. Section 112(d)(2)(A) 
of the statute states, without caveat, that MACT standards may be based 
on ``process changes, substitution of materials or other 
modifications.''
---------------------------------------------------------------------------

    \256\ H.R. Rep. No. 101-952, at p. 339, 101st Cong., 2d Sess. 
(Oct. 26, 1990).
    \257\ See 62 FR 24239, May 2, 1997.
---------------------------------------------------------------------------

    As noted above, our MACT approach in today's rule relies on metal 
and chlorine hazardous waste feedrate control as part of 
developing MACT emission standards. It should be noted, that we do not 
directly regulate raw material metal and chlorine input under this 
approach, although there is no legal bar for us to do so. Since raw 
material feedrate control is not an industry practice, raw material 
feedrate control is not part of the MACT floor. In addition, we do not 
adopt such control as a beyond-the-floor standard. We conclude it is 
not cost-effective to require kilns to control metal and chlorine 
emissions by substituting their current raw materials with off-site raw 
materials. (See metal and chlorine emission standard discussions for 
cement kilns and lightweight aggregate kilns in Part Four, Sections VII 
and VIII.) 258
---------------------------------------------------------------------------

    \258\ The nonhazardous waste Portland Cement Kiln MACT 
rulemaking likewise controls semivolatile metal and low volatile 
metal emissions by limiting particulate matter emissions, and did 
not adopt beyond-the-floor standards based on raw material metal and 
chlorine feedrate control--see 64 FR 31898.
---------------------------------------------------------------------------

    Although today's rule offers a petition process, we considered 
varying levels of metal and chlorine emissions attributable to raw 
material in identifying the metal and chlorine emission standards 
through our MACT floor methodology. This consideration helps to ensure 
that the emission standards are achievable for sources using MACT 
control. Therefore, we anticipate very few sources, if any, will need 
to petition the Administrator for alternative standards. However, it is 
possible that raw material hazardous air pollutant levels, at a given 
kiln location, could vary over time and preclude kilns from achieving 
the emission standards. We believe, therefore, that it is appropriate 
to adopt a provision to allow kilns to petition for alternative 
standards so that future changes in raw material feedstock will not 
prevent compliance with today's emission standards.
    Other commenters believe that alternative standards are not 
necessary because there are kilns with relatively high raw material 
metal concentrations already achieving the proposed standards. To 
address this point, and to reevaluate the ability of kilns to achieve 
the emission standards without new control of metals and chlorine in 
raw material and conventional fuel, we again estimated the total metal 
and chlorine emissions, assuming each kiln fed metal and chlorine at 
the defined MACT feedrate control levels.259
---------------------------------------------------------------------------

    \259\ When estimating emissions, the Agency assumed the kiln was 
feeding metals and chlorine in its hazardous waste at the lower of 
the MACT defining maximum theoretical emission concentration levels 
or the level actually demonstrated during its performance test. See 
Final Technical Support Document for Hazardous Waste Combustor MACT 
Standards, Volume II: Selection of MACT Standards and Technologies, 
July 1999, for further discussion.
---------------------------------------------------------------------------

    The following table summarizes the estimated achievability of the 
emission standards assuming kilns used MACT control. Our analysis 
determined achievability both at the emission standard and at the 
design level--70 percent of the standard. (To ensure compliance most 
kilns will ``design'' their system to operate, at a minimum, 30 percent 
below the standard.) The table describes the number of test conditions 
in our data base that would not meet the emission standard or meet the 
design level by estimating total emissions. For example, all cement 
kiln test conditions achieve the mercury emission standard, assuming 
all cement

[[Page 52964]]

kilns used MACT control. On the other hand, the table also indicates 
that four cement kiln test conditions out of 27 do not achieve the 
design level for mercury. In our analysis, if all test conditions 
achieved both the standard and the design level, we concluded that 
there is no reason to believe raw material contributions to metal and 
chlorine emissions might cause a compliance problem.

      Cement Kiln and Lightweight Aggregate Kiln Emission Standard
                          Achievability Results
------------------------------------------------------------------------
                                                         Low
       Source category         Mercury  Semivolatile  Volatile    Total
                                            metal       metal   chlorine
------------------------------------------------------------------------
No. of cement kiln test        \1\0/27     \1\1/38     \1\1/39   \1\2/42
 conditions in MACT data base
 not achieving standard......
No of cement kiln test            4/27        6/38        3/39      3/42
 conditions in MACT data base
 not achieving 70 % design
 level.......................
No of lightweight aggregate       0/17        5/22        2/22      3/18
 kiln test conditions in MACT
 data base not achieving
 standard....................
No of lightweight aggregate       0/17        5/22        4/22     10/18
 kiln test conditions in MACT
 data base not achieving 70%
 design level................
------------------------------------------------------------------------
*Number after slash denotes total number of test conditions.

    Our analysis illustrates that, subject to the assumptions made, 
some lightweight aggregate kilns and cement kilns have raw material 
hazardous air pollutant levels that could affect their ability to 
achieve the emission standard if no additional emission controls were 
implemented (e.g., additional hazardous waste feedrate control, or 
better air pollution control device efficiency). Nevertheless, we 
conclude that it is difficult to determine whether raw material 
hazardous air pollutant contributions to the emissions result in 
unachievable emission standards because of the difficulty associated 
with differentiating raw material hazardous air pollutant emissions 
from hazardous waste pollutant emissions. This uncertainty has led us 
to further conclude that it is appropriate to allow kilns to petition 
for alternative standards, provided that they submit site-specific 
information that shows raw material hazardous air pollutant 
contributions to the emissions prevent the kiln from complying with the 
emission standard even though the kiln is using MACT control.
    Many commenters dislike the idea of an alternative standard. They 
wrote that regulation of raw material metal content may be necessary to 
control semivolatile metal and low volatile metal emissions at 
hazardous waste burning kilns because: (1) These kilns have relatively 
high chlorine levels in the flue gas (which predominately originate 
from the hazardous waste); and (2) chlorine tends to increase metal 
volatility. We agree that increased flue gas chlorine content from 
hazardous waste burning operations may result in increased metals 
volatility, which then could result in higher raw material metal 
emissions.260 The increased presence of chlorine at 
hazardous waste burning kilns presents a concern. To address this 
concern, we require kilns to submit data or information, as part of the 
alternative standard petition, documenting that increased chlorine 
levels associated with the burning of hazardous waste, as compared to 
nonhazardous waste operations, do not significantly increase metal 
emissions attributable to raw material. This requirement is explained 
in greater detail later in this section.
---------------------------------------------------------------------------

    \260\ The potential for increased metal emissions is stronger 
for semivolatile metals (lead, in particular), but low volatile 
metal emissions still have potential to increase with increased flue 
gas chlorine concentrations. See Final Technical Support Document 
for Hazardous Waste Combustor MACT Standards, Volume II: Selection 
of MACT Standards and Technologies, July 1999, for further 
discussion.
---------------------------------------------------------------------------

    Many commenters also point out that the alternative standard, at 
least as originally proposed, could result in metal and chlorine 
emissions exceeding the standard to possible levels of risk to human 
health and the environment. We agree that this potential could exist; 
however, the RCRA omnibus process serves as a safeguard against levels 
of emissions that present risk to human health or the environment. 
Therefore, sources operating pursuant to alternative standards may 
likely be required to perform a site-specific risk assessment to 
demonstrate that their emissions do not pose an unacceptable risk. The 
results of the risk assessment would then be used to develop facility-
specific metal and chlorine emission limits (if necessary), which would 
be implemented and enforced through omnibus conditions in the RCRA 
permit.261
---------------------------------------------------------------------------

    \261\ RCRA permits for hazardous waste combustors address total 
emissions, regardless of the source of the pollutant due to the 
nexus with the hazardous waste treatment activities. See Horsehead v 
Browner, 16 F. 3d 1246, 1261-63 (D.C. Cir. 1994)(Hazardous waste 
combustion standards may address hazardous constituents attributable 
to raw material inputs so long as thee is a reasonable nexus with 
the hazardous waste combustion activites).
---------------------------------------------------------------------------

    c. How Do I Demonstrate Eligibility for the Alternative Standard? 
To demonstrate eligibility, you must submit data or information which 
shows that raw material hazardous air pollutant contributions to the 
emissions prevent you from complying with the emission standard, even 
though you use MACT control for the standard from which you seek 
relief. To allow flexibility in implementation, we do not mandate what 
this demonstration must entail. However, we believe that a 
demonstration should include a performance test while using MACT 
control or better (i.e., the hazardous waste feedrate control and air 
pollution control device efficiencies that are the basis of the 
emission standard from which you seek an alternative). If you still do 
not achieve the emission standards when operating under these 
conditions, you may be eligible for the alternative standard (provided 
you further demonstrate that you meet the additional eligibility 
requirements discussed below). If you choose to conduct this 
performance test after your compliance date, you should first obtain 
approval to temporarily exceed the emission standards (for testing 
purposes only) to make this demonstration, otherwise you may be subject 
to enforcement action.
    In addition, you must make a showing of adequate system removal 
efficiency to be eligible for an alternative standard for semivolatile 
metal, low volatile metal, or hydrochloric acid/chlorine gas. This 
requirement provides a check to ensure that you are exceeding the 
emission standard solely because of raw material contributions to the 
emissions, and not because of poor system removal efficiency for the 
hazardous air pollutants for which you are seeking relief. (It is 
possible that poor system removal efficiencies for these hazardous air 
pollutants result in emissions that are higher than the emission 
standards, even though the particulate matter emission standard is 
met.) This check could be done without the expense of a second 
performance test. The system removal efficiency achieved in the 
performance test described above could be calculated for the hazardous 
air pollutants at issue. You would then

[[Page 52965]]

multiply the MACT control hazardous waste feedrate level (or the 
feedrate level you choose to comply with) 262 for the same 
hazardous air pollutant by a factor of one minus the system removal 
efficiency. This estimated emission value would then be compared to the 
emission standard, and would have to be below the standard for you to 
qualify for the alternative standard.
---------------------------------------------------------------------------

    You may choose to comply with a hazardous waste feedrate limit 
that is lower than the MACT control levels required by this 
alternative standard.
---------------------------------------------------------------------------

    As discussed in the next section, this alternative standard 
requires you to use MACT control as defined in this rulemaking. For 
lightweight aggregate kilns, MACT control for chlorine is feedrate 
control and use of an air pollution control system that achieves a 
given system removal efficiency for chlorine. Thus, lightweight 
aggregate kilns that petition the Administrator for an alternative 
chlorine standard must also demonstrate, as part of a performance test, 
that it achieves a specified minimum system removal efficiency for 
chlorine. This eligibility requirement is identical to the above-
mentioned eligibility demonstration that requires sources to make a 
showing of adequate system removal efficiency, with the exception that 
here we specify the system removal efficiency that must be 
achieved.263
---------------------------------------------------------------------------

    \263\ The requirement to achieve an 85.0% and 99.6% chlorine 
system removal efficiency for existing and new lightweight aggregate 
kilns, respectively, together with the requirement to comply with a 
hazardous waste chlorine feedrate limitation, ensures that chlorine 
emissions attributable to hazardous waste are below the standards.
---------------------------------------------------------------------------

    For an alternative mercury standard, you do not have to perform a 
performance test demonstration and evaluation. We do not require this 
test because the mandatory hazardous waste mercury feedrate specified 
in Sec. 63.1206(b)(10) and (11) ensures that your hazardous waste 
mercury contribution to the emissions will always be below the mercury 
standard.264
---------------------------------------------------------------------------

    \264\ The MACT defining hazardous waste maximum theoretical 
emission concentration for mercury is less than mercury standard 
itself, thus hazardous waste mercury contributions to the emissions 
will always be below the standard.
---------------------------------------------------------------------------

    Finally, if you apply for semivolatile metal or low volatile metal 
alternative standards, you also must demonstrate, by submitting data or 
information, that increased chlorine levels associated with the burning 
of hazardous waste, as compared to nonhazardous waste operations, do 
not significantly increase metal emissions attributable to raw 
material. We expect that you will have to conduct two different 
emission tests to make this demonstration (although the number of tests 
should be determined on a site-specific basis). The first test is to 
determine metal emission concentrations when the kiln is burning 
conventional fuel with typical chlorine levels. The second test is to 
determine metal emissions when chlorine feedrates are equivalent to 
allowable chlorine feedrates when burning hazardous waste. You should 
structure these tests so that metal feedrates for both tests are 
equivalent. You would then compare metal emission data to determine if 
increased chlorine levels significantly affects raw material metal 
emissions.
    d. What Is the Format of the Alternative Standard? The alternative 
standard requires that you use MACT control, or better, as applicable 
to the standard for which you seek the alternative. MACT control, as 
previously discussed, consists of hazardous waste feed control plus 
(for all relevant hazardous air pollutants except mercury) further 
control via air pollution control devices. Cement kilns and lightweight 
aggregate kilns will first have to comply with a specified hazardous 
waste metal and chlorine feedrate limit, as defined by the MACT 
defining maximum theoretical emission concentration level for the 
applicable hazardous air pollutant or hazardous air pollutant group. 
This work practice is necessary because there is no other reliable 
means of measuring that hazardous air pollutants in hazardous waste are 
controlled to the MACT control levels, i.e., that hazardous air 
pollutants in raw material are the sole cause of not achieving the 
emission standard. (See CAA section 112(h).) To demonstrate control of 
hazardous air pollutant metals emissions to levels reflecting the air 
pollution control device component of MACT control, you must be in 
compliance with the particulate matter standard. Finally, we require 
lightweight aggregate kilns to use an air pollution control device that 
achieves the specified MACT control total chlorine removal efficiency. 
This work practice is necessary because there is no other way to 
measure whether the failure to achieve the chlorine emission standard 
is caused by chlorine levels in raw materials.265 See 
Sec. 63.1206(b)(10) and (11) for a list of the maximum achievable 
control technology requirements for purposes of this alternative 
standard.266
---------------------------------------------------------------------------

    \265\ There is no corresponding chlorine air pollution control 
device efficiency requirement for cement kilns since air pollution 
control is not the basis for MACT control of cement kiln chlorine 
emissions.
    \266\ See also ``Final Technical Support Document for Hazardous 
Waste Combustor MACT Standards, Volume IV: Selection of MACT 
Standards and Technologies'', Chapter 11, July 1999, for further 
discussion on how the maximum achievable control technologies were 
chosen for the hazardous air pollutants.
---------------------------------------------------------------------------

    There may be site-specific circumstances which require other 
provisions, imposed by the Administrator, in addition to the mandatory 
requirement to use MACT control. These provisions could be operating 
parameter requirements such as a further hazardous waste feedrate 
limitation. For instance, a kiln that petitions the Administrator for 
an alternative semivolatile emission standard may need to limit its 
hazardous waste chlorine feedrate to better assure that chlorine 
originating from the hazardous waste does not significantly affect 
semivolatile metal emissions attributable to the raw material. As 
discussed above, a kiln must demonstrate that increased chlorine levels 
from hazardous waste do not adversely affect raw material metal 
emissions to be eligible for this alternative standard. For this 
scenario, the alternative standard would be in the form of a 
semivolatile metal hazardous waste feedrate restriction which would 
require you to use MACT control, in addition to a hazardous waste 
chlorine feedrate limit.
    Additional provisions also could include emission limitations that 
differ from those included in today's rulemaking. For example, the 
Administrator may determine it appropriate to require you to comply 
with metal or chlorine emission limitations that are than the standards 
in this final rulemaking. The emission limitation would likely consider 
the elevated levels of metal or chlorine in your raw material. This 
type of emission limitation would be no different, except for the 
numerical difference than the emission limitations in today's rule 
because it would limit total metal and chlorine emissions while at the 
same time ensuring MACT control is used. If the Administrator 
determines that such an emission limitation is appropriate, you must 
comply with both a hazardous waste feedrate restriction, which requires 
you to use MACT control, and an emission limitation. A potential method 
of determining an appropriate emission limitation would be to base the 
limit on levels demonstrated in the comprehensive performance test.
    e. What Is the Process for an Alternative Standard Petition? If you 
are seeking alternative standards because raw materials cause you to 
exceed the standards, you must submit a petition request to the 
Administrator that includes your recommended alternative

[[Page 52966]]

standard provisions. At a minimum, your petition must include data or 
information which demonstrates that you meet the eligibility 
requirements and that ensure you use MACT control, as defined in 
today's rule.
    Until the authorized regulatory agency approves the provisions of 
the alternative standard in your petition (or establishes other 
alternative standards) and until you submit a revised NOC that 
incorporates the revised standards, you may not operate under your 
alternative standards in lieu of the applicable emission standards 
found in Secs. 63.1204 and 63.1205. We recommend that you submit a 
petition well in advance of your scheduled comprehensive performance 
test, perhaps including the petition together with your comprehensive 
performance test plan. You may need to submit this petition in phases 
to ultimately receive approval to operate pursuant to the alternative 
standard provisions, similar to the review process associated with 
performance test workplans and performance test reports. After initial 
approval, alternative standard petitions should be resubmitted every 
five years for review and approval, concurrent with subsequent future 
comprehensive performance tests, and should contain all pertinent 
information discussed above.
    You may find it necessary to complete any testing associated with 
documenting your eligibility requirements prior to your comprehensive 
performance test to determine if in fact you are eligible for this 
alternative standard, or you may choose to conduct this testing at the 
same time you conduct your comprehensive performance test. This should 
be determined on a site-specific basis, and will require coordination 
with the Administrator or Administrator's designee.
2. What Special Provisions Exist for an Alternative Mercury Standard 
for Kilns?
    See Sec. 63.1206(b)(10) and (11).
    a. What Happens if Mercury Is Historically Not Present at 
Detectible Levels? Situations may exist in which a kiln cannot comply 
with the mercury standard pursuant to the provisions in Sec. 63.1207(m) 
when using MACT control and when mercury is not present in the raw 
material at detectable levels.267 As a result, today's rule 
provides a petition process for an alternative mercury standard which 
only requires compliance with a hazardous waste mercury feedrate 
limitation, provided that historically mercury not been present in the 
raw material at detectable levels.
---------------------------------------------------------------------------

    \267\ The provisions in Sec. 63.1207(m) waive the requirement 
for you to conduct a performance test, and the requirement to set 
operating limits based on performance test data, provided you 
demonstrate that uncontrolled mercury emissions are below the 
emission standard (see Part 4, Section X.B). These provisions allow 
you to assume mercury is present at half the detection limit in the 
raw material, when a feedstream analysis determines that mercury is 
not present at detectable levels, when calculating your uncontrolled 
emissions.
---------------------------------------------------------------------------

    We received comments from the lightweight aggregate kiln industry 
expressing concern with the stringency of the mercury standard. 
Commenters oppose stringent mercury standards, in part, because of the 
difficulty of complying with day-to-day mercury feedrate limits. One 
potential problem cited pertains to raw material mercury detection 
limits. Commenters point out that if a kiln assumed mercury is present 
in the raw material at the detection limit, the resulting calculated 
uncontrolled mercury emission concentration could exceed, or be a 
significant percentage of, the mercury emission standard. This may 
prevent a kiln from complying with the mercury emission standard 
pursuant to the provisions of Sec. 63.1207(m), even though MACT control 
was used.
    We agree with commenters that this is a potential problem. In 
addition, it is not appropriate to implement a mercury standard 
compliance scheme that is relatively more burdensome for kilns with no 
mercury present in raw material, as compared to kilns with high levels 
of mercury in their raw material.268 Because we establish 
provisions that provide alternatives to kilns with high levels of 
mercury in the raw material, we are doing the same for those kilns 
which do not have mercury present in raw material at detectable levels.
---------------------------------------------------------------------------

    \268\ Kilns that comply with alternative mercury standards 
because of high mercury levels in their raw material are not 
required to monitor the mercury content of their raw material unless 
the Administrator requires this as an additional alternative 
standard requirement. Thus, absent the alternative mercury standard 
discussed in this section, a source that does not have mercury 
present in their mercury at detectable levels would be subject to 
more burdensome raw material feedstream analysis requirements.
---------------------------------------------------------------------------

    b. What Are the Alternative Standard Eligibility Requirements? To 
be eligible for this alternative mercury standard, you must submit data 
or information which demonstrates that historically mercury has not 
been present in your raw material at detectable levels. You do not need 
to show that mercury has never been present at detectable levels. The 
determination of whether your data and information sufficiently 
demonstrate that mercury has not historically been present in your raw 
material at detectable levels will be made on a site-specific basis. To 
assist in this determination, you also should provide information that 
describes the analytical methods (and their associated detection 
limits) used to measure mercury in the raw material, together with 
information describing how frequently you measured raw material mercury 
content.
    If you are granted this alternative standard, you will not be 
required to monitor mercury content in your raw material for compliance 
purposes. However, after initial approval, this alternative standard 
must be reapproved every five years (see discussion below). Therefore, 
you should develop a raw material mercury sampling and analysis program 
that can be used in future alternative mercury standard petition 
requests for the purpose of demonstrating that mercury has not 
historically been present in raw material at detectable levels.
    c. What Is the Format of Alternative Mercury Standard? The 
alternative standard requires you to use MACT control for mercury 
(i.e., the level of hazardous waste feedrate control specified in 
today's rule). This alternative standard for mercury is conceptually 
identical to the emission standards in this final rule, because it 
requires the use of an equivalent level of hazardous air pollutant MACT 
control as compared to the MACT control used to determine the emission 
standards.
    The mercury feedrate control level will differ for new and existing 
sources, and will differ for cement kilns and lightweight aggregate 
kilns. See Sec. 63.1206(b) (10) and (11) for a list of the mercury 
hazardous waste feedrate control levels for purposes of this 
alternative standard.269
---------------------------------------------------------------------------

    \269\ Also see Final Technical Support Document for Hazardous 
Waste Combustor MACT Standards, Volume IV: Selection of MACT 
Standards and Technologies, Chapeter 11, July 1999, for further 
discussion on how the maximum achievable control technologies were 
chosen for mercury.
---------------------------------------------------------------------------

    d. What Is the Process for The Alternative Mercury Standard 
Petition? If you are seeking this alternative mercury standard, you 
must submit a petition request to the Administrator that includes the 
required information discussed above. You will not be allowed to 
operate under this alternative standard, in lieu of the applicable 
emission standards found in Secs. 63.1204 and 63.1205, unless and until 
the Administrator approves the provisions of this alternative standard 
and until you submit a revised NOC that incorporates this alternative 
standard.

[[Page 52967]]

We recommend that you submit these petitions well in advance of your 
scheduled comprehensive performance test, perhaps including the 
petition together with your comprehensive performance test plan. After 
initial approval, alternative standard petitions should be resubmitted 
every five years for review and approval, concurrent with subsequent 
future comprehensive performance tests, and should contain all 
pertinent information discussed above.
B. Under What Conditions Can the Performance Testing Requirements Be 
Waived? See Sec. 63.1207(m).
    In the April 1996 NPRM, we proposed a waiver of performance testing 
requirements for sources that feed low levels of mercury, semivolatile 
metal, low volatile metal, or chlorine (see 61 FR at 17447). Under the 
proposed waiver, a source would be required to assume that all mercury,