TECHNICAL BACKGROUND DOCUMENT TO SUPPORT RULEMAKING PURSUANT TO CERCLA SECTION 103(f)(2) A Report to RELEASES CONTROL BRANCH OFFICE OF RESEARCH AND DEVELOPMENT And to EMERGENCY RESPONSE DIVISION OFFICE OF SOLID WASTE AND EMERGENCY RESPONSE U.S. ENVIRONMENTAL PROTECTION AGENCY Under Contract No. 68-03-3182 Prepared by ENVIRONMENTAL MONITORING & SERVICES, INCORPORATED A WHOLLY OWNED SUBSIDIARY OF COMBUSTION ENGINEERING, INC. NEWBURY PARK, CALIFORNIA 91320 June 1986 DRAFT DO NOT CITE OR QUOTE ------- 54019861 TECHNICAL BACKGROUND DOCUMENT TO SUPPORT RULEMAKING PURSUANT TO CERCLA SECTION 103(f)(2) A Report to RELEASES CONTROL BRANCH OFFICE OF RESEARCH AND DEVELOPMENT And to EMERGENCY RESPONSE DIVISION OFFICE OF SOLID VASTE AND EMERGENCY RESPONSE U.S. ENVIRONMENTAL PROTECTION AGENCY Under Contract No. 68-03-3182 Prepared by ENVIRONMENTAL MONITORING & SERVICES, INCORPORATED A WHOLLY OWNED SUBSIDIARY OF COMBUSTION ENGINEERING, INC. NEWBURY PARK, CALIFORNIA 91320 June 1986 DRAFT DO NOT CITE OR QUOTE ------- DISCLAIMER This document is a preliminary draft. It has not been formally released by the U.S. Environmental Protection Agency and should not at this stage be construed to represent Agency policy. It is being circulated for comments on its technical merit and policy implications. ------- FOREWORD This report was submitted to the Releases Control Branch, formerly the Oil and Hazardous Materials Spills Branch, Office of Research and Development, Edison, New Jersey, and the Emergency Response Division of the Office of Emergency and Remedial Response of the U.S. Environmental Protection Agency (EPA) to provide the technical base for the proposed rule on notification of continuous releases under CERCLA Section 103(f)(2). Comments on possible approaches under consideration by the EPA had been solicited in the preamble to a May 25, 1983 (NPRM) (48 FR 23559). This report was prepared by the technical staff of Environmental Monitoring & Services, Incorporated (EMSI), a wholly owned subsidiary of Combustion Engineering. Principal authors are Mr. Franz J. Burmann, Dr. Milton Kirsch and Dr. Alan Messing. Much of the material contained herein was presented in preliminary drafts by the staff of ICF, Inc., EMSI's principal subcontractor. The guidance of many members of the EPA staff was provided not only in a continuous fashion, but also in special working group meetings and by written and oral comments on the various drafts. The assistance of Mr. John Riley, Dr K. Jack Kooyoomjian, Ms. Barbara Hostage and Ms. Nancy Parkinson of the Emergency Response Division was especially helpful. The EPA project officer for this project is Mr. Rich Field of the Release Control Branch, Edison, New Jersey, who provided invaluable guidance during the project. The support and helpful suggestions of Mr. James Janis, Mr. Michael Goldman, Mr. James Bunting and other staff members of ICF, Incorporated, subcontractor to Environmental Monitoring & Services, Incorporated, are acknowledged. ------- TABLE OF CONTENTS EXECUTIVE SUMMARY ES-1 PURPOSE OF CONTINUOUS RELEASE REGULATION 1-1 OVERALL PURPOSE OF CERCLA SECTION 102 AND 103 1-1 NEED FOR CONTINUOUS RELEASE REGULATION 1-2 Need for Authority to Regulate Under CERCLA Section 103(f)(2) 1-5 Need for Clarifying Statutory Language 1-5 Provision of Civil Penalties in the Superfund Improvement Act 1-6 Usefulness of Continuous Release Data 1-7 EMISSIONS AND DISCHARGES FROM OPERATING FACILITIES 1-7 RATIONALE FOR NOTICE OF PROPOSED RULEMAKING 2-1 THE AFFECTED COMMUNITY 2-1 DEFINITION OF CONTINUOUS RELEASE 2-8 INTERPRETATION OF "STABLE IN QUANTITY AND RATE" 2-9 INTERPRETATION OF "FOR A PERIOD SUFFICIENT TO ESTABLISH THE CONTINUITY, QUANTITY AND REGULARITY OF THE RELEASE" ...2-10 INTERPRETATION OF "STATISTICALLY SIGNIFICANT INCREASE" (SSI) 2-11 The Nonparametric Test 2-13 Other Tests Which Were Considered But Rejected 2-15 Control Chart Test 2-15 Dixon Criterion 2-17 Grubb' s Test 2-19 Student t Test 2-20 INITIAL, ANNUAL, AND SSI REPORTING 2-22 Annual Reporting of a Continuous Release 2-22 Notification of a Statistically Significant Increase (SSI) 2-23 BASIS FOR ADDING TOGETHER CONCURRENT RELEASES 2-25 NEED TO MEET ALL FOREGOING REQUIREMENTS 2-25 DATA SOURCES EXAMINED TO ASSIST IN RULEMAKING 2-25 NPDES Permit Compliance Reports —2-26 Philadelphia Air Pollution Control Board Reports 2-26 Office of Air Quality Planning and Standards (OAQPS) Data Review 2-27 Summary 2-27 REFERENCES , 3-1 APPENDIX A A-l APPENDIX B B-l APPENDIX C C-l ------- LIST OF TABLES TABLE 2-1 - RELEASES REPORTED TO PHILADELPHIA AIR CONTROL BOARD THAT ARE POTENTIAL CANDIDATES FOR REDUCED REPORTING UNDER CERCLA SECTION 103(F)(2) 2-4 TABLE C-l DIXON CRITERION REPORTING TRIGGERS C-8 TABLE C-2 GRUBB'S TEST REPORTING TRIGGERS C-9 TABLE C-3 STUDENT'S t TEST REPORTING TRIGGERS C-10 ii ------- EXECUTIVE SUMMARY Section 103(a) of the Comprehensive Environmental Response, Compensation and Liability Act of 1980 (CERCLA or "Superfund") requires immediate notification from any person in charge of a vessel or an offshore or an onshore facility who releases an amount of a hazardous substance equal to or greater than its reportable quantity (RQ). Under CERCLA Section 102(b), the RQ of any hazardous substance defined in CERCLA Section 101(14) is 1 pound unless a different RQ has been established pursuant to Section 311(b)(4) of the Federal Water Pollution Control Act also known as the Clean Water Act (CWA). These are the statutory RQs for the hazardous substances unless and until the Environmental Protection Agency (EPA) promulgates regulations adjusting them. CERCLA also permits the EPA to establish a single RQ for each hazardous substance regardless of the environmental medium into which the substance is released. Section 103(f)(2) of CERCLA provides a mechanism for those who release CERCLA hazardous substances on a continuous basis to reduce their reporting burden. The proposed regulation on the reduced reporting requirements for continuous releases of hazardous substances clarifies the undefined terms in CERCLA Section 103(f)(2) so that the regulated community is fully apprised of its responsibility under the statute. "Continuous" is defined to include "during operating hours" and "during regularly occurring batch processes" as well as "without interruption or abatement." To qualify for the reduced notification provisions of CERCLA Section 103(f)(2), it must be demonstrated that the release is "continuous" and "stable in quantity and rate." In addition, the release must either be; 1) ESI ------- from a facility for which notification has been given under Section 103(c); or 2) one for which notification has been given under Section 103(a) or 103(b) for a "period sufficient to establish the continuity, quantity and regularity of such release." Documents and other relevant information used to support this qualification are to be kept on file at the facility where they may be inspected by EPA as necessary. The required annual notification shall be made in writing to the appropriate EPA Regional Administrator. For each hazardous substance for which the reduced reporting requirement is claimed, the annual report must include: o the identity of the hazardous substance, the Chemical Abstracts Service Registry Number, the media affected, the annual quantity released in pounds; o the number of times the amount released in a 24-hour period exceeded the reportable quantity; and o the size of both the mean release and of the largest release in pounds. Notification of a statistically significant increase, SSI, in the quantity released shall be made to the National Response Center as set forth in 40 CFR Part 302.6. Determination that a release represents an SSI shall be made using a simple Nonparametric test described herein, or a superior alternative test provided by the releasor and approved by the EPA Administrator. Multiple concurrent releases of the same hazardous substance occurring at various locations within a single facility shall be added together in deciding whether the releases constitute a continuous release or a SSI. ES2 ------- The proposed regulation is designed to provide the regulated community and the Agency with the maximum flexibility and minimum reporting burden while ensuring that sufficient information and timely notice is given to the Federal Government so that a response may be taken, if necessary, to protect the public health, welfare and the environment consistent with the CERCLA statutory mandate, and the National Contingency Plan (NCP). ESS ------- SECTION 1 PURPOSE OF CONTINUOUS RELEASE REGULATION OVERALL PURPOSE OF CERCLA SECTIONS 102 AND 103 The Comprehensive Environmental Response Compensation and Liability Act of 1980 (CERCLA or the Act) is an Act "to provide for liability, compensation, cleanup, and emergency response for hazardous substances released into the environment and the cleanup of inactive hazardous waste disposal sites." Section 101(14) of the Act defines a set of "hazardous substances" by reference to other environmental statutes, with authority also residing with EPA to designate additional hazardous substances under CERCLA Section 102. The CERCLA hazardous substance list currently contains 717 substances. Sections 103(a) and (b) of CERCLA require that a person in charge of a vessel or a facility from which a hazardous substance is released into the environment in an amount equal to or greater than its reportable quantity (RQ) must immediately notify the National Response Center (NRC). Under CERCLA Section 102(b), the RQ of any hazardous substance defined in CERCLA Section 101(14) is 1 pound, except for those hazardous substances for which RQs have been established pursuant to Section 311(b)(4) of the Clean Water Act (CWA). Section 102(a) authorizes EPA to adjust RQs for hazardous substances if deemed appropriate. The primary purpose of the CERCLA notification requirement is to ensure that releasers notify the government so that the need for a federal response can be evaluated and any necessary response can be undertaken in a timely fashion to protect the public health or welfare or the environment. Details of the notification requirements for release of hazardous substances appeared in a notice of proposed rulemaking (NPRM) on May 25, 1983. (48 FR 23552) and in a final rule on! April 4, 1985 (50 FR 13456). 1-1 ------- NEED FOR CONTINUOUS RELEASE REGULATION Section 103(f)(2) of CERCLA modifies the general notification requirements of Sections 103(a) and 103(b) for certain releases. Releases may be reported less frequently than otherwise would be required if they are "continuous," "stable in quantity and rate," and notification has been given under Sections 103(a) and (b) "for a period sufficient to establish the continuity, quantity, and regularity" of the release or under Section 103(c), which relates to notification of the existence of certain facilities that are or have been used for storage, treatment, or disposal of hazardous wastes. The alternative reduced reporting requirements imposed by Section 103(f)(2) are that notification of releases that are continuous and stable in quantity and rate must be given "annually, or at such time as there is any statistically significant increase" in the quantity of the hazardous substance being released. The main function of CERCLA's notification requirements is to alert government officials to the existence of a situation that may require a government response to protect the public health or welfare or the environment. Because episodic releases are almost always "unanticipated," they must be reported as they occur. Continuous releases, on the other hand, may be predictable and can either be literally continuous or recurring. Congress recognized in Section 103(f)(2) that CERCLA's notification objectives would be satisfied by less frequent reporting in the case of continuous releases. Thus, instead of reporting every release as it occurs, persons in charge are allowed to report certain continuous releases less often under Section 103(f)(2). 1-2 ------- The purpose of this section of the act is to reduce unnecessary reports of releases. This provision is designed to decrease the burden of reporting on both the regulated community and on the Federal Government. The rationale for this approach is that when a release occurs regularly and in relatively stable amounts, the Agency does not need to be notified each time such a release occurs to have the information necessary to decide whether a response to the release is necessary. In addition, this provision allows attention to be focused on those releases most likely to require a response by the government. Section 103(f)(2) provides, however, that notification must still be given (1) under Section 103(c) or "for a period sufficient to establish the continuity, quantity, and regularity of such release" and (2) "annually, or at such time as there is any statistically significant increase in the quantity of any hazardous substance or constituent thereof released, above that previously reported or occurring." These reports will provide EPA with sufficient information to evaluate a release to determine whether a response should be taken to protect public health, welfare and the environment. The annual report of continuous release will be evaluated by the Regional Administrator of the EPA region in which the releasing facility or vessel is located in order to determine whether or not a .response action is necessary. If no response is found necessary at this point, the notification of each statistically significant increase to the NRC will provide another opportunity for evaluation of that release and the necessity for response. 1-3 ------- The terms in Section 103(f)(2), continuous; stable in quantity and rate; for a period sufficient to establish the continuity, quantity and regularity of the release; and statistically significant increase, were not defined in CERCLA or in its legislative history. On May 25, 1983, EPA published a Notice of Proposed Rulemaking (NPRM) in 48 FR 23552, to clarify procedures for reporting releases and to adjust reportable quantities for 387 of the then 698 CERCLA hazardous substances (see Appendix A). In the preamble to that proposed rule, the Agency discussed the reduced reporting requirements for continuous releases and set forth a number of ideas that were under consideration in the development of EPA's position on such releases. The EPA specifically requested comments on the most feasible approach for continuous release notification, the information to be required, what the Agency should consider as a statistically significant increase in the release, and any other relevant issues. The Agency stated at that time it intended that such information would enable it to develop a system which would impose a minimal burden on both the regulated community and the government, while achieving the underlying statutory objectives. The EPA received 45 letters with comments on the reduced reporting requirement for continuous releases. (See Appendix B for a summary of these comments and a list of commenters.) In the preamble to a final rule adjusting RQs published on April 4, 1985 (50 FR 13456), the EPA noted that due to the complexity of the issues involved, the Agency had decided to study the continuous release reduced reporting requirement further. The Agency's definition of these terms associated with continuous releases are defined in Section 2 of this document and in the proposed rule. 1-4 ------- Need for Authority to Regulate Under CERCLA Section 103(f)(2) For many of the provisions of CERCLA that mandated action by the President, rulemaking authority was delegated through Executive Order 12316 either to the Administrator of the EPA, the Attorney General or other officials of the Executive Branch. Since Section 103(f)(2) was assumed originally to be self-implementing, no rulemaking authority was delegated to EPA for this section. Not only has the regulatd community expressed uncertainty in the operation of the reporting requirements in Section 103(f)(2) in comments submitted on the May 25, 1983 NPRM, but the Agency has received many letters and telephone calls requesting clarification of the provision. Furthermore, letters received by the Agency that assert to be annual notifications of continuous releases significantly lack uniformity in substance, level of detail, and form. The apparent confusion within the regulated community in regard to this section of the statute shows the need for regulation and, in turn, the need for rulemaking authority for the EPA to promulgate a regulation pursuant to this subsection. Need for Clarifying Statutory Language Because so many terms are undefined, and because of the desirability of exacting uniform notification requirements under Section 103(f)(2), clarification of the statutory language is essential for the regulated community to understand fully what and how continuous releases are to be reported to the NRC. The primary purposes of this document and of the proposed rule it supports, are: 1-5 ------- o to explain the meaning of the undefined terms in Section 103------- Usefulness of Continuous Release Data The major reason for a continuous release regulation is to clarify how the regulated community can comply with the provisions of CERCLA Section 103(f)(2) and thereby take advantage of the reduced reporting requirements for releases of RQs of hazardous substances under the circumstances described in that subsection of the statute. Peripherally, increased use of the reduced reporting requirement under CERCLA Section 103(f)(2) should supply, in the form of the required annual reports, data useful to air and water permit programs in determining levels of environmental pollution, and information for the Chemical Emergency Preparedness Program (CEPP). The EPA developed the CEPP to address accidental releases of acutely toxic chemicals as part of a comprehensive strategy to deal with the problem of air toxics in the environment. A notice announcing the availability of the CEPP Interim Guidance appeared at 50 FR 51451, December 17, 1985. This document is consistent with the right-to-know provisions of the CERCLA Reauthorization Bills and with the desirability of local contingency planning to handle releases for air toxic substances. EMISSIONS AND DISCHARGES FROM OPERATING FACILITIES Continuous releases may occur in operating facilities as a result of emissions from stationary and rotating seals. Pressure vessels, pipe joints, and pump seals are a few sources of emissions that contribute to continuous releases. Chimney stacks, scrubber vents, and baghouses may be larger contributors. Discharges of liquid wastes to Publicly Owned Treatment Works (POTWs) or waterways may contribute other continuous releases. 1-7 ------- Continuous releases may be expected to increase with time as a facility ages and fittings of all kinds loosen. Without increasing vigilance and continued improvements in maintenance, the sources of continuous releases will increase. Gradual creep in the quantity released continuously is part of normal operation. 1-8 ------- SECTION 2 RATIONALE FOR NOTICE OF PROPOSED RULEMAKING The large number and wide variety of comments received in response to the May 25, 1983 NPRM suggested that clarification was essential for uniform interpretation of CERCLA Section 103(f)(2). Several comments requested that this subsection be implemented soon to avoid the ambiguity that was perceived. Some commenters also attempted to associate the "federally permitted release" exemption from notification defined in CERCLA Section 101(10) with the reduced notification requirement allowed by CERCLA Section 103(f)(2) for continuous releases. The EPA is hoping to make perfectly clear that a federally permitted release as defined in CERCLA Section 101(10) is indeed exempt from the reporting requirements of CERCLA Section 103(a) and (b). The reduced reporting allowed by CERCLA Section 103(f)(2) is applicable to releases that are not federally permitted as defined by CERCLA Section 101(10). THE AFFECTED COMMUNITY Since the enactment of CERCLA in 1980, it has become apparent that industry is somewhat confused concerning the types of release that qualify for the reduced reporting requirement under CERCLA Section 103(f)(2). Thirteen applications have been submitted to the Agency from the regulated community since enactment of CERCLA requesting continuous release status and clarification of the law concerning continuous releases. In addition, a total of 45 public comment letters were received in the months of July and August, 1983, in response to the continuous release discussion of the May 25, 1983 NPRM on CERCLA notification requirements. This number of comments represents only a tiny fraction of those potentially able to claim reduced reporting status under CERCLA Section 103(f)(2). ------- The modification to the general reporting requirements of CERCLA allowed by Section 103(f)(2) includes releases that are "continuous," "stable in quantity and rate," and for which notification has been given under Section 103(c) of CERCLA or under Sections 103(a) and (b) of CERCLA "for a period sufficient to establish the continuity, quantity, and regularity" of the releases. In addition, in order for the facility to be affected by the provisions of Section 103(f)(2), the release: o must contain a CERCLA hazardous substance; o must not be a federally permitted release as defined in Section 101(10) of CERCLA; and o must equal or exceed the RQ for the hazardous substance in a 24-hour period. The Agency intends to publish a proposed rule which will address the federally permitted release exemption. The types of facilities which are included in this estimate and which will be affected by this proposed regulation include chemical and petrochemical manufacturers, metal smelters, dry cleaning establishments, users of CERCLA hazardous substances, such as aerospace industries, ordinance manufacturers, insesticide distributors, pharmaceutical manufacturers, degreasers, and others. 2-2 ------- Continuous releases occur as a result of processing, storing and transferring hazardous substances. Indeed many releases are a result of fugitive emissions. An analysis of the Philadelphia Air Pollution Control Board data is shown in Table 2-1, which illustrates the types of hazardous substances and sources of their release. These data show that over a 24-hour period the RQs of many substances are exceeded, in some cases by a great deal. Industrial practices involving solvent degreasers are obvious examples of continuous releases of hazardous substances. . In many industries, Metal-fabricated articles must be washed or degreased before electroplating, painting or other surface finishing. Most degreasing operations are carried out in units, called degreasers, in which a chlorinated solvent, such as trichloroethylene or perchloroethylene, either in the gaseous or liquid state, is used to wash the parts free of grease and oil. Some measureable amounts of solvent are released as a vapor or a liquid from even the smallest degreaser, and the large number of these units in any single big manufacturing organization makes the combined release substantial. In the manufacture of rocket engines vapor degreasers are used. Parts to be plated are suspended in conveyorized units. Solvent is vaporized and rises to fill the vapor space below water cooled condensers. At the condenser there is a definite vapor line. Condensed solvent runs through a collection trough and into a storage tank for later reuse. Articles to be degreased are conveyed into the vapor space. Vapors condense on the metal parts and hot condensate rinses oil and grease into the receptacle. It has been reported that daily emissions of solvents (trichloroethylene and/or perchloroethylene) vary from a few pounds to as high as 1300 pounds. Solvent is lost from the degreaser ------- TABLE 2-1 RELEASES REPORTED TO PHILADELPHIA AIR CONTROL BOARD THAT ARE POTENTIAL CANDIDATES FOR REDUCED REPORTING UNDER CERCLA SECTION 103(f)(2) Hazardous Substance Company Source Reported RQ Maximum Daily (Promulga- Release ted or Pounds Proposed Pounds) Benzene Carbon tetrachloride Carbon tetrachloride Chloroform Chloroform Chromium & compounds Ethylene oxide Ethylene oxide Lead Lead Lead Methyl chloride Methylene chloride Methylene chloride Nickel & compounds Gulf Refining Smithkline Smithkline Smithkline Smithkline Gulf Refining Penna. Hospital St. Agnes Med. Center Halpern Stein Assoc. lead Metal Bank Rohm & Haas Publicker Smithkline Gulf floating roof storage tanks fan exhauster from pellet pans fan & plenum from pelltet & copper pans fan exhauster from pellet pans fan & plenum from pellet & copper pans cooling tower gas sterilizers central supply baghouse Barton operation wire incineration process vents storage tank vent film coating stacks 100 50 100 106 351 17 18.5 U.33 1 10.32 2.8 210 1,355 1,295 20 10 10 10 10 10 1 10 10 1 1 1 100 1000 1000 10 7-4 ------- Contd. >• Hazardous Substance Company Source Reported RQ Maximum Daily (Promulga- Release ted or Pounds Proposed Pounds) Nickel & compounds General Electric metallizing 17.6 10 spray Perchloroethylene Coyne dry cleaning 165 100 Perchloroethylene Devon evap. dry 860 100 cleaning Trichloroethylene Botany dry cleaning 681 100 2-5 ------- tanks in essentially two ways: (1) vaporization (including diffusion) from the unit, and carryout with degreased articles. Estimates of 0.05 pound of solvent lost by vaporization per hour per square foot of open tank area have been made (Ref. 2). Although the releases discussed above exceed the RQs, the hazards posed by the operation need not be assessed every day by an On-Scene Coordinator (OSC). Indeed, the OSC needs to know the daily output only once a year to determine if a government response is necessary. However, if that daily release exceeds the anticipated value then the OSC must be apprised so that he can determine appropriate response action, if any. Because of the difficulty in obtaining the precise number of facilities continuously releasing CERCLA hazardous substances into any of the environmental media, it has been necessary to make estimates. The estimates are based upon the number of facilities regulated under the Clean Water Act, the Clean Air Act, and the Resource Conservation and Recovery Act, but do not include non-regulatd facilities that continuously release CERCLA hazardous substances for which the Agency does not have any data. Regardless, the Agency has attempted to make reasonable estimates approximating the number of facilities that may be affected by the proposed continuous release reporting regulation. The companion Regulatory Impact Analysis (RIA) to this report estimates that approximately 15,300 facilities in the United States which discharge CERCLA hazardous substances on a continuous basis to surface water in an amount equaling.or exceeding an RQ may qualify for CERCLA Section 103(f)(2) reduced reporting. In addition, the RIA estimates that 9,700 facilities releasing CERCLA hazardous ------- substances to air and 1000 facilities releasing CERCLA hazardous substances to groundwater may also qualify for the reduced reporting provisions of CERCLA Section 103(f)(2). The Federal government is also likely to benefit from the reduced reporting requirements set forth in the continuous release regulation. In the absence of such a regulation, continuous releasers must notify the National Response Center each time a release of a hazardous substance equals or exceeds its RQ until a demonstration has been made that the release is "continuous" and "stable in quantity and rate." Once the above mentioned demonstration has been made, telephone notification to the National Response Center will be required when there is a statistically significant increase in the continuous release. A written report containing summary information as described below must be sent to the EPA Regional Administrator. The EPA Regional Administrator will receive and process these annual reports. The National Response Center will also receive and process each telephone notification of a statistically significant increase in these releases. Though this represents an increase in the work load for the EPA regional office, this should be more than offset by the dramatic decrease in reports of individual releases to the National Response Center, and lead to a net cost-savings to Government. 1. Processing by the National Response Center includes the relay of information on the statistically significant increase to the appropriate EPA Regional Office or U.S. Coast Guard District Office for response. ------- DEFINITION OF CONTINUOUS RELEASE Section 103(f)(2) of CERCLA provides a reduced reporting requirement for releases which are "continuous" and "stable in quantity and rate." In order to demonstrate that the release is continuous in nature and stable in quantity and rate and therefore eligible for the reduced reporting requirement, a releaser must have: o notified the Administrator under Section 103(c) of CERCLA; or o notified the National Response Center under Sections 103(a) and (b) of CERCLA "for a period sufficient to establish the continuity, quantity, and regularity of the release." Releases that meet these criteria need only be reported annually, or when a statistically significant increase in the quantity of release occurs. Members of the regulated community have expressed concern in their comments on the May 25, 1983 NPRM over the definition of "continuous." (See Appendix B). Many commenters expressed the desire that the word "continuous" be defined as broadly as possible and that it contain any anticipated routine intermittent release that is necessarily associated with a manufacturing or treatment process. In response to these concerns, the EPA has developed a proposed regulation that would clarify the meaning and operation of the key terms and phrases of the reduced reporting requirement for continuous releases, but has chosen not to broaden the definition to include all "anticipated intermittent" releases. The Agency feels that such a broad definition may include releases that could involve a large quantity of hazardous substance (e.g., popping off of a relief valve) that are much larger than the quantity normally released. Indeed, since "accidents are bound to happen," and are thus anticipated, such releases of an RQ or more 2-8 ------- of hazardous substance might not be reported to the NRC resulting in a potentially significant threat to human health, welfare, or the environment. For this reason, the definition of a continuous release is, therefore, limited to a release that is: o literally continuous, without interruption or abatement; o continuous during operating hours; or o continuous during regularly occurring batch processes. Releases that are anticipated are not to be considered continuous for that reason alone, but must conform to the above definition. INTERPRETATION OF "STABLE IN QUANTITY AND RATE" The EPA has decided that the phrase "stable in quantity and rate" refers to a release that is predictable during normal operations and not the result of malfunction or upset. This interpretation of the phrase in qualitative terms is preferable to a quantitative definition because it allows the flexibility demanded by the wide variety of facilities, plants and vessels from which continuous releases can occur. Quantitative definitions based on an absolute magnitude of variation (e.g., 5 pounds, 10 pounds or one RQ from the daily average) or a relative magnitude of variation such as 5 or 10 percent from the daily average could be developed but would suffer from lack of flexibility and would be much more costly and burdensome both to the regulatd community and to the Agency with no apparent increased benefit in protecting human health or the environment. 2-9 ------- The person-in-charge must report in writing at least annually to the EPA Regional Administrator the magnitude of the continuous release that is stable in quantity and rate. A decision can be made at this time if a response to this quantity and rate is necessary. If no response is necessary to this base rate of release, then a response decision can be made at the time a statistically significant increase in the magnitude of the release is reported to the National Response Center. EPA has chosen a qualitative measure of the phrase "stable in quantity and rate." This is defined as a release that is predictable during normal operations and is not the result of malfunctions or upsets. INTERPRETATION OF "FOR A PERIOD SUFFICIENT TO ESTABLISH THE CONTINUITY, QUANTITY AND REGULARITY OF THE RELEASE" In the May 25, 1983 NPRM, the Agency asked for comments on approaches to defining the "period sufficient to establish the continuity, quantity and regularity of the release." Most commenters on this qualifiation for continuous release reporting were in favor of leaving the specific definition to the releaser on a case-by-case basis. Other commenters supported the establishment of a single time period during which all releases would need to be reported, while others proposed a specified number of release reports to determine whether the release was continuous and stable in quantity and rate. Since there are many types of facilities, plants and vessels involved, each of which differ in the kinds of operations which may produce releases, the EPA agrees that the person-in-charge may decide how to determine the "period 2-10 ------- sufficient." The Agency also believes that this period need not be long since the facility owner/operator is familiar with the operation and can draw on many sources to determine the "continuity, quantity and regularity of the release." It might be possible to use recent data calculated from mass balance, release monitoring, National Response Center reports or other sources for this determination. The Agency recognizes the need for flexibility because it would be impossible to anticipate all the possible conditions under which the various facilities operate that could legitimately claim the reduced reporting provisions of the continuous release exemption regulation. Just as the term "continuous" encompasses several types of operations and their associated releases, the "period sufficient" to demonstrate the continuity, stability and regularity must also be allowed to vary so as to be compatible with the type of release. In order to reduce the information collection burden on the regulated community and the EPA, the Agency has also decided to allow the owner/operator to record and file at the facility the data and other information upon which the demonstration that the release is continuous and stable in quantity and rate is based. The EPA reserves the right to inspect the releaser's file when necessary or desirable. INTERPRETATION OF "STATISTICALLY SIGNIFICANT INCREASE" (SSI) In the preamble to the May 25, 1983 proposed rule (48FR 23559), EPA reported that three alternatives were being considered for defining statistically significant increase: 2-11 ------- o Requiring reporting whenever a release falls outside some expected range based on statistical tests, such as the Student's t test; o Requiring reporting whenever the amount released exceeds the amount ordinarily released by some pre-established factor, such as 2, 5, or 10 times the daily average; or o Letting the releaser determine what is a statistically significant increase. Commenters were nearly equally divided in preferring each of these alternatives. The EPA has decided, however, to propose a method that is simple, understandable and easy to use in determing whether a SSI has been reached. This Nonparametric test identifies the top 5 percent of releases, and the Agency defines these to be statistically signifant increases. The Agency investigated a number of statistical tests including the Control r Chart test, Dixon Criterion, Grubb's test and Student's t test in order to determine their appropriateness for identifying a SSI. These tests are described below. The statistical tests described in this report do not represent an exhaustive list of potential statistical tests which might be suitable for identifying a SSI. As noted in the proposed continuous release reporting regulation, applicants who wish to use an alternative method may do so if they can demonstrate to the Agency's satisfaction that it is superior to the Nonparametric test (see Section 302.8(d) of the proposed rule). 2-12 ------- The Nonparametrie Test The Nonparametrie test is a simple method to determine whether an increase in a release is significant. The Agency defines a SSI to be any release that exceeds the largest of the 19 most recent releases. If fewer than 19 release events have been recorded, then the Agency defines a SSI to be the largest release to date. Each statistically significant increase in release, so defined, must then be reported immediately to the National Response Center. After the initial 19 releases, the newest release is compared to the immediately previous 19 releases in determining a statistically significant increase. For an example of the application of the Nonparametric test, assume that a facility has a daily release of a CERCLA hazardous substance with an RQ of 10 pounds. The following pattern' of daily releases are recorded. Release reports are made to the NRC as indicated. Event Event Event Event Event Event Event Event Event Event Event Event Event Event Event Event Event Event Event 1 = 2 = 3 = 4 = 5 = 6 = 7 = 8 = 9 = 10 = 11 = 12 = 13 = 14 = 15 = 16 = 17 = 18 = 19 = 50 70 40 150 60 10 80 20 70 50 100 50 60 30 70 90 40 95 100 pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds Report Report No Report Report No No No No No No No No No No No No No No No Report Report Report Report Report Report Report Report Report Report Report Report Report Report Report Event 20 = 50 pounds No Report Event 21 = 30 pounds No Report Event 22 = 40 pounds No Report Event 23 = 90 pounds No Report Event 24 = 60 pounds No Report Event 25 = 110 pounds Report — 1 7 ------- For release events 2 through 19, a release would be reported to the NRC if it exceeds the largest previous release event. Thus, events 2 and 4 would be reported to the NRC. Event 20 is then compared to releases in events 1 through 19 and is found not to be statistically significant because it is less than 150 pounds. Release event 20 is then added to the release history, and event 1 is deleted. Event 21 would be compared to releases in events 2 through 20. Because 30 pounds released in event 21 is less than 150 pounds in event 4 no report is required. Event 21 is next added to the release history and event 2 is deleted. For event 25, the relevant release history includes events 6 through 24. The 110 pound release in event 25 must be reported to the NRC because it exceeds the largest of the most recent 19 release events (i.e., 100 pounds). The principal advantage of the Nonparametric test is that it is insensitive to and indepenent of the underlying distribution of releases. Because the Agency has defined the largest five percent of all continuous releases as statistically significant increases in a release and therefore reportable to the National Response Center, it was necessary to select a test that was insensitive to the underlying distribution. The insensitivity of the Nonparametrict test to the underlying distribution of releases helps minimize underreporting and overreporting. The largest 5 percent of all releases from all facilities with continuous and stable releases of CERCLA hazardous substances will be reported to the National Response Center. ------- A second advantage of the Non parametric test is its simplicity. The test does not require any sophisticated calculations but is performed by recording releases and comparing the newest release to the largest of the preceding 19 releases. This analysis by inspection will not be burdensome to the regulated community. It is understood that no additional monitoring of emissions are required. Section 103(a) and (b) require notification of a release of a hazardous substance when the operator has knowledge that a release above athe RQ has occurred. Furthermore, if it was reasonable for the owner/operator to have knowledge of the release, then failure to report a release at or above the RQ may be a criminal offense. It follows, therefore, that industry practices involving handling, manufacturing, using and storing hazardous substances dictate knowledge of releases. Since the reporting requirements of Section 103 were self-implementary upon CERCLA enactment, then owners and operators of facilities involved with hazardous substances must have the capabilities of determining their daily releases in order to be in compliance with the law. Because the Nonparametric test is simple, non burdensome and implementable the Agency has selected this test as a means of defining statistically significant increase in quantity and rate. Other Tests Which Were Considered But Rejected Control Chart Test The Control Chart test requires a minimum history of 30 releases in order to be used effectively. A report is made when an observation exceeds 1.645 ------- Standard Deviations above the mean. If the new recorded release is less than 1.645 standard deviations above the mean (1.645*5), then the release is not considered a statistically significant increase and does not need to be reported to the National Response Center. The release is then added to the release history and a new mean and standard deviation are calculated. n - 1 n Where n is the number of releases; Y is the mean (average); and 2 1=1 represents the summation of the term in parentheses, which ranges from 1 to The Control Chart test will dependably identify the largest 5 percent of all releases only when the underlying distribution is normal. If the underlying distribution is lognormal, for example, the control chart test will trigger a" report for fewer than 5 percent of the releases. The number of reportable releases is very sensitive to the underlying distribution of releases. Because of this sensitivity, it is highly likely that this statistical test would result in significant under or overreporting of releases to the NRC, depending on the shape of each facility's release distribution. ------- Relative to the Nonparametrict test, the Control Chart test is more time consuming and difficult to perform. The test involves recalculation of the mean and standard deviation of the release data with each nonstatistically significant release. This recalculation (for some facilities, recalculation would be expected to occur quite often), could be time consuming and confusing, with the test yielding no apparent advantage over the Nonparametric test. Further, an alternative test to the Control Chart test would be required for facilities with fewer than 30 data points, resulting in a more complicated regualtory scheme. An example of the application of the Control Chart test can be found in Appendix C. Dixon Criterion The Dixon Criterion is a statistical test that can be used to test the largest value in a release distribution. The Dixon Criterion compares the newest release to the largest release in the data set and divides that difference by the difference between the newest release and the smallest release in the data set. That is: 2-17 ------- Where X is the most recent relese, X, _i\in the data set, X.. is the f smallest release in the data set, and C is the calculated value. The Dixon Criterion thus derived is then compared to a table relating reporting triggers to the number of releases in the data set. The Dixon Criterion table is reconstructed in Table C-l. If the calculated value is larger than the tabulated value, the release is considered statistically significant and must be reported to the National Response Center. An example of the use of the Dixon Criterion can be found in Appendix C. The Dixon Criterion has several disadvantages. First, it assumes normality of observations such that if the underlying distribution of releases is not normal, the test will not be sufficiently sensitive in identifying outliers in the distribution. Underreporting to the National Response Center of statistically significant increases in release will result. Second, if the largest release (X, _..) in the data set is actualy an extreme value, the test will mask the identity of other extreme values. Extreme values are more difficult to identify with smaller data sets and misclassification could .occur relatively often with use of the Dixon Criterion. The Dixon Criterion was not considered by the Agency for use with large data sets. It was only considered by the Agency for situations when fewer than 30 observations were available. After 30 data points were accumulated, the Aency would have required a differenct statistical test. Because such an approach would have employed a relatively complicated regulatory scheme, the use of the Dixon Criterion was rejected. _T 0 ------- Grubb's Test The Grubb's test is a statistical test used to identify outliers or anomalies. The Grubb's test involves the calculation of a test value and a comparison of this test value with the "Grubb's Statistic". The test value is calculated by subtracting the mean (average) of all past releases (except those determined to be statistically significant increases), from the newest release, and dividing this difference by the sample standard deviation of those same releases. That is, X - X Test Value • n S where X is the most recent release, X is the mean or average of all past statistically insignificant releases, and S is the standard deviation for all statistically insignificant releases. X and S are calculated using the formulas: + (X.-X)2 + ... + (X -X)2 X - -i £ i 11 and S - ' * 2 n n - 1 . h - 1 2-19 ------- where n represents the number of statistically insignificant releases in the data set. If the Test Value is less than the corresponding Grubb's Statistic shown in Table C-2, then the release is not considered a statistically significant increase and need not be reported to the National Response Center. If the Test Value is equal to or greater than the Grubb's Statistic in Table C-2, then the release must be reported to the National Response Center. After a report to the National Response Center, the value reported should not be included in future Grubb's test calculations. An example of the Grubb's test can be found in Appendix C. The Grubb's test assumes the underlying distribution is normal. If this is not the case, underreporting or overreporting to the National Response Center of statistically significant releases could result. As with the Dixon Criterion, the Grubb's test was not considered for use with large data sets. The Agency believed there were better tests available when a release history of 30 data points or more was available. The Grubb's test is sensitive to the underlying distribution of releases and does not perform so well as the Nonparametric test. A second statistical test would have been employed when sufficient data were accumulated, resulting in complicated the reporting requirements. Student t Test The Student t test is simlar to the Grubb's test. It involves the calculation of a test vlaue and a comparison of this test value with the t statistic. THe test value is calculated by subtracting the mean of all 2-20 ------- releases (excluding the newest release) from the newest release, and dividing this difference by the standard deviation. Student t Value where X is the most recent release, X is the mean or average of all releases excluding the newest release, and S is the sample standard deviation. X and S are calculated using the formulas described for the Grubb's test. The Student t test is identical to the Grubb's test except the mean and standard deviation do not include the newest release. The Student t value is then compared to the reporting triggers found in Table C-3. If the Student t value equals or exceeds the reporting trigger, the release is statistically significant and needs to be reported to the National Response Center. An example of the use of the Student t test can be found in Appendix C. The burden of continually recalculating the mean and standard deviation to formulate the t statistic for comparison to the tabulated value appears to be much greater than the burden associated with the Nonparametric test, it is a relatively well known statistical test, however, and facilities would be generally more familiar with it than with either the Dixon Criterion or the Grubb's test. 2-21 ------- INITIAL, ANNUAL, AND SSI REPORTING For those facilities which qualify for reduced reporting the Agency proposes that the annual report required pursuant to Section 103(f)(2)(B), be sent in writing to the Regional Administrator of the EPA for the Region in which the facility or vessel is located instead of to the NRC. The initial report is due within 90 days of the effective date of the rule. Details of the contents of these annual and initial reports are described below. Notice of an SSI is to be made to the NRC pursuant to Section 103(a) and (b) as soon as the person-in-charge has knowledge that such a release has occurred as set forth in 40CFR Section 302.6. Annual Reporting of a Continuous Release The annual notification is to include for each hazardous substance for which the CERCLA Section 103(f)(2) reduced reporting allowance is claimed— o the identify of the hazardous substance, the Chemical Abstracts Service Registry Number (CASRN) for the hazardous substance, the media affected, and the annual quantity released in pounds, o the number of times the amount of the release during any 24 hour period exceeded the reportable quantity, o the amount in pounds of the single largest daily release, and o the amount in pounds of the mean daily release. The Chemical Abstracts Service Registry Number (CASRN) is a unique identification number that is universally recognized in the chemical community. The number provides a useful code and insures accurate identification of the hazardous substance being released. ------- The Agency is requesting information on the media affected by the release. The medium or media into which the release occurs may be significant when evaluating whether the Federal government should respond to the release to protect human health, welfare or the environment. For example, a release of phosgene into air may be more hazardous than a release into water where phosgene is rapidly hydrolyzed into less toxic products. If possible, the releaser should not only identify the media to which the release is being made, but also identify the distribution of the hazardous substance into all media: air, surface water, groundwater and soil. The total quantity of the hazardous substance released annually is necessary to the Agency in assessing the potential long-term health and environmental impact of the release. The Agency will use this data to determine whether receiving waters are being overloaded or whether the air has become unhealthful requiring a response by the Federal government. Furthermore, the number of times the amount of the release during any 24 hour period exceeded the RQ will provide an additional trigger for investigating whether a response action should be taken by the Fedral government to protect human health, welfare and the environment. For example, if a facility is continuously releasing a hazardous substance that has an RQ of 1 pound (based on its potential carcinogenicity or chronic toxicity) at an amount that is 100 times this RQ on a daily basis, the Agency may determine that there is a need to respond to such a release. Notification of a Statistically Significant Increase (SSI) Notification of a SSI in the continuous release of a hazardous substance is to be made to the National Response Center as described in 40 CFR 302.6. The determination that a particular continuous release represents a SSI is to be based on: 2-23 ------- o a Nonparametric test which identifies the top 5 percent of releases as statistically significant, or o any alternative methodology specifically approved by EPA that can be demonstrated to the Agency's satisfaction to be superior to the Nonparametrict test. The Nonparametric test and the Agency's reasons for choosing it are described above. All SSIs in the amount of a continuous release that is stable in quantity and rate are to be reported to the NRC. Because SSIs represent the release of larger quantities of the hazardous substance resulting from an unusual occurence at the facility (e.g., a malfunction or upset condition), it is more likely that the Federal government will need to respond to the release to protect human health, welfare and the environment. Notifying the NRC initiates an assessment by the government of the circumstances surrounding the release and leads to the determination of the need for a Federal response. If the person in charge prefers an alternative test to the Nonparametric test to detect a SSI, the Agency may accept that alternative under the conditions below. EPA will require applicants who choose to use an alternatiave test to present two sets of background information to the EPA Administrator: o first, the applicant must describe any assumptions that its test would make regarding the distribution of releases that would be expected. EPA will consider the extent to which expectations regarding the release distribution are releastic and whether the alternative test depends on the accuracy of these expectations, o second, the applicant must present evidence that the alternative test is superior to the Nonparametric test described above for the applicants facility. The alternative test's effects on the rates of false positives and false negatives must be presented by the applicant. No alternative test is acceptable unless and until the EPA gives written approval. 2-24 ------- BASIS FOR ADDING TOGETHER CONCURRENT RELEASES Multiple concurrent releases of the same substance occurring at various locations within plants or installations upon contiguous grounds that are under common ownership or leasehold should be added together, both in determining whether the facility qualifies as a continuous releaser and in determining whether a SSI has occurred. Such additions are expected to smooth out variations that occur at single emission or discharge points such as valves, flanges, and unions. Realistic averaging of the releases to the environment takes into account the possibly large changes that may occur for a short period of time at a temporary leak in just one fitting or other unit of the entire plant. Such averaging also eliminates the otherwise burdensome need to keep detailed individual records for separate portions of a large facility. NEED TO MEET ALL FOREGOING REQUIREMENTS The reduced frequency of reporting is available only when all the conditions described in the above paragraphs of Section 2 are fully complied with. Failure to do so subjects the releaser to all the notification requirements for episodic releases of an RQ or more of a hazardous substance. DATA SOURCES EXAMINED TO ASSIST IN RULEMAKING In the development of proposed continuous release regulations, available data sources were examined in a search for relevant information on continuous releases. The information sought included ways of estimating the magnitude of the variation in the quantity of particular hazardous substances released to any medium, the time frame over which the release remained essentially constant, the frequency with which major excursions occurred in the amount released, and the size of those excursions. It was expected that the data could be used to determine the stability of the rate of release of some hazardous substances and possibly also the magnitude of 2-25 ------- significant increases in the quantity released. Although no precisely applicable data sources were available, a number of sources were investigated to determine what useful information could be extracted. They are described below. NPDES Permit Compliance Reports In the expectation that information on the constancy of discharge of some hazardous substance might be inferred from NPDES compliance reports, a printout of such reports covering a period of several months was obtained from EPA Region II. The reports were found to contain data that was very difficult to relate to the magnitude of discharge of individual hazardous substances or to the variation in the rates of discharge. Most of the data were expressed in terms of conventional pollutants, e.g., BOD, total suspended solids, rather than in terms of concentration or quantity of particular hazardous substances. Flow rate data were not generally relatable to concentration. Frequency of monitoring or measurement of any particular parameter was usually unstated and the measurements were apparently made at irregular intervals. Although the Agency hoped to utilize the data to define "stable in quantity and rate" and "statistically significant increase" it was not possible to determine which numerical values corresponded to variations within permitted limits that might be interpreted as continuous releases within normal operating conditions and which values represented out-of-limit upsets. Philadelphia Air Pollution Control Board Reports The possible applicability of data collected over a two-year period by the Philadelphia Air Pollution Control Board on 99 "toxic air contaminants" (84 of which were CERCLA hazardous substances) to continuous release regulation was examined in order to arrive at an estimate of continuous releasers in Philadelphia that could be extrapolated to the remainder of the United States. The maximum daily emission rate reported was compared with the proposed RQ for the CERCLA hazardous substances. It was assumed that releases which were equal to or larger than the proposed RQs might qualify 2-26 ------- for the Section 103(f)(2) reduced reporting requirement. No information existed on whether the reports represented releases which were "stable in quantity and rate" or whether notification had been given for a "period sufficient to establish the continuity, quantity and regularity." The data was very useful in assisting the Agency to estimate the number of facilities likely to qualify for reduced reporting. Office of Air Quality Planning and Standards (OAQPS) Data Review The OAQPS data provide information on air pollutants of concern to the Agency and are presented on a pollutant by pollutant basis. For each pollutant, facilities are identified which have emissions either from their stacks or from other source. The emissions data are presented as an annual total, with no information on the continuity (if any) of the process involved. Thus, only a single data point is available for each emission source, leaving no opportunity to determine the variability of the emission rate over time which would have been useful to the Agency in defining what is meant by the term "stable in quantity and rate." The data has proven useful, however, in identifying the universe of facilities that will benefit from a reduced reporting requirement for continuous releases. Summary As can be seen, the available data sources proved useful in identifying the need for a regulation leading to reduced reporting. The Agency could not use the data in deriving a quantitative definition of "stable in quantity and rate," and could not apply the data in deriving a percentage from the daily average for purposes of defining a statistically significant increase in the amount released. 2-27 ------- SECTION 3 REFERENCES ICF, Inc., Analysis of Continuous Release Notification Strategies for the Superfund Program, February 1, 1982. Air Pollution Engineering Manual, U.S. Department of Health, Education and Welfare, 1967. 3-1 ------- APPENDIX A TEXT OF MAY 25, 1983 NPRM PREAMBLE ON CONTINUOUS RELEASES 50 FEDERAL REGISTER 23559-60 A-l ------- D. Continuous Releases Section 103(f)(2) of CERCLA exempts certain releases from the general notification requirements of Section 103(a) and 103(b). Releases may be exempted if they are "continuous," "stable in quantity and rate," and notification has been given under Section 103(a) and (b) "for a period sufficient to establish the continuity, quantity, and regularity" of the release or under Section 103(c) (which relates to notification of the existence of certain facilities that are or have been used for storage, treatment, or disposal of hazardous wastes). Notification of continuous releases must be given "annually, or at such a time as there is any statistically significant increase" in the quantity of the hazardous substance being released. The main function of CERCLA's notification requirements is to alert government officials to the existence of a situation that may require a government response to protect the public health or welfare or the environment. Since episodic releases are by definition "unanticipated," they must be reported as they occur. Continuous releases, on the other hand, may be predictable and can either be literally continuous or recurring. Congress recognized in Section 103(f)(2) that CERCLA's objectives would be satisfied—in the case of continuous releases—by less frequent notification. Thus, instead of reporting every release as it occurs, persons in charge are allowed to report certain continuous releases less often under Section 103(f)(2). The purpose of this section is to reduce unnecessary reports of releases. The rationale for this approach is that when a release is regular and of stable quantity, the Agency does not need to be notified each time such a release occurs in order to have the information necessary to decide whether a response to the release is necessary. Section 103(f)(2) provides, however, that notification must still be given (1) under Section 103(c) or "for a period sufficient to establish the continuity, quantity, and regularity of such release" and (2) "annually, or at such time as there is any statistically significant increase in the quantity of any hazardous substance or constituent thereof released, above that previously reported or occurring." The Agency is considering development of a policy to identify: • The types of releases that qualify for the limited exemption under Section 103(f)(2), and • The Section 103(f)(2) notification scheme. Pending implementation of Section 103(f), EPA will focus its enforcement efforts on episodic releases (e.g., accidental, one-time, non-routine releases) exceeding the RQs proposed today which may present a serious risk of harm to human health or welfare or the environment, rather than on continuous releases (e.g., routine, continuous, or anticipated intermittent releases which are incidental to normal manufacturing or treatment processes or operations of facilities or vessels). A-2 ------- 1. Types of Releases. The Agency is considering defining releases that are "continuous" and "stable in quantity and rate" to include: • Literally continuous releases that enter the environment 24 hours a day, 365 days a year. Examples include hazardous substances leaking from pipes or lagoons into surface water, leaching into the soil or groundwater, or evaporating. • Release continuous during operating hours, such as releases from some continuous indiustrial processes; • Releases from batch operations; and • Routine, anticipated, intermittent releases of hazardous substances that are incidental to the normal manufacturing or treatment processes or operations of facilities or vessels. Examples include releases from relief valves, the maintenance of pollution control equipment, charging of coke oven batteries, and tank cleaning operations. In order for a continuous release which is stable in quantity and rate to be subject to Section 103(f)(2), notification must be given pursuant to Section 103(c), or under Section 103(a) and (b) for a "period sufficient to establish the continuity, quantity, and regularity" of the release. Reporting under Section 103(c) is addressed in 46 FR 22144 (April 15, 1981) and will not be discussed further here. The Agency is considering several approaches for defining the "period sufficient," including: • Establishing a single period (such as a week, a month, six months, or a year) of reporting that will be required for all releases; or • Specifying the number of reports, instead of a specific time period, to establish the "period sufficient." EPA solicits comments on these or any other approaches for implementing the Section 103(f)(2) notification provisions. 2. Section 103(f) Notification Scheme. The Agency intends to develop a notification system for continuous releases. An approach under consideration involves an annual written notification identifying the hazardous substance being released, the quantity and rate of such release, and the environmental media affected. The Agency does not anticipate that this written notification would be necessarily costly or elaborate; it does not anticipate that sophisticated analysis to determine the constituents of a release or that special monitoring to determine the quantity or rate of a release will be needde. A new notification (either written or by telephone) would be required if there is a statistically significant increase in the amount, or statistically significant change in the type of hazardous substance being released. EPA is also considering whether applications for federal permits containing information on the nature of these releases might suffice to show that a release qualifies for the continuous release exemption. A-3 ------- Three alternatives for defining "statistically significant increase" are under consideration. • Requiring reporting whenever a release falls outside some expected range based on statistical tests, such as the "Student's t" test; • Requiring reporting whenever the amount released exceeds the amount ordinarily released by some pre-established factor, such as 2, 5, or 10 times the daily averge; or • Letting the releaser determine what is a statistically significant increase. EPA requests comments specifically on the most feasible approach for continuous release notification, the information to be required, what the Agency should consider a statistically significant increase in the release, and any other relevant issues. EPA hopes that such information will enable it to develop a system which imposes a minimal burden on both the regulated community and the government, while achieving the underlying statutory objectives. A-4 ------- APPENDIX B SUMMARIES OF PUBLIC COMMENTS ON THE MAY 25, 1983 NPRM'S CONTINUOUS RELEASE PROVISIONS INTRODUCTION A total of 45 public comment letters were received by EPA during the months of July and August, 1983 in response to the continuous release provisions of the May 25, 1983 NPRM on Superfund notification requirements The majority of commenters were members of the regulated community, with four government agencies and two environmental groups comprising the balance. Most comments centered on the definition or implementation of such terms as "statistically significant," "period sufficient," "continuous," and "stable in quantity and rate." Most industry commenters urged EPA to adopt a broad interpretation of these terms to lessen the frequency of required reporting. Many industry commenters also recommended that responsibility for defining these conditions rest with the releaser. Comments by government agencies, environmental organizations, and one member of industry, however, indicated that this approach was unacceptable because of the inconsistencies and possible evasions that would result. Summaries of all of the comments are arranged in this Appendix according to the order in which the comments were received. KENDALL COMPANY Kendall suggested that "continuous releases covered by either state or federal permits should be left unchanged. There is no need for additional reporting." This comment seemed to confuse the reduced reporting requirement for continuous releases with the complete reporting exemption for federally permitted releases because any release covered by a federal permit is by definition a federally permitted release. In such instances, no reporting would be required unless the release exceeded the permitted level by an RQ or more. If such exceedences of permit conditions occurred repeatedly and the releaser could show continuity and regularity, the releases might be considered to be continuous releases. AIR PRODUCTS Air Products recommended that if a facility is covered by existing monitoring rules, it should not be covered by Section 103(f) of CERCLA, particularly in the case of continuous releases of volatile organic compounds (VOCs) from facilities governed by the SIP process or NSPS guidelines. Air Products requested that EPA make adjustments to 103(f) reporting requirements to reflect the wide variety of reporting rules already in place or in the process of rulemaking to avoid duplication of existing programs. B-l ------- VULCAN CHEMICALS Vulcan maintained that facilities which emit non-photoreactive VOCs, if not entirely exempt from the notification requirements of CERCLA under the federally permitted release provision, should at least fall within Section 103(f). To qualify for the continuous release exemption, such facilities should be required to file only one report stating that they emit these solvents. ADOLPH COORS Adolph Coors generally endorsed EPA's approach to the continuous release issue. Coors made the following recommendations: o EPA should give the releaser primary responsibility to determine the period of time sufficient to establish continuity. Coors suggested that such a practice will not lead to abuse because the Agency has adequate authority to request additional data, if required. o Whether there has been a "statistically significant increase" should be determined by the releaser. THE LIGHT COMPANY The Light Company supported EPA's proposal to emphasize reporting requirements for episodic releases exceeding RQs rather than continuous, routine, or anticipated intermittent releases incidental to normal operations. Such an approach would, in its opinion, reduce redundant reporting and promote efficiency in both the Agency and the regulated community. RSR CORPORATION RSR Corporation suggested that to avoid duplicative reporting requirements, continuous releases of substances subject to other legislation (e.g., CAA, CWA) should be governed by amendments to the appropriate legislation rather than by a new reporting system under CERCLA. Information contained in permits covering such releases should be sufficient to satisfy the notification requirements of Section 103(f)(2) of CERCLA. PENNZOIL Pennzoil approved of EPA's plan to broaden the definition of continuous release to include not only literally continuous releases but also those that are continuous during operating hours, those from batch operations, and routine, anticipated, or intermittent releases that are incidental to normal manufacturing or treatment processes or operations. B-2 ------- NATIONAL AGRICULTURAL CHEMICALS ASSOCIATION (NACA) — THE FERTILIZER INSTITUTE (TFI) With the exception of the last paragraph of each comment, the comments on continuous releases by NACA and TFI were identical. The identical portions of their comments included the following recommendations: o EPA's continuous release policy is heading in the right direction because it will provide the Agency with relatively few but comprehensive reports that analyze rates of release and project the likelihood of future releases. o EPA should define "routine, anticipated intermittent releases" broadly to cover all routine discharges "that can be reasonably estimated for the future based on experience." (The comment, however, did not specify whose experience.) o The Agency should take an ad hoc approach toward determining the "continuity and quantity" of releases. o A rule requiring a certain number of reports or period of time for reporting is inappropriate because some facilities already have substantial data from past operations enabling them to accurately anticipate future releases. o The determination of what constitutes a "statistically significant increase" should be left to the releaser. o EPA can correct any abuse of this discretion since the annual reports, which must be prepared in any event, would reveal the actual release history for the year. In addition, NACA urged EPA to design its final regulations to allow facilities to show that they are entitled to annual reporting on the basis of existing as well as new data. The Fertilizer Institute made a separate request that when continuous releases can be estimated reliably on the basis of material balances or other procedures, that continuous monitoring not be required. CONOCO Conoco.expressed the following views on the continuous release issue: o EPA has correctly construed congressional intent in its examples of releases considered to be "continuous" and "stable in quantity and rate" for purposes of Section 103(f)(2). o Under 103(f), the Agency should create a one-time notification system, because continuous releases by their nature are expected to be similar in composition, quantity, and rate from year to year. B-3 ------- OLIN Olin seemed to suggest that, in effect, almost any continuous release could be considered a federally permitted release. Olin's reasoning was as follows: o If a continuous release was not already federally permitted, the data necessary to establish a continuous release exemption could be used instead by the "permitting authority . . . to decide whether or not the release warrants permit limits, monitoring only, or no action. o "Unless the decision is 'no action,' regulation by a permit would qualify the release to be treated as a federally permitted release." o "If 'no action' is appropriate, this should satisfy the Agency that no further notification is needed." Coupled with this approach was a suggestion by Olin that a statistically significant increase be defined as "an increase great enough to change the release from 'no action' to 'federally permitted' status." Although Olin did not state the following explicitly, the net effect of its proposals would seem to be the elimination of all continuous release reporting requirements. AMERICAN TEXTILES MANUFACTURERS INSTITUTE (ATMI) ATMI made the following suggestions: o The "period sufficient" should consist of a single report where releases are "continuous" and "stable in quantity and rate." o As to "statistically significant releases," EPA should use a factor of 2, 5, 10, or 100 times the daily average, depending on the potential impact on the environment of the release. o The National Response Center (NRC) should receive and distribute to the appropriate state and federal agencies all hazardous substance reports. AMERICAN PAPER INSTITUTE AND NATIONAL FOREST PRODUCTS ASSOCIATION (API/NFPA) API/NFPA made the following comments: o EPA has properly decided to focus enforcement efforts on episodic releases exceeding proposed RQs. B-4 ------- o The Agency should propose implementing regulations for Section 103(f)(2) continuous releases as soon as possible. GENERAL BATTERY General Battery stated that EPA's proposed application of reporting requirements to routine operations (e.g., releases from relief valves, maintenance of pollution control equipment, charging of coke oven batteries, tank cleaning operations) will be "extremely difficult to compute in of reportabterms le quantities and will be extremely difficult to enforce." DOW CHEMICAL Dow made the following recommendations: o Continuous releases should be established through notification or from information already in EPA's possession; o EPA should not require additional and duplicative reporting; o EPA should focus on sudden and accidental releases, not routine releases; and o EPA should ensure that routine releases "contemplated as the normal occurrence of manufacturing operations" are exempt from reporting. STANDARD OIL COMPANY (INDIANA) Standard Oil suggested that EPA promulgate simple and flexible continuous release reporting requirements, and that for relatively stable or predictable releases, a one-time notification should suffice. AMERICAN PETROLEUM INSTITUTE (API) API made the following recommendations: o EPA should rely on the discretion of the releaser to determine when a "statistically significant release" has occurred. API maintained that this approach is supported by the discretion given the releaser under Sections 103(a) and (b) to determine whether a given release exceeds an RQ. o Any anticipated release should be eligible to qualify for a continuous release exemption. o The releaser should be permitted to establish the continuity, quantity, and regularity of the release through the use of available monitoring data or estimates based on best engineering ------- judgment. API stated that because an error in initial quantifica- tion will lead to a "change" in the release and since this "change" must be reported to the Agency, the releaser will have sufficient incentive to quantify accurately the release. o EPA should not set an arbitrary required number of notifications to establish continuity and regularity because in most cases the releaser will be able to identify the nature and quantity of the release in a single notification. o The Agency should not require releasers to file an annual report. Because the primary purpose of notification to the NRC is to facilitate a decision regarding emergency response, an annual report would be of little value. o The concept of "statistically significant increase" should be applied through a trigger mechanism; specifically, when an "upset" condition occurs (operational parameters exceed a predetermined range), and in the releaser's discretion a significant increase in emissions results, the NRC must be notified. o Whether a significant increase has occurred should be a case-by-case determination, dependent upon the substance released, media of release, normal amount of release, and increment of increase. o In determining an operational definition for "statistically significant," API recommended that formal statistical measurements and arbitrary multiples not be used because they do not accurately reflect the environmental significance of an increase in a release. ELI LILLY Eli Lilly made the following recommendations: o EPA should promulgate regulations for continuous releases as soon as possible. o The Agency's proposed definition of "continuous" and "stable in quantity and rate" is satisfactory. o The "period sufficient" should be one year, and the reporting frequency"should be annual. o In their reports to the NRC, releasers should not be required to include any quantities permitted by federal, state, or local authorities. B-6 ------- CHEMICAL MANUFACTURERS ASSOCIATION (CMA) CMA made the following comments: o EPA's proposed definition of releases that qualify as continuous (e.g., literally continuous releases, releases continuous during operating hours, releases from batch operations, and routine, anticipated, intermittent releases) is fair and comprehensive. o The "period sufficient" requirement should be satisfied by an initial report containing a description of the types of releases and an estimation of the levels of those releases for the year. o If a releaser underestimates significantly any release, he should be required to report such a release under the "statistically significant increase" provision of Section 103(f). o The terms "statistically significant increase" and "period sufficient" should be determined on a site-specific and release- specific basis using "best engineering judgment." SOUTHERN CALIFORNIA EDISON (SCE) SCE made the following suggestions concerning air emissions and the reduced reporting requirement for continuous releases: o Those substances (most stack gases) for which no de minimis policy has been established under CAA regulation but which are emitted under federal permits continuously and are stable in quantity and rate, should qualify for the 103(f)(2) reduced reporting requirement for continuous releases. o Only those CERCLA-designated hazardous substances released on an accidental, one-time or non-routine basis should be subject to 103(f)(2) reporting requirements. These releases, however, should not include releases incidental to normal operations. o These suggested exceptions are consistent with congressional intent to minimize reporting burdens on the regulated community and focus resources only on the most serious releases. MISSISSIPPI POWER AND LIGHT (MP&L) MP&L suggested that EPA give authority to the releaser to determine what is a statistically significant increase, because the "releaser has more intimate knowledge of his facility and thus could better decide what is significant." B-7 ------- DETROIT EDISON Detroit Edison made the following recommendations: o EPA should interpret broadly Section 103(f)(2) for stack emissions to exclude monitoring and reporting of day-to-day emissions resulting from normal plant operations. o In determining "period sufficient," only a very short time period should be necessary for fossil-fuel fired power plants. o To ease the reporting burden on industry while still gathering adequate information, a few typical plants could be selected to establish "continuity, quantity, and regularity." Then EPA would only have to require reporting on an annual basis or when emissions change significantly from the rest of the industry. KIMBERLY-CLARK Kimberly-Clark stated that: o It supports EPA's intention to require only annual reporting of continuous releases, unless there is a statistically significant increase in the release. o EPA needs to define more clearly the meaning of "continuous or routine intermittent discharges, "taking into account both point and nonpoint sources. o The Agency should clarify the meaning of the phrase "statistically significant increase" by using a pre-established, scientifically based factor to trigger a reporting requirement. DIAMOND SHAMROCK Diamond Shamrock made the following suggestions: o EPA should regulate continuous releases under existing permit programs; if substantial continuous releases do not fit within these programs, the programs should be modified. o The concept of "statistically significant increase" should not be based on the "Student t" test. GENERAL MOTORS CORPORATION (GM) GM believed that industrial discharges to POTWs are typically continuous releases, stable in quantity and rate, and should be exempted from CERCLA Section 103(a) and (b) notification requirements. Further, GM stated that the less burdensome requirements of Section 103(f)(2) may be unnecessary to B-8 ------- the extent that they duplicate reports required under local sewer ordinances as well as state and municipal permit programs. A yearly notification of continuous releases should be sufficient, coupled with the provision for reporting "statistically significant increases." However, of the three possible alternatives for defining "statistically significant increase" discussed in the preamble to the NPRM (page 23559), the third alternative, which allows the releaser to determine what is a statistically significant increase, is objectionable because "unscrupulous operators could be expected to take advantage of such a provision." ATLANTIC RICHFIELD COMPANY (ARCO) ARCO agreed that: o Annual reporting should be sufficient, provided that the release is "stable in quantity and rate," and that notification is not provided through some other mechanism; and o Releases which are "intermittent and anticipated" should be considered "continuous" because, in most cases, they are an unavoidable part of many industrial processes. ARCO suggested that: o The "period sufficient" to determine the continuity, quantity, and regularity of a release should be determined by the person responsible for reporting the release. ARCO suggested that EPA set an upper time limit of one year for establishing the qualifications of a release; and o The person responsible for reporting a release should specify or propose the method(s) by which a "statistically significant" increase will be determined, and EPA should include the statistical methods which would be acceptable in its final rule. PENNSYLVANIA POWER.& LIGHT COMPANY (PP&L) PP&L supported the Agency's continuous release policy but also suggested: o EPA should explicitly consider recognizing distribution capacitor failures (a "recurring" release) as a continuous release because most incidents result in little or no environmental impact and mechanisms exist to ensure that incidents needing EPA response are reported; and o The period for establishing that a release is continuous should be short and should be based on the number of reports, not some arbitrary time period. After a given number of occurrences (regardless of time period), a determination can be made concerning whether a release is continuous. B-9 ------- SOHIO The continuous release notification policy should be flexible and broadly construed. SOHIO suggested that: o Any anticipated release should be a candidate for the reduced reporting requirement for continuous releases; this includes any releases incidental to normal manufacturing or operation of a facility; o The releaser should be allowed to present the particulars of the release in its initial notification; o A twenty-four hour period for purposes of determining the amount of the release should be employed; o The releaser should not be required to aggregate releases across all facilities within a plant; however, such an aggregation should be allowed as long as there is an adequate description of the sources which are included in the filing; o The continuity, quantity, and regularity of the release should be established by the releaser through the use of actual monitoring data or estimates based on best engineering judgment. There will be substantial incentive on the part of the releaser to quantify accurately the release, since an error will necessitate reports to notify the Agency of "changes" in the release; o EPA should not specify an arbitrary number of notifications, because in most instances the releaser will be able to pinpoint the nature and quantity of the release in the initial notification based on historic information and normal operating practice. On the other hand, the Agency should not foreclose multiple notifications in appropriate circumstances; and o EPA should require the reporting of all upset conditions which will cause a change in emissions. The term "upset," has a clear meaning to those involved with industrial processes, whereas the term "statistically significant," can mean different things to different industries. This notification scheme, undertaken by the releaser, identifies the types of unanticipated, episodic changes which could prompt a government response. POTOMAC ELECTRIC POWER COMPANY (PEPCO) PEPCO supported the Agency's flexible definition of the kinds of releases that are considered continuous, but urged EPA to provide guidance concerning what period is sufficient to establish the continuity, quantity and regularity of a release. B-10 ------- o PEPCO suggested that a source's "potential to emit" a hazardous air pollutant be considered when determining if a release qualifies as "continuous." o A "statistically significant" increase should be defined as any emission of a hazardous pollutant from a source above its defined "potential to emit." (Examples include emissions increases resulting from changes in fuel characteristics or removal of government restrictions on a source's hours of operation.) ROOKS, PITTS, FULLAGER AND POUST on behalf of ILLINOIS MANUFACTURERS' ASSOCIATION o A release of a hazardous substance which is not a criteria pollutant or is not otherwise regulated should be exempt from notification requirements either under the federally permitted release exemption or under the reduced reporting requirement for continuous releases. o The continuous release definition should include: literally continuous releases that enter the environment 24 hours a day, 365 days a year; releases continuous during operating hours; and routine, anticipated, intermittent releases incidental to normal manufacturing or treatment processes or operations of facilities or vessels. o Facilities should have to notify EPA only once concerning their releases, with additional notifications to be made if the release substantially exceeds the continuous release rate. o The releaser should be allowed to determine what is a "statistically significant" increase. MIDDLE SOUTH SERVICES, INC. (MSS) MSS urged EPA to use other submissions besides NRC reports to determine the eligibility of a release as continuous. MSS was concerned, however, that the use of other submissions "may be somewhat constrained by the statute's requirement that reporting be under specific CERCLA sections." In addition, MSS suggested that: o EPA not establish a specific time period in which to demonstrate that a release is continuous, because some releases may be reported more frequently than necessary, and some releases, due to seasonal variations, may not be reported at all. Instead, the releaser should be allowed to.provide the Agency with the information it believes shows that a release qualifies for the reduced reporting requirement for continuous releases; and o The releaser should be allowed to determine what constitutes a statistically significant increase based on "his knowledge of his operation and what can be expected from it." B-ll ------- NATIONAL FOOD PROCESSORS ASSOCIATION (NFPA) NFPA offered the following comments regarding the continuous release issue: o NFPA concurred with EPA that episodic releases exceeding RQs present a more serious risk of harm to the public health or welfare or the environment than continuous releases. o The Agency should simplify reporting requirements for continuous releases by identifying specific releases which do not pose significant risks to human health and the environment and therefore do not need to be reported (e.g., releases of ammonia to the air during routine maintenance of refrigeration units). o The releaser should determine what is a statistically significant increase that triggers the reporting requirement. o Annual reports are unnecessary and burdensome. PHILLIPS PETROLEUM COMPANY Phillips reserved comment on continuous releases until specific regulations are proposed by EPA. COUNTY SANITATION DISTRICTS OF LOS ANGELES COUNTY (Sanitation Districts) The Sanitation Districts recommended that notification of continuous releases be in a written form because formal documentation of a problem is more important than immediate telephone notification. NATURAL RESOURCES DEFENSE COUNCIL, INC. (NRDC) NRDC suggested that EPA: o Provide specific guidelines for releasers on sampling plans to establish stability and continuity of releases, and to provide estimates of parameters (such as average flow rate and variation in discharge) useful in determining if a statistically significant increase has occurred; and o Clearly define what a statistically significant increase is by providing: "alternatives for sampling schemes for release monitoring; procedures for checking distributional assumptions about release data (such as the use of normal probability paper); and suggestions for appropriate statistical tests with desired level of significance." NRDC maintained that allowing releasers to define their own monitoring and reporting standards would almost certainly result in conflicting and inaccurate continuous release reports. B-12 ------- TEXAS DEPARTMENT OF WATER RESOURCES (Texas DWR) The Texas DWR suggested that the continuous relese notification system be structured as follows: o Continuous releases should be documented through a registration system; upon proper registration, all notification requirements would be satisfied. o The registration should document the location, source, impact, and character of the discharge. o If at any time after original notification a "significant change" in discharge occurs, updates should be filed with the appropriate regulatory agencies. o Once registered, continuous releases should be individually reviewed to determine whether periodic reporting or other response actions are required. EVANS, KITCHELL AND JENCKES, P.O. on behalf of PHELPS DODGE CORPORATION Phelps Dodge provided the following recommendations:. o EPA should give a broad interpretation to the term "continuous release" so that notifications can be reduced or eliminated in cases where a government response is unlikely. o Because continuous releases are predictable and can be monitored, a less frequent notification scheme is appropriate. o Depending on the definition of "stable in quantity and rate," continuous air releases from stationary sources may not qualify for the 103(f)(2) exemption. o The unique circumstances surrounding any continuous release argue in favor of determining "statistically significant" on a case-by-case basis. B-13 ------- HUNTON AND WILLIAMS on behalf of UTILITY WATER ACT GROUP (UWAG) UTILITY AIR REGULATORY GROUP (UARG) UTILITY SOLID WASTE ACTIVITIES GROUP (USWAG) EDISON ELECTRIC INSTITUTE (EEI) NATIONAL RURAL ELECTRIC COOPERATIVE ASSOCIATION (NRECA) AMERICAN PUBLIC POWER ASSOCIATION (APPA) The UWAG, UARG, USWAG, EEI, NRECA, and APPA made the following comments concerning continuous releases: o EPA correctly rejected an extensive monitoring system to document fluctuations in the nature or rate of releases. o For purposes of Section 103(f)(2), accidental releases cannot always be considered episodic. o "Period sufficient" should be defined as a certain time period (no longer than a month) or a certain number of reports, determined on a case-by-case basis to maintain flexibility. o Because Section 103(f)(2) calls for notification annually or when there is a statistically significant increase, a releaser that reports statistically significant increases should not be required to report annually. o As long as a source operates normally under the range of conditions existing when it established the continuity of a release, EPA should not require it to monitor for statistically significant increases. Imposing an affirmative monitoring requirement would defeat the purposes of Section 103(f). o Once a release is established as continuous, further reporting should be required only if existing monitoring reveals a statistically significant increase or if a material change in operations occurs. o The Agency should use its rulemaking authority to implement Section 103(f) for entire categories of sources based on the experience of representative sources within the category. The electric utility industry, for example, could provide representa- tive data documenting the quantities of various hazardous substances emitted under the full range of operating procedures and techniques applicable throughout the industry. An individual source could then establish the quantity of its continuous B-14 ------- releases by stating its relevant operating procedures and techniques and referring to the source-wide data. AMERICAN MINING CONGRESS (AMC) AMC generally endorsed EPA's approach to continuous releases. AMC made the following recommendations: o "Period sufficient" should not be measured by a specific time period or number of reports. Where a releaser knows, based on past experience, the identity, frequency, and rate of its releases, a single notification should satisfy the requirements of Section 103(f)(2). o EPA should allow releasers to qualify for the reduced reporting requirement for continuous releases based on information contained in applications for federal, state, and regional permits. Such an approach would reduce reporting burdens on both the Agency and the regulated community. KOPPERS COMPANY Koppers expressed the following views on the continuous release issue: o EPA should define "period sufficient" as an initial report describing the types of releases the releaser anticipates in the next 12 months; no single time period will work well for all types of possible releases. o The definitions of both "period sufficient" and "statistically significant increase" should incorporate the concept of "best engineering judgment." o Sites of past industrial activity which have been reported under Section 103(c) of CERCLA should be exempt from reporting continuous releases. AMAX ENVIRONMENTAL SERVICES AMAX generally supported EPA's continuous release policy. AMAX made the following suggestions: o EPA should define "continuous release" by type of activity and "significant increase" by significant change in the activity because of the difficulty of quantifying the amounts released by many activities. o Until the Agency establishes reporting procedures, it should not take enforcement action for failure to report continuous releases. B-15 ------- U.S. DEPARTMENT OF THE INTERIOR (DPI) i' DOI made the following suggestions: o "Period sufficient" must be clearly defined as a specific number of periodic reports rather than an arbitrary time period. o "Statistically significant increase" should be defined by a statistical test because such tests are specifically designed to determine significant variances from an average. ARMY CORPS OF ENGINEERS o The definition of a "period sufficient" should encompass both a specific time period and a specified number of reports. o "Statistically significant increase" should be defined in terms of a pre-determined range, not a pre-established factor, because even a factor of only two times the "daily average of certain hazardous substances" can be lethal for some aquatic biota. ENVIRONMENTAL DEFENSE FUND (EOF) EOF urged EPA to propose immediately a notification system for continuous releases. EOF believed that data on hazardous substance releases of a routine nature are extremely valuable in setting priorities for future regulatory activity. U.S. NUCLEAR REGULATORY COMMISSION The Nuclear Regulatory Commission recommended that for purposes of defining "statistically significant increase," EPA permit concentration levels to be averaged over periods of one year. ------- COMMENTERS ON CONTINUOUS RELEASES Kendall Company Air Products and Chemicals Vulcan Chemicals Adolph Coors The Light Company RSR Corporation Pennzoil National Agricultural Chemicals Association and The Fertilizer Institute Conoco Olin American Textile Manufacturers Institute American Paper Institute and National Forest Products Association General Battery Dow Chemical Standard Oil Company (Indiana) American Petroleum Institute Eli Lilly Chemical Manufacturers Association Southern California Edison Mississippi Power and Light Detroit Edison Kimberly-Clark Diamond-Shamrock General Motors Atlantic Richfield Pennsylvania Power and Light SOHIO Potomac Electric Power Company Rooks, Pitts, Fullagar and Poust on behalf of Illinois Manufacturers' Association Middle South Services National Food Processors Association Phillips Petroleum County Sanitation Districts of Los Angeles County Natural Resources Defense Council Texas Department of Water Resources Evans, Kitchell and Jenckes, P.C. on behalf of Phelps Dodge Corp. Hunton and Williams on behalf of Utility Water Act Group; Utility Air Regulatory Group; Utility Solid Waste Activities Group; Edison Electric Institute; B-17 ------- National Rural Electric Cooperative Association; American Public Power Association American Mining Congress Koppers Company AMAX Environmental Services U.S. Department of the Interior Army Corps of Engineers Environmental Defense Fund U.S. Nuclear Regulatory Commission B-18 ------- APPENDIX C EXAMPLES OF USING STATISTICAL TESTS CONSIDERED Control Chart Test The Control Chart test should be used when at least 30 releases have been recorded. The Control Chart test is a dynamic test in that it uses prior releases to form a data set in which the mean and standard deviation are calculated. When a new release is recorded, its value is compared to 1.645 times the standard deviation. If the new recorded release is less than (1.645)*(S), then the release is not considered statistically significant and does not need to be reported to the NRC. The release is then added to the release history and a new standard deviation is calculated. If the next new release is greater than 1.645 times the newly calculated standard deviation, then the release must be reported to the NRC and the release is not included in the release history; there is no recalculation of the standard deviation. Example Release Release Release Release Release Release Release Release Release Release 1 = 2 = 3 = 4 = 5 = 6 = 7 = 8 = 9 = 10 = 10 15 70 30 60 50 40 80 10 50 pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds Release Release Release Release Release Release Release Release Release Release 11 12 13 14 15 16 17 18 19 20 = 30 = 40 = 5 = 90 = 47 = 12 = 25 = 33 = 75 = 10 pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds Release Release Release Release Release Release Release Release Release Release 21 = 22 = 23 = 24 = 25 = 26 = 27 = 28 = 29 = 30 = 110 50 105 80 40 20 60 10 90 40 pounds pounds pounds pounds pounds pounds pounds pounds pounds pounds C-l ------- For the 30 releases shown above, the mean is 46.2 and the standard deviation is 30.25 (the general method for calculating a mean and standard deviation is shown in the section entitled "Girubb's Test"). Any release greater than 49.76 pounds (1.645 times 30.25) must be reported to the NRC. If release 31 equals 40 pounds, it is less than the trigger value of 49.76 pounds and is not reported to the NRC. This release is added to the data set and a new trigger value is calculated. The standard deviation now equals 29.77 and 'the trigger value is equal to 48.96. If release 32 is equal to 50 pounds, it must be reported to the NRC because. it is greater than the trigger value of 48.96 pounds; The trigger value remains the same for the next release because the 50 pound release is not added to the release history and no new standard deviation is calculated. Dixon Criterion The Dixon Criterion compares the newest release to the largest release in the data set and divides that difference by the difference between the newest release and the smallest release in the data set, as shown in the equation below: C-2 ------- The number so derived is compared to Table C-l relating reporting triggers to the number of releases in the data set. If the number is larger than the appropriate number in the table, then a report must be made to the NRC. Example If the first three releases by a continuous releaser were 10, 15 and 20 pounds, in that order, then: 20 - 15 C - 0.5 20 - 10 The reporting trigger for the third release is not exceeded and the release need not be reported to the NRC. A fourth release of 60 pounds would, however, exceed the reporting trigger for a fourth release and must be reported to the NRC. 60 - 20 C =—: = 0.8 60 - 10 C-3 ------- Grubb's Test The Grubb's test involves the calculation of a test value and a comparison of this test value with the "Grubb's Statistic." The test value is calculated by subtracting the mean of all past releases (except those determined to be statistically significant increases) from the newest release, and dividing this difference by the standard deviation of all releases. That is, where X is the most recent release, X is the mean or average of all past n nonstatistically significant releases and S is the standard deviation for all statistically significant releases. If the test value is less than the corresponding Grubb's statistic shown in Table C-2, then the release is not considered a statistically significant increase and need not be reported to the NRC. If the test value is equal to or greater than the Grubb's statistic in Table C-2, then the release must be reported to the NRC. A SSI value should not be included in future Grubb's test calculations. The general methods for calculation of X and S are shown below. n - 1 C-4 ------- Example If the first three releases by a continuous releaser were 10, 15 and 20 pounds, in that order, the standard deviation (S) is 5, the mean is 15 and the Grubb's test value is calculated as follows: A (10 - 15)2 + (15 - 15)2 + (20 - 15)2 !___________ L. 5 (3- 1) 20-15 Grubb's Test Value = — - 1 5 Since Table C-2 gives a Grubb's statistic of 1.15 for the third observation and the Grubb's test value just calculated is 1, no report need be made to the NRC. IF the fourth release is 60 pounds, the mean equals 26.25 and the standard deviation is 22.87. The Grubb's test value is calculated below. 60 - 26.25 Grubb's Test Value = = 1.48 22.87 C-5 ------- The Grubb's test value is 1.48 while the Grubb's statistic is 1.46 indicating that this 60 pound release must be reported to the NRC. Student t Test The Student t test is identical to the Grubb's test except that the mean and standard deviation do not include the newest release. All releases in the release history are used to calculate the mean and standard deviation. Student t Value -. S Example If the first three releases by a continuous releaser were 10, 15 and 20 pounds, and it was desired to calculate the Student t value for the fourth release of 25 pounds and the standard deviation is 5. 25 - 15 Student t Value = = 2 C-6 ------- The Student t value is less than the value in Table C-3 for the fourth release, and this release need not be reported ''to the NRC. A fifth release of 40 pounds yields a Student t value of 3.48, calculated in the same way, and would require a report to the NRC. 40 - 17.50 Student t Value = = 3 48 6.45 C-7 ------- TABLE C-l DIXON CRITERION REPORTING TRIGGERS Trigger Value at Number of Observations 95 Percent Confidence Level 3 4 5 6 7 8 9 10 12 14 16 18 20 25 30 0.941 0.765 0.642 0.560 0.507 0.468 0.437 0.412 0.376 0.349 0.329 0.313 0.300 0.277 0.260 C-8 ------- TABLE C-2 GRUBB'S TEST REPORTING TRIGGERS Number of Observations Grubb's Statistic 3 4 5 6 7 8 9 10 12 14 15 16 18 20 30 1.15 1.46 1.67 1.82 1.94 2.03 2.11 2.18 2.29 2.37 2.41 2.44 2.50 2.56 2.74 C-9 ------- TABLE C-3 STUDENT'S t TEST REPORTING TRIGGERS Number of Observations Student's t Statistic 3 4 5 6 7 8 9 10 12 14 15 16 18 20 30 2.353 2.132 2.015 1.943 1.895 1.860 1.833 1.812 1.782 1.761 1.753 1.746 1.734 1.725 1.697 C-10 -------