ML23156A475

PR-MISC - 59FR63389 - Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities; Proposed Policy Statement

ML23156A475
Person / Time
Issue date:	12/08/1994
From:	Bates A NRC/SECY
To:
References
PR-MISC, 59FR63389
Download: ML23156A475 (1)

Text

ADAMS Template: SECY-067 DOCUMENT DATE: 12/08/1 994 TITLE: PR-MISC - 59FR63389 - USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES; PROPOSED POLICY STATEMENT CASE

REFERENCE:

PR-MISC 59FR63389 KEYWORD: RULEMAKING COMMENTS Document Sensitivity: Non-sensitive - SUNSI Review Complete

STATUS OF RULEMAKING PROPOSED RULE: PR-MISC OPEN ITEM (Y/N) N RULE NAME: USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES; PROPOSED POLICY STATEMENT PROPOSED RULE FED REG CITE: 59FR63389 PROPOSED RULE PUBLICATION DATE: 12 / 08 / 94 NUMBER OF COMMENTS: 17 ORIGINAL D~TE FOR COMMENTS: 02 / 07 / 95 EXTENSION DATE: I I FINAL RULE FED. REG. CITE: 60FR42622 FINAL RULE PUBLICATION DATE: 08 / 16 / 95 NOTES ON: DRAFT POLICY STATEMENT SIGNED BY GARY HOLAHAN. FINAL POLICY STATUS STATEMENT EFFECTIVE 8/16 / 95. FILE LOCATED ON Pl.

OF RULE:

HISTORY OF THE RULE PART AFFECTED: PR-MISC RULE TITLE: USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES; PROPOSED POLICY STATEMENT PROPOSED RULE PROPOSED RULE DATE PROPOSED RULE SECY PAPER: 94-218 SRM DATE: 10 / 04 / 94 SIGNED BY SECRETARY: 12 /01/ 94 FINAL RULE FINAL RULE DATE FINAL RULE SECY PAPER: 95-126 SRM DATE: 10 / 04 / 95 SIGNED BY SECRETARY: 08/10 / 95

• CONTACT!: ANTHONY HSIA, NRR STAFF CONTACTS ON THE RULE MAIL STOP: 010-E4 PHONE: 415-1075 CONTACT2: MAIL STOP: PHONE:

DOCKET NO. PR-MISC (59FR63389)

In the Matter of USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES; PROPOSED POLICY STATEMENT DATE DATE OF TITLE OR DOCKETED DOCUMENT DESCRIPTION OF DOCUMENT

- 01/20/95 12/01/94 PROPOSED POLICY STATEMENT PUBLISHED IN THE FEDERAL REGISTER ON 12/08/94.

01/20/95 01/12/95 COMMENT OF CAROLS. MARCUS, PH.D., M.D. ( 1) 01/30/95 01/20/95 COMMENT OF PECO ENERGY CO (G.A. HUNGER) ( 2) 02/01/95 01/26/95 LETTER FROM RETA ROE, CENTRAL INTERSTATE LOW-LEVEL RADIOACTIVE WASTE COMMISSION, REQUESTING A 90-DAY EXTENSION OF THE COMMENT PERIOD 02/06/95 02/04/95 COMMENT OF PLG, INC (B. JOHN GARRICK, PH.D., PRESIDENT} ( 3) 02/07/95 02/07/95 COMMENT OF DEPARTMENT OF ENERGY OFC OF CIVILIAN RAD WASTE MGT (RONALD A. MILNER) ( 4)

- 02/08/95 02/07/95 COMMENT OF OHIO CITIZENS FOR RESPONSIBLE ENERGY, INC (SUSAN L. HIATT, DIRECTOR) ( 5) 02/08/95 02/07/95 COMMENT OF NUCLEAR UTILITY BACKFITTING &REFORM GRP (JAN A. MACGREGOR) ( 6) 02/08/95 02/07/95 COMMENT OF DETROIT EDISON (LYNNE S. GOODMAN) ( 7) 02/09/95 02/01/95 COMMENT OF NUCLEAR ENERGY INSTITUTE (THOMAS E. TIPTON) ( 8) 02/10/95 02/08/95 COMMENT OF WASHINGTON PUBLIC POWER SUPPLY SYSTEM (J.V. PARRISH) ( 9) 02/13/95 02/07/95 COMMENT OF CAROLINA POWER AND LIGHT CO (R.E. ROGAN) ( 10) 02/13/95 02/09/95 COMMENT OF VIRGINIA POWER (M.L. BOWLING) ( 11) 02/14/95 02/07/95 COMMENT OF TOLEDO EDISON (JOHN P. STETZ) ( 12) 02/15/95 02/10/95 COMMENT OF ILLINOIS DEPARTMENT OF NUCLEAR SAFETY (THOMAS W. ORTCIGER, DIRECTOR) ( 13)

DOCKET NO . PR-MISC (59FR63389)

DATE DATE OF TITLE OR DOCKETED DOCUMENT DESCRIPTION OF DOCUMENT 02/21/95 02/15/95 COMMENT OF NEW JERSEY BUREAU OF NUCLEAR ENGINEERING (KENT W. TOSCH) ( 14) 02/22/95 02/15/95 COMMENT OF WESTINGHOUSE OWNERS GROUP (ROGER A. NEWTON , CHAIRMAN) ( 15) 02/24/95 02/07/95 COMMENT OF BRUCE E. FISHMAN , PH .D. ( 16) 02/28/95 02/22/95 COMMENT OF NEVADA , STATE OF (ROBERT R. LOUX) ( 17)

- 08/11/95 08/ 10/95 FEDERAL REGISTER NOTICE FOR FINAL POLICY STATEMENT

NUCLEAR REGULATORY COMMISSION Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities; Final Policy St atement AGENCY: Nuclear Regulatory Commission.

ACTION: Final policy statement.

SUMMARY

This statement presents the policy that the Nuclear Regulatory Commission (NRC) will follow in the use of probabilistic risk assessment (PRA) methods in nuclear regulatory matters. The Commission believes that an overall policy on the use of PRA methods in nuclear regulatory activities should be established so that the many potential applications of PRA can be implemented in a consistent and pred i ctable manner that would promote regula-tory stability and efficiency. In addition, the Commission believes that the use of PRA technology in NRC regulatory activities should be increased to the extent supported by the state-of-the-art in PRA methods and data and in a manner that complements the NRC's deterministic approach. The pertinent comments received from the published draft policy statement are reflected in this final policy statement. This policy statement will be implemented through the execution of the NRC's PRA Implementation Plan.

foO FR-4-~6 g,,~

'-0~. ~ \ l~\C\5

~ \<o > \ C\C\s EFFECTIVE DATE: (l:fp011 pttb+i-e-ali-et, i" t he Federal Registe r .)

ADDRESSES: The proposed policy statement and the comments received may be examined at: NRC Public Document Room, 2120 L Street, NW. {Lower Level},

Washington, D.C.

FOR FURTHER INFORMATION CONTACT: Anthony Hsia, Office of Nuclear Reactor Regulation, U.S. Nuclear Regulatory Commission, Washington, DC 20555.

Telephone {301) 415-1075

,e SUPPLEMENTARY INFORMATION:

I. Background II. Summary of Public Comments and NRC Responses III. Deterministic and Probabilistic Approaches to Regulation IV. The Commission Policy V. Availability of Documents I. Background The NRC has generally regulated the use of nuclear material based on deterministic approaches. Deterministic approaches to regulation consider a set of challenges to safety and determine how those challenges should be mitigated. A probabilistic approach to regulation enhances and extends

this traditional, deterministic approach, by (1) allowing consideration of a broader set of potential challenges to safety, (2) providing a logical means for prioritizing these challenges based on risk significance, and (3) allowing consideration of a broader set of resources to defend against these challenges.

Until the accident at Three Mile Island (TMI) in 1979, the Atomic Energy Commission (now the NRC), only used probabilistic criteria in certain specialized areas of licensing reviews. For example, human-made hazards (e.g., nearby hazardous materials and aircraft) and natural hazards (e.g., tornadoes, floods, and earthquakes) were typically addressed in terms of probabilistic arguments and initiating frequencies to assess site suitability. The Standard Review Plan {NUREG-0800) for licensing reactors and some of the Regulatory Guides supporting NUREG-0800 provided review and evaluation guidance with respect to these probabilistic considerations.

The TMI accident substantially changed the character of the analysis of severe accidents worldwide. It led to a substantial research program on severe accident phenomenology. In addition, both major investigations of the accident (the Kemeny and Rogovin studies) recommended that PRA techniques be used more widely to augment the traditional nonprobabilistic methods of analyzing nuclear plant safety. In 1984, the NRC completed a study (NUREG-1050) that addressed the state-of-the-art in risk analysis techniques.

In early 1991, the NRC published NUREG-1150, "Severe Accident Risks:

An Assessment for Five U.S. Nuclear Power Plants." In NUREG-1150, the NRC used improved PRA techniques to assess the risk associated with five nuclear power plants. This study was a significant turning point in the

use of risk-based concepts in the regulatory process and enabled the Commission to greatly improve its methods for assessing containment performance after core damage and accident progression. The methods developed for and results from these studies provided a valuable foundation in quantitative risk techniques.

PRA methods have been applied successfully in several regulatory activities and have proved to be a valuable complement to deterministic engineering approaches. This application of PRA represents an extension and enhancement of traditional regulation rather than a separate and different technology. Several recent Commission policies or regulations have been based, in part, on PRA methods and insights. These include the Backfit Rule (§50.109, "Backfitting"), the Policy Statement on "Safety Goals for the Operation-of Nuclear Power Plants," (51 FR 30028; August 21, 1986), the Commission's "Policy Statement on Severe Reactor Accidents Regarding Future Designs and Existing Plants" (50 FR 32138; August 8, 1985), and the Commission's "Final Policy Statement on Technical Specifications Improvement for Nuclear Power Reactors" (58 FR 39132; July 22, 1993). PRA methods also were used effectively during the anticipated transient without scram (ATWS) and station blackout (SBO) rulemaking, and supported the generic issue prioritization and resolution process.

Additional benefits have been found in the use of risk-based inspection guides to focus NRC inspector efforts and make more efficient use of NRC inspection resources. Probabilistic analyses were extensively used in the development of the recently proposed rule change to reactor siting criteria in 10 CFR Part 100 (59 FR 52255; October 17, 1994). The proposed rule change invoked the use of a probabilistic approach to estimate the Safe

Shutdown Earthquake Ground Motion for a nuclear reactor site, instead of the purely deterministic method currently specified in Appendix A to 10 CFR Part 100.

Currently, the NRC is using PRA techniques to assess the safety importance of operating reactor events and is using these techniques as an integral part of the design certification review process for advanced reactor designs. In addition, the Individual, Plant Examination (IPE) 4lt program and the Individual Plant Examination - External Events (IPEEE) program (an effort resulting from the implementation of the Co11BDission's "Policy Statement on Severe Reactor Accidents Regarding Future Desigps and Existing Plants"} have resulted in coR111ercial reactor licensees using risk-assessment methods to identify any vulnerabilities needing attention.

The Commission has been developing performance assessment methods for low-level and high-level waste since the mid-1970s and these activities intensified using performance assessments techniques in the late 1980s and early 1990s. This has involved the development of conceptual models and computer codes to model the disposal of waste. Because waste-disposal systems are passive, certain analysis methods used for active systems in PRA studies for power reactors had to be adapted to provide scenario analysis for the performance assessment of the potential geologic repository at Yucca Mountain, Nevada. In regard to high-level waste, the NRC staff participates in a variety of international activities (e.g., the Performance Assessment Advisory Group of the Organization for Economic Cooperation and Development, Nuclear Energy_Agency) to ensure that consistent performance assessment methods are used to the degree appropriate.

The Co111T1ission believes that an overall policy on the use of PRA in nuclear regulatory activities should be established so that the many potential applications of PRA methodology can be implemented in a consistent and predictable manner that promotes regulatory stability and efficiency and enhances safety. In May 1994, the NRC staff forwarded a draft PRA policy statement to the Advisory Committee on Reactor Safeguards (ACRS) for review and briefed ACRS on the same subject. On August 18, 1994, the NRC staff proposed a PRA policy statement to the Commission in SECV-94-218, "Proposed Policy Statement on the Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities." In that Commission paper, the staff proposed that an overall policy on the use of probabilistic risk assessment (PRA) methods in nuclear regulatory activities should be established and that the use of PRA technology in NRC regulatory activities should be increased. Comments from the ACRS regarding the policy statement as documented in a letter dated May 11, 1994, were incorporated. On August 19, 1994, the staff forwarded

• SECV-94-219, "Proposed Agency-Wide Implementation Plan for Probabilistic Risk Assessment (PRA)," to the Commission. On August 30, 1994, the staff discussed the PRA policy statement and the PRA implementation plan in a public meeting with the Commission. On September 13 and October 4, 1994, the Secretary issued two staff requirements memoranda (SRMs) providing ColllTlission guidance regarding the draft policy statement. In these SRMs, the Commission directed the staff to revise the proposed PRA policy statement, publish the policy statement for public comment in the Federal Regjster, and conduct a public workshop on the PRA implementation plan.

As directed by the Commission, the staff conducted a public workshop on December 2, 1994, to discuss the PRA implementation plan. The purpose of the workshop was to inform the public of NRC activities related to increasing the use of PRA methods and techniques in regulatory applications and to receive public comments on these activities. After the staff incorporated the corrments from the SRMs, the proposed policy statement "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities" was published in the Federal Register on December 8, 1994 {59 FR 63389).

The public comment period expired on February 7, 1995.

II. Summary of Public Co111J1ents and NRC Responses In January and February 1995, the NRC received 17 letters commenting on the proposed policy statement on "Use of Probabilistic Ris~ Assessment Methods in Nuclear Regulatory Activities". These comments were from the following organizations: six utilities - PECO Energy Company, Detroit Edison, Washington Public Power Supply System, Carolina Power and Light Company, Virginia Power Company, and Centerior Energy; three State regulatory agencies - State of Illinois Department of Nuclear Safety, State of New Jersey Department of Environmental Protection, State of Nevada Agency for Nuclear Projects; two industry groups - Nuclear Energy Institute and Westinghouse Owners Group; two engineering firms - PLG, Inc. and !CF Kaiser Engineers, Inc.; University of California at Los Angeles; Ohio Citizens For Responsible Energy; Winston and Strawn, Counsel to the Nuclear Utility Backfitting and Reform Group; and the Department of Energy. Copies

of the letters may be examined at the NRC Public Document Room at 2120 L Street., NW. (Lower Level), Washington, DC.

General Comments Twelve co11111enters explicitly supported the basic tenet of the policy to increase the use of PRA technology in NRC's regulatory activities. The other convnenters did not object to the policy statement but provided recommendations for the NRC to modify and improve the policy statement and/or the PRA implementation plan. Five commenters indicated that they agreed with the NEI comments on the proposed PRA policy statement. The NRC staff has reviewed the comme~ts and summarized them in the following areas.

The staff response to the comments are also included in this final policy statement.

Use of PRA in Regulatory Decisions Several comments dealt with the scope of the PRA applications (where can PRA be used) and the implementation of the policy statement (how can PRA be used).

One commenter felt that neither the policy statement nor the PRA implementation plan provided consistent decision criteria for accepting PRA results as part of the justification for licensing decisions. The commenter was concerned that the short term effect of the policy statement would likely be an increased burden on the licensees. For the long term, the commenter recommended a systematic review of the rules and regulations

to identify opportunities for elimination of unnecessary regulations. The proposed policy statement directed the staff to use PRA and associated analyses, where appropriate, as part of the justification for licensing decisions. The PRA implementation plan describes how the stated policy is to be implemented. Appropriate decision criteria will be developed and documented as part of the PRA implementation plan. The Co11111ission has already perfonned a systematic review of the many current rules and 4lt regulations to identify opportunities for the elimination of unnecessary regulations. In 1993, the NRC established the Regulatory Review Group (RRG} to conduct a structured review of power reactor regulations with special attention on the opportunity to reduce unnecessary regulatory burdens. The RRG reconwnendations to reduce the regulatory burden included the suggestion to use more risk-based approaches in quality assurance, inservice inspection and testing, and the concept of a PRA plan. The RRG reco1T111endations were documented in SECY-94-003. To better focus the NRC's effort on the PRA related activities recommended by the RRG, the PRA Working Group, and the Regulatory Analysis Steering Group, the PRA implementation plan was developed in 1994. The implementation plan included a task to develop guidelines for determining when it is practical to use PRA technology and results in regulatory activities. The NRC has had discussions with volunteer licensees regarding the pilot applications of risk-based regulatory initiatives. Results from the pilot applications will be incorporated in the NRC's guidance for PRA applications in regulatory activities. A number of current regulatory requirements are being considered as part of the PRA implementation plan to determine if alternative risk-based approaches are practical. Over time, the Commission

~ 10 -

would expect some streamlining and refocusing of its rules and regulations as part of this process. The Co11111ission has implemented a continuing regulatory improvement program which is responsive to the comenter'*s recoD111endation of a systematic examination of marginal regulatory requirements.

Another cotnenter reco1111ended that the po7icy statement be amended to state that when backfitting analyses are performed, mean risk levels be the exclusive basis of regulatory decision-making when comparisons are made against the $1000/person-rem criterion. The Convnission does not feel this policy statement needs to address the issue regarding the use of mean risk level as the exclusive basis for applying the $1000/person-rem criterion because the Convnission's safety goal policy statement has already spoken to the use of mean values of risk in connection with the cost-benefit analyses. Furthermore, this issue is addressed in the proposed Revision 2 of NUREG/BR-0058, "Regulatory Analysis Guidelines of the U. S. Nuclear Regulatory Convnission, Draft Report for ColTffllent. 11 This commenter also recommended that the policy statement should direct the staff to use the relevant plant specific PRA in assessing the need for any backfitting action at that plant. For generic backfits, this co1111enter reco111111ended that the policy should allow licensees to take credit for plant specific information to justify relief from NRC imposed action. The Convnission believes that the use of the plant specific PRA in the backfit analysis to evaluate whether there is a substantial increase in the overall protection or to justify relief from NRC imposed action is acceptable when combined with other relevant deterministic considerations, as appropriate.

Regarding the use of safety goals, one co111111enter reco11111ended retention of the language in SECY-94-218 to effect that safety goa1s could be used in granting relief from unnecessary requirements. Another commenter reco111111ended that the safety goals should be used as a minimum goal, rather than the maximum 1eve7 of safety. As stated in the proposed PRA policy statement published on December 8, 1994, the Commission's safety goals are

" .*. intended to be generically applied by the NRC as opposed to plant specific applications,* and 8

• • • to be used with appropriate consideration of uncertainties in making regulatory judgements in the context of backfitting new generic requirements on nuclear power plant licensees." ~

In the Staff Requirement Memorandum (SRM) dated June 15, 1990, regarding,the implementation of safety goals, the Co11V11ission directed that "Safety goals are to be used in a more generic sense and not to make specific licensing decisions.* Therefore, at this time, the NRC would use the safety goals in making regulatory decisions regarding backfitting new generic requirements but not to make specific licensing decisions including granting reli~f from unnecessary requirements. Any changes to the safety goal policy are outside the scope of the PRA policy statement and would, therefore, need to be pursued independently.

Referring to paragraphs 1 and 2 of the proposed policy statement, a commenter suggested that it should include the application to NRG enforcement decisions, including the severity levels. As noted in NUREG-1525, "Assessment of the NRC Enforcement Program," the Commission does not support defining severity levels using PRA results. The NRC's basis for severity level categorization clearly is safety significance. In judging safety significance, the NRC considers {l) actual consequences,

(2) potential consequences, and (3) regulatory significance. It is recognized that PRA results may be helpful to provide risk insights on the likelihood and significance of potential consequences. The NRC plans to continue to consider the use of PRA results where relevant as part of the integrated process considering all facets surrounding the violation in support of enforcement decisions.

Several co/lllenters discussed the role of PRA in reducing the unnecessary conservatisms in regulations anq to support additional regulatory requirements. One commenter's concern was that the proposed policy statement appeared to be biased in the direction of using PRA to support deregulation. Another commenter was concerned with the implication that PRA could result in an additional layer of regulation. The policy statement addressed the need to remove unnecessary conservatism associated with regulatory requirements. It is not the Commission's intent to replace traditional defense-in-depth concepts with PRA, but rather to exploit the use of PRA insights to further understand the risk and improve risk-effective safety decision-making in regulatory matters. In doing so, the Commission is focusing its attention and resource allocation to areas of true safety significance. Where appropriate, PRA should be used to support additional regulatory requirements, according to 10 CFR 50.109 (Backfit Rule).

One commenter recommended that the policy statement should explicitly state that the use of PRA by licensees in regulatory matters is at the discretion of each licensee. The commenter also believed that the NRG should not prescribe how and when PRA methods should be used by licensees in regulatory matters, but should address the potential impact the expanded

use of PRA may have on regu7atory interactions with 7icensees. The Convnission's PRA policy statement is intended only to encourage the NRC staff and industry to use probabilistic risk assessment methods in regulatory matters. It is not intended to prescribe or require any of the many potential PRA applications. Any requirements for licensees to perform PRA analyses would be expected to occur through formal rulemaking.

One commenter's concern was that there was a wide range of applications for which PRA was being applied without consistency and standards. This commenter urged the NRG to insist on quality PRAs commensurate with the intended applications and to develop standards.which require rigorous and living PRAs by regulation for nuclear power plant applications. The commenter also questioned whether the PRA analyses for the IP£ may be used for other applications because of a lack of PRA standards. Another commenter expressed the concern that strict conformance to detailed PRA standards would not be desirabl~, and reco/11/Dended that flexibility in PRA models should be a11owed. The Commission issued- Generic ...J Letter {GL) 88-20 with the primary purpose of generating IPEs to identify severe accident vulnerabilities. The PRAs which supported the IPE efforts may be useful for other applications, however, this would have to be evaluated on a case-by-case basis under well-defined objectives. After the Convnission briefing on the IPE program, the Conmission recognized, as stated in the SRM dated April 28, 1995, that current industry IPE results do not provide a complete basis for supporting risk-based regulatory decision-making. The SRM suggested that 11

• • • the industry should, in coordination with the staff, initiate the actions necessary to develop PRAs that are acceptable for risk-based regulatory use (i.e., standardized

- I~ -

methods, assumptions, level of detail)." The industry is encouraged to formulate a general approach for performing PRAs acceptable for regulatory use. This approach should include guidance on standardizing approaches for use of PRA techniques for specific applications, narrowing some of the variability in the IPE results, and strengthening its usefulness in the regulatory and safety decision-making process. The Conmission is currently considering the quality level and scope of assessment necessary to justify use of specific PRAs for specific regulatory applications. The Conmission will require PRA quality conmensurate with the proposed application.

PRA Methodology One commenter agreed with the NRC that the probabilistic approach should be used to complement the deterministic approach and that PRA numbers alone should not be used to make regulatory decisions. The commenter also believed that uncertainties should not prevent or delay the implementation of PRA in regulatory activities. The Commission understands that uncertainties exist in any regulatory approach. These uncertainties are derived from knowledge limitations that are not created by PRA, but are often exposed by it. The PRA implementation plan has provided a framework to assess the significance of potential uncertainties and to develop a strategy to accommodate them in the regulatory process.

One commenter stated that probabilistic analysis is simply an extension of deterministic analysis. They are not separate and distinctive concepts. The Convnission agrees with this concept as the proposed policy statement stated that "The probabilistic approach to regulation is,

therefore, considered an extension and enhancement of traditional regulation by considering risk in a more coherent and complete manner. 11 The Commission believes that the PAA method plays a complementary role in relationship to the deterministic method. This was reflected in the policy statement that RDetenninistic-based regulations have been successful in protecting the public health and safety and PAA techniques are most valuable when they serve to focus the traditional, detenninistic-based, regulations and support the defense-in-depth philosophy. 11 One co1R111enter recommended that the most efficient use pf NRG resources should be to enhance or improve the existing methods, but not to develop new ones. The Co11111ission's principal focus will be on improving the existing methods, but some new methods development may also be useful.

Another commenter recommended that the PRA policy statement should seek a uniform and standard application of PRA,within the NRG, and begin with a co111111itment to ensure that PRA is used consistently and is not ignored when required by those unfamiliar

. or reluctant to apply it.\. The Corm1ission's PAA policy statement specifically emphasizes the need for consistent and predictable application of PAA within the Commission to promote regulatory stability and efficiency. The Comission believes that this goal can be achieved through the implementation plan which will ensure that the appropriate use of PAA is implemented by the staff.

Schedule of PRA Activities Two letters commented that the activities discussed in the PRA implementation plan appeared to be on a protracted schedule and recommended

that priority and urgency be stressed and reflected in the plan, including the use of PRA and PRA insights in the near term. The Commission's PRA implementation plan showed the target completion dates for all the tasks.

The Cormtission fully realizes the need for near term PRA applications and has included them in the implementation plan wherever possible. These milestones include examples such as pilot applications for risk-based initiatives and transfer of IPE insights to NRC staff members for use in regulatory matters in the near term. The Convnission plans to periodically review the progress of the "living" PRA implementation plan and, as appropriate, to adjust the priorities.

One letter commented that the NRC review and approval of licensing actions that are based on PRA insights should not be contingent upon the schedule for implementation of the plan. The plan should not be an impediment to moving forward toward the goals outlined in the policy statement. The Commission's implementation plan had been developed. to effectively and expeditiously establish a framework for increasing the use of PRA technology inside the Convnission. Since it is a "living" plan, new tasks could be added and existing tasks could be modified, as the plan progresses. The Commission agrees that the plan should not be an impediment to moving forward to achieve the goals stated in the policy.

The Co11111ission welcomes risk-based regulatory initiatives from the industry as the plan is being carried out and will adjust resources, as appropriate.

One commenter asked how the NRC wi11 propose to control the utilities, application of PRA and the timeframe to implement the consistent use of PRA within the NRC. The Commission's PRA implementation plan describes the activities and schedule to effect a coherent and consistent PRA application

~ 17 -

within the agency. As the plan is implemented, the NRC expects to interact with licensees and publish guidelines for the application of PRA in their submittal to the NRC.

PRA Training Two coffllllenters advocated PRA training for appropriate NRG and licensee staff as soon as possible to ensure proper application of PRA in regulatory matters. A PRA training program has been in place for the NRC staff for a number of years. As part of the PRA implementation plan, the existing training program is being enhanced. The existing PRA training curriculum serves as the basis on which to build a more comprehensive staff PRA training program. Six new courses have been incorporated in the training program to address the short term needs from the increasing use of PRA in regulatory activities. As a result of the PRA implementation plan, ~the number of NRC staff participating in the training program has increased significantly during the first half of fiscal year 1995.

One commenter recommended that NRC's PRA training should be extended to State agencies that can justify attendance. Historically, attendance at NRC courses has been routinely available on a space-available, no-cost basis to State personnel as well as for other non-NRC personnel {such as foreign regulators, EPA, DOE, and other Federal personnel). This has included training in the PRA area for a limited number of State regulators.

In courses that were under-subscribed by NRC personnel, many had sufficient available space to allow acceptance of outside personnel. Logistics for

these arrangements are handled by the NRC office responsible for interactions with the outside group (i.e., Office of State Programs for States or Office of International Programs for foreign personnel). NRC training currently is not available to NRC licensees. Because of recent budgetary constraints, as described in SECY-95-017 "Reinventing NRC Fee Policies," full cost reimbursements from States for NRC training is expected in future years. However, NRC will continue its space-available policy for all courses, including PRA courses.

Data Collection Several co111111enters expressed concerns about the potential data collection implications of the proposed PRA policy. They are summarized as follows:

One commenter. stated that the desire to collect detailed data related to equipment and human reliability should not prohibit the use of PRA for applications or support for decision-making. The collection of plant-specific data must be co11111ensurate with the benefit that specific information might have on the quality or insight from the PRA. Plant-specific information may not be statistically significant. Furthermore, requiring all plants to collect the same information without a focus based on plant performance, is counter to the concept behind the Haintenance Rule.

Another commenter stated that the discussion of uncertainties in Part 11.(B) of the proposed policy statement is appropriate. However, in the implementation of this part of the policy, care must be exercised to

restrain from requiring or implying the need for massive plant-specific component level failure rate data collection programs. Several commenters expressed concerns that a new or expanded-nuclear power plant experience data co11ection rulemaking could further burden the licensees and the resulting benefit may we71 be marginal.

The Commission agrees that it should make every effort to avoid any unnecessary regulatory burdens in connection with collecting reliability

• and availability data. Specific, comments on the types of data that should or should not be collected will be addressed in connection with proposed data collection requirements when they are published for comment.

Radiation Medicine *'l One commenter recommended that NRC should abandon the use of the linear hypothesis in estimating radiation-induced cancer and*mutation risk.

The commenter further stated that the NRC's PRA implementation planhrefers to risk analysis to analyze nuclear medical devices and that, " ..* there are no nuclear medicine devices that have risk to be analyzed."

The International Commission on Radiation Protection, the United Nations Scientific Committee on the Effects of Atomic Radiation, and the National Academy of Sciences' Committee on the Biological Effects of Ionizing Radiation believe that, in the absence of convincing evidence that there is a dose threshold or that low levels of radiation are beneficial, the assumptions regarding a linear nonthreshold dose-effect model for cancers and genetic effects and the existence of thresholds only for certain nonstochastic effects remain appropriate for formulating radiation protection standards. NRC follows their guidelines. Although some data suggest the possible use of other models, there are still many scientists

who believe there are insufficient data to deviate from the "linear" hypothesis. The issue of realism involved in continuing the use of the "linear" hypothesis is expected to be a matter of debate over the coming years.

The NRC regulates radiation medicine, which includes both nuclear medicine and radiation oncology. The intent of the policy statement concerning medical applications is to refer to medical devices containing byproduct material, in particular, those used in radiation oncology. The term *nuclear medical device" was revised in the recent status update on the PRA implementation plan {SECY-95-079) and clarified in the policy statement.

Nuclear Waste One commenter recommended that the NRG expand its use of PRA to other areas such as radiological dose assessment during the site decommissioning process. The NRC intends to consider expansion of PRA techniques into additional areas with the proviso that the application of these techniques to these facilities should be tempered according to the complexity of the disposal system, its uncertainties and the estimated risk.

One commenter provided comments on several aspects of the proposed policy statement in the nuclear waste area. Regarding the scope of the policy statement, the commenter recommended that the policy statement be amended to include risk assessment applications other than power reactors.

The Commission agrees with that comment. The use of PRA should be considered for those applications that involve projecting system

performance for very long time periods, such as hundreds or thousands of years. The policy statement stated that the use of PRA technology should be increased in all regulatory matters. Another recommendation was to temper the co111111itment to PRA to reflect inherent risk differences associated with different waste management facilities. Because of inherent differences in the regulations and practices associated with the licensing of waste management facilities, the application of performance assessment

• (PRA is called performance assessment for waste management systems) techniques to these facilities should be tempered according to the complexity of the disposal system, uncertainties surrounding the system performance, and the estimated risk. The Commission also agrees with the comments regarding uncertainties in projecting repository performance and the use of technical expert judgment in assessing these uncertainties, but feels the PRA policy statement is not the appropriate forum to discuss these items applicable only to waste management.

Regarding the suggestion of describing the reasons for using th.e PRA and the application of PRA in regulatory activities, the Commission included the reasons for using PRA in Section III of the policy statement and added a description of the impact of PRA on the rule changes to 10 CFR Part 100 in the background discussion.

Another commenter expressed concern that the proposed policy statement inappropriately encouraged the use of PRA in the licensing and regulation of nuclear waste disposal facilities. The Commission disagrees with this comment since PRA techniques are acceptable in a performance assessment for the geologic repository, but are only part of the requirements for a license. The commenter was also concerned that any new regulations

proposed by the Environmental Protection Agency (EPA) and the NRC's 10 CFR Part 60 for a high-1eve1 waste (HUI) disposal facility proposed for Yucca Mountain will probably prohibit use of PRA for these facilities because of Type I faults at this site. The Co1M1ission anticipates that both probabilistic and deterministic hazard assessment methodologies will be applied to assess the significance of faulting at Yucca Mountain.

Furthermore, the Convnission does not interpret 10 CFR Part 60 so as to preclude the use of PRA as a basis for licensing a proposed repository at Yucca Mountain. The commenter did not agree with NRC's characterization of the waste disposal system as passive and believed that, at this time, there is no alternative to the use of deterministic techniques for waste disposal application because PRA techniques are in the embryonic stage. The 11 Fault Tree Handbook" (NUREG-0492, January 1981) refers to "passive" as a

" ... mechanism (e.g., wire) whereby the output of one 'active' component becomes the input to a second 'active' component." "Passive" is generally used for "engineered" components that have no moving parts. Since there are no "engineered" components that are "active" (or causing motion in another engineered component) in the post-closure phase of the potential geologic repository at Yucca Mountain, the NRC has applied the traditional PRA concept to the waste disposal system and referred to it as a "passive system." The remanded 1985 EPA Standard, 40 CFR 190, required a probabilistic analysis for a geologic repository. The NRC has developed this type of analysis since 1970 and has attained a state of maturity for these analyses that is accepted by internationally-known organizations (e.g., Organization for Economic Cooperation and Development (OECD)/Nuclear Energy Agency (NEA.

A number of editorial co/llBents were received on the role of PRAs in the licensing of waste disposal facilities. The NRC has incorporated the appropriate co1M1ents in this final PRA policy statement. III. Deterministic and Probabilistic Approaches to Regulation (A) Extension and Enhancement of Traditional Regulation The NRC established its regulatory requirements to ensure that a licensed facility is designed, constructed, and operated without und~e risk to the health and safety of the public. These requirements are largely based on deterministic engineering criteria. Simply stated this deterministic approach establishes requirements for engineering margin and for quality assurance in design, manufacture, and construction. In addition, it assumes that adverse conditions can exist (e.g., equipment failures and human errors} and establishes a specific set of design~basis events. It then requires that the licensed facility design include safety systems capable of preventing and/or mitigating the consequences of those design-basis events to protect the public health and safety. The deterministic approach contains implied elements of probability (qualitative risk considerations), from the selection of accidents to be analyzed as design-basis accidents (e.g., reactor vessel rupture is considered too improbable to be included) to the requirements for emergency core cooling (e.g., safety train redundancy and protection against single failure). The approach by the Commission for the use of performance assessment to implement its regulations for disposal of radioactive nuclear

waste (10 CFR Part 60 for high-level waste disposal and 10 CFR Part 61 for low-level waste disposal) also contains implied elements of probability. The results of the numerous calculations obtained from a performance assessment for a given performance measure and for a particular type of facility (e.g., a spectrum of values for ground-water travel time or individual dose} are expressed in terms of statistical distributions that express the probability that a given measure of performance will be attained. When this distribution is compared to the appropriate deterministic standard in the Commission's regulations, the probability of not exceeding the standard can be obtained from the part of the distribution that falls below this standard. PRA addresses a broad spectrum of initiating events by assessing the event frequency. Mitigating system reliability is then assessed, including the potential for multiple and common cause failures. The treatment therefore goes beyond the single failure requirements in the deterministic approach. The probabilistic approach to regulation is, therefore, considered an extension and enhancement of traditional regulation by considering risk in a more coh~rent and complete manner. A natural result of the increased use of PRA methods and techniques would be the focusing of regulations on those items most important to safety. Where appropriate, PRA can be used to eliminate unnecessary conservatism and to support additional regulatory requirements. Deterministic-based regulations have been successful in protecting the public health and safety and PRA techniques are most valuable when they serve to focus the traditional, deterministic-based, regulations and support the defense-in-depth philosophy. In addition, PRA techniques are appropriately used when

considering regulations defined in probabilistic terms, and for estimating safety of systems with very large uncertainties such as waste disposal systems (Note that PRA is called performance assessment for these waste disposal systems). Beyond its deterministic criteria, the NRC has formulated guidance, as in the safety goal policy statement, that utilizes quantitative, probabilistic risk measures. The safety goal policy statement establishes top-level objectives to help assure safe operation of nuclear power plants. The safety goals are intended to be applied generically and are not for plant-specific applications. For the purpose of implementation of the safety goals, subsidiary numerical objectives on core damage frequency and containment performance have been established. The safety goals provide guidance on where plant risk is sufficiently low that further regulatory action is not necessary. Also, as noted above, the Commission has been using PRA in performing regulatory analysis for the proposed backfit of cost-beneficial safety improvements at operating reactors {as required by 10 CFR 50.109) for a number of years. {B} Uncertainties and Limitations of Deterministic and Probabi1istic Approaches The treatment of uncertainties is an important issue for regulatory decisions. Uncertainties exist in any regulatory approach and these uncertainties are derived from knowledge limitations. These uncertainties and limitations existed during the development of deterministic regulations and attempts were made to acconvnodate these limitations by imposing

prescriptive, and what was hoped to be, conservative regulatory requirements. A probabilistic approach has exposed some of these limitations and provided a framework to assess their significance and assist in developing a strategy to acco11111odate them in the regulatory process. Human performance is an important consideration in both deterministic and probabilistic approaches. Assessing the influence of errors of commission and organizational and management issues on human reliability is an example that illustrates where current PRA methods are not fully developed. While this lack of knowledge contributes to the uncertainty in estimated risks, the PRA framework offers a powerful tool for logically and systematically evaluating the sensitivity and importance to risk of these uncertainties. Improved PRA techniques and models to address errors of convnission and the influence of organizational factors on human reliability are currently being developed. It is important to note that not all of the Commission's regulatory activities lend themselves to a risk analysis approach that utilizes fault tree methods. In general, a fault tree method is best suited for power reactor events that typically involve complex systems. Events associated with industrial and medical uses of nuclear materials generally involve a simple system, involve radiation overexposures, and result from human error, not equipment failure. Because of the characteristics of medical and industrial events, as discussed above, analysis of these events using relatively simple techniques can yield meaningful results. Power reactor events, however, generally involve complex systems and human interactions, can potentially involve more than one adverse consequence, and often result

from equipment failures. Therefore, power reactor events can require greater use of more complex risk analysis techniques, such as fault tree analysis, to yield meaningful insights. PRA methods need to be adapted for waste disposal systems because they are passive systems subjected to interlocking natural and man-made processes and events that are dominated by complex phenomenology. Given the dissimilarities in the nature and consequences of the use of nuclear materials in reactors, industrial situations, waste disposal facilities, and medical applications, the Convnission recognizes that a single approach for incorporating risk analyses into the regulatory process is not appropriate. However, PRA methods and insights will be broadly applied to ensure that the best use is made of available techniques to foster consistency in NRC risk-based decision-making. {C) Defense-in-Depth Philosophy

• In the defense-in-depth philosophy, the Commission recognizes that complete reliance for safety cannot be placed on any single element of the design, maintenance, or operation of a nuclear power plant. Thus, the expanded use of PRA technology will continue to support the NRC's defense-in-depth philosophy by allowing quantification of the levels of protection and by helping to identify and address weaknesses or overly conservative regulatory requirements applicable to the nuclear industry. Defense-in-depth is a philosophy used by NRC to provide redundancy for facilities with "active" safety systems, e.g., a commercial nuclear power, as well as the philosophy of a multiple-barrier approach against fission product releases.

Such barrier principles are mandated by the Nuclear Waste Policy Act of 1982, which provides redundancy for a geologic repository to contain and isolate nuclear waste from the human environment. IV. The Commission Policy Although PRA methods and information have thus far been used successfully in nuclear regulatory activities, there have been concerns that PRA methods are not consistently applied throughout the agency, that sufficient agency PRA/statistics expertise is not available, and that the Commission is not deriving full benefit from the large agency and industry investment in the developed risk assessment methods. Therefore, the Commission believes that an overall policy on the use of PRA in nuclear regulatory activities should be established so that the many potential applications of PRA can be implemented in a consistent and predictable manner that promotes regulatory stability and efficiency. This policy statement sets forth the Commission's intention to encourage the use of PRA and to expand the scope of PRA applications in all nuclear regulatory matters to the extent supported by the state-of-the-art in terms of methods and data. Implementation of the policy statement will improve the regulatory process in three areas: foremost, through safety decision making enhanced by the use of PRA insights; through more efficient use of agency resources; and through a reduction in unnecessary burdens on licensees. Therefore, the Commission adopts the following policy statement regarding the expanded NRC use of PRA:

(1) The use of PRA technology should be increased in all regulatory matters to the extent supported by the state-of-the-art in PRA methods and data and in a manner that complements the NRC's deterministic approach and supports the NRC's traditional defense-in-depth philosophy. (2) PRA and associated analyses (e.g., sensitivity studies, uncertainty analyses, and importance measures) should be used in regulatory matters, where practical within the bounds of the state-of-the-art, to reduce unnecessary conservatism associated with current regulat,pry requirements, regulatory guides, license commitments, and staff practices. Where appropriate, PRA should be used to support the proposal for additional regulatory requirements in accordance with 10 CFR 50.109 (Backfit Rule). Appropriate procedures for including PRA in the process for changing regulatory requirements should be

• developed and followed. It is, of course, understood that the.. intent of this policy is that existing rules and regulations shall be complied with unless these rules and regulations are revised.

(3) PRA evaluations in support of regulatory decisions should be as realistic as practicable and appropriate supporting data should be publicly available for review. (4) The Co1M1ission's safety goals for nuclear power plants and subsidiary numerical objectives are to be used with appropriate consideration of uncertainties in making regulatory judgments on the need for proposing

and backfitting new generic requirements on nuclear power plant licensees. Policy Implications There are several important regulatory or resource implications that follow from the goal of increased use of PRA techniques in regulatory activities .. First, the NRC staff, licensees, license applicants, and Commission must be prepared to consider changes to regulations, to guidance documents, to the licensing process, and to the inspection program. Second, the NRC staff and Commission must be committed to a shift in the application of resources over a period of time based on risk findings. Third, the NRC staff must undertake a training and development program, which may include recruiting personnel with PRA experience, to significantly enhance the PRA expertise necessary to implement these goals. Additionally, the NRC staff must continue to develop new and improved PRA methods and regulatory decision-making tools and must significantly enhance the collection of equipment and human reliability data for all of the agency's risk assessment applications, including those associated with the use, transportation, and storage of nuclear materials. However, it is recognized that there may be situations with material users where it may not be cost-effective to use PRA in their specific regulatory applications. This policy statement affirms the Commission's belief that PRA methods can be used to derive valuable insights, perspective, and general conclusions as a result of an integrated and comprehensive examination of the design of nuclear facilities, facility response to initiating events,

the expected interactions among facility structures, systems, and components, and between the facility and its operating staff. The Convnission also recognizes, and encourages, continuation of industry initiatives to improve PRA methods, applications and data collection to support increased use of PRA techniques in regulatory activities .

• V. Availability of Documents r

Copies of documents cited in this section are available for inspection and/or for reproduction for a fee in the NRC Public Document Room, 2120 L Street, NW, (Lower Level), Washington, DC 20037. Copies of NUREGs cited in this document may be purchased from the Superintendent of Documents, U.S. Government Printing Office, P.O. Box 37082, Washington, DC 20013-7082.

• Copies are also available for purchase from the National Technical, Information Service, 5285 Port Royal Road, Springfield, VA 22161.

In addition, copies of I} SECY-94-218, "Proposed Policy Statement on the Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities,n 2) SECY-94-219, "Proposed Agency-Wide Implementation Plan for Probabilistic Risk Assessment (PRA)," 3) the Cot11llission's Staff Requirements Memorandum of September 13, 1994, concerning the August 30, 1994, Conmission meeting on SECY-94-218 and SECY-94-219, and 4) the Commission's Staff Requirements Memorandum of October 4, 1994, on SECY-94-218 can be obtained electronically by accessing the NRC electronic bulletin board system (BBS) Tech Specs Plus. These four WordPerfect* 5.1

documents are located in the BBS MISC library directory under the single filename "PRAPLAN.ZIP". The WordPerfect~ 5.1 file for the final policy statement on the "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities," is located in the BBS MISC library directory under the filename "PRPOLICY.ZIP 11

• The BBS operates 24 hours a day and can be accessed through a toll-free number, 1-800-679-5784, at modem speeds up to 9600 baud with communication parameters set at 8 data bits, no parity, 1 stop bit, full duplex, and using ANSI terminal emulation.

Dated at Rockville, Maryland, this /(} day of August, 1995. FOR THE NUCLEAR REGULATORY COMMISSION Andrew[. Bates, Acting Secretary of the Commission.

BOB MILLER STATE OF NEVADA ROBERT R. LOUX Governor Executive Director

                                                                             *95 FEB 28 P :33 AGENCY FOR NUCLEAR PROJECTS NUCLEAR WASTE PROJECT OFFICE                  Of FlLL (1f Capitol Complex OGCKE r:1 Carson City, Nevada 89710 Telephone: (702) 687-3744 Fax: (702) 687-5277 DOCKET NUMBER PROPOSED RULE~,_.!...:,.l~ :::-i-February 22 , 1995           (5°' rt{. los~-aq)

Secretary

u. s. Nuclear Regulatory Commission Wash ington DC 20555 ATTENTION: Docketing and Service Branch

SUBJECT:

Comments on NRC Proposed Policy Statement "Use of Probabilistic Risk Assessment Methods"

Dear Mr. Secretary:

The purpose of this letter is to provide the state of Nevada's comments on the Proposed Policy Statement regarding "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities", as published in the Federal Register (FR-59-235, p.63389-91, December 8, 1994). Based upon our review of SECY 218, SECY-94-219, and the transcript of the "Probabilistic Risk Assessment Implementation Plan Public Workshop," (December 2, 1994), we feel that the four-item Policy Statement as found in FR-59-235, p.63391, is well formulated and appropriate as it relates specifically to the licensing of reactor structures, systems, and components (SSCs); maintenance of reactor sscs; in the context of backfitting new generic requirements for sscs on nuclear reactor plant licenses; and for the examination of risk from other operating nuclear facilities. We are concerned, however, that the Policy Statement and supporting background material unjustifiably encourages the use of Probabilistic Risk Assessment (PRA) in the licensing and regulation of nuclear waste disposal facilities, both high-level and low-level. Although we recognize that the Pol icy Statement, per se, does not specifically address nuclear waste disposal facilities, the supporting background discussion and reference material implies that the NRC has equated PRA to performance assessment and therefore PRA will be acceptable for use by the U. s. Department of Energy (DOE) in licensing a proposed geologic repository at Yucca Mountain, Nevada (FR-59-235, p. 63389, last paragraph). MAR ~ 1 1995 Acknow1edged by card .......uunm UlUUl; I. 68

I .S. NUCLl:AR REGULA ORY COMMISSIOIY DOCKETING & SERVl(,E SECTION OFFICE OF THE SECRETARY OF THE COMl'v:ISSION ostmark Oat." __s_\-2C:l+-~ ---- pies Rec.. :-r; ___ \ _ __ _ __ Adcfl Copies f"i p. ~ -.3

                          ~ -----

Speclal Di~tril:lut;ori & , TR \3)5 7

The State of Nevada feels that it is inappropriate for the NRC Policy Statement to make specific reference to Yucca Mountain as "the geologic repository" (FR-59-235, p. 63390, first paragraph). The DOE is still a long way from establishing the suitability of Yucca Mountain under 10CFR960, their own siting guidelines. Given the documented evidence of the presence of Type I faults (NUREG-1451) transecting and bounding the proposed repository, it is questionable that DOE will be able to effectively satisfy the present NRC postclosure licensing requirements found in 10CFR60 using either a deterministic or a probabilistic approach. It is obvious that the NRC's present reactor regulations, 10CFRlOO, woul d effectively preclude the use of PRA as a basis for licensing an operating nuclear facility with a systems lifetime of approximately 40 years, if it was located over a capable fault. We would submit that it is highly likely that the NRC's present high- level waste disposal regulations, 10CFR60, will also effectively preclude the use of PRA as a basis for licensing a proposed repository at Yucca Mountain with a systems lifetime of 10,000+ years, since it is located on at least two known Type I faults. The State of Nevada believes that no useful purpose is served by having the last paragraph on page 63389,(FR-59-235) included in the Proposed Policy Statement, except possibly for the public recognition of the ongoing NRC program to develop performance assessment methods applicable to waste disposal facilities. If this is the case, we suggest that, at a minimum, the sentence in the last paragraph on page 63389, that begins with "Because waste disposal systems are passive *** " be removed or extensively modified. There are two reasons for this. First, as we have stated earlier, Yucca Mountain may be the only site presently under consideration as a geologic repository, but the DOE has a long way to go before demonstrating that the site is suitable and licensable. Second, it is misleading and technically inaccurate to classify a waste disposal system as passive given the long time frame being considered (10,000+ years). Although most of the conceptual engineered sscs being proposed for the repository may be relatively passive in a mechanistic sense during the pre-closure phase (50-100 years), they would be, in the case of Yucca Mountain, subject to uncontrolled and unpredictable interactions with an active geologic system during the next 99 00+ years post-closure phase. Another way to look at the issue is to think of a reactor as an active engineered system that is located on top of a relatively passive geologic environment that may be periodically disturbed by an extreme event during the short 40-50 year facility lifetime. A waste disposal system, on the other hand, is a passive engineered system that is located within a relatively active (at least at Yucca Mountain) geologic environment that will continually be modified by ongoing geologic processes and will be periodically disturbed by extreme events during the 10,000+ years facility lifetime. In the case of the reactor 2

system, a PRA technology could be used as a complement to the NRC's deterministic approach to the extent supported by the state-of-the-art PRA methodology a nd data. In the case of the waste disposal system, because the development of PRA methodology for this application is barely in t he embryonic stage, and because there is little, if any data, on the performance of engineered sscs within the natural geologic system over the time frame under consideration, the State of Nevada believes that there is effectively no alternative at this time to using a strictly deterministic approach. Thank you for the opportunity to comment on this subject. If you have any questions, please give me a call. Sincerely , Robert R. Loux Executive Director RRL:CAJ:jem 3

ICFKAISER ENVIRONMENT & E ERGY GROUP DOCKETED ICF Kaiser Engineers, Inc. us *~c Four Gateway Center Pittsburgh, PA 15222-1207 412/497-2000 Fax 412/497-2212 DOCKET PROPO *95 FEB 24 A;. :22 February 7. 1995 ( so._~R-- l.o33-iq) Secretary U.S. uclear Regulatory Commission Washington, D.C. 20005 Attn: Docketing and Service Branch

Subject:

Federal Register Vol.59 No. 235, December 8, 1994, pp. 63389-63391. Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities; Proposed Policy Statement

Dear Sir or Madam:

This letter presents a recommendation regarding an expansion of the above mentioned technology to additional applications. The NRC has undertaken a rational technical approach in preparing to apply probabilistic risk assessment (PRA) to evaluate the potential for reactor malfunction. I recommend the NRC expand its use of PRA (otherwise known as stochastic modeling or "monte-carlo" analysis) to additional areas. such as the radiological dose assessment. This approach would serve to identify potential threats to public health due to residual constituents in soil, and would be a superior focus for identifying residuals which do not pose a threat to public health. As is noted in the NRC statement:

           "A result of the increased use of PRA methods and techniques would be the focusing of regulations on those items most important to safety by eliminating excess conservatism."

Such approaches have been repeatedly considered for the evaluation of ~ites containing residual chemical constituents in soils and groundwater. In these cases as well. the technology serves as a measure of the conservativeness of a deterministic health risk assessment and could also serve a radiological dose assessment. This measure should be equally applicable to the analysis of the extent of necessary remediation as it has been to the risk assessment for the protection of public health. An effort to incorporate PRA into the site, decommissioning process would seem to be timely, particularly in light of Dr. E. Gail de Planque's presentation during the NRC Workshop on Site Characterization for Decommissioning, in which a definition of background was attempted as well as a focus on defining acceptable levels of risk. Dr. De Planque states (p. 10):

           "Given that the risk associated with a 15 mrem residual dose adds very little to the risk of exposure to background and indeed is buried in the noise of natural variations of that background. then how much money and effort should be spent not only to dean up to this level. but to assure compliance?"

M-2252.hth2

• MAR *= 1 1995 1 Acknowledged by card .........................""°...~

13SIOt'-. C

ICFKAISER U.S. Nuclear Regulatory Commission February 7. 1995 Page 2 The answer to this question would be clearly defined in a site-specific dose assessment with a PRA supporting the identification of potential health risks or lack thereof associated with the constituents at the site. I hope the NRC will consider this effective technology for dose assessment as well as reactor regulation and use it to evaluate the necessity of remediation at sites nationwide. Sincerely,

      /)          er;     /

I '!f--,,_,----~/~--- Bruce E. Fishman, Ph.D. DABT - Chief Toxicologist M-2252.htb:!

DOCVET[O u::.mc Westinghouse Owners Group Domestic Utllllles 95 FEB 22 InternationalP1 *3 1Utllltles American Electric Power Georgia Power Pacific Gas & Electric TU Electric Electrabel Carolina Power & Light Florida Power & Light Public Service Electric & Gas Union Electric Kansai Electric Power Commonwealth Edison Houston Lighting & Power Rochester Gas & Electric Virg~ l 1 , , ~r } [ Kl~~l¥tric Power Consolidated Edison New York Power Authority South Carolina Electric & Gas Wis~C"Jf'pl"er-

• 11'!ucilj~"1f"ctric pie Duquesne Light Northeast Utilities Southern Nuclear WisMn'sfn'Pl,b~ S9lvic~ ' t *iiiuli.1.lama Elektrana i

Duke Power Northern States Power Tennessee Valley Authority Wolf Creek Nuclea,t5 f ( , \..d, Spanish Utilities OG-95-11 Taiwan Power Vattenfall February 15, 1995 Document Control Desk DOCKET NUMBER U.S. Nuclear Regulatory Commission PROPO E Pl'LE ffi\~ - Washington, DC 20555 ( 5q ~R- Co33-S9) Attention: Docketing & Services Branch Mr. J.C. Hoyle, Acting Secretary

Subject:

Westinghouse Owners Group Review of NRC Policy Statement and Implementation Plan on PRA The Westinghouse Owners Group has reviewed the proposed policy statement, "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities", noticed in the Federal Register, Volume 59, No. 235, dated December 8, 1994, and the proposed implementation plan, "Proposed Agency-Wide Implementation Plan for Probabilistic Risk Assessment (PRA)", SECY-94-219. We have also reviewed the Nuclear Energy Institute's (NEI) comments on the proposed policy statement. Our Owners Group has used risk-based methods in previous programs to support resolution of industry issues as appropriate. We are currently leading the technology development of risk-based methods in a number of important areas, and plan on increasing the use of risk-based methods in the future, as appropriate, to support risk-effective safety and operational decision-making on emerging applications. The policy statement provides an excellent opportunity to increase the use of these methods to let us focus on those items most important to safety and efficient plant operation. We have no specific comments on the policy statement, but we do agree with NEl's comments and encourage the NRC to consider the identified issues. We appreciate the opportunity to comment on the proposed policy statement, and we welcome the opportunity to increase the use of risk-based methods in our industry. Very truly yours,

  ~t~                                                   -

Roger A. Newton, Chairman Westinghouse Owners Group RAN/dac cc: WOG Steering Committee Westinghouse Owners Group Primary Representatives R.F. Saunders, Virginia Power C.K. McCoy, Georgia Power N .J. Liparulo, W Acfm Jed .MA!1 -= 1 1gg5 - K.J. Voytell, W ow ged by card .......................... ~'W***o

                        .,- Cu ,MISSIOt-.

[., f SECTION RETARY

                     , S'ON Pos'm Co .

Ad 3 _ _ __ Spec ,1s ** :£:De,,~ ~- -..,...- -

                                 ~tab of ~.efu ID.ers.ey Christine Todd Whitman         Department of Environmental Protection
                                                                                 *95 FEB 21 P2Robert     :56 C. Shinn, Jr.

Governor Commissioner Division of Environmental Safety, l }q, . rt and Analytical Programs 1, L~-, Radiation Protection ProgramQOCKL 1*3K Bureau of Nuclear Engineering CN 415 KET NU 1 BER Trenton, New Jersey 08625-0415 DOC C * - Tel (609) 987 - 2032 PRO i...U ------ '1:lC- = Fax ( 609) 987-6354 (SOi re.lo33 ~9) February 15, 1995 Secretar y U.S. Nuclear Regulatory Commission Attention: Docketing and Service Branch Washington, DC 20555

Subject:

Use of Probabilistic Risk Assessment (PRA) Methods in Nuclear Regulatory Activities Proposed Policy Statement on December 8, 1994, the Commission proposed a policy statement regarding the expanded NRC use of PRA in nuclear regulatory matters. The proposed policy encourages the use of PRA and expands the scope of PRA applications in all nuclear regulatory matters to the extent supported by the state-of-the-art in terms of meth ods and data. The New Jersey Department of Environmental Protection's Bureau of Nuclear Engineering (BNE) reviewed the NRC PRA policy proposal and has the following comments. In general, our experience and interaction with the NRC and utilities on PRA has been positive and we encourage its use at the NRC but remain cautious about possible abuses. We recommend that the NRC move forward but the NRC should l ose s i te of the limitations of PRA. Specifically, it is not clear how the four policy statements wi ll address the inconsistent use of PRA within the NRC. Before incre ased use of PRA at the NRC is pursued, the NRC should address and remedy* inconsistencies in its current use. The first goal of the policy should seek a uniform and standard application of PRA wi th i n the NRC. Therefore, the policy should begin with a commitment to ensure that PRA is used consistently and is not i gnored whe n required by those unfamiliar or reluctant to apply it. MAR =1 1995 Ae1mow1edged by card .................................... New Jersey is an Equal Opportunity Employer Recycled Paper

U.S. NUCLE.. OMMISSIOt-.- DOCKi:. T EC1 ION OfFI ARY N D cs Po3 m rk Cop. s . . Addi C SfJ C ~ ,'. )

page 2 K. Tosch Secondly, because of the complexity of PRA technology, it will be much more difficult for State agencies to properly perform a review of changes resulting from the application of PRA unless State agencies have access to NRC PRA training. since New Jersey performs numerous review functions and participates in NRC inspections at nuclear power plants, NRC PRA training should be extended to State agencies who can justify attendance. Also, the NRC po l icy should ensure that utilities are forthcoming with all supporting documentation when PRA based decisions are made. The NRC should expect longer review times for PRA based changes, at least until more people are knowledgeable with the technology. Therefore, more review time should be granted for those parties who review PRA based changes at nuclear facilities. Thirdly, we have observed similar inconsistencies with the application of PRA within utility organizations. How will t h e NRC and the proposed policy control the utilities' application of PRA and its implementation at the operating plants? Similar standards of consistency, sought by the NRC, are needed in utility organizations. Also, we would like the NRC to minimize a potential approach by utilities to se l ect PRA data to support cost saving changes without proper consideration to the overall impact upon safety. The utility has many resources available when searching for reasons to delete "unnecessary conservatisms." The NRC has to ensure that proper oversight is performed in order to ensure that overall plant changes move in the right direction - more cost efficient operations that increase safety - a goal which can be done if prudent changes take place at operating facilities. Finally, how much time does the NRC expect to take in implementing the consistent use of PRA technology within the NRC organization? If you have any questions, please contact Dennis J. Zannoni at (609) 987-2037. Sincerely, Kent W. Tosch, Manager Bureau of Nuclear Engineering

page 3 K. Tosch c: Distribution J. Lipoti, Assistant Director Radiation Protection Programs, NJDEP D. Zannoni, Supervisor Bureau of Nuclear Engineering, NJDEP D. Chawaga NRC Region I State Liaison Officer A. Dromerick, us NRC Project Manager, Oyster Creek D. Moran, US NRC Project Manager, Hope Creek L. Olshan, us NRC Project Manager, Salem 1 and 2

DEPARTM AR SAFE

                                                                    'VE                                                             -

162704 j;;, d ' - Jim Edgar Th ~~'"*)/

                                                                                                                         \ ,,.

Governor ' ,.

                                    -~."\'t~}~s\'"\.~~':--,.*;-~*

Co py faxed on 2/ 15/95. February 10, 1995 DOCKET NUMBER Secretary PROPOSED AUi E c... U.S. Nuclear Regulatory Commission (~f'R..~?>C\) Washington, D.C. 20555 Attn: Docketing and Services Branch

                                                                                                         @)

The Illinois Department of Nuclear Safety (IONS) appreciates the opportunity to comment on the proposed policy statement for Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities . IONS has acknowledged in commenting on numerous rulemakings in the last five years that Probabilistic Risk Assessments (PRAs) are one of the best tools available for analyzing and managing risks to the health and safety of the publ ic from nuclear power plant operation. Because of their value to the NRC, IONS, and power plant licensees in fulfilling the common mission of protecting the health and safety of the public, IONS believes high quality PRAs should be required and kept current. This conclusion was reached based on the following:

• PRA is a proven technology for analyzing and quantifying the risks to the health and safety of the public, and can be applied to the operation of nuclear power plants.

Weaknesses and uncertainties in the technology are understood by the practitioners and can be dealt with consistently by end users when given guidelines to follow.

• The policy statement expresses a desire to involve the technology in all regulatory matters, limited only by the state-of-the-art. It is not clear whether this means the current state-of-the-art in the industry, or the state-of-the-art of the technology used in developing a PRA being considered for use in some application. This can vary widely. IONS hopes it means the state-of-the-art of the industry.
• The Nuclear Energy Institute realizes the utility of PRAs in power plant regulatory matters and is encouraging licensees to do PRAs and keep them current. Not all licensees want to do this. We think they should be required to.

FEB 2 4 1995 Acknowledged by card ............................... J recyclable

U.S. NUCLE. J..AlOR COMMISSIOt--. DOC <HI G & SERVICE SECTION Ot-FICE: OF Tl-iE SECRETARY OF T'it COM 1.SSION D Postmark n~ ~, l'Slgs

                     \ _f_
                              ~-'o.OJ

( Cop*es Ft " Add l Cop1~5 ~ Specii'l~D 'ib* 1ti

U.S. Nuclear Regulatory Commission Page 2 February 10, 1995

• NRC has acknowledged the uti l ity of PRAs in numerous studies; is anticipating the licensees will use them to comply with the maintenance rule; and recently inspected to see if licensees were using them in analyzing plant configurations for doing on-line maintenance. PRA use is becoming widespread in the nuclear industry, but is inconsistent. Firm guidance is needed.
• Each licensee performed a safety analysis to comply with the Independent Plant Evaluation requirements of Generic Letter 88-20, so a PRA of sufficient quality already will have been developed to satisfy those requirements. Because of a lack of standardization, these analyses may or may not be useful for other applications.
• The policy statement states that use of PRAs should be used in all regulatory matters, in conjunction with the safety goal policy, to complement traditional deterministic approaches. PRAs should be as realistic as possible using publicly available supporting data. They should be used to reduce unnecessary conservatism associated with regulatory requirements, regulatory guides, commitments, and staff practices. They should be used to support additional regulatory requirements. This is a wide-ranging list of applications for which to apply a PRA that only satisfies GL 88-20 requirements.

If much of both the industry and NRC see the value of PRAs in managing risks to nuclear power plant operation, and a deterrent to using them in all regu l atory matters is their quality, then for the above reasons, IDNS believes the NRC should insist on high quality PRAs . We bel ieve the NRC should, by rulemaking, require PRAs of sufficient rigor to allow their use in all regulatory matters pertaining to power plant operation. Following this course would have several advantages:

• NRC will have a good means to evaluate and quantify how much margin of safety and defense-in-depth is being sacrificed as conservatisms are removed from regulations.
• Licensees already have a considerable investment in their PRAs. Upgrading them and keeping them current should not cause undue addit i onal burden considering the offsetting cost savings from the reduced regulatory requirements they likely will justify in the future.
• A considerable amount of training on applying PRAs is necessary for both licensee and NRC staffs before effective application is possible. Given that the maintenance rule is

U. S . Nuclear Regulatory Commission Page 3 February 10, 1995 a risk-based regulation (if NUMARC 93-01 guidelines as endorsed by the NRC are used to comply), and it becomes effective in July of 1996, not much time for training is left. Training to a relatively consistent quality product will lessen the NRC training burden, without affecting the licensee training efforts.

• Under the present practice of requiring sufficient quality of a PRA for a desired application, quality guidelines for various applications are needed, if the NRC hopes to have a consistently applied regulatory process across the industry.

A requirement for consistently rigorous and relatively current PRAs would simplify the development of these guidelines. We believe to achieve a satisfactory degree of standardization of PRAs that the guidelines should be stated in a Standard Review Plan chapter, or a similar document. The technical information used in a PRA should be made available for public scrutiny. The cornerstone of risk management, hence risk-based regulation, is the risk analysis. Without a rigorous and current risk analysis, in combination with deterministic analyses to reduce their inherent uncertainties, the effectiveness of risk-management is reduced. IONS urges the NRC to alter the present practice of only requiring sufficient quality of PRAs commensurate with the intended application, to requiring rigorous and living PRAs by regulation for nuclear power plant applications. We also urge the NRC to codify the safety goals. IONS has major concerns, however, about the application of the PRA policy outside of the nuclear power plant operation arena. The policy statement was written in very general terms, at the NRC Commissioners' request, so that use of PRA technology can be used in all regulatory matters, if appropriate. We are very concerned that NRC not blindly require the application of PRA technology to all regulatory matters without the determin-ation of when it is appropriate to do so. IONS licenses a large number of diverse industrial, medical and academic licensees in the Agreement State program. Many, if not most, of these licensees have no familiarity with PRA technology and would find the development cost of PRAs to be prohibitive. IONS will also have the responsibility for reviewing a license application for a low-level radioactive waste (LLRW) disposal facility for LLRW from Illinois and Kentucky. There is currently no available disposal capacity for LLRW generated in Illinois and Kentucky, as well as many other states throughout the country. We are concerned that the continued absence of disposal capacity will spawn the creation of a large number of storage facilities and

U.S. Nuclear Regulatory Commission Page 4 February 10, 1995 may result in curtailment of uses of radioactive materials in socially beneficial practices. The PRA policy could further complicate and extend the efforts to develop new LLRW disposal facilities, if the policy is used as a delaying tactic by those opposed to the facilities. It could place a burden on licensees and Agreement States if applied to medical and sealed source licensees. If applied outside of the power plant arena, the training required could be a burden to both non-power plant l icensees and Agreement States. Therefore, before PRA technology is required beyond power plant regulatory matters, consideration should be given to the benefits to be achieved and the burdens to be placed on all classes of licensees and Agreement States. Finally, the proposed rule is silent on whether it will be a matter of compatibility for Agreement States.

                                  ~\ .

Thomas W. Ortcig r Director TWO/nth

tr..-.

 ~

CENTERIOR ENERGY DOCKETED us~mc 300 Madison Avenue John P. Stetz Toledo, OH 43652-0001 419-249-2300 *95 FEB 14 P4 :27 Vice President - Nuclear Davis-Besse Serial Number 2273 February 7, 1995 DOCKET NUMBER PROPO E R LE \SC-Secretary ( 5'\ f'R. 1,;!>3'!!'\) @ United States Nuclear Regulatory Commission Document Control o~sk ijashington, D.C. 20555-0001 Attn.: Docketing and Service Branch

Subject:

Comments Regarding the NRC Proposed Policy Statement on, "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities" (59FR63389) Gentlemen: Toledo Edison, a subsidiary of Centerior Energy, is partial owner of and is responsible for operation of the Davis-Besse Nuclear Power Station (DBNPS). As a 10 CFR part 50 licensee, Toledo Edison has a vested interest in any policies that the U. S. Nuclear Regulatory Commission (NRC) may adopt which can affect the management and operation of a commercial nuclear power plant. Industry Comments - Toledo Edison has reviewed the NRC proposed policy on "Use of Probabilis-tic Risk Assessment Methods in Nuclear Regulatory Activities", noticed in the Federal Register on December 8, 1994 (59FR63389) and the comments prepared on behalf of the nuclear power industry by the Nuclear Energy Institute (NEI). Based on th is review, Toledo Edison endorses the industry comments being provided by NEI in response to the 59FR63389 notice. Toledo Edison Comments Toledo Ed i son is supportive of the establishment of an overall policy on the use of probabilistic risk assessment (PRA) in nuclear regulatory activities, so that the many potential applications of PRA methodology can be implemented in a consistent and predictable manner that promotes regulatory stability and efficiency, and enhances safety. Operating Companies: Cleveland Electric Illuminating

FEB 2 4 1995 Toledo Edison Acknowledged by card ................................

U.S. NUCLE* HR vv:.. M'-....~101'. DOCKfTIN"' & '

• WE St:CTIO OFf" - ~ I tE 1 ARY 0 ION I\;,

Posti"'.' r ~\~\Q,s-Cop*as F'..." * \_ Add'! Cor, -~ ~- Spvci I , "37))e. I ~ - . J . o -- -

~:01

Serial Number 2273 Page 2 The proposed policy statement is general in nature, but does contain important elements in effectively utilizing PRA technology. Two poten-tial enhancements to the proposed policy statement are as follows:

1) Vhen applying a technique as general as PRA, the basic tech-nical approach and assumptions are as important as the sup-porting data in achieving a result which is "as realistic as possible." Therefore, the following change is proposed to paragraph 3 of the policy statement (proposed change under-lined):

(3) PRA evaluations in support of regulatory decisions should be as real istic as possible and appropriate supporting data should be publicly available for review, along with a mean-ingful description of the subject probabilistic analysis.

2) The use of PRA technology in all regulatory matters, as referenced in paragraphs 1 and 2 of the proposed policy statement, should include the consistent application of PRA results to NRC enforcement decisions, i~cluding the severity level of violations.

In summary, Toledo Edison believes that implementation of the proposed policy statement would improve the regulatory process in the areas of risk-effective safety decision making, efficient use of NRC resources and reduction in unnecessary burdens on licensees. Should you have any questions or require additional information, please contact Hr. Villiam T. O'Connor, Manager - Regulatory Affairs, at (419) 249-2366. Very truly yours, cc: L. L. Gundrum, DB-1 NRC/NRR Project Manager J.B. Hartin, Regional Administrator, NRC Region III S. Stasek, DB-1 NRC Senior Resident I nspector

5000 Dominion Boulemrd Glen Allen, Virginia 23060 DOCKE TED US 'Re DOCKET NUMB ER PROPOSED RULE..:-.::.~ --

                                                *95 FEB 13 p 1 :S9     (.5C\fta.~i9)

February 9, 1995  !.,._J~ Secretary, U.S. Nuclear Regulatory Commission Serial No. GL 94-083 Washington, DC 20555 NL&P/MAE RO Attention: Docketing and Service Branch

Dear Sir:

USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES: PROPOSED POLICY STATEMENT In the December 8, 1994 Federal Register, the NRC announced that it is proposing a policy statement regarding the use of probabilistic risk assessment (PAA) in nuclear regulatory matters. The proposed policy statement is intended to improve the regulatory process through improved risk-effective safety decision-making, through more efficient use of agency resources, and through a reduction in unnecessary burdens on licensees. Virginia Power encourages the NRC's efforts to solicit industry advice and recommendations in developing this proposed policy statement. However, Virginia Power is concerned that NRC implementation of the proposed policy statement could result in a new or expanded nuclear power plant experience data collection rulemaking, requiring station resources to provide equipment and human reliability data to the NRC. Additional comments are attached. We also endorse the comments submitted separately on this issue by the Nuclear Energy Institute in their letter dated February 1, 1995. If you have any questions, please contact us. Very truly yours, ~~::zger Nuclear Licensing and Programs cc: Mr. Thomas E. Tipton Nuclear Energy Institute 1776 I Street Suite 400 Washington, D. C. 20006-3708

                                                                                   ,FEB 2 4 1995 Acknowledged by card .........................n.-.

U.S NJCLEAh R \)uLAlORY COMMISSIOt--. 0Ovi\~ ,'-IG & SER ICE SECTION OF I f ~J- TI-.E SECRETARY OF THr ~OMVi1SSION Poslmarx o* * .0. Coples Rec 49.5___ J - - ---- Add1 Copies ~ -- Sp cial o*st: I~ ~l_ns_,_ ':).b.:,)

Comments on "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities; Proposed Policy Statement" Virginia Power welcomes the establishment of a Nuclear Regulatory Commission policy towards Probabilistic Risk Assessment (PRA) usage, and encourages the development of constructive PRA applications to reduce licensee burden while enhancing safety. Surry Power Station has been evaluated within two NRC sponsored PRA projects, with positive safety improvements as a result. Virginia Power has completed both Individual Plant Examination (IPE) and Individual Plant Examination External Events (IPEEE) analyses for Surry and North Anna Power Stations (excluding seismic analysis), and is developing PRA applications utilizing this technology. In support of the proposed policy statement, Virginia Power has two comments. First, the third item in the proposed policy statement states:

       "(3) PRA evaluation in support of regulatory decisions should be as realistic as possible and appropriate supporting data should be publicly available for review."

The wording "realistic as possible" is too absolute in a world of limited resources. Existing analyses such as NUREG-1150, "Severe Accident Risks: And Assessment For Five U.S. Power Plants," WASH-1400, "The Reactor Safety Study And Assessment Of Accident Risks In U.S. Commercial Nuclear Power Plants," IPE, and IPEEE reflect a more practical limit of realism. Hence, the wording "realistic as practicable" is suggested to reflect an efficient use of resources. Second, the above quoted item 3 indicates a possible change in public policy towards nuclear power plant experience reported in the public domain. Virginia Power recognizes that NRC may require additional plant experience information for agency use in implementing the proposed policy statement. We currently understand that the Office for Analysis and Evaluation of Operational Data is considering a Rulemaking to obtain system and equipment reliability and availability data. We are concerned changes or increases in plant experience reporting could further burden the nuclear power licensee.

                                                                      ....        DOCK ETED CP&L                                DOCKET NUMBER      PR                             US F'C PROPOSED RULE.__!..!!...lll,.l~ilallllf Carolina Power & Light Company           ( so..FR 6~~~~                       *95 FEB 13 A11 :10 PO Box 1551 411 Fayetteville Street Mall Raleigh NC 27602                                                           0 FF Ic- f ,    ... l :  ~ ~ , rf{

ODC Kc. T11 ,G ,.- . v;r~l

                                                                                    ~:k

• I *'

February 7, 1995 FILE: X-X-0529 SERIAL: NL&RAS-95-006 X-X-1075 Secretary U.S. Nuclear Regulatory Commission ATTN: Docketing and Service Branch Washington, D.C. 20555

Subject:

Comments on Proposed Policy Statement - Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities (59 FR 63389, December 8, 1994)

Dear Sir:

These comments are submitted on behalf of Carolina Power & Light Company (CP&L) and endorse the Nuclear Energy Institute (NEI) comments on the proposed Policy Statement - Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities (59 FR 63389, December 8, 1994). CP&L supports the increased use of Probabilistic Safety Assessment (PSA) by the NRC and concurs in the NEI comments dated February 1, 1995. In order to enhance the benefit of the use of PSA, the NRC policy statement should acknowledge the use by licensees of the forthcoming NEI PSA Applications Guide. This will ensure common understanding by licensees and by the NRC Headquarters and Regional staff of appropriate use of PSA in licensing applications, and will further the common goal of expanding the use of PSA into areas that will benefit both licensees and NRC staff.

                                                    ~R.E.Ro~

Manager, Nuclear Licensing and Regulatory Affairs FAE/alr cc: Mr. H. W. Habermeyer, Jr. Mr. T. E. Tipton (NEI) Mr. J. Presley Mr. K. Jury Mr. L. Rowell Mr. R. E. Oliver Mr. T. B. Clements Mr. F. A. Emerson ifEB 2 4 1995 Acknowledged by card ...............................~ n:\nad\nla\nria\2339nria

u.s. Nu

• c ,. ror e,o.~~i.11ss1or-.

DOC I INl1 .:,[l-1 ICE SECTION OF+i E OF nu= EC l "ARY OF TH::: COM l SSION D ,.. t "' istics Post:i, rk D"t Copies R * *

                  *' C-1~----

Add'l Copes Specia' p v, J ~ 13 \l>:'5,

WASHINGTON PUBLIC POWER SUPPLY SYSTEM P.O. Box 968

• 3000 George Washington Way
• Richland, Washington 99352-0968 * (509] 3~2-500Q February 8, 1995 GO2-95-029 Secretary U.S. Nuclear Regulatory Commission DOCKET UMBER Washington, D.C. 20055 PROPOSED RULE ffi\-.S<:- -

C5C\rR...to~-::i~~> Attention: Docketing and Service Branch (5)

Subject:

USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES, PROPOSED POLICY STATEMENT (59 FR 63389) The Supply System appreciates the opportunity to comment on the subject Nuclear Regulatory Commission (NRC) proposed Policy Statement. This policy statement is important to the industry in that it encourages the use of Probabilistic Risk Assessment (PRA) in the regulatory process. This use of PRA will lead to more efficient use of NRC resources and reduction of regulatory burdens on licensees. The Supply System supports the stated objectives of the policy statement and agrees that the proposed policy statement generally meets this intent. The Supply System has reviewed and endorses the Nuclear Utilities Backfitting And Reform Group (NUBARG) comments concerning the proposed policy statement. In addition, the following comments are provided for your consideration: Item ill(2) of the proposed policy states appropriately that 11 PRA and associated analyses . . . should be used in regulatory matters . . . to reduce unnecessary conservatism . . . " However, it also states ". . . PRA should be used to support additional regulatory requirements. " The implication is that PRA could result in additional regulatory requirements layered upon the existing conservative deterministic requirements. The philosophy of "defense-in-depth" is supported by the use of PRA processes in conjunction with more realistic deterministic analyses. Regulations that follow would result in cost benefit without compromising safety. This would justify the modification of existing regulations without distinguishing whether these modifications are additions or deletions. Therefore, the Supply System suggests that the second sentence of Item 111(2) of the proposed policy be deleted and replaced with the words "PRA should be used to support a consistent cost beneficial set of regulatory requirements that maintain the II defense-in-depth II philosophy. " FEB 2 4 1995 Acknowledged by card .................." .....w-.,., 0 0

U.S. NUCLE/.lR REGULATORY COMMISSIO~ DOvt<f SfRV LE SECTION Oft-ICE OF~ i: c- ~CRET~ o T ~c , s,m Po~ ITlc. Co *~:; Add' C

Page2 USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES, PROPOSED POLICY STATEMENT (59 FR 63389) In "Policy Implications," the last sentence states ". . . the NRC staff must continue to develop PRA methods . . . " Many PRA methods have been developed by the industry, et. al., and approved by the NRC. Development of new PRA methods, rather than enhancing or strengthening existing PRA methods is not the most efficient use of NRC resources. We recommend changing the word "develop" to "enhance" or "improve" to acknowledge the existence of previously developed and approved PRA methods. Again, we appreciate the opportunity to comment on this important Policy Statement and believe these comments, if incorporated, will enhance the final product. Sincerely,

&£J;,-t_

a;*. '::sh (Mail Drop 1023) Vice-President, Nuclear Operations AGC/ml

I OOC t, ETEO NUCLEAR ENERGY INST ITJ~::f :RC

                                                                           *95 FEB - 9 P2 :27                 Thomas E. npton VICE PRESIDENT orr ic                        , \i    OPERATIONS & ENGINEERING OO[)U I

• February 1, 1995 Mr. John C. Hoyle Acting Secretary U.S. Nuclear Regulatory Commission Washington, DC 20555-0001 ATTENTION: Docketing and Service Branch

SUBJECT:

Proposed Policy Statement, "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities" (59 Fed. Reg. 63389 - December 8, 1994) Request for Comment

Dear Mr. Hoyle:

These comments are submitted by the Nuclear Energy Institute (NEI)l, on behalf of the nuclear energy industry, in response to the request for comment on the proposed policy statement, "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities," noticed in the Federal Register dated December 8, 1994 (ref. 59 Fed. Reg. 63389). The industry has been in the forefront of recent activities to expand the use of probabilistic safety assessment (PSA). In this regard, the primary purpose of NEI's Regulatory Threshold Working Group is to promote the use of PSA technology and other new approaches to regulation as an aid to focus industry and regulatory attention and resources more effectively. We believe the Commission's proposed policy statement is supportive of this purpose. 1 NEI is the organization responsible for establishing unified nuclear industry policy on matters affecting the nuclear energy industry, including the regulatory aspects of generic operational and technical issues. NEI's members include all utilities licensed to operate commercial nuclear power plants in the United States, nuclear plant designers, major architect/ engineering firms, fuel fabrication facilities, materials licensees, and other organizations and individuals involved in the nuclear energy industry. NEI is the successor organization to the Nuclear Management and Resources Council (NUMARC). ffEB 2 4 1995 1776 I STREET . NW SUITE 400 WASHINGTON, DC 20006-3708 PHAiekrwwledgedty car&.,49.'i.i'.M,J,em1,,..,., ""*~

P r \ :.n

                             *95 FEB -7 AB :07 OIVISI H i:- CO PROPERTY 1 ,    ~

U. . ,~UCL': R ff' COMM'SSIO, OOCKEr * :1cE SECTION OF IC, EO'JC"':ARY OFT -.S:ON cs C' ' - ViJ ,

:\.L,: ,,) --

Mr. John C. Hoyle February 1, 1995 Page 2 Our comments on the proposed policy statement are focused in three principal areas:

• NRC staff expectations for use of PSA in the regulatory process by licensees;
• Deterministic regulation vs. risk-based regulation; and
• NRC implementation of the policy statement.

While the policy statement calls for expanding NRC use of PSA, it does not address the potential impact this expanded use may have on regulatory interactions with licensees. We believe this is an important aspect that should be included in the policy statement, given that there are no regulations requiring the use of PSA by licensees. The policy statement should be explicit in stating that the use of PSA by licensees in regulatory matters is at the discretion of each licensee. While encouraging the use of insights gained from PSA, the NRC should not prescribe how and when PSA methods should be used in regulatory matters. However, we acknowledge that it is appropriate for NRC to review licensee applications of PSA when it is used by licensees as the principal basis for licensing actions. Many licensees are already using PSA insights in the regulatory process today. In this regard, the industry is developing a PSA Applications Guide with the aim of establishing a common understanding of the "ground rules" for use of PSA. It is our intent that this guidance serve as a reference for both the development and review of PSA applications. We believe the Guide will complement the NRC's PRA Implementation Plan activities and will also serve our common goal of expanding the use of PSA. Our second area of comments focus on the differences between deterministic regulation and risk-based regulation. We believe the proposed policy statement is correct in stating that probabilistic methods should be used in a manner that complements current deterministic regulatory approaches. In our view, the complementary role of probabilistic methods is to aid in focusing licensee and NRC resources in areas of safety significance. If risk-based regulation evolves into a supplemental or additional layer of regulation atop the current deterministic framework, we believe this would only add burden to the regulatory process. The potential for conflict does exist, however, between the traditional deterministic licensing basis philosophy and risk-based approaches. For example, several industry studies have shown that there are many structures, systems and

Mr. John C. Hoyle February 1, 1995 Page 3 components (SSCs) in our plants classified as safety-related that have little safety content. In the current regulatory regime, these SSCs are viewed equally with other safety-related SSCs that have much higher safety significance. Another example of potential conflict is the use of conservative design basis assumptions in deterministic analyses versus the use of realistic, best-estimate assumptions in probabilistic assessments. The underlying concern is that these potential conflicts could divert resources and attention away from areas of true safety significance. We believe these potential conflicts are barriers that must be overcome in order to enhance the use of PSA in regulatory matters. As such, we believe the Commission should address these areas in the policy statement to ensure that finite NRC and industry resources are allocated to obtain commensurate safety benefits. One related idea that NEI's Regulatory Threshold Working Group has been considering is a recommendation to petition the NRC to modify the definition of "safety-related" in the code of federal regulations to include a risk-significance criterion. To ensure coherency in the regulatory process, insights from PSA must be reflected in both the intent and letter of the body of regulations and implementing guidance. Our final comments focus on NRC implementation of the policy statement. It is our understanding that the proposed NRC PRA Implementation Plan discussed in SECY-94-219 is essentially the plan for implementing the policy statement. As such, we believe the plan represents a significant commitment by NRC to move forward with the enhanced use of PRA in regulatory matters. We also believe that many current industry activities are complementary to the actions discussed in the plan. We look forward to coordinating our efforts, where appropriate, with the NRC activities discussed in the plan so as to make the most efficient use of both industry and NRC resources. Our principal concern with t he plan is timeliness. Many of the activities discussed in the plan are on an extremely protracted schedule. Priority and urgency need to be stressed and reflected in the plan in order to assure a more coordinated use of PSA in regulatory matters. Secondly, we believe that NRC review and approval of licensing actions that are based on PSA insights should not be contingent upon the schedule for implementation of the plan. In other words, we believe there is sufficient maturity in PSA technology to make sound regulatory decisions based on PSA insights today. Accordingly, the plan should not become an impediment to moving forward toward the goals outlined in the policy statement in the near term. The same can be said for the consideration of uncertainties. Uncertainties exist in the current deterministic framework where decisions are made daily. The capability of PSA to reveal and highlight uncertainties should not hamper its use as a decision-making tool. In addition, we believe the impact of uncertainties on decision-making is not as significant as is currently perceived, particularly when

Mr. John C. Hoyle February 1, 1995 Page4 mean values and relative risk ranking techniques are used in an evaluation. While we are generally supportive of research efforts to refine PSA methods and reduce uncertainties, these efforts should not forestall current PSA applications by licensees. In summary, we believe the proposed policy statement and implementation plan should address the above concerns. It is important to demonstrate progress in regard to the use of PSA applications by licensees in the short-term. Otherwise, it will be difficult to justify and sustain industry interest and resources in the long-term that are necessary to support greater use of PSA technology. We appreciate the opportunity to comment on the proposed policy statement. We look forward to future interactions with the NRC as we seek to expand the use of PSA in our industry. TET/jlm

OOC~ffTEO Detroit Edison Fermi 2 6400 "lorh [J1x e f-i14hway NEwport Micr1qar 48166 (313\ 586 5300 US !P.C

                                                                                          -~       Nuclear Operations
                                            *95 FEB - 8 P 1 :54 DOCKET NUMBER    PR PROPOSED RULE               .

(_iSqrR.b53~ct February 7, 1995 NRC-95-0007 Secretary U. S . Nuclear Regulatory Commission Washington, D. C. 205 55 Attn: Docketing and Service Branch

Reference:

Fermi 2 NRC Docket No. 50-341 NRC License No. NPF-43

Subject:

Detroit Edison Comments on the "Proposed Policy Statement on Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities," Federal Register Vol. 59, No. 235, December 8, 1994 The purpose of this letter is to provide Detroit Edison's comments on the sub j ect proposed policy statement regarding the use of probabilis ti c risk assessment ( PRA ) in nuclear regulatory matters. In general, Detroit Edison supports the comments of the Nuclear Energy Institute (NE I ) made in behalf of the nuclear industry on this document. Specific comments by Detroit Edison follow. First, Detroit Edison is pleased that a policy statement on PRA use in nuclear regulatory matters is being formulated by NRC. Such a policy statement is necessary for a continuing meaningful and consistent utilization of PRA in the regulatory arena to achieve a regulatory emphasis that is more commensurate with risk. Thus, the p olicy characterization of the use of PRA technology as complementing the deterministic licensing approach is appropriate so long as it is recognized that some deterministic based decisions will be subject to modification based on the application of best estimate risk considerations. It is assume d that this accommodation for change is what is meant by "the NRC staff, licensee, and Commission must be prepared to consider changes to regulations, to guidance documents, to the licensing process, and to the inspection program" as stated under "Policy Implications" in the proposed policy statement. Detroit Edison also agrees that application of PRA technology to regulatory matters should not extend beyond applications ifEB 2 4 199~ Acknowledged by card .......................... *....

CO .1MISSION SE;TION ETARY ION Pest "'I -- ~ I l

Secretary February 7, 1995 NRC-95-0007 Page 2 supported by the state-of-the-art in PRA methods and data. However, that limitation criterion is not precise. The existence of some uncertainty in either data or methods should not automatically be equated to exceeding the "state-of-the-art" and thus a priori preclude a particular application of PRA. Technical judgment in this matter is essential. The discussion of uncertainties in Part II. (B) of the proposed policy statement is appropriate. However , in the implementation of this part of the policy, care must be exercised to restrain from requiring or implying the need for massive plant specific component level failure rate data collection programs. Collecting such data in an accurate and consistent manner for PRA application is not an easy task, and the resulting benefit may well be marginal. Initiating event frequency and system unavailability data are more likely to drive the PRA results. Absence of a formal all inclusive component data gathering program does not preclude the identification and treatment of true component outliers ("bad actors") . Two additional concerns deserve comment. First, while recognition has been given in the proposed policy statement to the desirability of consistency in PRA use , this should not be construed to mean that all PRA models should be made uniform. Adherence to basic minimum standards as proposed by NEI's PSA Applications Guide (under development) may be a reasonable approach, but strict conformance to detailed standards would be unworkable. Allowance of model flexibility should be recognized in the policy statement . In fact , a t this period of time for currently operating plants , the option for a utility to completely forego the use of PRA should be maintained . Secondly, there is a concern caused in part by the statements of some NRC personnel that application of this policy could result in requirements for a shutdown PRA and for highly prescriptive methods for maintaining a "living PRA." A shutdown PRA model and "risk meters" could be useful, but they can also be very expensive. The former is not required to assure an appropriate level of safety during shutdown operations , and t he latter is certainly not required for useful and valid PRA application . Much can be done with current PRA technology in time frames much greater than the "on- line" implication of a risk meter. It is appropriate that the PRA model be a reasonable representation of the plant, but the very nature of a PRA model generally does not require the precision of update that is applied, for example , to a control room simulator used in operator training. In summary, Detroit Edison feels that PRA methods should indeed be used to provide a risk based component to regulatory matters, and a policy statement should help achieve this end. However , care needs to be exercised so that the end product is

Secretary February 7, 1995 NRC-95-0007 Page 3 not simply an additional layer of PRA related licensing requirements superimposed on inflexible deterministic requirements. Moreover , definite steps toward a more risk based licensing approach need to be taken in the near term and not wait until all PRA shortcomings are resolved. Much can be gained with current PRA technology and using generic data. Inordinate delay in achieving meaningful regulatory application will make it difficult for industry to maintain their current momentum in expanding the quality and utilization of PRA technology. Suggested editorial changes in the policy statement as i t now stands are given in Attachment A. If you have any questions, please contact Mr. Earl Page at (313) 586-4266. Sincerely, Lynne S. Goodman Director, Nuclear Licensing Attachment cc: T. G. Colburn J.B. Martin M. P. Phillips A. Vegel

• Attachment to NRC-95-0007 Page 1 Attachment A Editorial Changes Suggested for Section II of Proposed PRA Policy Statement Section II(A), third paragraph, sentences 2-4:

Comment - The important PRA characteristic of realism versus arbitrary failure assumptions is missed in the comparison as stated. Suggest new wording as follows for sentences 2 through 4. New Wording (Replace sentences 2-4) - "Mitigating system reliabi l ity is then realistically assessed without regard to safety classification. Single failure is not mandated as in the deterministic approach, but multiple failures are allowed including the effects of common cause. The probabilistic approach to regulation is therefore considered an extension and enhancement of traditional regulation by considering risk in a more realistic, coherent, and complete manner." Section II. (A), third paragraph, last sentence: Comma is needed after "public health and safety." Section II. (B), first paragraph, second sentence: Replace "are derived" with "result." Section II. (B), first paragraph, third sentence: Comma is needed after "deterministic regulations . "

WINSTON & S TRAWN 35 WEST WACKER DRIVE 1400 L STREET, N.W. CHICAGO, ILLINOIS 60601-9703 WASHINGTON, D.C. 20005-35 (312) 558-5600 FACSIMILE (312) 558-5700 (202) 371-5700 175 WATER STREET NEW YORK, NY 10038-4981 (212) 269-2500 FACSIMILE (212) 6584700 February 7, Secretary U.S. Nuclear Regulatory Commission ffi-'-~Q... =-- Washington, D.C. 20555-0001 ( 5C\ f~(o 3~ ~9) ATTN: Docketing and Service Branch (0 Re: Proposed Pol icy Statement Regarding "Use of Probabil i stic Risk Assessment Methods in Nuclear Regulatory Activities*

Dear Mr. Secretary:

On December 8, 1994, the Nuclear Regulatory Commission (NRC) published for comment a proposed Policy Statement entitled "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities." 59 Fed. Reg. 63,389 (1994). On behalf of the Nuclear Utility Backfitting and Reform Group (NUBARG), we appreciate the opportunity to provide our views on this important policy statement. The Commission states that its purpose in issuing the Policy Statement is to encourage the use of PRA and to expand the scope of PRA applications in all nuclear regulatory matters to the extent supported by the state of the art. The Policy Statement would thus improve the regulatory process by enhancing "risk-effective safety decision making," making more efficient use of agency resources, and reducing unnecessary burdens on licensees. 59 Fed. Reg. at 63,391. We support the NRC' s purpose in issuing the proposed Policy Statement and agree that it appears generally helpful. We provide below some specific suggestions for improving the Policy Statement. Use of PRAs in Support of Specific Licensing Actions Licensees have expended tremendous resources in the past few years in developing PRAs for their plants through their own efforts as well as the NRC-requested Individual Plant Examinations ifEB 2 4 1995 Acknowledged by card ........................., . ...,,,o

U.S. NUCLEAR REGULATORY COMMISS10r. DOCKETING i SERVICE SECTION OFFICE OF THE SECRETARY OF THE COMMISSION Postmarl-: D* 1* _;;:~~ '-'...., 9"'""'-- - - - - Copies Rece ;, _ _I ______ Add'l Cocics Re < Special 01 'or -=e.:D&.. 131Jx">1

PRA Policy February 7, 1995 Page 2 for internal and external events. It is clearly appropriate for the NRC to provide some specific policy guidance on how this vast store of information may be used for regulatory purposes. While we have no quarrel with the NRC's proposal, we believe the proposed Policy Statement could be substantially more beneficial to licensees and to the NRC if it more e xplicitly encouraged licensees to make use of their PRA information to support specific licensing actions. In the past, NRC practice has not been clear with respect to reliance on PRA insights for plant-specific licensing actions. More recently, the NRC has increasingly used PRA as a tool in support of specific licensing actions, but some policy guidance in this area would still be useful. For example, the Policy Statement should address whether, and to what extent, licensees may credit PRA information to justify licensing actions, e.g., an exemption or a change in Technical Specifications to expand an Allowed Outage Time. In this connection, we believe the proposed Policy Statement should state explicitly that PRA insights, coupled with other pertinent information, can be used to improve Technical Specification and other requirements by re-examining the risk significance of certain fundamental (and often very conservative) deterministic assumptions and design criteria -- for example, the assumed double-ended guillotine pipe break and some single failure assumptions. Risk-tempered interpretations of such assumptions and criteria could help correct certain distortions in a plant's risk profile that may be introduced by the use of such conservative assumptions. Use of PRAs for Backfittinq Initiatives The proposed Policy Statement correctly identifies that uncertainties exist in both deterministic and probabilistic approaches to regulation.!' Further, it is recognized that the approach to dealing with uncertainties in the deterministic process 1' We recommend that the Policy Statement language regarding uncertainty be revised to clarify that while some uncertainties do reflect limited knowledge of the phenomenon under consideration, other uncertainties reflect, not ignorance, but the inherent variability of pertinent parameters.

PRA Policy February 7, 1995 Page 3 was to build margin into regulat ory requirements by using conservative assumptions. The proposed Policy Statement also acknowl edges that the probabilistic approach permits "an improved framework to better focus and assess [the] significance" of uncertainties in regulatory requirements. 59 Fed. Reg. at 63,390. Comparisons performed by the NRC in NUREG-1150 and elsewhere show the considerable margin that exists today between the early and latent fatality Safety Goals and probabilistically calculated risk levels. Because of the considerable margin that exists, it would no longer seem necessary or appropriate to base new backfitting initiatives on cost-benefit analyses that use worst-case or unnecessarily conservative assumptions. Instead, we recommend that the Policy Statement be amended to state that when backfitting analyses are performed, the mean risk levels (i.e., mean person-rem averted as opposed to worst-case values) be the exclusive basis of regulatory decision-making when comparisons are made against the $1000/person-rem criterion. In addition, the NRC now properly relies upon probabilistic-based analyses to determine whether proposed backfits will produce the requisite "substantial increase" in overall safety under 10 C.F.R. § 50.109 (a) (3). The Policy Statement should explicitly direct the Staff to use the relevant plant-specific PRA in assessing the need for any backfitting action at that individual plant . With respect to proposed generic backfits, the Policy Statement should allow licensees to take appropriate credit for plant-specific PRA information to justify relief from the NRC-imposed action. These provisions would reflect the fact that nuclear plants now have completed their own plant-specific PRAs per the IPE and IPEEE programs, and that the results of those PRAs are ready to be used in applications such as these. These changes would further the purposes of the proposed Policy Statement by helping to reduce or avoid the imposition of unnecessary burdens on licensees. An additional related concern is a revision made in the proposed Policy Statement (in paragraph 4) from the version proposed in SECY-94-218 .Y Paragraph 4 of the proposed Policy Statement discusses the uses of Safety Goals, and states that they "are to be used with appropriate consideration of uncertainties in 1' SECY-94-218, "Proposed Policy Statement on the Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities" (August 18, 1994).

PRA Policy February 7, 1995 Page 4 making regulatory judgments in the context of backfitting new generic requirements on nuclear power plant licensees." The draft in SECY-94-218 would have provided additionally that Safety Goals could be used "in granting relief from unnecessary regulatory requirements." SECY-94-218, Enclosure 1 at 5. This phrase, however, was deleted from the Policy Statement as proposed. It is unclear why this revision was made. It would seem that deleting this provision coul d unduly hamper the effective use of Safety Goals in streamlining the regulatory process - - for example, through the adoption of risk-based improvements to regulations or needed relaxations to overly burdensome requirements. Accordingly, we recommend retention of the language from SECY-94-218 to the effect that Safety Goals could be used in granting relief from unnecessary requirements. NUBARG appreciates the opportunity to comment on this important Policy Statement. We believe that the changes proposed above will help ensure that it accomplishes its intended purposes. ely, as S. Daniel F. Ste Jan A. MacGregor Counsel to the Nuclear Utility Backfitting and Reform Group

February 7, 1995 COMMENTS OF OHIO CITIZENS FOR RESPONSIBLE ENERGY, ON PROPOSED POLICY STATEMENT, "USE OF PROBABILISTIC MENT METHODS IN NUCLE AR REGULAT ORY ACTIVITIES ," 63389 (DECEMBER 8, 1994 ) OCRE believes that PRA methods, if done correctly, can serve as a valuable supplement to the NRC's traditional regulatory philoso-phy. However, it is not capable of replacing these traditional practices. The danger of relying too heavily on PRA is that it can lead to an "accidents don't happen" philosophy in which vital concepts and systems (such as containment or emergency planning) are eliminated. OCRE believes that the uncertainties and limitations of PRA are simply too great to permit PRA to replace traditional concepts such as defense-in-depth. These uncertainties and limitations include: the virtual impossibility of identifying every accident sequence or component failure mode that can occur; difficulty in predicting human behavior; inadequate data on component failure rates; the difficulty in accounting for the effect of construc-tion defects and deviations from specifications; the sensitivity to analytical assumptions and methodologies; and the susceptibil-ity to manipulation and abuse. For example, consider the Watts Bar IPE, which initially showed a core melt frequency of 3.3E-4 per year. The IPE was subse-quently revised to yield a CMF of 8.0E-5/yr. The difference is attributed to "revised success criteria for loss of component cooling water . . . and credit for updated procedures and train-ing." (See SECY-95-004.) A cynic might ask whether the licen-see, after getting an unfavorable CMF at first, went back and applied some "fudge factors" to reduce that number. OCRE is concerned that the proposed policy statement appears to be biased in the direction of using PRA to support deregulation. The proposed policy statement declares: (2} PRA and associated analyses {e.g., sensitivity studies, uncertainty analyses, and importance measures) should be used in regulatory matters, where practical within the bounds of the state-of-the-art, to reduce unnecessary conservatism associated with regulatory requirements, regulatory guides, license commit-ments, and staff practices. Where appropriate, PRA should be 1 FEB 2 4 1995 Acknowledged by card ................................

U.S. NUCL~A , EG..:Lh., OR,' CO~/,MISSION DOC: f rn*r) & C:.E_R ICE SECTiON Offt,F OF- Tdf ECRE1ARY or r f C')'v1\,I SSION Pos ria ** ~~\qS ___ _ Copies R* _ _ l _ _ __ Add 1Cop*::s r -.=> Speci"' D1:,tr,b ~R..\-::R_\.I:6_ _ _ ~~

used to support additional regulatory requirements . . . . [empha-sis added.] Clearly, "appropriate" is a subset of "practical" in that not all things which are practical are appropriate. And, no definition of "appropriate" is given herein. Thus, the implication of the above excerpt is that PRA will principally be used, whenever possible, to reduce regulatory requirements, and maybe, in some undefined circumstances, PRA could be used to augment regulatory requirements. OCRE finds this unacceptable. If the NRC is truly a fair, inde-pendent regulatory body, it must be as willing to increase regulatory requirements as it is to reduce them. OCRE recommends revising the policy statement such that either the word "practi-cal" is used in both instances, or the word "appropriate" is used in both instances. Item (3) of the proposed policy statements states that "appropri-ate supporting data should be publicly available for review." OCRE agrees, and would add the requirement that the NRC make all data, as well as its computer codes and training materials on PRA, available to the public. Item (4) of the proposed policy statement would require the use of the NRC's safety goals and subsidiary numerical objectives in making judgments on backfitting new requirements on nuclear power plant licensees. It is apparent that the NRC intends the safety goals to be the maximum level of nuclear plant safety which the NRC will enforce. OCRE opposes the use of the safety goals in regulatory analyses. Core damage probability alone, or even with the consideration of conditional containment failure probability, should not be used to impede safety improvements. It is not rational to reject a regulatory requirement with an extremely favorable value/impact analysis on the basis of a PRA. Many high consequence accident sequences have low probabilities. For example. consider the sequence of events that led to the April 1986 Chernobyl accident. It would be instructive to inquire whether any PRAs were conduct-ed for the Chernobyl reactor and whether that exact accident sequence was contemplated. Most likely it would have been deemed incredible. But it happened, with devastating consequences. 2

OCRE believes that the use of the Safety Goals is not in the best interests of the nuclear industry. It is essential to consider the cumulative impact of individual plant risk. In the debate over safety goals, one risk appears to be ignored: the political risk to the nuclear option if another serious nuclear accident occurs. The General Accounting Office report "What Can Be Done to Revive the Nuclear Option?" predicts that one more serious accident could doom a second nuclear generation (GAO/RCED-89-67, pp. 5, 37). The accidents at Three Mile Island Unit 2 and Chernobyl have seriously eroded public acceptance of nuclear power. In order to preserve the nuclear option, the NRC and the nuclear industry should establish a minimum safety goal considering the aggregate risk, rather than the per-reactor-year risk, for the entire reactor population over time, of one severe core damage accident, which could imperil the survival of the nuclear industry. Both the current reactor population and the future, hoped-for population should be factored into this minimum safety goal. The probability of at least one core melt accident is given by: Pcm - 1 - e~(-LAMBDAcm

• t) where Pcm is the probability of one core melt over a certain time period; LAMBDAcm is the probability of a core melt per reactor year; and t is the number of reactors operating times the years in opera-tion (i.e., reactor-years).

(From NUREG/CR-2040, "A Study of the Implications of Applying Quantitative Risk Criteria in the Licensing of Nuclear Power Plants in the United States," p. 68). Assuming 109 reactors in operation, operating for 40 years, meeting the safety goal for core melt of lE-4 per reactor year, the probability of a core melt in the next 40 years is 35%. If the current reactor population is approximately doubled, to 200 reactors, the probability of a core melt in 40 years is 55%. If the circa-1970 prediction of 1000 reactors in the U.S. by the year 2000 had come true, the probability of a core melt in 40 years of operation is 98%. 3

For the risk of a core melt accident to be insignificant, Pcm must be a small number, certainly less than lE-2. In establish-ing the safety goal policy, the NRC assumed that lE-3 represents an insignificant risk to most people. Assuming Pcm is equal to lE-3, for 109 reactors operating for 40 years, LAMBDAcm must be less than 2E-7. Smaller values of Pcm would require even stricter safety goals for individual plants. The more reactors we have, and the longer they operate, the safer they must be. Thus, it would be to the benefit of the nuclear industry to seek more stringent safety goals to minimize the risk to the public and to the industry. Enhancing the protection of the public health and safety and preserving the nuclear option are not incompatible. Rather, improving nuclear plant safety is essen-tial to the industry's survival. OCRE believes that any safety goal the NRC establishes should be a minimum goal, rather than a maximum level of safety that plants should not be required to exceed. OCRE believes in a free market approach to nuclear safety. The maximum level of safety should be determined by this free market approach, i.e., what the public wants and is willing to pay for. Consider this automotive analogy. No one needs a luxury car. But if people want luxury cars, and are willing to pay for them, they can buy them in this country. The NRC may consider exceeding the safety goal to be a luxury, but in a free market economy, if the American people want this luxury and are willing to pay for it, they should have it. The NRC, as the regulator, should translate the public's expecta-tions into regulatory requirements. The safety goal is also inappropriate because it encourages complacency and discourages the pursuit of excellence. The NRC cannot credibly exhort the industry's poor performers to improve when it stifles its own staff from seeking safety improvements. The safety goal embodies a philosophy not of excellence but of mediocrity and retaining the status quo because it's "good enough." A free market approach to nuclear safety is consistent with our democratic traditions. Much of the public distrust of the NRC and nuclear industry stems from the fact that people feel isolat-ed from decision-making processes that affect them. Instead of being decided in a democratic fashion, issues such as what is an acceptable risk and how safe is safe enough are decided by bureaucrats in Washington. Most people feel that these are highly subjective and personal issues for which the right to self-determination should be paramount. The public might be more willing to accept nuclear energy if the industry were held to free market public expectations of safety. 4

OCRE recommends that item (4) of the policy statement be revised to establish the safety goals as the minimum acceptable level of plant safety, with the maximum enforceable level of safety to be determined by a free market approach. Respectfully submitted, Susan L. Hiatt Director, OCRE 8275 Munson Road Mentor, OH 44060-2406 (216) 255-3158 5

Department of Energy Washington , DC 20585 DOCKETED US RC FEB O7 1995 *95 FEB - 7 Al :27 Secretary U.S. Nuclear Regulatory Convnission Attention: Chief, Docketing and DOCKET NUMBER p Service Branch PRO OSE -~~~- Washington, D.C. 20555 (50,.fR.\o~~~

Dear Sir:

Enclosed are the comments of the U.S. Department of Energy's Office of Civilian Radioactive Waste Management on the Nuclear Regulatory Commission's Proposed Policy Statement on Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities. Our review produced convnents on several aspects of the Proposed Probabilistic Risk Assessment Policy Statement. Overall, we believe the document accomplishes the desirable goal of encouraging the use of Probabilistic Risk Assessment in regulatory activities. However, we have suggested the scope and discussion in the document be expanded to include aspects of the nuclear fuel cycle that are well suited (or only suited) to the use of Probabilistic Risk Assessment. The policy statement refers only once to storage, transportation, or disposal of nuclear materials, and repository-related issues are likely to be some of the most important issues considered by the Commission in the next ten years. Other convnents suggest that the Commission should clarify the statements that Probabilistic Risk Assessment is best used in support of deterministic analyses. - Should you have any questions, please contact Christian Einberg of my staff at (202) 586-8869. Sincerely,

                                             ~~a.,:;?r~

Ronald A. Milner, Director Office of Program Management and Integration Office of Civilian Radioactive Waste Management Enclosure

                                                                                  ,FEB 2 4 1995 ~

Acm wfedged by card ..............".......... ......

U.S. N CLEAR REG LATORY Ol\il'u>><<..*111 DO KETI G & SERVICE SE OFFICE OF THE SECRET OF THE COM S ON

cc: R. Loux, State of Nevada T. Hickey, NV Legislative Committee J. Meder, NV Legislative Counsel Bureau M. Murphy, Nye County, NV

0. Bechtel, Clark County, NV P. Niedzielski-Eichner, Nye County, NV B. Mettam, Inyo County, NV V. Poe, Mineral County, NV F. Mariani, White Pine County, NV R. Williams, Lander County, NV L. Fiorenzi, Eureka County, NV J. Hoffman, Esmeralda County, NV C. Schank, Churchill County, NV L. Bradshaw, Nye County, NV W. Barnard, NWTRB E. Lowry, NV Indian Environmental Coalition R. Holden, National Congress of American Indians

COMMENTS ON PROPOSED NRC POLICY STATEMENT ON USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES . {59 FR 63389, December 8, 1994)

GENERAL COMMENT

S

1. Co11111ent: The focus of draft PRA Policy Statement should not be limited to power reactors, because repos itory -rel ated issues are some of the most important issues likely to be considered by the NRC in the next ten years.

2. Co11111ent: The draft PRA Policy Statement should be amended to include risk assessment applications other than power reactors, including applications to transportation and storage of HLW materials (e.g., siting).

3. Co11111ent: The policy statement should temper the commitment to PRA to some degree. Without such a qualification, the regulatory requirements on some licensees could increase for regulated facilities with inherently low risks such as dry storage independent spent fuel storage installations or monitored retrievable storage facilities, for which a formal submittal and review of PRAs is not currently required. The policy statement needs to reflect the sense of limitation which is present in the supporting information. This i s true on page 63390, especially last paragraph on page, which states:

      "Given the dissimilarities in the nature and consequences of the use of nuclear materials in reactors, industrial situations, and medical applications, the Commission recognizes that a single approach for incorporating risk analyses into the regulatory process is not appropriate. However, *pRA methods and insights will be broadly applied within the NRC to ensure that the best use is made of available techniques to foster consistency in NRC risk-based decision-making."

4. Co11111ent: The draft policy statement should address the geologic technical uncertainties in projecting repository performance. The NRC should acknowledge that it may not be possible to reduce uncertainties completely. There wil l be residual uncertainties in site data, acquisition and analysis, development and validity of conceptual models and computer codes, and effect of future states on repository performance.

5. Co11111ent: The draft policy statement should acknowledge the use of technical-expert judgment in assessing the effects of residual uncertainties and unquantifiable uncertainties on repository performance and the use of decision-maker judgment in addressing the regulatory significance of residual uncertainties.

SECTION I, BACKGROUND

6. Co11111ent: The background information in the draft policy statement states that "PRA methods ... have proved to be a valuable complement to deterministic approaches." The draft policy statement should describe the reasons for using the PRA to complement the traditional deterministic approaches. (One reason is to handle inherently large uncertainties quantitatively such that results are sufficiently robust to support regulatory decisions. Another reason is to allow the consideration of risk in a coherent manner so as to permit the quantification of overall level of safety for a particular nuclear regulatory activity.}

7. Co11111ent: The third paragraph discusses how PRA has been applied in several regulatory activities. It does not, however, discuss the impact of PRA on the recent proposed rule change to 10 CFR Parts 50, 52, and 100 (59 FR 52255}. These proposed changes are significant considerations for reactor 1*icensing, and discussing them provides further evidence of the NRC's intention to encourage use of PRA.

Reco11111endation: Add the following two sentences to the end of the third paragraph:

      "Probabilistic analyses were extensively used in the development of t he recent proposed rule change to reactor siting criteria in 10 CFR Part 100 (59 FR 52255}. That same proposed rule change would require license applicants to use a probabilistic approach to specify the Safe Shutdown Earthquake Ground Motion for a nuclear reactor site, instead of the deterministic method currently specified in Appendix A to 10 CFR Part 100."

8. Co11111ent: Section I discusses the use of PRA in nuclear regulatory activities, but it does not discuss the fact that PRA has been used in a number of areas besides nuclear power. Adding such a discussion would strengthen the credibility of the PRA .concept.

Reco11111endation: Prior to the third paragraph of section I, add a paragraph that notes the use of PRA in various areas, such as aviation and space, chemicals, and environmental science.

SECTION II DETERMINISTIC AND PROBABILISTIC APPROACHES TO REGULATION (A) Extension and Enhancement of Traditional Regulation

9. Coment: The draft policy statement asserts that the existing "deterministic approach contains imp1 ied elements of probabi1 ity **. " (emphasis supplied). The only examples given relate to nuclear reactors. The draft PRA policy statement should acknowledge that this should be the case for HLW disposal as well. However, some of the performance requirements of 10 CFR part 60 are deterministic in nature and do not contain implied elements of probability.

1

10. Coment: Section II. (A) third paragraph, last sent ence states:

      " ... PRA techniques are most valuable when they serve to focus the traditional, deterministic-based, regulations and support the defense-in-depth philosophy." (run-on sent.)

This sentence is potentially misleading because deterministic methods cannot always be used. Some regulations (such as the expected Env*ironmental Protection Agency regulations pertinent to a proposed high-level waste repository) may be defined in probabilistic terms. Deterministic methods cannot be used to demonstrate compliance with such regulations. In addition, when discussing postulated events that may occur in the very distant future (as is often the case in nuclear waste issues), potential licensees must use performance predictions for hundreds or even thousands of years to demonstrate compliance with regulations that have never been successfully applied to any facility. The spatial and temporal uncertainties associated with these long-term analyses are very different from those uncertainties that are inherent in analyses of complex engineered systems such as nuclear power plants. The Commission should recognize the special nature of its regulations for low-level and high-level waste, and should expand its policy statement to include an endorsement of the use of PRA to deal with technical and regulatory uncertainties in this area. Recomendation: Add the following paragraph following the paragraph on which the com~ent is made:

     "In addition, PRA techniques are appropriately used where a deterministic approach is not feasible, such as when considering regulations defined in probabilistic terms, and for consideration of potential events in the distant future (e.g., postulated events and mitigating factors pertinent to a mined geologic nuclear waste repository). For such events, undergraded function of systems, components, and structures may not be assumed, because sufficient data may not exist to support this type of deterministic conclu~ion.

Instead, the probability of the system functioning at a given level of effectiveness may be determined using available data and assumptions, and this probability may be used in analyzing the probability and consequence of given postulated events. Deterministic analysis is performed to the extent it is valid, and probabilistic analysis is used to both complement the deterministic analysis and to replace the deterministic analysis where the deterministic approach becomes infeasible."

11. Coment: Section I II, i tern ( 1) states that PRA should be used "in a manner that complements the NRC's deterministic approach ... " The same comment applies to this sentence as is provided in the comment above.

Reconmendation: Add the phrase "where possible" following "approach" in the quoted sentence so it reads: "in a manner that is complemented by the NRC's deterministic approach where possible ... " Also, add the following passage as a new item (2), renumbering following items appropriately:

     "(2) The use of PRA should be emphasized for those applications that involve projecting system performance for very long time periods, such as hundreds or thousands of years. The use of PRA should also be emphasized for those applications where extensive experience applying existing regulations using deterministic criteria does not exist."

12. Coment: Section II, third paragraph states:

     "Mitigating system reliability is then assessed, including the potential for common cause failures. The probabilistic treatment therefore goes beyond the single failure requirements used in the deterministic approach."

This statement may be interpreted to imply that PRA addresses multiple failures only in the context of common cause failures. A proper PRA should address multiple failures other than common cause failures. For example, it might consider a mechanical failure of a safety system due to a worn-out part, coupled with a human failure (such as failure to correctly follow procedure) that affects another component or system that would mitigate an event. Recomendation: Modify the first sentence quoted above as follows.

     "Mitigating system reliability is then assessed, including the potential for multiple (including common cause) failures."

13. Co11111ent: Section II (A) third paragraph, first sentence states: " ... PRA addresses all credible initiating events ... " The statement is inaccurate in that no process can be stated to address all credible events. Imperfect human knowledge of possible events and outcomes prevents being sure that all possibilities have been considered. Instead, PRA starts with events and consequences of events that are postulated to be credible and helps define their credibility, or more accurately their estimated probability of occurrence and consequences.

Reco11111endation: Revise the phrase upon which comment is provided to read as follows: " ..* PRA addresses a comprehensive set of credible initiating events .** 11 {B) Uncertainties and Limitations of Deterministic and Probabilistic Approaches

14. Co11111ent: Section II (B) third paragraph discusses power reactor events but not nuclear waste-related events and issues.

Reco11111endation: Add the following sentence following the statement upon which the comment is made: "The same may be said of activities that involve consideration of complex physical phenomena that may occur in the distant future, as previously discussed." {C) Defense-in-Depth Philosophy

15. Co11111ent: The draft policy statement states that the "expanded use of PRA technology will continue to support the NRC's defense-in-depth philosophy ... "

The statement is accurate for nuclear power reactors. However, under the Nuclear Waste Policy Act of 1982, the Commission's technical criteria "shall provide for the use of a system of multiple barriers in the design of the repository .... as the Commission deems appropriate." The Commission in 10 CFR 60, has coupled the traditional defense-in-depth approach (cited in the draft PRA policy statement) with the multiple barrier approach (two engineered barriers [waste package and underground facility] in addition to the natural barrier provided by the geologic setting), and the unitary EPA standard, to provide greater confidence in its licensing judgments. The draft PRA policy statement should be changed to reflect this coupled philosophy for the disposal of high-level waste.

16. Co11111ent: Section II (C) first sentence states:

      "The Commission recognizes that complete reliance for safety cannot be placed on any single element of the design, maintenance, or operation of a nuclear power plant."

The same statement may be made about a high-level waste repository. Reco11111endation: Add the following phrase to the sentence upon which the comment is made: " ... or a nuclear fuel cycle activity such as a radioactive waste repository."

SECTION III, THE COMMISSION POLICY

17. Co11111ent: First sentence, item (2) should be amended to include the "reduction of uncertainties" as one of the reasons for using performance assessment (PA) in HLW disposal. The Commission should provide specific language, applicable to HLW disposal, which permits the use of proba_b ilistic analysis in showing compliance with existing disposal requirements. In addition, the draft Commission Policy should be amended to reflect the use of PA by the NRC in risk-effective safety decision making for HLW disposal.

18. Co11111ent: Section II (A) states that PRA techniques are most valuable when they "support the defense-in-depth philosophy." In Section III, Item (1) states that PRA should be used in a manner that "supports the NRC's traditional defense-in-depth philosophy." Section III, existing Item (2) states that PRA should be used "to reduce unnecessary conservatism associated with current regulatory requirements ... " The first two statements appear to contradict the third. It is not desirable to support defense-in-depth if, as the third statement implies, defense-in-depth is overly conservative.

Reco11111endation: In Section II, end the referenced sentence after the word "regulations." Add the following sentence following the referenced sentence: 11 PRA should be used in a manner that supports the NRC's traditional defense-in-depth concept and NRC's multiple-barrier concept, though it may also be used to reduce unnecessary conservatism these concepts may have introduced." In Section III, end the referenced sentence in Item (1) after the word "approach." Then add the following:

      "PRA should be used in a manner that supports the NRC's traditional defense-in-depth concept and NRC's multiple-barrier concept, though it may also be used as discussed in Item (2) (Item (3), if Comment 11 is incorporated) below to reduce unnecessary conservatism these concepts may have introduced."

19. Co11111ent: The importance of the development of hardware failure rates and human performance data should be emphasized in the policy statement. In Part III Policy, data is de-emphasized by the phrase: " ... and to the extent it can be supplemented by state-of-the-art in terms of methods and data."

Reco11111endation: This should be modified to read: 11 The PRA methodology development shall be accompanied by the appropriate development of data. 11

20. C011111ent: The Policy Implications Section, first paragraph, second sentence states:

"First, the NRC staff, licensees, and ... " The sentence needs to include license applicants as well as licensees. Reco11111endation: After "licensees," add: "and license applicants."

PLG, Inc., 4590 Ma-:Arthur Boulevard, Suite 400, Newport Beach , California 92660-2027 OOCl'F t- l.J Tel. 714 -833-2020

• Fax 714 -833-2085 U. !RC PLG, Inc., Bethesda, MD, Office Tel. 301-907-9100
• Fax 301 -907-0050 ENGINEERS
• APPLIED SCIENTISTS*

MANAGEMENT CONSULTANTS *95 FEB -6 P 2 *02 PLG, Inc., Albuquerque, NM, Office Tel. 505-881 -1424

• Fax 505-880-0727 OF I DO CK,- I DOCKET NUMBER p PROPOSED RULE..::...:.:._:..;...~ -

( 5C\ Fe Coeo~9) February 4, 1995 @ Secretary U.S. Nuclear Regulatory Commission Attention: Docketing and Service Branch Washington, DC 20555

Dear Sir:

PLG COMMENTS ON NRC PROPOSED POLICY STATEMENT: USE OF PROBABILISTIC RISK ASSESSMENT METHODS IN NUCLEAR REGULATORY ACTIVITIES PLG offers the following comments in response to the recent draft policy statement, "Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities; Proposed Policy Statement," published in the Federal Register on December 8, 1994.

GENERAL COMMENT

S ON CORE OF POLICY STATEMENT With reference to the core of the policy statement in Section III, PLG would like to make the following general comments: We agree with the basic tenets of the policy to (1) increase the use of PRA technology in all regulatory matters; (2) utilize PRA and associated analyses in regulatory matters, so as to reduce unnecessary conservatism; (3) incorporate PRA evaluations that are as realistic as possible; and (4) consider the Commission's safety goals and subsidiary numerical objectives in the backfitting of new requirements. We also acknowledge the significant milestone marked by the issuance of this policy statement on the path towards increased importance of PRA technology in the regulatory process. At the same time, the policy statement reveals some points of confusion and controversy that we believe must be sorted out if the full benefits of PRA technology are to be realized. These have principally to do with some confusion on the difference between deterministic and probabilistic analysis and the mixing up of analysis and decision criteria. ifEB 2 4 1995 Acknowledged by card .......................... *....

U.S. NUCLEAR REGUU1 TORY COMMISSIOt~ DOCKETINu & SERVICE SECTION OFFICE OF THE SECRETARY OF THE COMMISSION PostMark Da O

                ~ -Cl,:no9-,. ~ W' ,j Coples Recr.11,, d_ L Adcfl CopiJs ne*      , _... .e,_ _ _ __

Spec}al Disrr;but!J -3>J>..e...,-~-\PE,--... 1

Secretary February 4, 1995 U.S. Nuclear Regulatory Commission Page 2 Deterministic versus Probabilistic Analyses There are frequent references in the policy statement to the effect that deterministic and probabilistic analyses are separate, independent, and, at best, complementary methods of analysis. A more logical perspective is that probabilistic analysis is simply an extension of deterministic analysis. The nature of the extension is a higher state of knowledge than had we relied only on deterministic analysis. In particular, the probabilistic approach answers the "additional" question of how much confidence, realism, and conservatism we have in our deterministic calculations. Adoption of the PRA approach simply adds a new dimension to our safety analyses. They are clearly not separate and distinctive concepts and the policy statement discussion would be greatly improved if this distinction was eliminated. The difference between the methods comes in the technical questions being asked. Deterministic analyses answer the questions: "What can go wrong?" and "What are the consequences?" The former question is answered deterministically by specifying an essentially generic list of design basis accidents and the latter by the performance of the design basis accident analyses. The most important distinction between the two approaches is with respect to the third fundamental safety question (i.e., "What is the likelihood?"*) since it is only addressed implicitly in one approach but quantified in the other. Almost as important a distinction is the different approaches taken to address uncertainties: a set of point estimate calculations based on a set of "conservative" assumptions on the one hand, and a probabilistic analysis of uncertainties centered around a set of "realistic" assumptions on the other. PRA and Decision Criteria - We submit that there is no such thing as "deterministic criteria." The current decision criteria are the collective set of rules and regulations that are based on the safety philosophy of "defense in depth" and, prior to PRA and the associated severe accident research stimulated by the TMI-2 accident, made use of a nuclear safety knowledge base that was driven by the results of deterministic analyses. The principal question before us is how the additional information provided by PRA technology (PRA does nothing more than add information to our knowledge base) should be used to make more informed regulatory decisions, to reduce unnecessary regulatory burden, and to bring about a better allocation of resources. One of PRA's most important contributions is that it has clarified the questions that need answers. For example, the current regulatory framework has been somewhat preoccupied with "design basis" questions rather than the more fundamental "what can go wrong" question.

• Kaplan, S. , and B. J. Garrick, "On the Quantitative Definition of Risk," Risk Analysis, Vol. 1, No. I, March 1981.

Secretary February 4, 1995 U.S. Nuclear Regulatory Commission Page 3 Deterministic safety analyses have been confined to a preselected and largely generic set of design basis accidents that were believed to be too "limiting" so that most of the resources in the deterministic arena have been devoted to the "what are the consequences" question. A premise of the deterministic approach that the TMI-2 accident proved wrong is that the application of the defense-in-depth philosophy and use of conservative assumptions in the design basis accident analyses would preclude the need to consider the risks of accidents more severe than the design basis. If on the other hand, both approaches started from the same premise (i.e., that the risks associated with a nuclear plant are plant specific), meaning that answers to all three of the fundamental safety questions might be plant specific, then the concept that deterministic analysis is a subset of probabilistic analysis would be obvious. Neither the policy statement nor the implementation plan provides consistent decision criteria for accepting PRA results as part of the justification for licensing decisions. While there is an implicit offer to consider pilot applications of risk-based arguments to help justify licensing changes or relief, there is no commitment apparent in the policy statement to systematically review and modify the current regulations. Thus, the transition process is obscured. As a minimum, the short-term effects of this policy statement are likely to be an increased burden on the licensees, contrary to the stated intentions of the policy statement. The NRC should consider adopting performance measures that if achieved could, in fact, lead to reduced regulatory burden. In the long term, there needs to be a top-down review of the current rules and regulations to identify opportunities for the elimination of unnecessary regulations likely to result from the combination of deterministic- and probabilistic-based requirements currently being imposed. An important beginning, in this regard, would be to agree on the basic questions for which we want answers, whether they be developed deterministically or probabilistically. SPECIFIC COMMENTS ON TEXT OF POLICY STATEMENT In addition to the above general comments, we offer the following specific comments that refer to particular parts of the text of the policy statement: The Role and Treatment of Uncertainties We are pleased to see the acknowledgement in Section II(b) that uncertainties are fundamental to our knowledge base and not simply introduced by the PRA approach, in contrast with previous staff positions on the use of PRA. Both deterministic methods and PRA must deal with uncertainties and, as evidenced by the policy statement, some deterministic analyses that have been implied only have served to mask the role of these uncertainties. One of the principal advantages of PRA when it is done right is that it tends to expose sources of uncertainty and to provide a means of dealing with the application of realism and conservatism in a quantitative manner.

Secretary February 4, 1995 U.S. Nuclear Regulatory Commission Page 4 Beyond this observation, the proposed policy could be improved in its discussion and consideration of uncertainties and associated limitations of the analysis techniques. It appears to take the viewpoint that uncertainty analyses (along with sensitivity analyses) are somehow PRA specific. Uncertainty must be dealt with in both deterministic and probabilistic analysis. Specifically, problems in understanding human error rates is a determinant to the validity of all regulations, no matter how they were developed. Analyses that will assist in understanding and minimizing these errors will be beneficial. The fact that PRA methods are not fully developed does not make deterministic methods valid. Need to Revisit Defense in Depth Section Il(c) discusses the continuing need for the defense-in-depth philosophy while the applications of PRA are expanded. At one level, there is a basic consistency between these safety philosophies in the sense that plants that were designed and licensed with defense-in-depth were later found, from application of PRA, to result in acceptably low risks from severe accidents. At a more detailed level, there are insights from PRAs that suggest that the application of the defense-in-depth concept may, in certain cases, lead to a poor allocation of resources and thereby defeat the intent of the policy statement. For example, in considering the respective roles of the plant protective systems, reactor vessel, and containment in managing the risks of pressurized thermal shock scenarios, we know from probabilistic analyses of these scenarios that the degree of protection provided by each of these safeguards cannot be independently specified for all sequences. Thus, it may not be necessary, or in fact feasible, to ensure that same level of confidence in reactor vessel or containment integrity for any and all pressurized thermal shock challenges. Alternatively, a balanced degree of protection from each safeguard may only be possible for certain relatively high frequency sequences. An important lesson from PRAs, cited later in the policy statement, is that risks are largely dictated by dependent and common cause failures which, on specific sequences, may tend to defeat the idealistic concept of defense-in-depth, even though the risk levels may be acceptably low for such sequences. What this comment suggests is that defense-in-depth has to be generalized and reformulated for sensible application within a PRA framework to avoid overly conservative treatment of certain issues and unnecessary regulatory burden. The Commission Policy Section III, the end of paragraph (2) of the Commission Policy states:

       "It is, of course, understood that the intent of this policy is that existing rules and regulations shall be complied with unless these rules and regulations are revised."

Secretary February 4, 1995 U.S. Nuclear Regulatory Commission Page 5 The policy statement implies that the burden of proof for any reduction in regulations rests with the industry. An important lesson from previous PRAs is that generic rules do not always have the intended safety impacts on specific plants. Generic plants do not fail, specific plants may. We encourage the NRC to review the wisdom of regulating the safety of specific and unique plants through generic rules, unless it is understood in advance that such rules are indeed generically applicable. This should be an integral part of the top-down review recommended earlier. We are also concerned that the NRC staff may interpret the above statement to put the burden on licensees to demonstrate applicability of the PRA argument beyond a reasonable doubt without a comparable investigation into the basis for retaining current requirements. For example, can we currently justify the basis for the generic technical specifications? Simply classifying a rule or regulation as "deterministic" does not necessarily establish its basis vis-a-vis a PRA approach. We recommend that the following statement be added to this paragraph:

       "However, the objective is to produce a set of rational regulatory requirements whose bases can be understood in terms of an appropriate combination of deterministic engineering and probabilistic analyses. This entails responsibility on the part of licensees to seek changes where they believe them to be warranted and the staff to consider changes in the regulations if strongly supported by probabilistic arguments."

PLG hopes that the above comments are taken in the same constructive spirit in which they were made, and appreciates the opportunity to provide these public comments. Very truly yours, B. John Garrick, Ph.D. President

DOCl'TTED

                                                 ~
                                                  ~0
                                                    .,~,..,,,,,.,_    ._, , . ,
                                                              . "' As>**                     *cs    FEB -1         p ? ,c-Central Interstate Low-Level Radioactive                                                  ~  *..J 1

Waste Commission OFFi ,:,L c *-, *::: 1 r Y DOCKl 1 ,,u _ :~ h CC H. A. Caves t /f\.: Eugene Crump Chairman Executive Director & General Counsel January 26, 1995 DOCKET NUMBER Pl PROPOSED RULE...........~ = - Secretary, U.S. Nuclear Regulatory Commission ( ER f' R...lo~:f~'{) Washington, DC 20555 - Attention: Docketing and Service Branch On behalf of the Central Interstate Compact Commission, I request that you extend the comment period on the Proposed Policy Statement on Use of Probablilistic Risk Assessment Methods by ninety days. Our office just became aware of this proposed policy statement today, and feel that we cannot make a thorough assessment in the short time available. This additional time would also allow us to obtain the background documents. While this policy may be very appropriate in some instances, particularly those in specific power plant applications, a sweeping policy change covering all aspects including waste disposal may be premature, as some applications may not be desirable. These specific applications related to an operating power plant can frequently be handled internally between the facility and the - NRC. Siting and licensing a waste disposal facility is a different situation. An immediate concern is the difficulty of presenting probabilistic statistical information at a public meeting. Deterministic performance assessments are more easily explained and understood. Environmental data is almost always log-normal, and outliers are inevitable. Project opponents (and, in some cases, regulators) will focus on the outliers and have difficulty in understanding the method, especially if they are using NUREG 1200 for guidance and expect the performance assessment to be conservative and bounding. Consistency is used as a justification for developing this policy, but consistency should not be substituted for sound technical judgement and application. Industrial and medical uses are exempted because they are "simple systems". Low-level waste is also a simple system. 233 South 13th Street

• Suite 1200
• Lincoln, Nebraska 68508 * (402) 476-8247 Arkansas
• Kansas
• Louisiana
• Nebraska
• Oklahoma

U.S. NUCLEAR REGULATORY COMMISSIOl't DOCKETING & SERVICE SECTION OFFICE OF THE SECRETARY OF THE COMMISSION Document Statistics Postmartc Date -4.__..~ .....__ _ __ Copies Recoi~d_.....__ _ _ _ __ Add'! Copies Reproduced --"'-"' - - - - - Speci_al Distribution '.?l>R...1 e\'.~*6>

The major concern in regulation of nuclear facililties is health and safety. The present approach attains that goal and it is difficult to see how this policy will enhance safety or mesh with present guidance relative to llw disposal. Sincerely, Reta Roe Technical Director cc H.A. Caves

Station Support Department PECO ENERGY PECO Energy Company Nuclear Group Headquarters 965 Chesterbrook Boulevard Wayne, PA 19087-5691 January 20, 1995 Chief, Rules Review and Directives Branch Division of Freedom of Information and Publication Services Office of Administration U. S. Nuclear Regulatory Commission Washington, DC 20555-0001

Subject:

PECO Energy Company jT - Comments on NRC PRA Implementation Plan C-

c:,

z

Dear Sir:

                                                                                      \..,..)

0 PECO Energy supports the development and implementation of an integrated plan that .,._ '

• ci incorporates the increased use of Probabilistic Risk Assessment (PRA) in regulatory matters. We consider that PRA as a tool can focus the limited resources of both the NRC and utilities-toward those issues or existing requirements that are significant in terms of risk to the public. The NRC also recognizes that PRA can augment and perhaps reduce the conservatisms associated with the deterministic approach but that PRA insights blended with the current deterministic philosophy will provide the basis for many of the changes utilizing the PRA tool. We agree that a combination of deterministic and probabilistic approaches is important in the regulatory arena because the focus should not allow decisions to be made solely on PRA numbers.

We consider the Implementation Plan schedule outlined in SECY-94-219, "Proposed Agency-Wide Implementation Plan for Probabilistic Risk Assessment (PRA)," to be extremely protracted, and recommend that the schedule be adjusted to reflect the need to utilize PRA and PRA insights in the near term. Priority and urgency need to be stressed and reflected in the Plan in order to assure the coordinated use of PRA in regulatory matters in all facets of NRC staff interaction with utilities. We recognize that implementation can be difficult to achieve; therefore, we support continued NEI dialog and applications utilizing PRA with the NRC management to assure alignment with the proposed PRA Policy Statement. In addition, acceptance of PRA-based applications by the NRC should not be contingent on the Plan implementation within the NRC or the schedule. We also recommend that the NRC staff that interacts directly with utilities be trained in PRA aspects as soon as possible to assure the acceptance of PRA is not hindered by a lack of understanding of the tool or its application. The desire to collect detailed data related to equipment and human reliability should not prohibit the use of PRA for applications or support for decision making. The collection of plant specific data must be commensurate with the benefit that specific information might have on the quality or Insight from the PRA. Plant specific information may not be statistically significant. It may also be of limited incremental benefit in light of sensitivity analyses that can be performed utilizing existing models to demonstrate impact of data uncertainty without collecting actual plant-specific data.

                                                                                                      ,FEB 2 4 1995 Acknowledged by card.......................... .....~

u.~. i *ru:-1 R REGULATORY COMtvilSSION C , <. r , G & SERVICE SECTION l r- Tl-<[ SEGRETARY

                    -  C.'.")'Al !SS'O N Po"tmarl* c ~~-(J .\1..i=o til) rn Copies Recs .~ ,~

Add'l Cop*e 8e: Speci I Distr"bu!io:1 'J?J)e,. 1 Q \J>$f

January 20, 1995 Page2 In addition, Implementation of the Maintenance Rule will require monitoring and collection of specific system/plant Information. Since the Rule Is performance based and plant-specific, data collection and general site-maintained performance Indicators will be focused on those systems or attributes that are significant to that particular plant. Requiring all plants to collect the same Information regardless of the focus utilizing a performance basis Is counter to the concept behind the Maintenance Rule. The variability In PAA results may be due In part to subtle design differences between similar plants and plant specific assumptions. The Impact of the differences can be assessed by sensitivity analyses. Sensitivity analyses can be extremely useful In determining the variability of a given result with regard to data, modeling, and assumptions thereby serving to demonstrate a level of confidence of the overall conclusion or result. Uncertainties should not prevent or delay the Implementation of PAA into the regulatory framework. These uncertainties exist in the current regulatory approach and are not necessarily reduced by the existence or use of PAA. The use of PAA as a tool should not be hampered by the knowledge that uncertainty might exist. Decisions can be made within the framework of uncertainty. If you have any questions, please do not hesitate to contact us. Very truly yours, /;/!~ Director - Licensing

UNIVE RSITY O F CALIFORNIA, LOS ANGE LE§ QCKETE0 UCLA USHRC BERKELEY

• DAVIS
• IRVINE
• January 12, 1995 LOS ANGELES
• RIVERSIDE
• SAN DIEGO
• SAN FRANCISCO

                                                              *95 JAN 20 P3
                                                                                       ~
                                                                                 ~=--c. _,.,

D SANTA BARBARA

• SANTA CRUZ UCLA SCHOOL OF MEDICINE HARBOR - UCLA MEDICAL CENTER DEPARTMENT OF RADIOLOGY 1000 CARSON STREET TORRANCE, CALIFORNIA 90509 Secretary, us NRC DOCKET NUMBERpR --

Washington, DC 20555 PROPOSED RULE ~~ (.SC\ "Q....\o '=>3~°' Attention: Docketing and Service Branch CD Re: FR 59(235):63389-63391, 8 Dec. 94: Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities; Proposed Policy statement; also, PRA Implementation Plan, 12 Aug. 94.

Dear Mr. Secretary:

I wish to make a general comment about probabilistic risk assessment (PRA) and then some specific comments about PRA applied to Nuclear Medicine. First, a general comment. When NRC applies PRA to radiation risk (as opposed to risk from fire, explosion, chemical toxicity, or crush injury), NRC will have to make a decision about radiation-induced cancer and mutation risk. I would like to see NRC abandon the l inear hypothesis, which seems to be undergoing significant (and rightful) buffeting lately, and move to an "effective threshold dose" in estimating risk. I have recently sent materia l to the Commission and the EDO's office on progress made this past year in identifying DNA repair enzymes in humans. This new understanding requires a shift away from the linear hypothesis toward an effective threshold model for chemical as well as radiation hazards. This change, more than any other single analytical tool, should revolutionize the manner in which radiation regulation is approached in this country. Second, some comments about PRA and Nuclear Medicine. In NRC's PRA Implementation Plan, p. 7, Section IV.B, last sentence, NRC states, "In regard to nuclear medicine application, NRC contractors have recently completed the preliminary development of a relative risk-rank approach for analyzing nuclear medical devices." Again on p. 8, Section V.B., last sentence, NRC states, "Risk assessment capability (including specific training) to deal with emerging issues in rising risk analysis to analyze the use of nuclear medical devices will be augmented as required by the demands of the developing methodology". FEB 2 4 1995 Acknowledged by card ........................... "'"'~

u.S. '1i.!~ L.~AR RECULA10RY COMMISS, J ooc*<Ern.G & SERVICE SECTION OFFICE OF THE SECRETARY OF THE COMMIS" 1()N

                ~     rrent Statistics Postmark Dat(.,, ~ + - ' -.....~ - - - - - -

Copies Recei-...,__ i _ _.__ _ _ _ __ Adcrl Copies Rep nduced -""""' - - - - - - Speci I Distribution_?_,_,,.=1---""-'-"'~- -

January 12, 1995 secretary, us NRC Page As a board certified Nuclear Medicine physiclan with 35 years of experience in Nuclear Medicine and related areas, the Vice-President of the Society of Nuclear Medicine, President of the California Chapter of the American College of Nuclear Physicians, a two-term Advisor and Consultant to NRC, and present Chair of FDA' s Radiologic Devices Advisory Panel, let me assure you that THERE ARE NO NUCLEAR MEDICINE DEVICES THAT HAVE RISK TO BE ANALYZED . Is NRC spending our user fees to study non-existing technology again like the recent nonsensical effort to write regulations for a non-existent Nuclear Medicine technique which NRC named " Fractionated Radiopharmaceutical Therapy"? Is this "contractor" another bunch of underfunded DOE desperadoes who will churn out anything NRC wants in exchange for our user fees? We use cameras, probes, survey meters, ion chambers, computers, gurneys, hoods, and other "devices" but they haven't killed anyone so far as I am aware. Could it be that NRC does not even know what Nuclear Medicine is? For example, it has nothing whatever to do with Radiation Oncology, a completely separate medical specialty that does use radiation-producing devices. These two medical specialties need to be regarded as completely separate entities. The habit of unknowledgeable NRC employees and management of lumping the two specialties together because they are both "medicine" is like lumping aircraft carriers together with nuclear submarines because they are both "boats". I object to NRC wantonly wasting nuclear medicine user fees to "study" non-existent risks, by PRA or any other method. As NRC apparently does not understand anything about "radiation risk" in nuclear medicine, let me explain it. In terms of years of life lost, which is the preferred, sophisticated way of describing risk, the number of years of life lost from radiation accidents in nuclear medicine, after 59 years and approximately g quarter of a billion radiopharmaceutica l administrati ons , is probably less than one. I am thinking of two events which occurred many years ago, one involving a mistake by a factor of 1000 involving Au-198 colloid, and another involving the accidental substitution of sodium phosphate P-32 for chromic phosphate P-32 in a patient with only a matter of weeks left to live. The first patient died of cirrhosis, but I recall a serious underlying liver disease and the Au-198 would have been a contributory, but not the only, factor involved. The second patient had a normal complete blood count 10 days later and it is probably that much of the P-32 became loculated in pockets of malignant ascites. The patient appears to have died as expected of his underlying disease. All the NaI-131 events that NRC crows about have not been associated with any years of life lost.

January 12, 1995 Secretary, us NRC Page Patients at worst simply take synthetic tetraiodothyronine, and, with the exception of certain lawyers, are normal. This synthetic thyroid hormone is the 6th or 7th most commonly prescribed prescription drug in the United States, with millions of Americans taking it daily. As the average diagnostic nuclear medicine mistake results in 440 mrem ede, an absorbed dose in between yearly background in Washington, D.C., and Denver, co, no "risks" occur, which would be obvious if NRC finally got honest and stopped using the century old fantasy of a linear hypothesis. The one sr-89 chloride error I know of will almost certainly cause no years of life lost in the patient. The biggest risk to patients in Nuclear Medicine is being deprived of services because of NRC's preposterous regulatory requirements and ridiculously high user fees. I therefore suggest that whatever risk assessment tool NRC wishes to use, Nuclear Medicine should be completely deregulated except for making sure that those who are "authorized users" have valid qualifications in basic nuclear and radiation sciences and experience handling radioactive material. NRC regulation of Nuclear Medicine is doing far more harm than good. I do not propose to speak for Radiation Oncology---just Nuclear Medicine. Any effort on NRC's part to extort user fees from Nuclear Medicine to study its "risks" constitutes purposeful fraud on the part of the Agency at this point in time. Thank you for your attention and consideration. Sincerely, ~ Carols. Marcus, Ph.D., M.D. Director, Nuclear Med. Outpt. Clinic and Professor of Radiological Sciences UCLA cc: Chairman Ivan Selin, Ph.D. Commissioner E. Gail de Planque, Ph.D. Hugh Thompson David Nichols Sheldon Trubatch, Ph.D., JD. CSM:sfd aec:112.cam

DOCKET NUMBER -'"~ PROPOSED RULE Pl ~ ~c_ (S~ fR. lc ~~~C\J NUCLEAR REGULATORY COMMISSION Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities; Proposed Policy Statement AGENCY: Nuclear Regulatory Commission. ACTION : Proposed policy statement.

SUMMARY

The Nuclear Regulatory Commission {NRC) is proposing a policy statement regarding the use of probabilistic risk assessment {PRA) in nuclear regulatory matters. The Commission believes that an overall policy on the use of PRA in nuclear regulatory -activities should be established so that the many potential applications of PRA technology can be implemented in a consistent and predictable manner that promotes regulatory stability and efficiency and enhances safety. The proposed policy statement would improve the regulatory process through improved risk-effective safety decision-making, through more

- efficient use of agency resources, and through a reduction in unnecessary burdens on licensees. The use of PRA technology woul d be in~reased in all regul atory matters to the extent supported by the state-of-the-art in PRA J methods and data and in a manner that complements the NRC's deterministic approach and supports the NRC's traditional defense-in-depth philosophy. DATES: Submit comments by February 7, 1995. Comments received after this date will be considered if it is practical to do so, but the Commission is able only to ensure consideration for comments received on or before this date. ADDRESSES: Send comments to: Secretary, U.S. Nuclear Regulatory Commission, ~ l

                                                                                 --C..

Washington, DC 20555, Attention: Docketing and Service Branch. Deliver co1T111ents to: One White Flint North, 11555 Rockville Pike, Rockville, Maryland 20852, between 7:45 am and 4:15 pm Federal workdays. Copies of comments received may be examined at: NRC Public Document Room, 2120 L Street NW. {Lower Level), Washington, DC. FOR FURTHER INFORMATION CONTACT: Thomas G. Hiltz, Office of Nuclear Reactor Regulation, U.S. Nuclear Regulatory Co1t111ission, Washington, DC 20555. Telephone {301) 504-1105. SUPPLEMENTARY INFORMATION: I. Background II. Deterministic and Probabilistic Approaches to Regulation III. The Commission Policy IV. Availability of Documents I. Background The NRC has generally regulated the use of nuclear matenial based on deterministic approaches.

                             .Deterministic approaches\to regulation consider' a set of challenges to safety and determine how those challenges should be defended. A probabilistic approach to regulation enhances and extends this traditional, deterministic approach, by 1) allowing consideration of a broader set of potential challenges to safety, 2) providing a logical means for prioritizing these challenges based on risk significance, and 3) allowing consideration of a broader set of resources to defend against these challenges.

Until the accident at Three Mile Island (TMI) in 1979, the Atomic Energy Convnission (now the NRC), only used probabilistic criteria in certain specialized areas of licensing reviews. For example, human-made hazards(e.g., nearby hazardous materials and aircraft) and natural hazards (e.g., tornadoes, floods, and earthquakes) were typically addressed in terms of probabilistic arguments and initiating frequencies to assess site suitability. The Standard Review Plan (NUREG-0800) for licensing reactors and some of the Regulatory Guides supporting NUREG-0800 provided review and evaluation guidance with respect to these probabilistic considerations. The THI accident substantially changed the character of the analysis of severe accidents worldwide. It led to a substantial research program on severe accident phenomenology. In addition, both major investigations of the accident (the Kemeny and Rogovin studies) recorrmended that PRA techniques be used more widely to augment the traditional nonprobabilistic methods of analyzing nuclear plant safety. In 1984, the NRC completed a study (NUREG-1050) that addressed the state-of-the-art in risk analysis techniques. In early 1991, the NRC published NUREG-1150, "Severe Accident Risks: An Assessment for Five U.S. Nuclear Power Plants.M In NUREG-11~0, the NRC used l improved PRA techniques to assess the risk associat~d with five nuclear power plants. This study was a significant turning point in the use of risk-based concepts in the regulatory process and enabled the Commission to greatly improve its methods for assessing containment performance after core damage and accident progression. The methods developed for and results from these studies provided a valuable foundation in quantitative risk techniques. PRA methods have been applied successfully in several regulatory activities and have proved to be a valuable complement to deterministic

engineering approaches. This application of PRA represents an extension and enhancement of traditional regulation rather than a separate and different technology. Several recent Co11111iss1on policies or regulations have been based, in part, on PRA methods and insights. These include the Backfit Rule (§50.109, "Backfitting"), the Policy Statement on "Safety Goals for the Operation of Nuclear Power Plants," (51 FR 30028), the Commission's "Policy Statement on Severe Reactor Accidents Regarding Future Designs and Existing Plants 0 (50 FR 32138), and the Co11111ission's "Final Policy Statement on Technical Specifications Improvement for Nuclear Power Reactors" (58 FR 39132). PRA methods also were used effectively during the anticipated transient without scram (ATWS) and station blackout (SBO) rulemaking, and supported the generic issue prioritization and resolution process. Additional benefits have been found 1n the use of risk-based.inspection guides to focus NRC inspector efforts and make more efficient use of NRC inspection resources. Currently, the NRC is using PRA techniques to assess the safety importance of operating reactor events and is using these techniques as an integral part of the design certification review process for advanced reactor designs. In addition, the Individual Plant Examination (IPE)iprogram and the Individual Plant Examination - External Events {IPEE() program (an effort

                           '                                                  I resulting from the implementation of the Convnission's "Policy Statement on Severe Reactor Accidents Regarding Future Designs and Existing Plants 11 ) have resulted in commercial reactor licensees using risk-assessment methods to identify any vulnerabilities needing attention.

The Commission has been developing performance assessment methods for low-level and high-level waste since the mid-1970s and these activities intensified using performance assessments techniques in the late 1980s and

early 1990s. This has involved the development of conceptual models and computer codes to model the disposal of waste. Because waste-disposal systems are passive, certain analysis methods used for active systems in PRA studies for power reactors had to be adapted to provide scenario analysis for the performance assessment of the geologic repository at Yucca Mountain, Nevada. In regard to high-level waste, the NRC staff participates in a variety of international activities (e.g., the Performance Assessment Advisory Group of the Organization for Economic Cooperation and Development, Nuclear Energy Agency) to ensure that consistent performance assessment methods are used to the degree appropriate. The Commission believes that an overall policy on the use of PRA in nuclear regulatory activitie~ should be established so that the many potential applications of PRA methodology can be implementea in a consistent and predictable manner that promotes regulatory stability and efficiency and enhances safety. On August 18, 1994, the NRC staff proposed a PRA policy statement to the Commission in SECY-94-218, MProposed Policy Statement on the Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities." In its Staff Requirements Memorandum of October\ 4, 1994, the Commission directed the staff to revise the proposed PRA policy statement and

                          '                                                I publish the proposed PRA policy statement for public convnent in the Federal Register.

II. Deterministic and Probabilistic Approaches to Regulation (A) Extension and Enhancement of Traditional Regulation

The NRC established its regulatory requirements to ensure that a facility is designed, constructed, and licensed to operate without undue risk to the health and safety of the public. These requirements are largely based on deterministic engineering criteria. Simply stated, this deterministic approach establishes requirements for engineering margin and for quality assurance in design, manufacture and construction~ In addition, it assumes that adverse conditions can exist (e.g., equipment failures and human errors) and establishes a specific set of design basis events. It then requires that the licensed facility design include safety systems capable of preventing and/or mitigating the consequences of those design basis events to protect the public health and safety. The deterministic approach contains implied elements of probability (qualitative risk considerations), from the selection of accidents to be analyzed (e.g,, reactor vessel rupture is considered too improbable to be included) to the system level requirements for emergency core cooling (e.g., safety train redundancy and protection against single failure). -- In contrast to the deterministic approach, PRA addresses al1 credible initiating events by assessing the event frequency. Mitigati;f1g system reliability is then assessed, including the potential for common cause failures. The probabilistic treatment therefore goes beyond the single failure requirements used in the deterministic approach. The probabilistic approach to regulation is, therefore, considered an extension and enhancement of traditional regulation by considering risk in a more coherent and complete manner. A natural result of the increased use of PRA methods and techniques would be the focusing of regulations on those items most important to safety by eliminating unnecessary conservatism. Where appropriate, PRA can also be

used to support additional regulatory requirements. Deterministic-based regulations have been successful in protecting the public health and safety and PRA techniques are most valuable when they serve to focus the traditional, deterministic-based, regulations and support the defense-in-depth philosophy. Beyond its deterministic criteria, the NRC has formulated guidance, as in the safety goal policy statement, that utilizes quantitative, probabilistic risk objectives. The safety goal policy statement establishes these top-level objectives to help assure safe operation of nuclear power plants. The safety goals are intended to be generically applied by the NRC as opposed to plant-specific applications. For the purpose of implementation of the safety goals, subsidiary numerical objectives on core damage frequency and containment performance have been establi-shed. The safety goals provide guidance on where plant risk is considered to be sufficiently low such that further regulatory action is not necessary. Also, as noted above, the Cormnission has been using PRA in performing regulatory analysis for backfit of cost-beneficial safety improvements at operating reactors {as required by 10 CFR 50.109) for a number of years. I (B) Uncertainties and Umi,tations of Deterministic\ and Probabilistic Approaches The treatment of uncertainties is an important issue for regulatory decisions. Uncertainties exist in any regulatory.approach and these uncertainties are derived from knowledge limitations. These uncertainties and limitations existed during the development of deterministic regulations and attempts were made to acconvnodate these limitations by imposing prescriptive,

                                      -S-and what was hoped to be, conservative regulatory requirements. A probabilistic approach has exposed some of these limitations and provided an improved framework to better focus and assess their significance and assist in developing a strategy to acconnnodate them in the regulatory process.

Human performance is an important consideration in both detennin1st1c and probabilistic approaches. Assessing the influence of errors of commission and organizational and management issues on human reliability is an example that illustrates where current PRA methods are not fully developed. While this lack of knowledge contributes'to the uncertainty in estimated risks, the PRA framework offers a powerful tool for logically and systematically evaluating the sensitivity and importance to risk of these uncertainties. PRA techniques and models to address errors -0f commission and the influence of organizational factors on human reliability are currently being developed. It is important to note that not all of the Commission's regulatory activities lend themselves to a risk analysis approach that utilizes the same PRA tools (e.g., fault tree methods}. In general, fault tree methods can be more suitable for power reactor events that typically involve complex systems. Events associated with industrial and medical uses of nuclea~ materials generally involve simple systems, involve radiation\overexposures, and result from human error, not equipment failure. Because of the characteristics of medical and industrial events, as discussed above, analysis of these events using relatively simple techniques can yield meaningful results. Power r~actor events, however, generally involve complex systems and human interactions, can potentially involve more than one adverse consequence, and often result from equipment failures. Therefore, power reactor events can require greater use of more complex risk analysis techniques, such as fault

tree analysis, to yield meaningful insights. Given the dissimilarities in the nature and consequences of the use of nuclear materials in reactors, industrial situations, and medical applications, the Commission recognizes that a single approach for incorporating risk analyses into the regulatory process is not appropriate. However, PRA methods and insights will be broadly applied within the NRC to ensure that the best use is made of available techniques to foster consistency in NRC risk-based decision-making. (C) Defense-in-Depth Philosophy In the defense-in-depth .philosophy, the Commission recognizes that complete reliance for safety cannot be placed on any single element of the design, maintenance, or operation of a nuclear power plant. Thus, the expanded use of PRA technology will continue to support the NRC's defense-in-depth philosophy by allowing quantification of the levels of protection and by - helping to identify and address weaknesses or overly conservati~ regulatory requirements in the physical and functional barriers. 1

                                                      \

III. The Conmission Policy Although PRA methods and information have thus far been used successfully in nuclear regulatory activities, there have been concerns that PRA methods are not consistently applied throughout the agency, that sufficient agency PAA/statistics expertise is not available, and that the Commission is not deriving full benefit from the large agency and industry investment in the

developed risk assessment methods. Therefore, the Conuntssion believes that an overall policy on the use of PRA in nuclear regulatory activities should be established so that the many potential applications of PRA can be implemented in a consistent and predictable manner that promotes regulatory stability and efficiency. This policy statement sets forth the Commission's intention to encourage the use of PRA and to expand the scope of PRA applications in all nuclear regulatory matters to the extent supported by the state-of-the-art in terms of methods and data. Implementation of the proposed policy statement would improve the regulatory process in three areas: foremost, through improved risk-effective safety decision making; through more efficient use of agency resources; and through a reduction in unnecessary burdens on licensees. Therefore, the Comnission proposes the following policy statement regarding the expanded NRC use of PRA: * (l} The use of PRA technology should be increased in all regulatory matters to the extent supported by the state-of-the-art in PRA methods and data and in a manner that complements the NRC's deterministic approach and supports the NRC's traditional defense-in-depth philoso~hy.

                                                   \

(2) PRA and associated analyses (e.g., sensitivity studies, uncertainty analyses, and importance measures) should be used in regulatory matters, where practical within the bounds of the state-of-the-art, to reduce unnecessary conservatism associated with current regulatory requirements, regulatory guides, license commitments, and staff practices. Where appropriate, PRA should be used to support additional regulatory requirements. Appropriate procedures for including PRA in the process

for changing regulatory requirements should be developed and followed. It is, of course, understood that the intent of this policy is that existing rules and regulations shall be complied with unless these rules and regulations are revised. (3) PRA evaluations in support of regulatory decisions should be as realistic as possible and appropri~e supporting data should be publicly available for review. (4) The Convnission's safety goals for nuclear power plants and subsidiary numerical objectives are to be used with appropriate consideration of uncertainties in making regulatory judgments in the context of backfitting new generic requirements on nuclear power plant licensees. Policy Implications There are several important regulatory or resource implications that follow from the goal of increased use of PRA techniques in r~ulatory activities. First, the NRC staff, licensees, and Co,llllission must be prepared to consider changes to regulations, to guidance documents, to the licensing process, and to the inspection program. Second, the NRC staff and Commission must be co11111itted to a shift in the application of resources over a period of time based on risk findings. Third, the NRC staff must undertake a training and development program, which may include recruiting personnel with PRA experience, to provide the PRA expertise necessary to implement these goals. Additionally, the NRC staff must continue to develop PRA methods and

regulatory decision-making tools and must significantly enhance the collection of equipment and human reliability data for all of the agency's risk assessment applications, including those associated with the use, transportation, and storage of nuclear materials. This proposed policy statement affirms the Colllllission's view that PAA methods can be used to derive valuable insights, perspective and general conclusions as a result of an.integrated and comprehensive examination of the design of nuclear facilities, facility response to initiating events, the expected interactions among facility structures, systems and components, and between the facility and its operating staff. IV. Availability of Documents Copies of documents cited in this section are available for inspection and/or for reproduction for a fee in the NRC Public Document Room, 2120 L Street NW, {Lower Level), Washington, DC 20037. Copies of NUREGs cited in 4t this document may be purchased from the Superintendent of Documents, U.S. Government Printing Office, P.O. Box 37082, Washington, DC 2Q013-7082. Copies are also available for purchase from the National T~chnical Information Service, 5285 Port Royal Road, Springfield, VA 22161. In addition, copies of 1) SECY-94-218, 11 Proposed Policy Statement on the Use of Probabilistic Risk Assessment Methods in Nuclear Regulatory Activities," 2) SECY-94-219, NProposed Agency-Wide Implementation Plan for Probabilistic Risk Assessment (PAA)," 3) the Convnission's Staff Requirements Memorandum of September 13, 1994 concerning the August 30, 1994 Convnission meeting on SECY-94-218 and SECY-94-219, and 4) the Commission's Staff

Requ1rements Memorandum of October 4, 1994 on SECY-94-218 can be obtained electronically by accessing the NRC electronic bulletin board system (BBS) Tech Specs Plus. These four WordPerfect* 5.1 documents are located in the BBS MISC library directory under the single filename aPRAPLAN.ZIP". The BBS operates 24 hours a day and can be accessed through a toll-free number, 1-800-679-5784, at modem speeds up to 9600 baud with comunication parameters set at 8 data bits, no parity, I stop bit, full duplex, and using ANSI terminal emulation

• Dated at Rockville, Maryland, this 1st day of December 1994.

FOR THE NUCLEAR REGULATORY COMMISSION

                   ~~Hor.J£_,,Lo--

Di rector, Division of Systems Safety and Analysis Office of Nuclear Reactor Regulation

• \}}