Guidance for Performance-Based Regulation...NUREG/BR-0303 Guidance for Performance-Based Regulation...

NUREG/BR-0303

Guidance for Performance-Based Regulation

U.S. Nuclear Regulatory CommissionOffice of Nuclear Regulatory ResearchWashington, DC 20555-0001

NUREG/BR-0303

Guidance forPerformance-Based Regulation

Manuscript Completed: November 2002Date Published: December 2002

Prepared byN. P. Kadambi

Division of Systems Analysis and Regulatory Effectiveness Office of Nuclear Regulatory ResearchU.S. Nuclear Regulatory CommissionWashington, DC 20555-0001

iii

ABSTRACT

This document provides guidance on aprocess for developing a performance-basedalternative for consideration, along with othermore prescriptive alternatives, in regulatorydecisionmaking. The U.S. Nuclear RegulatoryCommission (NRC) Management Directive6.3, “Rulemaking,” calls for the considerationof a performance-based alternative. Such analternative differs significantly from aprescriptive one in which licensees areprovided detailed direction for obtaining safetyresults. Performance-based approaches focusprimarily on results. They can improve theobjectivity and transparency of NRCdecisionmaking, promote flexibility that canreduce licensee burden, and promote safetyby focusing on safety-successful outcomes.These attributes are reflected in the processdescribed in this document. The process is

set up to develop answers to questions that, inturn, provide the information to formulate analternative that can be compared againstothers in a management review process. Thefive steps in the process are (1) defining theregulatory issue and its context, (2) identifyingthe safety functions, (3) identifying safetymargins, (4) selecting performanceparameters and criteria, and (5) formulating aperformance-based alternative. Examples areprovided to illustrate the process. The formalhigh-level guidelines for performance-basedactivities are shown in Appendix A. Forbroadly scoped and complex issues, a morerigorous consideration of performance issuesmay be appropriate; accordingly, Appendix Bprovides supplementary guidance andbackground information.

v

CONTENTS

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Frame of Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 HIGH-LEVEL PERFORMANCE-BASED GUIDELINES . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.1 Viability Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2 Assessment Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62.3 Guidelines to Ensure Consistency with Other Regulatory Principles . . . . . . . . . . . . . . 6

3 PROCESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.2 Process Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

4 ILLUSTRATIVE EXAMPLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

APPENDICES

A HIGH-LEVEL GUIDELINES FOR PERFORMANCE-BASED ACTIVITIES . . . . . . . . . . . A-1

B SUPPLEMENTARY GUIDANCE FOR MORE COMPLEX ISSUES . . . . . . . . . . . . . . . . . B-1

FIGURES

1 Flow Chart To Address Regulatory Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11B-1 Overview of Objectives Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-3B-2 Reactor Oversight Process Objectives Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-4B-3 More Detailed Decomposition of Means Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . B-5

vii

ABBREVIATIONS

ACRS Advisory Committee on Reactor Safeguards

AFWS auxiliary feedwater system

ALARA as low as reasonably achievable

ASME American Society of Mechanical Engineers

BWR boiling-water reactor

CDF core damage frequency

CFR Code of Federal Regulations

DG draft regulatory guide

GPRA Government Performance and Results Act

LERF large-early-release frequency

NRC U.S. Nuclear Regulatory Commission

PRA probabilistic risk assessment

PWR pressurized-water reactor

ROP Reactor Oversight Process

SRM staff requirements memorandum

1

1 INTRODUCTION

1.1 Background

A performance-based regulatory actionachieves defined objectives and focuses onresults. It differs significantly from aprescriptive action in which licensees areprovided detailed direction on how thoseresults are to be obtained. For example, in thereactor arena, one can envision a U.S.Nuclear Regulatory Commission (NRC)regulatory concern involving the reliability ofemergency backup diesels during stationblackout accidents. A prescriptive approachwould direct the licensee to perform specificdetailed maintenance operations, testingprocedures, and inspections at precise timeintervals. A performance-based approachwould simply set a performance objective(e.g., diesel reliability of 95 percent) and allowthe licensee considerable freedom in how toachieve that reliability objective. Similarly, theas low as reasonably achievable (ALARA)provisions of Title 10, Part 20, of the Code ofFederal Regulations (10 CFR Part 20), whichimpact both materials and reactor licensees,are performance-based in that they allowlicensees to meet the specified dose limits in amanner that they deem most appropriate. Ifthe NRC had written a prescriptive Part 20, itmight have included specific time limits inspecific radiation areas, and required rigidrules concerning the use of respirators andprotective suits under defined conditions.

Performance-based regulatory approachespossess inherent strengths that can lead tomore effective regulation. Examples includeimproving the objectivity and transparency ofNRC decisionmaking, promoting licenseeflexibility in response to regulatoryrequirements that can reduce licensee burden,and promoting safety by focusing on safety-successful outcomes.

The history of NRC’s activity in performance-based regulation began with the staffrequirements memorandum (SRM) of January22, 1997 (Ref. 1), in which the Commissiondirected the staff to propose a plan to develop

performance-based objectives that are notamenable to probabilistic risk assessment.

Why must the staff consider performance-based regulatory approaches?

Performance-based regulatory approachesare considered for policy and effectivenessreasons. The policy considerations arebased on the 1993 GovernmentPerformance and Results Act (GPRA).NRC’s Strategic Plan, developed inresponse to GPRA, provides specificperformance goals that drive the agency’sregulatory program. Performance-basedapproaches are mentioned in the goals forreactors, materials, and waste. Regulatoryeffectiveness has been found to improvewhen such approaches are usedappropriately.

What is the basis for the guidance in thisdocument?

The basis for the guidance in thisdocument is the High-Level Guidelines forPerformance-Based Regulation (Ref. 4)(hereafter referred to as “the high-levelguidelines”). The high-level guidelinescover the three arenas (reactors,materials, waste) and a broad range ofissues within each arena. Accordingly, thehigh-level guidelines are formal andabstract.

How will this guidance documentaccomplish its objectives?

It provides simplified guidance for many ofthe regulatory issues that NRC staff maybe tasked to resolve. It may not covercomplex considerations such as defensein depth. Appendix B has been preparedas a reference to support formulation ofperformance-based alternatives to addressbroader or more complex issues.

2

The initiative was given further definition byDirection Setting Issue 12 (Ref. 2) and theWhite Paper on “Risk-Informed andPerformance-Based Regulation,” SRM toSECY-98-144 (White Paper) (Ref. 3). Thecontinuing efforts of the staff, which includedpublic workshops, led to the publication ofhigh-level guidelines for performance-basedactivities (SECY-00-191 (Ref. 4) and 65 FR26772 (Ref. 5)). These guidelines weredeveloped with interoffice participation by thePerformance-Based Regulation WorkingGroup. The developmental aspects of thisactivity will end with the issuance of thisguidance document.

This document describes the use of high-levelguidelines for determining whether aperformance-based approach can be appliedto a given regulatory activity. In addition, itprovides insights into the formulation ofperformance-based alternatives. Although atendency exits to characterize a regulatoryapproach as either performance-based orprescriptive, the reality is that the most likelyand preferred approach will often be a blendof the two. Thus, when this document refers tomaking an approach performance-based, theintent is actually to make it as performance-based as possible.

1.2 Frame of Reference

The NRC is heavily committed to theidentification and evaluation of regulatoryactions. Such activities are undertaken inresponse to a wide array of regulatory issues,and they typically focus on ways to improveperformance relative to NRC’s goals asarticulated in the Strategic Plan (Ref. 6).These activities tend to involve concerns overpublic health and safety; public confidence;regulatory effectiveness, efficiency, andrealism; and unnecessary regulatory burden. When a decision is made to evaluate aregulatory issue, it is standard practice for thestaff to use NUREG/BR-0058, Rev. 3,“Regulatory Analysis Guidelines of the U.S.Nuclear Regulatory Commission,” (Ref. 7) asguidance. The regulatory analysis prescribedtherein is designed to determine whether aregulatory action is needed, to provide

adequate justification for the proposed action,and to explain why a particular action wasrecommended. The heart of the regulatoryanalysis is a cost-benefit assessment thatconsists of a systematic evaluation of theconsequences associated with a range ofalternative responses and the selection of thepreferred alternative. It is clear thatconsideration of the alternatives is criticallyimportant in this overall decisionmakingprocess. NUREG/BR-0058 recognizes thedesirability of including a performance-basedapproach as one of the alternatives to beevaluated. In NUREG/BR-0058, Section 4.2,“Identification and Preliminary Analysis ofAlternative Approaches,” states –

If the objective or intended result of aproposed generic requirement or staffposition can be achieved by setting areadily quantifiable standard that hasan unambiguous relationship to a

What must the staff do?

In order to increase regulatoryeffectiveness, the Commission has directedthe staff to consider risk-informed andperformance-based alternatives when achoice is being made between differentregulatory approaches. The staff can usethis document to accomplish theCommission’s objective.

How should the staff proceed?

The staff can begin with this document tosee if concepts such as safety function andsafety margin are relatively simple toevaluate in the context of the specificregulatory issue. If the resolution issensitive to where and how performance ismeasured, Appendix B should beconsidered for supplementary guidance.The regulatory analysis, which considersthe costs and benefits of each alternative,can document the preferred alternative.

3

readily measurable quantity1 and isenforceable, the proposed requirementshould merely specify the objective orresult to be obtained rather thanprescribe to the licensee how theobjective or result is to be attained. Inother words, requirements should beperformance-based, and highlyprescriptive rules and requirementsshould be avoided absent good cause tothe contrary.

For many applications, the guidance in thepresent document will suffice to support aperformance-based approach. For issueswhose resolution is sensitive to where andhow performance is measured, or in casesinvolving cross-cutting issues, staff may needto undertake a more considered developmentas described in Appendix B.

1The “readily measurable quantity” does notnecessarily imply that only direct measures ofphysical parameters (such as length, weight,temperature, pressure, flow rate) are acceptable.Although direct measures (called naturalmeasures) are preferred, objective measures ofother sorts should also be considered. Thisbroader interpretation of the term “readilymeasurable quantity” is based on more recentwork on performance-based approaches toregulation.

5

2 HIGH-LEVEL PERFORMANCE-BASED GUIDELINES

In SECY-00-191, September 1, 2000, theNRC issued high-level performance-basedguidelines for identifying and assessingpotential performance-based regulatoryactions. “High-level” means that the guidelinesare applicable across the full spectrum of NRCregulatory activity, corresponding to the threeNRC arenas, reactor safety, material safety,and waste safety.

The guidelines are meant to raise questions,the answers to which should assist the staff indetermining whether and how to pursue aperformance-based alternative. The staff mayexercise discretion in applying theseguidelines, which are classified into threegroups (1) viability guidelines, (2) assessmentguidelines, and (3) guidelines to ensureconsistency with other regulatory principles. Asummary follows.

2.1 Viability Guidelines

Viability guidelines ask questions that enablethe regulator to determine whether a specificregulatory issue is amenable to aperformance-based approach based on howwell the regulator can construct a regulatoryalternative that has the four attributesdiscussed in the Commission’s White Paper.These attributes are:

• Failure to meet the predeterminedperformance standard will not result in animmediate safety concern. (Can margin beestimated realistically, and if so, what isknown about it?)

• Measurable or calculable parameters areavailable to determine whether theperformance standard is met. (Canperformance parameters be identified thatprovide measures of performance and theopportunity to take corrective action ifperformance is lacking?)

Can a “performance-based approach” haveprescriptive elements?

Appropriate regulatory decisionmakingcannot exclude the possibility ofprescriptive elements. The characteristic ofa performance-based approach, asdescribed in the Commission’s WhitePaper, is a reliance on performance andresults. This is evident from the followingstatement in the White Paper, “Aperformance-based regulatory approach isone that establishes performance andresults as the primary basis for regulatorydecisionmaking....” The focus of aperformance-based approach is the use ofprescriptive elements only whennecessary.

How does “margin” enter into a“performance-based approach?”

One of the White Paper attributes of aperformance-based approach is that, “...aframework exists in which the failure tomeet a performance criterion, whileundesirable, will not in and of itselfconstitute or result in an immediate safetyconcern.” Such a framework contains theconcept of “margin.” In this construct,“margin” is a quantity that expresses thedifference between performance within thelimits of a “criterion” and performance thatis representative of a “concern.” The word“immediate” requires that a time elementbe considered in the development of aperformance-based approach. The high-level guidelines incorporate thisunderstanding. They are also consistentwith the NRC’s regulatory responsibility tomonitor potential erosion of margin, as wellas licensee responsibility for promptcorrective actions. These interpretationshave been discussed with the public andpresented to the Commission.

6

• The performance standard is based onobjective criteria. (Can objective criteria bedeveloped that are indicative ofperformance?)

• The licensee or the NRC has flexibility in themethod used to achieve the desiredperformance level. (Is flexibility for the NRCor licensees available consistent with thelevel of margin?)

If a regulatory alternative can be designedwith these three attributes, a performance-based approach is judged to be feasible. Thisassessment would be applied on a case-by-case basis and would be based on anintegrated consideration of these guidelines,rather than on strict adherence to eachindividual guideline.

In terms of relative importance, the guidelineconcerned with performance failure leading toan immediate safety concern is pre-eminent. Aperformance-based requirement is justifiedonly if assurance exists that adequate safetymargins can be preserved to meet regulatoryneeds. A safety margin is adequate for thispurpose when, if there is a failure to meet theperformance objective, sufficient time will beavailable to take corrective action to avoid amore serious condition associated with asafety concern. The importance of safetymargin considerations justify placing thisguideline as the first among the viabilityguidelines. Hence, if sufficient margin existsunder the first viability guideline, a broadrange of less-prescriptive approaches becomeviable, including performance-basedapproaches. The three subsequent viabilityguidelines characterize the performance-based approach.

The White Paper associates flexibility with theconcept of incentives. It states that one of theattributes of a performance-based approach isthat licensees have flexibility to determine howto meet the established performance criteria inways that will “encourage and rewardimproved outcomes.” The coupling of flexibilityand licensee incentives has been addressedin the formal guidelines in Appendix A.

2.2 Assessment Guidelines

If a performance-based approach is deemedviable, the regulatory activity would beevaluated against guidelines that assesswhether such an approach results inopportunities for regulatory improvement.Regulatory improvement is a positivecontribution to NRC’s performance goals andachievement of a net societal benefit. Thus,the assessment guidelines question whetherthe regulatory alternative achieves thefollowing:

• maintains safety• increases public confidence• increases effectiveness, efficiency, and

realism• reduces unnecessary regulatory burden• results in a net benefit

Additional assessment guidelines include theability of the proposal to be incorporated intothe regulatory framework and the ability toaccommodate new technology. Thisevaluation is to be based on an integratedassessment of the individual guidelines withinthis grouping.

Many of the considerations that apply to thisset of guidelines are also pertinent to the cost-benefit evaluation performed as part of theregulatory analysis. Hence, informationdeveloped here could also support thealternatives analysis described in theRegulatory Analysis Guidelines.

2.3 Guidelines to Ensure Consistency withOther Regulatory Principles

These guidelines take into accountfundamental regulatory principles that havebeen articulated by the Commission, such as,the Principles of Good Regulation (Ref. 6).The intent is to ensure that a performance-based regulatory alternative that conforms tothe viability and assessment guidelines doesnot compromise any of NRC’s basic regulatoryprinciples. Although it is not generallynecessary to remind staff of these principles,this third set of guidelines provides areasonable check.

7

The third set of guidelines need only beapplied if the candidate activity passes the firsttwo sets of guidelines.

A complete presentation of the guidelinesappears in Appendix A.

9

3 PROCESS

3.1 Objectives

The objective of this document is to provide aprocess that can be widely used for decisionsconcerning performance-based regulatoryalternatives. As noted above, theperformance-based guidelines are articulatedat a high level and are applicable to regulatoryissues involving the reactor, materials, andwaste arenas. The process must be relativelygeneral for it to have such wide relevancy.More detailed guidance, tailored specifically toindividual arenas, is beyond the scope of thisdocument. It is anticipated that each of themajor arenas will have additional guidance inthe form of office procedures that willeffectively supplement this document.Nevertheless, for many regulatory actions, thenature and objectives of the regulatory issuewill be clear and the knowledge base largeenough that a relatively modest process, suchas the one described here, will be sufficient.

Although applicable across all regulatoryarenas, considerable leeway exists in the levelof adherence to the general process outlinedhere. For example, in certain instances it maybe so clear that a performance-basedapproach is not feasible that formal applicationof these guidelines could be largelyunnecessary. Circumstances are also likely inwhich the appropriateness of using aperformance-based approach and thecharacteristics of that approach are veryobvious and straightforward. Decisions onthese matters are almost always governed byexpert-judgment assessments of margins thateffectively mimic the substance of the processbeing spelled out here. In such cases, the staffmay simply choose to use these guidelines asa check on the performance-based approachthat is being proposed. Similarly, theguidelines may also be useful to test aproposed regulatory action that purports to beperformance-based to determine whether theaction, under a formal scrutiny, in fact meritsthe label.

Alternatively, some complex regulatory issuesmay require that a far more detailed process

be used than the one described here. If so,concepts and tools such as decision theoryand objectives hierarchies would likely beneeded. NRC staff in need of, or interested in,a more systematic and structured process aredirected to Appendix B and the referencestherein for decision theory concepts.

Regardless of the level of effort, somedocumentation for case-specific applicationsof this process should be maintained fortransparency and efficiency. Individual officeprocedures should specify the requirementsfor documentation.

What is the relationship between risk-informed and performance-basedapproaches to regulation?

Risk-informed and performance-basedapproaches to regulation complement oneanother. Risk information, when aprobabilistic risk assessment (PRA) isavailable, can be useful in finding the mostsafety-significant functions and systems. Ifa PRA is not available, operationalexperience may provide enoughinformation. Safety would be best servedby demanding the highest (i.e., mostaggregated) levels of performance fromthe most safety-significant structures,systems, and components.

In SECY-01-0218, “Update of the Risk-Informed Regulation Implementation Plan”(December 5, 2001) (Ref. 8), the staff hasstated that, to the extent appropriate,activities to risk-inform regulations shouldalso incorporate the performance-basedapproach to regulation. The corollary isalso true–performance-based regulationsshould be risk-informed when possible.

10

Figure 1 is a schematic representation thatcharacterizes the most important elements ofthe overall process of addressing a regulatoryissue. The guidance presented here isspecifically concerned with the development ofa performance-based regulatory alternativeand corresponds to an entry point at the“Performance-Based” or “Risk-Informed &Performance-Based” boxes of Figure 1. Theprocess contains a number of feedback loops,which is indicative of the iterative nature of thisprocess.

The process is intended to determine whethera given regulatory issue is suited to aperformance-based approach, and if so, toidentify an alternative that most stronglyreflects performance-based attributes. Inaddition, other competing objectives must beconsidered, such as setting performance andreliability measures at the highest (i.e., mostaggregated) possible level and, to the extentpracticable, using multiple parameters tosatisfy the defense-in-depth philosophy.Meeting these objectives may requireapplication of the methodology in Appendix B.This process is intended to enable the staff toidentify a preferred performance-basedalternative based on the consideration of allthese objectives. This is accomplishedthrough an iterative process, similar to thatdepicted in Figure 1. Each iteration shouldresult in more detailed and focusedinformation that can be used toward improvingthe performance-based alternative.

This document has been prepared asguidance to be used in the context ofrulemaking. However, the resolution of aregulatory issue may involve changes to anypart of the regulatory framework and may notresult in an actual regulation. For example,improvements obtained through the revisedreactor oversight process, which isperformance-based in nature, were realizedmainly by changes to the inspection,assessment, and enforcement aspects of theregulatory framework. Those improvementsare considered to be highly effective eventhough rulemaking was not involved.

3.2 Process Steps

For any given regulatory issue, the design of aperformance-based approach depends on fivebasic considerations that are addressed indetail below. Each of these elements istreated as a step in a process. The answers tothe questions provide the basis for developingthe required details on a performance-basedalternative, if it is feasible.

Step 1 – Defining the Regulatory Issue andits Context

Purpose – To define the regulatory issue withclear objectives that are well-understood.

• What is the arena and sub-arena for theregulatory issue?

• Which of the NRC’s performance goalsdoes the regulatory issue address?

• What are the expected outcomes andresults from resolution of the regulatoryissue?

Does a risk-informed and performance-based approach to regulation necessarilyinvolve rulemaking?

No, any part of the regulatory frameworkcan be considered for modification.According to the Strategic Plan (Ref. 6),the regulatory framework consists ofseveral interrelated aspects. They are (1)the NRC's mandate from Congress in theform of enabling legislation, (2) the NRC'srules in Title 10 of the Code of FederalRegulations, (3) the regulatory guides andreview plans that amplify those regulations,(4) the body of technical information,obtained from research performed by NRCor by others and from evaluation ofoperational experience, that supports thepositions in the rules and guides andreview plans, (5) the licensing andinspection procedures utilized by the staff,and (6) the enforcement guidance.

11

Figure 1: Flow Chart To Address Regulatory Issues

OperatingExperience

CommissionDirection

StakeholderSuggestion

StakeholdersInvolvement

NRC Identifies Proposed Modification toRegulatory Framework

Prioritization Using NRCPerformance Goals

Performance-Based

TraditionalApproach

Select an Option

Risk-Informed

Selected Option DoesNot Meet Viability

Criteria

Define ProposedModification

Develop Regulatory Framework (e.g. Rulemaking Package)According to Existing Office Procedures

Risk-Informed &Performance-

Based

StaffInitiatives

ApplyAppropriatePrinciples

Develop Approach and Technical Bases

1 Flow Chart To Address Regulatory Issues

12

Step 2 – Identifying the Safety Functions

Purpose – To identify the safety functions andsystems that affect the regulatory issue(directly or indirectly).

• What are the safety functions or conceptsthat can impact the regulatory issue?

This inquiry should be focused on the natureof the regulatory issue and the expectedoutcomes. For example, if the regulatory issueis directed at placing appropriate qualityrequirements on a set of components, thesafety functions (which are functions thatanswer the question, “How is risk mitigated ?”)served by the components would have to beconsidered, along with the sensitivity of thefunction to the quality level.

• What equipment/systems/procedures arenecessary to satisfy the safety function?

• What level of safety (based on appropriatemetrics) is required to meet the objectives ofthe regulatory issue?

For example, if the objective is to maintainsafety while relaxing the stringency of aregulatory requirement, an appropriate metricfor monitoring safety should be found andmonitored when the regulatory requirement ischanged.

Step 3 – Identifying Safety Margins

Purpose – To evaluate margins and identifyperformance parameters (if any) that satisfythe regulatory issue objectives.

• How much safety margin is available, andhow robust is it, for performance monitoringto provide a basis for granting licenseeflexibility?

As mentioned earlier, an effectiveperformance-based approach to regulationprovides flexibility to the NRC and licenseesprovided there is sufficient safety margin.Safety margin can be divided into two parts,physical and temporal. Physical margin is thedifference between two physical conditions,

the first of which represents expectedconditions and the second of which representsa performance-limiting condition. An exampleof a performance-limiting condition is the peakpressure capability of a pressure vessel.Physical margin in a pressure boundary is thedifference between the pressure-retainingcapability of the vessel and the expectedmaximum pressure during an accidentcondition.

A temporal margin represents the timeavailable to identify a concern and to takeactions, such as restoring a failed safetyfunction, implementing a corrective actionprogram, or initiating a regulatory responsethat mitigates the concern. A temporal marginin a spent fuel pool, for example, could be thedifference in time between when thetemperature of the pool water is detected tobe at some elevated level (caused by loss ofcooling) and the time needed to reach theboiling point of the water.

“Robustness” of a safety margin means thatthe margin between two performance levels issignificantly greater than uncertainty andnormal variability in performance. If thiscondition is met, a very low probability of theperformance parameter crossing the limitexists, unless performance changes in a verysignificant way. In any case, wherever there issubstantial uncertainty, achieving robustnessrequires that nominal performance levels beset more conservatively than when there isless uncertainty. Depending on the situation,uncertainty can be assessed using explicitmodels (e.g., PRAs), expert judgment, oractuarial methods based on operatingexperience.

The term “margin” is employed in thisguidance somewhat differently than in itstraditional use for health and safety regulation.The significance of “margin” is closelyassociated with the other factors in the viabilityguidelines, namely, performance parameters,objective criteria, and flexibility. If themagnitude of the safety margin is sufficient tosupport a performance-based approach, itcan, in concept, be subdivided andapportioned in such a way as to consider the

13

objectives of stakeholders. The NRC andlicensees are the principal stakeholders inresolving a regulatory issue.

For example, the initiating event indicator inthe reactor oversight process (ROP) forunplanned scrams ranges from 0–25 per 7000hours of critical operation. The performanceof a licensee would be declared asunacceptable from a safety standpoint if suchscrams exceed 25 in number. Thus,according to the action matrix, green levelperformance is limited to 0–3 scrams, white ischaracterized by 4–6 scrams, yellow ischaracterized by 7–25 scrams, and reddenotes more than 25 scrams. Given theadverse economic consequences ofunplanned scrams, reactor licensees have astrong incentive to minimize them and stay inthe green band. Almost all licenseesaccomplish this objective in practice. Thelicensee will likely view exceeding 3 scramsas a “serious” concern or condition. In thisway, a proxy objective has been created thatworks in NRC’s favor because, althoughcrossing the green/white threshold has only aminor adverse safety impact, the licensee hasa strong incentive to correct the underlyingcause expeditiously.

Hence, an evaluation of the available marginand its robustness should include a search forappropriate performance parameters thatprovide for operational flexibility as well as themeans to fulfill regulatory responsibilities.

• What observable characteristics,quantitative and qualitative, exist within thesafety functions identified in Step 2?

Measurable or calculable parameters aregenerally associated with quantitativeobservable characteristics. Parameters thatare observed directly, such as pressure,temperature, flow, incurred cost, and radiationexposure, are called natural measures. Somenatural measures require simple calculation,such as reliability, percentage, andconcentration. However, qualitativeparameters are also able to support aperformance-based approach under

appropriate conditions and should not beoverlooked.

• Can constructed measures be developedthat provide qualitative expressionscapable of observation with reasonableobjectivity?

Step 4 – Selecting Performance Parametersand Criteria

Purpose – To select a complement ofperformance parameters and objective criteria(if possible) that both satisfy the viabilityguidelines and resolve the regulatory issue.

• Can the identified observablecharacteristics, together with objectivecriteria, provide measures of safetyperformance and the opportunity to takecorrective action if performance islacking?

• Can objective criteria be developed thatare indicative of performance and thatpermit corrective action?

How can qualitative observations be used?

Qualitative observations present specialchallenges, but should not be ignored. Forexample, the quality of housekeeping in anuclear facility is an important aspect ofpreventing fire hazards. Qualitativeobservations can be quite effective inassessing such a characteristic. Alinguistically defined measure thatrepresents a level of impact orsignificance, called a constructed measure,is a way to represent qualitativeobservations. A constructed measurebecomes necessary when naturalmeasures do not exist or are too difficult touse. It is used to describe performanceneeded to satisfy higher-level objectives.Examples are (1) impact on publicconfidence is high, medium, or low, or(2) environmental significance is high,medium, or low.

14

• Is flexibility (for NRC and licensees)available consistent with level of margin?

The approach recommended by the AdvisoryCommittee on Reactor Safeguards (ACRS) isto apply the performance criteria at as high alevel as practicable (Ref. 9). Setting thecriteria at a higher level can allow moreflexibility. However, the need to assureopportunity to take appropriate correctiveaction requires that criteria be setappropriately for the regulatory issue, in a waythat depends on available margin. In general,this tradeoff between flexibility and the needfor prompt corrective action will require aniterative approach. If the complexity of theregulatory issue increases beyond a certainpoint, it may be necessary to apply themethodology described in Appendix B.

The set of questions in Step 4 may not beamenable for application exactly in thesequence shown. They should be viewed asrequiring iteration when questions of margin,corrective action, and flexibility stronglyinteract with one another. Strong linkages canexist between observable characteristicschosen as the performance parameters to beused in a performance-based approach andthe assessment of margin based on criteriaapplied to these parameters. For example, inthe area of quality assurance, the quality ofemergency backup power provided by adiesel generator would not necessarily bewell-reflected just by the criteria that areapplied to each component part of the dieselgenerator. Even if very strict quality criteriaare applied to each of the component parts,the overall diesel generator performance maynot meet regulatory standards. On the otherhand, a diesel generator could adequatelymeet performance standards even if thecomponent parts are only commercial grade.

Step 5 – Formulating a Performance-BasedAlternative

Purpose – To determine the appropriateimplementation of a performance-basedapproach within the regulatory framework.

• Does the performance-based regulatoryalternative provide necessary andsufficient coverage of the regulatory issueobjectives?

One of the important elements of coverage isconsideration of defense-in-depth. The ACRSrecommendations (Ref. 9) included one thatwould involve multiple performanceparameters that provide redundant informationto satisfy the defense-in-depth philosophy.The NRC’s defense-in-depth philosophy alsoincludes consideration of “prevention” and“mitigation” strategies which operate in properbalance. If the regulatory issue involvescomplex defense-in-depth issues, the neededguidance should be drawn from Appendix B.

• Of the performance parameters selectedin Step 4, which of them requires that aprescriptive approach be used to meetregulatory needs? Can a combination ofperformance-based and prescriptivemeasures be implemented such that theresolution of the regulatory issue is asperformance-based as possible?

• Has the regulatory alternative beenconsidered for implementation within each

What happens if incentives are excludedfrom the regulatory approach?

Experience shows that absence of properincentives, or the existence of “perverse”incentives, can result in such emphasis oncompliance that safety may be adverselyaffected. One example of such anoccurrence is when licensees, faced withthe approach of the end of an allowableoutage time for a safety systemmaintenance, may feel forced into actionsthat may meet compliance standards, butare not fully supportive of safety. TheCommission’s policy on such matters ispresented in detail in COMSAJ-97-008,“Discussion on Safety and Compliance”(Ref. 10).

15

of the levels of the regulatory framework sothat an optimum level is proposed?

For example, a prescribed parameter can beincluded in a Technical Specification or otherlicense condition. It may be possible toprovide flexibility in operation for parametersthat do not have to be strictly controlled. Also,as mentioned above, staff may consider oneor another type of incentives for licensees toincrease the likelihood of improved safetyoutcomes.

• Are licensees’ incentives appropriatelyaligned, considering the overall complementof performance measures, criteria, theimplementation, and the regulatoryframework as a whole?

Licensees’ flexibility can be coupled withpositive and negative incentives. Examples ofpositive incentives occur when licensees maybe able to reduce costs of operation if theymeet specified levels of safety or trends insafety of operation. Examples of negativeincentives occur when NRC’s enforcementpolicy causes licensees to experienceundesired consequences when levels ofsafety or trends in safety are unfavorable.

Regulation that is based on sampling licenseeperformance needs to be designed with care,in order to avoid incentivizing performance inone important area at the expense of another,with a net adverse outcome. As a hypotheticalexample, regulation that sought only tominimize the unavailability of componentsmight create an incentive to reducemaintenance to a level at which unreliabilityperformance would be adversely affected.When staff considers a change to theregulatory framework, they must also considerthe incentives created by the new overallframework.

• Is it worth modifying the regulatoryframework in the manner proposed,considering the particulars of the regulatoryissue?

Summary

This step-by-step process takes the staffthrough a series of questions that should leadto answers to the following fundamentalquestions (1) What is the regulatory issue? (2)Is it possible to use a performance-basedapproach to resolve the issue? (3) Does itappear worthwhile to pursue a performance-based solution? Although a formal response tothe last question requires application of theassessment guidelines, managementjudgment can be sufficient for considering aperformance-based alternative in a regulatoryanalysis or other appropriate process.

17

4 ILLUSTRATIVE EXAMPLES

For illustration purposes, this section presentsactions on three recent regulatory issues arepresented below to exemplify application ofthe process steps. By applying this process,staff can determine whether a regulatoryrequirement can be made more performance-based. These examples also demonstratehow different elements of the regulatoryframework can be targeted for the mosteffective and efficient application of theguidance steps and hence improve the netbenefit from a performance-based proposal.

The first example is a recent proposal torevise 10 CFR 50.44 concerning combustiblegas control in containment at nuclear powerplants (Ref. 11). The draft regulatory guideprovided with the rulemaking package, DG-1117, offers a performance-based approach tomeeting the requirement for hydrogenmonitors to be functional, reliable, andcapable of measuring hydrogen concentrationin the containment after an accident.

The second example is a proposed rule thataddresses geological and seismologicalcharacteristics for siting and design of drycask independent spent-fuel storageinstallations and monitored retrievable storageinstallations (Ref. 12). The regulatory analysisfor this rulemaking recommends aperformance-based approach in order to meetthe safety objectives in a cost-effectivemanner.

The third example was originally proposed asa risk-based performance indicator forshutdown conditions for use in the ROP. NRCstaff is considering Its essential aspects arebeing considered for use in the significancedetermination process relative to inspectionfindings during shutdown conditions (Ref. 13).

The process steps are presented below usingspecific factors that entered into incorporatingperformance-based concepts in each of theexamples. The examples (identified asExamples 1, 2, and 3) have been employed ineach of the process steps introduced inSection 3.2 to the extent that they successfully

illustrate the step. In other words, each stepdoes not employ all examples. Example 3 hasnot been incorporated into the ROP; however,it represents a useful illustration of a way tolook for the most effective performanceparameters.

Step 1 – Defining the Regulatory Issue andits Context

• What is the arena and sub-arena for theregulatory issue?

In the case of Examples 1, 2, and 3, theissues belong in the reactor and wastearenas, with the sub-arenas in rulemaking,spent fuel safety and inspection programs,respectively.

• Which of the NRC’s performance goalsdoes the regulatory issue address?

The examples potentially cover all four of theagency’s performance goals.

• What are the expected outcomes andresults from resolution of the regulatoryissue?

The expected outcome in each example is toidentify the performance-based alternativesthat optimally use the flexibility of theregulatory framework to obtain a cost-effectiveresolution of each regulatory issue.

Step 2 – Identifying the Safety Functions

• What are the safety functions or conceptsthat can impact the regulatory issue?

In Example 1, the safety functions areaccident management to maintaincontainment integrity and effective emergencyplanning. In Example 2, the safety functionsare stability against soil liquifaction duringvibratory motion, and cask sliding andresulting displacements during an earthquakeevent.

18

• What equipment/systems/procedures arenecessary to satisfy the safety function?

The technical evaluation for Example 1showed that non-safety-related commercialgrade hydrogen monitors that meet lowercategory requirements than the existingsafety-related systems would satisfy the safetyfunction. The evaluation for Example 2showed that more cost-effective analyses maybe possible to satisfy siting criteria thanexisting prescriptive criteria, which essentiallyrequire the same criteria as operation of anuclear power plant.

• What level of safety (based on appropriatemetrics) is required to meet the objectives ofthe regulatory issue?

For Example 1, the required level of safety isthat which meets the objective of reducing, toan acceptable level, the risk of earlycontainment failure from hydrogencombustion. Hence, the metric in this case isthe conditional containment failure probability.For Example 3, the required level of safety isto maintain at an acceptable level the coredamage risk associated with certainconfigurations typical of shutdown operations. Step 3 – Identifying Safety Margins

• How much safety margin is available, andhow robust is it, for performance monitoringto provide a basis for granting licenseeflexibility?

Example 2 exemplifies an instance for whichthe safety margin can be assessedqualitatively, yet reliably. The casks inquestion are designed to safely withstand theconditions associated with transportation andhandling of spent-fuel operations. The rigorsof such operations, and the associatedregulatory requirements, can reasonably beexpected to envelop, with considerablemargin, the challenges posed by vibratorymotion and sliding displacements during anearthquake.

A different type of margin can be identified inExample 3. The technical evaluation identified

four risk-significant categories ofconfigurations for pressurized-water reactors(PWRs) and three for boiling-water reactors(BWRs). For each reactor type, the categoriesrange from low risk to high risk. Baselineperformance levels were established usingoperating history. In the low-riskconfigurations, baseline performance isachieved by exiting a configuration within 20days in PWRs and 2 days in BWRs. At theother end of the spectrum, the threshold forunacceptable performance occurs at 1 day ina high-risk configuration. Applying theconcepts of the ROP and its action matrix,under nominal conditions, licensees wouldexpect to operate within the green band,which corresponds, for PWRs, to 21 days in alow-risk configuration and by totally avoiding ahigh-risk configuration. One day in a high-riskconfiguration would denote unacceptableperformance. Hence, monitoring the time atrisk-significant shutdown configurationsprovides the NRC and the licensee with aperformance-based approach for maintainingsafety margins at acceptable levels.

• What observable characteristics,quantitative and qualitative, exist withinthe safety functions identified in Step 2?

For Example 1, the observable characteristicscome from the results of periodic servicing,testing, and calibration of hydrogen monitors.The operating margin would come from acomparison between these results and thetarget values established by the licenseeunder the maintenance rule. For Example 2,the observations would likely be based onverification of design margins by post-earthquake inspections, and corrective actionsas necessary over the extended periods oftime over which such sites may be inoperation.

• Can constructed measures be developedthat provide qualitative expressionscapable of observation with reasonableobjectivity?

Among the examples considered, Example 2may require a constructed measure to

19

characterize the response of the storagefacility to an actual earthquake.

Step 4 – Selecting PerformanceParameters and Criteria

• Can the identified observablecharacteristics, together with objectivecriteria, provide measures of safetyperformance and the opportunity to takecorrective action if performance is lacking?

In Example 1, the recorded results of themaintenance program provide the observablecharacteristics and also provide theopportunity to take corrective actions if lack ofperformance is noted by the NRC or thelicensees themselves.

The performance criterion in Example 3 is thetime period in each of the shutdownconfigurations. This exemplifies a focusedapplication of the recommendation of theACRS that the performance parametersshould be identified at as high a level aspracticable. If this recommendation had notbeen followed, all systems and sub-systemsinvolved in risk-significant configurations mighthave been targeted for monitoring. Themanagement of risk in such configurationscould still be accomplished. Therecommendation to search for parameters at ahigh level would direct the analyst’s attentionto other, more cost-effective possibilities. InExample 3, time in risk-significantconfigurations fulfills the needed attributes.The second ACRS recommendation ofproviding for defense-in-depth by redundantobservations can also be implemented byseparately monitoring the safety functions ofcertain risk-significant components.

• Can objective criteria be developed that areindicative of performance and that permitcorrective action?

In both Examples 1 and 3, the regulatoryissues are highly amenable to establishmentof objective criteria that are indicative ofperformance, as well as permit sufficient timefor corrective action.

• Is flexibility (for NRC and licensees)available consistent with level of margin?

Example 2 exemplifies how the existence ofsubstantial margins enables consideration ofhigher magnitudes of flexibility. Because of therobustness of the casks, licensees canpotentially employ simplified analyses to showthat the uncertainties associated withseismicity and cask response can be reliablyenveloped to demonstrate conformance withsite suitability criteria. Simplified analysesprovide greater cost-effectiveness.

Similarly, in Example 3, a range of flexibility isconsidered, consistent with margin available ineach shutdown configuration category. Theleast flexibility would be available for the high-risk category, which is appropriate from aregulatory perspective.

Step 5 – Formulating a Performance-BasedAlternative

• Does the performance-based regulatoryalternative provide necessary andsufficient coverage of the regulatory issueobjectives?

In both Examples 1 and 2, the staffdetermined that sufficient margin is present forNRC staff to confirm directly that licenseeperformance is adequate in key areas, withoutrequiring compliance with prescriptiverequirements in all areas. Therefore, a less-prescriptive, more performance-basedapproach can provide suitable coverage of theregulatory issue objectives.

• Of the performance parameters selectedin Step 4, which of them requires that aprescriptive approach be used to meetregulatory needs? Can a combination ofperformance-based and prescriptivemeasures be implemented such that theresolution of the regulatory issue is asperformance-based as possible?

Example 3 offers an illustration of how specificprescriptive elements can be incorporated intoa less prescriptive regulatory approach. Theregulatory framework permits inclusion of

20

prescriptive elements through TechnicalSpecification or License Condition provisions.

• Has the regulatory alternative beenconsidered for implementation within each ofthe levels of the regulatory framework sothat an optimum level is proposed?

In both Examples 1 and 2, various alternativemeans for implementing a performance-basedapproach within the regulatory frameworkwere considered. A variety of internalstakeholder input was explicitly includedbefore coming up with the recommendationspublished in the Federal Register.

• Are licensees’ incentives appropriatelyaligned, considering the overall complementof performance measures, criteria, theimplementation, and the regulatoryframework as a whole?

Inappropriate incentives could potentiallyoccur in the case of Example 3. In NUREG-1753, for the case of PWRs in reducedinventory operation, the threshold timescorresponding to green/white and white/yellowthreshold are 1 day and 1.08 days,respectively. Licensees might perceive anincentive to inappropriately rush throughsafety-sensitive operations only to avoidcrossing a threshold. The staff is in a betterposition to structure regulatory provisions thatminimize perverse incentives by explicitlyconsidering the possibilities within theregulatory framework and the ROP’s actionmatrix.

• Is it worth modifying the regulatoryframework in the manner proposed,considering the particulars of the regulatoryissue?

In both Examples 1 and 2, the proposedperformance-based alternatives have beenpublished for comment. In both cases,preliminary indications suggest that theproposals enjoy considerable stakeholdersupport. Hence, the modifications appearworthwhile.

21

5 REFERENCES

1. Nuclear Regulatory Commission (U.S.) (NRC). “Staff Requirements - SECY-96-218 -Quarterly Status Update for the Probabilistic Risk Assessment (PRA) ImplementationPlan, Including a Discussion of Four Emerging Policy Issues Associated with Risk-Informed Performance-Based Regulation.” NRC: Washington, DC. January 22, 1997.

2. Nuclear Regulatory Commission (U.S.) (NRC). “Staff Requirements - COMSECY-96-061- Risk-Informed, Performance-Based Regulation (DSI 12).” NRC: Washington, DC. April15, 1997.

3. Nuclear Regulatory Commission (U.S.) (NRC). “Staff Requirements - SECY-98-144 -White Paper on Risk-Informed and Performance-Based Regulation.” NRC: Washington,DC. March 1999.

4. Nuclear Regulatory Commission (U.S.) (NRC). SECY 00-191, “High-Level Guidelines forPerformance-Based Activities.” NRC: Washington, DC. September 2000.

5. Nuclear Regulatory Commission (U.S.), Washington, DC. “Revised High-LevelGuidelines for Performance-Based Activities,” Proposed Rule. Federal Register: Vol. 65,No. 90. pp. 26772-26776. May 9, 2000.

6. Nuclear Regulatory Commission (U.S.) (NRC). NUREG-1614, Vol. 2, Parts 1 & 2,“Strategic Plan Fiscal Year 2000 - Fiscal Year 2005.” (Final Report) NRC: Washington,DC. September 2000.

7. Nuclear Regulatory Commission (U.S.) (NRC). NUREG/BR-0058, Rev. 3, “RegulatoryAnalysis Guidelines.” (Final Report) NRC: Washington, DC. June 2000.

8. Nuclear Regulatory Commission (U.S.) (NRC). SECY 01-0218, “Update of the Risk-Informed Regulation Implementation Plan.” NRC: Washington, DC. December 5, 2001.

9. Letter, Dana A. Powers (ACRS) to William D. Travers (EDO, NRC), September 8, 2000.Subject: Proposed High-Level Guidelines for Performance-Based Activities.

10. Nuclear Regulatory Commission (U.S.) (NRC). “Staff Requirements - COMSAJ-97-008 -Discussion on Safety and Compliance.” NRC: Washington, DC. August 25, 1997.

11. Nuclear Regulatory Commission (U.S.), Washington, DC. “Combustible Gas Control inContainment,” Proposed Rule. Federal Register: Vol. 67, No. 149. pp. 50374-50383.August 2, 2002.

12. Nuclear Regulatory Commission (U.S.), Washington, DC. “Geological and SeismologicalCharacteristics for Siting and Design of Dry Cask Independent Spent Fuel StorageInstallations and Monitored Retrievable Storage Installations,” Proposed Rule. FederalRegister: Vol. 67, No. 140. pp. 47745-47755. July 22, 2002.

13. Nuclear Regulatory Commission (U.S.)(NRC). NUREG-1753, “Risk-Based PerformanceIndicators; Results of Phase 1 Development.” NRC: Washington, DC. April 2002.

APPENDIX A

HIGH-LEVEL GUIDELINES FOR PERFORMANCE-BASED ACTIVITIES

A-1

APPENDIX A

HIGH-LEVEL GUIDELINES FOR PERFORMANCE-BASED ACTIVITIES

The U.S. Nuclear Regulatory Commission(NRC) staff developed, and provided to theCommission and the public, high-levelguidelines to identify and assessperformance-based activities in SECY-00-191,dated September 1, 2000 (Ref. A-1). Theseguidelines were developed consistent with thedirection in the Staff RequirementsMemorandum (SRM) to SECY-99-176, “Plansfor Pursuing Performance-Based Initiatives”(Ref. A-2). The staff made presentations onthe guidelines to the Advisory Committee onReactor Safeguards (ACRS) and the AdvisoryCommittee on Nuclear Waste and providedthem for information to the AdvisoryCommittee on Medical Use of Isotopes. OnSeptember 8, 2000, the ACRS (Ref. A-3)provided conclusions and recommendationsthat included the following:

• The guidelines should explicitly state thatthe performance levels and reliabilityparameters should be set at the highestpractical level.

• Guidance should be given on the extent to

which multiple performance parametersthat provide redundant information shouldbe used to satisfy the defense-in-depthphilosophy.

The staff has determined that, in order to meetthese recommendations, the guidelines needto be modified and the process that is followedto implement the guidelines must have thefollowing guidance:

1. A hierarchical structure of goals andobjectives should be developed thatreflects the decisionmaking values of aregulatory issue so that the term “highestpractical level” can be interpretedobjectively.

2. The development of redundant informationto satisfy the defense-in-depth philosophyrequires that the association ofperformance parameters with objectives

should be represented in a formal mannerso that attributes such as “prevention” and“mitigation” are explicitly treated.

The re-formulated high-level guidelines forperformance-based activities are presentedbelow in two versions. The first version is theformal one that maintains similarity andconsistency with the guidelines provided to theCommission in SECY-00-191 and published inthe Federal Register at 65 FR 26772(Ref. A-4). The only changes made are thoseneeded to implement the ACRSrecommendations. The second version is aplain-English rendition of the high-levelguidelines that relates more directly to thestepwise process described in Section 3.

Version 1 (Formal)

I. Viability Guidelines

A. A framework exists or can be developedto show that performance by identifiedelements will serve to accomplish desiredgoals and objectives. Margins ofperformance exist such that ifperformance criteria are not met, animmediate safety concern will not result.

(1) An adequate safety margin exists.

(2) Time is available for taking correctiveaction to avoid safety concerns.

(3) The licensee is capable of detectingand correcting performancedegradation.

B. Measurable, calculable, or constructableparameters to monitor acceptable plantand licensee performance exist or can bedeveloped.

(1) Directly measured parameters relatedto the safety objective are preferredand will typically satisfy this guideline.

A-2

(2) Calculated or constructed parametersmay also be acceptable if there is aclear relationship to the safetyobjective.

(3) Parameters that licensees can readilyaccess, or are currently accessing, inreal time are preferred and willtypically satisfy this guideline.Parameters monitored periodically toaddress postulated, design basis, orother conditions of regulatorysignificance may also be acceptable.

(4) Acceptable parameters will beconsistent with defense-in-depth anduncertainty considerations.

C. Objective criteria to assess performanceexist or can be developed.

(1) Objective criteria consistent with thedesired outcome are establishedbased on risk insights, deterministicanalyses, and/or performance history.

D. Licensee flexibility in meeting theestablished performance criteria exists orcan be developed.

(1) Programs and processes used toachieve the established performancecriteria will be at the licensee’sdiscretion.

(2) A consideration in incorporatingflexibility to meet establishedperformance criteria will be toencourage and reward improvedoutcomes, provided inappropriateincentives can be avoided.

II. Assessment Guidelines

A. Maintain safety and protect theenvironment and the common defenseand security.

(1) Safety considerations play a primaryrole in assessing any change arisingfrom the use of performance-basedapproaches.

(2) Adequate safety margins aremaintained using realistic safetyanalyses, including explicitconsideration of uncertainties.

B. Increase public confidence.

(1) An emphasis on results and objectivecriteria (characteristics of aperformance-based approach) canhelp NRC to be viewed as anindependent, open, efficient, clear,and reliable regulator.

(2) A performance-based approach helpsprovide the public clear and accurateinformation about, and a meaningfulrole, in the regulatory programs.

(3) A performance-based approach helpsexplain NRC’s roles andresponsibilities and how publicconcerns are considered.

C. Increase effectiveness, efficiency, andrealism of the NRC’s activities anddecisionmaking.

(1) The level of conservatism existing inthe currently applicable regulatoryrequirements would be assessed,considering analysis methodology andthe applicable assumptions. Anyproposal to use realistic analysiswould take into account uncertaintyfactors and defense-in-depth relativeto the scenario under consideration.

(2) The performance criteria and the levelin the performance hierarchy at whichthey have been set would beassessed. In general, performancecriteria would be set at a levelcommensurate with the function beingperformed. In most cases,performance criteria would beexpected to be set at the system levelor higher.

A-3

D. Reduce unnecessary regulatory burden.

(1) A performance-based approachenables the NRC to impose regulatoryburden that is commensurate with thesafety benefit and that effectivelyfocuses resources on safety issues.

(2) A performance-based approach willenable the costs associated with NRCactivities to States, the public,applicants, and licensees to befocused on areas of highest safetypriority and avoid burden imposed byoverly prescriptive regulatoryrequirements.

E. The expected result of using aperformance-based approach is an overallnet benefit.

(1) A reasonable net benefit test beginswith a qualitative approach to evaluatewhether there is merit in changing theexisting regulatory framework. Whenthe net benefit test is approached fromthe perspective of existing practices,stakeholder input would be sought.

(2) Unless imposition of a safetyimprovement or other societal benefitis contemplated, expending resourcesfor a change in regulatory practicewould be justified only if NRC orlicensee operations benefit from sucha change. Licensees themselves willbe the primary source of initialinformation and feedback regardingpotential benefits.

(3) For the limited purpose of screeningpotential performance-based changes,consideration of a specific result (suchas net reduction in worker radiationexposure) may be sufficient forweighing the immediate implications ofa proposed change.

F. The performance-based approach can beincorporated into the regulatoryframework.

(1) The regulatory framework may includeregulation in the Code of FederalRegulations, the associated regulatoryguides, NUREGs, standard reviewplans, technical specifications, andinspection guidance.

(2) A feasible performance-basedapproach would be directedspecifically at changing one, some, orall of these elements.

(3) The proponent of the change to theelements of the regulatory frameworkwould be responsible for providingsufficient justification for the proposedchange; all stakeholders would havethe opportunity to provide feedback onthe proposal, typically in a publicmeeting.

(4) Inspection and enforcementconsiderations would be addressedduring the formulation of regulatorychanges rather than afterwards. Suchconsiderations could include reducedNRC scrutiny if performance sowarrants.

G. The performance-based approach wouldaccommodate new technology.

(1) The incentive to consider aperformance-based approach mayarise from the development of newtechnologies, as well as difficulty infinding spare components and partsfor existing technologies.

(2) Advanced proven technologies mayprovide more economical solutions toa regulatory issue withoutcompromising safety, hence justifyingconsideration of a performance-basedapproach.

III. Guidelines for Consistency withRegulatory Principles

A. A proposed change to a moreperformance-based approach isconsistent and coherent with other

A-4

overriding goals, principles, andapproaches in the NRC’s regulatoryprocess.

(1) These principles are provided in thePrinciples of Good Regulation (Ref. A-5); the Probabilistic Risk AssessmentPolicy Statement (Ref. A-6);Regulatory Guide 1.174, “An Approachfor Using PRA in Risk-InformedDecisions on Plant-Specific Changesto the Licensing Basis” (Ref. A-7); andthe NRC’s Strategic Plan (Ref. A-5).

(2) Consistent with the high level at whichthe guidelines have been articulated,specific factors that need to beaddressed in each case (such asdefense in depth and treatment ofuncertainties) would depend on theparticular regulatory issues involved.

Version 2 (Plain English)

I. Viability Guidelines (Can a performance-based approach be developed?)

A. Estimate margin realistically, usingquantitative and/or qualitative factors. Isthe margin robust, or is it sensitive tounpredictable factors? Is there time to takecorrective action if expectations regardingthe regulatory issue or the licensee are notborne out? Can the affected licenseesreasonably be expected to reactappropriately if surprises occur?

B. Are there quantitative or qualitativeparameters, using natural or constructedmeasures, that can be observed and thatcan promptly reveal an unacceptablereduction in margin? Does the observationof these parameters support the objectivesof defense in depth and control ofuncertainty?

C. Can objective criteria, consistent with thedesired outcome, be established based onrisk insights, deterministic analyses,and/or performance history?

D. Is the flexibility afforded to licensees in theexisting regulatory framework consistentwith the realistic margin estimate? Can achange in the level of flexibility be effectedin a way that would encourage and rewardimproved outcomes, while avoidinginappropriate incentives?

II. Assessment Guidelines (Is it worthwhile todevelop a performance-based change?)

A. Can a performance-based change to theregulatory framework fulfill theperformance goal of “Maintain safety andprotect the environment and the commondefense and security”?

B. Can a performance-based change to theregulatory framework fulfill theperformance goal of “Increase publicconfidence”?

C. Can a performance-based change to theregulatory framework fulfill theperformance goal of “Increaseeffectiveness, efficiency, and realism ofthe NRC activities and decisionmaking”?

D. Can a performance-based change to theregulatory framework fulfill theperformance goal of “Reduceunnecessary regulatory burden”?

E. Does implementing a performance-basedapproach yield a net benefit?

F. Where does the performance-basedchange best fit into the regulatoryframework? What are the effects oninspection and enforcement functions?

G. Is accommodation made to employ thebest available technology, now and infuture?

III. Guidelines for Consistency withRegulatory Principles

A. Is any part of a performance-basedapproach inconsistent with basicregulatory principles?

A-5

REFERENCES

A-1 Nuclear Regulatory Commission (U.S.) (NRC). SECY 00-191, “High-Level Guidelines forPerformance-Based Activities.” NRC: Washington, DC. September 2000.

A-2 Nuclear Regulatory Commission (U.S.) (NRC). SECY 99-176, “Plans for PursuingPerformance-Based Initiatives.” NRC: Washington, DC. July 12, 1999.

A-3 Letter, Dana A. Powers (ACRS) to William D. Travers (EDO, NRC), September 8, 2000.Subject: Proposed High-Level Guidelines for Performance-Based Activities.

A-4 Nuclear Regulatory Commission (U.S.), Washington, DC. “Revised High-LevelGuidelines for Performance-Based Activities,” Proposed Rule. Federal Register: Vol. 65,No. 90. pp. 26772-26776. May 9, 2000.

A-5 Nuclear Regulatory Commission (U.S.) (NRC). NUREG-1614, Vol. 2, Parts 1 & 2,“Strategic Plan Fiscal Year 2000 - Fiscal Year 2005.” (Final Report) NRC: Washington,DC. September 2000.

A-6 Nuclear Regulatory Commission (U.S.), Washington, DC. “Use of Probabilistic RiskAssessment Methods in Nuclear Activities: Final Policy Statement.” Federal Register:Vol. 60, No.158. pp. 42622–42629. August 16, 1995.

A-7 Nuclear Regulatory Commission (U.S.) (NRC). Regulatory Guide 1.174, “An Approachfor Using PRA in Risk-Informed Decisions on Plant-Specific Changes to the LicensingBasis.” NRC: Washington, DC. July 1998.

APPENDIX B

SUPPLEMENTARY GUIDANCE FOR MORE COMPLEX ISSUES

B-1

What is an “objectives hierarchy?”

An objectives hierarchy is a diagramrepresenting the relationships anddependencies between goals, top-levelfundamental objectives, lower-levelfundamental objectives, and meansobjectives. Fundamental objectives areends in themselves; means objectives arethings that are desirable because theysupport fundamental objectives. Anexample of a goal is “protection of thehealth and safety of the public;” anexample of a fundamental objective is“protection of the public from excessiveradiological exposures;” and an example ofa means objective is “reliability of safetysystems.” These are examples from thestructure of the Reactor Oversight Process(ROP), which makes use of an objectiveshierarchy.

APPENDIX B

SUPPLEMENTARY GUIDANCE FOR MORE COMPLEX ISSUES

B.1 Background

In some cases, the general guidance offeredin Section 3 of this document is insufficient,because the treatment of the regulatory issueis affected by one or more of the following:

• complexity• uncertainty• multiple objectives, especially competing

objectives• different stakeholder perspectives

When these conditions are present, a moreconsidered approach, based on decisionanalysis, is warranted (Ref. B-1). Theapplication of this type of analysis to nucleartechnology has recently appeared in thetechnical literature (Refs. B-2, B-3, B-4). Thepurpose of this appendix is to discussselected elements of such an approach.

As mentioned in Section 3, the purpose of thisdocument is to support the development of aperformance-based regulatory alternative, ifone is appropriate. Generally, a performance-based regulatory alternative needs to:

1. allocate performance across relevantfunctions, systems, or barriers, in order toassess whether the target safetyobjectives are satisfied

2. then implement that allocation ofperformance which entails identifying thesteps to be taken by licensees and/or NRCto make the performance allocation “cometrue” in practice

Part of implementation is confirmation ofongoing performance.

Accordingly, this appendix is aimed at a morerigorous development of Step 2 in Section 3,which called for “identifying the safetyfunctions.” Depending on the individual case,this level of rigor may vary. For example,where an issue affects many areas and

different objectives, this added complexitymay justify the explicit development of a moredetailed objectives hierarchy (see text box).

With an objectives hierarchy in hand, itbecomes easier to assess the levels ofperformance needed from each element. And,in conjunction with the viability guidelines, onecan determine those cases in whichperformance can appropriately be monitoredthrough inspections or performance indicators,and where it cannot, which would suggest thata more prescriptive approach is warranted.

The approach discussed below is intended topromote adequate coverage of keyperformance areas with performancemeasures in a way that qualitativelyaddresses defense-in-depth considerations.

B-2

What value does an objectives hierarchyadd to the process of formulating aregulatory alternative?

Formulation of a performance-basedalternative is supported directly by anobjectives hierarchy. First, systematicconsideration of the objectives hierarchyfosters completeness in the set ofperformance areas considered. Second,the hierarchy structure segments theobjectives that are more suitable forperformance-based and prescriptivealternatives which typically correspond tothe upper and lower levels of the hierarchy,respectively. Finally, in the allocation step,the objectives hierarchy helps to keep trackof the satisfaction of each safety objectiveat each level in the objectives hierarchy.

B.2 Development of the ObjectivesHierarchy

Decisionmaking, and in particular thedevelopment of performance-basedapproaches to regulation, conventionallybegins with clarification of the objectives(Refs. B-1, B-4). This is especially importantwhen competing objectives are involved as isoften the case in regulatory decisionmaking. Acommon instance of competing objectivesoccurs when there is a need to weigh safetybenefits against burden (unless the issue isrelated to adequate protectionconsiderations). In the present case, tworeasons exist for emphasizing thisconsideration. First, the viability guidelinesfocus on parameters in terms of which criteriacan be specified; a clear picture of the desiredfunctional attributes is necessary in order tocarry out this step properly. Second,determining these objectives also establishesthe scope of the net benefit test applied underthe assessment guidelines, which constitutesthe second step in a three-step application ofthe high-level guidelines.

Even if an issue is not concerned with a safetyenhancement (that is, the current level ofsafety is considered appropriate), regulatoryproposals to resolve the issue must addresssafety so as to confirm that it is beingmaintained. Alternatively, if an issue initiallyinvolves only safety, regulatory proposals toresolve it must address burden, publicconfidence, and efficiency and effectiveness(except when the issue is deemed to be oneof adequate protection).

Figures B-1 through B-3, based in part on theROP (Ref. B-5), illustrate concepts of anobjectives hierarchy. Figure B-1 shows thegoal, fundamental objectives, and meansobjectives of an objectives hierarchy. FigureB-2 presents more detail of the ROP. Thecornerstone areas identified in Figure B-2 areintended to be a complete set of keyperformance areas affecting safety. The keyattributes identified within each cornerstoneare likewise intended to be a complete set.Completeness is one of the reasons to pursuesuch a systematic development. In Figure B-2,

consideration of the different cornerstoneareas also illustrates how the implicit underlying allocation of performanceaddresses defense-in-depth at a high level.Balance between prevention and mitigation isshown by the presence of cornerstonesaddressing initiating events, mitigatingsystems, and emergency preparedness; theadditional consideration of barrier integrityfurther reinforces defense-in-depth.

Analogous to logic tree development, eachlevel of the objectives hierarchy is derivedfrom the level above by decomposing eachnode into constituent elements. Each meansobjective relates to an objective above it onthe hierarchy, in that it answers the question,“How is the higher-level objective to beaccomplished?” (Question: How will safetyfunction X be accomplished? Answer: Byreliable function of systems A, B, and C.) Infact, a system reliability model developedhierarchically and expressed in “successspace” is essentially a partial objectiveshierarchy. It is “partial” because it addressesonly safety performance, and because, evenwithin safety, a logic model does not usuallyaddress cross-cutting programmatic issues ofthe sort that appear at the bottom ofFigure B-3.

B-3

��

��

Goal FundamentalObjectives

MeansObjectives

Figu

re B

-1O

verv

iew

of O

bjec

tives

Hie

rarc

hy

Key

Attr

ibut

es

Perf

orm

ance

Elem

ents

Cor

ners

tone

s

Stra

tegi

cPe

rfor

man

ceA

reas

PUB

LIC

HEA

LTH

AN

D S

AFE

TY A

S A

RES

ULT

OF

CIV

ILIA

N N

UC

LEA

R R

EAC

TOR

OPE

RA

TIO

N

B-1 Overview of Objectives Hierarchy

B-4

PUB

LIC

HEA

LTH

AN

D S

AFE

TYA

S A

RES

ULT

OF

CIV

ILIA

NN

UC

LEA

R R

EAC

TOR

OPE

RA

TIO

N

REA

CTO

RSA

FETY

RA

DIA

TIO

NSA

FETY

SAFE

GU

AR

DS

Stra

tegi

cPe

rfor

man

ceA

reas

Cor

ners

tone

sIN

ITIA

TIN

GEV

ENTS

MIT

IGA

TIO

NSY

STEM

SB

AR

RIE

RIN

TEG

RIT

YEM

ERG

ENC

YPR

EPA

RED

NES

SPU

BLI

CO

CC

UPA

TIO

NA

LPH

YSI

CA

LPR

OTE

CTI

ON

HU

MA

NPE

RFO

RM

AN

CE

SAFE

TY C

ON

SCIO

US

WO

RK

ENV

IRO

NM

ENT

PRO

BLE

MID

ENTI

FIC

ATI

ON

AN

DR

ESO

LUTI

ON

��Fu

ndam

enta

l Obj

ectiv

es ��M

eans

Obj

ectiv

es

Figu

re B

-2R

eact

or O

vers

ight

Pro

cess

Obj

ectiv

es H

iera

rchy

B-2 Reactor Oversight Process Objectives Hierarchy

B-5

Miti

gatio

n Sy

stem

s Per

form

ance

Func

tion

Syst

em

Trai

ns

Com

pone

nts

Hum

an A

ctio

ns

Line

Sup

ervi

sion

Man

agem

ent

Trai

ning

/ V

alue

sEn

gine

erin

g Su

ppor

tH

uman

Fac

tors

Engi

neer

ing

Proc

edur

esPr

ogra

mm

atic

Act

iviti

es

Cor

ners

tone

Lev

elObjectives Hierarchy

Figu

re B

-3M

ore

Det

aile

d D

ecom

posi

tion

of M

eans

Obj

ectiv

es

B-3 More Detailed Decomposition of Means Objectives

B-6

Figure B-3 illustrates levels of a hierarchyapplicable to many issues involving safetyassessments. A complete and explicitdevelopment of all hardware and programsinvolved in accomplishing high-level safetyfunctions would be a significant undertaking.The point of the present guidance is not tomandate such an explicit, detailed, andlaborious development, but to point out howthe concepts can apply even if thedevelopment exists only in the abstract or in aqualitative way. The value of this construct willbe clearer in light of the discussion in thefollowing subsection.

B.3 Allocation of Performance

Before selecting performance measures, it islogically necessary to determine what kind ofperformance and what level of performance isneeded from each performance area. Forexample, in the reactor arena, each of thecornerstone areas (initiating events, mitigatingsystems, barrier integrity, emergencypreparedness) receives attention.Performance is expected in each cornerstonearea. Strong performance in all areas providesan important defense-in-depth component,because to some extent, performance in onearea can compensate for lack of performancein another. For example, an increase ininitiating events frequency will not typically bea safety issue, if the mitigating systems’performance is satisfactory. This approach isfully consistent with the Commission’s WhitePaper definition of defense-in-depth, whichstates:

Defense-in-depth is an element of theNRC’s Safety Philosophy that employssuccessive compensatory measures toprevent accidents or mitigate damageif a malfunction, accident, or naturallycaused event occurs at a nuclearfacility. The defense-in-depthphilosophy ensures that safety will notbe wholly dependent on any singleelement of the design, construction,maintenance, or operation of a nuclearfacility. The net effect of incorporatingdefense-in-depth into design,construction, maintenance, and

operation is that the facility or systemin question tends to be more tolerantof failures and external challenges(Ref. B-6).

Generally, it is desirable to specify andmonitor performance targets as high on theobjectives hierarchy as possible, consistentwith the viability guidelines. Allocatingperformance too far down on the hierarchyreduces licensee flexibility. Arriving at animplementation that maintains safety, whileappropriately balancing licensee flexibility withthe need for regulatory assurance of ongoingperformance, will require some iteration withthe allocation step.

B.3.1 Example

In order to understand what it means toallocate performance, and why performancemay need to be allocated, consider thefollowing example. The quantitative analysispresented in the Reactor Safety Study (Ref. B-7) showed that resources were allocated in animbalanced way. Much attention hadpreviously been focused on plant responseconditional on a design-basis loss-of-coolantaccident, while somewhat less attention hadbeen focused on decay heat removal inscenarios initiated by transients or smallbreaks. This was addressed in a TMIrequirement (Ref. B-8), as a result of whichauxiliary feedwater systems (AFWS) inpressurized-water reactors (PWRs) licensedduring the 1980s were expected todemonstrate low unreliability “in the range of1E-4 to 1E-5 per demand” (Ref. B-9).Assuming reasonable initiating eventfrequencies, satisfaction of this requirementwould mean that a certain class of accidentsequences was being controlled reasonablywell.

This is a specific example of an allocation.Performance in the initiating events area wastacitly allocated at then-current levels, whiletarget performance in AFWS reliability was setat a level somewhat better than was thenbeing achieved at certain plants. Note that thiswas a design and licensing issue, not a“performance” issue.

B-7

What do we mean by “allocation?”

Given a top-level safety objective (such asCDF, or dose limits, or frequency ofoverexposure), the “allocation” ofperformance among barriers, structures,systems, and components (SSCs), etc., isthe choice of performance targets set by aregulatory alternative in order to assure thatthe safety objective is met. A choice amongpossible allocations must often be made,because it may be formally possible tomeet a safety objective in more than oneway. However, cost and practicalitysignificantly restrict the possibilities thatneed to be considered.

Does allocation always require settingexplicit numerical targets?

No. But the history of the development ofregulations at NRC suggests that it ishelpful to work with performance targetsthat are at least implicitly numerical.Otherwise, it is difficult to determine howstringent the performance requirementsneed to be, or whether performancemonitoring is a viable option in some areas.

This example serves to illustrate the previouscomments regarding licensee flexibility. Referagain to Figure B-3. Some flexibility was madeavailable in the application of this requirementto AFWS systems, provided that alternativemethods of core cooling were shown to beavailable (e.g., feed and bleed). Thus, the realallocation was pitched at the functional level(decay heat removal), with alternativeallocations available at the systems level(allocate all performance to AFWS, oralternatively, allocate some to AFWS andsome to high-pressure injection and primarydepressurization).

Carried down to lower levels, theserequirements have implications for such areasas inservice testing and inspection, andtechnical specifications.

B.3.2 Process

A thought process analogous to the aboveshould be carried out, at least implicitly orqualitatively, whenever regulatoryrequirements are being formulated. Thestringency of requirements at a given levelneeds to be related to the top-level safetyobjectives being addressed.

The starting point is an allocation ofperformance at the upper levels of thehierarchy. This allocation should appropriatelybalance prevention and mitigation, whilesatisfying agency safety objectives. In thereactor arena, this top-level allocation stepfrequently begins with core damage frequency(CDF) and large-early-release frequency(LERF) objectives, and then infers practicallimits on the frequencies of various accidentsequence families from the heuristic guidelinesuch that no single sequence type dominatesrisk. This sort of insight can informperformance allocation over initiating eventtypes, mitigating systems performance, andemergency preparedness. At this point, the allocation continues bysegmenting each high-level performanceobjective into sub-objectives corresponding tothe performance elements on the adjacentlower level. The collective satisfaction of thesesub-objectives assures performance at the

higher level. This entails making choices,because this decomposition is not unique.This process is continued on down theobjectives hierarchy, until a point is reached atwhich it is no longer appropriate to articulateobjectives, because there would be no netbenefit to measuring performance at that level.The operational character of this guidancemeans that meaningful allocation must bedone iteratively with the selection ofperformance measures and theirimplementation.

In the reactor arena, because of variations inplant design, it will be difficult to perform thiskind of allocation generically below thefunctional level. As illustrated in the AFWSexample cited above, the requirement wasactually formulated at the functional level, withthe expectation that either the AFWS alonewould meet it, or additional systems would be

B-8

How does defense in depth affectallocation of performance?

Part of defense in depth is avoiding over-reliance on any one performance area insatisfying the top-level safety objectives.Therefore, defense-in-depth considerationsfavor allocations that spread outperformance over redundant and diversebarriers and SSCs.

What is the relationship between nominalperformance, performance criteria, andallocated performance?

The allocated performance level is the levelof performance that is needed in order tosatisfy safety objectives. In general, normalperformance will be significantly better thanthis level. In order for a performance-basedapproach to be viable, the performancecriterion needs to be set at a level havingsignificant margin to the allocated level, inorder for problems to be detected andaddressed before a safety issue arises. Inorder for this approach to be practical,there must also be some margin must alsoexist between the normal performancelevel and the criterion.

credited in a more flexible evaluation. On theother hand, system-level, train-level, andcomponent-level requirements clearly derivefrom an implicit allocation of performance thatis not numerically precise, but derives fromimplicit consideration of safety objectives.Improving the alignment of regulatoryrequirements with real safety objectives is theessence of “risk-informing” regulatory practice.Significant progress can be made in this areawithout overly detailed developments ofobjectives hierarchies or allocations.

B.4 Formulation of an AppropriateImplementation: SelectingPerformance Measures

Given an allocation, one must still decide onan implementation, i.e., a suitable combinationof regulatory requirements, inspections, andperformance indicators, collectively aimed atassuring performance at (or better than) thestated target levels. These elements thenneed to be captured within appropriate partsof the regulatory framework.It is useful to illustrate this point by continuingwith the AFWS example introduced above, inwhich a target range for unreliability wasgiven. Once that objective was fixed, verydifferent implementations could have beenchosen. The implementation actually selectedallowed considerable flexibility in design, andthis flexibility is reflected in the variety ofsystem topologies to be found among U.S.nuclear power reactors. However, theimplementation was not fully performance-based. Instead, the implementation followedthe traditional regulatory approach; systemreliability was achieved through numerousprescriptive American Society of MechanicalEngineers (ASME) code requirements ontesting and inspection, and TechnicalSpecifications addressed allowed outage timein the usual way. Essentially, the effect of thetop-level requirement on unreliability was firstto drive performance to a level better than thatof a typical, active two-train fluid system withno backup, and second, to provide improved

guidance on the formulation of the traditionalprescriptive requirements.The aim of thepresent guidance is to foster choices inimplementation that are as performance-based as appropriate, in light of the high-levelguidelines. Therefore, given the objectiveshierarchy and the allocation, one uses theobjectives hierarchy once again to selectpromising performance areas that are high upthe hierarch as possible. If it is impractical todetermine performance at a particular level,the next level down is tried. In the case of theAFWS, it is readily established that it isimpractical to measure unreliability at thesystem level, because too few systemdemands occur to support a usefulmeasurement. One must measure orprescribe at a level lower than that of theAFWS, and the complement of measures andprescriptive requirements must collectively

B-9

provide the needed assurance ofperformance.

The actual process may have to rely on someamount of trial and error or iteration, makingrepeated use of the viability guidelines. Tryingto work directly at the level of a highly reliablesystem will typically fail the test of viability,because risk-significant changes in theoperation of such a system cannot beidentified and addressed in a timely fashion.Therefore, the process must seek a level onthe objectives hierarchy at which performancecan be objectively determined throughprescriptive requirements, and/or confirmedeither by inspection or through directconfirmation of performance (performance intests or in actual demands). Once suchperformance elements have been identified,performance parameters and associatedcriteria can be determined, and the viabilityguidelines can be applied as discussed inSection 3. Note that the “criteria” discussed inthe viability guidelines are not the targetallocations discussed above, but thresholdswhose violation signals declining performancewell before a significant safety problem hasdeveloped, i.e., the allocated performancelevels are violated.

When moving down the hierarchy, looking fora level at which an explicit allocation can beaddressed either by monitoring or byprescriptive requirements, it is important torecognize that it is possible to underperform ata higher level, even if redundant elements atlower levels nominally succeed. To see whythis is true, return once again to the AFWSexample. Even if all trains are highly reliableand available, performance at the system levelcan suffer if (for example) the trains’unavailabilities are sometimes concurrent, orthe contributors to train unreliability includeshared hardware or common causecontributions affecting more than one train.Concurrent unavailability is addressed throughtechnical specifications, and the possibility ofcommon cause mechanisms is one reason forprescriptive requirements on testing andinspection. This example serves to illustratethe ineffectiveness of excessively prescriptiveapproaches without a disciplined analysis of

the outcomes and results sought to beachieved.

B.5 Summary

Three major activities have been described:

1. formulation of an objectives hierarchy2. allocation of performance3. selection of an appropriate complement of

prescriptive requirements andperformance measurements

The regulatory alternative developed by thisprocess is intended to have the followingproperties.

1. The top-level safety objectives aresatisfied by the performance levelstargeted by the performance allocation.

2. The performance allocation is consistentwith defense in depth: no singleperformance area is relied on excessively.

3. The complement of prescriptiverequirements and performance measuresalso reflects defense in depth, in that itsamples performance in enough areasand in sufficient depth to support a reliableassessment of safety performance.

4. Prescriptive requirements are imposedonly when considerations of viability or netbenefit preclude using performancemeasures.

5. Performance measures and prescriptiverequirements are implemented as high onthe objectives hierarchy as possible, inorder to maximize licensee flexibilityconsistent with the viability guidelines.

B-11

REFERENCES

B-1 Clemen, R. T. Making Hard Decisions. 2nd Edition. Belmont, California: WadsworthPublishing Company. 1996.

B-2 Weil, R. and G. E. Apostolakis. “A Methodology for the Prioritization of OperatingExperience in Nuclear Power Plants.” Reliability Engineering and System Safety.Vol. 74, pp. 23-42. Elsevier. 2001.

B-3 Apostolakis, G. E. and S. E. Pickett. “Deliberation: Integrating Analytical Results intoEnvironmental Decisions Involving Multiple Stakeholders.” Risk Analysis. Vol. 18, No. 5,pp. 621-634. Society for Risk Analysis. 1998.

B-4 Keeney, R. L. Value-Focused Thinking. Cambridge, Massachusetts: Harvard UniversityPress. 1992.

B-5 Nuclear Regulatory Commission (U.S.) (NRC). SECY 99-007, “Recommendations forReactor Oversight Process Improvements.” NRC: Washington, DC. January 8, 1999.

B-6 Nuclear Regulatory Commission (U.S.) (NRC). “Staff Requirements - SECY-98-144 -White Paper on Risk-Informed and Performance-Based Regulation.” NRC: Washington,DC. March 1999.

B-7 Nuclear Regulatory Commission (U.S.) (NRC). WASH-1400, NUREG-75/014, “ReactorSafety Study: An Assessment of Accident Risks in U.S. Commercial Nuclear PowerPlants.” NRC: Washington, DC. October 1975.

B-8 Code of Federal Regulations, Title 10, Energy, Part 50.34(f)(1)(ii)(A), “Contents ofApplications; Technical Information.”

B-9 Nuclear Regulatory Commission (U.S.) (NRC). NUREG-0800, “Standard Review Plan,”Section 10.4.9. NRC: Washington, DC. July 1981.

Guidance for Performance-Based Regulation...NUREG/BR-0303 Guidance for Performance-Based Regulation...

Documents

Transcript of Guidance for Performance-Based Regulation...NUREG/BR-0303 Guidance for Performance-Based Regulation...