ML20058E712

Transcript of 901030 Briefing on Nonprescriptive Nuclear Safety Regulation in Rockville,Md.Pp 1-111.Related Documentation Encl

ML20058E712
Person / Time
Issue date:	10/30/1990
From:	NRC COMMISSION (OCM)
To:
References
REF-10CFR9.7 NUDOCS 9011070288
Download: ML20058E712 (153)

Text

{{#Wiki_filter:- - i N 4d%AnW6%%%ed%%%%%%%df%dtffffg g g TPAMSMITTAL T0: Y Document Control Desk. 016 Phillips r T ADVANCED COPY TO: The Public Document Room l // / f/ '70 OATE: FROM: SECY Correspondence & Records Branch g 6 5 Attached are copies of a Commission meeting transcript and related meetinq g document (s). They are being forwarded for entry on the 04tly Accession L'st and g placement in the Public Document Room. No other distribution is recuented or required. Meeting

Title:

b(l/ A A x #tMA$AAlo'/Lu d ud a f 8 be fa*6jv.) p /MIo/'M Open )( Closed Meeting Date: / Item Description *: Copies Advanced DCS

• 8 to PDR C,p,gy, i, :.

1. TRANSCRIPT 1

1 75 ful/. del > J a lx A M k J' V-Adswe.s.)/-& ) i' /. [

i V

2. /0b b e rlA /;f k M d e> V

} l l c f it b 4 > ! bh W 00 3. j d = 4 L F ti. LV 5. !-i c i s. .op sv l

• PDR is advanced one copy of each document, two of each SECY paper.

re.!! Seanch files the original transcript, with attachments, without SECY papers. D Fo 1 9011070288 901030 I! 1 PDR 10CFR Una PT9.7 PDC ais asis uinilla trinnnNMMMMMMMNNNnnnNNNNNnnMNNnnnnMMNnpNnnMnnMMMNni ;

e .4 UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMIS SION I $k16I BRIEFING ON NONPRESCRIPTIVE NUCLEAR SAFETY REGULATION LOCitiO!!l ROCKVILLE, MARYl.AND h&IS -OCTOBER 30, 1990 .Pages: tii RAGES 4 NEAL R GROSS AND C0,, INC, covet Eremtras Asp t ANSCRl8ERS 1323 Rhode Island Avenue, Northwest Washington, D.C. 20005 (202) 234-4433' ) i -+

DISCLAIMER This is an unofficial transcript of a meeting of the United States Nuclear Regulatory Commission held on October 30, 1990. in the Commission's office at one White Flint North, Rockville, Maryland. The meeting was open to public attendance and observation. This transcript has not been reviewed, corrected or edited, and it may contain inaccuracies. The transcript is intended solely for general informational purposes. As provided by 10 CFR 9.103, it is not part of the formal or informal record of decision of the matters discussed. Expressions of opinion in this transcript do not necessarily reflect final determination or beliefs. No pleading or other paper may be filed with the Commission in any proceeding.as the result-of, or addressed to, any statement or argument contained herein, except as the Commission may authorize. HEAL h. GROSS COURT 999otT945 ANO TRANSCRittR$ 1833 RHoot ilLANO AVINUI, N.W. (POR) 234-4433 WA&MINGToN. D.C. 20005 (202) 232-6600 ........_.__....-_m,___

UNITED STATES OF AMERICA I NUCLEAR REGULATORY COMMISSION I BRIEFING ON NONPRESCRIPTIVE NUCLEAR SAFETY REGULATION i PUBLIC MEETING h Nuclear Regulatory Commission One White Flint North r ? Rockville, Maryland Tuesday, October 30, 1990 The Commission met in open session, pursuant to notice, at 10:00 a.m., Kenneth C.

Rogers, Commissioner, presiding.

1 COMMISSIONERS PRESENT: KENNETH C. ROGERS, Commissioner JAMES R. CURT 1SS, Commissioner FORREST J. REllTCK, Commissioner i NEAL R GROSS l COURT REPORTERS AND TRANSCR$ERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D C. 20005 (202) 2324400

2

e STAFF AND PRESENTERS SEATED AT THE COMMISSION TABLET l

SAMUEL J. CHILK, Secretary MARTIN MALSCH, Office of the General Counsel DOCTOR DAVID ROSSIN, President, Rossin and Associates PROFESSOR MICHAEL GOLAY, Massachusetts Institute of Technology MARSHALL BREGER, Chairman, Administrative Conference of the United States DOCTOR ROGER MATTSON, Vice President, Scientech, Inc. i NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE IDLAND AVENUE, N W. (202) F344433 WASHINGTON, D C. 20006 (202) 232 4600 i

y, 3 V. 1 P-R-0-C-E-E-D-I-N-G-S 2 10:00 a.m. i 3 COMMISSIONER ROGERS: Good morning, ladies 4 and gentlemen. 5 Chairman Carr will not be with us today. 6 He's been asked to attend a hearing of the House 1 7 Subcommittee on Oversight and Investigations of the } 8 Committee of Interior and Insular Affairs being 9 conducted this morning in Plymouth, Massachusetts. 10 COMMISSIONER CURTISS: I'm sure he'd 11 rather be with us. 12 COMMISSIONER ROGERS:

Well, he's I

13 representing us. 14 In the past year or so, the NRC has become i 15 increasingly interested in the concept of 16 nonprescriptive regulation. We've come to realize -17 that the prescriptive regulatory approach we've 18 traditionally applied and which appeared to~ be i .19 appropriate during the construction of nuclear power-l '20 plants' has deficiencies when the focus of regulation f 21 is on operations.- In fact, prescriptive regulation 1o 22 may have unintended negative effects. It may stifle m 23 innovation and divert attention f rom substance to 24 process. It may make it more difficult for a licensee 25 to use an approach optimized to its own culture and NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENVE, N W. (202) 2344433 WASHINGTON, D C. 20005 (202) 232 4600

4 1 circumstances. 2 Therefore, we've been giving more thought 3 and considering current rulemaking and other -4 regulatory activities to approaches that are less 5 prescriptive. These are boing considered in several 6 major Commission initiatives, including the 7 development of regulations for advanced reactor 8 designs and for plant maintenance. l 9 As we attempt to do so, we find there is a 10 large gap between a perception that nonprescriptive il regulation is good and actually what is involved in 12 implementing it. There appear to be very different ~ 13 concepts of what nonprescriptive regulation means and. 14 how-it can be applied. There is considerable ' 15 uncertainty over how the Agency can create regulations 16 that - are nonprescriptive, yet still ' demonstrably 17 provide adequate assurance of-safety and. appropriate 18 ways for us to track performance and take enforcement 19 actions where necessary. 20 Therefore, we.'re pleased today to hear a i 21 discussion of the concept of nonprescriptive nuclear 22 safety regulation by four distinguished individuals 1 23-from outside the Agency who can offer us several 24 different perspectives on what the possibilities for l l 25 nonprescriptive regulation are for the NRC. We will. l-NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS L 1323 RHODE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON. D C. 20005 (202) 232 6600

m j 5 1 1 hear first from Professor Michael Golay of the j i 2 Massachusetts Institute of Technology who's done 3 ~ considerable work in this area. We'll then hear from j 4 Mr. Marshall Breger, Chairman of the Administrative 1 5 Conference of the United States who has an overview of i 6 what is taking place in all the federal regulatory 7 agencies; Doctor David Rossin of Rossin Associates who 8 has worked closely with the licensee committee; and I 9 Doctor Roger Mattson, Vice President of Scientech, j 10 Incorporated, who was formerly an NRC staff member and 1 11 can therefore comment on this issue from both NRC and 12 external perspectives. i 13 Gentlemen, we would like to welcome all of 14 you-and to thank you for joining us today. I hope in 15 briefing us you will touch upon the scope of 16 applicability of - the nonprescriptive approaches you 4 e - 17 would like to see the Agency pursue and that you will 18 give us your views of both the benefits and the -l 19 potential drawbacks of a nonprescriptive approach. '20 copies of presentation slides used today 21 are available at=the entrance to the meeting room. 22 Professor Golay, I believe you're the [ J 23 first on the agenda. Would you please proceed. 24 PROFESSOR GOLAY: Thank-you, Mr. Chairman. 25 The things that I want to do today I NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON, D C. 20005 (202) 232 6600

6 l 1 believe are familiar to' you in part from material 2 which I've submitted previously where at MIT, since 3 about 1983, we've been engaged in research projects 4 focused on the so-called advanced reactors which 5 people hope to use in the next generation of power 6 plants. From tbst, we have done a lot of thinking l 7 about the role of the safety regulation system and 8 effects which it has on the directions which 9 technology. can take and. how good safety performance 10 can be achieved. 11 As I think you're aware, we also undertook 12 a project with NRC sponcorship during the late 1980s 13 examining one aspect of how nonprescriptive regulation 14 could be pursued. From that we gained enough j .15 confidence to suggest that the scope of that kind of 16 investigation should be broadened ~ to see if an 17 alternative practical - system could be implemented. 1: 18 What I'really want to--- am going to end up suggesting 1. 19 today is that given the time table for.the development I 20 of the - so-called. advanced concepts, that it is ~21 appropriate. f or the NRC to undertake some internal' l 22 action to ' investigate in a timely way whether. a l 23 nonprescriptive safety approach could be established 24 for use with these reactors. 25 What I'm going to do is follow the order NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W. (202) 234 4.33 WASHINGTON. 0.C. 20005 (202) 2324600

7 4 1 of the information packet which I've provided to you. I 2 So, I'd like to just ask you to follow along with me. 3 The first point I want to make concerns 4 the question which was raised in introduction, which 5 are the benefits and costs of nonprescriptive l 6 regulation. I would say that among the benefits are j 7 that it provides, at least potentially, a means of 8 escaping some of the limitations of the current '9 prescriptive approach. In this first page, I've 10 listed here some of them. 11 I would say that probably the most 12 important point is the second one where the current [ 1? approach has been shown through PRA investigations to 14 produce a non-uniform level _ of safety and, in fact, 15 one which is substantially uncertain. That is, for 16 example, one of the questions before the Commission 17 today is whether the current fleet of nuclear power 18 plants, 'in f act, satisfies the safe'ty goal where at 19 the moment it's ' impossible to provide a definitive 20 answer to that question. 21 So, the overall reason for wanting to j 22 undertake an alternative method is the hope that it 23 can provide a' means for producing more uniform and l L 24 higher levels of safety. t 25 The fourth' item is one which is also very NEAL R. GROSS ( CoVRT REPORTERS AND TRANSCRIBERS + 1323 RHoDE ISLAND AVENUE, N W. l. l: (20M 2344433 WASHINGTON, D C,20006 (202) 232 MOD I

8 1 important and tied to the presc"iptive approach to 2 regulation in that we have seen examples with the 3 large number of license holders for nuclear power 4 plant licenses in the U.S. of cases where the full 5 assumption of responsibility for safety has really not 1 6 been undertaken by license holders, in part because of 7 the very heavy burden of complying with the large 8 number of regulations which issue from the NRC. If we 9 go to performance based or nonprescriptive regulation 10 without detailed guidance from the commission 11 concerning how to satisfy every single item in detail, 12 suggestions about how to do this, we can expect a 13 clearer transfer of responsibility and authority for 14 actual safety performance. ^ 15 One aspect of the current system which has 16 been very important to us in examining technology-17 improvements is that the. current-system, f avors 18 imitation of what's been done before in both 19 technology and in terms'of methods and models. As a 20 consequence, it inhibits technological innovation, 21 both with existing plants in terms of makinct small S 22 improvements, and in considering the future 23 technology. 24 Finally, I'd like to really ask a question 25 to yourselves, which is that my observation in dealing ~ NEAL R. GROSS q CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON. D.C 20005 (202) 2324600 s

9 1 1 with the commission and the commissioners over many 2 years has been that there is a continual stream of 3 detailed questions which have to be resolved at the j J 4 highest level with the approach to regulation which we 5 have today, with the consequence that the resources t 6 for broad policy making are much harder to bring to 7 bear than if we had a system which could segregate 1 8 some of the detailed questions so that they could be 9 handled at lower levels in the organization. l 10 I'd like to turn to the next page and go i 11 to the need for nonprescriptive regulation. Here I [ i' 12 violated the guidance which I was issued by the staff 13 and there are two pages which are different from what i 14 I've given you before and they're before you. The 15 headings are all the same. i 16 But one of the main reasons why we feel 17 that nonprescriptive regulation is worthwhile is that -18 in the work that's gone on since the Three Mile Island 19 accident and-especially since the Chernobyl accident, 20 a major feature of this work has been to improve tho 21 safety of_ future reactor. concepts, especially both the avoidance and [ 22 concerning the mitigation 23-the mitigation of severe accidents. Today'we have a 24 void. .We have reactor designs which are initially l '25 being formulated with the goals which I just mentioned NEAL R. GROSS court REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE. N W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232-6600

10 i ', 1 in mind, yet the hoops that they must jump through in 2 order to obtain regulatory approval are still focused 3 primarily on the DBA level of thinking. 4 So, there's a real mismatch between what 5 these concepts are trying to achieve and the actual 6 challenges which they must satisfy in order to get an 7 operating license. I would argue that the potential 8 for a degradation of safety levels which could be 9 achieved is fairly strong here. 10 The third point on this page which I want 11 to bring attention to is that I think that any action 12 concerning nonprescriptive regulation really has to be 13 dealt with at the level of the commission. 14 I ci te here testimony presented by 15 Commissioner Palladino back in 1985 before the Marilyn 16 Lloyd Committee in which the logical necessity for 17 nonprescriptive regulation and the expectation that it 18 would be used in future reactors was stated and yet-- 19 and so, in a sense, there's been this recognition in 20 the Agency for a long time, but it doesn't gel into 21 practice. In dealing with the staff and the 22 Commissioners, I felt basically it requires a mandate 23 from the commissioners to have a focused effort to see 24 what would be practical in this area, that without 25 that nothing is going to happen. NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIDER$ 1323 RHODE ISLAND AVENUE, N W. 1 (202) 234 4433 WASHINGTON O C. POM (202) P324600 i - - _ _ - - - - _ _ - - _. _. - _ _ _ _ _ - _ _ _ - _ _ _ - _ - - - _ - _ - _ _ - - _ _ _. - _ _ -, - - - - - _ - - _ _ _ _ _ _ _ _ _ =

33 1 the bottom line on this figure is that 2 this is really the right time to do it. The agenda is 3 served up by new reactor concepts coming forward, but 4 only the Commission can really take action in this 5 way. So, that creates the reason for considering the 6 question here. 7 I'd like to turn to the next page. 8 What I've tried to do here is to summarize 9 the regulatory situation in which these new reactor 10 concepts are facing which are, if we take the light 11 water reactors, they're being basically designed 12 according to the prescriptive -- the large literature 13 of prescriptive regulation. There is negotiation 14 going on between NRC and EPRI in the context of the 15 requirements document which could perhaps alter whr.t 16 those regulations are. In the severe accident area, 17 at I've mentioned before, there is basically a V ACuum. 18 That is, there is not new guidance to the de61gners ~ 19 concerning how they can improve the designs and also 20 offering any kind of reward for safety improvements. 21 With the non-light water reactors, the 22 situation. is. even cloudier in that even an existing I 23 body of regulation other than that created by Fort St. '24 Vrain and Clinch River really doesn't exist. There 25 have been reviews within the Research Branch really NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON. O C. f.05 (m7 2324600

82 1 looking for vulnerabilities and show stoppers with 2 some of the concepts which have been offered, but 3 these are very far from constituting a body of 4 regulation which the designers can use in confidence. 5 In the same fashion the severe accident vacuum exista 6 for these concepts as well. 7 So, this leads to the next page where I 8 wish to make a suggestion, which is that the NRC 9 undertake an in',ernal project to investigate what 10 would be involved in a practical syrtem of 11 nonprescriptive regulation and, as the product of 12 this, to produce proposals to the Commissioners for 13 their consideration. I've outlined what I see as the 14 elements of the project's scope. I'm not going to 15 read them to you. I think that they are things which 16 most of you would create with ten minutes' 17 consideration as well. But it's essentially to create 18 some proposals, try them out, refine them and get 19 external review and present them for consideration. 20 There is one element though that we have 21 found to be very useful in working on questions of 22 this kind, which is what I'm mentioning at the bottom 23 of this page, which is a method of working. That is 24 what we have found in our efforts on related topics at 25 MIT is that working in parallel where in one part of NEAL R, GROSS CoVRT REPORT [RS AND TRANSCRfDERS 1323 RHODE ISLAND AVENVE, N W. (20?) 234 4433 WASHINGTON. D C. 20005 (202) 232 4600

I 13 1 your effort you worked on a formulation of a general 2 approach and in a second effort you try to apply that 3 general approach as a means of mutual refinement of 4 the two efforts tends to work very well. 5 So, what I would suggest in this case is J 6 to try to formulate a general approach to regulation 7 and apply it to one of the advanced reactor concepts 8 which has been proposed. For example, the gas reactor 9 concept has come in with some proposals of their own 10 where that could provide a starting point, if that 11 were desireable. Frankly, I don't think the choice of 12 the reactor concept is all that important, but I do 13 think that this mode of working actually is. 14 COMMISSIONER CURTISS: Before you go on, 25 l 1et me ask you -- 16 PROFESSOR GOhAY: Sure. 17 COMMISSIONER CURTISS: -- a question about 18 your first bullet on the adequacy of safety goals.. Do i 19 you envision there f ocusing on how t% safety goals 20 might be incorporated into a program like this'or is 21 the -- 22 PROFESSOR GOhAY: Yes. the use of the 23 COMMISSIONER CURTISS:' i 24 terminology " adequacy of safety g o als designed to s 25 focus on whether they're internally consistent and NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIDER$ 1323 RHoDE ISLAND AVENUE, N W. (202) P344433 WASHINGTON, D C. PC005 (202) 232 4600

14 I sufficient? 2 PROFESSOR GOLAY: No, no. I view the 3 safety goals as being effectively a vector or a set of 4 elements which express different safety concerne. The 5 question would be are they formulated in a way that 6 forms a good basis as the start of a regulatory 7 system? There may be things that you hasn't thought 8 of that will become revealed as you go into this 9 exercise. So, that's more the spirit in which I'm 10 thinking of it. 11 COMMISSIONER CURTISS: Okay. 12 PROFESSOR GOLAY:

However, I would say 13 that the safety goals, the use of the safety goals as 14 the fundamental element in terms of the performance 15 requirements that you would hope to get the reactors 16 regulated under this approach, that they play that 17.

role was the basis. 18 COMMISSIONER ROGERS: You would start with 19 them as they presently are stated, at least as -- 20 PROFESSOR GOLAY: Yes. 21 COMMISSIONER ROGERS: Not question that,. 22 per se, but whether there might - be another way of 23 using them. or making them more useful for this 24 purpose? Is that -- 25 PROFESSOR GOLAY: That's right. NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIDERS 1323 RHODE ISLAND AVENUE, N W. l (202) #34 4433. WASHINGTON, D C. 20005 (207) 7324400

15 1 COMMISSIONER ROGERS: Rather than the 2 fundamental numbers or concepts that are -- 1 3 PROFESSOR GOhAY: Exactly. But what we've 4 seen is the safety goals have evolved. The particular 5 codification changes about every three years as people 1 6 think more about it. So, we would expect more 7 refinement of that kind. 4 8 Moving to the next page, there's one thing 9 that is built into this approach, which is the use of 10 PRA. So, I'm trying to say something up front about 11 that, which is that, as you can read here, the safety l 12 goals are stated in terms of expected risks. So, this 13 implies some means of quantifying expected risk where f 14 the standard tool for this is probabilistic risk 15 analysis and there has been some reluctance to try to i 16 use probabilistic risk analysis in safety regulation 17 except at some distance from detailed questions. So, 18 I'm going to spend some time trying to focus on that 19 particular concern that people have. 20 .I'd like to turn to the next page. In our 21 work for the NRC, one of the things we examined is how 22 you.would use PRA in regulation. To get a little bit 23 ahead of the story on this page, the big concern that t 1-24 people have with PRA really comes in two ways. One is 25 that the methods and models used in it are still l NEAL R. GROSS l' COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. l (202) 234-4433 WASHINGTON, D C, 20005 (202) 232 4600 L

l'f 16 C 1 evolving. It's not an infantile analytic approach, 2 but it's one which hasn't reached its full maturity. 3

However, the aspect which I think causes greater 4

concern to people is the element of uncertainty, which 5 is recognized as being inherent in probabilistic 6 analyses. 7 The point that I want to make is that you 8 must consider both the result of any such analyses and 9 recognize up front that uncertainty is associated with 10 them and that at some means, I would propose through 11 conservative treatments, has to be also incorporated 12 to deal with uncertainties. You can't pursue the 13 lines which I'm suggesting without recognizing that 14 uncertainty exists and you've got to have some means 15 with safety factors and conservatism in taking them 16 into account. s 17 And on this page are listed some aspects .18 of what we see as being necessary elements in use of 19 PRA within the space where it can be used with 20 . acceptable accuracy. These are standardized NRC 21 approved models, databases and means of updating these 22 tools, but recognizing.that-refinement will just be a 23 way of; lif e in both the databases and the models as 24 our knowledge increases. 25 Second, means have to be worked out to NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D C. 20005 (202) 232 6600

87 1 make use of the results f rom these kinds of analyses, 2 not only for design but f or operation and reflecting 3 what we learn through the evolving industry A experience. These are al?. things which I think can be 5 done, but I'm going out that it will be a changing 6 tool which you would e::pect with an approach which is 7 maturing. 8 on the next page, I want to draw attention 9 to another of the concerns which people sometimes 10 have, particularly using PRA, which is the idea of 11 using rigid decision criteria, either go/no-go kinds 12 of things, for example, on the core damage frequency. 13 The issue has been raised, well, what if you end up 14 with an expected core damage frequency which is 15 1/100th less than a limit as opposed to being 1/100th 16 greater than a limit? Do you suddenly change your 17 answer, which in. the face of associated uncertainties 18 is'something which people aren't comfortable with? 19 The point that I want to make is that it 20 doesn't have to be that way, that there are judgmental-21 criteria which can be used which have bans for saying 22' yes or no to proposals where further investigation and 23 judgment is required. So, this was-one of the things 24 which might reflect back onto the safety goal, for 25 example, or subsidiary goals derived from them. 1 NEAL R. GROSS i CoVR1 REPORTERS AND TRANSCRIDERS 1323 RHoDE ISLAND AVENUE, N W. (202) P34 4433 WASHINGTON O C. 20005 (202) 232 M40

b^ ',. 18 i

Now, this is the second page which I've 2

changed from what I've given you and the revised page 3 is there because there's a typographical error in one 4 of the bullets, just a logical error. So, we'll pass 5 onto the next page. 6 What I've done 1 re is to try to give you 7 some examples of how a deci. ion making structuro could 8 be formulated. The point here is that the best 9 structure isn't really clear because the issue is how t-10 to handle uncertainty. So, I've outlined here three 11 with the expectation that what one would really do 12 would probably result as a blending of these proposals 13 or others based on the experience derived from the 14 investigation which I'm suggesting. 15 One way to do it would be to use 16 probabilistic risk assessment as the basis for 17 formulating a set of deterministic accident sequences 18 to be analyzed, in effect use;them as the basis for 19 extension of the design ' basis accident approach, but 20 always being able to go back to a higher level of 21 standards for insurance of consistency in the 22 formulation 'of whatever is done, and as a means of 23 always evaluating the marginal effect of any changes 24 which are introduced.. So, that would be one extreme, i 25 which is basically to use the safety goals and PRA as NEAL R. GROSS CoVAT REPORTTR$ AND TRANSCRIBERS h 1323 RHODE ($'.AND AVENUE, N W. l= . (202) 2344433 WASHINGTON. D.C 2000$ (202) 232 6600 i

19 1' I the basis for constituting a set of design basis 2 accidents going completely through the severe accident 3 space. 4 For example, EPRI, in their proposal in 5 the requirements document for the passive plants is in 6 effect making this kind of proposal where they're 7 saying that, "For severe accidents, let us take a B source term from the reactor coolant system coming 9 into the containment and then let us analyze the 10 containment performance in preventing whatever 11 material comes from the reactor coolant system and 12 getting into the biosphere, but where the source term 13 is a deterministically specified quantity." So, that 14 would be one example that would fit into this 15 particular suggestion. 16 COMMISSIONER ROGERS: Excuse me. Just a 17' point there. It's. been my opinion that -the 18-determini: tic approach to engineering evaluations 19 really in a sense does that intuitively. You start 20 off with a model and then you do a deterministic ~ 21 - ' calculation on that model. Where did you get the '22 model in the first place? Well, it came from some 23 intuitive ' $udgments. But it seems to me is what 24 you're suggesting here that that be a more formalized 25 probabilistic basis for determining the starting NEAL R. GROSS COURT REPORTERS ANr TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) ?)44433 WASHINGTON, D.C. 20005 (202) 2?2 4600 'I

20 1 point. 1 2 PROFESSOR GOLAY: That's right. That's 3 right. And the uncertainty that is, in effect, built 4 into the conservatisms in the cases you deal with is, 5 in fact, exactly the same uncertainty that makes 6 people nervous in using PRA. It's simply we don't 7 confront it quite so explicitly in the current 8 approach. 9 COMMISSIONER ROGERS: Absolutely. 4-10 PROFESSOR GOLAY: I think that that's 11 really an important point to remember, that is that we 12 have implicit assumptions concerning uncertainty 13 riddled throughout the current approach. There's 14 nothing different in going on nonprescriptive 15 regulation in terms of uncertainty except how explicit 16 you make your treatment. '17 Okay. I'm going to skip over the second 18 illustration and simply go-to the third which,is more 19 at the other limit, in the interest of time. 20 Another way to do this would be to use PRA 21 as the basic decision tool. That is, do a level 3 PRA 22' and then search through for design vulnerabilities, + 23 all the things which would make you doubt the result. 24 I've listed some of the more important factors here, 3 l 25 such as vulnerabilities to common mode failures, which NEAL R. GROSS CoVRT REPORTFHS AND TRANSCRIBERS 1323 RHOF4 ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON D.C. 20005 (207) 2324600

21 1 are very hard to analyze, or human failures and, in 2 effect, revise the expected risks on the basis of 3 conservatisms introduced to deal with those 4 uncertainties and then ask does the resulting design 5 satisfy the safety goals with all of those 6 adjustments. Where it doesn't, the applicant would be 7 told where the design is deficient and would be told 8 to go back and do more homework. 9 There's one item that I really want to 10 draw attention here to on the stage 3 where in 11 examining the regulatory systems in other countries, l 12 one of the things which we were impressed with was the 13 greater detachment in some cases, not in all, between 14 the regulator and the designer, where in effect the 15 responsibility of coming up with an acceptable fix to I 16 a deficiency was really the responsibility of the 1: 17 designer. This idea of having the regulator working 18 together to help him do this was, in fact, something 19 that was not part of the approach. On Jooking at it, 20 we think that that's actually a very important feature 21-to keep in. I bring it up because I know that this 22 kind of thing has been under discussion with the 1 23 Commissioners recently. j 24 The reason is to make very clear that the I 25 responsibility for whatever is proposed is the NEAL R. GROSS t l COURT REPORTERS AND TRANSCRIBERS 1323 RHoOE ISLAND AVENUE, N.W. (202) 234-4433 WASHINGTON, D.C. 20005 (202) 232 6600

28 1 responsibility of the designer or the operator. So, 2 there's no, in effect, co-option of the Agency by 3 implicit assumption of the responsibility for the 4 particular avenue which is selected. 5 COMMISSIONER ROGERS: There is a problem 6 though. That is that if the regulator is trying to 7 set these objectives totally independently of what is 8 achievable, what is technically achievable, there can 9 be a disconnect there that what is expected just 10 simply can't ever be arrived at. 11 ' PROFESSOR GOLAY: That's right. '12 COMMISSIONER ROGERS: State of the art 13 doesn' t perr.it it. 14 PROFESSOR GOLAY: That's right. And this 15 implies, I think, some means of technical appeal which 16 you could see how you could have because that's always 17: a possibility. However, you also have the discipline 18 introduced by the top level performance goals. In 19 effect, the real question is can the design satisfy .20 those goals? That's the kind of argument that has to 21 be offerod and it can be offered _to more than just one 22 staff member, if the right decision structure isn't 23 set up. 24 COMMISSIONER REMICK: I saw what you said, 25 Mike, in a slightly different way. I thought that NEAL R. GROSS I COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (P02) 2344433 WASHINGTON, O C. 20005 (202) 232-6600

4-e n :_ 23 Y. _4 1 you you were not precluding in my mind the 2 opportunity for the designer and the regulator to get 3 together to establish a reasonable performance 4 standard, but when it came to specifically how do you 1 c 5 meet that standard, that that be the responsibility of 6 the designer and not the regulator. 7 PROFESSOR GOLAY: Exactly. 4 8 COMMISSIONER REMICK: I read into your 9 words that viewpoint, but I must admit I was reading 10 into it. 11 PROFESSOR GOLAY: No, that's exactly 12 right. 13 COMMISSIONER REMICK: You would not see 14 anything wrong with them working on a technical. basis >15 to establish reasonable performance standards? 16: PROFESSOR GOLAY: That's right. Yes, I i 17 ' agree. 18 On the next-pace, I've listed a number of 19; practical problems which arise here and I'm sure that l 20 this list'isn't encyclopedic, it's merely impressively 21 long, which have to be addressed in working out any 1 22 kind of practical system. 'I'm not going to read these P ~23 to you. I think that they ' re. almost all self-1: l 24 ' explanatory. 25 COMMISSIONER REMICK: I'm not sure I NEAL R, GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. 3 (202) 234 4433 WASHINGTON, D.C. 200r? (202) 232-6600

f 34 W, 1 understood what the-third bullet meant. You seem to 2 be questioning the philosophy of defense in depth, but 3 maybe I just' misunderstood. [. L 4 PROFESSOR GOLAY: No, no, no. The sense-5 was-that if you recognize the need for defense in 6 depth,,how do you avoid a situation where the where you end up with a design where the 7-overall 8 overall performance goals are met with very heavy 9 reliance on one single design. feature. For - example, 10 with the gas reactor today, we' have that with the 11 reliance on very tough fuel providing it's fabricated 12: to high quality standards. ' 13 The thing that I think you have to 14 recognize is that defense in depth, the use of i 15 containment for example, arises basically as a 16 treatment of uncertainty. That is, because of some of 17. the things that you simply cannot lay away. So,- the 18 question that has 'come up previously. in discussions - 19 with the staff-is, "Well, if you go this way, how do 20 ' we.also make - sure that defense in depth is insured?" 21 'My answer to it is the way you do it is to say that 22 it's. necessary as a means of controlling-the 23 implications of uncertainty. But I think that it's 24 .something we have to have. F 25 COMMISSIONER REMICK: Another way I have NEAL R. GROSS COURT REPORTERS AND TRANSCRIDERS 1323 RHODE ISLAND AVENUE, N.W. .-(202) 2344 433 WASHINGTON. D.C. 20005 (202) 232 4600

25 '1' often looked at defense in depth, it does not require 2-absolute perfection in any particular means of defense 3 because you have an alternate in case that one fails, 4 -so.you don't have to have complete assurance that the 5 -one defense system will work because you have a 6 backup. 7 PROFESSOR GOLAY: Right. a 8 COMMISSIONER REMICK: That's another way-9= of saying the same thing. 10 PROFESSOR GOLAY:

Yes, I would agree with 11 that.

3 12 Yes, the one bullet that I want to draw 13 your attention to is the third from the' bottom. That 14 is how would-the NRC oversight for inspection.and 15 enforcement-change under nonprescriptive regulation. 16' I focus on this because I was-struck by a letter from

17 Larry Minnick recently concerning the - business of 3

i 18. going beyond the level.of diminishing. returns. 19 One of-the problems which I. think-the s 20 Agency has continually had in~ facing-new questions is s 21' determin'ation of just ho'w much effort is really called 2'2~ for. I think one of the'-reasons why'this thing keeps ,23 coming up is that there isn't a global standard to go 24 to in looking for the marginal - ef f ect on safety of a .25 new approach, some new wrinkle ' in dealing with a NEAL R. GROSS CoVRT REPORTEdS AND TRANSCRIBERS 1323 RHoOE ISL AND AVENUE, N W. (202) 234-4433 WASHINGTON, D.C. 20005 (202) 232-6600 tu m am m

i 4 26 .l "'.3 1 particular issue or the need to formulate a new 2 policy. I think that if we had this-nonprescriptive l -I 3 approach in place, we would have a consistent standard 4 to which any new activity could then be judged in j S terms of. 6 So, I suggest that that's one place that I 7 might be relevant in terms of a current concern, but I' l 8 don't want to dwell on these practical aspects. I 9 t '. link they can all be dealt with, but some require 10 work. And I want to really wrap up because I've taken i 11 all my time. I 12 On the aspects of the suggested work, y 13 which is the next to the last page, I would.say that 14 the first thing that I've tried to convey is'that,the 15 best nonprescriptive approach isn't obvious, that it 16 takes some refinement to really come up with a -17 practical system that you can use and that that's what 4 i 18 really ought to:be done. 19 The third bullet is that the use of PRA is 20 often resisted because of concern about the 21 uncertainties. As I've said previously,. the 22 uncertainties are no different. They're determined.by 1 .23 . the : physics of the problem. It's only a question of 24 how you address them. 25 The last thing which is also important is NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232 4600

27-1 that'in principle with the nonprescriptive approach it 2 should: be possible to refine the regulations as we 3 learn more, one of the things which we have seen in p L 4 the absence of a global standard to appeal to is that 5 the current system is very hard to refine. We've seen 6 this, for example, in the emergency planning area = 7. where there have been demands from the industry to: k 8 revise the standard for the emergency planning zone. 9 We've seen it, I would suggest, in the ~ 10 impasse that developed over Seabrook and Shoreham 11' where, again without a global risk standard, it was 12 very hard to take some of the objections on some of f 13 .the outlying accident cases, like evacuating the beach 14 at Seabrook, and put that in a context for decision 15 making. Eventually the Agency did make its decisions, - 16 but it was a very hard and expensive process, as I'm ' 17 ' sure I don't need to tell any of you. 18-So, finally, I'd like to turn to the last 19 page which is sort of a recapitulation'on why I-think 20 we 'ought.to undertake the project which I'm-21 suggesting, which -is it promises greater and more 22 ' uniform safety which is fundamentally the goal of the .23 Agency. Hopefully it can do it in a way that doesn't 24 ' kill the industry, which I would suggest -is what's 25 happening today. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232-6600

28 1 One of the mandates of the NRC is not to 2 promote nuclear power, but implicit in the existence 3 of the Agency is also the obligation to provide i 4 acceptable safety in a way that doesn't kill the 'S -nuclear enterprise. I would suggest that the way it's 6 going.today as the outcome of many, many interactions 7 and decisions is effectively to put it into a death 8 ~ spiral. That's, I think, another reason for decisive 9 action. 10 We're in a situation today where there are 11 all kinds of reasons to-keep nuclear power available 12 as a vital option for the country. We had today, for 13 example, Japan announcing a new policy to - increase 14 their use of. nuclear power as a means of dealing with ~15 oil concerns and greenhouse concerns. 16' In the U.S. we're having a much harder-17-time formulating similar policies for which I think 18 there would be broad support. I think that we can 19 also get better safety performance by. making it clear. 20 that 'the responsibility and authority for producing

21-safety rests with the applicant and this system, I 22 think, will promote that.

It should be able to. 23 provide swifter and clearer regulatory decisions and 24 it should free the Commissioners up to focus more on 25 the policy level questions rather than on the very NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W. (202) 234 6 WASHINGTON, D.C. 20005 (202) 2324600

1 29 h i detailed issues which I believe constitutes much of 2 the agenda today. 1 3 Then finally I would suggest that it would 1 4 greatly. stimulate technological innovation which would 5 also give us a good result. 6 That's as much as I'll take time to say. t 7 Thank you. 8 COMMISSIONER ROGERS: Well, thank you very q 9 much. I think what we'll do is we'll hold our 1 10 questions until everyone has had

a. chance to make a 11 presentation and then we can go at you all and maybe 12 some of you will answer. some of our questions as we 13 proceed.

14 Next we'd like to hear from Mr. Marshall 15~ Breger. L16 MR. BREGER: Thank you very much, Mr. 17. Chairman, members ' of - the Commission.. I'm pleased to i 18 be invited to share my thoughts concerning the use of ' 19_ nonprescriptive regulations by federal-agencies. p L ~20 Let me take this opportunity to thank the -21 Commission for its continuing support of the 22 Administrative Conference and its. activities. Due to ' 23 p! the support of this Commission and a number of its l' 24 sister agencies, the Conference has been able-to 25-develop round tables, printed materials and training NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISt.AND AVENUE, N.W. (202) 234.4433 WASHINGTON, D.C. 20005 (202) 232-6600

30 1 programs in the area of alternate dispute _ resolution. 2 These initiatives:are helping reduce costs and-delays 3 associated with traditional adversarial proceedings. 4 I'd also like. to commend the Commission's 5 representative to the conference, General Counsel 6 William Parler, who.is a respected and diligent member 7 of our rulemaking committee. 8 Although I have read Professor Golay's 9 paper on nonprescriptive nuclear safety regulation, 10 the example of loss of off-site power, with interest, 11-I have a disclaimer at the outset. I am here as a .12 1 rank amateur. I think ignorance was my ticket to be 13 invited. I'm not a technical expert, nor do I profess 14 any keen insight into whether the Commission's 15 experience 'with regulation of loss of off-site power 16 is - typical _ or atypical of the issues commonly 17. addressed.in! this. Agency, t 18-I do note,

however, that there. is a 19_

. spirited debate concerning the use of performance -20 standards L-as a regulatory alternative to prescriptive -21 requirements which has been the subject of numerous 22 Law Review articles since the early 1970s when 23 _Congr e s s, began enacting them. as part of the 24' environmental statutes such as the Clean Air Act. 25 As Chairman of the Administrative NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 2344433 WASHtNOTON, D.C. 20005 (202) 2324600 l

( - 3 31 i 1 Conference of the United States over the last five c 2 -years, I have seen this issue debated'in a number of 3 contexts in a host of agencies and I hope my 4~ observations concerning those debates drawn from an 5 administrative law perspective will'be useful to you. 6 The Administrative Procedure Act, passed 7 in 1946 as a compromise between New Deal: advocates who i 8 made belief in the superiority of technical expertise 9 an article of faith, and' conservative opponents of y 10 administrative government who believed in the 11 superiority of generalized courts of -law to provide t 12 independent checks on agency experts whose decisions-D 13 were viewed as partisan advocacy. 14 In that same year, Congress established T 15 the. Atomic Energy Commission, the first federal agency v 16 authorized to engage in research, ' development and n }'- f' 17-industrial production. The LCommissi'on had the best 3 W 1 82 scientific, engineering and ; management skills. In h (19 short,- everything the ardent-advocate of technical .c 20 expertise 'in decision making could > wish for. In N \\ { j, ^. ti 21 addition, the AEC-had'the flexibility provided by the-122' newly-enacted APA, Administrative. Procedure Act, to

M g.

.23 promulgate rules and inf ormally ' resolve disputes, 124 Thirty years later, in the midst of charges that y o p: 25 reactor development had pushed aside siting and safety (i h . MEAL R. GROSS f m ^ COURT REPt RTERS AND TRANSCRIDERS 1323 RHODE ISLAND AVENUE. N W. l. H. (202) 234 4433 ' WASHINGTON, D.C. 20005 (202) 232 6600 o l: } '\\

ti 32 l.3,', 1l decisions of. immense public importance, the AEC's r 2 ' regulatory authority was separated and given, as you 3 know, to the newly created NRC and here we are. 4 I cite this short historical synopsis to 5 make a point about the first sentence in Professor 6 Golay's article, which states that current U.S. I 7-nuclear regulatory system has been criticized as 8' ' arbitrary, inconsistent and' highly legalistic. "he 'W 9 unstated inference is that an overly legalized s,ystem l 10; i s. somehow inevitably tied to arbitrary and' - 11 inconsistent regulation. The article further implies '-12

that nonprescriptive ' regulation could provide 'a 1

13 technically more acceptable situation, j x 44 -14 If nuclear regulation is arbitrary and: 15 inconsistent,-and it may well be, I would: submit that 16 it cannot be,so because of sn overly legalized system [ L17 -in, the_ traditional sense. The English; and American 18- ' common law system may be ponderous, it may be costly, i 19 it. may be - overly reliant on' adversarial procedures, 20 but it is not normally criticized for being arbitrary 21 nor' capricious. Indeed, all of what I just mentioned, + 22 cost,. length, adversariness, is designed-to prevent 23 capriciousness and arbitrariness. q ^24

Further, it's unlikely that nuclear 25 regulation is arbitrary and inconsistent because n

NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234-4433 WASHINGTON, D C. 20005 .(202) 232 4600

i f '). : "'.-{ 33 1 technical expertise has been given short shrift in the 2 a'dministrative. processes - the Agency must follow. In ~ 3 -fact, one could argue that the history of nuclear 4 regulation demonstrates the fallacy of New Deal 5 thinking concerning the superiority of technical 6 centered decision making over the constitutionally 7 mandated fact-based decision making by general' courts 8 of law which are bound by precedent'and are therefore 9. not arbitrary orl inconsistent, but have their own 10 f aults '. They're legalistic. 11 I make = these observations to establish a 12 context f or ' dis'cus sing what Professor Golay calls' 13 nonprescriptive. ~ regulation or, f unctional goals, or > i, 14 - what is termed performance standards ~ lin much.of the a 15 legal literature. These approaches are usually s

16 distinguished?from decision standards or prescriptive

,y ,i 17 regulations and have become increasingly popular in y ly l .18 regulatory contexts. M c& 19-The advantages of performance standards are said to,be that they leave regulate <t entities free 20. 21 to = choose.- or ' invent "least cost solutior:s." They are 1 i; "W!' 22 said to foster innovation, produce more flexible [" > 23 results-oriented policy, and to be less damaging? to y 24 competition in the free market. On the other hand, 25 performance standards are frequently harder to write, L NEAL R. GROSS COUNT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W, (202) 2344433 WASHINGTON, D.C. 20005 (202) 232-6600

34-l'- . harder to administer and may give competitive -2 advantages to larger and more' sophisticated firms who 3 have the capacity to navigate the system, l 4 Let me underscore this point. It is often 5 missed. The design flexibility afforded by 6 performance standards may or may not lead to more 7 efficiency and greater technological innovation. .I 18

think, in fact, it often does.

But there's a 9 correlative. It creates greater challenges to 10 regulatory review and compliance and likely more work 11 for the legal profession. Whether that's a cost or 12 not I leave to you. 13 . Evidence for this can be seen: in the 14 -degree to which performance standards are suspectable 15 to accurate fact-based decision making. Ultimately ~ L16 given-the litigious environment.. in which we live, Si performance standards will.be tested in sone form of 18 legal proceeding where the-decision maker will have to -19 determine whether the standard has been properly 20' applied.. How easy it is to adduce proven compliance -21 will be a major contributor to any analysis of whether 22 a prescriptive or performance standard is the most 23 cost effective or rather was the most cost effective 24 and consistent means of achieving a ;:ar ti cular 25 regulatory result. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE. N W. (202) 234 4433 WASHINGTON. D.C. 20005 (202) 232-6600

i 35 1 In evaluating this

issue, some 2

-misconceptions about how putformance standards are 3 treated.in an APA adjudicatory proceeding might arise. 4' Professor Golay states that NRC regulations, and I f 5 quote, " exist in diverse formats such as the Code of l 6 Federal Regulations, the NRC Standard Review Plan, the 7 NRC regulatory guides and branch technical positions." 8 While I have no doubt that your licensees may view all 9 of these. sources as regulations in fact, the courts 10 clearly do not and they will not give them the same 11 weight.- The Commission may not be happy about that 12 fact.. l 13 The Administrative Conference has recently '14-completed R'ecommendation 89-5 concerning which agency. ( 15' interpretation should t bind citizens and courts based ( 16 on a study by; f ormer Conference Chairman' Robert i 17' Anthony. While courts will clearly give deference.to i .t 18 Agency. interpretations of its mandate, which are 19 codified-in= its formal regulations following an 20 opportunity for public comment, it is not'so obviously 21 bound with respect to other devices used by agencies 22' to set forth its views.

The Conference therefore l

23 recommends that agencies use notice and comment 24 rulemaking or formal adjudication when they wish to-25 promulgate definitive interpretations of their NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE. N.W. (202) 234-4433 WASHINGTON D.C. 20005 (202) 232 4600 l

36 D', m 1 statutory mandate. 2 In short, if agencies want their views to 3 he given the deference owed official agency 4 pronouncements, they should act with the requisite for 5 malady owed such _ positions. As I continue in my 6 discussion, the importance of that may become clearer 7 to the discussion of performance. standards. 8 This recommendation was made because of 9 the' Supreme Court opinion in 1984 called Chevron USA 10-ve! sus the Natural Resou;*ces Def ense - Council. -The 11 court in Chevron gave substantial new directions to 12 resiew in courts concerning whether to give deference e 13: to agency interpretations of their governing statute. a f 14 Chevron now requires that a court accept an agency's 15 _ interpretation if it is, one, not contrary _to statute 16 or specific statutory intent and, two, is reasonable. 17 In April, the Administrative Conference 18 completed a study, an empirical study of federal 19 administrative law to examine the effects of the 20 Chevron. decision on reviewing courts. That study by 21 Peter Shook and Donalf }lliot,.who is now general 22 counsel at the EPA, finds that the rate of reversal-of 23 agency decisions declined but the rate of remand did 24 not. This suggests that chevron makes it harder for 25 courts to reverse agency decisions for errors of law. NEAL R. GROSS COURT REPORTERS ANDTRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON, O C. 20005 (202) 232-6600

37 1- 'In'short, chevron makes a difference that works. 2 Further, as this Agency well knows, in 3 'another' case, Vermont Yankee, the Supreme Court ended a 4 the meddling by reviewing courts of the development of' 5 procedures for rulemaking. The court stated clearly _6 that procedures going beyond requirements of the t 7 Administrative Procedure Act and of any particular 8. statute in, question could not be required by the 19. courts. The' ' Con f erenc e s recognizes that reviewing 10; courts tend to. find new ways to reverse or remand 11' cases to agencies when they. suspect the agency's - 12 record is'less adequate than it should be and itL l 1 -, e 13 should be expected ~ that much less deference will be 14L given to agencies,who _ don't place their 15 interpretations and policies on the record.through 16 notice and' comment rulemaking -or formal adjudication 4 17 than.may-have_been the case-in the past. 18 These recent develops.in judicial. review 19 are_ important to the discum on 'of performance 20 standards-because they may affect the conclusions.the 21 . decision maker reaches concerning whether a 22 performance standard or a prescriptive standard is the Y 4 23 most appropriate in a particular case. For example, 24 if you were to' decide to require that accident 25 sequences involving loss of off-site power must have a NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND /WENUE, N W, j (202) 234-4433 WASHINGTON, C.C. 20005 (202) 2324600

-____.m..__;_. 38 ' '[, 1 - probability of occurrence of less than one in a 2 million as. opposed to requiring two on-site gas 3 turbine generators, you should recognize that one 4 approach gives a reviewing court much more latitude to 5 examine the basis of your decision than the other. 6 On the Vermont Yankee, either requirement 7 implementing your statutory authority may be rather 8 easily promulgated, assuming there is a reasonable 9 technical basis for both through the notice and 10 comment rulemaking procedure. On the Chevron, both 11 rules,would be given deference, assuming that they are 12 clear -and. track-the statutory mandate and they would 13 be considered as definitive interpretations of the 14 statutory.act's requirements. 15 However, the adjudicatory record necessary 16 to show compliance with.the performance standard-could 17 'be-much more difficult to develop than'that necessary '18 to-show compliance with the prescriptive rule. 19 These. observations concerning the possible 20 drawbacks of performance standards apply both to 21 initial -licensing and to inspection.- and enforcement. 22 While they are more difficult for the agency to 23 administer, they provide the regulated entity.with the 24 advantage of being able to choose or invent the least 25 cause solution for its particular situation. In this NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENVE N W. (202) 234 4433 WASHINGTON. D.C. 20005 (202) 232 6600

39 +g .1, o 1 respect, I'd agree with Prof essor Golay that such an '2 approach encourages innovation and design 3 simplification and technological development and you 4 have to take that aspect into account in your 5 decisions as well, p 6 The Conference has had several occasions 7 to consider alternatives that might. assist technical 8 ' agencies in improving their decisional processes. In 9

1987, as part of its Recommendation 87-10, the 10-Conference encouraged the occupational Safety and li Health Administration to use performance standards 12

-whenever they would provide equivalent protection as y 13' that provided by design standards, those that 14 prescribe a specific technology or precise procedure 15 for compliance. 16 The Conference added that in deciding i 17 -which type of approach = to use, it should c'onsider + 18 whether the-standard could be readily understood and 19 monitored.and whether it could lower compliance costs. 20 In the preamble to the recommendation, the Conference 21 included an important caveat. The Conference has 22 found no alternative regulatory approach that is 23 always appropriate or better than the traditional 24 presumptively prescriptive regulation. So, while our 25 OSHA investigations encourage performance standards, NEAL R. GROSS OouRi REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUS, N W. l (202) 234 4433 WASHINGTON. D C. 20005 (202) 232-6600

40 1 we wouldn' t say or we couldn't say that one should .y 2 jettison the traditional mechanisms completely. 3 3 The OSHA example is similar in many { i 4 respects to the criticisms made by Professor Golay in 5 his underlying paper. OSHA critics objected to design 6 standards because they could impose greater costs and 7 equally effective alternatives.- They required 8 revision every time a technological change took place, 9 they = were-dif ficult to apply in varied work machine 10 configurations and they were difficult to understand. 11 Their advantages however were precise expectations for 12 employers, you knew what you had to do;. easy 13 compliance monitoring, you either fit in the rule or L 14 you'didn't fit in the rule; and the ability to ensure 15 that new technologies were implemented. 16 While the Conference supports use of 17 performance standards in appropriate circumstances, I .18 would.-be remiss-if I did not mention the. desirability 19 of developing consensus standards, whether they are of 20 the performance or prescriptive variety. You didn't 21 expect me to not mention alternate dispute resolution. 22 As you may know, the Conference has been a strong 23 advocate of alternate dispute resolution techniques 24 and the most obvious example in the standard setting 25 context is negotiated rulemaking. Since the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D.C. P0005 (202) 232 4600

41 1 Commission has already conducted negotiated 2 rulemakings, I will not belabor this point. The 3 Conference encourages the use of that technique in 4 appropriate circumstances, which I hasten to add is 5 not every circumstance for even a large number of 6 situations. In difficult policy areas where consensus 7 is needed and where issues can be defined in an 8 appropriate manner for consensus resolution, 9 negotiated rulemaking is the tool of choice. 10 -I might add that the Congress just passed 11 a negotiated rulemaking act which is up before the 12 President to encourage all agencies to explore greater 13 opportunities for negotiated rulemaking. If that is '14 signed. by the President, then we will have a - new 15 fr'amework'for the use of that technique. 16 The Conference has had occasion to 17 . consider as well hearing procedures for the resolution 18 of scientific issues. In 1985, the_ Food and Drug 19 ' Administration established a unique procedure for the 20 Public Board of -Inquiry, PBOI in their alphabetic 21 parlance, to obtain independent scientific review of 22 particular regulatory decisions. The PBOI is one of-23 three alternate informal methods of proceeding that 24 the FDA offers applicants in lieu of formal ALJ type 25 on-the-record proceedings or APA 554 proceedings NEAL R. GROSS i COURT REPORTERS AND TRANSCRIBERS g 1323 RHODE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON. D.C. 20005 (202) 232-6600 7

42 ,= 1 similar to that when considering approval of food 2 additives and new drugs. 3 They also are allowed to choose a hearing-4 before an advisory committee similar to the NRC 5 Advisory Committee on Reactor Safeguards or an 6 informal hearing before the Commissioner of FDA which 7 .is probably like the informal proceedings now codified-8 in NRC's Rules of Practice. 9 As a result of the Conference'.s 10 examination of the FDA experience, we concluded in our 11 statement ' number 11 entitled, " Statement on Hearing 12 Procedures for the Resolution of Scientific Issues," 13 that agency experimentation with alternate types of 14 hearing procedures for the resolution of scientific - 15 issues is justified. I should note that in these FDA 16 ' examples, the purpose of the procedure was to develop 17 a record to be used in-successive - stages of ' the 18, administrative and legal process whether that was a 19 rulemaking or an adjudication. These flexible type 2 0 -- formats could be used by ' the NRC to develop the type 21 of standard, i.e. performance or prescriptive, as well. 22 as the technica1' basis for the standard involved. 23 I would urge that you consider the 24 possibility of some of these flexible formats to 25 assist you in determining whether a particular rule NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W. (202) 234M33 WASHINGTON D.C. 20005 (202) 232-6600

43 i should be of a performance or a prescriptive variety 2 because I have to underscore that your decisions in 3 these matters will depend as. much on the particular 4 scientific context or regulatory context that you're i 5. dealing with, the specific rule you want to make, as 6 much as some c:,neral miasma, we like performance 7 standards, we.like prescriptive standards. 8 -I've named these different approaches 9 because practical considerations usually prevent the i I 10 use of one theoretical model to the exclusion of 11 others. That is to say whatever is in the air, you 12 still have to'look at the particular problem at hand. i 13 For example, the Consumer Product Safety ,p .14 Commission adopted'a performance standard requiring a i r 15- " sharp points" lab test'for toys and toy blocks. This 1 16 was an easier standard to write and certainly more 17 cost' effective than trying to specify myriad design

.18 and material. options acceptable for toys.

That's a 19 performance standard,' don't have sharp edges. 20 On the.other hand, if a regulator tries to-I i l. l 21 adopt many performance standards in lieu of design L 22-standards, the result may be unacceptable confusion. 23-For, example, sanitary plumbing must meet structural 24. strength requirements, thermal response requirements, 25 mech ~anical features, chemical features, including NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE. N.W. (202) 234 4433 WASHINGTON. D.O. 20005 (202) 2324600 t

44 (, 1 odorlessness, color stability and stain resistance, 2 biological effects and noise control regulations. 3-Separate performance standards for each factor might 4 make conformance prohibitively complex and expensive. 5 The regulated entity may also find 6 situations where a performance standard is less l .7 desireable than a prescriptive one. For example, when 8 OSHA changed its fire safety rule dictating the exact 9 height for mounting fire extinguishers and 10 substituting a performance standard stating that the 11 extinguishers must be " accessible," it opened an 12 Pandora's box. The burden of compliance for the 13-industry became more difficult. Basically whether or 14 not the fire extinguisher was " accessible" became a 15 matter that had to be litigated continuously and more f 16 regulation resulted rather.than less. 17-My conclusion,- after having observed the 18 debate over regulatory philosophy for a number of 19

years, is that while I urge ' experimentation with 9"

~20 performance standards, again I underscore no one 21 solution -is - a panacea. Regulators' have to examine-22 each approach and;veign the cost. benefits and 23 ~ associated tradeoffs in specific Lases. While-24 performance standards are often easier to write..they 25 are often harder to regulate, i.e. both practically as NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232 4600

45 1 in the accessible example with OSHA, and as a matter 2 of legal process. 3 If the time necessary to adjudicate i 4 compliance therefore is a major factor, the more 5 prescriptive the requirement the less they will be to 6 adjudicate. If adjudication is not a problem, but the 7 state of technology is rapidly changing and that's 8 your cutting edge issue, the performance requirement 9 may be indicated. Compliance measurement may be 10 controlled, public acceptance may be critical with one 11 approach more easily accepted than another. The ease li ' h 12 of writing one standard or the other may be important 13 .and indeed dispositive. Each of these consider -- I 14 mean I think again looking back on it'the OSHA people 6 15 would say, "We failed to capture what we wanted when 16 we-used the word ' accessible.'" Each of these 17 considerations ~must be balanced in order ~to determine ~ 18 how to proceed. ,~ f 19 One further specula. tion. While -20 _ performance standards themselves have historically 6 m"a 4g '21 been promulgated through rulemaking, there was at s, K' s} 22 least some evidence that the common law of the S M; m how they are in fact implemented, will. 4 23 standards, e.g. f JE' 24 be worked out in compliance hearings and 25-adjudications. To some extent, therefore, performance x NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS /p q; 1323 rho 0E ISLAND AVENUE. N W. (202) 234-4433 WASHINGTON. D c. 20005 (202) 232 4600

46 1 standards allow implementation policy to be made in 2 adjudication context and you the commission will have 3 to struggle how much you like that or don't like that 4 approach. 5 I hope I've been of some small assistance 6 to your deliberations and if at any time the 7 Conference can be of help in more specific ways 8 concerning fairness, efficiency or management of 9 initial processes, please do not hesitate to ask. 10 Thank you very much. 11: COMMISSIONER ROGERS: Thank you very much .12 for a very interesting perspective. It's a little 13 different from a purely technical one. J 14 Shall we turn now to Doctor David Rossin? 15 DOCTOR ROSSIN: Thank you, Commissioner. 16 This presentation was based on some work '17 that George-Sauter did at NSAC back in the mid ' 80s. 18 I left NSAC-in '86 to go to DOE. George is now at DOE 19 and this hearing offered us the opportunity to review 20 and rethink some of the work that we did at that time. 21 In an effort to shorten this, I'm going to 22 hope that whoever is running the slides can stick with The slides have numbers on them and I'll try and 23 me. '24 use them. So, if you want to use the first slide, at 25 least I'll know where you're starting. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLANO AVENUE, N W. b (202) 234-4433 WASHINGTON, D.C. 20005 (202) 232 4600

47 1 Presc-iptive regulation, we've heard about 2 these terms today. Prescriptive regulation focuses on 3 a means of accomplishing a desired end and tends to 4 down play the end in itself. Nonprescriptive 5 regulation recognizes that there are more ways than 6 one to meet a functional requirement. The appropriate 7 measure is the result, the performance, and not how it 8 is achieved. The how can be studied and emulated, but 9 should not be prescribed. 10 Ultimately, sets of words are used to tell 11 the designer and operator what is to be done. We call 12 these prescriptions. Somebody has to write every-13 prescription. In nonprescriptive regulation, it means 14 that the regulator, which would be the NRC in this 15

case, doesn't do it.

The licensee writes the 16 prescription-and the NRC reviews it. This is because 17 it is the licensee _who is ultimately responsible for 18 safety and performance. The one _with the 19 responsibility 'should be the one who determines how 20. the desired-result is to be obtained. 21 Nonprescriptive regulation also means that 22 theflicensee can change a prescription subject to the 23 applicable rules and subject to NRC review under those 24 'same rules. If another party, either the regulator or 25 a third party, proposes a change, the burden of proof NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234 6 WASHINGTON, D.C. 20005 (202) 232 4600

L' 48 ] 1 .that is necessary falls on the one that proposes the 2 change. The proposer may make a showing that a change 3 would be cost effective, but that would not be 'h 4 sufficient to mandate the change. The decision to 5 implement remains that of the licensee. 6-This presentation is not offered today as 7 an academic exercisa. The ultimate viability of 8 standardization in nuclear plant licensing may hinge 9 on success in implementing conc 9 pts suggested in this 10 presentation. I'll return to that point in a few 11 moments. 12 Historically, NRC has been heavily 13 involved in writing prescriptions. The regulatory 14 guides are the most obvious examples. They say that 15 following the guidance will be acceptable and that 16 another approach can also be acceptable but the burden 17 of proof of acceptability of any alternative 18' prescription is on the licensee. The result, though 19 not intended, has become a barrier to -innovation and 20 improvement.- As the Kemeny Commission' noted after 21 TMI, the control and safety systems at that relatively 22 .new plant were.old. They did not represent state.of 23 the: art at all.- For a complex and changing 24 technology, it does not make sense to be locked into 25 old and outdated equipment. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234-4433 ' WASHING 10N D.C. 20005 (202) 232 6600

49 1 I'd like to comment there. I agree with a 2 lot of the things that Michael Golay presented. I 3 think our thinking tracks very well in many.of these 4 areas. I do think oack to the Kemeny Commission's 5 report which focused heavily on the over attention of 6 the nuclear industry on compliance and also an over 7 emphasis on designing for extremely severe accidents 8 and not looking enough at accidents of somewhat higher 9 probability but less consequence. Of course that's 10 what'got us into the problem with TMI in a way. 11 Anyhow, I think there has been improvement 12 over the years. I think the trend is turning around. 13 I think'there's more attention being. paid to what it 14 is we're tryi's to. accomplish in safety and I. do 15 believe that 'ne backfit rule has permeated the 16 thinking of the staff and is helping to accomplish .17 this. 18 I'd like to just hark back to one thing ' 19. that' occurred-to me during your earlier presentation. 20 That is an example of a regulatory guide that was 21 . proposed without much attention to the state of the 22 art. This was the severe ' accident instrumentation 23 guide that was proposed back in about '77. It said, i 24' "Here's what'information the plant is going to have to 25 be.able to provide in the event of the following type NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON, D.C. 20005 (202) 232 4600

50 1 of accident." The only probably was there wasn't any 2 technology to deliver that. The instrumentation did i 3 not exist, it wasn't on the market. In fact, the t 4 people didn't know how to do it to meet all the 5 requirements. 6 We argued about that for a couple years. 7 Meanwhile, TMI happened and it would have been very l l 8 handy to have had instrumentation not to meet that c 9 regulatory guide, but to provide post accident 10 instrumentation. The ' whole. industry was sitting 11 around waiting to aee what the negotiations would 12 result in. 13 Anyhow, let me turn to the problem, and i 14 that is the regulatory structure in the U.S. has been 15. . subjected to heavy political pressures and has 16 responded by becoming very rigid and legalistic. I 17 There's a widespread recognition throughout the j 18 industry that there can be a heavy penalty ' f or 19 attempting to change, to improve or'to' innovate in any 20 plant that has. its license and is operating i 21 satisfactorily, s 22 The regulation of any technology to assure -.e 23 that it is' acceptably safe should have three: principal 24 components: risk policy, risk assessment, and risk i 25 management. I'll comment briefly on these. My olide NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON. D.C. 20005 (202) 232-6600 i

51 1 number 6 is about where I am in this presentation. 2 How safo should nuclear power plants be? 3 What is an acceptable level of risk? These questions 4 have been the subject of endless debate and this will 5 beep going. The answers cannot be obtained from 6 technical and scientific principles alone.

Rather, 7

the answers should reflect societal value judgments 8 which are indicated by what people are willing to 9 accept. Now, ideally, these judgments should reflect 10 due consideration of all the factors, the benefits, y 9 11 the costs and the risk, and the risk of the / 12 alternatives to nuclear power for electrical 13 generation, the need for diverse and reliable sources 14 of electricity. 15 For example, the risks to society from 16 inadequate electrical generating capacity are real and 17 should certainly be considered in setting our top 18 level performance objectives. Professor Golay 19 commented on the situation in Japan, or somebody did, 20 that Japan says, "We're going to have to increase our 21 amount of nuclear power in the future." I live in 22 California now. Los Angeles has said, "We're going to 23 start introducing electric cars out here and we're 24 going-to have a goal of introducing a substantial 25 fraction of the vehicles powered by electricity by NEAL R. GROSS court REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENVE, N W. (202) 234 4433 WASHINGTON, D C. 20005 (202) 232 4600

52 1 2006," I believe. That's all fine unless we end up 2 charging these batteeries by running gas turbines day 3 and night in order to charge the things. We're going 4 to be self-de2 eating in our goal unless we have the 5 technology to do what it is we're trying to do. 6 I believe that over the years NRC has been 7 schizophrenic in its consideration of non-safety 8 factors. On the one hand, it's often stated that its 9 mission is safety only. On the other hand, it has 10 claimed that it is required by the Nationt.1 ~ 11 Environmental Policy Act to consider

costs, 12 environmental impacts, energy needs and diversity of 13 supply in its environmental analyses, and I think it's 14 done this.

The purpose of NEPA, as I recall it, was 15 to make sure that these arguments are heard during the 16 licensing process and that the agency granting the 17 license has had the opportunity to hear this input. 18 It doesn't mean thet they've got to obey everything 19 that they're told, but to hear the input. And I think 20 this process is maturing gradually. 21 Because they may be subjective, however, 22 things like no undue risk to any individual or because 23 technical uncertainties may make it too unsure that = M-24 they're being complied with, such as no more than a 25 one percent increase in latent cancer risk, statements NEAL R. GROSS COURT REPORTERS AND TRANSCRIDERS 1323 RHODE ISLAND AVENUE. N W. (702) 234 4433-WASHINGTON. D C. 20005 (202) 2324600

i 53 o 1 of overall levels of acceptable risk are not easy for i 2 the regulator to apply. The regulator still needs a l 3 set of supporting criteria at lower levels, down to 4 the level of plant functions and systems, to allow him 5 to implement top level objectives in a useful way. 6 To be useful, these supporting criteria 7 should be consistent with top level objectives and the 8 degree of compliance with them should be measurable 9 with reasonable certainty. Given the top level 10 objectives, appropriate lower level criteria can be 11 established through the application of scientific 12 principles. Therefore, their establishment should be 13 done by technically capable persons with input from 14 the regulatory body itself, the regulated and the rest 15 of the technical community. 16 To follow Professor Golay's presentation, 17 I'd make the point here that this is an opportunity to 18 use PRA to test the effectiveness or the validity of 19 the lower level criteria. You set a lower level of 30 criteria to see how a system would look, how it would 21 be designed, and conduct a probabilistic risk 22 assessment and see if you get consist results that say 23 that the criteria you have established makes sense in l 24 terms of the results that you get. 25 in terms of risk assessment, I would note NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIDERS l 1323 RHoDE ISLAND AVENUE N W. (702) 2344433 WASHINGioN, D C. 20005 (202) 232-6600

54 1 that based on perf ormance, one can evaluate them and 2 ask to see if our top level safety objectives and 3 supporting criteria are being met. We haven't 4 explicitly stated all these, but I think we can say 5 that safety objectives are bei.ng made in licensed 6 plants in this country and certainly we have spent an 7 awful lot of time learning from them. That doesn't 8 mean things can't be improved. 9 In the performance based or 10 nonprescriptive regulatory structure we outlined, the 11 NRC would do all the things it does now except 12 prescribing what is to be done.

Thus, it would 13 continue to conduct safety analyses, review iisk 14 analyses and other submittals by licensees and license 15 applicants, evaluate operating experience and analyze 16 abnormal operating events, identify actual or 17 potential safety issues and carry out plant 18 inspection.

19 Risk management, as we define it, is 20 prescribing of a means to be used to assure that 21 nuclear plants comp.y with their criteria established 22 for acceptable levels of risk. 23 The NRC role in risk management would be 24 essentially limited to enforcement. However, under 25 its risk e.::::: rent authorities, the NRC could review NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 2144433 WASHINGTON. D C. 20005 'e02) 032 6600

55 o 1 and evaluate any aspect of the licensee's prescription 2 at any time to test, for example, if assumed component 3 reliabilities were actually being achieved or if 4 prescribed operator training programs were actually 5 being conducted. 6 In summary, under the regulatory concept 7 outlined here, the NRC could carry out all the i 8 activities it does now except to prescribe in full or 9 in part how the established safety criteria are to be i 10 met. The licensee, who is ultimately responsible for 11 the safe operation of his plant, must be allowed to 12 write those prescriptions and any necessary revisions 13 as best suits his particular circumstances. If the 14 NRC or any other party forces part of the prescription 15 on the licensee, then it should be responsible for any 16 consequences resulting from that part of the 17 prescription being faulty. 18 I would comment on the side that I don't 19 visualize the success of that particular set of 20 circumstances. 21 I believe the critical test which may make 22 or break the future of nuclear power in the U.S. lies 23 before us. That test is whether we can successfully 24 apply the concept of standardization to nuclear plant 25 licensing. For standardization to work, there must be NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND kVENUE, N W. (202)2344433 WASHINGTON. D C. 20005 (202) 232 4600

56 'o 1 a viable mechanism for making necessary changes to 2 .'icensed plants, making improvements, meeting 3 functional requirements with different components and 4 methods. This mechanism must also permit changes at 5 one plant to remain unique or make it easy to 6 implement the same changes at some or all similar 7 plants. 8 The key to standardization is an agreement 9 on the purposes and the objectives. Standardization 10 is important for safety certification and regulatory 11 review. Standardization has great potential value for 12 improving reliability and economics. But 13 standardization is not important for the sake of 14 standardization itself. I'll repeat that. 15 Standardization is not important for the sake of 16 standardization itself. 17 The NRC needs to find a way to handle the 18 modifications and continual work projects that must go 19 on in operating plants. The relevant issue here is 20 not standardization per se, but the need for a good 21 mechanism which allows changes to be made at plants 22 initial approved as standardized plants without 23 requiring that the same change be made at other 24 plants. Standardization is needed for safety review 25 and certification of designs, but maintaining strict NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS I 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232+600

l 57 1 standardization for the sake of the concept is neither 2 needed nor appropriate. 3 How can safety be assured if plants are 4 allowed to do different things once they are licensed 5 as standardized plants? The answer is nonprescriptive 6 regulation. What is important is that the functional 7 performance objectives need to be maintained. There 8 may be a variety of ways to assure that they continue 9 to be met even though changes are made in hardware and 10 procedures and that these changes may differ from 11 plant to plant. 12 Where safety considerations dictate that a 13 change must be made in all plants of a class, NRC can 14 mandate it, just as NRC can do today. 15 The operation of standardized plants 16 offers a logical opportunity to apply nonprescriptive 17 regulation. Once constructed satisfactorily, 18 nonprescriptive regulation based on performance 19 criteria can be applied to evaluate both proposed 20 changes and possible requests for conformance based on 21 changes made in other plants licensed under the same 22 standardized plant certification. 23 I would comment as an example of the 24 problem that I see, in the past there have been 25 instances in which design changes have not be approved NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHoOE ISLAND AVENUE, N W. (?O2) P344433 WASHINGTON, O C. 20005 (202) 2324400 l

58 1 even though they still meet the performance criteria 2 required for safety on the argument that the changes f 1 3 would result in a decrease in the margin that 4 previously existed. This approach has resulted in a 5 stifling of design improvements and efficiency l 6 initiatives. In standardized plants, there would be a 7 temptation to apply this same policy, regulating on I 8 existing margin rather than on meeting actual safety 9 and functional requirements. The proper test should 10 be the ability to meet performance requirements 11 without limit on how they are to be met or what margin 12 may be claimed by previous or alternative designs. 13 I think since TMI there have been both in 14 the NRC and the industry, with INPO and EPRI and NSAC, 15 they've devoted extensive effort to increased 16 understanding of all the aspects of design and 17 operation and in-depth analysis of operating 18 experience. 19 on the basis of safety, the present body 20 of regulations appears to have been successful. 21

However, the fact that no plant has been completed 22 that has been ordered since 1973 is a clear indicator 23 that the balance is not correct.

The fact that it is i 24 taking the system so many years to review the design 25 of a plant based on 20 years of experience serves 1 NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W l (?O2) 234 4433 WASHINGTON, D C. 2000$ - (202) 232 4600 ~

i 59 1 1 notice that the present approach must be improved 2 upon. 3 The application of performance-based, 4 nonprescriptive regulations can be achieved without 5 legislation, and could go far toward building the 6 necessary understanding to permit new plant orders 7 without compromising the high standards of safety that 8 have distinguished the American nuclear power program. 9 Thank you. 10 COMMISSIONER ROGERS: Thank you very much, 11 Doctor Rossin. 12 I think we'll turn now to Doctor Roger 13

Mattson, 14 DOCTOR MATTSON:

Thank you, commissioner. 15 (Slide) If I could have my slides, I'd 16 start with the first one, please. 17 I guess it's an advantage to be the clean-18 up hitter here. What I've heard going before me is 19 some views on both sides of the question about whether 20 we should move towards less prescriptive regulation 21 and it kind of fits in with what I had planned to say 22 to you because I'm going to give you some views on 23 both sides. 24 Let me say at the outset that I come down 25 at a bottom line thinking there are some advantages to NEAL R, GROSS CoVRT REPORTERS AND TRANSCRmERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D.C 20006 (202) 232-6600

60 1 nonprescriptive regulation and that we ought to 2 undertake some work to try to realize those 3 advantages. But there also are some advantages to 4 prescriptive regulation. You've heard some of those 5 listed and I'll try to put my spin on those with some 6 of my experiences with prescriptive regulation. 7 (Slide) If I can have the second slide. 8 I think it would help us all to remember 9 how we got to this point. Marty Malsch and I were 10 talking before we started this morning that we 11 remember back in the early days of the AEC, when we 12 first started working in regulation, that all of our 13 requirements were very general and we decided we 14 nseded to make them more prescriptive so that we knew 15 how to enforce them and that people knew what we were 16 intending. The call of the day was to be more 17 prescriptive, not less prescriptive. 18

Well, prescriptiveness grew as the 19 industry grew and I've listed some of the reasons on 20 this second slide.

We had a number of suppliers to 21 deal with, we had a wide variety of licensees, they 22 had different size, they had different experience, 23 they had different technical abilities and different 24 management abilities and regulating such a diverse 25 industry required us in many instances to be specific NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D C 20005 (202) 232-6600

62 1 about what it was that was needed for safety. 2 As time went on, technology improved and 3 operating experience accumulated. It was necessary to 4 change regulations, to interpret them differently or 5 to add regulations to address new things that we 6 learned about. Also as time went on, higher levels of 7 safety were required. The societal, political forces 8 in our country decided that higher levels of safety 9 were neoded, higher levels of assurance of safety were 10 needed and that led to increased prescriptiveness as 11 to how those levels were to be attained. 12 Then, of course, by the 1980s, after there 13 was already quite a body of prescriptive regulations 14 in place, there were additional information that 15 became available. The TMI accident and all of the 16 specific requirements that flowed from that accident 17 and, of course, the resolution of a large backlog of 18 generic and unresolved safety issues both led to 19 specific changes aimed at special problems and usually 20 problems of a fairly narrow nature. What this 21 resulted in was a patchwork of specific prescriptive 22 requirements that was very costly to backfit, it cost 23 plants under construction, it cost plants in 24 operation. And I think it is that a legacy of the 25 '80s that causes us today to say, "Su.ely there must NEAL R. GROSS CoVRT REPORTERS AND TRAN9CRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON D C P0005 (202) 232 4600

62 1 be a way to do this better." I 2 Another legacy of the '80s is the growth 3 of safety goals and the acceptance of PRA and perhaps 4 the tools by which we can do better in the future than 5 we've done in the past. 6 (Slide) on slide 3 I've tried to remind 7 us that there are a variety of for:ns of prescriptive 8 regulation. It's not just in the written requirements 9 like regulatory guides or Appendix K or Appendix R or 10 some of the national standards. You can also get 11 prescriptive regulations by the actions of individual 12 staff members, both reviewers and inspectors. You can 13 get collective decisions from the staff either on 14-specific licenses or on generic issues that are very .15 prescriptive. For example, the TMI action plan was 16 very prescriptive or the station blackout rule or the 17 ATWS rule. 18 You can get actions of licensing boards as 19 they try to arbitrate the resolution of contentions on 20 a specific case that can be very prescriptive and the 21 actions of the commission itself can ae prescriptive, 22 such as Appendix R or the inerting of *WRs, things 23 like that that were actions initiated by and taken by 24 the commission. 25 All of these forums of prescriptive NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON D.C. 20005 (202) 232 4600

63 1 regulation are amenable to management by the 2 Commission. We don't need to just do rulemaking to 3 manage prescriptive requirements, that is to put in 4 place through rulemaking some performance standard. 5 You can also manage the kind of prescriptive actions 6 that are taken by the staff individually or 7 collectively or the kinds of actions that are taken by 8 the Commission itself. 9 (Slide) Slide 4 talks about some of the 10 reasons for prescriptive regulation. Maybe we've 11 touched on these all today, so I won't spend a lot of 12' time with them. But often in the licensing process 13 there's a time pressure to not be on the critical path 14 for completion of construction or to not be on the 15 critical path f or the resumption of operations. So, 16 in the negotiating that goes on in the granting of a 17 license or the approval to restart a plant, there's 18 pressure to come to resolution and prescriptiveness 19 sometimes flows from that pressure of time. 20 Remember that our licensing process in'the 21 United States is an adjudicatory or adversarial 22 process and there are advantages in such a process for 23 being prescriptive about what the agency expects. The 24 less prescriptive you are, the more subject you are to 5 a delay in an adversarial licensing process. More NEAL R GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D C. 20005 (202) 232-6600

64 1 contentions will be brought against you, more debates 2 will have to occur, in all likelihood, if you are 1 3 nonprescriptive. 4 I would just give you two examples to show 5 you what I'm talking about. Compare Appendix B on QA, 6 which is quite general, with Appendix K on ECCS, which 7 is quite specific technically. Then think back over 8 your own experience about how much adjudication has 9 gone on in the adversarial licensing process involving 10 Appendix B, especially in the '80s and there was a lot 11 of it. And think back about how much adjudication has 12 gone on about Appendix K since it was issued in the 13 early '70s and the answer is not much since it was 14 issued, but a whole lot before it was issued. That's 15 the simple point I'm trying to make there. 16 There's been prescriptiveness because of 17 the technical complexity of some of the issues that we 1 18' deal with. You cat't be superficial with some of 19 these issues. You nave to get down to details. When 1 20 people get down to details in an adversarial, 21 negotiating type process 'ike our licensing process, 22 then you have to write the details down in order to 23 deal with them. That leads to prescriptiveness. l l 24 There ' s also variability among plants and 25 there you see prescriptiveness in the implementation NEAL R. GROSS COURT REPORTERS AND TRANSCRIDERS 1323 rho 0E ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON. D.C. 20005 (202) 232-6600

65 o. 1 of NRC requirements. For

example, 50.49 on 2

environmental qualifications is quite prescriptive as 3 a rule in and of

itself, but it gets even more 4

prescriptive when it's implemented plant by plant. 5 The rising level of safety I've already 6 mentioned, but just let me say that when we regulate 7 as we do in the United States by stating a minimum 8 standard for safety, and then if we raise that 9

minimum, it leads to a kind of incrementalism and 10 incrementalism, I think, leads to being specific and 11 being specific leads to being prescriptive.

So, 12 that's another reason for why we've come to this point 13 of being a very prescriptively regulated industry. 14 Another reason that I haven't put on a 15 slide, and I thought about this morning in preparing 16 to talk to you, is.thist sometimes the industry itself l 17 has come to the NRC anc begged you to be prescriptive. 18 They do that as individual licensees.saying, "Tell me 19 what it takes and I'll do this much and I'll get my 20 license." or sometimes a vendor will come with a 21 standardized plant, for example, or a standardized i 22 component, and want you to do a very prescriptive 23 technical review, approve their component for use in 24 very prescriptively controlled situations so they can l 25 go sell it in the marketplace. So, those are some of NEAL R. GROSS f CoORT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, O C. 20005 (202) 232 6600

l 66 j 1 the reasons. l 2 (Slide) The advantages, on slide 5, I 3 think we've kind of covered those. There's less 4 manipulation in an adversarial process. Appendix K 5 worked, but Appendix B didn't, Appendix K being much i 6 more prescriptive. There's less susceptibi3ity to 7 reinterpretation. 8 I ought to pause on that one for a moment. 9 My experience at NRC and since leaving NRC has been 10 that a lot of people in the nuclear industry have been 11 burned by reinterpretation of NRC requirements. 12 Notably the people that had plants under constrN tion 13 at about the time of Three Mile Island when nationally 14 we decided to have a higher level of safety and higher 15 assurances that that level of safety was being met. 16 So, nonprescriptive regulations, like Appendix B and 17 quality assurance and some other regulations, were 18 interpreted differently in the '80s than they had been 19 in the '70s by the NRC, which lead to risir.g costs and 20 delayed schedules for plants under construction. 21 So, if we go to less prescriptive 22 regulation, we have somehow to guard ourselves against 23 the reinterpretation of those nonprescriptive 24 standards sometime in the future, j 25 Another advantage is the binding record of l NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D C. 20005 (202) 232 4600

67 1 negotiated settlements. This is a process where the 2 government and the industry have to agree to do 3 something on a plant in order to get it to operate and 4 keep it in operation, sometimes because the issues are 5 complex and because the negotiation takes place at one 6 point in time and implemented over a long period of f 7 time. If there's not some binding record of what was 8 agreed upon, the agreement can erode. People can 9 forget what the safety basis was, what the reasons 10 were for the negotiated settlement. 11 That really leads to the last bullet on 12 this slide. Configuration control, we all know, in 13 our nuclear plants has not been what we had hoped it 14 might be. Our safety system functional inspections 15 that NRC performs and that some licensees perform for 16 themselves have shown that the licensing basis has not 17 been well protected in older plants, in some older 18 plants as we had hoped it might be. Part of the 19 reason might be that the older plants were licensed to 20 some very general regulations, Appendix A and Appendix 21 B. They weren't backfit to the standard review plan 22 over the years. They didn't have detailed safety 23 analysis reports. That is, they were 24 nonprescriptively regulated in the early days of the 25 industry. That could contribute to the difficulty of NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4 433 WASHINGTON, D C. 20005 (202) 232&>00

68 1 configuration control later in life. 2 So, I would urge us as we think about 3 nonprescriptive regulation, which I said at the outset 'I 4 I think there are some advantages we should work on, 5 we've got to be careful that we don't throw the baby 6 out with the bath water. There are some things to 7

keep, i

8 (Slide) Okay. Slide 6 talks about the 9 advantages in regulating safety function performance. 10 Nobody else has called it that today. I did that on 11 purpose because of some experiences that I've had both 12 at NRC and since leaving the commission when I did .13 some work for the IAEA. What Professor Golay is 14 talking about is setting some goals for the for example, 15 performance of safety functions and then using some modern 16 removal of decay heat 17 tools like PRA and safety goals to set the goals and 18 to determine whether you've met the goals for these 19 safety functions. 20 What happens when you do this kind of 21 safety function regulation is you get a better 22 counting f or system behavior because you're using l 23 system level models, PRA-type models of the plant. 1 24 You get better understanding of the interdependencies 25 of systems within the plant than you do'when you're NEAL R. GROSS i CoVRT REPORTERS AND TRANSCRIDERS 1323 RHODE ISLAND AVENUE. N W (202) 234 4433 WASHINGTON, D C. 20005 (202) 232M00

69 1 narrow and prescriptive and you regulate one component 2 or one system at a time. 3 You get an improved treatment of plant-4 specific factors because you've by this approach 5 delegated responsibility to implement the approach-- 6 that is, to figure out the details down to the 7 licensee and the licensee has command of the 8 individual plant factors. You provide for some 9 rationalization of your safety goal. Others have 10 called that the " disaggregation" of a safety goal. 11 That is, you break the overall performance standard 12 for the plant -- that is, the safety goal -- down into 13 its aggregate parts, the safety goals for subsystems 14 or functions, safety functions within the plant. We 15 did that when we solved the ATWS problem for light 16 water reactors and when the Commission put out its 17 rule on anticipated transients without scram. The 18 basis for that rule was a disaggregation of a yet 19 unapproved safety gual. 20 We've already mentioned that this approach 21 puts increased responsibility on licensees. That's 22 good. That causes licensees to be more technically 23 capable and it causes the details to be better 24 handled, but you can't be naive as you go into that 25 placement of greater responsibilities on licensees. NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W. (232)2344433 WASHINGTON, D.C 20005 (202) 232&M

70 1 We know from past history that there are some who are 2 good at accepting that responsibility and there are 3 some who are not as good, and you have to find a way 4 to protect yourself and protect the public for the 5 utilities that aren't able or ready to accept that 6 responsibility. 7 Probably the biggest advantage as an 8 engineer that I see in this approach is it provides 9 flexibility for optimization. You can take into 10 account both the power production purposes of the 11 plant along with the safety needs of the plant in 12 deciding what's the best way to design a system or to 13 put in place an administrative procedure or whatever. 14 That is, systems reliability becomes a common currency l 15 of both safety and production when we move to this 16 kind of approach. 17 (Slide) On slide 7, I've tied to recall 18 for you, as Professor Golay did in his paper, that l 19 there are some places where we've made forays into 20 this field before and I would just mention a couple of l 21 them. The aux. feedwater reliability goal that's in 22 the standard review plan of course came out of post-23 TMI PRA work on PWR aux, feedwater systems, and that 24 goal is a disaggregation of the then still unapproved 25 safety goal. I've said that twice now. You've NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433' WASHINGTON. D C. 20005 (202) 232 6600

71 1 approved the safety goal since then.

And, some 2

understanding of grid reliability in PWRs across the 3 United States. 4 You may have forgotten, but at the time 5 the ATWS rule was proposed for rulemaking it was the 6 so-called second NRC proposed rule or the Hendrie 7 Rule. Commissioner Hendrie and I think Frank Rowsome, 8 who was on the staff at the time, drafted this rule. 9 I'd read to you a paragraph from the statement of 10 considerations, if you'll bear with me. It's the 11 second paragraph under the so-called Hendrie Rule. 12 It says, "The NRC is exploring the 13 possibility that the regulation of reactor safety may 14 evolve toward regulating the process by which 15 licensees ensure public health and safety and away 16 from licensing the details of plant design and 17 operation. Programs like the reliability assurance 18 program and this proposed rule offer promise of 19 growing into a formal, auditable way the NRC can 20 determine that licensees are doing a satisfactory job 21 of ensuring public health and safety." 22 That's very close to what Professor Golay 23 is recommending in safety function regulation, as I've 24 called it, reliability assurance regulation as Hendrie 25 called it, nonprescriptive regulation as we've titled NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W, (202) 234 4433 WASHINGTON, D C 20005 (202) 232 & %

72 1 it in today's meeting. We all know the Hendrie Rule 2 didn't win in that rulemaking, but out of that 3 rulemaking came a final rule on ATWS and I'll bore you 4 one moment more by reading to you a paragraph from 7 5 that rule's statement of consideration. There's a 6 page and a quarter on reliability assurance in that 7 statement of considerations. 8 It says, "One of the principal findings," 9 and this is of the Salem ATWS event, "was the lack of 10 adequate attention being paid to the reliability of 11 the reactor trip system. The Salem Generic Issues 12 Task Force recommended to the Commission that a 13 reliability assurance program be included in the final 14 ATWS rule," and it gives a reference. "While this 15 rule," the one that was finally issued, "does not 16 require such a program, the Commission urges the 17 voluntary development of a reliability assurance 18 program for the reactor trip system." 19 It goes on to list four components of such E 20 a reliability assurance program, and if I could I'll 21 read you those four components. They're going to 22 sound familiar in the context of today's discussion. 23 First, an analysis of the challenges to 24 and failure modes of the reactor trip system. 25 Second, a numerical performance standard NEAL R. GROSS CoVRT REPORTERS AND TRANSCR$ERS 1323 RHODE ISLAND AVENVE, N W. (202) 234 4433 WASHINGTON D C 20006 (202) 232M00

73 1 for the reactor trip system challenges and the reactor 2 trip system unavailability. 3

Third, a process of evaluating plant-4 specific and industry-wide operating experience to 5

provide feedback to assess whether the reactor trip 6 system is performing reliably enough. 7

Fourth, procedures within quality f

8 assurance programs to assure that the reactor trip 9 system performs satisfactorily in service from a 10 reliability perspective. 11 And that sounds to me very close to a

l2 trial application of what Professor Golay has 13 recommended in his paper.

What this statement of 14 considerations says, in essence, is the industry will 15 be given some time to take that approach for the 16 reactor trip system, and if it does not the Commission 17 will examine whether rulemaking should be undertaken 18 to develop such an approach for the reactor _ trip 19 system. 20 On Sizewell-B in the United Kingdom they 21 had a safety goal, if memory serves me, of 10- 6 for a 22 significant release. They disaggregated that safety 23 goal for safety function performance in the design.and 1 24 regulatory review of Sizewell-B. Your staff has done 25 a review of that process. It's in a NUREG report and NEAL R. GROSS CouaT REPORTERS AND TRANSCRIBERS l 1323 RHODL ;SLAND AVENUE, N W (202) 234 4433 WASHINGTON. D C. 20005 (?o2) 232-0600

1 74 1 I believe that is an implementation by another country 2 for an advanced PWR design that would be fairly 3 typical of what you would do if you applied this 4 technique to an advanced reactor in the United States. 5 INSAG-3 is a document you've probably 6 seen. It came out of the IAEA. I had the privilege 7 of having some role in that. I'd remind you that in 8 there, in INSAG-3, there's a section on reliability 9 targets under the section on design. And, it was a 10 section that was put there not based on U.S. 11 experience. It was put there at the insistence of the 12 French, the Germans, and the British, because it was 13 their judgement that this kind of regulation was the 1 14 best way to move in the future. And, without reading 15 it to you, it says to establish reliability targets 16 for safety functians, and I believe that's very 17 similar to what we're doing here or what we're talking 18 about here. 19

Finally, I'd introduce a new thought I l

20 think to today's discussion, and that is it isn't only 21 in design that you should consider using 1 22 nonprescriptive regulation. You can also use 23 nonprescriptive regulation in operational matters, and 24 I'll give you one example where we've done that. 25 After TMI there was a move within the NRC NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIDERS 1323 RHoDE ISLAND AVENUE, N W. (202) 234 4413 WASHINGTON. D C. 20005 (202) 232 4600

l 75 1 staff to get fairly prescriptive about occupational 2 exposure regulation. There was a counter move within 3 the industry to say, "We see the problem," that is, 4 occupational exposures are rising year by year and 5 they shouldn't be. They should be reducing year by 6 year. Naval reactors experience showed that that 7 could be done. Industry said, " Don't regulate us 8 prescriptively. In fact, don't even change your 9 regulations. Tell us generally what you want, and 10 that is exposure reduction, and we through INPO will 11 go do that." And we know that there's been some 12 success in that area. 13 And, basically what NRC did was to assure 14 itself that there was an active critical safety review 15 ongoing in plants, and they did that through 16 . inspections and through promises by INPO, that there 17 wereL active ALARA. programs, as low as reasonably 18 achievable exposure control programs. And, those the 19 NRC did have a handle on. They could assure that 20 there was such a program. They assured themselves '21 that there was analysis of exposure data and snat -22 there were' exposure reduction goals set through the 23 INPO process. That safety function is now achieved by 24 nonprescriptive' regulation. 25 It's possible, in my mind, looking in from NEAL R, GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENVE, N W. (202) 2344433 WASHINGTON, D C. 20005 (202) 232 4600

76 1 the outside as you struggle with the problem of 2 maintenance regulation, that there may be an analogue 3 in exposure regulation through nonprescriptive 4 regulation to solve the maintenance problem. 5 (Slide) On slide 8, I've tried to list 6 some prerequisites to moving to some safety function 7 performance type of regulation. We've already 8 mentioned today that you'd have to do come plant-9 specific PRAs. If you confine your initial efforts in 10 this field to advanced reactors, you're going to get a 11 PRA for those reactors anyhow, so there's not an 12 increment by going this way. It's not an incremental l 13 cost. 14 You've got to have some way to credibly 15- -maintain the performance data. I believe in Professor 16 Golay's paper-he recommended you do it. I don't 17 recommend _ you do it. I recommend you have somebody 18 else'do it for you, like'INPO, like NPRDS, snd that 19 you have oversight of it. But, if industry is gning 20 to use it, then it's got to be ~ their system ' and. 21 they've got to believe in it and they've got to. 22' provide data to it. 23 PROFESSOR GOLAY: That's actually my e 24 recommendation. 25 DOCTOR MATTSON: Okay. I misread it. i NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. .(202) 2344433 WASHINGTON. D C. 20005 . (202) 232 6600 .~

77 I 1 You've got to have a durable consensus on 2 methodology. The last time I think as an industry we 3 looked at the methodology was with some guides that 4 were written in the early '80s with the cooperation of 5 t.he American Nuclear Society, the AIF then, and the 6 Commission. I think there's probably some change 7 that's occurring in PRA as you go about your 8 individual plant examination program right now. There 9 are some efficiencies, I suspect, that people are 10 beginning to learn about in probabilistic risk 11 assessment, the use of PCs, the use of master plant 12 logic diagrams. Things like that are making PRA 13 technology advanced and I don't know that you need to 14 worry about standardizing methodology right now. I'd 15 probably wait a couple years before I got into that 16 business. and I've 17 You also have to worry about ' the ability of 18 mentioned this before 19 licensees to implement the new approach. People are 20 going to have to be trained. You've got to make sure 21 that.everybody shares the same attitude and the same 22 philosophy about the move in this di: on.

Again, 23 if you decide your initial trials art Ath advanced 24 designs, then the people you'll have to deal with are 25 more limited.

You only have the advanced reactor NEAL R, GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON, D.C. 20005 (202) 2324600

• 18 1

designers. 2 (Slide) There are a number of interfaces 3 that you have to consider. I've listed those on slide 4 9. I won't spend much time here. You've got to find 5 a way to deal with the qualitative factors, such as 6 management performance and human factors, as you move 7 into this kind of domain. But, you can do it, because 8 you've shown you can do it with exposure control, as i 9 low as reasonably achievable occupational exposure. 10 Common sense, the last bullet, means that 11 all of these new analytical high technology ways to 12 regulate you've got to make sure that the bottom line 13 still makes sense. 14 I think the defense in depth discussion 15 that you were having earlier, Commissioner Remick, in 16 looking at advanced reactors where people have 17 encouraged that they be of. all passive character in -- 18 safety systems to reduce the load on operators, I've 19 noticed a tendency to build containments and decay 20 heat removal systems that require the reactor system 21 to go either to high pressure.or high temperature and H b, L 22 stay there for a long time and to transfer heat out 23 through the containment through some passive means to 24 the environment. 25 That looks pretty good, but then you stop NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W. _ 202) 232 4600 ( (202) P34 4433 WASHINGTON, D C. 20005

79 1 and you say, " Wait a minute. In the Chairman of the - 2 Nuclear Regulatory Commission and I want to explain to 3 the governor of state XYZ that I've had an accident at 4 the plant and it's all very passively safe, but it's - 5 going to take 30 days for the containment pressure to 6 decay." There's something wrong with a reactor that .7 does'that. 8 It's nice to have that passive capability '9 in.the background and you might get that out of some . 10 of'these new approaches, but it would also be nice to 11' have' an active. system that under most circumstances 12 lwould -rapidly reduce the pressure inside a 13 containment. So, common sense has to be there and 14 defense in depth may be a good way to. provide that = 15 1 ' common sense. 16-. (Slide) In -summary, on my last slide, I '17 think 4 prescriptive reaulation has been necessary. I 1 "18 ' think< it' will ccc.uinue, to be necessary -in the: future .19 for some-of the same: reasons it was.necessary'in the 20 past.

I don' t see you g i v i n g '. a w a y - t h e licensing 121 process that you have toduy, at least not overnight.

22-There are a' lot of people who-wouldn't want you to do . 2 3' 'that. 'And, that licensing process demands some 1 24 ' prescriptive statements of what it is you expect, + 25 otherwise' 'it 's too susceptible to reinterpretation NEAL R. GROSS CoORT REPORTERS AND TRANSCRIBERS 1323 RHODE l$ LAND AVENUE, N.W. ,(202) 734 4433 WASHINGTON. D.C. 20006 (202) 232-6600

80 f 1 over time and from case to case. 2 Nonprescriptive regulation has some 3 supt lementary benefits. You've heard some of them 4 described. I hope I've listed some. You shouldn't 5 feel that you have to be static in regulation. This 6 is the Kemeny Commission speaking in '79, as Doctor 7 Rossin has described. They're saying keep up with 8 technology in '79 and there are some aspects of that 9 in this initiative that you're considering. I would 10 encourage you to try it a couple places. I've 11 mentioned one, maintenance, and others have mentioned 12 advanced reactors and those are the two that'I 13 intended by this last bullet. 14 on advanced reactors, I think you have a 15 unique opportunity-to lay your prescriptive 16' regulations alongside your-nonprescriptiv6-analysis 17 and optimization that-the designer and the NRC could-18 work on, and then by certifying the plant through 19 rulemaking you can-replace the old prescriptive 20-regulations that are not optimi7ed by some new 2i prescriptions. that were d e t e r.a '. n e d through a 22 nonprescriptive process, but still have the advantages 23 of-prescription - that is, they're durable over time. 24 They're not subject to reinterpretation by you or by 25 the licensee. NEAL R. GROSS COURT REPORTERS AND TRANSCRIDERS 1323 RHODE ISLAND AVENUE N.W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232 4600 ) m.,

4 .( 1, 81 1 And with that, I'll conclude. Thank you. '2 COMMISSIONER ROGERS: Well, thank you very 3 much. .'4 I know Mr. Breger has to leave essentially 5 'now, but I want to make sure that we thank him very 6 much for being with us and giving us a very 7 interesting perspective that's a little different from 4 4 8 the technical one. 19 I wonder, Commissioner Curtiss, if you 10' have perhaps a question? 'll COMMISSIONER CURTISS: I just have a 12 comment, I guess. 13 COMMISSIONER ROGERS: I know Mr. Breger-l t 14 ~ has to' move, and that's why i'm -- P s-- 15 MR. BREGER: I have un appointment on the 16-Hill'as well.'. 17 COMMISSIONER CURTISS: I would, I ' guess,1 18' . congratulate you on the work that you've done in'the y [i - 1'9 ; alternate dispute resolution area - I know you've put p' L20! a lot of' effort into that initiative over time. .It's P

21

not squarely, on the subject that. we ' re discussing _22 today, but the contributions that you and ACUS-have - (g 23 made in that area have been:I think of benefit to us 24 ,and all' agencies. 25 - -MR. BREGER: Thank you. NEAL R. GROSS l COURT REPORTERS AND TRANSCRl0ERS 1323 RHODE ISLAND AVENUE, N W. -(202) 2344433 WASHINGTON, D.C 2000$ (202) 2324600 t

82 1 COMMISSIONER ROGERS: Well, we thank you 2 very much. We're sorry that you have to leave us, 3 because I think we've got some interesting questions 4 that we'd like to explore, but we'll just have to do i 5 it a little bit separately. 6 MR. BREGER: Well, please, if you give any 7 of those to my office.or my staff, I'd be happy to try g 8 to answer them in detail for you. 9 COMMISSIONER ROGERS: Thank you again. 10 MR. BREGER: Thank you. ? 11 COMMISSIONER ROGERS: All right. I think 12 now I'd turn to questions. I'm sure we all have some. 13 Commissioner Remick? t 14 COMMISSIONER REMICK: First, Professor 15 Golay, unquestionably I'm a believer in performance 16 standards where it makes sense to do so and I did have' 17 some questions about= some of the concerns I have in 18 implementing it, but Mr. Breger did an outstanding job i 19 in great depth and-with some examples of-expressing '20 those, and then Roger came along and added some, and + 21 basically they relate to:the fact that-we live-in an t 22 advers trial adjudicatory licensing process. And, as: 23 was po;nted out particularly by Mr. Breger, I think it 24 is more difficult in that environment to show 25 compliance with performance-based standards than NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W, (202) 234 4433 WASHINGTON, D.C 20005 (202) 232 4600 e

-83 =. - 1 -perhaps prescriptive. I also think that probably 2 applies in.our inspection process. 3 For inspectors, prescriptiveness helps 4 them to check-off whether a licensee is meeting I 5 don't defend it, but it's an observation, makes it 6 easier. And I think performance-based standard 7 probably requires greater technical capability of i 8-2nspector staff to be able to put away a licensee is-9 doing_something and whether that meets the overall-10. performance standard. E 11 And then my fear, as I. think Mr. Breger 12 indicated, that this could lead to continuous 13 litigation -- but, he 3id'a thorough job of expressing

14 :

.those limitations or what I express as my concern. I 15 wasLgoing to ask you, do you have any reaction to the ^,. 16i things that he said or, as I've tried to describe, d.' 17 ' general concerns? -18 . PROFESSOR GOLAY: Well,- there,is one 19 point, which was ref erring to the article which I'd '20 circulated _with you and the criticisms of the existing g -j 21l systems. I'd mentioned being arbitrary, inconsistent, a;.. 8 N' ~ legalistic. They do.not all flow from thel same> 22 and m m 23 sets of causes, in my view, and the. legalistic side-cf ^ a. 24' ' things will be there no matter what we do, as you've -25 pointed - out. Rather, the hope here is that you can NEAL R. GROSS COURT REPORTER $ AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. ~-(202) 23+4433 ' WASHINGTON, D.C 20005 (202; 232-6600 l, s \\.-

84 1 make at least the structure of the questions that you + or try to answer,

rather, more 2

try to ask 3 consistent and with the basis of always appealing to 4 alternative interpretations to a single standard. 5 I guess I don't really know or have a 6 sense of whether this would be better or worse from 7 the point of view of litigation. I do believe that 8 you can take almost any logically-based system and 9-make a mess of it, you know, given enough help from 10 enough people, and from that point of view it might be 11 worthwhile to try it in a smaller example first and 12' =make some decisions on that basis. 13 It seems to me that, in a way, the 14 litigious aspect of what we have in our nuclear safety 15 regulation is less a consequence of whether we have a 16 prescriptive or nonprescriptive system than it is a 17 consequence: 'of the f act that many of our social 18 questions tend to get. f ocused on - the licensing of 19' nuclear power plants. That is, we have questions 20, .which range f rom should we have a nuclear power-21- . program to is the layout of the bolt circle on the 22; pressure vessel the right one. 23 I would probably cast the question 24 differently, asking in effect how can we create 25 alternative forums to resolve the broad social NEAL R GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE N.W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232-6600

85 ~ F; - 1 questions on the acceptability of nuclear power so 2 that the nuclear safety regulation becomes focused 3 much more on just whether the hardware will do what -4 you intend it to do. If you look at the systems in 5 most other countries, they have that kind of 6 separation and I've always felt that that explained a ? 7 lot of why our procedures are very time-consuming and 8 complicated. 9 DOCTOR ROSSIN: 'Forrest, could I add 10 something to that? 11-COMMISSIONER REMICK: Most certainly. 4 1 - 12 DOCTOR ROSSIN: I think Marshall Breger e-i; 13 pointed. out that in general the Supreme court and a 14 where it's been settled at a lower level have 15 s u pp o r t e d a dmi~ni s t r a tiv e agencies in their 16 determinations on - these issues. when the~ record is a 17 sound one. And,.you look at some of the most serious 18 delays in nuclear plants in the case history. -When-19. they finally got to_the Supreme Court, the Agency and 4 20 the~ licensee were upheld-in'their decision. The only 21 problem was that-the years of delay resulted in the w 22 economic disasters that have pretty. much put nuclear 23-power on the shelf-in this country. 24, I can think back all the way to the Bailey 25' case where I think 10 or 11 years went by while that NEAL R, GROSS CoVRT REPORTERS AND TRANSCPIDERS I 1323 RHODE ISLAND AVENUE, N.W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232 4600

i _ i. 86 1 was litigated.-

Well, the final decision of the 2

Supreme Court said that the Agency was perfectly in 3 the right in making the decisions it made. By that 4 time, the baby was dead and the issue was gone.

And, 5

I don't think this really is a function of-6- prescriptive or nonprescriptive regulation. If 7 somebody wants-to take advantage of the system, they 8 may find a way and it may be with a prescriptive rule. 9 It may not. So, I don't think that's a valid argument 10 for not applying nonprescriptive regulation where you 11 can. 12 And, the' OSHA example I think is a.Very '13 important one, because I think it's widely recognized L. 14-that' OSHA is an indication of a lot-of our problems of 15' being overly' prescriptive in legislation and in rules 16 that'were so prescriptive that in general they were 17 overkill. in many, many dir e c tic.'s. It' seems to me s 18 that: what was missing there was: h r.dership in_ an 19 agency that told its enforcement people, "Here is what l= 20 you mustLdo and ' here 's a record and a basis f or it," L 21 an'd that I think' would have stood the court test, l 11.. 22 rather than finding that individual inspectors were-y 23 fighting with a licensee about whether the thing is -24 this high or that high. That shouldn't have-happened. 25 The leadership was missing. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W. (202) 23&'A33 WASHINGTON, D.C. 20005 (202) 232 6600 L

'., = 87 1. COMMISSIONER REMICK: Well, certainly, I 2 feel that performance standards are worthy of an 3 ' attempt to-use them. There's just no question about 4~ that. I found Roger's suggestion that perhaps the 5 certification process is one where we can take 6 existing prescriptive regulations perhaps and put in 7 another format. You have the prescriptive regulations 8-that you want to make sure things meet. That's an 9 intbresting thought. 10-DOCTOR MATTSON: By having the 11-certification rulemaking, you draw all of the 12 attention to.'a national level proceeding where you put 1. .13 on the record all of your PRA basis for the optimized 14 design that you've chosen or that you've decided to - 15 license for'an advanced plant. And then, once you've 16 been through that process at a national level, then in 17-. . theory. you don' t. have - to. go through it again on-an w 18' individual plant level, but you could~still make the 4 Y . 19 outcome ' of l that national:' level rulemaking be a very 20, . prescriptive definition of' what you expected to see i (!; 21 -over the lif'e of' that design in' terms o' ne _ 22 performance of that design. .I-don't know. The 23 lawyers would have to advise us how to do that, but I _ 24 would think it would be through some reference to the 7' 25 safety analysis report on which the rulemaking was NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON. D C. 20005 (202) 232-6600

J 88-1 based to. certify the design. 2 PROFESSOR GOLAY: Let me add two things to 3 build on that, and that is in my proposals I've always F 4 suggested that we try this out for new plants, that 5 trying to take an alternative approach and apply it to 6 the existing license I think is just asking for an t 7 incredible amount of trouble. And, the idea of doing 8 it.on standardized designs I think is really the right 9 way to do it, because this is a more labor-intensive 10 approach. You really have to go in much greater 11. depth. And so, the justification, if you've got a few 12 examples where you're really going to try and do it gs, t right,,I think makes sense in that way as opposed to 13 14 taking some operational examples, n . 15 COMMISSIONER REMICK: One of the example 16 'that-I think of performance-based -- and I hesitate to 1 17 use-the word " standard.'" I'should say a performance- .18 based goal -- is the safety goal. q w .19 But,. Dave,.I noticed on one of your slides -l .i 20 you indicated the saf ety.- goal as an example of t 21 prescriptiveness and that really surprised me because > c'.- w '22 it doesn't say how-to-do it. -It just states a goal

23 from a perspective of:public health and doesn't'tell 24 you how --

25 DOCTOR ROSSIN: Let me clarify that. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON. D.C. 20005 (202) 232 6600 y

89 i e 1 COMMISSIONER REMICK: Okay. 2 DOCTOR ROSSIN: It's very simple. On that 3 one slide I listed a number of topics for discussion 4 and I wanted to discuss and compare. Please ignore 5 that particular slide. 6-COMMISSIONER REMICK: Okay. 7 DOCTOR ROSSIN: It needed a lot more 8 discussion than just the bullets. 9 COMMISSIONER REMICK: Okay. And both in 10 your paper and you mentioned a one percent increase in '11 latent cancer risk and I wasn't clear if you were 12 referring to the saf ety goal. If that's the _ case, it's a tenth of a. percent increase, but I thought -- I 14 didn ' t. know. The one percent increase in -latent 15 cancer risk,nboth in the-paper and you mentioned this = 16 -

morning, I'm not sure what you're referring to 17 specifically.-

18 . DOCTOR ROSSIN:. I'was just using that as 19' an-example of=a thing that could.be use'd, if one-tenth 20 is the right number. But, all I'was getting at is 21 that ' if you.use a calculated. goal like that, using 22 uncert 4nties of. the kind you have in the linear 23 theory, you 2e going to get a number, but-it's just a 24 number. 25 COMMISSIONER REMICK: Do you have any NEAL R. GROSS COURT REPORTERS AP'O TRANSCRIBERS - 1323 RHODE ISLAND AVENUE. N W. (202) 234 4 33 : WASHINGTON D C. 20005 (202) 232 6600

'90 ^ 1-reaction to - my statement that I think if we had 2 performance-based stanlards that it would require of. 3 our

staff, certainly our inspection

staff, 4

considerable technical capability to make the 5 judgement of what they saw in the plant, because it 'P 6 would ba many different ways of approaching the same 7-performance standard.

And, wouldn't it require 8

greater technical capability on their part to make.the 9 judgement that that meets that performance standard, 10 versus, if it's very prescriptive, saying yes they 11 meet it or not? 12 DOCTOR ROSSIN: Well, obviously, if it's 13 meeting a simple prescription, you can say yes or.no. .14 But, ny impression is that the engineering capability i 15 of the staff is adequate to do the kinds of things + 16 we're -talking about. There aren't that.many nuances 17 here. I think-we're talking about trying to establish i 18 criteria and then saying,

Look, I'm trying. to do 19 exactly.what.I've agreed-to do.-

.I'm' going-to do it a 20. -slightly different way'and I can show you why this way '21 is just as. good or better than the way-I did before." 22' 'That should be no-problem. 23 I'm scared about rigidity where someone 24

says, "Uh-oh, no.

That's no good because it's 25 'different from what we did bef ore," or, "I can show NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHoOE ISLAND AVENVE, N.W. (202) 234-4433 ~ WASHINGTON, D.C. 20006 (202) 232 4600

91 U,. 1 you that I've got a safety margin of 50 percent here 2 and I can see that we can do this much better. The 3 safety margin, we still meet the goal. We meet it by 4 30 percent instead of 50 percent. That should be 5 perfectly good." 6 In fact, one of the examples Roger gave

• /

~ was Appendix K,

i. p p e n d i x K was arrived at by 8

litigation, not by technology, and years later-- 9 well, and at the end of that proceeding I think-one of 10-the reasons you didn't have a lot of litigation 11 following the decision on Appendix K was, number one, 12 the ~ industry lost and they'd had a belly full of it 13-and they were really-not interested in going-back and 14 going through this again. 15 Back-in the mid '80s at'NSAC we went back 41 6' to Appendix K and we looked;at the margins that were i o 17 there based on the safety basis for. Appendix K. I 18 mean, it says in the role, "Here's what wo set out tc 19 do," Now, you looked at.the numbers with the kinds of 1 20 computer capability we had' and the. experimental: 21-evidence and the operating.- experience that = we ' d-22 accumulated ' over the previous seven or eight. ' years 23 since the rule.and saw that the limiting temperatures 24 were maybe 200 or 300' or 400 degrees ' more stringent 25 than they needed to be to accomplish all the safety NEAL R. GROSS i COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. . ASHINGTON, D C. 20005 (202) 232 4600 W (202) 2344433 i

T 92 's. 1~ goals or all the safety bases that were in there. 2 Appendix K is too rigid to changa without rulemaking 3 again. 4-If it said do it so that it meets the 5

rules, then I think any competent regulatory 6

engineering staff could review what.you did and 'see 7 that with NRC approved. codes and everything else the 8-requirements are met, and f rankly ' this has made a 9-difference of a substantial amount in terms of core 10 performance throughout the United States and ' 11 throughout the rest.of the world where they adopted a 12 out standards just because it was easier to do it. . 13 I think that a tremendous price has been 141 paid, but as a result of that litigation it was so 'h ' 15 contentious and so extended: and so brutal that nobody 16: really was in the. mood to contest it for quite some 17; time.. '] 18 DOCTOR MATTSON: Commissioner Remick, - l'9 ' could. I go back to. your. question about~ whether-. you 20 -need more capability in th'e inspection staf f ? 21 It seems to me. that you would implement- + '22 nonprescriptive regulation like-you' do prescriptive 23 regulation, with a 50.59, Part ~ 50.59 process. You i a 24 depend on that implementation of that process by the 25 licensee the same way you do today. My experience NEAL R. GROSS - q CoVRT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234M33 WASHINGTON. D.C 20005 (202) 232 6600

i cc,<. 93 1 with implementation of 50.59 is that it takes quite a 2 lot of technical skill by licensees today to know when i 3 they have an unreviewed safety question. 4 Many licensees are required to run their 5 own analysis codes. Some of them are very 6 sophisticated codes. The licensees that choose not to 7 do that themselves, there are NSSS suppliers or others 8 that supply. those analysis capabilities for them. 9

And, the fact-that you're moving to a

more 10 analytically.. dependent form of regulation I wouldn't 11 think would be particularly troublesome. I think that 12 licensees can provide that kind-of skill,today. They s 13 could provide it in the future. That's why I 14 mentioned training. They may - not have all of the 15 skills today that would be required by that different. .16 kind of analysis that we're talking'about. i 17 Then, I'think NRC would,need two kinds'of o .18 insp'ectors: one to make sure that the process, the i n. the determination of 19-50.59 process, theLreview of 2'O unreviewed saf ety questions', that kind of thing, was Y 21 being followed, that is somebody-to make sure the n 22-records were there, that the organizations were in a i 23 place, that there was paperwork to show that the ';- g 1 ,24 organizations were reviewing the right questions using 25 the right codes; then, there might have to be another NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W. (202) 234-4433 - WASHINGTON, D C. 20005 (202) 232 6600 -

4 94 1 level' of inspector who knew how to do that kind of 2 analysis himself, that would on a less frequent basis j 3 make sure that the right technology was being applied 4 by licensees in reviewing changes to their plants. 5 COMMISSIONER REMICK: _Very good. 6-I appreciate your mini history of the 7 development of prescriptive regulations in the AEC and 8 the NRC, and certainly agree. You basically said it i h 9' was inevitable that we became more prescriptive, y 10 because we were very general, and for many reasons you 11 gave they became more prescriptive. Some of the 12 reasons were that new applicants weren't familiar with-13 what the AEC or NRC intended by their general 14-regulations, and you also mentioned that there were E 15 later developments in technology or methods like PRA. 16. I wonder, then, if you drew a curve of-17. amount of-prescriptiveness in regulations versus time, 18 if it wouldn't increase with ' time. But, should it 4 .x 6 19 decrease at some point, then, _as people ---more, people 20 know. what it is the Agency intended by -its general 21 regulations and we develop new tools like PRA, 4 r-22 statements of saf ety' goals and so forth? If that 23 curve - should bend over with time, should it or, 24 should it always increase? Should it level off?- Have L 25 you given any thought to that? NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHOOC ISLAND AVENUE. N W. (202) 234 4433 WASHINGTON, O C. 20005 (202) 232 4600 - i,,

95 r. 5,. .=.- 1-DOCTOR MATTSON: 'Yes. I had one idea of 2 that that I was going to mention, but I left it out 3 and I'm reminded of it now. 4 I think high energy line-break is a little 5 bit that way, the decision in the early '70s to go 6 with arbitrary pipe breaks and then to describe i 7 exactly where those pipe breaks are going to-be. And 8 then, the plants became very rigid over the '70s and 9 industry put its heads together and i.ame up with a-- 10 .or helped NRC come up *dith a technical basis for 11 allowing-credit for leak before break and saying that

12:

there are ways to protect against this that are 13 ' preferable to the prescriptive way that was developed 14 in' Reg ' Guide.1.46~ and some letters that went out in-15 connection with that in the early '70s prescribing 16. exactly;how to do this with esary kind of pipe in the .17 plant. .Instead, an equivalent level of protection- '5 .18 taking into account the bad effects.of rigidity came 19 into being in the '80s and the commission amended its 20' rules to'take account for that, so in a sense that's '21 = moving back from the overly-prescriptive, based on 22 lessons learned. 23 COMMISSIONER REMICK: Do you think we've 24 done enough of that? '25 DOCTOR MATTSON: Think we ' ve. done enough NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N.W. (202) 2344433 WASHINGTON, D.C. 20005 (202) 232-6600 . =, _ _ _

fi. i i / * !... l. 96 1 if that? 2 COMMISSIONER REMICK: Enough of that type 3 of thing, that example you gave that we've -- l 4 DOCTOR MATTSON: Oh, no. I am convinced 5

that, if you apply this technique to advanced 6

reactorr, you will find places in your current 7

regulations that you will want to waive through 8 rulemaking because this approach shows you better ways 9 to design those plants. No question that this will 10 show you more systematically, more uniformly, in a 11-more optimal way hcw to build plants than the old 12 approach. It has to, because the old approach grew up 13 piece by piece through many different Commissions, 14 th' rough many different members of the staff over-a 15 long period of time and it's a patchwork, 16 DOCTOR ROSSIN: I can support that 17 strongly. I was deeply involved with that whole 18 process et EPRI throughout the early '80s. There is a. and I think targets of c19: . point where I think 20 opportunity like this-are there-and good work can be 21 done that can move this in that direction. It's not 22. an across the board thing. -It's selective and should J; 23 be applied where it makes sense. 24 I think one of the things that could be 25 done better and could have been done better in that NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N.W. (202) 234-4433 WASHINGTON, D c. 20005 (202) 232 6600

97 t 1 ' case and I think will be done better in the future is 2 that as examples like this come up I think the 3 Commission can adopt much less rigid ways of 4 implementing these improvements and changes. It seems 5' to me that that was a_very good example, but the l .A 6 amount of analysis that had to be done in order to get ? 7 back to a reasonable level where everybody knew they 8 had to be was extremely huge, took an awful lot of 9 time and probably could have been done in a much more 10 streamlined way. l{ 11 I think the negotiations between the 12' industry and the Commission-over that period of time 13-moved in-the rightMdirection, but it was very hard I 1 l. 14 believe for the commission to exercise the flexibility 15 that really they could have exercised as long as they y 16 were'very clear on what they were doing and documented 17 'the' basis on why they were doing it; I think we could 18 have moved-ahead much faster. i m COMMISSIONER REMICK: I would thank 'all 19 20 four of the speakers'for a very -interesting and 21 Linformative -- I know you've all spent a considerable l' I 22 . amount of time' giving - thought to this and it's been 23 very helpful with the different insights that you've -1 L, 24 provided. l 25 That's all, Mr. Chairman. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISuND AVENUE, N W. (202) 2344433 WASHINGTON, D.C. 20005 (202) 232 4600 i

l 98 1 C O M M *.'S S I O N E R ROGERS: Commissioner i 2 Curtiss? l 3 COMMISSIONER CURTISS: I guess I have a v 4 lot of questions, but I'm not going to ask them - all-5 since it's after lunchtime. 6 Roger, let me just ask you one question. 7 You talked about the potential for using an approach 8' like this in the maintenance area. I wonder if you 9 could cmballish on what you see as a nonprescriptive-10 based approach to maintenance along the lines of what 11. you envision? 12 DOCTOR MATTSON: Let me' try. I have a 13-f airly' simplistic view of it, so you'll have to 14 forgive me for being somewhat distant from it. Maybe 15 it will help. 16 You want better assurance that plants are 17 maintained in-the state that ' you licensed them and x 18 that you f eel comfortable about them. The industry 19 seems not to want you to meddle in their business any-t p more than you already do. It seems to me that you 20 3 4 21 could speak to the process by which maintenance is l 22' provided 'in' nuclear plants and speak to the p l. 23 ' performance you expect out of maintenance in keeping a 24 plant in the condition that you understood it to be L 25 when you 'icensed it witheno getting into the details. NEAL R. GROSS i COURT REPORTERS ANO TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234 4 433 WASHINGTON, D C. 20005 (202) 232 4600

99 = ' l-1.

And, to me.

that's a form of nonprescriptive =2 ~ regulation. 3-I think that's the same thing that the 4-Commission did when it decided not to get prescriptive 5 about occupational exposure. It just said, "We want 6 .to see exposures go down, individually, collectively, 7 across the industry and at individual plants." I 8 believe you'_ve got some record to show that they have 4 9 gone down. -10 COMMISSIONER ROGERS: You've got a 11 measurable number that you can point t o ', a . 12 . quantitative result, and with maintenance that's part 13' .of:the big problem. '14 DOCTOR MATTSON: Okay. I understand that, 15 and that's one of' the reasons'I brought reliability 16 -- assurance to the table.again today, because an element, ~ I <17~ of ' a reliability assurance. program is maintenance. 18 There's also quality = assurance. There's also 19 . replacement of parts.or whatever, housekeeping, 20-whatever goes into _ assuring the reliability of an C '21 -already built system. p 22 Maybe the ' way to go at maintenance is 23 along with some other things, quality assurance during 24 . operations, maintenance, housekeeping. A lot of the '25 ex-Naval. officers that show up on the Commission seem NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, D.C. 20005 (202) 232 4600

100 'm l' to. have a strong interest in housekeeping. All of 2 those things go into making up a reliability assurance 3 program'and maybe you want to go back and revisit that 4 idea and.say all of these things are aimed at making 5 certain systems be of high reliability, the ceram 6 tsystem for example, and we have an interest in the way 7 neram systems and certain other safety systems are 8 riaintained and-the only way we know to measure that is 9 through. some reliability, some performance standard 10 for those systems. Help us develop what constitutes a 11 reasonable performance standard for your plant. 12 The Hendrie Rule said let the plants 13 themselves develop the performance standard, NRC only e, 14-look at the methods that they were using to develop-15 the standard, because it will-vary design by design-- 16 that is, the performance system for the scram system. '17 And you could, through that door, influence not only 18 maintenance but-also housekeeping and quality 19 assurance and' replacement of parts and surveillance 20 testing and a variety of things that go into a 21 reliability assurance program. 22 PROFESSOR GOLAY: You know, I think one of 23 the problems that you get into with, say, if we take 24 the maintenance or scram reliability thing is that in 25 a lot of these things there's judgement required. And NEAL R. GROSS CoVHT REPORTERS AND TRANSCRIDERS 13M RHOOE ISLAND AVENUE, N.W. (202) 234 4433 WASHtNOTON. D.C. 20005 (202) 232-6600

101 1 one approach that could be suggested is, well, let's 2 do what seems reasonable to try and push people 3 through ' th'eir_ voluntary efforts acting in their own 4 self interests to do things which would give a good 5 result and I've always felt that one of the reasons a 6 why you r.eed a formal process or the kind that we're 7 talking about today is that it is in fact difficult 8 for the Commission to act with discretion where 9 there's_ judgement involved because of the spectrum of i 10 licensees involved. You always-have to worry.about 11 - those who won't perform in a reasonable way, even 12 .though the majority may do so. And so, overall,'I 13 think tha't-would be one good reason for trying to get i 14 some experience with a more formal functionally-t 15 oriented approach. 16

Well, that's as much as. I'll say about 17 that.-

Y 18 COMMISSIONER CURTISS: Well, I guess the v, 19 concept that you've articulated here, I. thought m =20 ' Marshall Breger and, Roger, you did a good job of 21 summarizing some of the advantages and disadvantages, m V q 22 We'll al1 go back to our offico now and we've got '23' specific initiatives sitting on our desk and. it 's a 24 question about how do we come up with a performance-25 -based approach ' to fitness for duty or fuel assemblies NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON. DC 20006 (202) 232-6600 L d m y ;(;, '

w ..q. 102 1 that get hung-up on the top or on the bottom of ) i 2 reactor internals or day to day problems. I do, I 3 think, share a number of the desires that you all have 4-articulated. 1 5 I guess I look at the issue with I guess a l 6 healthy dose of skepticism in terms of the 7 disadvantages. It does seem to me that the kinds of 8 problems that you get into that Mr. Breger identified, 9 the additional time that it will take to develop the 10-

approach, the additional detail that the record is 11 going ~co have to be developed in a n d,- - t h i r d,

the 12 particular point that we've discussed in the context - 13 of the. regulatory impact,- the extent to which the . 14 approach lends itself to interpretation. We spent two 15 hours here two weeks ago talking about what a problem i 16 it'is that our regulations are interpreted in such a j 17 diverse range of ways by the inspectors and by-the - 18 regions and so forth, and the licensees are looking at l i 19 least-in a lot-of respectc for some clear and uniform 20 answer to regulatory questions. l 21-I do happen to believe that the approach, - 22 Doctor Mattson, that you've suggested on maintenance 4 23' is one that deserves a careful look. It's in that 24 area, perhaps, as we have in training, found that difficult to 25 it's -- my own personal view here NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234-4433 WASHINGTON, D.C. 20005 (202) 232 4600 l t

103 1 prescribe an approach that would work for everybody in 2 the maintenance area. You go to plants where the 3 maintenance program seems to work because you've got 4 people who-pass down the maintenance program from year 5 to year and from decade to decade. You go to other 6 plants where it works because it's got a lot of 7 procedure and a lot of paper and everybody follows the 8 procedure and'the paperwork. 9 Stepping

back, though, and asking 10 ourselves what is it about maintenance that we would 11 like to ensure, I guess, and maybe oversimplifying it 12

'beyond what you said, you'd like to see the system ^13 perform when called upon. We can all sit -down and 14 identify the various systems at a given plant that are RHR and HPSI and go down the list-- 15 key systems 16 and I suspect every plant in this country could sit 17 down with our people and identify the key systems that 18 are important to that plant. '19 I guess the question 'is whether you can '20 establish -some sort of performance-based results-21 oriented reliability goal that you'd like to see those 22 systems-meet and it's for that reason.that we've asked 23 che staff.to take'a look at that very question.. Can 24 you develop a regulatory framework around that 25 principle, results oriented, as opposed to what we've NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE, N W. (202) 234-4433 WASHINGTON, D.C. 20005 (202) 232 4600

104

e.,

1 sort of loosely referred to here as a prescriptive 2 type approach. I do think this has some merit. 3 Dave? 4-DOCTOR ROSSIN: Jim, you said you were 5 going to look at something and see how to apply it. 6L First, the question is,.is this a case where it should 1 7 apply? A couple of the examples you gave are, and a .8 couple of the examples I would say, no, you don't 9 apply it there. 10-You know, the Commission has been arguing 11 among itselves about a maintenance rule for about 12 seven or eight years now, I mean the current wave of. 13 argument about maintenance rule. And, meanwhile, the 14' industry is maintaining plants. They've got-to do it. 15 It would seem to me that we may be reaching a point 16 here with the maintenance rule where a W, o f things 17 are in place now. 18 I think one of the things that the 19 ' industry over the years has been concerned about is a 20 rule coming in and undoing. something that they've 21 worked very hard to make work in their own 22 circumstances, and that's why this may be a very good 23 example of where details and prescriptive.. stuff coming 24 f rom the Commission would be counter-productive at 25 this point. You've got to really ask yourself what is NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE, N W. - (202) 234 4433 ' . WASHINGTON, D.C. 20006 (202) 2324600

105 i.- 1-it we're trying to accomplish here. 2 PROFESSOR GOLAY: Let me go to one other 3 thing, and that-is another thing that is going to be 4 on your table that you just can't avoid is severe 5 accidents, because right now the polici.a in place are 6 so much in disconjunction with what people think they 7 can achieve technologically and the kinds of arguments 8 that are going to be offered to you on advanced 9 reactors that I think you're unavoidably going to have 10 to be forming some sort of policy, a new severe 11 accident policy, and it's< the heart of safety. You 12 know, if you look at how the public is threatened by 13. nuclear power, reactors, it's through severe accidents. 14 And so, I think-that you're basically not going to be 15 able to avoid this question of what's the right, the -16 logically correct way to approach nuclear safety.

17

-In this discussion, we 've got into a L18 spirit of saying, "Well,'if we go to nonprescriptive - 19. regulation, we throw out all of the beneficial aspects 20 of deterministic criteria," and I think it's important 21 not-to accept that thinking because in truth you're 22-going t o: have to have a large component of 23 deterministic standards.-because uncertainty is. an ' mportant aspect of the question. And so, first, I'd i 24 25 like to make sure that that clarification is accepted. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVENUE N.W. . ASHINGTON, D.C. 20005 (202) 232 4600 W (202) 234-4433

106 1 The second is I want to build on Mr. 2 Breger's point, which is that when he compared these 3 'different examples the main argument that came down on were essentially-4' the prescriptive approach was 5 arguments of expediency in terms of decision-making, 6 timeliness, clarity in a detailed way. And his most 7 telling point I thought was that when you end up in a 8 situation where.you have a plethora of issues which 9-are linked to one another and must be handled in this 10 way, that's when the deterministic approach collapses. c 11 And that's what we've got with nucler.r power, and 112 that's the reason, a fundamental reason why I think we I 13 really ought to explore this and see what we can do to 14 .make it work. t 15 The important point, though, is that only 16 .the'NRC can change itself. 17 COMMISSIONER ROGERS: Did you mean 18T deterministic or prescriptive? 19 PROFESSOR GOLAY: See, the prescriptive i p 20 ~ approach we have has been deterministic wholly, but 21 the two are not synonymous. That is, telling people 1 1 5 '22 what'you expect from a particular system out of v.a ' 23 context and doing this many times over and over is P' m 24 what I would call the prescriptive approach, doing it 25-in fine detail, as opposed to starting from a global NEAL R, GROSS yi COURT REPORTERS AND TRANSCRIBERS 1323 RHOOE ISLAND AVENUE, N W. (202) 734 4433 WASHINGTON, D.C. 20005 (202) 232 4600

I 107 l 1 performance goal and then trying to work that down t 2 into a set of decision rules th.t people can use, 3 which is the fundamental goal here. 4 COMMISSIONER CURTISS: That's all I have. 5 COMMISSIONER ROGEk$1 We'1, I've got a 6 long list too, but I think I've run out of time. So, 7 I'm going to just ask one question because I know the B Chairman had an interest in it. 9 What do you see as the impact of these 10 ideas on standardization? I think Professor Golay and 11 Doctor Rossin might both respond to that. I think you 12 may have different -- you ' ve said some things about 13 different aspects of it. For example, how does one 14 maintain a program of configuration control in the 15 absence of some prescriptions? 16 DOCTOR ROSSIN: Well, I think we agree-- I 'l,. at least as far as I know, Michael, we agree -- that, 18 in answer to that question, I think it's very obvious. 19 Configuration control means configuration control. It 20 does not mean don't change anything or don't fix 21 anything, don't replace anything and don't make things 22 work. It goes back to what it is you're trying to do. 23 You're trying to run plants safely, and configuration i 24 contrM means you keep a record of everything you do 25 so that you know what you've got out there. You know NEAL R. GRCSS COUAT REPORTERS AND TRANZRIBERS 1323 RHODE ISLAND AVENUE, N W. (?02) 234 4433 WASHINGTON, D C. 20006 (202) 232Roo 4

108 1 the configuration that you're dealing with. 2 There is no reason to confuse 3 tonfiguration control with a stan.1+cdized plant that 4 can't be modified and cl.'.nged. Standardization is in 5 order to certify the safety of a design. Once that 6 design is certified, if it's built, you're going to 7 have to ope. ate it. You're going to have to maintain 8 it. You're going to have to modify it and change it. 9 Every nuclear plant in the country probably has 100 10 mods going on right now and there's no way that that's 11 going to change if we're going to have nuclear power, 12 and that has to be understood going in. 13 PROFESSOR GOLAY t Let me add to that. I 14

think, if anything, this approach would be more 15 93reeable to standardization and maintenance of a good 16 database.

As Dave pointed out, the essential thing is 17 to have a record, and for the kinds of analyses which 18 we're talking about you have to have a much more 19 thorough record. And implicit in using this approach 20 is over the plant life refining your analysis as ycu 21 learn more about your plant and as the database from 22 experience changes, so in fact I think it would be an 23 inherent aspect of it. Northeast 24 We have found with utilities 25 Utilities is a good example -- where they have gone to NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIDERS 1323 RHODE ISLAND AVENUE, N W. (202) 234 4433 WASHINGTON, O C. 20005 (202) 232 4000

109 1 PRA basis for their safety management and their 2 negotiation with the Agency that they have reported 1 1 3 that one of the big benefits of doing that is they i 4 have a much better understanding of the nature of i 5 their plant and the things which they think should be 6 done. That's part of why I make my suggestion to 7 focus on the standardized plants as the primary target 8 or the first target here. 9 COMMISSIONER ROGERS:

Well, I'd like to 10 thank you all for coming today and providing the il commission with this informative briefing.

I know you 12 all took time from busy schedules and that some of you 13 also traveled a considerable distance to ioin us and 14' we appreciate your willingness to assist us as we 15 begin to consider this potentially important departure 16 from our traditional way of doing business. 17 You've given us considerable food for 1 18 thought. We must now digest what you've given us 19 about different approaches to nonprescriptive 20 regulation and the areas of applicability of each and 21 about legal, institutional, and other considerations. 22 We must assess both the potential benefits and 23 possible drawbacks you've identified. Our challenge-i 24 will be to select the most appropriate type of 25 regulation for different areas of Agency NEAL R. GROSS CoVRT REPORTERS AND TRANSCRIDERS 1323 RHoDE ISLAND AVENUE, N W. (202) 2344433 WASHINGTON, D C 2000$ (202) 232 4600

110 1 responsibility and to avoid the pitfalls you've 2 mentioned. 3 We're particularly interested in examining 4 the potential use of nonprescriptive regulation for 5 the advanced reactor designs and f or maintenance, so 6 we'll be giving strong consideration to the design 7 approaches that might be appropriate in the former 8 case and to performance measures that might be 9 suitable in the later case. 10 In the longer-term, we may want to 11 consider other areas es possible candidates for 12 nonprescriptive regulation and even perhaps consider 13 converting some of our existing prescriptive 14 regulations to a less prescriptive form. Prescriptive 15 regulation may remain necessary in some areas, while 16 different types or degrees of nonprescriptive 17 regulation. may be appropriate in other areas. We'll 18 have to address not only what is best in each case. 19 but how a body of regulations that may incorporate 20 different degrees and types of prescriptiveness can 21 work effectively as a whole. 22 Thank you again for your help in getting 23 us started on this process. 24 Do any of my fellow Commissioners have any i 25 other remarks? d NEAL R. GROSS COURT REPORTERS ANC TRANSCRIBER $ 1323 RHODE ISLAND AVENUE. N W. (202) 2344433 WASHINGTON, D C. 20006 (202) 232-6600

4 i I 111 -1 1 We stand adjourned. 1 .I 2 (Whereupon, at 12:22 p.m., the above- ) i 3 ' entitled matter was concluded.) I l 4 i 5 6 'j i 7 1 .c j 9 i ( 10 i 11- ^ 12 13 1 14 -i 15. 116 ~ 17- .ig: 1 b - i t f: 20. 4 -21 l- ' 3 22-t 23' -24 '25-NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS ll l'- .1323 RHODE ISLAND AVENUE. N W. - l (202) 2344433 ' WASHINGTON, D.C. 20006~ '(202) 232 8000 l '- l'

CERTIFICATE OF TRANSCRIBER This is to certify that the attached events of a meeting of the United States Nuclear Regulatory Commission entitled: TITLE OF MEETING: BRItFING ON NONPRESCRIPTIVE NUCLEAR SAFETY REGULATION PLACE OF MEETING: ROCKVILLE. MARYLAND DATE OF MEETING: OCTOBER 30, 1990 were transcribed'by me. I further certify that said transcription is accurate and complete to the best of my ability, and that the transcript is a true and accurate record of the foregoing events. OA %v u Reporter's name: Peter Lynch k HEAL R. GROSS COUtf 99900T19$ AND TRANSCRittR$ 1313 BMODE ISLAND AYINUt, N.W. (202) 234-4433 WA8HINGTON, D.C. 20005 (202) 232 6600

10/30/90 SCHEDULING NOTES

Title:

Briefing on Nonprescriptive Nuclear Safety Regulation Scheduled: 10:00 a.m., Tuesday, October 30, 1990 (OPEN) Duration: Approx 1-1/2 hrs

Participants:

- Professor Michael W. Golay 30 mins Massachusetts Institute of Technology - Mr. Marshall J. Breger 20 mins Chairman Administrative Conference of the United States - Dr. A. David Rossin 20 mins President Rossin and Associates - Dr. Roger Mattson 20 mins Vice President Scientech, Inc.

NON-PRESCRIPTIVE REGULATION OF-FUTURE NUCLEAR POWER PLANTS presented before U.S. NUCLEAR REGULATORY COMMISSION by Professor Michael W. Golay Department of Nuclear Engineering Massachusetts Institute of Technology Cambridge, MA 02139 Tel: 617/253-5824 Fax: 617/258-8863 30 October 1990

LIMITATIONS OF CURRENT NRC REGULATORY APPROACH IS HIGHLY PRESCRIPTIVE, OFTEN PRESUMING THE NATURE OF PLANT FEATURES PRODUCES NON-UNIFORM SAFETY RESULTS 9 PRODUCES UNCERTAIN LEVELS OF SAFETY i INHIBITS ASSUMPTION OF FULL RESPONSIBILITY FOR SAFETY BY LICENSE HOLDER FAILS TO ENCOURAGE SAFETY IMPRO.VEMENTS INillBITS TECHNOLOGICAL INNOVATION FAVORS REPLICATION OF EXISTING TECHNOLOGIES AND ANALYTICAL APPROACHES BOMBARDS NRC COMMISSIONERS WITH BURDEN OF DETAILED DECISION-MAKING, THEREBY INHIBITING TIMELY FORMULATION POLICY k n ~s-,,m

NEED FOR NON-PRESCRIPTIVE REGULATION PRESCRIPTIVE, DB A-B ASED, REGULATION WAS USED IN THE PAST FOR LWRS, HTGRS AND LMRS NON-PRESCRIPTIVE REGULATION LONG RECOGNIZED BY NRC AS LOGICALLY ESSENTIAL FOR NEW REACTORS, e.g.,1985 TESTIMONY OF COMMISSIONER PALLADINO BEFORE HOUSE ENERGY R&D SUBCOMMITTEE OF SCIENCE COMMITTEE LITTLE PROGRESS TOWARD NON-PRESCRIPTIVE REGULATION ACHIEVED TO-DATE THE CHANCE OF A GENERATION TO RATIONALIZE NUCLEAR SAFETY REGULATION WILL BE MISSED IF NRC REMAINS PASSIVE t

REGULATORY SITUATION OF NEW REACTORS NEw cWRS (EVOLUTIONARY AND PASSIVE):

• BEING DESIGNED IN ACCORDANCE WITH EXISTING REGULA-TIONS i
• MAYBE WILL ENCOUNTER NEW REGULATIONS FOLLOWING NRC-EPRI NEGOTIATIONS I
• FACING REGULATORY VACUUM REGARDING SEVERF JENTS i

i NEW HTGRS AND LMRS: l PRELIMINARY CONCEPT REVIEWS FINISHED l

• MOST RELEVANT PRECEDENT AVAILhBLE IS FROM:

FT. ST. VRAIN CLINCH RIVER i

• FACING REGULATORY VACUUM REGARDING SEVERE ACCIDENTS 1

t

---

.-m m

..-..-,m,

RECO'MMENDED ACTION ESTABLISH AN INTERNAL NRC PROJECTTO FORMULA'tE A METHOD FOR NON-PRESCRIPTIVE REGULATION (USE EXTERNAL HELP AND REVIEW, AS NEEDED) FOR PROPOSAL TO COMMISSIONERS PROJECT SCOPE

• ADEQUACY OF SAFETY GOALS

. REGULATORY REVIEW APPROACHES AND DECISION CRITERIA . DECISION TOOL 3 - MODELS AND DATA - MEANS FOR REVISION - METHODS FOR VERIFYING COMPLIANCE THROUGHOUT PLANT LIFE

• PROPOSALSTOCOMMISSIONERS

. PROJECT REVIEW METHOD OF WORK, PARALLEL EFFORTS MUTUALLY REFINING EACH OTHER FOR:

• FORMULATIONS OFGENERAL APPROACHES

APPLICATION TO A SPECIFIC POWER PLANT EXAMPLE

__-__-..--.s BASIS OF NON-PRESCRIPTIVE REGULATION: NRC SAFETY GOALS, INCLUDING FUTURE AMENDMENTS ) THE SAFETY GOALS ARE STATED IN TERMS OF EXPECTED RISKS. THUS, SOME USE OF PROBABILISTIC RISK ASSESSMENT (PRA) IN NON-PRESCRIPTIVE REGULATION IS IMPLIED. l o L SAFETY GOALS CAN GUIDE FORMULATION OF RISK ACCEPTANCE l l CRITERIA. s a e 9 -I 5 o- ~ -.ew-v-m-

1 NECESSARY FEATURES FOR PRACTICAL PROBABILISTIC RISK ASSESSMENT (PRA) USE STANDARDIZED, NRC-APPROVED:

• Data Bases
• Models
• Means of Ref~ ming Data Bases and Models w

STANDARDIZED METHODS FOR ASSESSING ACCEPT-ABLE COMPLIANCE IN TERMS OF:

• Design

. Operations

• Evolving Industry Experience BOTH EXPECTED RISKS AND ASSOCIATED UNCER-TAINTIES MUST BE ADDRESSED WHEN USING PRA-

I ALTERNATIVE ACCEPTANCE CRITERIA ) RISK ACCEPTANCE CRITERIA COULD BE STATED IN VARIOUS . WAYS, e.g., MEAN, MEDIAN OR CONFIDENCE-LEVEL BASED i DECISION CRITERIA NEED NOT BE RIGID, e.g., RIGID CRITERION: EXPECTED RISK 2 LIMIT =$ REJECTION EXPENDED RISK < LIMIT => ACCEPTANCE FLEXIBLE CRITERION (USES A BAND OF POTENTIALLY ACCEPT-ABLE PERFORMANCE): EXPECTED RISK > UPPER LIMIT => REJECTION EXPECTED RISK'< LOWER LIMIT => REJECTION EXPECTED RISK BETWEEN UPPER AND LOWER LIMITS =* FURTHER INVESTIGATION AND JUDGMENT i ( 4

[. EXAMPLES OF ALTERNATIVE METHODS OF ) REGULATORY IMPLEMENTATION I. PROBABILISTICALLY-BASED DETERMINISTIC EVALUATION OF SAFETY: STAGE 1 - PERFORMANCE OF LEVEL 3 PRA i STAGE 2 - EXAMINATION OF PRA RESULTS FOR DESIGN VULNERABILITIES STAGE 3 - USE OF STAGE 2 RESULTS IN TERMS OF SAFETY l GOALS TO IDENTIFY A SET OF DETERMINISTIC TRANSIENTS, DBAS AND SEVERE ACCIDENTS FOR DESIGN SAFETY ASSESSMENT, BIASING ASSUMED I ACCIDENT FREQUENCIES TO ACCOUNT FOR-L UNCERTAINTIES STAGE 4 - ANALYSIS OF PLANT PERFORMANCE IN TERMS OF STAGE 4 RESULTS STAGE 5 - IDENTIFICATION OF AREAS WHERE PLANT IS DEFICIENT (WITHOUT GIVING GUIDANCE REGARDING HOW TO FIX THE DEFICIENCY) o I - - -m m . m --+.....-w...m.- w.m.-e,- +,w. -..,v._. -,w w

L EXAMPLES OF ALTERNATIVE METHODS OF REGULATORY IMPLEMENTATION II. COMBINED PROBABILISTIC AND DETERMINISTIC EVALUATION OF S AFETY: STAGE 1 - PRA-BASED RISK ANALYSIS WITHIN RANGE OF CASES WHERE ACCEITABLE ACCURACY CAN BE OBTAINED (e.g., LEVEL 1 PRA) STAGE 2 - DETERMINISTIC ANALYSES OF A SET OF CONSERVATIVELY-FORMULATED ACCIDENT SITUATIONS AND EXPECTED FREQUENCIES, EXTENDING BEYOND THOSE ADDRESSED AT STAGE 1 (TRANSIENTS, DBAS AND SEVERE ACCIDENTS) STAGE 3 - COMPARISON OF COMBINED RESULTS TO SAFETY GOALS STAGE 4 - IDENTIFICATION OF AREAS OF DEFICIENCY (WITHOUT GIVING GUIDANCE REGARDING HOW TO FIX THE DEFICIENCY) 4

EXAMPLES OF ALTERNATIVE METHODS OF REGULATORY IMPLEMENTATiT j i i III. PRA-BASED REGULATION OF SAFETY: STAGE 1 - PRA-BASED SCREENING OF ACCIDENT SEQUENCES AND CONSEQUENCES WITH RESPECT TO S AFETY GOALS t L STAGE 2 - EXAMINATION OF DESIGN VULNERABILITIES, e.g., OMirrED ACCIDENT SEQUENCES VULNERABILITIES (e.g., COMMON MODE, HUMAN FAILURES) OPTIMISTIC DATA & MODELINO ASSUMPTIONS STAGE 3 - IDENTIFICATION OF AREAS WHERE THE PRO-- POSED PLANT IS DEFICIENT (WITHOUT GIVING L GUIDANCE REGARDING HOW TO FIX THE DEFICIENCY)

IMPORTANT QUESTIONS CONCERNING THE PRACTICAL IMPLEMENTATION OF NON-PRESCRIPTIVE REGULATION How will non-prescriptive regulation b!cnd use of deterministic and probabilistic regulatory criteria in different areas of decision making and enforcement? How will the NRCs traditional approach to the regulation of design, construction and operational aspects of a plant be changed? What is the rationale for " defense-in-depth" and how is it to be implemented and maintained? What other deterministic consideration (such as redundancy and diversity) shall be maintained? Where is non-prescriptive regulation consistent, or not consistent, with current NRC policies on standardization, backfitting, severe accidents, safety goals? - What industry acceptance / preference exists for non-prescriptive regulation? How will NRC oversight ofinspection and enforcement change under non-prescriptive regulation? What documentation will a license applicant submit for review? How does the proposed framework improve safety and why?

ASPECTS OF SUGGESTED WORK . THE BEST NON-PRESCRIPTIVE REGULATORY APPROACH IS NOT OBVIOUS . AN INVESTIGATION OF ALTERNATIVE APPROACilES APPLIED i REALISTICALLY IS NEEDED TO IDENTIFY BENEFICIAL REFINEMENTS i . USE OF PRA IS OFTEN RESISTED BECAUSE SOME RESULTS ARE SIGNIFICANTLY UNCEF.TAIN i . THE UNCERTAINTY REVEALED THROUGH PRA IS NOT DIFFERENT FROM TH AT INHERENT IN THE CURRENT REGULATORY APPROACH i

• NON-PRESCRIPTIVE REGULATIONS AND IMPLEMENTATION CAN BE REFINED AS KNOWLEDGE IMPROVES

. THE CURRENT REGULATORY APPROACH IS DIFFICULT TO . REFINE 1 ? ...m. .r-.. m

) i n POTENTIAL BENEFITS OF NON-PRESCRIPTIVE REGULATION (APPLIED TO NEW POWER STATIONS) j ) GREATER, MORE UNIFORM S AFETY SAFETY RESPONSIBILITY AND AUTHORITY RESTING WITH LICENSEE SWIFTER, CLEARER, MORE CONSISTENT REGULATORY DECISIONS INCREASED ABILITY OF NRC COMMISSIONERS TO PROVID'E .c POLICY GUIDANCE j GREATER TECHNOLOOICAL INNOVATION t

-...- _ = ~ - - _.. -. i 1 PERPORMANCE BASED REGULhTION f A. DAVID ROSSIN EEPORE TEE U. S. NUCLEAR REGULATORY COMMISSION j TUESDAY, OCTOBER 30, 1990 2 PRESCRIPTIVE REGULATION ^ FOCUS ON MEAMS l PRESCRIPTIONS ON HOW THINGS ARE DONE 3 NON-PERSCRIPTIVE REGULATION FOCUS 'ON IJ[DE i PRESCRIPTIONS STILL NEEDED 4 EEANFLES OF HPRESCRIPTIONSH SAFETY GOAL i DESIGN CRITERIA O PERFORMANQE OIL 7ECTIVES, SYSTEM FUNCTIONAL REQUIREMENTS 1 COMPONENT PERFORMANCE REQUIREMENTS 5 SOMEONE EAS TO WRITE RACE PRESCRIPTION PRESCRIPTIVE: NRC NRITES NON-PRESCRIPTIVE: LICENSEE WRITES L 5 NO ACCIDENTS OR SERO RISK AS & GOAL l " YEARNING" MAY BE UNATTAINABLE IN PRACTICE l OPEN-ENDED l l - + -. --~.,w.. _. -.....,,,.,.,a a_. .,,..w .--..v.,mm,.....,,, ,,,y,.w_,e ,g. ,,,e

UNDERSTOOD THAT ATTAINMENT CANNOT BE MEASURED IN ADVANCE 7 TCP-LEVEL ON ECTIVES SOCIETAL VALUE JUDGHENT BENEFIT / COST / RISK BALANCE ALTERNATIVE TECHNOIDGIES RISK OF BEING WITHOUT ADEQUATE ENERGY 8 AN ACCIDENT DOES NOT INDICT TEE TECHNOLOGY SERIOUS ACCIDENT NOT " ACCEPTABLE" IDW PROBABILITY EVENTS DO HAPPEN 9 LOW PROBABILITY EVENTS ARE TOLERABLE UNLESS TREY RAPPEN 10 TCP-LEVEL SAFETY O N ECTIVES SET BY NRC OPEN PROCESS WITH PUBLIC COMMENT 'o e 11 %CWER-LEVEL CRITERIA TO IMPLEMENT TCP-LEVEL ON ECTIVES SET BY NRC AND LICENSEE GUIDANCE, NOT REQUIREMENTS 13 CRANGES TO PRESCRIPTION SY LICENSEE BURDEN OF PROOF THAT REVISED PRESCRIPTION - WILL MEET SAFETY OnkTECTIVES AND/OR FUNCTIONAL PERFORMANCE OMECTIVES 13 CRANGES SOUGHT BY NRC OR OTHER PARTY BURDEN OF PROOF ON PROPOSER TO SHOW THAT CHANGES ARE NECESSARY 4 m.__ m

t - i 14 RISR POLICY . RISK ASSESSMENT RISK NANAGEMENT 15 ROLE OF NRC EVERYTHING IT DOES NOW EXCEPT PRESCRIBE 16 APPLICATION TO STANDARDISATION PERFORMANCE-BASED REGULATION OFFERS A VIABLE NAY TO MAKE NECESSARY CHANGES 70 LICENSED STANDARDIZED PLANTS 17 F.TANDARDISATION: 1 IMPORTANT FOR SAFETY CERTIFICATION AND LICENSING 1 l 18 STANDARDISATION: ) i POTENTIAL VALUE FOR IM*< ROVING RELIABILITY AND ECONOMICS i 19 STANDARDISATION: MQ1 IMPORTANT FOR THE SAKE OF STANDARDIZATION ITSELF o .... ~ to ONCE LICENSND: ', IMPORTANT THAT. PLANT NEET OE7ECTIVES I NOT THAT IT REMAIN " STANDARD" L

31. ASSURANCE OF SAFETY FOR STANDARDISED PLANTS

? L NON-PRESCRIPTIVE, PERFORMANCE-BASED REGULATION L L 32 PERFORMANCE-BASED REGULATION f-EVALUATE THE PERFORMANCE L OF LICENSED NUCLEAR POWER PIANTS PERF01GANCE-BASED REGULATION WOULD HOT COMPROMISE SAFETY, AND COULD FACILITATE PERFORMANCE IMPROVEMENTS . ~

3 33 ADVANTAGES OF PERFORMANCE-BASED REGULhTION STABILITY TLEXIBILITY DEFI' DES A REGULATORY END POINT BETTER DEFINITION OF ROLES 24 DEALING WITN UNCERTAINTY USE IDWER-LEVEL CRITERIA NEETING TOP-LEVEL OIL 7ECTIVE WITH REASONABLE CONFIDENCE, i NOT' ABSOLUTE CERTAINTY t 25 UNCERTAINTY 4 UNCERTAINTY MEANS LACK OF DETAILED KNOWLEDGE, NOT NECESSARILY INCREASED RISK 25 FURTHER EVALUATION OF NON-PRE 80RIPTIVE REGULATION' POTENTIAL BENEFITS OF IMPLEMENTATION OBSTACLES AND LIMITATIONS DEVE14P ttIT'LINE OF IMPLEMENTATION STRATEGY 27 DETAILED DEVELOPMENT OF NON-PRESCRIPTIVE REGULATION ESTABLISHING 14WER-LEVEL CRITERIA ' ACCEPTABLE TESTS BASED 0 " REASONABLE CONFIDENCE" '" REASONABLE CONFIDENCE" AS A BhSIS FOR COMPLIA'NCE INCORPORATING DEFENSE-IN-DEPTH ' DETAILING NRC ENFORCEMENT ROLE TRIAL APPLICATIONS

j l i NONPRESCRIPTIVE NUCLEAR l SAFETY REGULATIONS PRESENTATION TO: i U. S. NUCLEAR REGULATORY COMMISSION i l l ROGER J. MATTSON j SCIENTECH, INC. OCTOBER 30,1990 i I i SUDE 1 4 1

i i F i I I I I i l i l DEVELOPMENT OF PRESCRIPTIVE REGULATION l l I l 1960s: EARLY AEC REQUIREMENTS WERE VERY GENERAL - 1970s: PRESCRIPTIVENESS GREW AS INDUSTRY GREW j I - VARIED SUPPLIERS AND LICENSEES - IMPROVED TECHNOLOGY AND EXPERIENCE - HIGHER LEVELS OF SAFETY - 1980s: RESPONSE TO TMI AND USI RESOLUTIONS i suoe2

i i FORMS OF PRESCRIPTIVE REGULATION l WRITTEN REQUIREMENTS l ACTIONS OF INDIVIDUAL STAFF MEMBERS l ACTIONS OF THE STAFF i 1 ACTIONS OF LICENSING BOARDS l l [ ACTIONS OF THE COMMISSION i 4 i SUDE 3

^ i h REASONS FOR PRESCRIPTIVE REGULATION 'i l TIME PRESSURES l f ADVERSARIAL LICENSING PROCESS TECHNICAL COMPLEXITY OF ISSUES l VARIABILITY AMONG PLANTS RISING LEVEL OF SAFETY i l i SUDE 4 f i

a_ ,; jj ADVANTAGES OF PRESCRIPTIVE REGULATION LESS MANIPULATION IN ADVERSARIAL PROCESS LESS SUSCEPTIBLETO REINTERPRETATION BINDING RECORD OF NEGOTIATED SETTLEMENTS FIRM BASIS FOR CONFIGURATION CONTROL SLIDE 5 -

4 ? ADVANTAGES IN REGULATING SAFETY FUNCTION PERFORMANCE i k BETTER ACCOUNTING OF SYSTEM BEHAVIOR i IMPROVED TREATMENT OF PLANT SPECIFIC FACTORS PERMITS RATIONALIZATION OF OVERALL SAFETY GOAL l PLACES INCREASED RESPONSIBILITY ON LICENSEES i PROVIDES FLEXIBILITY FOR OPTIMlZATION i l SLIDE 6 i

e EXPERIENCE IN REGULATING SAFETY FUNCTION PERFORMANCE AFW RELIABILITY GOAL IN STANDARD REVIEW PLAN HENDRIE RULE PROPOSED FOR ATWS SCRAM SYSTEM RELIABILITY ASSURANCE PROGRAM SIZEWELL-B DESIGN IN UNITED KINGDOM INSAG-3 SAFETY PRINCIPLES OF IAEA RADIATION PROTECTION PROGRAMS FOR LICENSED NPPs SUDE 7

i I PREREQUISITES TO REGULATING l SAFETY FUNCTION PERFORMANCE s i PLANT-SPECIFIC PRAs FOR SYSTEM PERFORMANCE . CREDIBLE MAINTENANCE OF PERFORMANCE DATA DURABLE CONSENSUS ON ACCEPTABLE METHODOLOGY ABILITY OF LICENSEES TO IMPLEMENT NEW APPROACH r f i i o SUDE 8 i

,,I i i d l' REGULATORY INTERFACESTO CONSIDER POWER PRODUCTION RELIABILITY MANAGEMENT PERFORMANCE HUMAN FACTORS EXISTING LICENSING BASIS COMMON SENSE SLIDE 9

I

SUMMARY

PRESCRIPTIVE REGULATION HAS BEEN NECESSARY NONPRESCRIPTIVE REGULATION HAS SUPPLEMENTARY BENEFITS AS SAFETY EXPERIENCE EVOLVES, SO SHOULD REGULATION TRY USING SAFETY FUNCTION APPROACH IN ADVANCED REACTOR DESIGN AND OPERATIONAL PERFORMANCE REGULATION I l l l ~ SLIDE 10

o PERFORMANCE-BASED REGULATION A. DAVID ROSSIN-AND GEORGE D. SAUTER BEFORE THE U. 8. NUCLEAR REGULATORY COMMISSION TUESDAY, OCTOBER 30,-1990 10:00 AM e O 0

3-PERFORMANCE-BASED REGULATION h.: D&Yl.Q BQ3AJ.M &HQ GEORGE D.3 SAUTER BEFORE THE U.

8. NUCLEAR REGULATORY COMMISSION TUESDAY, OCTOBER 30, 1990 10:00 AM WHAT DO WE XEAN BY NON-PRESCRIPTIVE REGULATION?

Nuclear safety philosophy in the Unites States has evolved through statements of safety policy, ACRS letters, rulemakings, regulatory guides and design safety standard review plans. Attempts have been made to set a safety goal based on probabilistic risk assessment. Ultimately, sets of words are used to-tell._the designer and operator what is to be done. These are: prescriptions. Prescriptive regulation focuses on a means of accomplishing a desired and, and tends-to downplay the end itself. This discourages-improvement and innovation.- A better approach is non-prescrJptive or' performance-based regulation, which focuses on attainment of the desired ends.- Non-prescriptive regulation recognizes that there are more ways than-one to meet a functional requirement. The appropriate measure is the result: the performance, and not how it is 4 achieved.~-The' "how" dan be studied and emulated, but should not be; prescribed.. Performance-based regulation recognizes that ,there can-be more than'one way or one set of means capable of achieving the desired end. .somebody has to write every prescription. Non-prescriptive regulation means that the regulator (the NRC in this case) does-not; the licensee writes.the prescription and the NRC reviews it. This is:because it is the licensee who is ultimately responsible for safety and performance. The one with'the responsibility should'be the one who determines how-the. desired result is to be obtained. 'Changes can be:made today under mechanisms like Sec. 50.59E. The . concern is that-the need for flexibility and innovation is growing,-but the atmosphere, communication and degree of mutual understanding are all becoming more rigid.

4 Q Non-prescriptive regulation also means that the licensee can change a prescription, subject to the applicable rules, and subject to NRC review under those same rules. If another party, either the regulator or a third party, proposes a change, the burden of proof that it is necessary falls on the one that proposes the change. The proposer may make a showing that a = change would be cost-effective, but that would not be sufficient to mandato the change. The decision to implement remains that of the licencee. The fundamental objective, and the ultimate test of the success of regulatory policy, is the attainment of desired levels of safety and performance, not whether everyone follows a h predetermined set of prescriptions for attaining.them. This presentation is not offered as an academic exercise. The ultimate viability of standardization in nuclear plant licensing may hinge on success in implementing concepts suggested in this presentation. T Historically, NRC has been heavily involved in writing J prescriptions. The Regulatory Guides are the most obvious example they say that the following the guidance will be acceptable, and that another approach can also be acceptable, but the burden-of proof of acceptability of any alternative prescription is on the licensee. The result, though not intended, is a barrier to innovation and improvement. As the Kemeny Commission noted after TMI, the control and safety. systems at that relatively new plant were old. They did not represent " state of the art" at all. For a complex and challenging technology, it does not make sense to be locked into old and outdated equipment. - e ThenLwhat11s--the problem? 'The regulatory structure in the US has been. subjected to heavy political pressures, and has responded.by becoming very rigid and-legalistic. There is a widespread recognition throughout the industry that.there can be a heavy penalty'for attempting to change, improve or innovate in any plant that has its~ license and is operating satisfactorily. SAFETY GOAL 8 Nuclear. power plants, and for that matter, most major engineering efforts, are designed and operated with the goal that no serious accident will occur. That is a proper design goal, but people .are'not" infallible, practical judgments must be made, and ultimately it must be recognized that the world is not risk-free. .Some degree of risk.must be accepted. However, that does not give either the designer or the regulator.a clear and simple basis for safety design. i 2 .m ,..~

1). '

Setting a goal to strive for, such as no serious accidents or zero public risk, is laudable, as long as everyone realizes that they are goals, or " yearnings", as Dr. Baruch Fischoff calls them.- For example, "There should be no accidents at nuclear power plants" expresses a desire for zero risk, something that cannot be achieved. Likewise, "The risk of offsite health or environmental impacts from the operation of nuclear power plants should be made as low as possible" is an open-ended objective. It is not quantitative or measurable, so compliance cannot be demonstrated. Such safety goals or yearnings are not in themselves useful as regulatory standards or criteria. To be useful, regulatory objectives must address measurable and achievable levels of plant design and operation, so that compliance (or lack of compliance) with them can be demonstrated, although perhaps with some uncertainty. They are criteria for acceptable performance, and recognize that', as with any human endeavor, there will always be some non-zero level of projected risk from the operation of nuclear power plants. Decisions must be made in the design and operation of the plants about-where to draw the line beyond which further steps to reduce the projected risk are not feasible or even desirable from a societal point of view. Once drawn, this non-zero risk line can be translated into acceptable, useful regulatory objectives through-cost / risk / benefit considerations, engineering judgment, or just_ plain common sense. Progress irt improving 'the regulation of nuclear plants will be serious 1-y-impeded, if not stopped, unless everyone concerned recognizes that zero projected risk is not a practical regulatory objective. Neither is every accident or mistake a disaster, a failure of management or a breakdown of the quality assurance program,.or an indictment of the technology, implying that only something new and different will be acceptable in the future. If an accident does happen, it does not mean that the accident is acceptable. In fact, it is not. Every effort must then be made to' determine why it happened and what can be done to prevent it or anything related to it'from happening.again. But by the same token, if an accident happens, does that mean that the technology itself, or its regulatory safety structure has been a-failure, and should therefore be abandoned? This is neither a foregone conclusion nor an absurdity. The premise has to be that a low-probability event can happen. This recognizes the basic fact that there is no such thing as a risk-free world, or a risk-free energy techrology. In recognizing this, it still does not mean that one accepts a 3 )

w L* serious accident as inevitable. Rather, it neans that society can accept a technology even with a non-zero probability that a serious accident can occur. THE CONCEPT: RISK POLICY, RISK ASSESSMENT, RISK MANAGEMENT The regulation of any technology to assure that it is acceptably safe'should have three principal components: risk policy, risk assessment, and risk management. Risk Policy "How safe should nuclear power plants be? What is an acceptable level of risk?" These questions have been the subject of endless debate. The answers cannot be obtained from technical or scientific principles. Rather, the answers should reflect societal value judgments, which are indicated by what people are willing to accept. Ideally, these public judgments should reflect due considera-tion of factors such as the-appropriate balance of the benefits, costs, and risks.of nuclear power; the costs and risks of.the alternatives to nuclear power for electrical generation; and the need for diverse, reliable sources of electricity. For example, the risks to society from inadequate electrical generating capacity are real, and should certainly be considered in setting top-level performance objectives. In theory, since risk policy should reflect societal value judgments about the right balance of benefits, costs and risks, 311 interested-persons should have a voice in the setting of overall levels for acceptable risk. This would require a high level of individual education, information and interest, which has not been seen in our society. In practice, the NRC will:have to play a lead role in stimulating discussion, publishing proposals for public comment, and' -issuing criteria, just as it has done in establishing its current safety policies. However, to date the NRC has been schizophrenic in its consideration of non-safety factor. On one' hand, it has. stated'that its mission is safety only; on the other hand, it ~ has claimed that'it i;s required by the National environmental Policy Act to consider costs, environmental impacts, energy needs, and diversity of supply in its environmental analyses. Because they may be subjective (e.g., no undue risk to any individual) or.because-technical uncertainties may make it too unsure they are being complied with (e.g., no more than a one percent increase 11n latent cancer risk), statements of overall levcis of acceptable risk are not easy for the regulator to apply. The regulator still needs a set of supporting criteria 4

4 at lower levels (e.g. down to the level of plant functions and . systems) to allow him to implement the top-level objectives in a useful way. To be-useful, these supporting criteria should be consistent with the top-level objectives, and the degree of compliance E with them should be measurable with reasonable certainty. h Given the top-level objectives, appropriate-lower level criteria can be established through the application of scientific-principles. Therefore, their establishment should be done by technically capable persons with input from the L regulatory body itself, the regulated and the rest of the technical community. In this case, the NRC would take the lead role. Risk Assessment L 'How safe are nuclear power plants now? Is this sufficient,. L based on a comparison with established top-level safety p objectives and supporting criteria? If not, what degrees of l improvement are needed, and in what areas of plant design and operation? Providing' answers to these questions would be the l predominant activity of the NRC in the future, both in design review of'new plants and in continuing evaluation of operating -experience ~with current' plants. The process would involve participation of the licensees and opportunity for input from the public. But there is another question: "How are.any needed degress of improvement to be achieved?" This is the province of the ' licensee, not the regulator, and not the public.. The NRC role L would not include prescribing what modifications in-plant design,~ construction, and-operation were to be made. That r would fall within the risk management component (described below). In carrying out its risk assessment responsibilities, the NRC l would do all the things it does now except prescribing what is i to be done. It would continue, for example, to. conduct safety l-analys's, review risk analyses and other submittals by licensees and license applicants,' evaluate operating experi-ence, analyze abnormal operating events, identify actual or . potential safety issues, and carry out. plant inspections.- Risk Management This component addresses questions such as, "How are the plants to be designed, constructed, operated, and managed so that the established safety criteria are complied with? If needed degrees of improvement are identified, how are they to be achieved?" In short, risk management as defined here is the prescribing of the means to be used to assure that the nuclear plants comply with the criteria established for acceptable levels of risk. 5

...-,.i,......... q Each plant licensee is ultimately responsible for the safe operation of his plant, and it is he who bears the primary burden of poor risk management c-the benefits of good risk management (although the public, particularly those nearby the-plant, may have a large stake as well). Accordingly, each should be allowed to prescribe how his plantHis to meet the established safety criteria. The prescription will include such things as plant design, operating procedures, personnel training, and safety analyses, as well as the assumptions on which any of these are based (e.g., component reliability, human factors, external event occurrence frequencies). It is possible, of course, that a licensee could simplify his l prescription by opting for a previously approved standardized plant design or set of operating procedures. It may be to his advantage to. join with other plants and industry groups to take advantage of joint research and to adapt or even adopt things that have already demonstrated success. The NRC role in risk' management would be essentially limited. to. enforcement. However, under its risk assessment authori-ties, the NRC could review and evaluate any aspect of the ' licensee's prescription at any time to test, for example, if assumed component reliabilities were actually being achieved or if; prescribed operator training programs were actually being conducted. In summary, under the regulatory concept outlined here, the NRC could carry out all the activities it does now except to

prescribe, in full'or in part, how the established safety criteria are-to be met.

The licensee,-who is ultimately responsible for the safe operation of his-plant, must be allowed'to write thot-prescription, and any necessary revisio'hs'of it, as best. suits his particular circumstances. If:the NRC, Tnt'any other party,- forces part of the = prescription on the licensee, then it should be responsible for.any consequences resulting from that part of the . prescription >being faulty. STANDARDI5ATION The critical test, which may make or break the future of nuclear power in the U.S., lies before us. That test is whether we can .successfully apply the concept of standardization to nuclear plant licensing. For standardization to work, there must be a viable mechanism for makingfnecessary changes to licensed plants, making improvements, and meeting functional requirements with different components and methods. This mechanism must also permit. changes at'one plant to remain unique or make it easy to implement the same changes at some or all similar plants. The. key to standardization is an agreement on the purposes and objectives: i 6

a. l

• Standardization is important for safety certification and regulatory review.
• Standardization has great potential value for improving reliability and economics.
• Standardization is NOT important for the sake of standardization itself.

The standardized plants of the future are the opportunity for . application of non-prescriptive regulation to a reel. issue where a break is needed. The'NRC needs to find a way to 1.andle the modifications and continual work projects that must go on in operating plants. The relevent issue here is not standardization, per se,-but the need for a good mechanism which allows changes to be made at plants initially' approved aus standardized plants without requiring that the same change be made at the other plants of the same standardized family. Standardization is'needed and proper for safety review and certification of designs. But maintaining strict standardization for the sake of the' concept of standardization is neither needed nor appropriate. It is evident that every plant will have to be maintained. While it is important to have assurance that safety is maintained, it is : legitimate to recognize idiat every component does not need to remain the same throughout the plant operating lifetime. How can sp.fety be assured if plants are allowed to do different y things once they are licensed as standardized plants? The answer .is non-prescriptive regulation. What is important is that the functional performance objectives be maintained. 'There may be a variety of ways to assure that-they continue to be met, even though changes are made in hardware and in procedures, and that ~ these changes may differ from plant to plant. Assuring that the performance-based objectives are met gives assurance of safety. Changes can be made under 50.59E, or based on submittal of proposals if the 50.59E criteria are not applicable. Where safety considerations dictate that-a change must be made in all plants of a class, NRC can mandate it, just as NRC can do now. 3 Operation of standardized plants offers a logical opportunity-to apply.non-prescriptive regulation. Standardized designs can be approved'and plants can be licensed under this method, once constructed satisfactorily, non-prescriptive regulation based on . performance criteria can be applied to evaluate both proposed 7

0 Operstien of standardized plants offers a logical opportunity to apply non prescriptive regulation. Standardized designs can be approved and plants can be licensed under this method, once constructed satisfactorily, non-prescriptive regulation based on performance critoria can be applied to evaluate both proposed changes and possible requests for conformance based on changes made in other plants licensed under the same standardized plant certification. In the past there have been instances in which design changes have not been approved even though they still' meet the performance criteria required for safety, on the argument that the changes would result in a decrease in the margin that previously existed. This approach has resulted in a stifling of design improvements and efficiency initiatives. In standardized plants, there would be-a temptation to apply this same policy: regulating on existing margin rather than on meeting actual safety and functional requirements. The proper test should be the ability to meet performance requirements, without limit on how they are to be met or what margin may be claimed by previous or alternate designs. CONCLUSION -Particularly since the'Three Mile Island accident, both the NRC and the nuclear industry, with INPO, EPRI, and NSAC, have devoted extensive' effort to increased understanding all aspects of design and operation, and to in-depth analysis of operating experience. Can we judge how well the present body of regulations is performing? Certainly on the basis,of safety, they appear to have been a success.< *However, the fact that no plant has been completed that has been ordered since 1973 is a clear indicator that the balance is not correct. TheLfact that it is taking the system so many _ years to-review the design of a plant based on twenty years of experience serves-notice that-the present approach must be improved upon. The' application of performance-based, non-prescriptive regula-tions can be achieved without legislation, and could go far toward building the necessary understanding to permit new plant orders, without compromising the high standards of safety that have-distinguished the American nuclear power program. 1 8 ._____ _. _}}