ML20059L987

Transcript of ACNW 24th Meeting on 900919 in Bethesda,Md. Pp 1-129.Viewgraphs Encl

ML20059L987
Person / Time
Issue date:	09/19/1990
From:	NRC ADVISORY COMMITTEE ON NUCLEAR WASTE (ACNW)
To:
References
NACNUCLE-T-0027, NACNUCLE-T-27, NUDOCS 9010020363
Download: ML20059L987 (155)

Text

{{#Wiki_filter:j_LeQ m A, e, \\ A fc.gy7".-- 00W7 = N' . OFFICIALTRANSCRIFT OF PROCEEDINGS p n .-..a-....______.,, e 0 g~ w(ACNW) g. J TR08' Return Original to.B. Uhite ii ~ACRS-P-315 Thanks! Barbar Jo White l; 27288 - ~ - - - - ~ - I l' M6!)@ U.S. Nulcear Regulatory Commission Ei Advisory Committee On~ Nuclear Waste (ACNW) Tide: 24ra astr1No i Docket No. 6' i ~ l '. ( ( ll -mng Bethesda, Maryland i.[ .mg-Wednesday ~, September 19, 1990 Pacts: 1 -f129 c i- [. 'f L 1. I I { X6NW OFince: e a. ' THE LIFE op]Il l. L 4 FOli .~ ANN RIIEY& ASSOCIKES, LTD. 'y 0 1 1612 x. s..se - >o. OMC9 Wahingen,D.C 20006 (202) 295-3950 +o 100:0 x 3 w,. r PDfy, 27ADycM pgo_g n L F-1-vv pgg t l-,

t .1 Yi 'I '2. r,. '3 i -4 PUBLIC NOTICE BY THE 5-UNITED STATES NUCLEAR REGULATORY COMMISSION'S -6~ ADVISORY COMMITTEE ON NUCLEAR WASTE 'l 7, 8 DATE: Wednesday, Septemb'er 19, 1990 9 10 -11 12 lp) 13 The contents of this transcript of the e, t 14 proceedings of the United States Nuclear Regulatory 15-Commission's Advisory Committee on Nuclear Waste, 16 (date) Wednesday. September 19. 1990' i: I' 17. as reported herein, are a recordlof the discussions recorded at ' 18 : ,the aceting held on the above date. i 19 -This transcript has not been reviewed, corrected '20 or edited, and it may contain inaccuracies, j 'I L, .21-22 L 23 '24 li

1 -l' 7 1s_/ ' 2 UNITED STATES OF AMERICA. I -3 NUCLEAR REGULATORY COMMISSION t 4 ~*** ~5: ADVISORY COMMITTEE ON NUCLEAR WASTE (ACNW) i 6 7. 24th MEETING 8 1 9 10 Nuclear Regulatory Commission 'll Room P-110 12 7920 Norfolk Avenue j-13 Bethesda,-Maryland 14 Wednesday, September 19, 1990 1 ,15

16 '

The 24th meeting of the. Advisory Committee on i 17 Nuclear Waste (ACNW) commenced, pursuant to notice, at 8:30= ,18 O' clock a.m., Dade'Moeller,. Committee Chairman, presiding. l b ~ x19' l 20 PARTICIPANTS: L l ~ 21: p. '22 D. MOELLER,-Chairman, ACNW L23 W. J. HINZE, Member, ACNW L n L,{ ~ 24 P. POMEROY, Member, ACNW ( 25 lt 1. l-l I i l

f 2, .? 'l- -PARTICIPANTS (Continued): [ s 2

3.

R.. MAJOR, ACNW Staff 4-H._LARSON, ACNW Staff-l: '5 C. ABRAMS, ACNW Staff 6 H. SCHOFER, ACNW Technical-Secretary-l 7-R. FRALEY, Executive Director L 8 l 1 i 9 R. McGUIRE, Consultant to EPRI l-10' M. SHERIDAN, Consultant to EPRI ill R. SHAW, EPRI L 112 7; l: '- [ L13' 14 '15 16 17-185 19

+

20 "21-22 o - 23. 24 bp. 25

3 l ~- PROCEEDINGS 2: [8:30 a.m.) 3 MR. MOELLER: The meeting will now come to order. 4 This is the first day of the 24th meeting of the Advisory 5 Committee on Nuclear Waste. I am Dade Moeller, Chairman of f 6 the Committee. The other Committee members present are 7 William Hinze and Paul Pomeroy. 8 During today's meeting, the Committee will do: 9 .three or four things. 10 First, we will review or hear a presentation from i 11 representatives of the Electric Power Research Institute on 12 their performance assessment methodology for a high-level h' 13 waste repository. And that will carry us up until Noon. A 14 And then after the lunch break, we will go into 15 Executive Session, and initially have a discussion in which 16 the Committee will attempt to define a strategy for 17 responding to several recent requests that we received from 18 Commissioner James Curtiss of the NRC.for the Committee to 19 analyze and deliberate on a couple of questions that he has. t 20 And then we will begin preparation of a response- '2 11 - to a letter that we received from Richard Guimond at EPA in 22 . which that organization has asked for clarification of some 23 of the comments of this Committee in our earlier letter 24~ reports in which we were commenting on the EPA's high-level (O 25 waste standards. l n .E

I. 4 in And then the last item we will take up'today will ( be to discuss whether the Committee, or how the Committee 2 -- 3 wants to approach the NAS/NRC National Research Council 4 report on,' quote, " Rethinking High-Level Radioactive Wasta 5 Disposal," unquote.. We know that the NRC Staff is preparing 6 comments on that. Mr. Bernero spoke yesterday at their 7 symposium and oftered comments on it. 'And our discussion 8 will be whether this committee wants, at this particular 9 time, to prepare an independent set of comments. 10 This meeting is being conducted in accordance with 11' the provisions of the Federal Advisory Committee Act and the .12 Government in the Sunshine Act. j g 13 Charlotte Abrams is the designated Federal 14 -official for the initial portion of the meeting. 15 The rules for participation in the meeting have 16 been announced as part of the notice that was published in 17 .the Federal Register. ~18 We have received no written comments or requests 19 from members of the public, or others, to make oral 20 statements at today's meeting. However, if there is someone-21-here in the audience that has a contribution or wants to 22 make a comment on something we are discussing, simply check 23. with us.and we will be pleased to provide time to you to do 24 so. e 25 A transcript of portions of the meeting will be i i iiimis- -m--mi-i-um-i-

k 5 ~ 1 kept, and it is-requested that'each-speaker step to one of f im. t-2' the microphones, identify himself or herself, and speak with^ 3 ' sufficient clarity and volume so that she or he can be 4 readily heard. 5 Before proceeding, I have a few brief remarks of 6 possible-interest to.the members and the people who are 7 witnessing our deliberations today. 8 I had prepared a summary of some recent 9 developments of possible interest. We will make sure that 10 everybody has a copy. Perhaps they do. But let me go 11. through a few of them. 12 One item that I believe it would be useful for the j ) 13 Committee to discuss at this meeting is the proposal 1for 14 possible PRA topics that have been suggested to us by 15 Stewart Long. In the main, what Stewart has dons ocewart 16: Long is-an ACNW/ACRS fellow -- and what he has done is to 17 look at the capabilities of the Staff here, our supporting 18 Staff, for doing probabilistic assessments, and he is simply i 19 calling that to our attention and asking whether there are 20 areas that we might want to join or to call upon them for '21 support. 22L In fact, this morning he gave me a memo with some 23 more information on that. And I hope that we can discuss it 24 later in the meeting. 25 But they do have the capability, and if it would

6 1 be-helpful-to us to-tie into it and to call-upon them to.do' ,- s) ' -(N / 2: 'certain; types of studies, we should consider doing.so. 3 We will'be taking up human intrusion at the next l 4 . month's meeting. And I had made a note of an item here that 5 the ACNW has been provided with the two-volume report '6 prepared by the Commission of European Communities on this 7-subject. AndLthat may be helpful to us. And then, through i 8 Howard Larson, who has been in contact with D.R. Anderson at 9 Sandia, we have:other information on the~ topic. 10 I noted that the Low-Level Waste Management and .11 - Decommissioning Program issued its-semi-annual report just 12 recently. - And it is in SECY 90-223, published on June 21 of r~'{

l. [

13 1990. 14 An interesting part of che report, to me at least, 15= was-that I found that they expect to take up four items or. '16 to bring four items before the Commission within the next 17-six months for approval and comment-before they move 18 forward.- And three of the four pertain to uranium mill 19 tailings.- So it is quite possible we need.to keep in. mind -20 . whether we should be looking at little more at that subject. 21 And then they do have their four-volume report, 22

which they have issued on, quote, " Background Information 23 for the Development of a Low-Level Performance Assessment 24 Methodology," unquote.

Of course, today we-are hearing 25 about high-level waste performance methodology. t

7 1 'Are we scheduled to discuss that at a-future' . ~.. \\_- -. 2 meeting? That is NUREG 5453. 3' MS. ABRAMS: The Staff will be talking to the-4- Committee next month'about their Phase 1 performance 'S assessment, which is just.a beginning. 2 6 MR. MOELLER: Okay. So at least we are into it. 7 All right. ~i 8 Commissioner-Rogers suggested to us several months L 9 ago that we might want to look at this NORDIC consultative L 10

report on high-level waste disposal.

I simply. wanted to-l. 11 . mention that, whether we should keep it in mind. What they. 12-published was, quote, " Disposal of High-Level Waste - (} 13 Consideration of Some Basic Criteria A Consultative. 14 LDocument."' And there is background information on that in 15 SECY 90-229, which was issued on June 27, 1990. 16 Now, the Staff has also recently published a r L-17 Federal. Register Notice on the availability.of a, quote, r t I 18 " Draft Technical Position on Regulatory Considerations in 19_ the Design and Construction of the Exploratory Shaft 20 Facility," unquote..The Committee will be taking that up, 21 or is scheduled to take that up at our February 1991 22 meeting. 23 And then there are a variety of other things, 24 MR. HINZE: Will we also be looking at the f. 25 alternative study at that same time? J

4-4% ,} 8? d i

1-MR. MOELLER:

Okay.' We'll'need to get some Staff ,q - k, ^E 2 report. 3 Do you know, Richard? j 4 MR. MAJOR: No, I don't. 1 5 MR. MINZE: It would seem to be appropriate, and i 6 it would be-good timing, I would suspect, from the. '74 alternative studies. 8 MR. POMEROY: I certainly would like to re-listen '9 to the performance assessment portion of the ESF study that 10 we heard some report on yesterday at the NAS symposium. 11 -MR. MOELLER: We can make a note of thase things 12 and talk about them in terms of future agenda. I also

(

13 notice, of course, DOE has petitioned for ruler.aking on 14 accident dose criteria for the high-level wasta repository. l l - 15 Is that -- done anyone know the schedule of when we may be' 4 o t _16 involved in that -- the. accident dose criteria rulemaking?. . 172 Is it quit 6-a ways off? 18 MS. ABRAMS: Yes. ~ l 19 MR. MOELLER: Okay. Then Mr. Fraley attended the 20 National State Liaison Officer's Meeting,.which was held in 21 Rockville on September the lith and 12th. And included in - 2;E - onfthat agenda was a presentation by commissioner curtiss 23 s on " Regulatory Dilemma: NRC and EPA-Perspectives on Mixed 24- . Waste Issues." And, of course, that's one of the two topics Gt 25' he's asked us'to talk on, or to review and evaluate. i

i 9 1 So, I hope that.the staff can obtain a copy of ?,r s \\_s 2 that talk for us. y J '3! Yes? 4 MR. FRALEY: Just a minor correction. -Howard 5 Larsen actually attended that meeting. 6 MR. MOELLER: I hope later in the meeting we'll 7 have time for-a report on what occurred there and follow up. 8 on it. 9 One other thing that was called to my attention 10 this morning is that the Senate'is holding -- and it's Senator Robert Graham from Florida, plus other Senators, but. i 1 -. [y, 12 he's one of them -- they're holding hearings on a range of l-L '[} 13 nuclear matters; and they've scheduled hearings on October U 14 the 2nd on the role of' standardization in radioactive waste. L 15 management and disposal.- And I think we ought to flag that-16 and be sure that semebody goes down and at least, you know, 17-years it and makes a summary of what was said. 18 MR. POMEROY: I plan to attend that. 19 MR. MOELLER: Great. Okay, well we're already 20 covered way ahead of us.

1 Okay, well, we'll move now -- or, well are there 2

22 any comments; Bill or Paul, other comments? 23 (No response.] 24 MR. MOELLER: Okay. Well, we'll move on then to f~s I 25 the first item on the agenda. And Paul Pomeroy will be =

10 ./, 91 handling that. It's the EPRI performance assessment p(s, ~, 2 methodology for an HLW repository. Paul? l 3 MR. POMEROY: Thank you. 4 The presentation this morning, by representatives 5 of the Electric Power Research Institute, will center on an-6 EPRI-sponsored performance assessment methodology.' This is,. 7 in essence, the first presentation, I believe, to the ACNW 8 of a new generation, if you will, of performance 9 assessments. And I expect that the Committee will hear at 10 least a few more presentations on the. findings of these new 11 -generation assessments in the coming months. 12 Before I introduce the speakers this morning, I j[ 13- 'would like to make a'few comments and pose more-than a few L 14-general questions that perhaps can be partially answered 15 within the context of the presentation. 16 Prior to the break this morning, our, agenda calls t j' 17 for a discussion of the objectives. And I'm sure that the 18; speakers will deal more than adequately with that. 19 The second part calls for a presentation of the i 20 methodology, itself. As part of that presentation, the Committee would like to highlight several areas. 21 '22. .Specifically, how does the EPRI methodology differ from 23 other -- others of these new generation of performance 24 assessments that are coming online right now? O 25 I'm sure you will tell us about the strengths of

11 1 the methodology.- Wo.would like to know about the weaknesses O 2 as well.. 3 We are, of course, very concerned about 4 uncertainties and I'm sure we'll talk'a lot about that. 5' But, I'd like to know how the uncertainties are treated _and-6 reduced within the methodology, and then how the residual 7 uncertainties are treated.. 8 A-fourth item -- since, as I understand it, teams 9 of experts are used, at least in the initial stages, and 10 perhaps throughout the process, it would be useful to know, 11 in essence, all of the points where expert judgment is 12 applied and exactly how it is applied; that is: specifics f) 13 on the assignment of probabilities for the logic tree 14 branches, for example, or the scenarios, if you will -- 15 whatever and where ever the expert judgment is applied. 16 The next comment has more to do with some of the 17 perhaps more standard approaches to performance assessment; 18 but how do you ensure completeness and independence of the 19 scenarios or the logic trees? 20 Another point that is somewhat related to another 21 one is just what can, in addition to what the models are 22 .that you've used, what can the models do, and what can't 23 they do? 24 We would like to now your thoughts on criteria 25 that you might suggest for credibility of these assessments.

~c + 12 r 1= Does'the methodology truly allow,for a computation of a ~. [,' (' ' ) 2- . credible performance assessment -- and.I suspect you will i 3 agree that it does ---and a defensible performance 4 assessment? f 5-And I'm concerned here about credibility within. .6 all the communities: the scientific and technical 7 community, the political community, and perhaps most l 8 importantly, the public. t 9 And-finally, it's my understanding that the 10-methodology centers primarily on issues that are related to 11 the earth and atmospheric sciences, if you will -- on the 12 broad geotechnical issues, related to siting. But certainly (5 ) 13 there are other issues. And I'm hoping that you'll provide ? 14 us some thoughts about how human intrusion, either is 15 incorporated or can Ima incorporated into the methodology 16 that you're proposing. 17. I understand that this approach-is based on the 18' decision modeling approach previously applied to seismic 19 hazard assessment. And I would like to ask our speakers to 20 explain the methodology as completely as possible. 21 The members of the committee and its consultants 22 are extremely interested'in learning, in detail, about the 23 methodology and one of our -- the members of the committee 24 and our principal consultant in this area, are not able to ~ 25 be here today. They will be avidly reading the transcript

13 7- ~ and they would certainly like to be able to~ read as complete 1 2 a version of the methodology as possible. 3: As for the rest of the members of the committee, I 4 we.are all at different points, I believe, on'the learning l 5 curve in this area, and the presentation will be most useful 6 to you and the committee, if we clearly understand, in '7 detail, the logic' involved. 8 Although both Dr. Hinze and I participated in the 9 seismic hazard assessment, on the EPRI program on seismic-10 hazard assessment, please don't assume that we are 11 completely familiar and that the entire committee is 12 completely familiar with the methodology. (N)! 13 WitP. J.ase comments and prompts for the speakers I 14 would like to introduce Bob Shaw of the Electric Power 15 Research Institute. Bob is a Senior Program Manager.for I 16 High Level Waste and Spent Fuel Storage in the Nuclear Power 17' Division of EPRI. 18 Bob, would you like to start? 19 I would'like to thank you for coming today. We 20 deeply appreciate your coming to talk to us and we're 21 certainly eagerly looking forward to what you have to sav. 22 MR. SHAW: We welcome the opportunity and we're i 23 pleased to be here as well. Is this mike working okay? 24' MR. MOELLER: Yes. b 25 MR. SHAW: Thank you,

14-l' - MR. MOELLER: -Do you need a-light or.'something up 2 -there, Bob? -MR. SHAW: It would be more fun-than eye strain.. MR. MOELLER: Could we turn on the lights. There 5 we go. 6 [ Discussion off the record.] 7 (Slide.] 8 MR. SHAW: There's a few items that you mentioned 9 in your preliminary remarks that I might make some reference 10 to now. 11 I have a nice opportunity to speak with you last 12-October and tell you what it was we were planning on doing l( 13 and at that time I covered in'some detail.the history of the 14 Seismicity owners Group because that was used'as the basis

15 and the background for.some of the work that we have done 16 here.

17 We have chosen this morning not'to repeat that, 18 since I think it is a matter of-your record and you could 19 make reference to that, but rather this morning we wanted-to 20 emphasize and spend time on what we-have done since that 21 time. 22 I am going to Ossume that that background is 23 available to you but even without it I think the picture 24 this morning will be complete in and of itself as we go 25 through this description.

15-11 We particularly appreciate the amount of' time that s you have made availabic to us because I think it gives all, 2 3 of us an opportunity to in some thoroughness go through what 4 it is that we have done. 5 The project -- at least what I will call the first 6 phase of the project -- is completed as of the end of July. 7' We made a presentation on August 1st to a group of DOE 8 people at Yucca Mountain at the project. office in Las Vegas. 9 There was a staff member and a TRB member at that time who 10 also were present. 11 Subsequent to that on the 5th of September we had-12. an opportunity to present it at DOE headquarters-for the [ ) 13 people who had the opportunity to join us there. Then 14 subsequent to that I have had two opportunities to give -15 presentations to utility representatives, one from the-EEI 16 and others from EPRI that we have met with in the last two-17 weeks, so we have had about four presentations to give.this-18 to what we consider the core groups. 19 EPuI of course and EEI have been financially 20 supportive of this as well as supportive in other ways and 21 DOE of course is the primary. body of interest to which we 22' want this material to be transferred. 23 As you will see here, a brief description or our 24 objectives is listed on the first viewgraph and that is to 25 develop an integrated methodology for early site performance

+ l L '16 ), 1 assessment and to identify and prioritize critical issues. / I' -24 'I hope _you'll agree with us that we have done that-

and that it is a part of our methodology as it-is being 4

' described-today. 5' The second part is to involve.OE in this 6 -- methodology development and its-implement <ci"n. I feel on j 7 this one we have made progress but I am L at yet where I want i . EL to be. 9 MR. MOELLER: Excuse me, Bob. What do you mean by 10 . involving them? You said you've made' presentations to them. - 11 Are1they actually taking part or do you hope they'll take - 12 part? .r' J As we move into phase 2 and phase 3 13-MR. SHAW: 1 14 that I'll describe lator, I think it will be very clear as 15 to what I mean exactly by that, so I would'like to defer on 16 that question -- 17 MR..MOELLER: Fine. 18 MR. SRAW: -- because-I think it is a part of the 19' complete presentation here; 20 It might be important to say a couple things that - 21' this methodology does not do. This methodology is meant to 22 be illustrative and therefore the results that you see today 1 23J are just that. They are illustrative results. -s 24 What we have attempted to do is develop a . (( 25 framework within which performance assessment can be done. e-b

i F f 17 .1 In the:very limited time you will-see very soon that.we have ,_s }'N/ ) _2i been. working:en.this'. We make no claims to being completely 3: understanding of. Yucca Mountain or even having knowledge-4 that:that is part of what-the DOE people'already-have. 5 It is limited in the number of people and the 6 number of' interactions that we have had so our emphasis.was [ 7 on developing an integrated framework that allows one to do 8 in a sense a back-of-the-envelope kind of calculation,.to 9 .say how do we pull all this stuff together? Can it be done? 10 What insightsLdo you get as a result of that, so that the 11 result of all that is I ask you not to focus so.much on the r 12 numerical results but rather to focus on the process that we .( -13 ~have evolved in developing this performance assessment 14 methodology. 1 15 The'second aspect I would say is there are certain 16 -parts that have come to the forefront of performance 17 . assessment that are not a part of our analysis. 4 18" Two examples that are primary -- we have no gas 19 phase transport'in the model -- it's something that could be ~ 20. included but we chose not to. We also had no' human 21 intrusian in the model and it could also be included and I 22 'think you'll agree as you see the assessment that it is 23 fairly straightforward to do that. 24 What we are going to do today is after I have 25-given some of these initial introductory comments we are

18 i L l' going to give you a brief perspective of what the results j ym \\ t g\\s / ' 2-look like so that you can see the format in which they are l[l 3 displayed. Then we are going to go back and we are going to 4 choose two of the particular technical areas that we will 5 cover in some depth, namely the seismic area and the volcano 6 area, to give you an indication of how we developed each of 7 our technical areas. 8 These are meant to be examp.\\es that would be representative of the range of technical areas that we have 9 l 10 cover d. I 11 Then having covered those two, we'll come back and 12 we'll go through ic some detail the results, showing

f)

13 sensitivity analyses and also then showing how these results 'w) 14 can be used to develop priorities for the various aspects 15 and then I will return and tell you what we have in mind to 16 go from here. 17 That is the general outline of the presentation. 18 (Slide.) l 19 MR. SHAW: We do have a methodology development l 20 team which was a group of experts that we pulled together. 21-They're all listed here and on your second slide. I'm not 22 gollig to read through them, but you can see the names, the 23-affiliation, and the expertise of each. 24 About two-thirds of the way down, you'll see the -j sU 25 namos of the three EPRI people who have been involved here: -l

19 ) 1 myself, Carl Stepp, and Bob Williams. At the very bottom, ,_s [' /) 2 you'll see the name of Russ Dyer, who, from the Department 1 3 of Energy, was their observer at our meetings. He was a ) 4 regular attendee. He is in charge of performance assessment J 5 under Carl Gertz and Max Blanchard at the Yucca Mountain 6 project office. So this was one of the ways in which we, 7 right from the start, said we need to have effective 8 interaction with the Department of Energy cn this particular 9 project. 10 (Slide.) 11 MR. SHAW: The next viewgcaph shows the history of 12 what we did in order to get where we are. We started last [ ) 13 July with a brainstorming session that we held in Palo Alto, 14 where we collected together about a dozen-and-a-half people 15 and said, basically, We have this history from the 16 seismicity owners group; we think it is applicable to the 17 methodology that could be developed for performance 3 18 assessment for Yucca Mountain. What do you think? So we J 19 bandied ideae around the table and brainstormed it, and 20 decided that we really felt we could do this. 21 So we moved at that time to identify appropriate q 22 experts to collect together that would be at our performance 23 assessment methodology team, so that we could use these i 24 expertise, and integrate and tie them together. 25 The second meeting, I've called a Qualification

20 i i check. This was, after we had identified what we felt were ,N j <~ \\) 2 appropriate experts, we wanted to in a sense interview them, s-3 bring them together, make sure it was working. I must 4 confess-that, a little to our embarrassment, we found that f i 5 it wasn't working as well as we thought it would, and we i 6 needed to have people who were team players; we needed to l i 7 have people who were open to new ideas, especially to a probablistic approaches. f 9 In some instances, the people we had chosen didn't 10 quite fit in that fashion, so we had to go back and work a l 11 little more. As a result of that, we came up with what I, 12 especially in retrospect, can say was a very successful team ( } 13 that worked very well together. 14 So our first kickoff meeting, then, was in i 15 December of 1989, where we got together and said, okay, 1 16 let's define the problem. Our second meeting was about a 17 month later, in which we formulated the problem. We had 18' used influence diagrams at the first meeting to generate, at 19 the second meeting, logic trees. of course, those logic 20 trees that you'll see today are not the same that we started 21 off with, but they were still decided as being the 22 appropriate basis for the description of our methodology. i 23 As a result of that meeting in January, we then 24 met in April, where the model was actually presented. In 7sQ 25 between each of these meetings, of course, we asked each of .,,. ~... ., ~...,..,. _ ...r,

21 1 our members of our methodology development team to go back V) f i 2 and work rather industriously on their particular aspect, 3 and the integration of that aspect with the others. 4 That turned out to bo one of the key points, of 5 course, is the interfaces between the technologies. Each of j 6 these people is quite capable of defining their own 7 technology, but to ask them to also interface with the l 8 others is the real challenge. l 9 So the model was presented, then, in April, and I i 10 that gave us an opportunity to critique it and refine it, 11 and we came back then in July with the model completion. [ 12 Go back one slide, please. l (f ) 13 (Slide.) l v 14 MR. SHAW: One thing that I neglected to point [ 15 out, and let me do so now, is that on this list, you'll see l 1 16 about halfway down the name of Robin McGuire of Risk 17 Engineering, who is our expert in risk analysis.

Robin, 18' who's my viewgraph projector here, is the one who integrated 19.

this whole package. He's responsible for the software that 20 you'll see the results of today, and really for pulling this 21 whole thing together. 22 The other person here whom I'll introduce -- who's 23 about three names below him, just above my name -- is Mike 24 Sheridan of the State University of New York at Buffalo, who f.( 25 is our expert on volcanology. .,r---, _~

22 ) 1 MR. MOELLER: A question, Bob. In selecting these (3 1 2 people, you were attempting, as you've just said, to find 3 people who are open, people who can work together, and 4 people who had the expertise to do the job. Have these same ) 5 people been working -- are they working for DOE or someone 6 else on the same thing, or was there any ttempt to have j j 7 independent, fresh approaches? Could you comment on that? 8 MR. SHAW:' Yes, I certainly can. My initial 9 thought was, Let's have people who are completely 10 independent, who haven't been involved. That turned out to 11 be inappropriate in some areas, especially, for example, i 12 waste package, where it just doesn't work. l . '([ 13 So what you'll see up there is really a mix. For 14 example, or waste package expert, Dan Bullen, who is now at 15 Georgia Tech, spent time at Lawrence Livermore and really is L 16 experienced in the program. 17 There are a couple other people who have been 18 contractors to the program, and there are a number of people 19 who have not been involved in the program. Mike Sheridan is 20 an example of that, in the area of volcanology. 21 So we ended up with a mix, and we certainly wanted 22' to get at least a core of people in there who were not 23 involved, and it became evident that it was also to our 24 advantage to have a core of people who had been involved so n . (t 25 that we could have both fresh and experienced points of y.

l-23 1' view. I 3/s I ISi 2 That completes all my preliminary and introductory l 3 comments here, and so I'd like to turn it over to Robin j i 4 McGuire now. 5 MR. HINZE: Bob, if I may ask, in that list, I 6 don't see structural geology tectonics represented. Is that 7 because of a lack of interest in that, or is that covered by 8 someone else in that group? i i 9 MR. SHAW: Well, it's covert:1 by the other areas, 10 I'd say. Robin, maybe you could comment better than I on 11 that particular technical area, t 12 MR. McGUIRE: I think Kevin Coppersmith covered J. 13 that area, and I'll give some indication of some of the 14 results he came to. .15 MR. SHAW: I will now introduce Robin McGuire who 16 will give us an overview of the results. 17 (Slide.) 18 MR. McGUIRE: The approach that we took in this 19 project was a structured representation of what is known and 20 unknown about the technologies that are required to estimate 21 possible releases at a repository. There are several 22 approaches available, and among those that were considered 23 by the group were fault tree approaches, Monte Carlo n 24 approaches, event tress, influence diagrams and logic trees. U 25 The one that was found to be most appropriate and

24 1 most useful by the group was the logic tree approach wherein '2 the understanding and uncertainties associated with each s-3 area of the problem can be represented in a straightforward 4 fashion. I show an example of a logic tree here as an 5 indication of the kinds of things that are represented. 6 In a logic tree, we have an uncertain input or an 7 uncertain state of nature which is important to our problem 8 or our calculations, and that's represented by a node. 9 Possible values of that parameter or that state of nature 10 are represented by discrete branches. Each of those has 11 associated with it a value, and then a probability. 12 The way the logic tree is set up, the tore (( ) 13 independent parameters or assumptions are put to :he left 14 side; the dependent parameters or assumptions are put to the 15 right side, so that, for instance, what's listed on this 16 side as a source term, can be dependent on an external 17 impact, so that the interpretations here will be dependent i 18 on whether one is conditional on the first branch of the j 19 external impact or the second branch of the external impact, 20 and similarly with the hydrological properties, for example. i 21 once we have the entire input specified in this 1 22 way, then we can look at each end branch and for that end 23 branch, first calculate the probability of that set of 24 parameters. That's just the product of the branch, the {I 25 probabilities on each branch that lead to that end branch,

i 4 25 1 and we can organize that set of parameters for set of l - (V i 2 calculations, for example, for radionuclide release at a i 3 repository. 4 The rules that are followed are the standard rules 5 for probability theory; that is, that these probabilities, [ 6 if appropriate, must be conditional on previous branches, 7 the sums of probabilities coming from each node will sum to 8 unity and the sum of the probabilities for the end branches 9 must be unity. { 10 (Slide.) 11 MR. McGUIRE: That is very conveniert for i 12 calculations, because it organizes the alter;atives and 13 their probabilities in a way that makes it easy to perform '. 4 calculations to represent the totality of that set of 15 possibilities, and also, as you will see later, to conduct i 16 sensitivity studies and uncertainty studies. i 17 Specifically, for each set of parameters, we can 18 make source and hydrological transport calculations for a 19 repository to calculate cumulative levels of radionuclide i 20 release to the accessible environment. 21 So-for set of parameter 3 one, we might have a one 22 curve, indicating that concentrat.an is a function of times 23 for a second set of parameters, we may have a second curve, l q indicating that concentration is a function of time. 24 Ik/ 25 With each of these curves is associated a ,m

i 26 1 probability, which is just the probability of that end V> \\- 2 branch, or that set of parameters. 3 At some time, "t zero," we can look.at that set j i 4 of curves and their associated probabilities, and accumulate 5 those, for example, from the top, to form the CCDF, the 6 complementary cumulative distribution function, of the level i 7 of release of some chemical or radionuclide representation. 8 And that can be, for example, if that is the 9 parameter or the representation upon which the safety of a 10 specific repository is to be judged, that can be compared to 11 a criterion, or a standard to determine whether or not it is n 1 12 acceptable or unacceptable, '(f } 13 In terms of one of Dr. Pomeroy's questions at the 14 beginning, how does our methodology differ from others, I 15 would say this is one of the important way" in which it 16 differs. 17 We set up logic trees for all of the inputs, and 18 require that of our participants. 19 (Slide.) 20 MR. McGUIRE: Of course, the total logic tree for 21 the entire system is rather large, and in our case, for this 22 demonstration, includes 11 nodes, and their associated 23 branches and probabilities. 24 I will come back to this slide a few times during 25 my presentation.

i. 27 '(._ 1 The advantages of doing that over alternatives is ) 2 that it really facilitates sensitivity studies and 3 uncertainty studies. 4 (Slide.) 5 MR. McGUIRE: The alternatives that are often used 6 are Monte Carlo approaches, wherein one selects or specifies 7 distributions on parameters and then goes through a set of 8 calculations selecting values of those parameters by 9 simulation and then conducting source and hydrological 10 transport calculations for that set of parameters. 11 That will lead to the same result. However, the 12 advantages of a logic tree approach are that once you have (( ) 13 calculated, made these calculations for the set of 14 parameters that represents this set of curves, one can very 15 easily change the probabilities on the branches. What that 16 does is change this column, that is, the probabilities 17 associated with each of those curves, but it doesn't change 18 the curves. 19 So one can set up a database of the calculations, 20 specifically the curves, and very easily change the 21 probabilities, thereby changing the CCDF, wi+.hout having to 22 redo all of the calculations. 23 This can be done very efficiently, almost in real 24 time, and can be a real tool at a workshop, for example, to r~g V 25 determine what are the important variables in the problem,

n i i 28 1 and what are not so important, with respect to, for example, l 2 the level of the ccDF, which probabilities are the most s-3 influential and which are less influential, and thereby l 4 concentrate the energy and the goal at that workshop or that 5 interaction to those parameters that are most influential. 6 MR. POMEROY: Robin, excuse me. You named several 7 possible alternative methodologies at the beginning of your l 8 presentation. Logic trees and the Monte Carlo methods were 9 two of them. i l i 10 How did you, can you give us an insight into, is 11 it the same kind of logic that prevails for choosing this as 12 the most promising methodology? ([ } 13 In other words, you have illustrated how you 14 compare this with the Monte Carlo methods and the logic for 15 choosing this over the Monte Carlo methods. Would the same 16 kind of logic prevail over all the other methods that you 17 named? 18 MR. McGUIRE: The other :sethods would have this 19 advantage as well. They were found by our team not to be 20 quite to convenient in terms of specifying what the team } 21 members knew. So that that was more of a group dynamics 22 decision, that they were more comrur+able with a logic tree, 23 for example, than a fault tree. 1 24 MR. HINZE: Was that just the experience of the kr 25 group with the logic trees in terms of the previous study?

i i 29 l 1 MR. McGUIRE: I think so. Yes, I think so. 2 It also, I think, is a result of looking at new 3 problem with these kind of, applying these kind of 4 probability methods, and not knowing beforehand what all of 5 the potential failure mechanisms are. i 6 Fault trees work very well if you have a good 7 understanding of what the potential failure mechanisms are 8 and can track those and represent those, then, with a fault 9 tree. 10 But if you are not familiar or don't have much j 11 experience with those, and maybe nobody has much experience i 12 with those, then it is difficult to be sure that you have [ ) 13 incorporated or represented all the possible fault trees. 14 With a logic tree, you, in a sense, represent the 15 entire system. And as you will see, I think, from some of ? l 16 the examples, from that you can get a sensa of what are the 17 most critical combinations of parameters that might lead-to 18 high levels of release. That comes out of the logic tree 19 representation, not because you put it in, but because you 20 are representing the entire system. So it may come out of 21 the problem even though you didn't foresee it. And that 22 gives you additional insight into what combinations of 23 parameters are driving the results, r-24 That is an additional advantage of the logic tree .- (- 25 approach, but certainly fault tree approaches could be used. ..., ~

) 30 1 In a sense,'they are equivalent if you identify all of the i IN-2 potential faults in a system. 3' MR. HINZE: Yes, Bob, go ahead. 4 MR. SHAW: Just to add a little bit to that, we 5 didn't do a careful analysis to say, What are the various 6 options that we might consider for this. 7 One of the areas that Robert mentioned was 8 comfort, and it was important that the technologists that 9 were involved were comfortable with what we used. The other { 10 certainly is the experience that we had from the seismicity 11 ownero group that enabled us to say, This worked in a 12 similar situation. Let's not try and reconstruct the rN j( ) 13 methodology as well as the technologies that go into this. 14 But I would also offer that I currently right now 15 have under contract an effort to look at the variety of ways i 16 that one could look at this. Robin listed about five 17 different techniques that can be used, and I have a 18 contractor who is trying to look at the whole variety of 19 things on the whole spectrum, and then to identify the 20 strengths and weaknesses of each of these as they would be 21 applied to a performance assessment methodology for a high 22 level waste repository. 23 MR. POMEROY: Bob, we'd certainly 1. Very 24 interested, in the future, when you do get some results from g-~ k( / 25 that, to hear that.

31 i a 1 MR. SKAW: Okay. 7 x ( ~ ) N- ~ 2 MR. POMEROY: Robin, could you expand a little bit i 3 perhaps on how you determined the completeness of the logic 4 tree? In other words, are you sure that you have all the 1 5 branches? 6 [ Slide.) 1 7 MR. McGUIRE: I think a more critical question is 1 8 are we sure we have all the nodes. ) l i 9 MR. POMEROY: Well, that's certainly a more j 10 critical question. 11 MR. McGUIRE: There are some specifically that are J l 12 missing -- for instance, human intrusion -- some that we did i [) 13 not have expertise nor time to get into. 14 I think we would say, as Bob expressed at the J 15 beginning, that those that we have deliberately neglected 16 could be added in the same vein, in the same methodology 17 that we were illustrating here. 18 MR. MOELLER: Excuse me. i 19 MR. McGUIRE: Yes? 20 MR. MOELLER: Could you go back to the logic tree 21 again and clarify something for me. 22 MR. McGUIRE: Certainly. 23 MR. MOELLER: The previous slide. fs 24-(Slide.)

(

25 MR. MOELLER: No, the one before that.

I 1 32 I 1 MR. McGUIRE: Okay. ,_s l ) l 2 MR. MOELLER: The logic tree itself. j 3 (slide.) ] 4 MR. MOELLER: I follow that, you know, you have an i 5 external impact or stress, and it produces a source term, 6 or, you know, it causes something to be released, and then, 7 though, I guess what I need explained, you next have the l 8 hydrologic properties, and if I see the node in the two 9 branches, what I'm confused on right at the moment is can't 10 there be more than two characteristics of the hydrologic 11 properties? I mean, you're giving me two choices. Am I ( 12 making sense? It either can go up or down, you know, one of () 13 the two splits. 14 MR. McGUIRE: Yes. i 15 MR. MOELLER: Aren't the hydrologic properties a 16 continuum of just a host of things that could occur at that 17 node? l 18 MR. McGUIRE: Certainly. One of the -- 19 MR. MOELLER: Maybe if you'd just give me an r 20 example, it will help straighten me out, e i 21 MR. McGUIRE: This particular slide is just a 22 cartoon representation. t 23 MR. MOELLER: Right. i l 24 MR,. McGUIRE: In each of these nodes, we can Os - T 25 represent with each branch first a set of properties. For ~

1 I 33 i 1 hydrological properties, we could represent the porosity of I V) i (_/ 2 the system, the number and degree of fractioning, the I 3 chemical retardation of -- 4 MR. MOELLEnt Fine. That's what was troubling me. l f 5 MR. McGUIRE: Yes. i 6 MR. MOELJER: So you can do all of that. 7 MR. McGUIRE: We can do all of that, and we are 8 not restricted to just two nodes coming out of each branch. 9 MR. MOELLER: Fine. Okay. Okay. 10 MR. POMEROY: But let me follow on to that, then, t 11 because that is my question with regard to the completeness 12 issue. That is how do you determine, or how have.you !{ } 13 determined in these illustrative examples that you have 14 incorporated that complete spectrum of possibilities that 15 Dr. Moeh11er refers to in this process? 16 MR. McGUIRE: Do you mean with respect to the 17' number of parameters, or the range, the range of values? l 18 MR. POMEROY: I'm thinking.of the number of 4 19 possible branches at the moment. 20 MR. McGUTRE: I look on the entire process here as 21 an evolutionary one. What wo have done is, as you'll see, 22 is to make a first cut at the overall logic tree to get the 23 parameters of a system, and r,ome first cut at the range of 24 possible values that those parameters might take. O_ 25 (Slide.) e m.

34 1 MR. McGUIRE: As part of the evolution of the p_ j \\ \\~ / 2 understanding in this kind of probablistic context, one goes 3 through the calculations and determines what nodes -- that 4 is what values of parameters and their probabilities +- are 5 most critical for the calculations, and what are less 6 critical, which ones are less critical. One then can 7 collapse the ones that are less critical and expand the ones 8 that are more critical. 9 MR. POMEROY: Right. And I understand that some l 10 of that is very well known, and some of that may not be l 11 known at all, and I'm getting to the question, I think, of 12 how is that determination made? .( ) 13 MR. McGUIRE: It's made -- l. 14 MR. SHAW: Let me suggest that as we get to the 15 more detailed presentations of the individual technologies, 16 that I think a lot of that will come out in that particular 17 discussion. 18 MR. POMEROY: Fine. 19 MR. McGUIRE: All right. We have then an overall i 20 logic tree that represents hydrological flux; potential 21 earthquake occurrences causing canister failures; potential l. 22 water table changes from fault rupture associated with those 23 earthquakes; potential volcanic activity and associated 24 water table changes; affects of borehole stability; affects l g ~ 25 of mean canister lifetime, or, in general, canister design;

L- ] p 35. I 1 affects on waste solubility, the rock fracture modell d 2 affects on porosity and affects on retardation of the 3 nuclides. 4 That represents our first cut at the system 5 analysis, and what I would like to do is then walk you 6 through the illustration of results so that you see the 7 entire picture. Then we'll go into some of the details on 8 how some of these branches -- in particular, volcanic 9 activity and tectonics, earthquake activity -- have been 10 derived. 11 The logic tree and representation of the branches 12 at some places may look very simple. ( 13 (Slide.) 14 MR. McGUIREt As an example, I will show you the 15 first node here, which is hydrologic flux. In our 16 illustrative calculation, the node has three branches with 17 three values of flux and associated probabilities. That 18 looks rather simple, and it is. One must realize that these 19 values and their associated probabilities are themselves a 20 result of quite careful and detailed analysis which may also 21 have been done by the logic tree approach which is 22 synthesize and collapsed into just those three values and i 23 their associated probabilities. 24 If you're familiar with PRA analyses, this D 25 represents, in concept, the so-called pinch point in a PRA

i 36 l 1 analysis, in which you have a large number of branches and 2 you pinch them down into one or several variables so that l 3 you don't continue to expand the tree geometrically beyond 4 reason. 5 Each of the inputs here to the logic tree has i 6 behind it, some very careful analysis, and we'll illustrate 7 some of that. 8 (Slide.) 9 MR. McGUIRE: What I will do now is show some l 10 results which will correspond to these two cartoons here, t 11 the set of concentrations versus time for the end branches 12 that were calculated, for the set of parameters and the [ ) 13 associated CCDF, or representation of the range of possible 14 radionuclide releases and their probabilities. 15 (Slide.) 16 MR. McGUIRE: What we will be illustrating are the 17 results for Neptunium 237 as an illustration. We've 18 conducted analyses for other nuclides and if one were to 19 compare that CCDF to a standard, one would have to look at 20 all nuclides and sum the total of all releases from all 21 nuclides. 22 We can do that and we can illustrate that, but the 23 simplest way is to illustrate that with one nuclide to start t'. 24 with. That total logic tree that we have developed has b 25 1,022 end branches, and plotted here are the results of that l

i 37 i set of calculations shown as cumulative curies on the left .,m ./ ) b'# 2 side, versus time going from zero to 100,000 years on the l 3 abscissa. 4 We have, in this case, color-coded the curves to 5 the value of flux, so that, for instance -- if you remember 6 that first node of the logic tree, we had three values of i i 7 flux.- The highest value of 4 millimeters per year is shown, 8 results of and branches that initiate with that branch are 1 9 shown with the dashed blue line, and the values that -- the i 10 end branches that result from an assumption at the beginning ] 11 of 1.6 millimeters per year are shown with red lines, and 12 results of the 0.5 millimeters ars off-scale on this curve; 'l [ that is, they don't show up, because most of those are zero 13 I 14' release. { 15 We show also a dashed line at 100 curies, because i 16 that's-the proposed EPA limit for Neptunium 237. So, this ] 17 represents the 1022 sets of calculations of source Term and 18 hydrological transport and, of course, associated with each 19 of those branches is a probability. Now, we're most ) 20 interested in looking at a comparison of this level of 100 21 curies at the year / time 1/10,000, at the year 10,000, 22 because that's when that standard is to be applied in that 23 standard, the EPA standards. G 24 What we can do then is expand this time scale from , b 25 the year 00 to 10,000 and show the curves over that range. l

38 j 1 While that's plotting, there are two special cases -- l 7-4 2 MR. HINZE Excuse me. Where are you at, what 3 kind of distance? J 4 MR. McGUIRE: We're at the accessible environment. f 5 MR. HINZE At the accessible environment; and l 1 6 where did you place that? i 7 MR. McGUIRE: That's at the carbonate Aquifer. 8 MR. HINZE Vertically below the site? [ 9 MR. McGUIRE: Vertically below the site about 150 1 10 meters at one end and about 250 meters at the other end. l 11 MR. HINZE You just averaged that out then? 12 MR. McGUIRE: No. As part of the hydrological [ ) 13 calculations, we have six paths that represent six possible 14 paths of water flow through the site, some of which are 15 through the vitric welded tuffs, some through the zeolitized 16 welded tuff. Correct me if I'm wrong, but I believe the l [ 17 vitric tuff has thn pathway of 250 meters, so what -- we've 18' used six specific pathways with their associated distances 19 between them and the repository and the Carbonate Aquifer. 20 MR. POMEROY: Just expand that a little more, 21 Robin, if you will, in terms of how the values of the flow 22 are determined, who determined them, what are the 23 probabilities involved? Bob is going to be telling me that r"'$ 24 that's going to be coming shortly. I can hear that. (d 25 I think you've gotten to the 1,022 end points,

39 1 perhaps too fast, Robin, and we will wait. ,_(D \\/ 2 MR. SHAW: Please look at this. It is our intent i 3 to simply put in front of you the results, because we felt i 4 that the detailed discuasion would be more meaningful to you 5 if you first saw a brief summary of the results. We'll come 6 back and go into more detail. I ask you to look at this l 7 just simply an overview at this stage. 8 MR. POMEROY: Fine. [ 9 MR. MOELLER: On the previous chart which, you l 10 know, covered many more years than this one, you've shrunk 11 it down and you're showing us the detail of the first 10,000 12 years, but you had a number of curves that went far above ( 13 the hundred curies or whatever it was, limit for Neptunium. 14 You have several curves here that are abovo 4 or 5,000 15 years. i e 16 Can you talk about that a little bit? 1 17 MR. McGUIRE: Yes, we will show through the 38 sensitivity studies, what sets of combinations of parameters 19 caused that. 20 MR. MOELLER: That's fine. I'll wait. l 21 MR. McGUIRE: There are two special curves on here l l 22 which are shown as almost vertical and they are special 23 cases of volcanic release that are shown in that way purely 24 arbitrarily, but so that they stand out from the shape of 25 the other curves. Professor Sheridan will talk about what --c-e. m

..3 .l 40 i those represent during his presentation, j (3 '. 'y) 2 So, here we have the set of 1022 curves and 3 they're representing the cumulative release over the period 4 of zero to 10,000 years. Again, these don't represent in i 5 any sense the probabilities associated with those sets of 6 combinations. To represent that, what we can do is form a 7 histogram at 10,000 years on the righthand side of this 8 scale, by putting into bins, the cu'.ves and their associated 9 probabilities. i 10 (Slide.) 11 MR. McGUIRE: This represents the histogram, again 12 color-coded by.the level of flux on that first node. And we (/] 13 see that those very high levels of radioactive release v 14 associated with levels of 1, 10 and 100 and higher curies, i 15 have very low probabilities. In other words, those curves ,i 16-have low probability associated with them. l 17 They represent -- here the blue curves, again, 18 represent the dash blue lines, or the level of flux of four 19 millimeters per year; the red curves represent a level of 20 flux of 1.6 millimeters a year; and the green curves 21 represent the level of flux of 0.5 millimeters per year. 22 From that we can sum from the top end to form a 1 23 CCDP. 24 (Slide.) -{ 0 \\ V -25 MR. McGUIRE: On linear probability scale, that

41 1 looks like the entire distribution is off to the lefthand j . (V) i 2 sider so what we can do is show that on logarithmic 3 probability scale, starting at 1 and going to 10 to the 4 minus 5, with normalized release on the abscissa. That is 5 the value of cumulative curies divided by 100, which is the 1 6 EPA -- the proposed EPA standard for Neptunium 237. 7 [ Slide.) 8 MR. McGUIRE: And this curve shows that CCDF 9 plotted on that scale, and also the proposed EPA standards i 10 which are -- that normaliza release of 1 or 100 curies, is 11 not to be exceeded with the probability of 10 to the minus 12 one, and 10 times that release is not to be exceeded with a 13 probability of 10 to the minus three. 14 So this represents that CCDF on log log scale and 15 can be compared to a standard, such as proposed EPA 16 standard, or any other one that one wishes to compare to to 1 17 get a feel for where that set of assumptions leads in tenns 18 of absolute levels of release. l 19 We'll be looking -- to look ahead, we'll be 20 looking at sensitivity studies of that color coding type, 21 with respect to cumulative curies released, versus time,.and 22 also with respect to how much this CCDF changes, as we 23 change input assumptions. In other words, how sensitive is ~ l. 24 that CCDP to some of the input parameters and their -25 probabilitics that we've chosen. And we will illustrate l l

l 42 1 that when we get into the details of the calculations. p_ N 2 (Slide.) 3 MR. McGUIRE: The purpose here is really to show 4 the overall methodology, where it leads, and then we're 5 coming back to some of the inputs and their associated 6 interpretations. 7 And so, just to summarize where we've been -- we i 8 have sets of curves of cumulative release versus time. We i 9 can form a histogram of those at 10,000 years, and from I i 10 that, calculate CCDF's on linear probability scale and i 11 logarithmic probability scale, and ultimately can compare 12 them to any proposed standard that is used to judge the ( ) 13 acceptability or unacceptability of a proposed repository. 14 So, I'll stop there, and give the podium to l 15 Professor Sheridan, who will describe the inputs that were i 16 developed for the volcanic node of that logic tree. And 17 then I'll come back and give you some additional insight 18 into the inputs that were developed for the tectonic node. 19 So, we'll talk about node number 4 and node number 10 5 now, a.id then come back and talk about node number 2 and 21 3, 22 MR. SHAW: As Mike is coming up, let me just make 23 an intermediate comment here. I did mention a couple of l gg 24 times, the

s/t that this was meant to be illustrative.

At V 25 the same tite,e, we also asked our experts to be reasonable -- l

43 1 reasonable in the sense of choosing values that would be 2 reasonable for Yucca Mountain, so that we were trying to 3 generate a methodology and a result that would be reasonably 4 within the range of parameters for that particular site. 5 It's adaptability to other sites, I think will be -- is and 6 should be obvious as we move along. 7 But, we would ask you to note focus too much on 8 the details of what particular parameters were suggested, 9 but we still feel and hope that they're in the sense of 10 reasonableness. 11 Mike? 12 (Slide.) ( 13 MR. SHERIDAN: For Yucca Mountain, the volcanology 14 component of the analysis is very important because some 15 researchers have suggested that the volcanism could be a 16 potentially disqualifying effect at Yucca Mountain. 17 So, the main objective for -- for the volcanology 18 mode is to be able to accept -- accept models by various 19 researchers, or in fact, any models that are currently 20 available and to apply them to the general logic tree. So, 21 as I'm describing the reasoning that went into the 22 volcanology node on the general logic tree, you should keep 23 in mind that there are these two trains: one train that is 24 connecting all of the possible components of the analysis 25 and the other train la the method for determining the h

t 44 ) 1-specific events related to a node and their potential l 7x 2 effects. 3 And also pisase realize that although there may be 4 many potential effects, for example, from volcanism that 5 only those most -- considered to be most critical, or those 6 that would affect the system are included; so that the 7 volcanology node is compressed into a few elements. And i 8 I'll present this -- the elements of the volcanology node in 9 this diagram which you have in your -- in your handout. But l 10 the other parts of this presentation will not be in your 11 handout. 12 And the main -- the main part of the volcanology r ( 13 node considers, first of all, that there would be no effect l i L 14 of volcanism on -- on the repository or that the volcanism 15 would affect the hydrology in some way that would lead to a l 16 change in the release rate from the ambient condition. 17 There are two additional branches from this nwie, 18 that lead to -- could lead to direct release of the surface, 19 and they are not connected to the other nodes, so that l' 20 they're more or less independent. One of those would be 21 direct release thrcagh magnetic activity. In other words, 22 red hot magma arrives within the repository and affects the 23 canisters or the material that was original in the canisters l l 24 in a direct way. lb 25 Another potential effect is the hydromagmatic --m... .m,-. a 3 w s-c y+-~.

45 1 release of materials related to a volcanic event. Now, a ( b \\ 2 hydromagmatism refers to steam explosions generated by fuel 3 coolant reactions, such that the hot magma comes into i 4 contact with ground water near the repository and causes 5 steam blant explosion. j 6 Each of the indirect and each of the direct j 7 releases can -- these effects can then be calculated and j 8 given some values. And for the ladder two, just arbitrary 9 values have been given of -- of the amount of release. ] 1 10 Considerable more study needs to be made, so that the actual 11-loss of material is probably not very well represented by 12 the model, but the effect on hydrology is probably quite i ( 13 accurate. 14 The two significant effects here would be -- the 15 - first type of effect is if there was a conduit -- a vertical-16 table or a conduit of magma intruding into a saturated zone 17 that would cause some compression and a rise of the magma 18 table -- water table associated with this event. 19 The second type of effect would be a combination 20 of these intrusions causing a general blockage of the 21 subterranean flow of water, leading to general rise in magma 22 -- in the water table. And in the area of Yucca mountain, 23 there is a peculiar configuration of the water table with a 7-' 24 strong drop-off upstream in the subsurface, b 25 And if that plateau were to move past this site, m .m m-m -m.

• v-

=w+ =

M k 4 f i l-461 t>,, 1 'it could cause considerable rise in the' water table'. jr-w e . s,...I /2 These were collapsed into to potentialt effects on s ~ ~3 -the' water table: the first being a very smallyeffect,'which~ 4 {;[ C 4 we consider to be essentially zero, for the sake of the 5: argument; and the other, a considerable blockage tha't would-

)

/" 6 -cause an 80 meter rise in the water table. V 7 These were arbitrarily selected. 1 8 MR. HINZE Is it reasonable to ask at this point D 9 -- if it's not, just tell~me -- but is it reasonable to ask, l[ 10-for example,-how you interact climatological predictions on 11 the hydromagmatic release, and how those all interact? I 1 12 think that's important. l L ' (()N ' 13' MR..SHERIDAN: No. This is a perfect time to ask l ' \\. l= 14 that ques'. ion. I should have actually said something about t -15 that. ' this diagram, I haven't really connected the i l 16 hydromagmatic to this node because actually, the l I l17 hydromagmatism'follows. 18 The probability of hydromagmatism depends on the 19 ' hydrologic condition. Under the current conditions of a 20 ?very dry region, the probability for a hydromagmatic event 21 is much lower than it would be, say, during pluvial times. .22 What we did was take a factor of ten difference between the L 23 . probability of hydromagmatism during-a time of heavy

N

. -24 rainfall and a higher water table. i E - 25 Let me get back to how these were -- these are y l L

l 47 1 r 1-numbers that'are the best possible numbers that I, being the 1 u'/'it \\ #U

2:

only member ofLthe team,'could come up with at this point.. f 3: But' Bob'will point out-later how we would hope to reduce the q l 4 uncertainty in this type of estimate and come up with a'more 1 l 5 accurate estimate of what this could be. 6 MR. HINZE: And-so what you're'doing is setting . 7. off different branches, depending upon the various l 8 climatological conditions that would be, in the experts' l \\ 9 opinions, possible within X number of years? L 10 .MR. SHERIDAN: That's correct. 4 J l 11-MR. HINZE: And how about things like the thermal 1 l 12 effects of the package itself and its affect upon the J q ) 13 presence of water and so forth? Is that also taking into l 14 account those nodes? 15 MR. SHERIDAN: 'All of these factors have been 16 taken into account, even the thermal effects of dikes on the l-water table' causing expansion. But it turns out that many, P

17

~ 18 many of these effects cause such a small change that they 19 could be put into this category of zero la the water table, ~ 1 -20 and we just collapse those arguments for the present and l 12 1 . consider those to be non-essential arguments for the l 22 preliminary. evaluation site suitability. l. L 23 MR. HINZE: But that's a subjective decision on 24 the part of the experts. L( 25 MR. SHERIDAN: For sure, it is. l? 1 1 j

a. P 48 e a f- .1 MR. HINZE: Okay. '2-MR. SHERIDAN:.And I'think that'one person is not b 3 capability of making a valid decision-in this' case', but this 4 was my decision. Actually, the groupidiscussed these possibilities, and this was the conclusion of this 5 n 6 particular group who worked on the problem. '7 MR. SHAW: I would just reemphasize that we made 8 the conscious' decision to have one expert in each area and 9 asked that person to develop what he thought was a L p 10 reasonable description of what was' going on. Phases beyond m.

ll-what we've done so far would look in more detail with more 12 experts at the particular of what would go into such logic 4

13 diagrams. 14 MR. HINZE:. Illustrative. 15 [ Laughter.] L l '16 MR. HINZE: Thank you. 17 MR. POMEROY: Snat's a nice word. But1to Bob, I ~18 want to further ask one question. In the illustrative results that you're showing us here, one person is [ 19 20 responsible for the assignment of probabilities within his 21-area of expertise, and there's no change as a result of the' i 22 team interaction? 23 MR. SHAW: No, that's not entirely true. 'There 24-was team interactions. What you're seeing here with Mike b,s 25 Sheridan is an example of a person who essentially worked in )

i ~ 'it 49 1 isolation, but he talks about the climatology, so he had p ~ \\s 2 discussions with our climatologist. In some of the:other- = 3: areas, there was very intimate interaction. A good example 4 is hydrology and gaochemistry, where the interaction was. 5-extremely strong between the two individuals. So there was l 1 6 ~ an' interaction in some cases, but in some cases, that's a 7 stronger influence; in some cases it's a very week' influence 8' on the results. '9 MR. POMEROY: In those cases where there was .10 strong interaction, the illustrative-probabilities that ] 11' you're using are the result of some. agreement between those 12; ' experts in this case? l l 3 13 MR. SHAW: That's correct. l IV-1 is MR. SHERIDAN: Let me try to respond l 14 l -15 question, also, in that the primary cbjective.- ,s l' L 16 particular node is to incorporate allrof the models that are y 1 17L currently existing, and to be flexible enough to incorporate 18' new models as these new models develop 19 In my compilation of the general scheme, I did 20 have conferences with Bruce Crowe, who is a DOE principal 21 investigator in volcanology on this problem. I also had s -22 some conferences and field trips with Gene Smith, who is the '23 state of Nevada principal investigator in a volcanology 24- -component. IV 25 From my independent perspective, I perceived that i ,l-1 N

o.. s. R 4 'L 1 j' 50 (,g. 1 both Bruce and Gene =were moving-in the same' direction I'm .e y kk k

2

, moving in that both of'their models could be incorporated ] 3 vary. easily into this and could be tested. 4-I think that their models would be actually quite ] Si different, and the curves that we saw would be quite H'

i 6

different, but the probability assigned to those curves I 7 think is something that a-group of experts could decide on-- and come to some agreement on. 8 L 9 So the volcanology takes into a case rather 10 extreme -- all of the extreme views, from no effect to a 11-very serious-effect, but the probability assigne'd to those ll-l -12 effects'is what is critical and what I-think would be a-t l# ) ' 13 logical direction to proceed from here. ' (' J +- 14 (Slide.)~ 15 MR. SHERIDAN: Now, the volcanology question is 16~ one that is a conditional probability, so that these y 17 branches, the probabilities we saw on.the branches, I'll'put l W' 18 this back up, put this diagram back. 19 The probabilities, the numbers that you see in 1 l l 20 here are a result of a conditionalLprobability that there is .21 a volcanic event, and conditioned on the occurrence of an j 22 event,-there is a spatial probability. So there is a 1 p 23 temporal probability and a spatial probability. 24 For the temporal probability, there is some lk 25 disagreement. But I should say that there is a feedback l l

l 7,

51. V 1! effect between temporal and spatial that sets boards on the 1 . prm: 1 $ Q ). p: .2' volcanology-. question. And I want to bring.that'up at the e l l.i :. [Q 3' very end, that as-a result of this analysis, we've set i 1 4~ limits on'what the extreme probabilities can be. And 1. i 5-actually, the estimates made by_our model-are~very j L 1 6-conservative estimates. And I think I will.'show why that -7 is. 8 The temporal estimates are.taken directly from L 9 Crow's estimate of what the frequency of a volcanic event is U L [ 10' ' going to be in this area. And let me say that is not going 11 to make a large amount of difference if that estimate is off 12 by an order of magnitude or a factor of 10, in terms of the 1l b 13 total analysis.

1. -

u '1 14 (Slide.] l 15 MR. SHERIDAN: The spatial question, however, is a 16 critical one. And up to'this point,.I don't think that l 17 there was a good solution for the spatial determination of a i 18 problem, because it seemed like the probabilities depended l I 19 on the initial geometries that the various investigators ) 20 were accepting. i t. j 21 So that we took a different perspective to look at-22 just-what is the distribution of volcanoes in a typical 23 ' volcanic field. And for illustration, I've shown here the 24 Pina.Cate volcanic field that's in the Basin and Range-25 Province. This happens to be in sonora, Mexico, or just

5 4 52 l 1 South of'there, it's on the border. But it is typical'.of-

(ry Q

2-others.. 3 And I studied a large number of volcanic fields, I ) L L4 think over 40, in the Great Basin, to look at the general' 1o 5 properties. And the general property that we can see for a l 6 volcanic field is that there is a concentration, each dot on g H 7 here is.a conduit, a volcanic conduit, and there is n' 8 concentration of these conduits towards the center of a s. 9 field, and a diminution in number, and an increase in 10 spacing of conduits moving away. l 11 It is not easy to draw a border, an accurate l 12 border, around a volcanic field. It is not easy to D , (/ 13 calculate the area of a volcanic field, because it is not 14 well-suited to this. It is like calculating the area of a 15 cloud.- 16 MR. MOELLER: Excuse.me. What are the dimensions

17 here?

Is this'a mile or five miles? 18 MR. SHERIDAN: Oh, this is huge. Yes. This would 19 be probably I-believe 50 kilometer, top to bottom. 20 MR. MOELLER: Okay. 21 MR. SHERIDAN: Something like that. 22 MR. MOELLER: Thank you. s 23 MR. SHERIDAN: So the first characteristic is a '24 concentration of conduits towards the center of the field. 25 The second is that the distribution ha a more or =

I l 23

v tI.

1-lessLelliptica11 pattern, so that we could then define r,7.c}.- 5 . \\-n 2. ' elliptical patterns-for all of the fields for comparison, to .3 -see what'we might expect at Yucca Mountain. And we can see 4 that there is a general access of elongation of the fields, 5 and a general aspect ratio, a general length to width ratio. 6 And I made these calculations for all of the fields r.xtent 1 7 .in the Great Basin, and this seems like a two to cne ratio 8 for the aspect ratio. It is fairly good. They go as high L 9 as five to one. \\ 10 Now, within this field, there are some alignments 1 11 of volcanic conduits that are very strongly aligned, and L 12 they'are on fracture systems. So we can tell what the .{ ) 13 orientation of the feeding fracture system is for the 14 volcanoes. L 15-And for this case, for most of these elongated 16 zones, the feeding fracture aystem is parallel to the 17 . elongation of the; field. And-the general interpretation in 18 volcanology is that the stress fitid, the minimum stress L 19 direction, then, would be perpendicular to the-access of the -20 feeding conduits, and'also perpendicular to the general 1 21 access of elongation to the field. This is going to be an 22-important component of the model as we come back, because -23 different investigators have proposed different shapes for l L 24 their fields. (h V. 25 The model that we used in this calculation of the l. 1 I

t s

54 ~ r-i i probability was then a two-dimensional Gaussian distribution ~\\_sb i ( 2-of cones,Jassuming some center. And using this. sort of 3 distribution, we could use a standard deviation, which would I 4 give-a general length' dimension and aspect ratio and 5 elongation of the field, and then, by a: Monte Carlo 6 simulation,' generate a large number of events, and count the {

(

'7 number of events that intersect the repository, given a 8' specific location orientation aspect ratio of a volcanic i 9 field to determine the probability. And this is how the 10 probability was decided for this problem, i:L 11 Another interesting aspect here, probably not q L12. visible from the back of the room, are the craters that ~ 13 represent hydromagmatic events. In this case,-there is one q { 14 here, and all of these green ones we see, actually there are 15-11-in this-field. And for each of the volcanic fields in 16 the Great Basin, the numberLof hydrovolcanic. cones was L L 17 calculated as a' ratio that gave the probability of a 18 hydrovolcanic event. It turns out to be about 1.in 100. 11 9 But that also depends on the climate. 'And during Winter 20 climates, apparently, some of the cones-in this field, for 21-example, occurred between 10,000 years ago and 18,000 years 22 ago, which was during a pluvial period, and apparently there l '23 is an increase in frequency of this type. 1 [T ~24 So that is the type of basis for information. - (\\.j 25 (Slide.)

55 i 1 MR.'SHERIDAN: 'Now what is apparent from'this _ ';q s(A / ; T 2 'analysistis that-the orientation of the volcanic field-is { 3 'not independent of the general state of stress in the crust. g 4 and that the general-interpretation, the assumption made for 5 this model is that the elongation of the volcanic field 6 would be controlled by deep stresses in the crust above a 7 zone of generation of magma. J l 8 (Slide.) 9 MR. SHERIDAN: Towards the surface the local 1 10 stress variations could cause different alignments of i 11 fracture systems towards the surface. 12 Now I'll just give an illustration of this problem ' [VI 13 with the next diagram, which shows the three dimensional i-34 form of a dike or a feeding conduit for a volcano with the 15 least principal stress being perpendicular to this feature j '16 atLdepth but due to changes in the stress field near the l [ 17 surface caused by topography, caused by differences in rock l 18 type, caused by local fractures or faults or whatever, there L 19 can be rotation of the dike segments and typically multiple 20-fractures of smaller dimension due to local stresses. l. != !21 (Slide.] -22 MR. SHERIDAN: The. reason for considering this is I L '23-that-the repository will not be at the surface. The l 24 repository will be at a few hundred meters depth and 7-s .l 25 -therefore the interaction of a dike or volcanic conduit has y c ..,. - ~..

? l e 56-j - 1l to be. considered. I '2. Then data was gathered from'all the published data

3

-that I could find on the length of dikes, the number of dike 4 segments,[the orientation of dike swarms was taken into 5 account and the model can then incorporate an intrusion-6 model based <n) the dikes or volcanic conduits that'uses the 7 average length of the dike and the average orientation of 8 the dike and given the standard deviation, so it is. 9 completely based on the statistical aspect which again the 10 confidence in this type of data can be improved from just g l 11 one person reading a little bit of the' literature to having 12 input from many investigators and using also level of ) 13 confidence developed from scientific discussions. ( l c l 14 [ Slide.]- - 15 MR. SHERIDAN: The basis for this_ type of j 16 interpretation comes from_other studies where actually fault 17-mechanism showing the orientation of the.least principal ~ 18' stress to be perpendicular to the axis of dikes. This is 19 one caso from: Iceland and here is another case for dikes -- 20 this would be in a zone of pure tension and this would be in 21 a zone where there is a transformed faults strike slip ' 22 -faults but the orientation of the dikes actually is 23 concentrated in a zone between offsets. 24 Techtonic considerations are important in the 25 volcanic evaluation also but these can be incorporated into f e-m

x i 'i 57 j l 1L _the model and the model is actually a very simple geometric: , 18;: n I;, model. L '- 21 ~ 1 3 MR. HINZE; Mike, as you did this, did you make an 4 ' appraisal of how well you knew what you were estimating? 1 5 MR. SHERIDAN: For example, you mean the average. 1 6 length of dikes, how good is that knowledge? j 7. MR. HINZE: And -- right, and the -- l 'B' MR. SHERIDAN: It-was poor, absolutely poor for- ] 9 good reasons. JU) MR. HINZE:- But did you assign a numerical-11 probability to your estimate? i l 11 2 MR. SHERIDAN: What I did actually was determine l -(

13.

bounding values and I took the worst possible case. 14 MR. HINZE: But you may end up from zero to a L .15 hundred with that sort of situation, isn't that -- 16; MR. SHERIDAN: But I didn't. 1 1' l '17: I'll show you. An amazing result _came out of this, L L 18f absolutely amazing. 19-MR. HINZE: Good. l L 20. MR. SHERIDAN: At least to me it was amazing f 21 because of arguments I had heard before that logically one T 22 would say, would argue in one way, we'll find that there 1 -23 actually is a maximum effect of intrusion and a maximum '2 4 f effect of conduit location. 25L MR. HINZE: I know going back to the old Eastern i l

-~ ' k 'I, % }f 5g i l' earthquake study one of the things we had to do was'to k 2 estimate how well our guess was-and you are doing that, do I 3: understand ~that correctly? 4 MR. SHERIDAN: I could give you for example in the 5' parameters of l'ength of dikes I could the average.and-t 6 standard. deviation of published literature, if that is a ) 7. good dike. We-could look at the focal mechanisms for faults 8 ir. terms of the orientations and saying that within certain-9 limits what the probability of the dikes having such-and-10 such an orientation could be so there are ways of doing that i 11 but our emphasis was really not in the analysis and not with r 12 coming up with the best possible answer we can but coming up ((I 13-with a reasonable answer based on. data that.is easily 14 available so that we can demonstrate the methodology. 15 MR. SHAW: Let me take:that response a step. 16 further. ? 17 In my mind part of the purpose of the first cut is 18 to avoid doing what you just asked. -Our criticism of DOE is. ( 19' that they are going into-extensive detail and then 20 eventually they'll get to.a model that ties everything - j 21 together. Our point is that you ought to have a model right 22 -now that ties everything together so that you make the first 23 cut assessment at what is important. 24 Rather than ask Mike to go into the detail that h \\~ 25 you just asked, we go through a first cut and say, is that

1 <4 59-1 j ~ q =1 iniportant to. ask that particular question. If.it is not,:if '?\\ u AI 2 volcanism-and the effects that he calculates = don't appear to 3; be-particularly important and he uses bounding values and. 41 things of.that nature then it may not be vital to answer 5 that'particular detailed questions and-so part of this 6. ' attempt is to say what are the important detailed questions 7 that'need to be addressed and what are those that at least 8 ~ right now based on what we know are not particularly 9-important to be addressed? 10 MR. HINZE: If your estimate is incorrect -- do 11 you make an estimate -- and you have extreme bound 12 conditions on that, then the results of your modelling may j( ) - 13 be totally incorrect, or at least they may. lead you in the 14 wrong direction. 15 MR. SHERIDAN: It depends on the-probability 16' assigned to-the extreme values, and 16 values have very low-17_' probabilities. They will have a very low effect on the 18 overall model. 19 MR. HINZE: My opinion on it is, that it's very 20 important to have some feeling for how well you know these 21 parameters. I'm sure that Mike has some excellent data in l i 22 some areas that he can really constrain, and there are other 23 areas -- and those areas may be important -- where you have 24 to make a judgement, but nonetheless, there is an extreme l 25 bound condition which may have a profound effect upon the

m .~ 60 l _- results:and mayflead in the wrong _ direction at this early ,s N~/ 2' ~ stage. I think that's something_that I'm concerned about. L 3- [ Slide.) ll MR. SHERIDAN: There's another important question 5 that Dr.~Pomeroy mentioned at the very beginning was the 6 effect of public opinion, and I think that volcanism is an '7 . area that is very sensitive to public. opinion. It may be 8 quite possible for nine out of tren scientists sitting around 9 a table to come to'a conclusion of what the probable effect' 10 of the volcano is, but if we took the opinion of people, 11- 'we'd find quite a different effect. It's quite amazing 12 also, depending on whether that effect of the volcano is /-k L{ 13 directly-due to nature, or whather that effect of the 114 volcano is due to something that man did.- 15 For example, people live -- I just returned from- .l 16. Naples -- 3 million people live right at the base of an . l 17 active volcano that, if it were to erupt, would kill them, '18 and they're not concerned about this. However, if some sort t '19 of man-made facility were put at the bottom of that volcano, 20-immediately there would be protests about the danger of that 'l 21 facility. So, that's not an issue we address. '22-MR. MOELLER: Back on the previous question, in 12 3 other words, you're trying to determine bounding values for y fs. 24 the key parameters. You have enough confidence in these (Q 25 bounding values that if it comes out that that parameter m

. t :i - b b 61 . p t e i 1-isn't important, you're fairly confident'in the result. i ss .( o 2 Mk. SHERIDAN: That's correct.' b ' 3- -MR. MOELLER: Okay. i 4 PGt. SHERIDAN: That's right. Actually, there is a 5 great deal of data on the volcanicLquestion if we take it 6 from the point of view of geometry, because we can see j 7 exactly what the geometry is. I want to address this 8 problem of geometry for volcanos, because it's my view that jg i 9 for volcanos,-some orders of magnitude follow a' fractal 1 type l 10 of distribution in terms of geo Ihy. 11 -In.other words, there are large clusters. For 12 ~ example, we could consider all of the volcanos in the p ) 13 Western United States, but we~wouldn't consider that Mt. St. 1 14 Helen's would have a direct effect. An eruption at:Mt. St.

15 Helen's would have a direct effect on the repository.

So 1 16 the question;is, in what proximity would there be feeding i 17 systems that would have a direct effect on the repository?

l 18-The assumption made by this model and one that 19 would have to suffer peer review, or undergo peer review, is 20, what actually is the feeding system.

In this diagram -- 21. this is one from Bruce Crowe showing general distribution of 22 volcanos and Yucca Mountain is the area in green with the 23 Nevada test site outlined, California / Nevada border here -- _e 24 and areas of fairly recent volcanism less than 5 million are 25 shown. What's obvious on here is that there's a spread of m . m m m

• +a-e- -Wh

.i 62 -1' volcanism throughout_the whole~ area.- ,c I J 2 It's very difficult to identify a volcanic. zone, 3 or so on, and I think that there's been a lot offunfruitful [ 4 discussion following this line of reasoning. So'I think, 5 taking into account the magma is generated in the upper part-s. 6 'of the mantle so that probably a distance greater than the 7 crustal distance is not a reasonable sort of distance to f 8 consider for the magma to have generated a plate at a single 9 place, you should really only consider those volcanos near 10 the test site. a 11 (Slide.) i L 12 MR. SHERIDAN: This is another diagram of Bruce ( ) 13 Crowe, and there's been considerable discussion also among 14 the various researchers on whether sites as far away as the 15 one labeled BB, or SB on here, would be, we can see that 16 here is Yucca' Mountain in Green and within -- there isn't a ~ 17 vertical scale on here -- but within about 10 kilometers, .18, there are these volcanos, what's-interesting.is they seem to 19 _ occur in groups. You have this clustering again. 20 The clustering is one of these double derived 21 problems that can lead to false analysis and because -- even 22' on a single cinder cone, and you've probably heard this, a-23 single Scoria cone -- there may be 30 or 40 different vents. 24 But what we'll find is taat, if we take the geometry of the. k 25 vents, we can see that for each clustering, there will be

3 63 J l' this'?.wo-dimensional Gaussian-type of distribution of: vents, i ~ 2 and11n time, they could be considered -- we have'to consider 3-what is.the geometric clustering that is most likely to have 4 an effect:on Yucca Mountain. 1 5-Now we could see that, for example, done here at 6 Lathrop' Wells that if there were repeated events at Lathrop 7 Wells, that the geometry of the past events would probably l 8 predict the future events. And if we take what I did 9 essentially for what I considered to be the most reasonable i E 10 model was that all of these volcanos -- these with an age of 11 about 1 million years to the North, these with an age of 12 about 4 million years to the center, and these with an age-l O-( 13 of less than.1 million years to the south -- all belong to 14 the same general grouping; and they can be considered as a 15 model to begin with. 16 This is'the sort of.model that I think Bruce E17' crowe, who is the DOE principal. investigator on this ~ 18 problem, has proposed. The state of Nevada's state 19 intastigator has taken this approach. He has taken the 20 approach that down here at Lathrop Wells, since these are L 21 the.most recent volcanos, these are the ones that are' going i L 22-to be most critical for the site, and he has then developed l L 23 a model that would take volcanic access from Lathrop Wells L r 24 more or less in this direction following fracture patterns 25 on the surface and said that this has to be. considered. i l !~ i. l.

a 64: 1 But I think all of these various models could be. 1 - vs l 2 taken into account. I'll just demonstrate'one of these-and s-7 3 how this works.: For_the next part of the. discussion, I used .4l a Monte Carlo simulation assuming a center of magmatic: I production: in the center of this field, and then'a-two- '6 dimensional-Gaussian normal distribution about this,.and we-7 can see' that the field has an access, more or_ less,- 8L - North / South and 'the aspect ratio is about 2 'So using'that 9 model generated a number of cases. 10 It's very easy in.that model to say: No, that's 't 11 not correct. It may be the State the Nevada argument is a 12 good one, and we could take-this particular field down here u(/N 13 and assume then, that the access is, say, to the: Northeast - d 14 Hor North / Northeast, and has a certain aspect ratio and then 15 generate a number of volcanic events and calculate then.the 16 probabilities of intersections. And that's -- 17 MR. MOELLER: Would there be any -- and please, _182 ,you know I'm a layman.in the field -- but is there any 19 explanation of the difference in the' ages of the three -20 groupings there, or when they were active if, indeed, you 21' might say they're part of the same family, or -- 22 MR. SHERIDAN: -- Yes. There could be a long 23 discussion-about this topic. We consider the lifetime of a 24 typical volcanic field of this sort in the Basin Range the .,s 25 lifetime is generally about 5 million years. The average i i '{.' ((.! i .~

65

1

?w. ' field:has--aLlifetime of 5 million years, but volcanic fields 11 '-I 2 J11ke Feapo have:a birth, and they have a death, and-theyL l 3 _have a' period of; immaturity, and then they have a period of 4 old age, and they have.a period of'the maturity, and we find-S 5 that, like humans, they are most active in their period of 6 muturity. t 7 If we considered this to be like all of the other 8 fields,.then it is in its period of maturity,_and.the. 9: production rate-is at its maximum during the period of~ i 10 maturity, and that is an upper bounding. If we consider, 11 instead,- that we have three different fields, we don't have 1 12 a migration here. If 4 million erupt here, and then'we had l p p ), 13 al2 million, and then we had a less than 1 million,_then 0 14 there would be some cause to say there is a movement of some 15 mechanism at death. But what we see is sort of -- actually l 16' what'I see is four different events, and these events.are 1 '17 . sporadic, and during the event -- which may last for a few 18 -hundred thousand years--- volcanos appear and then there's a 19 pause, later on some more magma is generated. Well, magma u 20 might be; generated all the time, but not reaching surfaced 21 too. 22-So, I would say that I'm not impressed by any 23 argument of migration in this volcanic field, but even if we 24 said so, then we'd have to say that this particular field is ("Sp . (G' 25 in its youth; and being in its youth, it's not very active.

.-..... ~

iGbh

,x '4 66 .o i il So that the production rate is actually very low,.and if 3 2 what we see here represents some, say,'2 sigma distribution,

, ',' a 3-even 1 sigma or even half sigma, tenth of a sigma, we.can 4..

4 see that it would'take an extraordinary projection from this a, 5 area to have soms probability.here. 4 ' [-: 6 So that is a very low probability, and'we could 7 actually then make, for this particular node, four or five 8 different branches following'different volcanic hypotheses, h l, 9 assign different probabilities as those, and-then determine i \\' 10 what effect that would have on the overall model. That's 11 very easy to do except for the assignment of the 12 probabilities,.which would take expert judgment. It could b (( 13 be done, i 14 (Slide.] y 15 MR. SHERIDAN: So, I will show'you just a few 16 slides of what these results look like, and all of these 17 slides will be in the same sort of aerial arrangement, a .18 scale bar for 10 kilometers. The repository is located here 19-in the green and.the distribution of volcanic material. p In each 'onte Carlo simulation, I used 5,000 20 d 21 vents, gave them an aspect ratio and gave them a certain 3 22 standard distribution or a certain size, and from that, we i q*. can calculate the frequency of events of conduits inside the 23 24 repository. The same tc't of model was then run for the 25 subsurface dikes and their distributien, giving them a

c N...

~-

a...

u., v. 67' c 1 - certain length, aLcertain orientation, a certain standard i 2 - deviation from that orientation, all of which can be varied. '3 The pattern is what's important_and not the= j 1 -4' ' results.. The problem with volcanology.is that some 5 investigators have said that since there are volcanoes so. 6' close.to the site and their so young that'they present'a 7 hazard such that we should stop consideration of this site-j 8. and look for another one, j 9 I think that what is not taken into consideration 10 is the probability of that particular logic line. So, here - 1 i 11 are'few results. First of all, assuming an-aspect ratio.of l 12 2, the North / South. elongation and a standard deviation of 13 events of 4,000 meters, we find that the simulated models 14 completely cover the existing volcanoes. It's a very good 15 . representation of what we've seen over the last 4 million 16 years at'this' location. 17 In that, there e e no intersections with the 7 l-18 - repository. Then, taking dykes, for each conduit, taking a 1 19. feeder dyke for that conduit, giving it the azimuth of 45 L 20 degrees,-in other words, oriented to the dykes directly 21-towards the repository, the worst possible case.

Again,

-22 - there were no intersections, so the probability is very low '23 ' for that particular geometry to have produced an l 24 intersection. 25 MR. HINZE: What happens when you have a five to h I

i i 68 _1 one aspect ratio? Lj f3 (%- 2-MR. SHERIDAN: Okay, let's take a look.. Oh,-five 3 to one aspect ratio;-that would just squash it down.

1. ith a 4

five to'one aspect ratio, you have.to have good airing. 5 It's the difference between a 30.30 shotgun. It depends on 6' what you're aunting. .7 MR..HINZE:- Have you looked at the probabilities [1 8 between five to one? L 9 MR. SHERIDAN:- I haven't used five to one, but l-L 10 five to one seems to me to be illogical because the average l 11 aspect ratio for the basin rings is two to ons. Five to one 12 is rather extreme. We could use five to one -- }() 13-MR. HINZE: The probability will less on'five-to ~ 14 one? 15 MR. SHERIDAN: For sure, it's less. 16 MR. HINZE: But it's still there.

17. -

MR. SHERIDAN: It's still there. 18 (Slide.] 19 MR. SHERIDAN: What if, however, we take the 20 orientation of the field at 45 degrees? Now, this is an 21 interesting case. In this case, this is what the state of 1 22 Nevada is proposing, but proposing the site to be down here. i 23 Then there are some intersections; in this case, 24 of 5,000 events, there were two intersections of the conduit U[y 25 of-the site and if we look at the dykes and their six ,+

,e 69-r . yi l' g y

1' intersections of the dykes, and it turns out there's.about

. Q(,.R%_j; ~ si; 21 double the number of intersect' ions of the subsurface-3 features of the-surface features in this type of 4 representation. -~ at 5 So, what this is illustrating is that'it's i ,6 possible to take, say, the DOE general model and put it in '7 or take the state of Nevada general model and put them in,. 8-and then determine what is going to be the more likely t 9 situation. What you can see is actually the probabilities i 10 are still very, very low, because this is.0004 for the-11 conduits. 12 (Slide.) '( 13-MR. SHERIDAN: So then you can say; what happens ( 14- 'if you increase the size of the field? Suppose you're 15 looking.at a field that is very small and if you do that, 16 this-is for doubling the size so that the distribution of 17 potential volcanoes is much larger than the present one, 18 saying that somehow we're looking.at a very early. stage. We C 19 can see that -- let me get the data in view ---that there 20 would be more intersections and so on. 2.'. Then the argument can go; you could increase the 22 size and maybe the probabilities will get very large by 23 increasing the size. Suppose we increase the size to all of 24 the Southwestern USA. 25 MR. MOELLER: What are the time factors here?

4 3 os i a: s. 70 g_ y l' MR. HINSE:.What are the time factors-here? 7__ }{"

• a 2:

MR. SHERIDAN: If-you increase the-size; I-think -3 the time factors are going to decrease. .4 (Slide.] i 5 MR. SHERIDAN: So, there is a playoff of the size 6 -.the temporal versus the spacial frequencies also that set W 7' a bound.. Just taking the North / South orientation, 8 increasing the size, we find out that below about five 9 kilometers of standard deviation; in other words, J sut 30 10 kilometers North to South in the field, there would not -- -l .11 it wouldn't be a-measurable probabili' >>ith 5,000 events. L 12 . You'd have to use many more. I don't know how many more, (i 13-but many more. 14 Then there's a linear increase in the frequency, 15 up to some quite large size. This would be 15 kilometers, 16 so we can say four times that might be the general sort of '17 distribution which would be about 60. kilometers North to l 18 - South. But after that, there is a' decrease, because you're-19 getting a playoff between increasing the size and the number 20. of those actual events that are going to intersect the (21 repository, specifically from this site. L22 .So, w at happens on this is that there is an upper h 1 23 bound to the frequency of event or conduit intersections. 24 (Slide.] K 25 MR. SHERIDAN: We look at the same type of diagra

P 71

1-for tha' intersection of,the' feeding systems *or the dikes.-

(~ 1 s'

2 We see the same:effect. There is a linear increase in 4 3 probability of that event, up to some point, and-then sort-4 of tailing off. 5 So, what this means -- this was a very surprising 6 result to me,'because I thought that as the size of the 7 fia'id got larger that there would be a greater probability 3 8 of a hit or a direct intersection of a volcano. It turns 9 out that it's not the case. 10 So, the value that we used in our calculation was s a 11 1 in 100; 1 out of every 100 events would have an 12 intersection on the site. [ ) 13 Then-to calculate the effect on water table, we 14 used an area surrounding the repository, for which the 15' intrusion of a dike would have some sensible effect on the 16 water. table within the site and calculated the number of 17 dikes that would lie within that. area, and that gave'the 18 value for the hydrologic mode.- I'll.put that back up. 19-(Slide.) 20 MR. HINZE: Excuse me, Mike. Just for information 21-purposes, what kind of relationship did you use between size .22 and number of events? 23 MR. SHERIDic "or the Monte Carlo simulation? for the simulation. f-4 24 MR. HINZE: 25 MR. SHERIDAN: Because I was doing this on a

qf < ;) ~ ~ 4 k i l ' ;t;; 72 + h I l' simple microcomputer, I-just used the size that would give a -f i NJ 2i result in a. reasonable amount of time, within a few minutes.- 3-So, I used 5,000. 4 4 MR.-.HINZE: So, you had a constant number of 5-events. 6 MR. SHERIDAN: For each one of these calculations. 7 MR. HINZE: I see. Okay. 8 MR. SHERIDAN: So, I used 5,000 events. j 9 MR. HINZE: Is there any support for that. 10 MR. SHERIDAN: Let me say.that the probability of 11 .alsingle event.per year is 1 in a million. So, the 12 conditional probability -- f 13 MR. HINZE: And that's based on? 14 MR. SHERIDAN: That's Bruce Crowe's' estimate. 15 It's a little bit more than 1-in a million. It's about 5-16 timesL10 to'the minus 6. 17 So, for 10,000 years, then you're talking about a >] 18-1 in 100 chance of a volcano erupting in this area,'and.then 19 of that.-1 in 100 chances, it's about 1 in 100 that is going a 20-to intersect the site. '21 So, that.gives this probability, but that's'just t -22 from this base calculation. Now, other models can be 23 incorporated and can be tested. .But it's my general l24 feeling, in terms of the quality of this particular data, (s 25 that it's probably within an order of magnitude of being

'U 73 1; correct.- I don't think it will change by much more than an .j_ f. IT 5/ ! 2' order of magnitude up or down. t 3 MR.- POMEROY: Okay. Thank you very much, Bob. - Do 4 -you have something more you'd like to do,_or is this an i 5 appropriate time for a break? 7 6 MR. SHAW: It's an appropriate time for a break. 7 MR. POMEROY: Very good. -With the Chairman's 8 permission, we'll break for 10 minutes rather than 15. b 9 (Brief recess.) I-10 MR. MOELLER: Let-a reconvene, please. I 11-MR. SHAW: Just by way of updating where we are, I'll remind you that we've covered the introductory 12 1.. j ) 13 comments, we had'the overview of what the results looked 14-like,.and we had the first of our detailed presentations, '15 which was on volcanology. l ,16 From here, we'll get into a second detailed j 17 presentation on the tectonics, and then we'll return to'the- ~ 18 results to look at the sensitivities, we'll talk about how 19 this methodology can be used to develop priorities, and then i 20 I'll come back and talk a bit more about where we go from 21. here. 22 MR. POMEROY: Fine. 23 (Slide.) g-sj 24 MR. McGUIRE: Let me discuss some of the k/ 25 considerations that were made in developing the input for

5 74 1 nodes 2 and 3. This work was done by Kevin Coppersmith and k_ I 2 Bob Youngs, and neither of them could be here today, so I am 3 representing that part of the technology. 4 (Slide.) 5 MR. McGUIREt As a result, if I have to beg off on 6 some of the answers to your detailed questions, please 7 understand. m m 8 The objective of this part of the analysis was to 9 represent the magnitude, location, and likelihood of 10 earthquake occurrences in the vicinity of the Yucca Mountain 1 11 repository and the potential resulting effects on repository 12 performance. 1 i _13 The primary hazard that was examinoid was that of 14 fault rupture through the repository with the potential l ,9 15 effect of rupturing canisters and providing early release of l 'A the contents to the immediate vicinity, which would then 17 entsr the yi.cential chain of events; that is, release to the p 18 water table or reinsAs, to the surrounding environment and L 19 then transportation via hydrologic means to the carbonate 2 20 aquifer. i n -21 Also considered there were co-seismic rises in ) 22 water table due to changes in the local stress field caused 23 by earthquakes, and I'll talk about that a little bit. 7-Specifictlly not considered here was the effects 24 l i 25 of vibratory ground motions and their potential effects on i

75 I 1 the canisters. It's, I think, assumed at this point that [ 2 that is a relatively minor effect, and if there are large 3 ground motions expected, they would not have potential 4 impacts on the canisters. But that assumption obviously 5 could be revisited in a more detailed analysis. (S.ide.) \\ 6 7 MR. McGUIRE: The proposed repository boundary has 8 in it and curround it several hypothesized faults, as shown 9 here. The area of the Yucca Mountain proposed repository is 10 in the central part of the western U.S. and is generally in 11 an area extensive in tectonics, wherein the faults are 12 generally north-south, and the extension, primary direction if 13 of extension is east-west or northwest-southeast. 14 So, these faults are steeply-dipping normal 15 faults, and their rates of activity have been es.timated, for .ow Ridge fault and the 16 some of them, particularly for the 17 Paint Brush Canyon fault by Gibst., That rate of activity 18 was assumed here. In addition, rates of activity for the 13 other faults were estimated based on the displacement of the 20 Paint Prush tuft and as displaced by each of those faults 21 and inferred from relationships to the Bow Ridge fault and 22 the Paint Brush Canyon fault. 23 Those are obviously preliminary estimates fo: the 24 purposes of this demonstration or this illustration and, 25 again, could be revisited.

.3 + 76 m, 1 (slide.) 2 MR. McGUIRE: Within the Basin and Range, there 3 are some examples of historical faulting, and I show here 4 examples of surface faulting from the 1915 Pleasant Valley 5 earthquake and the 1932 Cedar Mountain earthquake. One of 6 ths features of those events is that one observes not a very ( 7 straight or distinct line of faulting at the surface but a 8' zone of faulting associated, probably, with secondary 9 faulting near the main rupture and induced by the main 10 rupture. 11 Perhaps the example on the right, the cedar 12 Mountain event, is the best example of that, in which there 13 is no single line of faulting observed but a wide kind of 14 zone or so-called " halo" of faulting that is induced by the 15 mhin event. r 16 That's one of the main features that has to be considered in terms of the possible effects of earthquakes 17 Em 18 at a repository and, as you can see a little bit later from E IV these preliminary calculations, that is a more important 2 20 feature than the primary faulting through the repository, at 21 least from the preliminary calculations. 22 (Slide.) 23 MR. McGUIRE: To represent those faults, a logic 24 tree was developed that, for each fault, represents the (( 25 possible activity or inactivity of that fault, the possible

77 1 1 geometries of that fault, the possible maximum magnitudes, 2-the possible slip rates, and the possible recurrence models 3 that might be appropriate. 4 Again, the slip rates were estimated from work of 5 Gibson, extrapolated by using the displacement of the Paint 6 Brush tuft, and the maximum magnitudes were estimated from 7 fault-length relationships. 8 All faults were assumed to dip at 60 degrees and 9 to dip for -- P,sve a depth of 15 kilometers. 10 (Slide.) 11 MR. McGUIRE: With that kind of model, one can 12 develop relationships that show the length of secondary ( } 13 faulting to the length of rupture length as a function of 14 the width of the fault zone. 15 These are developed based on observations in the 16 Basin and Range and the relationship that was used, in this 17 case bounded the data and, as a first cut, is what is us d 18 to make the calculations here. 19 (Slide.) 20 MR. McGUIRE: We also have data on the fault-zone 21 width as a function of magni ude. That data is much more 22 scattered. And we show here the rupture width as an upper 23 bound on the fault-zone width, which would be the case if 24 each of these earthquakes occurred sub-horizontally, so that O 25 the entire rupture width was considered to be the fault-zone 1

78 l 1 width at the surface. That, of course, is an extreme upper 2 bound and will not be the case for a steeply-dipping fault. 3 As a reasonable upper bound, we have taken the 4 second line, which bounds the data, and provides some sort 5 of more rehsonable estimate of what this zone of faulting -- 6 the width of this zone of faulting might be, given any 7 magnitude, up to 6 1/2 or 7. 8 (Slide.) 9 MR. McGUIRE: So, with those kinds of geometrical 10 relationships, then if we have a f ault plane of a primary '11 fault or a secondary fault, then we can hypothesize or model 12 the potential location of the rupture on that fault plane (( ) 13 and determine the probability that ruptures occurs within 14 some area or crosses some area here; that if a canister is 15 sitting there with a length of about five meters, thus would 16 intersect that canister and have an affect on that canister. 17 That sort of model can be applied, as I say, both 18 for the primary faulting and for the secondary faulting, 19 and, of course, is a function of the dip of the fault, which 20 here was taken to be 60 degrees for all faults. 21 The model that then was developed by our experts 22 in this area is illustrated conceptually here. It's the 23 Monte Carlo method that models the location of the 24 repository, models all primary faults that I showed in one IO 25 of the first slides, shows the dip of that fault with

. _ _ _ _ ~ t 79 l [ 1 respect tt the repository, and, with those geometric s-2 relationships, then hypothesizes or models all possible j 3 magnitudes and locations on that fault line, and for each of i 4 those magnitude and locations, calculates a halo, a possible l f 5 secondary faulting around that primary faultir.g, and 6 calculates the probability, then, given a number of 7 secondary faults, of an intersection of a secondary fault, i 8 and a primary fault, of course, with the repository area. 9 So generation of these kind of calculations can be 10 done in the Monte Carlo sense for all faults, and were done t 11 for all faults shown on that first map. [ 12 In addition to just the intersection of a possible l ) 13 rupture of a primary or a secondary fault with the g\\_/ 14 repository horizon, we need to estimate the potential 15 displacement that would occur along that rupture, and that t 16 estimate we can constrain with historical data; here shown 17 with some empirical data of average displacement on a fault 18 versus magnitude for all slip types. These are data taken 19 from Wells and others in preparation. 20 So with that kind of model being developed, then 21 we apply that to the existing faults or known faults in the 22 region of the Yucca Mountain repository, or proposed 23 repository, and can calculate the probability of an 24 intersection of a primary or secondary rupture with the s P 25 repository and a distribution of the potential displacement

80 l 1 along any rupture. p, ( \\ 2 (Slide.) 3 MR. McGUIRE: This summarizes that calculation 4 for the parameters that were chosen in terms of a histogram 5 showing annual frequency of canister failure versus 6 probability, and canister failure was defined for purposes 7 of this illustration as the occurrence of a primary or 8 secondary fault rupture through the repository horizon that 9 has a displacement greater than one centimeter. 10 So these are annual frequencies per year that t 11 account for the frequency of occurrence of earthquakes on 12 each of those faults, and, through the Monte Carlo [%d') 13 simulation, show the distribution of fault dispiscement from 14 primary faulting through the site, from secondary faulting, 15 and from the total of both. 16 As I was intimating earlier, it is the secondary 17 faulting, potential secondary faulting through the 18 repository horizon that really governs the ca'culation of 19 possible rupture through canisters. 20 (Slide.) 21 MR. McGUIRE: We can additionally use those 22 models of fault occurrence on the primary faults, of rupture 23 occurrence on the primary faults, and some inputs from some f-~ 24 of the rock mechanics expertise we had on the project to get Q 25 an estimate of what the coseismic water table rise might be l

81 1 given that each of those events has associated with it a 2 certain stress drop. 3 The results of that integration are shown here, 4 with a range of water table changes from zero to 100 meters 5 in the extreme, and with probabilities of ten to the minus 6 one to ten to the minus five. 7 We show an average result and also a range of 8 results which result from a change or variation in some of 9 the potential parameters of the rock properties that might 10 be assumed in the model. What we've used for the purpose of 11 these calculations to see if there's an important effect of 12 coseismic water tables on our predicted concentrations of 13 radionuclides is to use the mean curve. 14 MR. HINZE Is that assuming that the gradient to 15 the north breaks down and that we have flooding in the rise 16 of the water table associatef with the breaking down of that 17 gradient? Is that what we're looking at here? 18-MR. McGUIRE: No. This is just the possible 19 changes in the water table caused by changes in the strain 20 conditions following an earthquake; so release in stress 21 causing changes in strain. 22 MR. HINZE So it's pervasive, not specific to the 23 water fall? 24 MR. McGUIRE: That's correct, yes. 25 MR. MOELLER: And to refresh me, how far is the

82 ~ repository horizon above the existing water table? 1 () i \\_/ 2 MR. McGUIRE: It's 150 meters in the zeolitic i 3 portion, and 250 2eters in the vitric portion. 4 MR. MOELLER: So you're 100 meters here is 5 approaching it? 6 MR. McGUIRE: Yes. That's correct. I 7 MR. MOELLER: Thank you. 8 MR. SHAW: I might just add to that point that as 9 you look at the logic diagram that we have, there are a 10 number of different events that can cause water table 11 changes, and we have added them where appropriate, t 12 MR. MOELLER: Thank you. (/'N 13 MR. HINZE: Excuse me. They are added in this 14 diagram, then? This is just the one? 15 MR. McGUIRE: Yes. 16 MR. HINZE: All right. 17 (Slide.) I 18 MR. McGUIRE: So as a result, this represents an 19 enlargement of those two nodes of the logic tree with their 20 values shown. Note, too, we have two branches, and for 21 purposes of determining if earthquake-caused ruptures have a 22 large affect on our estimates on our estimates of 23 radionuclide release, we've taken two branches, one with the 24 rate of an earthquake taken into account, and a second with 25 the rate of earthquakes assumed to be zero just so we can i y w a w e ---ee e + n n--+-.., -m.-- ww s-+ ,,a ms w .w-r u, 4r's v f

83 1 look later -- and I'll illustrate -- at what the effect is ( 2 of including or excluding earthquakes in the analysis. 3 That rate of occurrence is.01 per thousand years, 4 and that's been normalized, as all these calculations will 5 be normalized, by 500 canisters, which represents 1,000 6 metric tons of initial heavy metal. The reason we do that 7 is because that's the normalization proposed by the EPA 8 standard. So all of these calculations have been normalized 9 to that set of numbers. 10 We, for purposes of illustration, assigned 11 probabilities of 50 percent to those alternatives, and then 12 for this one used three branches on the water table node, [ } 13 representing zero, 30 or 50 meters rise, with a probability 14 of 99 percent, a probability of.009 or a probability of 15 .001. So we've discretized that distribution that I showed 16 earlier on water table into those three values. 17 Again, as Bob Shaw was indicating, if we have the 18 80 meter rise caused by the volcanic disturbance in 19 conjunction with any of these three we would add them to get 20 the total change in water table. 21 That gives some background as to some of the 22 modelling and considerations that were made in the tectonics 23 part of the model. 24 (Slide.) O 25 MR. McGUIRE: What I will do now is go into some

84 1 of the sensitivity studies that we have performed and some l 2 of the insights that can be gained from reprecenting the 3 performance of the system, performance of a repository i 4 system with these kinds of models and show you again in an l 5 illustrative sense the kind of insights one gets. 6 (Pause.] I 7 MR. McGUIRE: What I'll illustrate first are the i 8 sensitivities of the concentrations versus time, color coded 9 by certain of the parameters in the logic tree. 10 I'll call those out as I go. 11 The first one is the sensitivity to the flux 12 level. I've put that up. It's the one I've showed earlier l 13 -- just to get your eye calibrated to one of the parameters l i 14 that is most influential in the problem, at least at this l 15 illustrative stage. 16 Again, the dashed blue line represents the highest 17' value of 4 millimeters per year. The red lines represent 18 the value of 1.6 millimeters per year. These blue lines are 19 volcanic releases which are not considered in the subsequent 20 illustrations and the green lines would be down off-scale at 21 the bottom for the case of.5 millimeters per year flux. 22 (Slide.) 23 MR. McGUIRE: So that represents a large effect on 24 a potential water table and let me just bring that one up g k 25 again.

85 1 (Slide.*j ,() (N/ 2 MR. McGUIRE: We need to look at, etncentrate on '3 two things. One is how segregated are the curven with 4-respect to that color coding -- in this case they are quite 5 well segregated. That is, all of the high curves represent 6 the high fluxes end these low red curves represent the 7 intermediate fluxes and low fluxes. 8 The second thing that is important is what drives 9 the curves up to these high levels around 100 curies hera 10-because if that is a standard which we should be aware of we i 11 need to be particularly sensitive to the cases and 12 combinations of parameters that lead to those high releases. t (' } 13 What I'll do for the rest of the set of 14 illustrations is just use a subset which represents only 15 that set of and branches correspanding to the fluxes of 4 16 ' millimeters per year. That would represent just these '17 curves that are shown here on the dashed blue line. 18 The reason is that we first are interested -- 19 those are the only cases for this example which lead to 20 large releases and also I would like to avoid the confusion 21 that comes with putting curves frem different fluxes on the 2 same plot. 2 23 [ Slide.) 24 MR. McGUIRE: What I will show next is the ( 25 sensitivity to Node 3, which is that potential earthquake-

86' 1 caused canister failure. Ik 2 In this case we see blue and green curves 3 representing the two cases of earthquakes included or 4 excluded from the analysis so that the green curve would 5 represent the higher value or the lower branch which is 6 earthquakes included and the blue curve representing 7 earthquakeo excluded from the analysis. 8 I think if you look closely you'll see some of the 9 curves are blue and then they turn green, meaning that there 10 are blue and green curves there and one overlays the other. 11 In that case it is showing that the calculation 12 whether we include or exclude earthquakes has absolutely no 13 difference on our calculated radionuclide releases. There 14 are cases in which there are differences, some small 15 differences between earthquakes included or excluded. 16 So we will conclude from that in this illustration 17 that that particular node is not a very influential one in 18 terms of determining or causing high releases or not causing 19 'high releases of radionuclides. 20 (Slide.) 21 MR. McGUIRE: Let's go to parameter 5, which is 22 the vel m nic representatio'.1 and I think while I was turned 23 around the colors changed on me. 24 Again, this is a case where we see the colors 25 changing, meaning that the calculations are not governed by i

i i i 87 1 that node and remember these are only the hydrologic I (rv (,) 2 releases. I have not shown on here the hydromagnatic 3 eruption nor the case of the magma entraining canisters and [ 4 bringing it to the surface. I 5 All we have here are the two cases of no volcanoes 6 or a volcano causing a change in the water table. From our 7 calculations in this application of the model that would i 3 indicate that there is no difference between those two, j 9. meaning that the potential change of water table caused by a 10 volcano is really -- has littin' significance or little j 11 effect on potential releases of radionuclides. 12 (Slide.) ( 13 MR. McGUIRE: Parameter 7 represents bore hole 14 stability and here we adopted a very conservative model and 15 again had a node which represents including it or excluding 16 it from the analysis. l 17 The conservative model said that there might be 18 bore holes falling during the heat pulse that is associated 19 with emplacement of canisters and we said with one percent 20 probability that that spalling would cause rupture of two of 21 the canister designs. 22 We proceeded to look at that as a possible effect. 23 It is clear that in this case those curves are intermixes 24 and that potentially has very little effect on the 25 calculations. _... ~.. _ _ _ _..., _ _ _,.,

88 1 Of course that is a very conservative criterion. 2 If you have spalling you probably would not rupture all of 3 the canisters. There might be some effect there that might 4 perhaps need to be studied in more detail. 5 It does have a relationship with the canister 6 design that I'll come back to in a minute when I get to the 7 canister node because it illustrates one of the reasons why 8 one needs to look at combinations of parameters. One can't 9 look at things in the abstract or segregated alone. One 10 needs to look at the combinations of effects to see what 11 their combined effect is on the calculations. 12 (Slide.) 13 MR. McGUIRE: Let's look at the effect of the q 14 total water table change on calculated levels of release. 15 In this case, we have bifurcated or used two values of water 16 table change based on the sum of all possible changes. One 17 is that there is zero water table change, and the other is 18 that the water table increases by 150 meters. { 19 So effectively, there are saturated conditions for 20 one half of the repository. And so the hydrologic flow path 21 is through a saturated medium. That is represented by the 22 green curves. And the lower curves represent a case when 23 there is no water table change, and the entire hydrological 24 path is through an unsaturated medium. 25 For our set of assumptions and calculations, there

l l 89 1 is little difference between those two. That is the 2 possible saturation of the tuff would not have a major 3 effect on causing larger releases. It does have some 4 effect. And I think you can see here some of these curves 5 increase as a function of the water table, but not a major 6 effect. 7 (Slide.) 8 MR. McGUIRE: Let's look at the effect of canister 9 design. 10 We had three possible canister designs. All of 11 them have a probabilistic potential for failure. And we 12 used a Weibull distribution to represent that. So the three 13 designs correspond to threv sets of values for the Weibull 14 distribution. 15 The easiest way to represent that is just with the 16 mean canister lifetime. And the worst-case canister design 17 would be a canister design with a mean lifetime of 1,000 18 years. That is represented with the green curves. We call 19 that a poor design. 20 The moderate canister design has a mean lifetime 21 of 5,000 years. And that is represented by the red curves 22 here. And finally, the good canister design or excellent 23 canister design is represented i'y the dashed blue curves 24 over in the corner. And those curves have a mean lifetime, O 25 represent a mean lifetime of 50,000 years.

o 90 1" The difference between the first and second set of 2 curves, or the green and red curves and the dashed blue '3 curves is that we assume for the excellent canister design, 4 that neither the volcanoes nor the earthquakes nor the 5-borehole stability would cause failure of that canister. 6 So there is no possible early release from what I i 7 have called unnatural causes. And I got some negative 1 8 feedback from that in the sense that who says that a 9 canister failing by itself by corrosion is natural. I just ] 1 10 was making a human analogy, I guess, to that. If we die .11 from internal causes, that is called a natural death. 12 So these other curves represent the inclusion of 7 13 canister failures caused by earthquakes, volcanoes, and g i 14 borehole instability. 1 L i 15 So that is a case, in particular with the borehole i 1 16 instability, in which there is a combination between the 17 assumptions made there and the canister design that would, 18 aven for the moderate canister design represented by the red curves, would lead to high early releases that one would not 19 l" 20 get if you did not include that scenario of boreholes i il failing early during the heat pulse and leading to canister 22 failure. It is an area in which further development work 1 23 would be appropriate, and further modeling work. 24 (Slide.) 25 MR. McGUIRE: The next perameter is solubility of l a e.nn ,m ~ e

91 1 the waste. And let me explain the slide first, and then ,_s ' 's-2 come back with some more assumptions that we have made. 3 We used two values of waste solubility, 1 times 10 4 to the minus 5 moles per liter represented by the green 5 curves, and 2 times 10 to the minus 7 moves per liter 6 represented by the blue curves. And that has a major effect 7 on our calculated levels of release. 8 That leads me into the model that was used. It is 9 a solution-limited dissolution model that represents the 10 solution of radionuclides in and near the source or in and i 11 near the canisters as a function of solubility of the waste 12 material. [ f 13 That is put into a hydrological model that solves i 14 the diffusion equation analytically and as a result, l 15 requires, for simplicity, for this application, we used an 16 analytical solution that requires as a result that one has a 17 constant flow of water past the canisters. 18 So we do not include in any of these calculations 19 the possibility that the hydrological flux stays at low 20 level for one or two thousand years and then increases for 21. the remainder of the repository lifetime. 22 So, for example, that flux of 4 millimeters a year 23 assumes.that the flux or currently pluvial conditions start 24 tomorrow, and run past the canisters for their entire fq '~) I 25 lifetimes. i ~ ~.

we 92 1 That obviously is a simplifying solution that ' l ,) 2 allows us to do these calculations on a PC. One could l 3 generalize that and use the more sophisticated models in a l 4 broader application. I 5 For that hydrological model, we did use multiple l 6 pathways that represent pathways of water flow through the 7 vitrified paintbrush tuff and through the zeolitized 8 paintbrush tuf f, and also included matrix and fracture flow 9 through both of those units. 10 So this is, the solubility, then, of the waste 11 material, is an important factor in estimating the levels of 12 tadioactive release. ( 13 (Slide.) 14 MR. McGUIRE: Another important factor is the 15 representation of fractures in the media. We hed a node i 16 that represented possible fracturing of the welded tuft and i 17 had two values, depending on which of those zones, I won't 18 quote the numbers, because they depend on which of those L 19 zones one is considering, in terms of the flow. l 20 But we had high -- for -- we had high levels or l. L' 21-high areas-of large fractures and lower levels of large 22 fractures, as the two alternatives with respect to that 23 node. And the dash green curves here represent the high r. 24 areas of fracturing; and the dash blue curves, the low areas 25 of large fracturing. Those are also a function of the flux e -r we-'3w w --w w w w

1 1 93 1 level. And, again, those represent an important factor in 2 leading to large radioactive releases from the repository. 3 We're building a case, as you can see, of 4 parameters that are necessary but not sufficient in 5 themselves, to lead to largs levels of release. And I'll 6 come back and review those at the end. 7 MR. MOELLER: Excuse me. 8 MR. McGUIRE: Yes. 9 MR. MOELLER: You use the words high areas of 10 fracturing and low areas? 11 MR. McGUIRE: Yes. 12 MR. MOELLER: Are you talking about elevation or ( 13 numbers? 14 MR. McGUIRE: Numbers. 15 MR. HOELLER: Okay, you mean lots of areas 16 fractured? 17 MR. McGUIRE: Yes, lots of areas fractured. 18 MR. MOELLER: Okay. 19 MR. McGUIRE: And those numbers are typically a 20 half percent of two percent, for example. Those might be 21 two extremes. 22 MR. MOELLER: Okay. 23 MR. McGUIRE: Thank you for that clarification. (Slide.) - 'k ) 24 25 MR. McGUIRE: The porosity of the medium and its

e '? 1 94 I l 1 effect is shown here. And we see several distinct sets of (p-s\\ -) I 2 curves. The low porosity, shown in the green, and the high 3 porosity, as shown in -- I'm sorry -- the low porosity, as 4 shown in the red and the high porosity, as shown in the 5 green -- both of thoss -- those -- that has a major effect l 6 on -- on some of the releases, but both of those can lead to 7 large releasest so we find that that's not, say -- one of 8 those values is not, in itself, indicative of whether or not 9 one has large levels of release. One can get those for l 1 10 lower or high porosities in the medium. r 11 MR. HINZEt Does your model include the 12 possibility of interrelating porosity with the flux? ( ) 13 MR. McGUIRE: Yes, certainly. Yes. I 14 [ Slide.) 15 MR. McGUIRE: And finally, the last illustration l 16 I'll make on this set of representations, is with respect to i i 17 retardation. And that has an important influence, and you 18 can see the green cases representing low retardation, and 19 the red representing high retardation. So that, again, is 20 particular for this level of 100 curies, is a low 21 retardation would be a factor, or a condition that is 22 necessary but not sufficient to lead to large releases. And 23 clearly, one sees the effect of higher retardation -- it's r 24 on tha curves. (q 25 So, that concludes that set of presentations. We

l 95 1 would find then that large releases from these illustrations 2 are caused by a combination of high fluxes, high 3 solubilities, large areas of fracturing in the rock and low 4 retardations. 5 That's one way to look at the sensitivity of the 6 results to some of the inputs. Another way is to look at th 7 sensitivity of that CCDF to changes in some of the 8 assumptions. We have the ability to, with the software, to 9 either change probabilities in the input list and create a 10 new CCDF for those change probabilities, or we can condition 11 on one or more sets of branches, and perform or calculate a 12 CCDF, conditional on that set of branches -- on the end 13 branches associated with that branch. 14 (slide.) 15 MR. McGUIRE: Let me illustrate that first with 16 the fluxes. I've set up some input files here so that wo 17 don't have to just wait for the computer to pull up and 18 generate the plot files. So, I'll pick up the appropriate 19 input files for this -- this demonstration. 20 (Slide.] .21 MR. McGUIRE: I will also plot the base case that 22 I showed you earlier, which is this one here -- the dash 23 blue curve. And what I'm plotting are the CCDFs, 24 conditional on levels of flux. O 25 The red curve represents the CCDF, if a flux of

96 1 tour millimeters per year is the " correct value." In other (\\-) 2 words, if we give that probability of unity an give the 3 other altet..atives probabilities of zero. 4 If the flux of 1.6 millimeters a year is given a 5 probability of unity, then we get the green curve. And if 6 we give the flux value of 0.5 millimeters per year a 7 probability of unity, then we have a curve that's -- that's 8 so low, it's not shown here. 9 So, we would see that, again, that flux is an 10 important parameter in the calculations. We would also 11 infer from this that if the stair-step function is the 12 ultimate test by which 'de have to evaluate a repository, (( ) 13 then, even if we were at the worst case, in terms of the 14 flux, we would not exceed that bound -- we would not go 15 beyond it. And even in that worse case, if that were the 16 worse case interpretation, the repository would be judged 17 acceptable under that criteria. 18 MR. POMEROY: Robin, just for clarification, this 19 is fof' one radionuclide? 20 MR. McGUIRE: This is for one radionuclide. Thank 21 you, yes. 22 MR. HINZE: Robin, what is the cause of the high 23 wave number variations on the curve? Do those have any (s 24 significance? Those minor irregularities? D 25 MR. McGUIRE: These, here?

97 i 1 MR. HINZEt Yes. 2 MR. McGUIRE: No, they don't. No. That's purely 3 a function of the way we plotted the curves and picked 4 points. I think we -- we just plotted 35 points on the 5 graph. 6 MR. HINZEt Okay, it's the interpolation function? 7 MR. McGUIREt Yes, that's right. 8 MR. HINZE Okay. Thank you. 9 MR. McGUIRE: Yes. 10 (Slide.) 11 MR. McGUIRE: Let me show you a similar comparison 12 for the three canister designs. In this case, the blue f ) 13 curve represents the base case, the green and red curves 14 represent the poor and moderate canister designs and this 15 dashed blue curve which is over here represents that, quote, 16 " excellent" canister design. 17 In this case, as we saw earlier, because of the 18 potential effect of earthquake induced ruptures and other 19 effects, bore hole instability, for example, the first two 20 canister designs, the green and red are very close to the 21 base case design -- excuse me, the first two designs have a 22 CCDF that are very close to the bhse case CCDP. 23 If in this case we went to the excellent canister 24 design, we would achieve a very high degree of additional 25 safety or a reduction in that CCDF. Again, that is a

I 98 1 function of the way we've treated those two canister designs f 2 and, of course, it deserves review. 3 MR. POMEROY: What's the base case number here? I 4 4 have notes that say that the poor design was the 1,000 year 5 canister lifetime. That was the poorest number that you 6 considered? 7 MR. McGUIRE: Yes. What we call the poor canister i 8 design had a zero failures for 300 years, a mean canister f f 9 lifetime of 1,000 years and a Weibull exponent of one. The i 10 moderate canister design had zero failures for 1,000 years, l l 11 a mean lifetiae of 5,000 years and a Weibull parameter of 12 two. The excellent design had zero failures for 1,000 I yT 13 years, a mean lifetime of 50,000 years and a Weibull Y 14 parameter of five, and the base case is the mix of those. 15 Those were given, respectively, the probabilities 16 of.25, .5 and 25, so that represents the base case, the 17 entire combination of that set of assumptions. L 18 MR. MOELLER: Once again, the reason that there's l 19 not that much difference in the different canisters is that 1 20 you're assuming an earthquake destroys the integrity of a t 21 certain number of canisters? 22 MR. McGUIRE: The.t's right, yes. In particular, i 23 .the reason that these are very similar, the first two 24. canister designs are very similar, is that they're affected ( 25 similarly, Jn particular by the heat pulse and the potential l 1

i 99 I failure of the bore hole causing failure of the canister. V"'T \\s / 2 Under the set of assumptions we've used, that would be an 3 equal set of failures for either the poor or moderate 4 canister design. I 5 As a result, you get very similar CCDFs for those 6 two cases. I'm not sure that ussumption is realistic. It's 7 just used for illustration here. .i 8 (Slide.) 9 MR. McGUIRE Let me go on to one final set of 10 presentations that one could make. To introduce that, let. 11 me consider two alternative areas in which we might do 12 research at a repository. One might be in evaluating the l (( ) 13 possible levels of hydrologic flux that can occur in the 14 future. 15 We might have a set of probabilities as we've 16 assumed here for our base case which gives 8 percent i 17 credibility to.5 millimeters a year, 90 percent to 1.6 18 millimeters per year and 2 percent to 4 millimeters per l 19 year. We could have a hydrologist and/or a climatologist 20 propose some research of a certain type and at a certain i 21 level that would take place over a period of five years, 22 let's say. Thcy might estimate that if they were allowed to 23 do that research and expend those resources, they could 24 effectively eliminate one of these end members. es U 25 In other words, at the end of that time, they

i 100 I would be able to say, well, 4 millimeters a years is an ["s ') I 2 incredible value and I will give it zero, or 1.5 millimeters t is an incredible value and I would give it zero, but I would 3 4 still be pretty uncertain about the other remaining two. 5 That would be the outcome of expending those resources on j 6 that research. 7 So, with equal likelihood, I might get this set of i 8 numbers, this set of probabilities for those fluxes, or a 9 second set of probabilities for those fluxes, eliminating 10 the end member in those two, and I would like to do that 11 research. 12 Alternatively, we could have a rock mechanics i i ( 13 person and a hydrologist, perhaps, say, well, I would like s_ l 14 to do some research on the areas of large fractures and how i l 15 the fracturing occurs in the welded tuff and what its 16 distribution is in space. I could use those resources and 17 expend that time and money to do that research, and I think { t 18 that instead of 50/50 probabilities on low and high areas of 19 large fractures as currently defined, I could refine that to 20 an 80/20 probability or an 20/80 probability, again, with 21 equal likelihood. 22 Again, I could expend those resources and improve 23 this probability estimate from a coin toss to a preferred r3 24 value by a factor of four. of course, I would still have . li l 25 some residual uncertainty about all of that. That's an 1

l 101 1 alternate area in'which we could expend resources to do i -,s c'-) 2 research. t 3 Well, with the model that we've developed, it's 4 very easy to put these alternative probabilities in and 5 compute the resulting CCDF and determine what we would gain 6 from each of those sets of resources, so let me illustrate l l 7 that. ~ 8 (Slide.) 9 MR. McGUIRE: Again, I will show the base case, 10 and first, I'll show the effects of potential research in 11 the area of flux. 12 The blue represents the base case and the green [ 13 and red represent the alternative resulta that come out of 14 that research on hydrologic flux, and we might have 15 suspected, that would improve our understanding and it would 16 change our CCDF. In this case, if we eliminate the low 17 value of flux, we would get higher CCDF. In this case, if 18 we eliminated the high value of flux, we would get a lower 19 CCDF. And we would draritically change our interptotation 20 of what the CCDF is and this relationship of the CCDF to a 21 proposed regulatory limit. 22 Fix in your mind that relationship b etween those 23 curves for a moment, and let me pull up the equivalent 24 results for the alternative set of research, which is the (i 25 research on areas of large fractures.

- _ _ __~ i g, 102 1 (Slide.] j. 4 f f 2'~ MR. McGUIRE: Here, we don't see much change in -- 3 potential change in that'CCDF.. We have the base case and L 4 the two alternatives representing, perhaps, the total change 5 of a fatetor of 4 or so in the CCDF in terms of probability 6-at -any } evel of release.- 7 So, we "ould conclude that doing that 7 search in 8 the area of flux would have a much higher benefit in terms 9 of refining our CCDF and gaining vesults and gaining 10 understandi.g with respect to the site than would that 11 illustration of the research in the area of fractures, and 12 that's purely an illustration just drawn out of the air in 13 terms of the probability values but one designed to show 1 L how, again, these kinds of results can be'used to make -- 14 15 put priorities on research and show what they might lead to-j 16 in terms of future interpretations.: 4 l 17 MR..HINZE: How did the likelihood come into that, 18 Robin? You've got'a 50-percent likelihood. 19 MR. McGUIRE: It doesn't come into that in these -i 20-curves. The only reason'that's put'up there is that one, in 21. doing these kinds of predictions of what w;uld come out of-22 this kind of result, needs to be careful to make sure that 23 the mean-of this combined distribution is the same as the fL 24 current mean. If it isn't -- in other words -- 25 MR. HINZE: Do that again, please.

(: 103 1~ MR.-McGUIRE: In other words, what you're 21 predicting about the future and what you could reeolve in 3 terms of future research has to have the ca e caan ve).ue as 4 your current uncertainty. That's what yc.ur current 5_ uncertainty is supposed to represent, is if you and perfect knowledge, that's where my possible values would come out, 7 with their associated probabilities. So, the mean of that 8 is where I think the average of that distribution is. 9 If I went and had a research program to absolutely 10 resolve all uncertainties in that parameter, then, by 11. definition, you would have the same parameter shape and 112 values as my current uncertainty, because that's what my j l 13 current uncertainty is supposed to represent, is my 14 uncertainty in the value of that parameter. 15 MR. HINZE: It shifted. 16 MR. McGUIRE: Yes. So, if I were to proposed 17 research.to resolve that parameter, the distribution of that 18 research must have the same mean as my current uncertainty, 19 .because that's what my current uncertainty is supposed to 20 represent. 21 In this case, we're not resolving all of the

22 uncertainty in that parameter; we're only resolving some of 23 it.

But we need to be careful that -- to be consistent that 24 that mean of those -- that projected distribution has the (9 25 same mean as our -- has the same value as our current mean.

y vi o' w i j 104 1; And that's thelonly reason I put these probabilities'up_ 2 here. 3 MR. HINZE: Deja vu. 4 MR. POMEROY: All over again. 5-(Slide.) 6 MR. McGUIRE: One final note here: We can also do 7' these calculations for different nuclides separately, plot 8 them on the same scale. The blue curve shows the curve for 9 Neptunium 237',- the green curve for Cesium 135, and the red 10 -. curve for Tecnesium 99, and compare the levels of release of 11 each of those, normalized, of course, by the appropriate ~ 12 standard with respect to the proposed EPA standards. In

(

13_ particular, Cesium 135 has a proposed standard of 1,000 14-curies, not 100 curies. So, one would normalize by 1,000, 15 and'that's why_this axis is in-terms of normalized release. 16 -(Slide.] 17 MR. McGUIRE: We can.also, of course, total those 18 curves to get that total le. vel of release. In this case, 19 it's dominated by Neptunium 237 for that set of nuclides. -20 But one, in theory, can do that for all nuclides and perform 21' .a calculation of the total releases and than compare that, 22 as was pointed out, to a proposed standard. 23 So, that concludes my presentation of the 24 sensitivities. 25 MR. HINZE: If I could, one question: Do you 1

'f G, iii jid' g 'y V 105 I I .. A Lli think-that it would be. helpful.-to use to hava-a proprint'of y l2 the article that you and Walt wrote as some kind of 3 background on this for the committee? Would that be-E '4 helpful? ~ -5' MR. KcGUIRE: That could certainly be done. 6 MR. HINZE: I'really would suggest that that be made available. I think that that article is an excellent 8 one that you have written, and it would provide an excellent w 9 background for all of us, if that could be made available. 10 MR. McGUIRE: Okay. 11 MR. HINZE: The book is not out yet. 12 MR. McGUIRE: The-book is out. ( -13 MR. HINZE: The-book is out? 14 MR. McGUIRE: Yes. I have a copy of it. 15 MR. HINZE: Okay. 16 MR. McGUIRE: If somebody gives me an address, 17 I'll be happy to send a copy of that paper. 18-MR. POMEROY: Robin, also, is it possible -- I 19. realize that all these are computer-generated curves, but is '20 it possible to obtain more copies of what you'have given us 21 than what we currently have, or is that not possible? ~22 MR. SHAW: The reason we did what we did is that, 23 in a sense, we're still working on some of these results 24 and, also, that we feel that the numerical results are not 25 really significant, significant in the sense that we use

106 1 them for relative, rather than. absolutes. 2. The report that EPRI is publishing on this work 3 will include some of these curves. I expect that it will be 4 iwsued by the end of October. So, I think, within that .5-time-span,'you will have those results available to you, and 6 I prefer to do it on that basis. 7 MR. POMEROY: That's fine. 8 MR. SHAW: I think at this str.ge in going through 9 what we have done, we have responded to ost if not all of 10 the questions that you posed at the beginning ' the 11 session. What we have presented is what-we feel is a 12 simplified model that'is illustrative. ,( l -13 We feel, nonetheless, it is reasonable in the-. 14 sense of having the assential parameters that are important 15 for the evaluation of the site, that it takes vales of those 16 parameters that are reasonably close-to what we feel is 17 reality. 18 We certainly, I will reiterate, used one 19 individual for each of the technical areas and asked him to 20 represent as best he could that area. Robin noted and I 21 will emphasize that the-parameters are time independent in 22 our model. 23 We would not expect that the climate conditions 24 would suddenly change and be the same for all time. q 25 Nonetheless, that is what we assumed. A suitable revision

w. f 107-1- of the'nodel would be to update that but we did want it to , 7-q i]J-2 be on~a PC'so that it would be transportable for i .3' presentations of-this nature and we wanted something that we -] 4 could do within a reasonable time span, i 5 -So those were some of the assumptions that we'made q 6 and the conditions that we made on it in order to do it in -1 7 this time span that we made available to us. It uses 1 8 certainly background material from Yucca Mountain. In some 9 cases, it builds on that kind of material and in other 1 10 cases, it generates new perspectives such as Mike's 11 -perspective on volcanology. 1 1 -12 Certainly our models have deterministic aspects to -( ) 13 them but it is all done in a probabilistic setting as you 14. 'can well see. We now view this as a tool, as an important 15 tool that'can be used and expanded. 16 (SLIDE.] E '17 MR. SHAW: Let me start'now and we return now to 18 some of the material that you have in front of you and let 19 me make a few observations. Keeping in mind that these 20. results are illustrative, we have found that the following 21 are more influential in contrast to.less influential onsite 22 performance. 23 Hydrology, in particular, infiltration or recharge-24 from precipitation, water flow pathways which are influenced .((g v 25 by the extent of rock fracture and porosity and a

1 .108-i = b 1. 'significant rise in water table and then the second general O' l2. . class of geochemistry where uranium solubility'as influenced ? 3 by. dissolution chemistry and temperature and chemical 4 retardation of released radioisotopes. 5 (SLIDE.) l 6 MR. SHAW: Some conclusions that we would make in F 7 addition to.those observations are that the use of a multi-8 disciplinary scientific and engineering expertise to conduct 9 fa risk-based evaluation of a high level waste repository is 10 achievable with current. knowledge and technology. H11 A structured approach is required. The workshop 12 format worked very well for us and we find it suited to this _'( ' 13 approach. The use of logic trees is a convenient and we 14 feel a credible format in which to pose this methodology. ~15 The results of the methodology should be obtained-16 during the process of model development, that is, the 17 process should be iterative. I think' Robin and all of us 18 would say that if we had another meeting.of this group, we .19 would probably have a slightly different logic tree with 20 slightly different values. It is a continuing process and 21 the more we learn, the more we modify and change it. 22-But that emphasizes the need to do it in this 23 fashion, to have an overall methodology so that.you are . r~ 24 looking at the broad perspective of everything so that as 'b 25 you come in with new inputs, you say, "Oh, how does that I

u 109 1-affect this?" and so on and so forth. You determine the O 2 sensitivities and makes those changes as you move along. 3l A methodology of this type certainly can be 4 applied on a larger scale in which a larger body of 5 expertise participates. This application will lead, meaning 6 the application of a larger body of expertise, leads to a 7-realistic rather than a simple demonstrative results. 8 (SLIDE.] 9 MR. SHAW: That leads me to say, where do we go 10 from here? A quick summary and I will come back to each'of 11 these in a little more detail, a quick summary here is, 12 first,-our effort was to prepare a working version of the ) 13 Methodology Development Team performance assessment model 14 and issue a report. The first is done and it was presented 15 to you here-this morning. The'second, the report has been = 116 completed and it is in the publication process at EPRI and I 17 expect.that it will be ready for circulation by the end of 18 October. 19 Phase Two. I look at phase two as joining with R20 DOE in sponsorship of workshops on performance assessment 21 methodologies to identify crucial technical topics for 22 workshops and I will come back to that in a moment. 23 I see phase three that EPRI would support DOS in 24 conducting expert workshops on crucial technical topics 25 which could readily be identified in phase two.

r 110 1-(SLIDE. ) n .b 21 MR. SHAW: Returning to phase two, we propose a T3-series.of workshops'on the performance assessment E 4 methodologies with four key participants: DOE Yucca Mountain 5 project office contractors who have been for a while in the 6 process of developing performance assessment methodologies; 7, second, the DOE ~ headquarter contractor who is Golder 8 Associations who is putting together a similar or not 9 necessarily similar but also putting together'a performance L 10 assessment methodology; NRC who has also developed this 11 methodology and the EPRI/UWASTE Methodology Development Team-l 12 .that you have heard here. 13 We would certainly include the State of Nevada as 14 a participant or observer in these but we have seen no l; 15 evidence that they have a methodology as such, rather '16 scenarios that they are concerned about. 17 The objectives of this'is-that I would see that we 18-would change detailed explanations of each of these 19 methodologies between the four.Various parties. It would be- 'l 20, appropriate then for each of these parties to revise.the 21 methodologies where it is appropriate and to obtain 22 consensus then on highest priority technical areas. ~23 To be a little more descriptive, I see this as a ~24 series of maybe three workshops where the first one might . Q 25 take a few days and each party could tako on the order of a

n 111-i '1 : ' half la. day to.-explain.as we have done today the details of N-- 2-

the methodology, what are the assumptions, what are the

-i 1 3 limitations, what are the emphases and so on-and so forth-4 strictly from the point of the view of the other parties' ] 5 understanding what it was that we did, what it is that-NRC 6 has done and so on so that each of us understands the basis 7 and the~ assumptions that have gone into the various 8 methodologies. 9 Then we might sit there and say, " Whoops, we 10 forgot about that issue" or " Hey, that is an even more -11 interesting way to do this" or "That that is an easy way to 12 include human intrusion" or so on and so forth. There are a (l 13 variety of things that might come out of this discussion 14 that would lead to revision of the methodologies. 15-Then the second and very important aspect of this -- 16 is that we would ask each of these then based on this to go-17 back and rivise if you feel it is appropriate and then using 18. that revised methodology come back and tell us what do you 19 think are the key technical areas we need to emphasize 11n 20 ' order to improve our performance assessment estimates. 21 I just gave our preliminary assessment to you H22 which talked about hydrology and geochemistry as two of the 23 key areas that we see, but to have each of these parties .f-24 .come back then in a second workshop and say, "Here is where j (E 25 we think the emphasis ought to be; one, two, three, and so I

1 1 112 1 on" and then try to drive that towards a consensus-to maybe, i O (Q 2-you say, "Here is a set that we feel are the most important 3 issues-that.need to_be dealt with first and here'is a '4-secondary set." 5 It may require a third meeting to come back and ~6-revisit that based on the second one to come to some 7 consensus amongst these four parties as to what are the key J 8 ingredients or the workshops. 1 9 We have offered and are.willing to sponsor these -l '10 workshops and as a matter of fact, we have scheduled the 11 first one for the fourth through the sixth of December with-12 as far as.I can determine right now all four of these ( 13 parties in concurrence and in agreement that they will 14 participate. q 1 .15 [ SLIDE.] q .16 MR. SHAW: Phase three then we see as a series of 17 -workshops on these highest _-priority technical areas. '18-identified in phase two. Those of you who are familiar with -19 the seismicity owners group know the patters that was used 20 there, to take on one technology.at a time but build that in 21 an appropriate fashion. 122 We would see that happening here. One could, for 23 example, have a series of workshops on hydrology so that the 24 hydrologist experts not only from within the project area .{f 25 but others from without would collect together in a series 1 l ,.. ~. -.a

yx {y t 113 S,_ 1 of meetings to say, "What are the objectives? What are the (' )i 2' key-ingredients? What are the data? What are:the models?" ,J ' 23 2 at' cetera. 4: You would go through this-whole pattern of trying. 5-to describe where are you, these need to be justified, 6 validated as best they can ane' then you drive this group i 7 towards again a consensus. { 8 A consensus just simply means, what are the 9 ranges, let's take the example of different hydrologic 10 models. I~think we could well envision that there may be v 11 three or-four hydrologic models that different people.would 12 --say are_ appropriate descriptors of what might happen at j '( 13 Yucca Mountain. 14 I am not sure that it is appropriate at this point L l 15 to narrow those. To me, it is more-important to carry those 16 along with us until we have more information so that we can u 17 narrow that probability-in the. sense that Robin indicated. I L 18 More research gives "1 better information on which of these L l 19 models might be more appropriate. l. 20 The logic diagram that we have here certainly l 21 allows us to carry multiple models along, too, with certain 22 probabilities so that you do different calculations on l 23 different models to get the results in a similar fashion. l 24 So in that fashion, we would have a series of f i i 25 workshops or we would actually ask DOE to sponsor these q

i 114 '1 series of workshops and we-would see them used by us to

/^^{

t S l: 72 -update and revise our performance assessment methodology and 3; certainly would hope that the other three parties would do 4-the same.. 5 I would see it as one to three workshops per year-6 and that.really means one to three topics per year because I 7 think each topic could well be a series of workshops driving j 8 towards that consensus and we could contribute, I think, 9 significant independent technical expertise input to DOE.as 10 .a result of these workshops. So we really want to drive this in the direction 12' of saying, we have done a lot of what.we wanted.to do. We Lk '13: wanted to be a catalyst here. We wanted to prime the pump. 14 We wanted to get things going because we felt that this was 15 an area in which progress was lacking within DOE; that is, 16 having an overview model that would allow them to make these 17-sortlof quick'and dirty determinations. 18 We have done that but we have sensed that it could .19 be of real advantage now to take this second step and 20 sponsor phase two but it is very clear that the utility 21 attitude in general is that we have done this, let's move it-

22 along but let's phase ourselves out of it, not completely

'23 'out of it but phase it back. 24 The utility industry is not interested in '(g q G 25 sponsoring EPRI to have these detailed technical workshops. ~

115 1 The strong'and I'think correct feeling is that that is' DOE's. f\\. i bx 2- -2: job and.the utilities are already putting money into DOE to 31 do-that. They_ don't want to have to pay for'it twice. f 4 That completes our presentation as we had put.it 5 together.and we are certainly open-to any more questions you 6 might have. 7 MR. POMEROY: Bob, let me say first, that is an-8 outstanding contribution and EPRI is certainly to be 9 complimented for the speed and the magnitude-of the 10 contribution. 11 MR. SHAW: That you. That is very kind of you. I i 12 appreciate the comment. Like any others, we like it when we } ) 13 hear a nice comment. 14 -)CR. HINZE: I will second Paul's comments. Bob, I 15 _am curious if you would be willing to share with us some of-16 your thoughts and impressions of-DOE's 0 ponse to your -17 activities and for the past couple of-days we have both sat 18 in a room listening to flexibility'versus rigidity.- 19-You have talked here about a system which will not' 20. be operative unless it-is extremely flexible. The study 21-plans are in place, being reviewed ad nauseam and present a 22 rather. rigid structure for investigation in many peoples' / ~ 23 minds. i 24 The results of the MDT, the work that is going on 25 here, would suggest that there be a lot more flexibility and 'l

1 116 ,g { 1 with the interactive nature of it would lead to changes on a i 2- .rather frequent interval. 3 How-do you see this coming about within the 4 licensing and regulatory aspects and'what kind of support 5 are you getting from-DOE? 1 6 MR. SHAW: The DOE response has two facets to it. i 7 The first facet has been verbally supportive and their 8 participation as we moved ahead was very good, our requests u 9-for information, for access to contractors, was very 10-promptly adhered to, you know, "Yes, who do you want to talk 11 to? Yes,'what information can we get you" and so on. 12 So we had very nice assistance as we moved through L' 13 this whole process. As we have come to the end of it and 14 made presentations'to DOE, they were, I think I can say 15 almost unife.rmly, very well received. 16 But at the same time, I don't want to be naive and 17_ say that DOE is going to fall.over backwards and take the 18 EPRI process and say, "Oh, this is wonderful. This is just-19 what we needed." They have been working for years and spent 20 a lot of money and it'would be almost-inappropriate-to say, 21 "Oh,.yes,.what EPRI has done for a small amount of money in 22-a short time is just what we need and we will use it." 23 Nonetheless, I see many acpects of the' DOE plan 24 and of some of their activities that pattern or are pattered J 25 very mu'.:n af ter what we have done.. They have four primary

-<.,o p, m $N: 1 AR ~' 117 i o 1 activities they are-working'on right now, the surface-based -l y-j_ 0 \\- 2 testing task-force. They have an alternate licensing p 3 strategy task force.- They have a Calico Hills task force b 4 ' and they have an exploratory shaft facility task force. 5 Almost each one of those is using logic diagrams.

• t O

6 They are doing what I would call "back of the envelope" 7 performance assessment kinds of calculations. They are 8 really not adhering to the same philosophy that we have seen 9 for years, take it slow, take it easy, make sure everything } 10 is in line before we come to resultc. 11 They are really saying, " Hey, let's move ahead. 12 Let's do some cost benefit analyses. Let's do these (r - 13 - things. Let's come up with some conclusions and let's move 14 ahead on that basis." 11 5 -There is also this sense of flexibility that I see-16 in it too,-where it says that this'is a tool, we have ~ 17 developed it but we are going to change it as things moved '18 along. I think the attitude of Bartlett, John Bartlett, is 19 pervasive here in that I think that.he is very supportive of-20 this type of an approach but it.is also a very large 21 structure that he has out there that has dealt with things, 22 I think, in a different fashion-in the past and it takes 23 time to make those changes. /s= 24 With regard to.our phase two, I have every b 25 indication ~from key people who report directly to' Bartlett r r - +

w ~ ) o ' ok 1 l*e 'r j W ". 118 j u l 1 that DOE: will b supportive, that they will send whatever ,~ 2 ~ people are appropriate to'our workshops in phase two, that L 3 they will come and present their models and listen to and j 4 hear the oths? models. .) 5 The'other picture I have is that right now I am I 6 told that DOE has no money in their budget to conduct I 7 workshops beyond phase two. So I have tried to lay it out L 8 so that those could be started roughly fiscal year 1992' 9-start so that hopefully now the consideration of their q 10 budget as it proceeds through fiscal 1992 starting the first L 11-of October 1991, that there will be consideration.in the p. 12 budget for those kinds of things. 1 (! ) 13 The concept of expert judgment, of using 14 workshops, seems to be highly supportive within DOE. I 15' think it is a recognition that they have done a lot of vsry 16 good, very sound work but that it is very valuable both . 17 internally and from a public perspective for them to have j 18 other-experts participat a and interact, that that is 1 19 healthy, I think, on both sides. 20 So the. general impression I get-is that they are. 21' favorably inclined'towards that but I am still waiting to 22 see them make the moves that say, "Yes, we will put our 23 money where out mouth is." r-- 24 MR. HINZE: That is an excellent response. Let me kJ 25 ask another question. As you well stated, this is a i I \\

'119 y 1~ probabilistic analysis which is deterministically based, i 2 . presumably deterministically based. There has been a lot of j 3 work done over the last decade: or more in the - Yucca Mountain 4 region, a lot of data produced and that data and the i 5' information that is derived from it'are distributed widely 6 in the literature. 7 I worry -- no, I am concerned about the person 8 that, well, all of us even those that started at the very -9 base, having access to the information, particularly new l 10. experts that come into the procedure, having access _to all 11 of this'information and data because your results as you 12 have put it,.and I think Robin did as well, are based upon 13 the deterministic aspects of it. 14 I am wondering what EPRI is doing to make certain l-15 or will'do in their conversations with DOE and others about L l 16 making-certain that the data and information that'are needed 17 to come to these probabilities are available in an efficient 18 manner. 19 MR. SHAW: I am not sure EPRI can do anything. l-I' 20 along that line. It seems to me that it is more 1. 21 appropriately handled within peer meetings where I think the p L 22- -people who have the judgment, who have the expertise in a ( 23 particular technical atea can very much demand that results l' ^ 24 which are for whatever reason not available be brought to [ u 25 the table because if some of their peers acrosr the table L i i is

7 120 i 11 are saying, "Well, we know this is so.because we have the .,rq 2 data" and the peer on the other side of the table should be 3 saying, "Show me that' data. Demonstrate to me, prove to me 1 4 .that the model that you have in mind really~does represent-5- the data and unless you can present *.he data, there is no 6 justification for the model that you are arguing for." 7 So I would-see that these pter workshops, these i 8 expert workshops, are the perfect arent in which pressure 1 9 can be applied to have such data released and brought to the ( 10-table in an appropriate fashion. 11 MR. HINZE: I am not certain the corporate memoryL zi2 is really-there tecm some of this information and data but- /^N H ] ): 13 so be it. 14 MR. POMEROY: Bob, to pursue the question a little 15 bit, my last question that I had, and that was a question of 16 credibility to the public, could you just expand a little. 17 bit on-your thoughts in that regard especially. .18 - I can always see a group at the far end, at one 19 end or another of a spectrum of intervenors, coming up with. 20 a set of experts who might come to quite different 21-conclusions and'are in essence by their nature or for one. 22 reason or another excluded from the process up until a time 23 when it is too late to incorporate them. 24 I am concerned about the large number of people k("'g ' "U 25 that are experts or will be expert in any given area and

fS j 121 2 whether-you can bring all of those into a consensus that, I 1~ 7 2 presume, you are looking at as the credibility _to the 3: public. l' -4 MR. SHAW: We see that the process of expert L'

5 judgment used in workshop 8, M ving towards a consensus with 6

the right participants leads you to better credibility than -7 you would have if you don't have such a process. 8' To the extent that DOE has been isolated from the L 9 public, I think the credibility is lacking there or at least (- 10 the perception of the lack of credibility. We would t 11 certainly emphasize to DOE that in such workshops, the i' l i 12 opinions of others who are particularly outspoken such as 13 the State of Nevada, the scenarios that they have developed _- 14 should be brought to that-table, should be presented in l. L 15 exactly the same format as other scenarios and'should be l 16 argued'on the same basis. L 17 What is the real likelihood that such things will 18 . occur? I'think inevitably.there will be some scenarios or 19 some positions that will not be dealt with but one would 20 hope that most of these are handled in that appropriate 21 fashion in these workshops. 22 I also don't feel that once you have conducted a 1' 23 workshop, let's say, the first one happened to be t L, 24 volcanology and you conduct a workshop, you convene this and l '(L' ,25 you complete it, let's say even in calendar year 1991. That

n m, y) i pt. -122. 1 it {,. 1-doesn't mean we are done. The license application is not ] l.. 2 ready at the end of 1991 and it very well could be that new 3 propositions arise, new data arises and so on so that one 4 would reconvene or have another workshops on a particular 5 topic based on the fact that things have changed s ince the-6 last one. 'I 7 So I do see that this iterative approach. that ,{ 8 certainly was advocated in the Board of Radioactive I 9 Management at the National Academy of Sciences, is very-10 appropriate for this workshop scenario as well. ? [ '11 I think another factor that is important here is l 12 that in conducting a workshop.that brings together expertise 13 particularly some from academia who are in general ~more well 1' l: 14 respected by the public than those who are considered to . 15 have a particular corporate bias or a government bias or '16 whatever-it might be, bringing _ experts of this nature in and j l

17 having them say, "Yes,-we agree on the position that-has>

l l 18: been presented" will help in the public arena.. I think the 19. right word is " help." 20L MR. POMEROY: Thank you. I only have one small. 21- 'other question. There are, of course, a large number of 22; details that are incorporated into some of the things that '23 you have presented here this morning. Are the details of 24 those models going to be incorporated into your report? In 25 other words, will there be a detailed description?

123 l' MR. SHAW: Yes. The report is divided into 2 chapters, each chapter of which is essentially a node. For 3 example, all of-the details'that you heard from Mike 4 Sheridan-are in a chapter that he has written within the 5 report and the same thing can be said for each of the other '6 people who developed their particular expertise. 7 So it is developed in chapters on the particular 8 technologies and then Robin has a few chapters where he goes-9 into the overviews and the sensitivities so all of those 10 assumptions are laid out, I think, with great transparency 11 in the report. 12 MR. POMEROY: Thank you. 13. MR. MOELLER: A couple of comments. You said 14 early~and I just wanted to confirm it that human intrusion 15 and gas flow such as carbon-14 are not incorporated _in your 16 model at the moment. 17 -MR. SHAW: That is correct. 18 MR. MOELLER: So they will not be in'this~ upcoming 19 report? R20 MR. SHAW: That is correct. 21 MR. MOELLER: In terms of your observations, of 22 course, you gave observations for hydrology and geochemistry 23 and as I listened to your presentation, I was putting down 24= some observations-and these, I gather, well, I want to ask, 25 will they be in the report? Let me give you some examples

7 k 124 1. and maybe they are of.such secondary importance that they x \\2 2 are not worthy but when I was listenir.g.about faulting,.your-3 diagram showed that secondary faulting dominates versus <4 primary. Now will that be in there? 5-MR.-SRAW: Yes. I 6 MR. MOELLER: And things like that.- Fracturing, 7 you showed how important the fracturing of the welded tuff 8' would be and on and on. All right. But this hydrology and 9 geochemistry, were they dominating the observations here? .10 MR. SHAW: That was simply my synopsis as a Way of 11 collecting together the major features in let's say a single 12 viewgraph and so that was my attempt to classify those in jl ) 13 two general classes for most of what Robin correctly i l referred to'as the necessary conditions that would lead to l 14 l l. 11 5 failure. l 16 MR. MOELLER: If it is true that the canister y ~17: design has little, and I may be using the wrong words,- 'l .18 effect as you say, to me that is a very significant finding. 19 I have been hearing the statements-well the-Swedish people, 1 20 they ctn design a 10,000 year canister and' in fact, I have 21 heard statements that all you need is a good canister and l l 22 you could just dump it in Lake Michigan and leave it there 23 and everything would be fine. l l 24 MR. SHAW: I think you got the wrong message. 25 MR. MOELLER: All right. What is the right l i

125 i [ 1 message? L2-MR. SHAW: There were three canister designs and 3 we said there was very little difference between the two 4 more short-lived but the one that was very long, the super 5 design, so-to-speak, was way down in the bottom. 6-MR. MOELLER: Oh, i Lissed it then. All right.- 7 MR. SHAW: We found a significant advantage for 8 the Swedish Cadillac, so-to-speak. 9 MR. MOELLER: All right. Thank you. Also, one 10 last question that ' had, you have pointed out that this is 11 illustrative, it is not quantitative and so forth but you 12 know you can't help but want to ask if you, for example, the f ) 13 NRC staff recently announced and I guess have published the 14 results of their "back of the envelope" assessment. Have 15 you looked at that and compared it at all to your effort? 16 MR. SHAW: We have looked at it but we have not 17 made a strong, comparison and that is one of the things I 18 wanted to do at this first workshop. Let me talk a little 19-bit more about this word, " illustrative." 20 I think the important features that we feel should 21 be looked at in this is that we can make relative 22 comparisons and the observations and conclusions we came to 23 show that, that we have said, "Yes, we think some 24 conclusions can be made with regard to what are the more 25 important features in this performance assessment." i

33Mg._l^ r;n y t' O Rf ) o i 126 3 1 But it is' illustrative in the sense that I would 2. not want: to and we have not added together all-the isotopes j 3 we have calculated on a CCDF and said, "How close does that i 4 come to the limit?" i 5. So that the particular numbers that go into that, 6 we are not sufficiently confident of from just single 7 experts in the various areas to say, "Okay, this should be 8 compared for the licensing or the regulatory aspects." 9 But we do feel that it is appropriate to take the 10 performance aspects that are described here and on a 11-relative basis we can come to' conclusions with regard to 12 where we think the most important aspects and the less -/'*\\ l 13: important aspects are based on our analysis. {) 14 .MR. POMEROY: Bob, can I ask one more question along that line. Robin listed three or four factors that = i 16 were factors that clearly-showed that they were areas of 17 large concern..As he;did that quickly, I didn't recognize 18= that any of them were major surprises in my mind. Perhaps I 19 , missed something that'should have'been but could you comment 20 or could Robin comment on were there any surprises to you-in 21 the illustrative example in terms of the importance of some 2 2 '. of the parameters? Large fracture flow, for instance, 23 'doesn't surprise me that it is a very important factor. 73 24-MR. MOELLER: Or the waste solubility. hs_/. 25 MR. POMEROY: Right.

m - 127 L.j s .1 MR. McGUIRE: One of the perspectives that I 2 gained from.this that I think is-an important one is that 3 what leads to large releases, again for these illustrative 4 . calculations, are not really bad or incredible scenarios i L 5 like.hydromagmatic eruptions. What can lead to large p l 6 releases are combinations of things that are realistic but 7 which have a low but not very low probability. 8 They are not ten to the minus four or ten to the 9 minus five kinds of events, they are kinds of values or 10 parameters or assumptions that might have a probability of' ] 'I 11 ten to the minus one. 12 But they have to occur in combination. So if you j .( ) H13 need six of them, then the probability of that combination 14 is ten to the minus six. So the perspective I get is that l 15 it will require a combination lof bad things and maybe the 16 safety in a site is in ensuring that the probability of that .l f 17. combination.is low. 1 18 So you need to look at the combinations of. p [ 19 parameters not just fix on one potential cause of failure or l one potential fault in the sense of a fault tree to evaluate 20' 21 the site and I think that I was not expecting. I was 22 expecting more to be able to pin or pointLto a particular 23 failure mechanism or two or three failure mechanisms and a ,rx 24 that, at least, from these illustrative calculations has not A 25 been the case. ]

,, ~ 123 1 MR. SHAW: Let me just add a point to that and 's_/ 2 maybe it is clear but I think the system consideration is a l 3 very important result that we see coming from this, that to-4 a great extent the analysis has been done on an individual 5 technology in the past and to look at the total system 6 considerations.is an important aspect and one that we feel .7 needs to be done even if not in great detail, it still-needs 8 to be done to get a general sense. 9 The iteration between the various technologies 10 here as we brought these experts together was an exciting (, 11 venture to see it happen and to see the questions that were L 12 raised and the conflicts that came up because of the )- 13 interfaces between them and not individual conflicts but 14 just the technical kinds of conflicts of'how do you express 15 these results from one output so that they are-useful to the 16 next one and are they really consistent. '17 Robin went into his point five likelihood to show j l '18 'that'it'was appropriate to make surc those were consistent L 19 so you really can compare things and that is an important 20' part of the system's analysis certainly, that.you make'sure 21-that the things-that you are comparing are being compared on -22 a reasonable basis. L MR. MOELLER: Let me thank you once again on 23 L L;g-s 24 behalf of the committee. I think it is astounding how much L / {, ~ you have accomplished in 15 months. I was totalling up the ( 25 L l l

I 129 1. time and-it has!been very refreshing for us to hear-this. I ) ~l \\ 2 l report. ,3' Before we recess for lunch, let me just ask the g L4-committee, in terms of recording the rest of the day, I see l 5' no need for items four and five, the executive session, when L 6 we are responding to the EPA's comments or discussing the 7 NAS/NRC Report of having that recorded. Now is there any 8 value in'having the 1:30 to 2:30 session recorded when we i 1 I 9-talk L'ucut responding to commissioner Curtiss' questions? L l 10 (No response.) i' 11 MR. MOELLER: All right. I tend to think that l 12 there is no benefit in that so with that then, we will .f f 13 conclude the recorded sessions for today. We will be 14 recessing for lunch until 1:30 and let me stress to any L 15 members of the public who are here that this afternoon's V 16 sessions although not recorded will definitely be open to if -the public and you are invited to=come back and hear them if-18 you desi.ie. Let me thank our reporter for being.here and 19 once again thank our visitors or guests from EPRI for a 20 tremendous amount of work on their part and a very 21 beneficial presentation for us. Thank you, d 22 (Whereupon, at 12:15 o' clock p.m., the committee 23 was adjourned.)

24.

),3 25 I

y i s i( ) REPORTT 'S CERTIFICATE This is to certify that the attached proce<d-ings before the United States Nuclear Regulatory Commission in the matter oft NAME OF PROCEEDINGS 24th Meeting DOCKET NUMBtRt FLACE OF PROCEED 77#J: Bethesda, Maryland { were held ac hereht appears, and that this is th: 6tisinst transcript thereof for the file of the United States Nuclear Regulatory Commission taken by us and thor::Icee reduced to typewriting by me or under the direction of the co..rt report-ing company, and that the transcript'is a true and acciarate record of the foregoing proceedings. N J_ Nf) - 77 L Official Reporter Ann Riley & Associates. Ltd. ? e r e l l l' T f a , - -, +,

~W <-.,nre-. EPRI g i c . j , -,. n-o n,.e seME EPRI mgwi., M HIGH LEVEL WASTE ~ PERFORMANCE ASSESSMENT METHODOLOGY I O ~ i Presented to ACNW Se 3tember 19,1990 i Washington, D.C. l t I ): l Presented by l L R. A. Shaw h Robin McGuire ) Michael Sheridan 1-m,-."- .m, ..m,. ~ +

EPRLNPO EPBI HLW Project Objectives . To develop an intogleted method ( Oy for oorly site performence esensmont end to identrty end priornize crucialissuos To involve DOE in this methodology Oevelopment and its 6mplementation 4 HLW / SFS hhteCarp Me.D 1 kPRLUPD EPRI H6gh Level Weete Project Methodology Development Team same. nenman De w B.Bolen George Tech Weste Pecma06 Novee Coos Urw.of Calif,Brhemy Rock Mechance Keve Coseersmith Geomains Consuinna Seemic Geo6agy Raeh L. Keeney Urw.of Souvem Calitome RiaWDecean Awys.: John M Komony L%ersty of Argona Rock Mechance Ausen Long thorsty of Argona Cismemogy Rotun K. McGuve R4k Engineering R.ek Ansero.s F Joseph Peerson.Jr. Corowinnt Geocherrestry F rent W. ScNwa1: Ohe Smie Umversity Hyersegy W heetthenaan SaleUnN of NY.Battalo Votano*0y Roten A.Show EPRI Protect Manager J. Cart Sieoo EPRt - - -., & Geoonytes Rotort F. Wisems EPRI HLW Scences Roesq Youngs Goematna Consultants Geoeschtwani Engineenne De ben S. Barin UNLV/E RC Observes a Russ Dyer Desertrent of Eno<gy Otwver HLW / SFS ncee

't l

h t t EPMMPO Methodology Development Team Meetings i l i t 7/24 25/89 Brainstormmg j 11/28/89 Ovaldication check 1 12/19 20/09 Problem definnion 1/15 17/90 Modelformulation 4/24 26/90 Modelpresentsten j 7/30 8/1/90 Modelcompleton i b ? ? HLW / SFS m ese s t F l

s I g; END R'A ANCH s g! NO. PRC5. PARAMETERS H a EI 1 P,, s P;, x R, x P., E,, S, , H, 5 S, gi .g k*. P,3 2 P.,z P,ix P,,x P., E,S, , H, p 3 E,, S:, H, - E, P. P, H. E,, S, , H. pu H. E* P* Ss H. p, E,, S, , H, p, p S H, 4 E,, S. , H, p,, P., H. P. E,, S. , H. E=1 Figure 91 Example logic tree, ^*

I ? I I i P 1 f i END BRANCH aj" i NO. PROB. PARAMETERS $w mRANCH a" w a w" i 1 P., E,, S,, H, g SOURCE AND O 2 P, E,,S,,H, HYDROLOGIC $C I TRANSPORT C4 C ALCUL ATION S 2 3 P,, E,,S,,H, / i llo '4 P, E,, S,, H. o m ,8 TIME t r U l 1 10' ' - b 10 - l

f e

< 10- a. i eoE 10**- a. 10** CUMULATIVE RELEASE i i Figure 9 2. Illustration of use of logic tree parameters to form CCDP of cumulative chemical concentration released. i 1 I 1-l l l~ l n.. m U

j 9O p s N \\ OI Y IV \\ OUV g G ISU / gr ~L i I'SG AIISO-O g Y \\ N U-SU / DIOV O +? d U d b, A 1 'I8'O M 1I OU 's. SWI13 nI \\\\ d' OS II 31S - 4' M-@ USISINV Y 4 A ~ I'I x II O EV NV IS 3 SS 3'O M-O I b'- \\ NVD' H3H 0 \\ IO O A U M g ? AI O I U d AI1OV W S3MV 1 A SS DD OI E 'I 3 H NV h s I IEV 7 + O \\. Vd 3 0 M A-USISI x s OU d NY I O MV G3S (N s nV O11V OD ulnI 3 \\ dNO \\ X ( s n7 3 \\ \\ \\ 0 s I . ~

I NET FLUX 6 g3 r S*'V Ah Y Node 1 Net Flux = 1.6 mm/yr p=0.90 Net % 4 %y,r 0 Figure 9 4. Logic tree values and probabilities for Node 1. i W

f i j e x i WATER TABLE CHANGE 1 VOLCANO t M \\ 0-m Rise in Water Table t pol.0 VOLCANO EFFECTS i &c .i 91 set 08

• 06 s

&*/ge V Bo,* Nise. I 61*\\ Node 4 g 8% y,3fe p ***i. # a He + 5%4 N. 4,4 p= 1.0 0-m Rise in Water Table I P 0-m Rise in Water Table pol.0 Figure 9 6. Logic tree values and probabilities for Nodes 4 and 5. 6 0.0

1 i i I r' t i I h t L r i t. t i 5 Base Case Post Research Prebabilitics: i Flux value, mm/yr Probabiliticg 1ihelilioncl=0 5 Likelihonri=n s i 0.5 0.08 0.3 0.0

t. '

1.6 ~ 0.90 0.7 0.7 4.0 0.02 0.0 0.3 l Area of Large Base Case Post Site Investigation Probs. 1 Fractures Probabilities Likelihood =n.5 Likelihond=0 5 ~ ~ Low 0.5 O.8 0.2 High ' O.5 0.2 0.8 t b l'l. 1 1 l' t D 9-4 u. .m .4

c 1 t 3 4 Obserystions

• Keeping in mind that the MOT resuns are illustrative, the j

following are found to be more influential on she performance i ' 1 Hydrology - - Infitration (recharge) from preciphaten - Water flow pathways Influenced by extent of rock fracture and porosity - Significant rise in water table Geochemistry e - Uranium solubility, as influenced by dissolution chemistry 'j and temperature { , Chemical retardation of released radioisotopes j ( HLW / SFS 2 m ese s l 4 i ) EPRtWPD Conclusions The use of muhbdisciplinary scientific and engineenng expertoo to q conduct a rek based evaluaten of a HLW reposnory is achievable wkh current knowledge and technology. A c'ructured approach is roouired; the workshop format is l surted to this approach. The use of logic trees is a convenient and credible format Results of the methodology should be obtained during the e process of model development, i.e., the process should be Rorative. A methocology of this type can be applied on a larger scale, in L. which a larger body of expertos parteipates. This appiscation will i lead to realetic (rather than simple demonstrative) resuhs, l.' HLVV / SFS L l t i-L

) i 1 l 1 EPftFNPO ) Near-Term Plans

• Prepare working version of Methodology Development Team j

performance assessment model and report (940) i I Phase 2: Join with DOE in sponsorship of workshops on e performance assessment methodologies to identWy crucial technicaltopes for workshops i Phase 3: Svaport DOE in conducting expert workshoos on crucial techncal topics identifed in Phase 2 -l 1 HLW/SFS I 1 i Phase 2 ) p

• Series of workshops on performance assessment metho6 ologies

- Participants DOE YMPO contractors DOE HQ Contractor, Golder Associates NRC EPRt!UWASTE's Methodology Development Team - Objectrves Exchange detailed explanations of each P/A methodology Revise methodologies where appropriate Obtain consensus on highest pronty technical areas ( HLW / SFS meum - e b /

q. 'l EPRLNPD Phase 3 Serlos of workshops on highest priority technical areas idemded in Phase 2 - Sponsored by DOE - Used by EPRI to update and revne P/A methodology - One to three workshops por year - Significant 6.sdependent technical expwt inout to DOE Industry's technical representative in Bartlett's proposed technicalforum HLW / SFS m ese s 6

l f I b r 1 EPRL'NPD i EPRI HLW Project Objectives 2 . To develop an integrated methodology for early site performance assessment and to identity and pnntnize l crucialissues } P . To involve DOE in this methodology covelopment and its implementation l HLW / SFS 5 EPRl/NPD EPRI High Level Weste Project Methodology Development Team Nana. .Allillation Esanosas Daniel 8. Buben Geoge Tech Weem Pacha0e Nevil6e Cook Urw.of Cant.Berheey Rock Mechanics Kevin Coppersmith Geomstna Consuriants Seemic Geology Ralph L. Keeney Unw.of SouthemCahfome RasivDecisen Analysis John M. Kemeny Universty of Aruona Rock Mechew Ausen Lon0 Unworsty of Anzona Clanaimogy Rotun K. Mtuire Risk Enginee ing Risk Anaysis F. Joseph Pearson,Jr. Consuitant Goochemistry Frank W. Schwartz Oho Siste Arworsity Hydronogy Mchael Shoneen State Unw of NY,Buttalo Vcicanoso0y Robert A, Shaw EPRI Protect Manager 5 J. Catt Sieon EPRI Seempo0y & Geophys.cs Robert F. Wisems EPRI HLW Scences Robert Youn04 G*omains %nsultants Genescencel Engireenno De4bert S. Barth UNLV/ERC - Observer Russ Dyer Deperiment A Ene gy Observer HLW/ SFS O- .m.c m e i e E

j. g. EPRLWPD Methodology Development Team Meetings. 7/24 25/89 Brainstorming 11/28/89 Gualdication check 12/19 20/89 Problem definnion 1/15 17/90 Modelformulation 4/24 26190 Modelpresentaten 7/30-8/1/90 Modelcompleton

• HLW / SFS O

O 4 1

'i t h \\ c( I .i I t i + i. END BRANCH 3: H, NO. PROB. PARAMETER 81 1 P., z P,,x P,,x P E,, S, , H, { p,, ai H, y 1 P,, 2 P., = P,,x P x P, Es,S , H, g p n i He E,, S, , H, E, 1 1 8, H. E,, S, , H, p, H. 'H 8 e L E, Pu S H. f' P, E,, S, , H, j Pg p Sa H, E,, S , H, l 4 p,, P., H. E,, S, , H, P, E=1 r 1 Figure 91. Example logic tree, t O 5 b ~- + W

g l

O aJ ' END BRANCH q=< BRANCH: NO. PROS. PARAMETERS 9w lw 3 1 P,, E,, S,, H, ~ SOURCE AND o 2 P,, E,,S,,H, HvoRotooiC ,c g e< 2 TRANSPORT m. 3 P,, E,,S,,H, / CALCULATIONS aw 4 P,, E,, S,, H. ggo o i 8 TIME y 1 i o- ' - t: 10 =f I s o - Eg t o* i to CUMUL ATIVE RELEASE Figure 9 2. Illustration of use of logic tree parameters to form CCDF of cumulative chemical concentration released. O 4 9-5

.4 e e, e O Sa NV a

1. '%)[ @

h$ <<n,:; r N i !e ,,@g I 29 N a

1. D

\\ 8h 0 g*s'4't &g%, 3 4 T@ff N j O 3 *** I + f 'V 'tq N o 9 yg 'I'o N 2 4,4 4 n %y %f 0p 1 N [9 %ip, Vo ip yo

• sy,N,, x g 3

.'0 ~,k Nke ~ 4n,4 e o0 x d'/.) l'n7 N Q 4 4 N /e

E h NET FLUX 1: 4 g,50 9% 4*'9\\ I Node 1 g Net Flux = 1.6 mm/yr p=0.90 Net %g Og %/r Figure 9 4. Logic tree values and probabilities for Node 1. G l'- u

h j ] U. i l i WATER TABLE CHANGE I VOLCANO l 1 0 m Rise in Water Table ) pol.0 VOLCANO EITECT_S 4#@D ) 08 M t- '!! Rise g

• te Nade 4 ry he i

4 9 0 m Rise in Water Table 4 \\ 0 m Rise in Water Table t p= 1.0 Figure 9 6. Logic tree values and probabilities for Nodes 4 and 5. t f -.. M -

t i 1, l O' i i i l i ? i I Base Case Post-Research Probabilitier,: l Flux value. mm/yr P_rohnhilitics Likelihood =0.5 Likelihnnd=0 s i 0.5 0.08 0.3 0.0 1.6 0.90 0.7 0.7 4.0 0.02 0.0 0.3 f g.i Area of Large Base Case Post Site Investigation Probs: Fractures Prohnhilities Likelihond=0.5 Likelihnnd==0 5 Low 0.'5 O.8 0.2 High 0.5 0.2 0.8 + L l~ l s l t'

_. ~ i .(; EPRLWPD Observations

• Keeping in enind that the MOT resuns : e i!ustrative, the following are found to be more inf tential on sne performance Hydronogy

- Infihrstion (recharge) from preciphation - Water flow pathways . Influenced by orient of rock fracture and porosity - Significant rise in water table Geochemistry e - Uranium solubikty, as influenced by dissolution chemistry and temperature 4 Chemical retardation of released radioisotopes HLW / SFS ~~~" O Conclusions The use of muni4isciplinary scientific and engineering expertise to conduct a risk based evaluation of a HLW reposnory is achievable wth cxJrrent knowledge and technology. . A structured approoch is required; the workshop format is suited in this approach.

• The use of logic trees is a convenient and credible format

. Results of the methodology should be obtained during the process of model development, i.e., the process should be herative. A methodology of this type can be apphed on a larger scale, in which a larger body of expertise participates. This application will lead to realistic (rather than simple demonstrative) results. HLW / SFS O.

O EPRINPD Near Term Plans 1 i l Prepare worxing version of Methodology Development Team ) performance assessment model and report (9,90)

j Phase 2: Join with DOE in sponsorship of workshops on J

performance assessment methodologies to identWy crucial -j technicaltopes for w)rkshops ) l Phase 3: Support DOE in conducting expert workshops on crucial technical topics identified in Phase 2 I HLW / SFS e:! EPRLHPD Phase 2 Serse of workshops on performance assessment methodologies - Participants DOE YMPO contractors M)E HO Contractor, Golder Associaias NRC j EPRIAJWASTE's Methodology Development h. m c - Objectives Exchange detailed explanations of each P/A methodology - Revise methodologies where aporopriate Obtain consensus on highest pronty technical areas 4

!O EPRLWPD Phase 3 Series of workshops on highest priorny technical areas identifed in Phase 2 - Sponsored by DOE - Used by EPRI to update and revise P/A methodology - One to three workshops por year - Significant independent technical expert input to DOE Industry's technical representative in Bartlett's proposeo technicalforum HLW / SFS ~ ~ ' ~ ' ' O O}}