ML18004A157
| ML18004A157 | |
| Person / Time | |
|---|---|
| Issue date: | 10/18/2017 |
| From: | Hossein Nourbakhsh Advisory Committee on Reactor Safeguards |
| To: | |
| Nourbakhsh H | |
| References | |
| NRC-3327 | |
| Download: ML18004A157 (198) | |
Text
Official Transcript of Proceedings NUCLEAR REGULATORY COMMISSION
Title:
Advisory Committee on Reactor Safeguards Regulatory Policies and Practices Docket Number:
(n/a)
Location:
Rockville, Maryland Date:
Wednesday, October 18, 2017 Work Order No.:
NRC-3327 Pages 1-168 NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers 1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005 (202) 234-4433
NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 www.nealrgross.com 1
1 2
3 DISCLAIMER 4
5 6
UNITED STATES NUCLEAR REGULATORY COMMISSIONS 7
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS 8
9 10 The contents of this transcript of the 11 proceeding of the United States Nuclear Regulatory 12 Commission Advisory Committee on Reactor Safeguards, 13 as reported herein, is a record of the discussions 14 recorded at the meeting.
15 16 This transcript has not been reviewed, 17 corrected, and edited, and it may contain 18 inaccuracies.
19 20 21 22 23
1 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION
+ + + + +
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS (ACRS)
+ + + + +
REGULATORY POLICIES AND PRACTICES SUBCOMMITTEE
+ + + + +
WEDNESDAY OCTOBER 18, 2017
+ + + + +
ROCKVILLE, MARYLAND
+ + + + +
The Subcommittee met at the Nuclear Regulatory Commission, Two White Flint North, Room T2B3, 11545 Rockville Pike, at 8:30 a.m., John W.
Stetkar, Chairman, presiding.
COMMITTEE MEMBERS:
JOHN W. STETKAR, Chairman RONALD G. BALLINGER, Member DENNIS C. BLEY, Member CHARLES H. BROWN, JR. Member MICHAEL L. CORRADINI, Member JOSE MARCH-LEUBA, Member DANA A. POWERS, Member
2 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 JOY L. REMPE, Member GORDON R. SKILLMAN, Member MATTHEW SUNSERI, Member DESIGNATED FEDERAL OFFICIAL:
HOSSEIN NOURBAKHSH ALSO PRESENT:
ANDREA D. VEIL, ACRS Executive Director NATE BIXLER, Sandia National Laboratories KEITH COMPTON, RES HOSSEIN ESMAILI, RES ED FULLER, RES TINA GHOSH, RES TREY HATHAWAY, RES SALMAN HAQ, RES DONALD HELTON, RES WILLIAM ORDERS, NRR EDWARD ROACH, NSIR PATRICIA SANTIAGO, RES AMY SHARP, RES TODD SMITH, NSIR CASEY WAGNER, Dycoda LLC KIMBERLY WEBBER, RES
3 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 C-O-N-T-E-N-T-S Introductory Remarks...............................6 Patricia Santiago, RES Focused Pressurizer Safety Valve Study (New Appendix I of Draft Report)..................26 Trey Hathaway, RES Summary of Report Updates in Response to ACRS Member Comments from Prior Subcommittee Meetings.............................89 Tina Ghosh, RES Casey Wagner, Dycoda LLC Nathan Bixler, SNL Discussion.......................................143
4 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 P R O C E E D I N G S 1
8:31 a.m.
2 CHAIR STETKAR: The meeting will now come 3
to order. This is a meeting of the Advisory Committee 4
on Reactor Safeguards, Regulatory Policies and 5
Practices Subcommittee.
6 I'm John
- Stetkar, Chairman of the 7
Subcommittee Meeting. And you're not.
8 Members in attendance today are Ron 9
Ballinger, Matt Sunseri, Dick Stillman, Dana Powers, 10 Mike Corradini, Jose March-Leuba, Dennis Bley, I 11 forgot you, Charlie Brown, and Joy Rempe.
12 Hossein Nourbakhsh is the Designated 13 Federal Official for this meeting.
14 The purpose of today's meeting is to 15 discuss the State-of-the-Art Reactor Consequence 16 Analyses Project for the Sequoyah Integrated 17 Deterministic and Uncertainty Analyses.
18 Today we have Members of the NRC Staff and 19 Sandia National Laboratories to brief the 20 Subcommittee.
21 The ACRS was established by statute and is 22 governed by the Federal Advisory Committee Act. That 23 means that the Committee can only speak through its 24 published letter reports.
25
5 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 We hold meetings to gather information to 1
support our deliberations. Interested parties who 2
wish to provide comments can contact our office 3
requesting time after the meeting announcement is 4
published in the Federal Register.
5 That said, we set aside ten minutes for 6
spur-of-the moment comments from members of the public 7
attending or listening to our meetings. Written 8
comments are also welcome.
9 The ACRS section of the NRC public website 10 provides our charter bylaws, letter reports and full 11 transcripts of all full and Subcommittee meetings, 12 including slides presented there.
13 The rules for participation in today's 14 meeting were announced in the Federal Register on 15 October 10, 2017. The meeting was announced as an 16 open meeting.
17 No written statement or request for making 18 an oral statement to the Subcommittee has been 19 received from the public concerning this meeting.
20 A transcript of the meeting is being kept 21 and will be made available as stated in the Federal 22 Register Notice.
23 Therefore, we will request that all 24 participants in this meeting use the microphones 25
6 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 located throughout the meeting room when addressing 1
the Subcommittee.
2 The participants should first identify 3
themselves and speak with sufficient clarity and 4
volume so that they can be readily heard.
5 We have a bridge line established for the 6
public to listen into the meeting.
7 To minimize disturbance, the public line 8
will be kept in a listen-in-only mode, and I'll open 9
it at the end of the meeting for public comments.
10 To avoid disturbance, I request that all 11 attendees and everyone else in the meeting room put 12 your electronic devices like cell phones, beepy things 13 and whatever, in the off or noise-free mode.
14 Now, I'll proceed with the meeting, and 15 I'll call upon Pat Santiago of the NRC Office of 16 Nuclear Regulatory Research to begin today's 17 presentations. Pat?
18 MS. SANTIAGO: Thank you, good morning.
19 We appreciate all the feedback that we got 20 from the Subcommittee Members in the May of 2016 21 Subcommittee Meeting, as well as the June meeting of 22 this year.
23 And we worked the last few months to 24 address all the comments that we received.
25
7 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 On this second slide, I want to talk a 1
little bit about our schedule.
2 Our Sequoyah Analyses is going to be 3
published as a NUREG/CFR report, and it's due to the 4
Commission November 30th. And we'll be sending it to 5
the NRC Publications Branch at the same time.
6 We're relying on the ACRS Members to 7
service our peer reviewers and we're requesting a 8
letter documenting your review. To facilitate that 9
process, we're briefing the full Committee in November 10 of this year.
11 And this works well with our schedule 12 since the NUREG is due to the Commission November 13 30th. So, we appreciate your support for that.
14 We hope to be able to address any final 15 comments from today's Subcommittee meeting in the next 16 few weeks, but as you can see from our schedule, any 17 additional analysis would not be documented in this 18 report.
19
- Rather, we could possibility add 20 additional clarifications on what was done within the 21 scope of this study. And time to include it within 22 the NUREG will be submitted for the Commission, and 23 for publication.
24 Also, separately, much of the same team is 25
8 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 in the process of updating the Surry Uncertainty 1
Analysis. And currently, that NUREG report is due to 2
the Commission June of 2018.
3 Previously, the Subcommittee had indicated 4
that you may not need to review the updated 5
Uncertainty Analysis.
6 However, we can discuss whether you'd like 7
the team to come back with the updated Uncertainty 8
Analysis next year.
9 MEMBER BLEY: Pat?
10 MS. SANTIAGO: In addition to that --
11 MEMBER BLEY: Pat? Excuse me. Can you 12 say something about the extent of that update?
13 And is it to make things at least 14 methodologically consistent with what's been done 15 here?
16 MS. SANTIAGO: Okay, so the Sequoyah 17 Update or the Surry --
18 MEMBER BLEY: The Surry Uncertainty.
19 MS. SANTIAGO: We'll talk a little bit 20 about that --
21 MEMBER BLEY: Is that in the presentation?
22 MS. SANTIAGO: Okay, yes.
23 MS. GHOSH: So, there are a couple of main 24 things.
25
9 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: Move a little closer, 1
please.
2 MS. GHOSH: Is this better?
3 You're aware of some of the major updates 4
we made to the Sequoyah Analysis between last year and 5
this year. Some of them are very relevant to Surry as 6
well.
7 All of the safety valve parameters are 8
important for Surry too, because of the induced steam 9
generator tube rupture in that case.
10 And so for a different reason but still 11 very important. So, we definitely wanted to make that 12 update.
13 So, basically, all of the relevant updates 14 in the Sequoyah Analysis in the last year, we are 15 implementing in Surry, and that will change the 16 results somewhat.
17 The other thing is at the Subcommittee on 18 Surry, we got some comments from Bill Shack and others 19 on our steam-generator tube rupture modeling. So, 20 right now, we are currently in the throes of updating 21 that modeling.
22 So, we've gone and gathered more 23 information, talk to more experts, and we're in the 24 process of updating that modeling. So, those are the 25
10 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 two main areas.
1 MEMBER BLEY: Well, since you brought up 2
that last one, are you going to talk about the reason 3
why you don't have any steam-generator tube rupture in 4
the Sequoyah Analysis?
5 Is that part of your presentation?
6 MS. GHOSH: No, we had tried -- okay, so 7
we did not address induced steam-generator tube 8
rupture.
9 MEMBER BLEY: So, you don't have to update 10 it?
11 MS. GHOSH: For Sequoyah. No, that was 12 not within the scope of the study, and we tried to 13 make that clear in the introduction that we did not do 14 that.
15 MEMBER BLEY: You made it very clear.
16 What wasn't clear was why the scopes are different, 17 why you didn't do it here and you've done it there.
18 And now you're going back and redoing it 19 over on the other plant.
20 MS. GHOSH: That's what we say in the 21 Sequoyah study, is look at the Surry study for 22 insights on tube rupture because --
23 MEMBER BLEY: Yes, you do.
24 MS. GHOSH: -- because those should be 25
11 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 consistent. The scope of the Sequoyah study was really 1
focused on containment issues.
2 MEMBER BLEY: You say that kind of at the 3
end, that's where the focus was, but you don't say 4
anything about why.
5 Is there a simple why, or it was just the 6
way the scope was set up?
7 MS. GHOSH: You know, at the time --
8 MEMBER BLEY: This was added in to look at 9
an ice condenser, I guess.
10 MS. GHOSH: Yes, when we went to the 11 Commission and said we should finish this third pilot 12 study for these reasons.
13 Some of the big motivators at the time 14 were still doing the post-Fukushima regulatory 15 actions, and we wanted to make sure we weren't missing 16 anything in terms of hydrogen challenges for the ice 17 condenser containment.
18 So, that was a major focus of this third 19 plant. It was really focused on the unique aspects of 20 the containment.
21 MEMBER BLEY: This kind of makes sense 22 but, at least to this reader, it didn't jump off the 23 page at me.
24 In fact, we're focused on the containment 25
12 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 as kind of an afterthought at the end. It's like we 1
didn't do these things but that's because we're 2
focused on the containment.
3 So, you didn't really introduce it with --
4 MS. GHOSH: Ah, okay.
5 MEMBER BLEY: -- why you're doing this.
6 MS. SANTIAGO: I would re-look at the 7
Executive Summary. I think that's an important point, 8
to make sure that it's clear why we did what we did.
9 MEMBER BLEY: Maybe it is, but it wasn't 10 clear to me.
11 MEMBER REMPE: Let's go a bit further.
12 If you had looked at it -- and since you 13 are doing it at Surry, maybe you have some insights on 14 what would have happened if you had looked at 15 Sequoyah, especially with the Level 3.
16 It's a Westinghouse plant, right? Is 17 there something you think you might be able to say?
18 For example, you did -- most of the time, 19 as I recall, and Don's in the audience and he can 20 correct me, but when they were doing the Level 3, they 21 decided, well, because it's a Westinghouse plant, it 22 wouldn't have been a big deal, except for if you had a 23 loop seal that was about 480 gallons per minute.
24 And do you think you might have the same 25
13 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 kind of conclusions that they had in the Level 3? And 1
then, there's only that one sensitivity study you did, 2
where you did have that high of a loop seal leak.
3 And I'm just wondering if you can kind of 4
make some comments in your Executive Summary, if 5
you're updating it, that could be based on engineering 6
judgment from all the ongoing evaluations you've done.
7 And the only place I think you might want 8
to think about acknowledging it makes a difference is 9
how what happens with this 480 gallon-per-minute leak.
10 Because instead of talking about the 11 containment failure, you might have had a bypass. And 12 it looks like that someone is agreeing with me by --
13 up and down.
14 But anyway, it seems like there's some 15 things you might want to go a bit further and say that 16 it's probably -- again, you guys should say it, not 17 me. But maybe it won't be that important except for 18 that one case.
19 MS. GHOSH: That's a good comment. We can 20 modify it. We can work on modifying it.
21 CHAIR STETKAR: I have to be cognizant of 22 time. We have a 12:00 p.m. hard stop today.
23 We shouldn't try to speculate about 24 differences between Surry and Sequoyah because they 25
14 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 are difference plants. Period. They're different 1
plants.
2 So, trying to speculate about lessons 3
learned at Sequoyah based on what we know for Surry is 4
against what we should be doing.
5 These are plant-specific studies and a 6
conditional probability of a high-dry-low situation.
7 And Sequoyah could be substantially different than 8
high-dry-low.
9 At Surry, we don't know that, and that's 10 the important insight from SOARCA, is that you must do 11 plant-specific and site-specific, integrated, 12 uncertainty analysis.
13 MEMBER REMPE: But then I would argue with 14 you, sir, that the Executive Summary is really coming 15 up with conclusions that may not be valid for a large 16 lead rate.
17 CHAIR STETKAR: That's correct.
18 MEMBER REMPE: And again, the Vogtle 19 analysis did something based on generic knowledge 20 about --
21 CHAIR STETKAR: We're not talking about 22 Vogtle here.
23 MEMBER REMPE: That's true, but again, I 24 think you may want to -- without trying to extrapolate 25
15 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 things from other plants, you may have some 1
conclusions that are not valid for this particular 2
plant --
3 CHAIR STETKAR: That's right.
4 MEMBER REMPE: Because you did not 5
consider the consequential steam-generator tube 6
rupture.
7 So, something needs to be acknowledged in 8
that Executive Summary on the downsides of not 9
considering it.
10 CHAIR STETKAR: It's fair game to talk 11 about the Executive Summary for this study.
12 And the Executive Summary should, in my 13 opinion, clearly delineate what was done, why it was 14 done, what was not done, and why it was not done.
15 MEMBER REMPE: And what are the downfalls 16 of not doing it? I would go a bit further on 17 potential downfalls of not doing it, because some of 18 the conclusions may not be valid because they had a 19 limited scope.
20 And they should have some acknowledgment.
21 And without extrapolating, they could say, hey, we've 22 done a bunch of analyses and this may be important, 23 you may have to do it plant-specific to determine it.
24 But I think you need to have some caveats 25
16 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 in your conclusions. Okay?
1 CHAIR STETKAR: Okay, good. Pat, we kind 2
of interrupted you midstream. Pick up the ball.
3 MS. SANTIAGO: One other thing that may 4
help some of this discussion, actually, is we plan on 5
doing -- basically, after the Surry RAY's done, we 6
want to work on a compendium of important insights 7
that we've gained from the three Uncertainty Analyses.
8 And we're going to publish that as a summary NUREG as 9
well.
10 But I do what to point out that in each of 11 the SOARCA studies, each bottom study during Sequoyah, 12 we do say that this is a site-specific study, and you 13 may have to look at the specific site and other 14 designs of other plants.
15 CHAIR STETKAR: And I think that whenever 16 you come to that sort of general insight, NUREG or 17 whatever, we would be very interested in seeing that.
18 Because I'm very concerned about trying to 19 make the entire pressurized-water nuclear power 20 industry look like Surry, and the entire boiling-water 21 industry look like Peach Bottom. And the entire ice-22 condenser plants look like Sequoyah.
23 So, trying to develop broad generic 24 insights from these very site-specific, plant-specific 25
17 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 studies is dangerous and misguided. And that's on the 1
record.
2 So, that's why I'd be really interested in 3
strength seeing that Overall Insights study.
4 MS. SANTIAGO: Okay, and maybe when we do 5
the Surry, we can give you an outline of what our 6
plans are for that compendium.
7 CHAIR STETKAR: Yes, that would be good.
8 MS. GHOSH: Can I just make one quick 9
clarification?
10 I don't think the point is to try to draw 11 some broad-brush -- you know what, the main issue is 12 just to make something that's more practical than 13 having -- I mean, at this point, we have like 1900 14 pages of Uncertainty Analyses.
15 And people go in and they'll pluck out 16 what they need, but we get a lot of complaints about 17 just how much material you have to go through to pick 18 out that nugget.
19 So, we're trying to make something that's 20 a little bit easier to digest in one place, rather 21 than having to go to the library and dig through all 22 that material.
23 CHAIR STETKAR: Pat? Since you still have 24 the ball to carry.
25
18 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 (Laughter.)
1 MS. SANTIAGO: Next slide, I did want to 2
compliment the team members that worked on this 3
because they worked very hard and diligently on this 4
particular project, at the same time that they're 5
working on the Surry Uncertainty Analyses and other 6
projects, as you heard me talk about on October 3rd, 7
when the Division of Systems Analysis briefed.
8 So, I did want to thank them all. Dr.
9 Tina Ghosh and Doug Osborn were the co-leads. Doug 10 Osborn's from Sandia National Labs.
11 We have a handful of folks in the audience 12 to help us answer any questions as we brief, as well 13 as those on the phone line.
14 On the last slide, I just wanted to talk a 15 little bit about the meeting today, and the focus is 16 going to be on the Sequoyah draft report.
17 Since our last June 6th Subcommittee 18 Meeting, there's two significant additions that we've 19 added to this draft report.
20 The first change is the new introductory 21 material in Section 4 on accident progression, which 22 was presented in June by Hossein Esmaili. And at that 23 time, it wasn't included in the report.
24 We don't plan to brief on that material 25
19 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 since we did brief in June, but I just wanted to 1
mention that.
2 The second addition is a new Appendix I 3
in the draft report, which documents the supplemental 4
Uncertainty Analyses that was focused on a range of 5
safety valve parameter values where an early 6
containment failure is possible.
7 Dr. Trey Hathaway will present this work 8
today.
9 After's Trey's presentation, the team will 10 walk through Member comments and discuss how they were 11 investigated and addressed in this new draft report 12 that we provided you last month.
13 And lastly, we'll seek the Subcommittee 14 Members' feedback on how we should focus our 15 presentation to the full Committee in November, since 16 we'll only have 90 minutes at that particular meeting.
17 And I'll now turn it over to Trey and my 18 Staff to present.
19 CHAIR STETKAR: Pat, thanks. A couple 20 things.
21 First of all, in your introductory 22 remarks, you characterized the ACRS's role in this 23 project as providing a peer review. I personally take 24 issue with that.
25
20 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 The ACRS has not performed a peer-review 1
service. We do not delve into the depth of technical 2
details that is implied by a full-scope technical peer 3
review.
4 We focus on integration, we focus on 5
consistency, we focus on high-level issues related to 6
technical scope content, things like that.
7 But by no means do we perform a peer 8
review service in the way that those words are 9
typically understood throughout the industry, and 10 throughout the Staff.
11 So, I would appreciate it if you don't 12 characterize what we're doing as a peer review.
13 MS. SANTIAGO: Okay, so for Surry and 14 Peach Bottom, we did have a full, complex, external 15 peer review group.
16 And so for Sequoyah, based on knowing what 17 the comments were for that and addressing all of that, 18 we have guidance that says we can use or ask the ACRS 19 to support review of the next document, which was 20 Sequoyah.
21 So, I apologize for suggesting it's a peer 22 review.
23 It's an internal review, I guess, that we 24 would like to have a letter from the Committee, so 25
21 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 that we can accompany or send it to the Commission 1
about the same time as the report is going to the 2
Commission.
3 So, thank you.
4 MEMBER BLEY: I have to add a little bit 5
to what John said. We're giving the some kind of 6
oversight in this study as we did to the others.
7 To imply that we're providing any sort of 8
detailed technical review of the sort you'd had from 9
your peer reviews is just wrong, and we don't do that.
10 And you ought to not tell people that that's what ACRS 11 has done for you.
12 CHAIR STETKAR: Also, please don't 13 characterize this as an internal body. We are not 14 internal to the NRC, we are an independent advisory 15 committee.
16 MS. SANTIAGO: I forgot the right words.
17 CHAIR STETKAR: Are you going to discuss 18 the Executive Summary?
19 I didn't see your slides until this 20 morning, so I haven't had a chance to look through 21 them.
22 Are you going to discuss the Executive 23 Summary at all?
24 MS. SANTIAGO: We weren't planning on it.
25
22 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: Okay, let me ask one 1
general question and just get it out on the table, and 2
kind of think about it.
3 As I read the Executive Summary and a few 4
of the Chapters that have the new material in the 5
current version of the report, I found kind of a 6
different tenor from the previous version of the 7
study.
8 And the difference seemed to be more 9
emphasis on comparisons between the current version of 10 the study, and what I'll characterize as the April 11 2016 version of the study that we saw a year and a 12 half ago.
13 There's a lot of comparative stuff saying, 14 well, we had this in the April 2016 version, and look, 15 the same trends are here, but the results are 16 different.
17 That obviously had to be a conscious 18 decision because it's pervasive throughout the new 19 material.
20 Why do you feel it's necessary to compare 21 the results, insights, conclusions, models, and so 22 forth from this current version of the study, to an 23 outdated, non-technically-supported interim version of 24 the study?
25
23 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 I mean, when I've done projects in the 1
past and have made mistakes in the middle, I don't go 2
back and say, hey, I made a mistake but, look, you 3
know, this other stuff is all the same.
4 I just don't see the benefit of it. You 5
changed models, you changed uncertainty distributions; 6
those changes made a difference.
7 Why go back and say, well -- I mean, there 8
are statements that say, well, the uncertainty 9
distribution moved the results around but, look, the 10 general trends would have been the same.
11 That's obvious. You can present that same 12 level of insights and conclusions, just given the 13 results of the current study.
14 It's obvious to a reader that if a 15 pressurizer safety valve sticks open, big, early, it's 16 not a good day.
17 You don't need to compare that to the fact 18 that it used to stick open more, bigger, in a previous 19 version of the study that was found to have some 20 technical questions.
21 So, I'm not sure, because it was a 22 conscious decision. The only reason I bring it up is 23 why was that decision made to compare and contrast to 24 the previous version of the study?
25
24 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 Because to me, it was not only from a 1
philosophical perspective that I tried to mention 2
here, it was actually distracting.
3 MS. GHOSH: Okay, so I think we understand 4
your comment.
5 I mean, as to answering the why, I think 6
in part, we've put out this big study, it's publicly 7
available. It's going to be forever in Adams and --
8 CHAIR STETKAR: When you say this big 9
study, you mean the 2016 version of the study?
10 MS. GHOSH: Yes, because it was out there.
11 It was the first time we did this with the ice 12 condensers. A lot of people had looked at it.
13 So, maybe it was in recognition that 14 people might have been familiar with last year's 15 study. But I understand your comment and I think we 16 can take a look at --
17 MEMBER CORRADINI: I have a slightly 18 different opinion than my colleagues. It's your 19 study, you can do whatever you want. Then you'll get 20 criticized for it.
So, my view of it is 21 that John's saying if you were reading it today, this 22 would be his reaction.
And another way to do it 23 is to write the Executive Summary in such a manner 24 that nobody had seen anything before, and say, oh, by 25
25 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 the way, we had a draft version.
1 Go look at Appendix blah-blah and you'll 2
be able to compare and contrast in blah-blah, but 3
don't do it in the main body.
4 One, it detracts from what I assume you 5
think is better. And two, it would confuse the reader 6
that never took the time or the energy to read the one 7
before.
8 But that's your style, it's your report.
9 MEMBER BLEY: And it sounds very 10 defensive.
11 We would have certainly asked and we would 12 have expected in this presentation you would do those 13 sorts of things and show us what changed.
14 But I agree with John and, I think, with 15 Mike.
16 CHAIR STETKAR: Certainly, as I read it, 17 quite honestly, because I read the Executive Summary 18 first, it's the first thing you come to, and quite 19 honestly, it's what most folks will pay most attention 20 to.
21 So, I wanted to kind of see what flavor of 22 information was in there.
23 I found it very, very distracting in the 24 Executive Summary. Once I got past that, it is 25
26 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 infused in parts of Chapter 4 and in Chapter 7 in the 1
results.
2 There, because those are -- well, Chapter 3
7 is kind of summary of results, so that's a companion 4
to the Executive Summary. But Chapter 4 is more of a 5
technical detail thing and I was less confused by it 6
there.
7 But I think Mike's idea of pointing to an 8
Appendix that compares and contrasts it -- I 9
understand the desire to capture the differences and 10 draw insights from the differences.
11 But, anyway, you've heard our comments.
12 MS. SANTIAGO: That's a good comment, 13 thank you.
14 CHAIR STETKAR: And with that, I don't 15 know who's -- Tina?
16 MS. GHOSH: Trey.
17 CHAIR STETKAR: Trey, you're up.
18 DR. HATHAWAY: Okay, what I'm going to 19 present today is the third results of this sort of 20 Focus Study that we performed kind of after the full 21 UA --
22 CHAIR STETKAR: Is your mic on?
23 DR. HATHAWAY: I've got it here.
24 CHAIR STETKAR: You're just a very soft-25
27 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 spoken person.
1 DR. HATHAWAY: Sorry about that.
2 CHAIR STETKAR: Scream at the mic.
3 DR. HATHAWAY: So, the point of this was 4
to really try to reinforce the conclusions and the 5
study.
6 We knew we only had a very few number of 7
early containment failures so we wanted to kind of 8
focus in on this region of ventures to see if it 9
reinforced conclusions of the full UA.
10 So, that's what I'm going to present to 11 you today. Also, kind of see if it provided any 12 additional insights, focusing in on this region.
13 So, the Figure on the right is the results 14 of the 2016 UA, and we just kind of -- we were drawing 15 conclusions -- not drawing conclusions, we were trying 16 draw insights from these previous analyses, all the 17 analyses that were performed to really hone in on what 18 region of interest led to early containment failure.
19 And the reason we chose this is because we 20 did have -- well, I'm presenting this because we had 21 fewer early containment failures than the current 22 study.
23 So, what you see here, the Figure on the 24 right, plots the number, the total number of 25
28 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 pressurizer safety valve cycles plus the open area 1
fraction.
2 The points in blue represent the -- excuse 3
me, all the points represent the various individual 4
realizations, the Monte Carlo realizations. And the 5
points in red represent cases that had an early 6
containment failure.
7 And what you see is those sort of cases 8
that begets early containment failure kind of housed 9
in this region of cycles greater than 0.3 and less 10 than 65, is what was chosen.
11 Really, my particular interest was less 12 than hot leg failure, try to have a dent before hot 13 leg failure.
14 So, yes, again, we were trying to use the 15 previous studies to inform what we were going to look 16 at for this sort of focused look.
17 So, we took the current model and just 18 sort of used the identical distributions as before, 19 except for two.
20 We bounded the distribution for the open 21 area fraction to 30 percent, and then we also 22 constructed a distribution based off the sample data 23 from the previous study to try to focus in on this 24 range of 165 cycles.
25
29 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 For this focused look, originally, it was 1
for whether or not we had early containment failures.
2 So, we limited the calculation to 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br />, and that 3
was just really to -- again, this was informed by 4
previous studies.
5 Previous studies indicated that the early 6
containment failures were before 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br />. The model 7
was identical to the previous study.
8 The only difference is there was this 9
minor error in the study having to do with the fabric 10 seal pressure.
11 That was corrected; it was a formatting 12 error on a text file that was written so that was 13 corrected for this study.
14 MEMBER CORRADINI: So, I'm going to ask a 15 general question and you can address it whichever way 16 you want. But I've been looking through the new 17 version, and again, I could be missing it.
18 I'm just looking for data, test data, that 19 helps decide if this is realistic, optimistic, or 20 conservative.
21 I'm assuming you were trying to buy us at 22 conservative, but I'm looking for safety-valve test 23 data.
24 DR. HATHAWAY: While this particular study 25
30 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 wasn't looking at --
1 MEMBER CORRADINI: No, I know you weren't.
2 I'm just looking for a reference that tells me to go 3
look over here.
4 So, your distribution of fraction of valve 5
open after how many cycles bounds it, or represents 6
it, or whatever.
7 MS. GHOSH: So, from the perspective of 8
having the potential for early containment failure, 9
we're looking at a more conservative range for this 10 two-dimensional sample space.
11 Because that black box is basically what 12 we're sampling, and that only represents about four 13 percent, four or five percent, of the entire sample 14 distribution.
15 But it's in that four or five percent 16 where the early containment failure is possible.
17 So, we're artificially constraining it to 18 explore this area where you can get early containment 19 failure. But it's only a small percentage of the 20 total sample distribution, which is --
21 DR. HATHAWAY: Okay, yes.
22 Again, this was to really try to focus in 23 on the full study and just look at that region where 24 early containment was most probable to really see if 25
31 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 there were any additional insights.
1 CHAIR STETKAR: Trey, your slides, I read 2
through this, I see what you're trying to do and what 3
you did. But I got confused a bit.
4 Your slides don't have the Figures that 5
I'm going to talk about, and the Figures I'm going to 6
talk about right now are Figures I-2 and I-3 of the 7
report, that develop the values and the range of 8
values that you sampled from for this study.
9 I got confused by those because they do 10 not, to me, anyway, seem consistent with the numerical 11 values that are used in the main body of the study.
12 So, I'd like somebody to explain to me where I'm 13 wrong.
14 In particular, I look at your fifth-order 15 polynomial fit to the valve failure rate for the valve 16 data in Figure I-2.
17 And I noticed the first term is 3.26 times 18 10 to the -2, which, to me, ought to be the mean value 19 for the conditional probability that a valve sticks 20 open on the first demand.
21 If I'm wrong about that, please tell me 22 because that's kind of fundamental.
23 Because in the main body of the report, we 24 don't use that value. We use a value of 2.65e to the 25
32 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
-2, not 3.26. So, I'm curious why the first-order 1
term is 3.26 versus 2.65.
2 The next-order term for failure is per 3
demand. If I use just a linear model, it's 1.83e to 4
the -3, and in the main body of the report, we use a 5
value of 2.23e to the -3 for the demand.
6 So, the shape is -- the intercept at zero 7
is slightly different and the shape is somewhat 8
different.
9 So, somebody fit a fifth-order polynomial 10 to stuff that ostensibly uses data from the main body 11 of the report, and I can't draw that link.
12 Now, why is that important? Well, 13 overall, I'm not sure, but it's certainly going to 14 affect the number of samples in different regions.
15 The second -- let me get through the 16 second part of it.
17 The second part of it is when you do the 18 differentiation of that curve to get the density 19 function in Figure I-3, that density function does 20 indeed start out at 1.83e to the -3, at a value of 1.
21 Now, that's actually at a value of one-22 plus because if you differentiate the delta function 23 on the first demand, you get infinite. So, you can't 24 do that.
25
33 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 So, I want to make sure that the actual 1
sampling used the correct probability for failure on 2
the first demand, that it didn't use 1.83e to the -3, 3
that it used something between 2 and 3e to the -2 in 4
the actual runs that you made.
5 Why is that important? Because that's 6
going to force more earlier failures.
7 Certainly, a lot more earlier failures 8
than might be implied by the density function here in 9
I-3.
10 So, if somebody could confirm to me where 11 the numbers came from, and in particular, when you did 12 your sampling?
13 Give me assurance that you used something 14 on the order of two or three times ten to the -2 for 15 the probability that it fails on the first demand?
16 The first demand, number one, not 1.83e to 17 the -3.
18 I'd be a lot happier.
19 DR. HATHAWAY: I'll try to address your 20 comment, I'll take a look at the comments, I can look 21 at them.
22 But the distribution extended well past 23 what I fit, because I was interested in just this 24 range. I think that's what I said.
25
34 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 I was interested in the range of 0 to 250.
1 So, I was really trying to get a fit to the data in 2
that range.
3 The problem is, as you extend past that 4
range, the shape of the distribution is different. It 5
starts to curve.
6 CHAIR STETKAR: You're starting to talk 7
about somebody who really likes to do curve-fitting.
8 I'm asking you a more fundamental question. I'm 9
asking a question that your first term says 3.26e to 10 the -2.
11 In the main body of the report, there's a 12 table that says the mean value for failure on the 13 first demand is, and I've lost it again, --
14 MEMBER BLEY: It's about 3 --
15 CHAIR STETKAR: It's about 2.65. Now, 16 it's about 3, but why is the value different? And the 17 first term is also different.
18 So, why didn't you use the values that are 19 tabulated, and then you can do all of your fifth-order 20 polynomial curve-fitting, whatever you want to do, and 21 extrapolate out?
22 It's the same mathematical problem, it's 23 just a different curve, and I don't know why it has to 24 be different. It's the same lambda.
25
35 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER BLEY: You make a big, and I think 1
reasonable, case that it's a different failure rate 2
for the first actuation than for the rest?
3 CHAIR STETKAR: Yes.
4 MEMBER BLEY: So, you have --
5 CHAIR STETKAR: And that's what's used in 6
the study. That's what's used in the real study.
7 MEMBER BLEY: Yes.
8 CHAIR STETKAR: This is supposed to be 9
something that focuses on the first, nominally, 65 or 10 so demands, and tries to force samples within that 11 regime, that are consistent with the first study, 12 given the fact that the first study extends -- uses 13 all three valves.
14 I mean, the first study, the baseline 15 study if you want to call it, that is very elegant in 16 the way that it characterizes the valves and multiple 17 failures, and things like that.
18 But the intent of this shouldn't be to 19 develop different failure rates.
20 MS. GHOSH: So, I think we're going to 21 take your comment and re-check the numbers after this 22 meeting, because we're not going to be able to resolve 23 it right here.
24 CHAIR STETKAR: Yes.
25
36 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MS. GHOSH: So, we understand the comment.
1 The one thing I will say, I don't think we 2
should attribute too much precision to all of this, 3
because our main focus was to make sure we were within 4
this range of sampling where we could get early 5
containment failure.
6 So, in terms of the high-level insights 7
that we can draw from this study, we feel that we 8
accurately explored the two-dimensional sample space.
9 There may be some discrepancies between 10 the exact frequency and in parts of the sample space.
11 We'll have to go back and check that.
12 CHAIR STETKAR: Tina?
13 MS. GHOSH: Yes?
14 CHAIR STETKAR: I get it.
15 But you spend a lot of time in the 16 Appendix, like a couple of paragraphs, and this plot, 17 saying you fit a fifth-order polynomial because, well, 18 if you tried to fit a fourth-order polynomial, you had 19 a little bit of different glitches when you did the 20 differentiation.
21 So, you had to fit a fifth order. If you 22 wanted to use a simple model, why don't you use the 23 linear model? Failure on demand at time zero plus 24 lambda X. It would have been a straight line.
25
37 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 The differentiation is pretty good; it's 1
got the same value at every demand. That's a real 2
simple model.
3 It's not as elegant, it doesn't have this 4
nice curvy shape to it.
5 MS. GHOSH: I understand that.
6 CHAIR STETKAR: But it's pretty simple in 7
the sense of we're just trying to get an approximation 8
and focus on this range of what might be important in 9
that range.
10 MR. FULLER: Ed Fuller, from the Office of 11 Research.
12 I want to make a comment on what John was 13 saying because it was an issue that we encountered 14 years ago, when I worked at Pole Star, doing a 15 steam-generator tube integrity risk assessment for a 16 variety of plants.
17 And we looked at operational data and test 18 data at the time, to come up with some notion of 19 whether or not there was a distinct difference between 20 what would happen on the first demand versus all the 21 rest.
22 And we concluded that at the time, anyway, 23 most of the time when the data was pertaining to one 24 demand right away failure, it was due to some 25
38 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 maintenance errors.
1 And so I think, my own view is, take what 2
John said and refit, starting with Demand 2; exclude 3
the first demand from your Focus Study.
4 CHAIR STETKAR: The main report does that, 5
and it does it in -- I tend to be not the most 6
pleasant person in the world -- but in a very elegant 7
way, the main body of the report.
8 The way it treats three individual demands 9
-- three individual valves, sequential demands, 10 sampling of the number of cycles to failure, is a very 11 elegant model.
12 The only questions that I had on this 13 Focus Study was the values that are used to plot that 14 curve.
15 And then as a separate, non-related error, 16 a question to make sure that you did use that initial 17 demand failure for sampling for Demand 1 in the Focus 18 Study.
19 DR. HATHAWAY: I don't know if this will 20 clear this up or not --
21 CHAIR STETKAR: It's a factor of ten 22 higher.
23 DR. HATHAWAY: In the UA, it was two 24 distributions. It has one on the first sample and 25
39 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 then it has one on the second.
1 What I'm trying to do by using the data is 2
fit to that so it convolves those two distributions 3
together.
4 CHAIR STETKAR: But your curve I-2 shows 5
the concept of what's in the baseline study.
6 It shows a delta function at one, which 7
corresponds to the conditional probability that it 8
fails on the first time.
9 And then it shows some sort of shape in a 10 accumulative form as you accumulate demands.
11 Now, the shape of that curve kind of 12 depends on the model that you use, and the slope of 13 that curve depends on the failure rate that you use 14 per successive demand.
15 The same failure rate is used for each 16 successive after the first, in the baseline model.
17 And depending on -- it's not a linear 18 model if you account for the accumulative effects from 19 failures.
20 But in a simple term, you could use a 21 linear model, especially over the first 50, 60, 70 22 demands or so. This didn't.
23 And then when you differentiate this 24 curve, you'd better account for that delta function at 25
40 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 one.
1 And if you're not, you've got a real 2
problem because you're ignoring a factor of ten or so, 3
or more than a factor of ten, conditional probability 4
that it fails on the first demand, which is, indeed, 5
much more closer to the regime that you're trying to 6
test in this Focus Study.
7 You're trying to see how sensitive the 8
results are to earlier failures with a large open 9
area.
10 MEMBER BLEY: Just maybe Figure I-3, which 11 is the density function, was drawn without showing the 12 delta function either.
13 CHAIR STETKAR: That's the reason why I 14 ask it. It could be.
15 MEMBER BLEY: Maybe it's just the way you 16 drew the Figure, and you didn't show that three minus 17 two, right, at one.
18 That's kind of what John was hoping you 19 would say.
20 CHAIR STETKAR: Yes, I was really hoping 21 you would say, well, of course we used the three minus 22 two for the first demand. Or two and a half minus 23 two, or whatever it should have been.
24 But I didn't hear that.
25
41 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MS. GHOSH: If you go to Figure I-4.
1 CHAIR STETKAR: I-7?
2 MEMBER BLEY: I-4?
3 MS. GHOSH: Yes, if you have that.
4 It has the actual sample then successful -
5
- we were showing this for a different reason, to show 6
that there wasn't a huge dependence in terms of the 7
completed runs.
8 But at least to the first order of 9
magnitude, we did sample that first demand higher.
10 CHAIR STETKAR: You did? I'm trying to 11 pull it up right now.
12 MS. GHOSH: But I think in terms of the 13 larger question of the exact numbers, we'll have to 14 get back to you on that and we'll double-check.
15 MEMBER BLEY: Somehow, I'm not seeing it.
16 CHAIR STETKAR: Yes, I mean, if I had seen 17 it in that Figure, I would have probably mentioned it.
18 Yes, I'm not seeing it here.
19 MS. GHOSH: So, maybe we can talk during a 20 break. We should probably move on and we can get back 21 to you on that.
22 MEMBER BLEY: Are you sure you meant I-4?
23 Okay.
24 BB: I know there's a bunch of --
25
42 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: Turn your mic back on, 1
Tina. Don't --
2 MS. GHOSH: Sorry, the bottom right of 3
that six-figure -- there's a bunch of charts in that 4
Figure.
5 Yes, I apologize, I meant to release all 6
of them. Maybe we'll catch up later, and we can move 7
on. Yes? Yes, sorry.
8 MEMBER BLEY: Thanks.
9 DR. HATHAWAY: So, we ran the models, and 10 what we really looked at is when we constrained our 11 distributions to this small portion of the sample 12 space, where early containment failure was the most 13 probable, we got about that 17 percent of our 14 realizations resulted in early containment failure.
15 And looking at it, about 15 percent of the 16 realizations, the BOC core conditions had an early 17 containment failure of 16 percent of the MOC and 19 18 percent of the EOC.
19 So, the table on the right shows just the 20 statistics for the various realizations. The mean of 21 the BOC was about 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br />, the mean of the MOC was 22 about 6.6, and the EOC is about 6.6.
23 And what you also see is the BOC's kind of 24 constrained closer to the mean, where the MOC and the 25
43 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 EOC are spread out further.
1 And we'll see this a little later on when 2
we look at the cesium iodine release rates as a 3
function of time.
4 CHAIR STETKAR: Trey, I don't know where 5
to ask this so I'll ask it now, and tell me if I 6
should ask it later. Let me get my notes here.
7 See, you see this number 17 percent? 17 8
percent of the realizations in the Focus Study 9
resulted in early containment failure.
10 What you really mean is 17 percent of the 11 361 successful realizations had that behavior.
12 Out of 600 realizations, there were 239 13 that did not run to completion, which is 40 percent.
14 40, 4-0, percent did not run to completion when you 15 tried to force this.
16 It's curious, and I went back and I did 17 the numbers, that in the baseline study, 40 percent of 18 the runs in the area that had a large early open 19 fraction did not run to completion. Same percentage.
20 So, something about MELCOR in this region 21 is not very good. 40 percent of the runs are not 22 running to completion in this particular region.
23 A much higher percentage run to completion 24 everywhere else.
25
44 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER CORRADINI: But I thought we had 1
asked that.
2 CHAIR STETKAR: Yes, and we never got an 3
answer.
4 MEMBER BLEY: We talked about it a lot.
5 CHAIR STETKAR: We never got an answer.
6 MEMBER BLEY: The way it is is kind of --
7 CHAIR STETKAR: Yes, that's the way it is.
8 Yes, we looked at it and it doesn't finish. Now --
9 MEMBER CORRADINI: But I thought -- if I 10 just want to interrupt you, John -- I thought you guys 11 promised, I thought, that you were going to go back 12 and try to unravel the reasons for failure in that 13 area?
14 CHAIR STETKAR: That's why I brought it up 15 now, because I don't think they're planning to discuss 16 it.
17 The reason I bring it up now is that in 18 the results from this Focused Safety Study, they were 19 reflected back through the overall insights and 20 conclusions.
21 There is an inference that the same 17 22 percent early containment failure and the same 23 conditional consequences apply for that 40 percent of 24 the runs that did not finish.
25
45 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 And I do not know why that must be true. I 1
have no evidence --
2 MEMBER BLEY: Or at least one wonders --
3 CHAIR STETKAR: I don't know if it is 4
true.
5 MEMBER BLEY: -- if there's something 6
going on where there could be a lot more of them.
7 CHAIR STETKAR: I would have been really 8
happy if like 95 percent of these runs ran to 9
completion and you saw that same fraction, but they 10 didn't.
11 MEMBER BALLINGER: To turn it around, if 12 you have 40 percent, I think I made this comment 13 earlier.
14 If you have 40 percent that don't run to 15 completion, what's to say that the numbers that did 16 run to completion are actually good numbers?
17 I mean, if there's something fundamentally 18 going on here, the fact that you've got runs that went 19 to completion and you get numbers kind of makes you 20 feel good, but what if there's something fundamentally 21 wrong?
22 So, that's why I think we ask that you 23 folks go back and find out exactly why the runs that 24 didn't go to completion, what caused them to do that?
25
46 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 Was it some convergence issue? Or whether 1
it's some flat-out mistake in a model or something.
2 Because it does reflect back on these numbers.
3 MS. GHOSH: So, I'm going to just note 4
something that I --
5 MEMBER BALLINGER:
- Remember, I'm 6
metallurgist so...
7 MS. GHOSH: Yes, I'm not going to give a 8
satisfactory
- answer, but one of the primary 9
motivations to do this Focus Study, to do a bunch more 10 realizations in this area was that we knew from the 11 overall UA that we had 20-some, I think, realizations 12 that we had tried to sample in this area.
13 And we had a 30-some percentage failure 14 rate, which is much higher than the overall UA, which 15 is pretty small.
16 So, we knew that there was something going 17 on in this sample space, that we were getting higher 18 incompletion rates.
19 So, that was one of the major motivations 20 to do this study, is just to explore the response 21 surface and the smaller area much more.
22 So, we have 361 additional points in this 23 area, so we have more information but it doesn't 24 answer the second part of your question.
25
47 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER BLEY: But you didn't put any 1
diagnostics in the code or anything to see where it 2
was failing, what was going on?
3 MS. GHOSH: So, I'll let Casey, and if 4
Hossein wants to speak, speak next. We did have our 5
Code Guides.
6 We look at the reason for the failure in a 7
bunch of these cases, and in many of them, it was 8
frustrating.
9 I think we had documented some, or maybe 10 we didn't end up putting it in the Appendix --
11 CHAIR STETKAR: It is not documented 12 anywhere in the report, I can guarantee that.
13 MS. GHOSH: Okay, so that's probably 14 something we could add because at least we know what 15 the error meant. We know what the error message was, 16 and in some cases it's easily explainable.
17 I think some cases, things are just 18 happening too fast, the time steps are too small and 19 you get hung up.
20 But in a bunch of cases, it was this 21 executive abort signal detected, which is really 22 frustrating, because it's very hard to untangle.
23 But let me turn it over --
24 MEMBER BALLINGER: I'll say the same thing 25
48 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 as I did last time, and that is sometimes, exploring 1
why these things didn't go to completion provides you 2
with what I would call surprise, which actually means 3
something.
4 And so I may be a metallurgist, but if a 5
student comes to me with results from a model he's 6
written -- he or she, excuse me, has written -- and 7
says that 40 percent of the time, the thing didn't 8
converge, then I would reject the results out of hand.
9 I would say there's no reason for me to 10 believe that the ones that did run to completion were 11 not fortuitous.
12 MEMBER CORRADINI: So, you guys don't have 13 to try to explain more. My interpretation of what 14 you're saying is you don't know, even though you tried 15 to look.
16 So, if there is a documentation of however 17 many it is and the failure message, it would be, to me 18
-- and this is not a criticism because these are 19 highly non-linear calculations that go off the 20 reservation, potentially more easily, and the fact 21 that you had the small percentage of the bigger 22 population is a good thing.
23 But I do think you've got to do some sort 24 of postmortem as to why.
25
49 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER BLEY: I'd like to hear what Casey 1
has got to say.
2 MEMBER BALLINGER:
He's a
thermo-3 hydrologist guy so he gets away with more.
4 MR. WAGNER: Casey Wagner, Dycoda.
5 There's a couple of things that I have to point out.
6 This was already done and it used the same 7
code as the last calculations.
8 So, whatever problems we were running into 9
on the last set of calculations, we were going to run 10 into it on this set of calculations.
11 MEMBER BALLINGER: See, I rest my case 12 then.
13 MR. WAGNER: Well, more calculations were 14 done there.
15 It went to a small time step and 16 eventually, if you have enough time steps that are 17 very small, it would give up and say that it was not 18 able to converge. And so that was the stoppage.
19 The convergence criteria is there, I don't 20 know if this will satisfy you, to assure that you do 21 get convergence, and if you don't have convergence, it 22 keeps on cutting the time step, trying to get there.
23 If it doesn't get
- there, those 24 calculations are stopped, and we are considered 25
50 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 failed.
1 The other ones, presumably, did satisfy 2
that convergence criteria, and so we do accept them as 3
successful and reasonable calculations.
4 MEMBER BLEY: It just seems there's no --
5 since you don't know quite what's going on, I don't 6
see any basis for assuming that the same percentage of 7
containment failures are in this group that died as in 8
the other calcs.
9 There might be something that's happening 10 on the way to containment failure that's blowing up 11 the code.
12 MEMBER MARCH-LEUBA: Casey, can I ask you 13 a couple of questions?
14 You mentioned you're using the same code.
15 Do you mean the same version of the code, or the same 16 number?
17 MR. WAGNER: It is the exact same version.
18 MEMBER MARCH-LEUBA: And all these 19 variables didn't happen on initialization? It 20 happened in the middle.
21 It did not happen during initialization, 22 it happened in the middle of the transfer?
23 That typically happens when one of the 24 nodes goes into other reservations that we mentioned 25
51 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 before. Another's supposed to have water, it doesn't.
1 A little one goes through somewhere, and 2
that happens all the time in other codes. But if it's 3
water, somebody dies.
4 CHAIR STETKAR: When you say all the time, 5
it's, to me, knowing nothing about the codes, it's 6
just curious that in the baseline study, at least 7
according to what I understand from the report, they 8
only had 40 chances to get into this regime.
9 And out of those 40 chances, 17 didn't 10 run. Here, they forced it to get 600 chances to get 11 into this regime, and the same fraction of times 12 didn't run.
13 MEMBER MARCH-LEUBA: You give these codes 14 too much credit.
15 CHAIR STETKAR: But, okay, you folks are 16 familiar with it, but we are in fact relying on this 17 code to draw conclusions and insights about kind of a 18 fundamental behavior of this machine.
19 MEMBER MARCH-LEUBA: The experience of 20 people that run this code will tell you that they run 21 10 runs when it crashes for every run when it 22 survives. And you converge on the model and you fix 23 everything.
24 CHAIR STETKAR: But that wasn't the case.
25
52 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 In more of the well-behaved regime of this whole 1
study, they got something, I don't know, Casey, 90 2
percent success rate or better?
3 MEMBER MARCH-LEUBA: Eventually.
4 CHAIR STETKAR: No, not eventually.
5 MEMBER MARCH-LEUBA: No, when you're 6
trying to develop the input there, you're trying to 7
describe your plant, you try once.
8 You run it, it crashes. Then you fix it.
9 You run it, it crashes. And eventually, you fix it --
10 MEMBER BLEY: We kind of get that, but 11 what John's talking about is there's this whole space 12 they did calculations, 90 percent of the time, the 13 code runs great.
14 There's this little vulnerable area for 15 early containment failure. When they run in that 16 regime, 40 percent of the time it crashes.
17 So, there's something in there that's 18 causing trouble.
19 MEMBER MARCH-LEUBA: My point that I 20 obviously didn't do correctly is that's worrisome 21 because it happens in the middle of the run.
22 Something is happening.
23 MR. WAGNER: Yes, I can explain what's 24 happening to get it to make sense to you.
25
53 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 For those cases where the valve sticks 1
open, we're moving, we're generating more hydrogen, 2
we're going into early containment failures, we're 3
getting more core damage.
4 And then the accumulator dumps, and so 5
we're dumping an accumulator into a much more degraded 6
core than the cases that are high-pressure, that we 7
get the hotleg failure, and the core's relatively 8
intact, and it quenches it pretty quickly.
9 And then we move onto a boil-off severe 10 accident. This is much more challenging because the 11 core has formed a debris bed and then we did a hotleg 12 failure, or it seeps into the accumulator dump.
13 MEMBER CORRADINI: So, let's just backup.
14 So, I don't do this anymore because I probably 15 couldn't do it right. I couldn't even get the model 16 to run at times.
17 So, if I were to do a detective job on 18 this, the first question I ask is of the 40 percent 19 that failed, did they all fail at approximately the 20 same time in terms of event sequence?
21 You are hinting that the event of an 22 accumulator, essentially, hotleg creep rupture, and 23 then an accumulator dump, is a common time-to-receive 24 failure.
25
54 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 Is that the case?
1 CHAIR STETKAR: This is a step up in valve 2
but I just think --
3 MR. WAGNER: Yes, we also get the hotleg 4
failure, even if the primary system pressure has come 5
down.
6 The one thing that I don't think Trey 7
shared with you is that we did forensics because we 8
had so many more samples here on our uncertain 9
parameters.
10 And was there in space in them where we 11 didn't have representative successful samples?
12 And Trey did a very thorough job of going 13 through all the different parameters, and they're 14 interspersed. They're like this, failures and 15 successes. And so there wasn't a common theme there.
16 On our sample parameters, we were missing 17 a space.
18 We unfortunately got the 40 percent, which 19 is very
- high, but the interspersed end of 20 representation of what we were trying to look at for 21 uncertain parameters was very good.
22 And we judged that as success in a lot 23 more cases.
24 MEMBER CORRADINI: Okay, I get your point.
25
55 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 But what I'm asking, from the standpoint 1
of just rudimentary failure times, is there an event 2
that MELCOR records that at, or right after, that 3
event, everything goes to hell in a handbasket 4
regardless of the uncertain parameters or the variable 5
parameters.
6 MR. ESMAILI: This is Hossein Esmaili, can 7
I just say something? I think we don't know right now 8
what caused this error.
9 We can going back and look at it, and when 10 Larry was here on April 18, we went through the whole 11 thing. We said that this is our quench model, this is 12 how we do this, this is our comparisons with the 13 experiments, et cetera.
14 When you apply those models to 15 sophisticated whole-plan models, you can run into 16 whether it's on the core side or whether it's the 17 input model side.
18 We cannot answer this until we go to the 19 actual source code to try to find out what caused this 20 problem. But in most cases, we can resolve it.
21 It was not important, I guess as the panel 22 here discussed. It was not important because we had 23 an overall high success rate, and most of calculations 24 that we did was over 99 percent success rate.
25
56 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 It's just that we got into this issue 1
here. And we are going to find out exactly what 2
happened in these calculations, and maybe we can let 3
you know later.
4 Right now, I don't think we can tell you 5
with certainty what happened in these calculations.
6 It's often one thing, one thing in the code, that once 7
you get over it, the calculations would go forward.
8 But I don't know until I talk to my 9
colleagues about it.
10 MEMBER REMPE: And so your diagnostics 11 don't tell you it's in --
12 MR. ESMAILI: No.
13 MEMBER REMPE: It doesn't tell you which 14 nodes? It doesn't give you any clues? It just --
15 MR. ESMAILI: It gives you some idea but 16 it doesn't tell you everything about it.
17 So, you really have to go run the code in 18 a de-bug mode, find out what happened there, and find 19 out exactly what was going on.
20 And in some cases, it could be a small 21 error in how things are done, and --
22 MEMBER REMPE: And in the diagnostics that 23 you've had, you've not gone through and the 17 --
24 MR. ESMAILI: It doesn't --
25
57 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 (Simultaneous Speaking.)
1 MEMBER REMPE: And the 17 or however many 2
hundred cases, you've not ever tried to say, oh, it's 3
always in that same place?
4 You've not had the opportunity to --
5 MR. ESMAILI: I don't know if we have done 6
that. I don't know if we have put in the tracing 7
model back on to the reference, but we can go back and 8
look at it more.
9 I'm just saying that we cannot resolve it 10 here because we don't have the right people, we don't 11 have them here.
12 MEMBER BALLINGER: So, you're on record 13 now as saying you're going to run this to ground?
14 CHAIR STETKAR: They said they'd get back 15 to us. From an academic sense, not that I understand 16 any of this stuff, I'm really interested to understand 17 why the code is blowing up.
18 From a bigger picture regarding this study 19 and its report, the study and its report does not 20 highlight the conversation we just had.
21 It basically says look, look, look, we did 22 this Focus Study, which is a really good idea.
23 The results from the -- we focused the 24 sample space on a regime that we really wanted to 25
58 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 test, that was a) blowing up more frequently in the 1
baseline study, and b) generally more important for 2
offsite health consequences, early containment failure 3
anyway.
4 It has a good chance, but then we observe 5
the same behavior, that 40 percent of the runs didn't 6
run to completion. That's mentioned, it's documented, 7
but it's just sort of mentioned in passing.
8 And the thing that concerns me most from 9
the perspective of the study is the inference that 17 10 percent of that 40 percent of the runs that did not 11 complete would have behaved the same as the 60 percent 12 that did complete.
13 Ergo, yes, we might pick up two or three 14 more early containment failures from the runs that 15 didn't complete.
16 And therefore, our overall study results 17 and conclusions and insights aren't affected by those 18 run failures.
19 That latter point is the thing that 20 bothers me, and drawing that inference, I believe, is 21 misleading.
22 I think that you should just own up to the 23 fact that the runs didn't run to completion and you 24 don't really know what fraction of those incomplete 25
59 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 runs may have gone into early containment failure.
1 You know from the baseline study that it 2
can't be -- because I've got confidence in the 3
baseline study sampling -- you know that it can't be 4
more than I think 17 additional runs in the baseline 5
study.
6 So, it can't be 200 more samples in the 7
baseline study that would have gone into an early 8
containment failure.
9 But you don't really know what fraction of 10 those additional runs can go to early containment 11 failure, and I think the overall study should 12 highlight that.
13 It's just something we don't know right 14 now.
15 MS. GHOSH: Okay, so I think we understand 16 the comment. The one other thing I'll point out again 17 is this Figure 1-4, which is in the Appendix.
18 This was our attempt to, at least from a 19 statistical standpoint, to look at is there a region 20 of the inputs that we're putting in?
21 So, that's only just the initial 22 conditions. It doesn't get to what's happening 23 midstream at all.
24 But at least, was there a region of even 25
60 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 this focus input space where we were getting excessive 1
failures.
2 So, the I-4, the one I think that Trey 3
summarizes, is, once again, the one area where you see 4
a little bit of a correlation in terms of the 5
failures.
6 And we already knew this, because we knew 7
this from the leger studies, that if you have very few 8
cycles dealing with big, open areas, you continue to 9
be more likely to have an incompletion rate.
10 But if you look at the successful 11 realizations from those Figures which are shown in 12 red, the blues were the samples that were attempted, 13 and then the red ones are -- so the blue is the CDF of 14 what we attempted to do.
15 The red is the cumulative distribution 16 function of what actually succeeded in completing.
17 You can see that what Casey was saying, 18 there's very good coverage across the range of the 19 inputs, and for most of the input and variables, we 20 get a very good match.
21 But one exception, again, is there's a 22 little bit -- you can see a little bit of a difference 23 in the CDFs for the number of cycles, and that's kind 24 of pushed to success for a higher number of cycles.
25
61 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 So, you do see that same correlation.
1 But we, at least from a statistical 2
standpoint, not having had been able to do the full 3
forensics on these realizations, from a statistic 4
standpoint, we felt comfortable to say that we thought 5
it was likely that in the runs that didn't complete, 6
you would see the same percentage of runs going into 7
early containment failure.
8 So, again, from a purely statistical 9
standpoint with the input space --
10 MEMBER MARCH-LEUBA: And that's why I 11 wanted to get a little time, because I do know a 12 little bit about this topic.
13 Based on the comments we receive here, I 14 see likely unconcerned that the Staff is going to go 15 into a five-year effort into de-bagging all those 40 16 percent run.
17 I'm thinking, using engineering judgment 18 of the guys that know what's going on, can you look at 19 the run and it fail, but I see the passage going, this 20 is going to be okay.
21 And then I have to look at that one.
22 CHAIR STETKAR: That would be really good, 23 wouldn't it?
24 MEMBER MARCH-LEUBA: Our guys who have 25
62 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 looked at the other 362 can look at one and say, yes, 1
it crashed, but this isn't going anywhere.
2 Good. Don't go into a five-year effort.
3 CHAIR STETKAR: And, Tina, I'm sorry, but 4
don't muddy it up with statistics on sampled 5
parameters. Make it an engineering problem.
6 MEMBER MARCH-LEUBA: In fact, it would be 7
perfectly good for this exercise. And you know it, 8
you look at the run and say I'm wasting CPU.
9 CHAIR STETKAR: At 239 runs, 57 of them 10 had this flavor, 52 of them had this flavor.
11 MEMBER BLEY: That would be much more 12 convincing than just saying it's got to be the same 13 percentage.
14 CHAIR STETKAR: Okay, I think we beat that 15 horse. That's not a dead horse, it's beaten.
16 DR. HATHAWAY: So, I'm going to be doing 17 the next part. The next Figures present the CDF, so 18 the hydrogen-generated, because that was sort of that 19 interest.
20 Really, when I first started out, I was 21 just kind of interested in what was the early 22 containment failure potential in this region.
23 So, just looking at the data a little bit 24 more, we went in and pulled out the mass of hydrogen 25
63 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 generated up to the time of first deflagration, and 1
the mass of hydrogen reaching the dome up to the time 2
of the deflagration reaching the dome.
3 Because those tend to be the important 4
aspects of the -- that tends to be one of the 5
important insights from the study.
6 So, what this Figure shows, the dots in 7
blue represent the BOC realizations, the dots in 8
purple represent the MOC, and the dots in green 9
represent the EOC.
10 In these Figures, the dots that are 11 highlighted in red represent the early containment 12 rupture cases, and what you can see looking at 13 hydrogen generation, there was slightly more hydrogen 14 generated in the BOC realizations.
15 But the EOC and MOC realizations overlap 16 pretty well.
17 But then when you look at the amount 18 reaching the dome, that was consistent across all the 19 realizations, and also, what you see is when you have 20 a larger fraction of hydrogen reaching the dome, those 21 tended to have a higher fraction of the cases leading 22 to early containment failure.
23 MEMBER BROWN: Excuse me, for the 24 uninitiated, what do you mean by early? I looked 25
64 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 here, and you see the hydrogen buildup, is that over a 1
6-hour, 10-hour, 20, 1 day, period? After?
2 DR. HATHAWAY: No, it's within the first 3
15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br /> of what the simulation.
4 What I meant by early is if you have a 5
failure due to the hydrogen deflagration, not 6
overpressurization, which is lead failure.
7 MEMBER BROWN: Okay, so it's about 15, in 8
the ballpark of 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br /> for the hydrogen, for the 9
deflagration --
10 DR. HATHAWAY: So, if you looked at the 11 previous study, the times are presented on the right.
12 So, six up to ten hours is what was the result.
13 MEMBER BROWN: I couldn't connect the dots 14 after I listened for 30 minutes or 45 minutes on that 15 last interchange of why the runs didn't work.
16 So, I lost the ball on this one. So, 17 thank you, I appreciate it.
18 DR. HATHAWAY: -- the first 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br /> for 19 this part of the analysis.
20 So, the next thing we kind of looked at is 21 what is the pressure response in the dome?
22 So, what this Figure shows, the blue dots 23 represent realizations and the redundant Focus Study.
24 So, you can see that those have the early valve 25
65 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 failures on that range of 1 to 65.
1 The purple represent the results from the 2
full UA, and what this demonstrates is in the full UA, 3
the axis on the right represents the difference 4
between the peak dome pressure and the sample 5
fragility.
6 So, what that really shows you is how 7
close that early burn gets to actually rupturing the 8
containment due to the hydrogen burn.
9 So, you can see that in the full UA, a lot 10 of the results don't get within 25 PSI of the sample 11 fragility, so they had smaller burns initially.
12 But in the Focus UA, you kind of fill out 13 that region between the -- well, going back, a lot of 14 the full UA studies have a hotleg failures, what the 15 initiating event is first.
Or what the 16 depressurization then is.
17 So, when we were de-pressurizing before, 18 you would kind of fill out that Region between that 19 hotleg failure and the admission event.
20 And what you also see in the Uncertain 21 Figure is a lot of the burns were still -- when they 22 failed containment, it was when the sample fragility 23 was less than the mode of the distribution.
24 But also, just looking back, if you look 25
66 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 at the distribution, 42 percent of the fragility 1
distribution is above the mode, but only 13 percent of 2
the cases gave you a burn large enough to get into 3
that region of the distribution.
4 So, after we did this, and really, this 5
was after the last ACRS Meeting, we looked at it and I 6
tried to look at the consequences analysis.
7 So, what we did is we went back and 8
grabbed all the cases.
9 Since we truncated these cases to 15 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />, we went back and ran them all from the 11 beginning and extended the problem 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />, to really 12 be able to look at what the consequences of these 13 cases are.
14 So, we did this for two sets. We looked 15 at all the early-containment rupture cases, but we 16 also looked at a subset of the late-containment 17 rupture cases.
18 And we randomly selected late-containment 19 rupture cases but tried to keep the same breakdown as 20 the time cycle.
21 So, if there were 11 percent in the 22 sample, try to pick 11 percent of the BOC cases, for 23 example.
24 So, what this Figure is, you can kind of 25
67 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 see where the BOC realizations clustered around six 1
hours, but the MOC and the EOC kind of spread apart.
2 Also, these late-containment failures in 3
the UA, a lot of them were smaller releases.
4 But we kind of looked at this, this came 5
about very late, that some of these, when you had 6
valve failures, produced larger late releases.
7 So, that's why we selected a subset of 8
these late releases.
9 And this is just really showing the 10 results, the average cesium release for the early-11 containment ruptures was 2.2, and the iodine was 6.3.
12 The late-containment ruptures were 0.004 and 0.021.
13 MEMBER REMPE: So, can you go back to that 14 slide?
15 I was going to bring this up later, but if 16 you had done a sensitivity study where you assumed at 17 the beginning of the cycle you had a higher eutectic 18 melt relocation temperature, how do you think it would 19 have affected all these results?
20 I mean, everything always says, okay, the 21 cesium release fraction's kind of similar if it's all 22 at the beginning of cycle.
23 And, again, if you do kind of a thought 24 exercise on how all these things are interrelated in 25
68 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MELCOR, what would have happened?
1 Because you would have had a longer time 2
for the fuel to heat up, it would have gone to higher 3
temperatures, you would have had a lot of more 4
release.
5 And it would have really kind of thrown 6
things off kilter, and I was really disappointed that 7
you didn't do a sensitivity study.
8 I mean, you've got this 2500 plus or minus 9
83, instead of 200, which is what the Europeans 10 recommend.
11 And I kind of think that's an important 12 sensitivity technology, some of the limitations and 13 uncertainties associated with this plan.
14 And all plans, frankly. But any comments 15 on that?
16 MR. WAGNER: So, it's a timely comment 17 because we're getting ready to do Surry and we 18 struggled with your comment on how do we address it?
19 It makes sense that BOC is going to have a 20 higher melting temperature, and maybe that's something 21 we could look at in Surry, certainly at the 22 sensitivity at a minimum.
23 But I agree with your conclusions but the 24 releases would probably be a little bit higher at BOC 25
69 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 potentially because --
1 MEMBER REMPE: Hydrogen generation could 2
be higher. It could really kind of be an outlier 3
there, that it's not being explored.
4 And so, again, it's kind of like you get 5
up in the Executive Summary when you come up with 6
these general insights from these results.
7 I'd really like to see some limitations to 8
those insights, because of things like that, so that 9
you don't get yourself over a bind, in this case, 10 because there's some uncertainties.
11 In the case of the consequential tube 12 rupture, your scope of your study is bringing you to a 13 certain conclusion.
14 Whereas, if you consider other ongoing 15 studies even, there's some knowledge that you guys 16 have that ought to be acknowledged as a limitation.
17 And so that's why I kind of wanted to 18 bring it up here because --
19 MR. WAGNER: To your comments there, maybe 20 our standard deviation about that is too narrow. The 21 inverse is maybe 200 and we really ought to go back 22 and rethink that BOC again.
23 MEMBER REMPE: Yes, at this point, I'm not 24 trying to say do more calculations, although, I would 25
70 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 like to see what that result looks like because of the 1
way the models have been constructed.
2 But just to acknowledge the limitations 3
and uncertainties.
4 MEMBER POWERS: I've tried diligently not 5
to make comments during this, but I can't anymore.
6 Do you really, really honestly think that 7
melt relocation depends at all on how long the fuel's 8
been burned up?
9 MR. WAGNER: I'd be looking for Randy at 10 this moment. I thought so.
11 MEMBER POWERS: Do you have any solid 12 evidence that the accumulation-efficient products as 13 the fuel burns up changes the melting point of the 14 fuel significantly?
15 MR. WAGNER: That's why we weren't able to 16 address Joy's comment. We said we agreed with the 17 hypothesis but we didn't know, we didn't have data to 18 19 MEMBER POWERS: Well, there are data in 20 the literature.
21 There are several people who have tried to 22 measure the melting point of irradiated fuel, and they 23 usually come back with results that are not wildly 24 different from the melting point of unburnt fuel.
25
71 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 The fact of the matter is that in that 1
particular range for LWR fuel, there's sufficient 2
uncertainty and variability just by small changes in 3
straight geometry and whatnot, that it's very hard to 4
say.
5 But the overall phenomenon of melt 6
relocation really doesn't depend very much on the 7
melting point of the fuel. It really has to do with 8
liquefaction with the clad interacting with the fuel.
9 And it seems to me that if you were to 10 hypothesize burn-up has an effect, it would be because 11 of something happening on the inside of the cladding 12 to give you a larger region of oxide formation there, 13 that inhibited liquefaction.
14 It would have had nothing really to do 15 with burn-up, but it had to do with internal oxidation 16 of the cladding.
17 MEMBER BALLINGER: But that is a function 18 of the lambda, right?
19 MEMBER POWERS: It can be depending on 20 what's happening on the inside --
21 MEMBER BALLINGER: So, the film on the 22 inside of the cladding is going to be a function of 23 the potential?
24 MEMBER POWERS: It depends on how the fuel 25
72 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 chooses to behave.
1 And in particular, what it seems to depend 2
most upon is how much of metallic modules you'd get 3
in.
4 Because the molybdenum acts as a buffer 5
for the accumulation of oxygen, and the availability 6
of the oxygen to go oxidize that cladding.
7 So, it depends on a lot of things, but not 8
which are sensitive enough, since the code works in 9
nodes that are like this, as opposed to individual 10 fuel rods, you're averaging over a lot of things here 11 to get liquefaction.
12 And liquefaction is treated in the code in 13 a relatively non-mechanistic fashion.
14 So, you do what they've done.
15 It's they take some breadth of that 16 relocation number based on matching a variety of 17 experiments, and they say, okay, you match these 18 experiments and we take some temperature range as a 19 criterion for relocation.
20 Honing that number down based on some 21 hypothesis about what's going on strikes me as kind of 22 futile.
23 MEMBER BALLINGER: They say about the 24 great tragedy of science, the slaying of a beautiful 25
73 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 hypothesis by an ugly fact.
1 MEMBER REMPE: So, there's also thermal 2
cracking, there is cracking during the radiation. But 3
the thing is that --
4 MEMBER POWERS: Joy, what happens --
5 MEMBER REMPE: Let me finish just for a 6
minute, sir.
7 There's uncertainty with the phenomena but 8
there's also what's embedded in the code, and because, 9
again, there's uncertainty in the phenomena, I think 10 it's a useful exercise to bump that thing up a bit.
11 Because, again, they only had plus or 12 minus 83 degrees, and I've been reading up a lot of 13 stuff from France, where they're saying there needs to 14 be a larger uncertainty.
15 But just to see how it's all connected, 16 because I would challenge or query that I think that 17 if you did bump it up, you might see that because of 18 the way the models are interspersed in that code, 19 you're going to see some significant differences.
20 MEMBER POWERS: And presumably, if I take 21 the ranges large enough, I will get some pretty 22 radical differences. I'm sure of that.
23 MEMBER REMPE: Yes.
24 MEMBER POWERS: That one I'm confident of.
25
74 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER REMPE: And I would like to see the 1
sensitivity to explore that because they're getting 2
some very similar things.
3 And I think it would be worthwhile to 4
understand if it's realistic if you go a bit higher, 5
even if your range is up --
6 MEMBER POWERS: I guess I don't understand 7
how increasing your understanding improves your --
8 increasing your uncertainty improves your 9
understanding.
10 A little bit of a mystery to me but I'm --
11 MEMBER REMPE: But, again, they've limited 12 it to VERCORS for their plus and minus 83, and if you 13 consider other things like PHEBUS, the French are 14 saying go to plus or minus 200.
15 MS. GHOSH: Can I just clarify --
16 MEMBER POWERS: I think they're centered 17 around PHEBUS.
18 MEMBER REMPE: The report indicates that 19 they've centered a lot on VERCORS.
20 MS. GHOSH: That's true, but I do want to 21 clarify --
22 MEMBER POWERS: VERCORS didn't involve any 23 relocation at all.
24 MEMBER REMPE: It's their eutectic melt 25
75 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 temperature, and that's what they based it on.
1 MS. GHOSH: I just want to clarify that 83 2
is the standard deviation.
3 MEMBER REMPE: Right.
4 MS. GHOSH: So, our bound's actually go up 5
to 200-plus, up and above. You just have less samples 6
in that area.
7 MEMBER REMPE: Right. But I think an 8
uncertainty would be worthwhile to explore.
9 MEMBER POWERS: I don't understand at all 10 now. If the sigma is 83, then 3 sigma is huge.
11 MEMBER REMPE: I don't think we ever got 12 up to that Figure. What was the little red dot?
13 Was it 2800 or something, and you never 14 had any sampling up there at all?
15 MEMBER POWERS: The 3 sigma would...
16 CHAIR STETKAR: But again --
17 MEMBER POWERS:...violate the --
18 CHAIR STETKAR: Please.
19 MEMBER POWERS:
annihilate the 20 molybdenum issue, annihilate everything.
21 CHAIR STETKAR: This is an Uncertainty 22 Analysis, and if believe the uncertainty distribution, 23 if you get a small number of samples out in that tail, 24 you get a small number of samples out in that tail.
25
76 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 As you should, if you believe the 1
distribution.
2 MEMBER REMPE: But then the conclusions on 3
the Regression Analysis are that the temperature is 4
very important, that it affects hydrogen, and 5
eventually, it becomes pretty important in the follow-6 on documentation.
7 And that's why I am emphasizing it.
8 MEMBER POWERS: I guess I'm still at a 9
loss that you concluded it's important.
10 MEMBER REMPE: But the models are --
11 CHAIR STETKAR: Well, it is important.
12 Tina, I don't remember your Regression Analysis. I 13 mean, it showed up in the mix.
14 Was it very important?
15 MS. GHOSH: The eutectic melt temperature 16 is clearly very important for hydrogen generation.
17 We've see that in all the studies.
18 It shows up in combination with other 19 variables for things like cesium release.
20 So, it's got high interaction effects, 21 again, for obvious reasons, because it has a big role 22 to play in the hydrogen generation, which then affects 23 early containment failure, et cetera, et cetera.
24 MEMBER POWERS: Also, I would think that 25
77 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 the analyses of transport wouldn't mediate the 1
importance of fuel relocation on cesium release to the 2
containment.
3 Just because cesium escapes from the fuel 4
does not ipso facto mean that it goes into the 5
containment.
6 There are a lot of slips towards the cup 7
in the lip on that, and the uncertainties inherent in 8
that transport process might work to mitigate the 9
importance of fuel relocation.
10 Fuel relocation is very important for 11 hydrogen because you cook things for a long time, and 12 that does give you high cesium releases from the fuel, 13 but doesn't necessarily affect the transport process.
14 So, it doesn't lead -- there's not a one-15 to-one correspondence between every location, and 16 cesium release, just because of the transport process.
17 And quite frankly, you're putting a lot of 18 heat in the fuel and you haven't got heat to put into 19 the piping system. So, you end up with a cool piping 20 system which leads to high deposition rates.
21 MR. WAGNER: Our volatiles are pretty 22 close to 100 percent release.
23 MEMBER POWERS: You can't get --
24 (Simultaneous Speaking.)
25
78 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER POWERS: That's released from the 1
fuel, but it's not released into the containment.
2 MR. WAGNER: Right.
3 So, over most of these cases, we do get 4
the volatiles out, but maybe I misspoke that our 5
releases might change a lot.
6 Our hydrogen could change and that has 7
dramatic effects, but maybe not so much what gets off 8
the fuel.
9 The release coefficients that we have from 10 the fuel, but most of the volatiles come off.
11 MEMBER REMPE: I'll have to look where I 12 saw it, but I think there's a discussion somewhere on 13 the report that does kind of have a different 14 perspective.
15 But the higher relocation temperatures 16 would increase the releases from the fuels somewhere.
17 And I'd have to find where it is, but that may be 18 something to look for.
19 DR. HATHAWAY: Okay, so I'm just sort of 20 moving to the Consequence Analysis.
21 We took all of the source terms that were 22 run to 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />, and we put them in the uncertain 23 MACCS model of the site to see what the latent cancer 24 fatality risk for the zero to ten-mile interval.
25
79 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 And again, as in the full study, these are 1
conditional on the occurrence of the assumed --
2 CHAIR STETKAR: Trey, I want to make sure 3
I understand both of these tables, but 11 is the one I 4
want to understand.
5 These results, to me, mirror the results 6
in the baseline study, which is encouraging in the 7
sense that you're showing a bimodal effect on the 8
conditional risk, the way the dots on the lower left 9
are clustered.
10 And the numerical ranges are similar.
11 The difference is that the first column in 12 11 that you called really Containment Rupture is a 13 subset of the scenarios in Figure -- I'll give you the 14 Figure number -- 6.1 and 6.3 -- so that the right-hand 15 side of those Figures, the 87 percent that goes to 16 early containment failure.
17 The first column in this table is a subset 18 of them, right?
19 The second column, the thing that's called 20 late containment rupture, is a combination of so-21 called late containment failures that are also in the 22 right-hand side of those figures in Chapter 6.
23 And the small set that never go to 24 containment failure which is the far left-hand side, 25
80 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 where the risks are really, really small, the other 1
mode of that bimodal behavior.
2 And therefore, comparing the numbers in 3
this table to numbers that appear in other sections of 4
the report is a little bit misleading. I mean, you 5
really have to kind of look through that stuff.
6 I was surprised that you didn't cast these 7
results in the same way as the results in Chapter 6, 8
so you could see that, yes, when we turned up the 9
microscope in this area, we basically got the same 10 shape, we got the same fractions in the bimodal, and 11 roughly the same ranges in the conditional latent 12 cancer fatality risk, coming through the zero to ten-13 mile snapshot here.
14 You can kind of see it from this, but you 15 really have to, I think you really have to, study it.
16 Have I mischaracterized anything, Tina? I 17 mean, you're really familiar with all of those curves.
18 MS. GHOSH: I think it's a good comment 19 and we can supplement the --
20 CHAIR STETKAR: I mean, basically, on the 21 Figure on the lower left, if I take a horizontal slice 22 in that Figure at about 1e to the -6, everything above 23 that horizontal slice, is what's on I'll call it the 24 right-hand side of the Figures in Chapter 6.
25
81 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 Everything below the horizontal slice is on the 1
left-hand side of those Figures approximately, right?
2 I just want to make sure that I got it.
3 MS. GHOSH: Yes, I think that's right.
4 These results are spread out a little bit more.
5 CHAIR STETKAR: Oh, yes, they're going to 6
be spread out a little bit just because what you did 7
here. But in the big-picture concept --
8 MS. GHOSH: Yes, I think that's a good 9
idea. We can add a paragraph to compare back to the -
10 11 CHAIR STETKAR: Because I like what you 12 did in Chapter 6, by the way.
13 We talked about that bimodal behavior, and 14 then you split it out to more explicitly show -- and I 15 really liked that, that kind of -- emphasize some of 16 the points you were trying to make in the text.
17 And you can draw the same conclusions from 18 this but it's a lot more difficult. I mean, you have 19 to really study this. Okay.
20 DR. HATHAWAY: Okay, I probably won't 21 dwell much on this.
22 The top table just is the sort of 23 statistics on the total risk where the bottom is sort 24 of statistics but looking at the emergency-phase 25
82 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 contribution, and the long-term phase and 1
intermediate-phase contribution individually.
2 So, finally, because we had more of these 3
early containment ruptures, we were interested in the 4
early fatality risk.
5 Previously, we only had four realizations 6
that had early fatality. So, we just looked at these 7
results. Past four miles, there was zero calculated 8
early fatality risk.
9 And here are just the statistics, and 10 again, this is a conditional long-term event 11 occurring.
12 And what we tried to do, and this sort of 13 all is dependent on the discussions from earlier, but 14 we tried to scale these results to consider the full 15 distribution because everything outside of this 16 parameter space would be to zero.
17 So, we tried to just scale it out to 18 consider the entire distribution, and that's what that 19 second column is.
20 It would be what the risk would 21 potentially be if you were to essentially add a lot of 22 cases that had late containment.
23 CHAIR STETKAR: A comment on Chapter I 24 think it's 6, where these similar tabulations are 25
83 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 shown.
1 It would help a lot of people to provide a 2
little bit more of a tutorial in that section about 3
how one can have a mean value that's something on the 4
order of ten to the minus, I don't know, small number, 5
eight, or whatever those numbers are, nine.
6 And yet, have precisely zero values for 7
the median 5th and 95th percentile.
8 The only reason I dwell on that is that 9
people who use very broad and very skewed 10 distributions routinely understand that.
11 Not necessarily everyone who reads this 12 study will have that level of understanding, and a 13 little bit of a tutorial will help to those folks to 14 not get the impression that this is simply some sort 15 of statistical mumbo-jumbo that has no basis in 16 reality, kind of anchoring them back to say that, yes, 17 I can get 95 percent of my samples to have precisely 18 zero, and yet, the mean value can still be non-zero.
19 And show them why that is.
20 Otherwise, the naysayers will say, well, 21 obviously, this is just statistics and numbers and 22 numbers games.
23 So, I think that would help in Chapter 6.
24 Here also, but here, at least, most of them, the 95th 25
84 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 percentile is above the mean, and that's reassuring to 1
a lot of people.
2 DR. HATHAWAY: So, that was all I was 3
going to cover in this.
4 CHAIR STETKAR: We're at a break time 5
here. No, any more questions, first of all, on 6
Appendix I? Because we're going to shift gears here 7
entirely.
8 If not, let's take a break, and I'm going 9
to try to hold us to ten minutes. I'll probably fail 10 but try to get back here by 10:20; we'll recess until 11 then.
12 (Whereupon, the above-entitled matter went 13 off the record at 10:08 a.m. and resumed at 10:21 14 a.m.)
15 CHAIR STETKAR: We're back in session. We 16 are back in session. Trey, Tina, I don't know, it 17 says Casey. Anybody?
18 MEMBER BLEY: I don't know when to ask the 19
-- thank you. I don't know when to ask this, but I'm 20 going to ask it now, as we move into more general 21 things. Sure.
22 (Laughter.)
23 MEMBER BLEY: But it's kind of the same 24 cast of characters here and who we meet in the Level 3 25
85 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 PRA work. And over in the Level 3 PRA work, I think 1
we've learned that some of the things you're doing in 2
SOARCA aren't making its way over there, because it's 3
not state of the practice.
4 But something comes up over there, I'm 5
curious about it, and over there, in our recent review 6
of the Level 2 PRA, I find a whole section of their 7
report on model uncertainties that's really pretty 8
thoroughly laid out, at least the thinking of things 9
that might not work the way we expect them to work and 10 ways that they might consider them in the report.
11 I don't find anything quite as thorough, I 12 mean, you do a tremendous job on the parametric 13 uncertainties, but a few sensitivity studies that are 14 kind of picked here and there, but nothing as 15 systematically thorough, of saying, where could there 16 be things about the model that aren't right, that I 17 see in this new part of the Level 3 PRA.
18 Have you thought about that? And probably 19 the work you've done here led to how that evolved, but 20 I would sure like to see it here. It seems that part 21 of the uncertainty analysis that we've always kind of 22 talked about and never gone very far on over here.
23 MR.
FULLER:
Yes.
So,
- Dennis, 24 unfortunately, Don and Hossein, who were, along with 25
86 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 myself --
1 MEMBER BLEY: They were here all morning.
2 MR. FULLER: They just walked in and they 3
missed your whole question.
4 (Laughter.)
5 MEMBER BLEY: You want them to handle it, 6
you don't want to handle it.
7 MR. FULLER: So, what I will say is this, 8
the team that worked on that document and that work, 9
obviously Don Helton is the lead for all of that.
10 MEMBER BLEY: Right.
11 MR. FULLER: It's the same team, so you 12 could say that --
13 MEMBER BLEY: It kind of seemed that way to 14 me.
15 MR. FULLER: Yes. So, the portion of the 16 SOARCA team that's been thinking about, we kind of 17 focused our efforts in supporting Don in the Level 3 18 PRA study for that more systematic thinking. And, of 19 course, the scope is much broader for the Level 3 PRA.
20 MEMBER BLEY: It is, but --
21 MR. FULLER: Yes.
22 MEMBER BLEY: -- SOARCA makes a claim that, 23 one, it finally acknowledges that it isn't a risk 24 assessment and you pick out particular scenarios. But 25
87 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 then you do a thorough consequence analysis and now, 1
you've added to that a thorough uncertainty analysis.
2 And, in fact, the parametric uncertainties 3
affect the results substantially, and we see that.
4 But the potential of the modeling uncertainty is to 5
overwhelm what we see in the parametric uncertainties 6
has always been an issue and we've talked about it 7
occasionally.
8 But this new part of the Level 3 PRA on 9
model uncertainties and the Level 2 analysis is the 10 first attempt at being thorough in a qualitative 11 examination of the things that might affect it.
12 It seems to me that at least some 13 acknowledgment of that approach and that it hasn't 14 been done in the SOARCA analysis would add to the 15 comfort level of people who have been uncomfortable 16 with SOARCA to this point.
17 MR. FULLER: I think -- thank you for that 18 comment. I think it's a good comment and we can 19 certainly add pointers to that work in our executive 20 summary and the introduction and when we talk about 21 the scope of our work being limited to the parameter 22 uncertainty.
23 We can certainly refer to that work as an 24 example of more thorough, at least qualitative 25
88 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 thinking through of all the modeling uncertainties 1
that we're not explicitly addressing.
2 MEMBER BLEY: And that's the first step.
3 Eventually, one would like to be quantitative about 4
it, but in SOARCA, we haven't even been qualitative 5
about it, in a thorough way.
6 MR. FULLER: Okay. Yes, thanks for the 7
comment. And I see Don is at the mic.
8 MR. HELTON: I just want to -- Don Helton, 9
Office of Nuclear Regulatory Research. I do just want 10 to point out, though, that that work is non-public and 11 will remain non-public for, at least the current plan, 12 well after this document would be finalized. So, 13 there is --
14 MEMBER BLEY: The engineering thinking --
15 MR. HELTON: -- a challenge there.
16 MEMBER BLEY: -- beyond it has been around 17 a long time. The application hasn't. And --
18 MR. HELTON: Yes, I mean, it can certainly 19 20 MEMBER BLEY: -- maybe I shouldn't have 21 mentioned it, because that is not --
22 MR. HELTON: No, I'm not worried about you 23 mentioning it and it can be described from a 24 conceptual standpoint as what was done.
25
89 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER BLEY: That's what I'm after.
1 MR. HELTON: I'm just reacting to the idea 2
that it's --
3 MEMBER BLEY: To have a pointer to it.
4 MR. HELTON: -- it can be cross-referenced 5
6 MEMBER BLEY: Yes.
7 MR. HELTON: -- and that just presents some 8
challenges, that I'm sure we can work through. But --
9 I don't know.
10 MR. WAGNER: Casey Wagner, Dycoda. So, I'm 11 going to go through, not all, but we thoroughly went 12 through the comments from the last meeting and I'm 13 going to address some of the highlights of the more 14 important ones and how we addressed them.
15 So, we're on the first slide here. First 16 one had to do with the fabric seal failure criteria.
17 And we had a range of different types of comments that 18 were relatively extensive discussion on that.
19 And they concerned the error itself, a 20 confusing description of the failure parameter, and 21 then, there was also some questions about how we were 22 handling the seal failure error. We tried to address 23 these by adding new documentation to the next version 24 of the report.
25
90 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 And in particular, we did a little bit 1
more work on the impact of the fabric seal on early 2
containment failure. And added some new text in 3
there. In particular, the flow through the ice chest 4
and the other leakage pathways allowed hydrogen to 5
move around, whether that seal had failed or not.
6 So, there were other mechanisms for, over 7
the time frames that we were looking at, that hydrogen 8
could be moving throughout the containment. So, 9
certainly, if the seal had failed, which with the 10 error introduced, it failing more early, it flowed a 11 little bit more thoroughly.
12 But nevertheless, there was other leakage 13 paths that allowed that to occur. And so, there's 14 some discussion about that. In particular, the 15 focused value study had many cases and in all those 16 calculations, the fabric seal had been corrected.
17 And so, we looked at the response of those 18 and tried to compare that to the UA calculations and 19 drew conclusions that the responses were similar on 20 key metrics.
21 And then, finally, Hossein had presented 22 last meeting a comparison to the 2016 UA results, 23 which did have the correct -- well, it had a different 24 distribution on the fabric seal failure, but it was 25
91 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 not prone to failing early. And that discussion is 1
also in there.
2 CHAIR STETKAR: Casey, when I read that, 3
and I -- the discussion of this issue, by the way, is 4
much, much improved. But I know nothing about the 5
physics, but I thought that in the current study, when 6
you looked at the limited cases that you ran for long-7 term, that you saw some -- I think a couple of those 8
kind of transitioned to an early containment failure.
9 So, the question that I had is, how 10 significant is that in the context of the full study?
11 I mean, what fraction -- if you had the corrected 12 model in all 600 realizations in the real study, do 13 you expect a similar number to transition to early 14 containment failure?
15 Because we're not finding many early 16 containment failures, so, like two or three can double 17 the number that we've seen. And I think I understand 18 why that's happening, but I'm trying to get a sense of 19 looking at the full spectrum of the baseline study.
20 You only ran a small number of these long-21 term or late containment failure scenarios, right? In 22 the --
23 MR. WAGNER: Yes.
24 CHAIR STETKAR: -- focus study, there's --
25
92 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. WAGNER: Yes.
1 CHAIR STETKAR: -- about 18 or 20 or 2
something like that, that you ran, to examine it.
3 MR. WAGNER: Right. If you combine that 4
with the 2016 calculations that had a more robust seal 5
6 CHAIR STETKAR: But a different --
7 MR. WAGNER: Yes, it's --
8 CHAIR STETKAR: The reason why I dislike 9
this notion of the comparisons is that the uncertainty 10 difference -- the differences on the uncertainty 11 distributions make a difference.
12 MR. WAGNER: They do, yes.
13 CHAIR STETKAR: So, you have to kind of 14 focus in the regime of the 2016 study, where --
15 MR. WAGNER: This was about a -- in the 16 couple cases I looked at, it was about a two to three 17 PSI difference if the seal was there during a burn 18 that was -- the ones that flipped over --
19 CHAIR STETKAR: Before the --
20 MR. WAGNER: -- that were just below the 21 failure pressure and if the seal was intact, it was 22 just above.
23 CHAIR STETKAR: Okay.
24 MR. WAGNER: And so, you had to be very 25
93 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 close. So, we judged --
1 CHAIR STETKAR: Yes, but I was going to 2
say, but there are two. If I ran an infinite number 3
of samples in the baseline study, such that you had 4
the full distribution for everything, I'm trying to 5
pulse what fraction, a guesstimate. Is it one 6
percent, ten percent --
7 MR. WAGNER: Okay.
8 CHAIR STETKAR: -- half of a percent might 9
transition?
10 MR. WAGNER: I certainly can't answer --
11 CHAIR STETKAR: Yes, okay.
12 MR. WAGNER: -- the percentage, but this 13 plot might be kind of useful. So, we're looking at --
14 and unfortunately, the blue ones have the corrected 15 seal behavior --
16 CHAIR STETKAR: Yes.
17 MR. WAGNER: -- versus the purple. But the 18 purple ones that are within a couple PSI could have 19 flipped over. So, I see maybe two there.
20 CHAIR STETKAR: And --
21 MR. WAGNER: I think that --
22 CHAIR STETKAR: But, again, they're -- when 23 you say within a couple PSI, you're still limited by 24 the number of samples from both the failure pressure 25
94 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 distribution.
1 And, if you're not hitting many samples, 2
that could either be fortuitous or -- I mean, you 3
would expect those samples to be somewhere in the 4
middle of the distribution, if there's only a small 5
number of samples, right?
6 MR. FULLER: Yes, so, I guess just --
7 CHAIR STETKAR: The containment failure 8
probability --
9 MR. FULLER: Right.
10 CHAIR STETKAR: -- distribution is what I'm 11 talking about.
12 MR. FULLER: Right. So, if we get back to 13
-- so, the focused study was looking at the range 14 where you could have an early containment failure.
15 So, you saw in 80-some percentage, it still went to 16 the late containment failure versus the early. And 17 what gives us some confidence -- and all of those have 18 the fabric seal failure criteria fixed.
19 So, what gives us some confidence is that, 20 in those realizations, the 360-something, with the 21 fixed seal pressure, we still saw the same, roughly 22 the same percentage of those cases where it's 23 possible, actually going to early containment failure.
24 So, that's what -- that's kind of where we're --
25
95 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: That might, yes, that might 1
be reassuring. I have to think about it that way, 2
thanks. Thanks.
3 MR. WAGNER: I verbally gave you some 4
information on late containment failure, because in 5
the last set of documentation in June, you didn't have 6
that and so, I documented that.
7 The Committee expressed some confusion on 8
this parameter in general, because there was a thermal 9
part and then, there was sampling. And so, that was 10 rewritten and hopefully, that's much clearer now. And 11 12 CHAIR STETKAR: On the record, it is.
13 MR. WAGNER: Great, great. It was a good 14 comment and I'm glad we could improve the 15 documentation. And then, there was also some 16 questions about the seal failure area related to that 17 comment, because we have some where it's sort of a 18 cold mechanical failure and then, some where there's a 19 thermal presence that's maybe more widespread and 20 maybe that should have a different failure area.
21 And we hadn't thought about that and that 22 was -- it was good that it was pointed out. It wasn't 23 a controlling factor on flow, if you look at the K 24 over A-squared and you look at the thermal hydraulic 25
96 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 resistance, even the small holes that we were sampling 1
allowed a pretty robust natural circulation to occur.
2 So, we concurred that that probably was 3
likely, that there's maybe a wider, larger area that 4
we should use, but it wouldn't impact the results, was 5
my engineering judgment.
6 And then, if you look at the seal, it 7
winds around the nooks and some of it's well-shielded 8
from thermal radiation effects. And so, depending on 9
where the fireball was, it -- there certainly is some 10 uncertainty on what parts would be affected.
11 Next slide. The ice condenser doors, and 12 we're going to call this AJAR parameter. The AJAR 13 parameter in context of the next couple of bullets, is 14 whether we fully opened up the doors to form the 15 hinges somehow and they stuck open. We satisfied that 16 criteria.
17 CHAIR STETKAR: Casey, just for the benefit 18 of everybody else, make sure, these are the lower 19 condense inlet doors?
20 MR. WAGNER: That's right.
21 CHAIR STETKAR: Okay.
22 MR. WAGNER: Exactly. And that's the only 23 place we had done the sampling, because we thought 24 those were by far the most important. And just a 25
97 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 little background, those are modeled with five MELCOR 1
flow paths, representing, I think it's about 20 doors.
2 They take about 46 pascals to fully open, 3
so it's not a lot of pressure, but surprisingly, that 4
took a bit of a pressure pulse in order to open them, 5
all of them, because they open up such a large flow 6
area, about ten square meters each. And so, you get 7
these opening up and they start relieving a lot of 8
pressure.
9 And the general comment was, we didn't 10 have much discussion about how they behaved. We had 11 the sample parameter, so we added discussion about how 12 the valves or how the doors behaved and FSEM plots and 13 tried to look at a range of different responses.
14 Where we might have had a hydrogen burn 15 early, prior, due to the PRT, one that might have been 16 delayed until hot leg failure. And then, one where 17 there wasn't even enough hydrogen there, we had hot 18 leg failure and looked at the response.
19 And the interesting thing that we also 20 got, Trey was able to dig out a few more diagnostics 21 out of his study, was that if we had a hot leg 22 failure, we almost got 100 percent, all five doors 23 sticking open.
24 So, that pressurization from the hot leg 25
98 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 failure was enough. The pressurization from the PRT 1
wasn't. And that is just a smaller tank, that's 2
blowing down.
3 And even though I can imagine that's 4
pretty violent down there, it was triggering some of 5
the doors sticking all the way open, but not all of 6
them. And in general, I call that a partial AJAR, 7
because it's not all five MELCOR flow paths fully 8
jammed open.
9 And so, we were able to get some really 10 interesting insights, digging into the door behavior a 11 little bit more. It's all documented in the report.
12 We talked a bit about ACRS member comment on or it 13 came up that -- we talk about burn direction, inside 14 the report.
15 And we have known ignition sources and you 16 said, we know the geometry, we know where they're 17 located, does it make sense that we are randomly 18 sampling the direction? That made maybe more sense 19 when we had the random sampling, a couple studies ago.
20 And I'm going to try and make a 21 distinction to you that, although we know the location 22 of those ignition sources, they're coming from the hot 23 leg or the PRT or from the ex-vessel debris, the 24 distinction I want to make is where the ignition 25
99 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 actually takes place may be a little ways away from 1
there and elsewhere inside the compartment, because of 2
the hot plume has to come out and it's got to mix with 3
some oxygen.
4 It's coming out as just pure hydrogen and 5
steam and that's not combustible until it mixes and 6
maybe ends up in different locations. And so, there 7
are -- or it's aerosols and hot things coming out of 8
the ex-vessel debris. And so, we were -- we 9
recognized the point that you were trying to make, but 10 we felt like there is still uncertainty where ignition 11 takes place.
12 And then, we kind of go one step further, 13 recognizing the comment that you had is that, let me 14 just read that last line, identifies the flame 15 directions as it propagates within the compartments as 16 surrogates for uncertainties in the ignition location.
17 So --
18 MEMBER CORRADINI: So, I interpret that to 19 mean you didn't change anything?
20 MR. WAGNER: We didn't change the words 21 that we used. The upward, horizontal, and downward is 22 pervasive through the document. And we tried to give 23 some motivation that there's some uncertainty on where 24 it was located.
25
100 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: They -- as I read it, you 1
changed nothing in the model --
2 MEMBER CORRADINI: Right.
3 CHAIR STETKAR: -- you added some words in 4
the text to --
5 MEMBER CORRADINI: Justify what you're 6
doing.
7 CHAIR STETKAR: -- impart this notion of 8
perhaps a more randomization of the ignition location, 9
despite the fact that you had three nominal point 10 sources.
11 MR. WAGNER: We --
12 CHAIR STETKAR: Is that -- that's what I 13 kind of hear you saying this morning, that --
14 MR. WAGNER: Yes, the --
15 MEMBER CORRADINI: I --
16 MR. WAGNER: -- that --
17 MEMBER CORRADINI: I'm sorry.
18 MR. WAGNER: Go ahead.
19 MEMBER CORRADINI: No, you finish, I'm 20 sorry.
21 MR. WAGNER: We understood, we could --
22 that was, I think, Mike's comment at the time. We 23 know where these things are, shouldn't we -- but 24 that's where the hot gas leaves from. And then, it 25
101 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 isn't ignited yet, it's -- you have to have adequate 1
oxygen, adequate hydrogen, and you have to have that 2
mixing and the source, the high temperature, in order 3
for it to ignite. And so --
4 MEMBER CORRADINI: But -- okay. So, I'm 5
not going to -- I mean, I don't think this is -- well, 6
I don't know if this is a big effect, it may not be a 7
big effect.
8 But in my simple mind, if I know where 9
it's coming out and I know the geometry, and we're 10 talking about a specific reactor, then I'm going to 11 essentially entrain the constituents so that I could 12 estimate when I'm going to get the right concentration 13 and, therefore, decide whether it's up, down, or 14 sideways, versus letting it be just a casting.
15 So, that's kind of the source of my --
16 right? If I'm dumping it out from hot gas to the hot 17 leg, then I know how it's going to, not know, but I 18 can, from an engineering judgment standpoint, decide 19 how it's going to mix.
20 I'm -- I seem to remember, my memory is 21 not so good, I seem to remember, I was just very 22 confused about the directionality issue. And it just 23 seems to add confusion, particularly when you know 24 where it's coming.
25
102 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 So, I'll ask it a different way. Did you 1
do a side calculation to confirm that it didn't make a 2
big effect or was nothing done calculational, you just 3
re-explained it? Re-rationalized it, excuse my 4
English.
5 MR. WAGNER: It was the later.
6 MEMBER CORRADINI: Okay.
7 MR. WAGNER: It's coming out at sonic 8
speeds, it's not combustible, what's leading --
9 MEMBER CORRADINI: Well, I mean, there's an 10 old study -- so, I only remember things that are old.
11 There was a study about 35 years ago from Spalding, 12 which has a very set of simple experiments of a gas 13 jet, whether it's a light gas or a heavy gas, he does 14 everything.
15 And he basically does a very beautiful way 16 of showing how far down the pathway, how the 17 concentrations mixed with something, and you can use 18 Spalding's study as a way to estimate, if I have the 19 hot debris coming up PRT, it's so many L over Ds and, 20 by God, that's where it's going to go and it's going 21 to propagate in that direction and then, we're done.
22 MR. WAGNER: Yes.
23 MEMBER CORRADINI:
Versus doing it 24 stochastically. That's kind of what was going through 25
103 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 my mind.
1 MR. WAGNER: Yes. Our dominate source is 2
the hot leg failure, which we have a lot of 3
uncertainty on --
4 MEMBER CORRADINI: Where it's going to 5
happen.
6 MR. WAGNER: -- the working -- and the PRT, 7
I think that probably makes more sense on. But most 8
of our ignitions came from the hot leg. And then, the 9
cavity is a little bit different animal, because it's 10 hot debris --
11 MEMBER CORRADINI: I think --
12 MR. WAGNER: -- it's kind of closed up --
13 MEMBER CORRADINI: -- the reason I focused 14 on this is, I have to admit, I still don't understand 15 the logic of the directionality. Maybe other people 16 do.
17 CHAIR STETKAR: I thought it got it when we 18 had a uniformly distributed set of random ignition 19 sources.
20 MEMBER CORRADINI: Because then they --
21 CHAIR STETKAR: Because they could --
22 MEMBER CORRADINI: -- can go anywhere in 23 the volume.
24 CHAIR STETKAR: -- go anywhere in the 25
104 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 volume.
1 MEMBER CORRADINI: But when I know where it 2
is in the volume, it seems to me, I've narrowed my 3
possibilities. That's, in my --
4 CHAIR STETKAR: I think the truth is, 5
there's still uncertainty, but it, in my opinion, it's 6
much more constrained. It ought to be much more 7
constrained in the current model. But what you said 8
earlier, does the difference -- would it make a 9
difference? I can't, I certainly can't answer that 10 question, I have no idea. I have absolutely no idea.
11 MEMBER CORRADINI: I mean, I know you have 12
-- sorry, my green light's on. I know you have 13 limited resources and the uncertainty study is fairly 14 complex, but I guess a side calculation, just to 15 verify that directionality doesn't make a big 16 difference once I have the known location of the 17 source, would be an interesting way to kill this guy 18 off. And I see we have a helper.
19 MR. ESMAILI: Let me -- well, I don't know.
20 CHAIR STETKAR: Hossein? Identify yourself 21 first.
22 MR. ESMAILI: Let me --
23 CHAIR STETKAR: Hossein? We don't know 24 you.
25
105 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. ESMAILI: Oh, okay. Hossein Esmaili --
1 CHAIR STETKAR: Thank you.
2 MR.
ESMAILI:
Office of Nuclear 3
Regulatory Research. So, I hope I'm not going to 4
confuse you further. That's why we wrote down here, 5
the location of ignition within the compartment. So, 6
in other words, this is tied to the flammability 7
limit.
8 So, we have this control volume, and you 9
can shake your head if I'm going in the right 10 direction or not, so, we have this control volume, 11 right? Hot gases are coming from one source, right?
12 I don't know whether my flame -- the way we are 13 modeling it, we are modeling it as a flame, right?
14 Propagating either, within this control 15 volume, not going to other control volume, whether 16 it's going to go down, going sideways, or going 17 upwards, right? So, this is one way of capturing 18 where that flame is going to propagate from and 19 whether I have satisfied my flammability criteria.
20 Typically, what we do is that, by default, 21 in medical, for example, we say, if in a control 22 volume, if you have more than ten percent hydrogen and 23 other flammabilities are satisfied, then you have 24 ignition. Or if the ignitor is available, it's seven 25
106 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 percent.
1 Here, we don't know, because we don't know 2
how this gas, as they are coming out of this hot leg 3
or as they are going upwards, they are mixing within 4
that control volume, so I don't know exactly where 5
that frame front is propagating from.
6 So, this is not only the location of the 7
ignition sources, but it's tied to how I am doing the 8
flammability limit within that control volume. This 9
is not propagating from this control volume to that 10 control volume. That one, we know already.
11 MEMBER CORRADINI: Okay. So, within the 12 volume, we're speaking?
13 MR. ESMAILI: Within the volume, we're 14 speaking. So, as Casey said, this gases are coming up 15 right at the location where the hot leg is or where 16 the PRT is, that does not -- that may not satisfy the 17 flammability limit, because he says that it is mostly 18 hydrogen and steam. It has to mix with this volume.
19 And we are working within a control volume 20 approach, right? So, what we really do is that, we 21 are, in the control volume, we are trying to model a 22 flame that propagates in that control volume and for 23 that, we are just changing the ten percent to four 24 percent, six percent, nine percent, because we don't 25
107 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 know where it occurs. Did I --
1 MR. WAGNER: Yes. It originally went back 2
to the Kumar data and if you ignite at ten percent, it 3
can go downward. If you ignite at seven percent, it 4
can go sideways and upward. And if you ignite at four 5
percent, it only -- you're the expert, you can --
6 (Laughter.)
7 MEMBER POWERS: I simply relate an example 8
I have actually used in a class I taught, where having 9
explained how well we know these ignition directions 10 and whatnot. That at 12 percent, you get kind of a 11 uniform expansion of a flame. At nine percent, it has 12 a tendency to go up. At seven percent, it has a 13 profound tendency to go up.
14 And then, I show an example of a seven 15 percent ignition and it does exactly as I said, the 16 flame propagates upwards. Everybody watches and 17 amazed that I'm so accurate in my prognostications.
18 And promptly, the flame comes down. And it's because 19 the gas gets heated up and pressurized.
20 Directionality of flame once it's ignited 21 is an extremely complicated thing to model and doing 22 it in a lump node code, where you've got all kinds of 23 structures actually there, that are vacuous in the 24 code, it seems to me to be braver than I am at these 25
108 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
- things, especially when you're talking about 1
relatively lean flames.
2 It directionality is easy and 3
reasonably concentrated, where reasonably concentrated 4
is like ten to 11 percent. Once you get down to lean, 5
around seven, it's going to depend on things other 6
than just the concentration of the hydrogen in there, 7
everything else is going to make a difference, too.
8 MR. ESMAILI: This is Hossein Esmaili, 9
again, from Office of Research. One thing we want to 10 make clear is that, when we say within the 11 compartment, we don't mean every compartment that we 12 model.
13 This one, we only imposed this in the 14 lower compartment region, correct, Casey? Because 15 that is where the location of the hot leg is. So, 16 this only means that we do this uncertainty in the 17 lower compartment, this control. So, the other 18 compartments are still allowed to propagate. So, yes, 19 this was another --
20 MR. WAGNER: Yes. That's a good point to 21 be made. After we ignite it, in some ways, you could 22 look at this as surveying a range of ignition 23 strengths and that first place that it ignites. And 24 then, it will propagate according to the default 25
109 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 criteria, which uses Kumar also. I understand your 1
point, Mike.
2 MR. FULLER: Okay. So, we'll move on to 3
the next one. I don't know about you all, but after 4
the last few years of struggling with the safety valve 5
modeling, every time I see those words together, 6
safety valve, I have this sense of dissatisfaction 7
deep inside my core.
8 But we took your comments from the last 9
Subcommittee meeting and we tried to address them as 10 best we could. Oh, you missed my -- dissatisfaction 11 deep in my core, every time I see those words 12 together, safety valves.
13 (Laughter.)
14 MR. FULLER: But we're trying, we keep 15 trying.
16 CHAIR STETKAR: You're very trying.
17 MEMBER POWERS: You're an obsessive-18 compulsive, that's all there is to it.
19 MR. FULLER: You're right, it takes a 20 certain amount of OCD to keep going and soldiering on.
21 We keep trying.
22 (Laughter.)
23 MEMBER POWERS: This is a psychosis that 24 can be treated.
25
110 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: Tina?
1 MR. FULLER: Yes?
2 CHAIR STETKAR: In all seriousness --
3 MR. FULLER: Yes.
4 CHAIR STETKAR: -- I'm really disappointed.
5 I mean, I'm really, really disappointed. I asked, in 6
June, a simple question. In June, we said, well, we 7
had this data -- I'm sorry, I'm not even going to call 8
it data anymore. We had these numbers from NUREG/CR-9 7037.
10 MR. FULLER: Yes.
11 CHAIR STETKAR: Those numbers are 15 12 failures in a total of 769 demands. That's all it's 13 in. Then, you said, well, okay, NUREG/CR-7037 only 14 goes up through September of 2007, so, gee, we went 15 back to see what more recent experience is.
16 So, we looked at experience from October 17 of 2007 through March 2016 and, lo and behold, we come 18 up now with 16 failures, one more failure, in a total 19 of -- I have to look up my numbers here, so that 20 they're on the record. I can't find them and I'm not 21 going to waste my time. A total of 75 more demands, 22 which you've subdivided into initial demands and 23 subsequent demands.
24 So, my question in June was, tell me how 25
111 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 you estimated the additional 75 demands and where the 1
additional failure came from, because that, at least, 2
was done by people who are still living today and 3
perhaps have a memory, and probably documented what 4
they did, so that we could understand, at least from 5
that snapshot, the thought process.
6 That does not at all come through in the 7
current report. In the current report, all it says 8
is, we needed to use engineering judgment and, yes, 9
there's still a lot of uncertainty on what the failure 10 rate should be. That's all it says. That's all it 11 says.
12 So, the report does not answer any of my 13 questions. So, I'm hoping that some physical body 14 sitting in this room can answer my questions.
15 MR. FULLER: Yes.
16 CHAIR STETKAR: Because if you can't, I'm 17 willing to say that the entire study is flawed. This 18 is a big issue, you make assertions in the report 19 that, oh, yes, well, things could slide to the left, 20 things could slide to the right, but you see the same 21 thermohydraulic behavior whether a valve fails to 22 open.
23 That's great if you're doing a conceptual 24 study. It's not good if you're drawing insights and 25
112 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 conclusions about the behavior of the Sequoyah plant.
1 MR. FULLER: Okay. So, what we did add to 2
the report was a summary paragraph. I realize, again, 3
it's not satisfying, because a lot of engineering 4
judgment does have to go in from the people who are 5
reading these reports and capturing both the demands 6
and the failures in the database.
7 So, we tried our best. We did talk to the 8
people who actually do this, the coding of the 9
operating experience, to figure out what they -- how 10 they decide. So, everything is reliant on licensee 11 reporting.
12 CHAIR STETKAR: I --
13 MR. FULLER: So, they have a requirement to 14 report failures of the codes they see. They certainly 15 don't have to report all the demands.
16 But typically, the operating events, these 17 are post-scram events, you can try to guess how many 18 demands were placed on the safety valves, because you 19 do have at least information that the scram event 20 happened and then, you can read the description of 21 what happened.
22 Sometimes, there's information about 23 system pressures, sometimes there isn't. The level of 24 information just varies a lot from LER report to LER 25
113 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 report.
1 So, in terms of the numerator, at least we 2
have more confidence that, because of the reporting 3
requirement, that we know how many failure events 4
happened for the safety valves. For the denominator, 5
the number of demands, we don't have a definitive 6
answer.
7 And I think what we said in the report is 8
that we may be undercounting the demands, because we 9
have to use some judgment on --
10 CHAIR STETKAR: But --
11 MR. FULLER: -- when those valves were 12 demanded and how many.
13 CHAIR STETKAR: Let me read verbatim what's 14 in the report, so it's on the record. Many LERs 15 describing the plant response to scram report the 16 operation of SVs, but a significant portion of the 17 LERs just use phrases like, all systems operated as 18 expected. Hence, for the number of demands, expert 19 judgment must be used to interpret information 20 provided in the LERs and may be undercounted.
21 My assertion is that you may have 22 substantially over counted the number of demands, 23 because it's very, very surprising to me that between 24 October of 2007 and March of 2016, we had a number of 25
114 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 scrams that resulted in 75 demands to open main steam 1
safety valves and/or pressurizer safety valves, but I 2
would be really surprised if we opened any of them.
3 So, my assertion is that you may have 4
substantially undercounted the denominator, making the 5
whole failure rate substantially optimistic. And if 6
the same process was used in NUREG/CR-7037, the 7
numbers from that report may be substantially 8
optimistic, the failure rate.
9 So, what I was trying to -- what I was 10 hoping to achieve, and apparently will fail, is to get 11 someone who actually did the work from 2007 to 2016 to 12 tell us what criteria were used when you examine the 13 operating experience to say, this event I judge today 14 to result in a demand to open the main steam safety 15 valves at Plant X and I assumed that n number of 16 safety valves were demanded to open.
17 I'm not hearing that answer. And without 18 that answer, I have absolutely no confidence in the 19 failure rates that are used. And I firmly believe 20 that they're optimistic. I can be convinced 21 otherwise, but I need to be convinced otherwise by 22 somebody who tells me what they did.
23 MR. FULLER: Yes, okay. So, I apologize.
24 As I said, I summarized the input we got from the team 25
115 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 in that one paragraph we inserted, but maybe we can 1
arrange a side conversation with the folks --
2 CHAIR STETKAR: No, no, we don't do things 3
4 MR. FULLER: -- who actually --
5 CHAIR STETKAR: -- in side conversations, 6
this is a -- the Subcommittee has raised this --
7 MEMBER BLEY: At least, not technical 8
information.
9 MR. FULLER: Right, okay.
10 CHAIR STETKAR: Not technical information.
11 MR. FULLER: So, maybe what we can do is 12 try to set aside some time at the full Committee 13 meeting to --
14 CHAIR STETKAR: Well, we have to decide on 15 16 MR. FULLER: Okay.
17 CHAIR STETKAR: -- the problem with the 18 full Committee meeting is that we have pretty good 19 representation of Committee Members here and have had 20 21 MR. FULLER: Yes.
22 CHAIR STETKAR: -- in previous meetings.
23 But I don't think that we've had the full complement 24 of the full Committee. And the Members who haven't 25
116 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 attended the Subcommittee meetings need to get briefed 1
on the whole scope of this study.
2 MR. FULLER: Yes.
3 CHAIR STETKAR: And even those of us who 4
have attended the Subcommittee meetings have to have a 5
better feel for how you're presenting the full study 6
to the full Committee.
7 MEMBER BLEY: Other than an informal 8
communication, which we can't do, if you wrote 9
something up and sent it to us, that we could look at 10 ahead of time.
11 MR. FULLER: Oh, okay, I see.
12 CHAIR STETKAR: Yes, if you want to write a 13 little report --
14 MR. FULLER: I see --
15 CHAIR STETKAR: -- we can do that.
16 MR. FULLER: -- and then put it --
17 CHAIR STETKAR: But you --
18 MEMBER BLEY: A memo.
19 CHAIR STETKAR: A memo.
20 MR. FULLER: -- put it in the public 21 record, yes. Okay. Actually, that's a good --
22 CHAIR STETKAR: And that might be a way, if 23 we can get it -- the full Committee meeting is in less 24 than two weeks.
25
117 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. FULLER: Yes. Okay, yes. So, yes.
1 MEMBER SUNSERI: While you --
2 CHAIR STETKAR: But we can't do it 3
informally.
4 MEMBER SUNSERI: While you --
5 MR. FULLER: Okay.
6 MEMBER SUNSERI: While you're thinking 7
about that question, the question I had in association 8
with this is, does the failure analysis include 9
testing? I mean, you test these safety valves every 10 refueling outage --
11 MR. FULLER: Yes. So, we had -- yes. We 12 had previously talked about that quite a bit. We 13 discovered through talking to the folks who actually 14 conduct the testing that the testing setup doesn't 15 quite test the reclosing probability, which is what we 16 care about.
17 The testing requirements are focused on 18 relieving pressure under design-basis overpressure 19 accidents. So, they test very well the ability of 20 that valve to open when you demand it to open, but 21 they don't really test it under the conditions that 22 we're looking for, repeatedly passing hot steam and so 23 on. So, there is a --
24 MEMBER SUNSERI: Okay, so you don't 25
118 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 consider a valve that may be in an as-found test 1
condition low on its setting to be a valve that would 2
be open during an accident?
3 MR. FULLER: So, yes. So, the setpoint 4
drift is a separate issue and you're right that we 5
don't worry so much about the setpoint drifts that are 6
typically found during testing.
7 It's true that a valve may open at a 8
slightly lower pressure than it's supposed to, but in 9
the grand scheme of things, we had looked a bit about 10 the effect of that in the -- through sensitivity 11 studies in the Peach Bottom study, which of course is 12 a different plant and we were looking for different 13 things.
14 But the setpoint drift issue is not as 15 important as all the other things that are going on in 16 the integrated analysis.
17 MEMBER SUNSERI: Okay, thank you.
18 MR. FULLER: And so, yes. So, we are not 19 using the testing data, because if you look at the 20 distributions are very different from the operating 21 experience data. And we did kind of talk to the 22 testers, we figured out that it's not applicable.
23 That's why there's so much focus on just using the 24 operating experience data, which is from actual scram 25
119 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 events.
1 CHAIR STETKAR: Tina, if you -- the other 2
thing that we brought up back in June is that, and I 3
don't want to get into details of the uncertainty 4
distribution, but you -- we've revised the uncertainty 5
distribution for the open fraction, I forgot the 6
parameter value, but given the valve sticks open, how 7
big is the area that it sticks open, that was revised 8
from 2016 to current time.
9 That distribution says that there is a 30 10 percent probability, for example, that a valve remains 11 open more than 90 percent. And that -- the shape of 12 the distribution seems intuitive, the distribution and 13 the discussion of the distribution in the report is 14 quite good.
15 It says that most likely it's just going 16 to weep, it's going to stick open just a little bit.
17 There's a non-zero, in this case 30 percent, chance 18 that it could stick open big. And then, there's kind 19 of a small probability that it could stick open 20 somewhere in-between, and that equal probability was 21 distributed over those open area fractions.
22 That all sounds really good, if indeed we 23 have information about the numerator, because people 24 report these things in LERs --
25
120 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. FULLER: Right.
1 CHAIR STETKAR: -- and we have something on 2
the order of 15 or 16 failure events, 30 percent of 3
those would be something on the order of four, five, 4
six events where we would expect to see a valve stuck 5
open pretty big.
6 So, we would also ask the question, is the 7
operating experience, even in the numerator, the 8
things that you have more confidence in, because you 9
can go read those things, does that operating 10 experience support the uncertainty distribution for 11 the open fraction?
12 MR. FULLER: Yes.
13 CHAIR STETKAR: And you said, yes, we're 14 going to go back and look at that.
15 MR. FULLER: Yes. So, that's what my 16 second bullet is about. We did go back and reread all 17 of the LERs for those events and we actually had a 18 valve subject matter expert also review both the LERs, 19 as well as we had additional information from a 20 proprietary database, which we can't get into the 21 details of that in the study, but we explained in the 22 report that we have access to this and we looked at 23 it, but we can't explicitly refer to the information 24 from that.
25
121 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 Unfortunately, in relooking at all of 1
that, there was nothing new that we could tease out to 2
kind of add additional details in the report that 3
would be worthwhile. We continued to rationalize, I 4
guess, the distributions that we had come up with.
5 And for the open area fraction, it's even more 6
frustrating because some of the reports had no 7
information on it.
8 But, however, I will say for that large 9
open area fraction, there were a handful of events, I 10 can't remember if it was four or five, I want to say 11 on the order of four, that were due to latent 12 conditions, assembling errors, maintenance errors 13 where the valve was left in the position that, when it 14 was open, the internals got completely misaligned, 15 something fell out of place, and it was basically 16 stuck open with a pretty large area.
17 So, the mechanisms for the failing with 18 the large open area versus the weeping area are kind 19 of different. And what we saw in some of these events 20 was, when there was a maintenance or assembly error, 21 sometimes it got stuck open with a very large area and 22 couldn't be recovered, even by lowering the pressure.
23 So, we had data, again, hard to quantify 24 very specifically what percentage should fall in these 25
122 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 different bins, but we did have data to support these 1
two ends of the distribution.
2 MEMBER BLEY: You give, in a couple places, 3
I think, a description of qualitatively the kind of 4
people you talked to and how you tried to go at this 5
and the data you looked at. I didn't see anything 6
real clear on how you picked the 90 percent and the 30 7
percent.
8 I have a question about the design of the 9
valve. I think the design of the valve stem is such 10 that if you're 30 percent open on this kind of valve, 11 but I'm not sure of this, that you get something like 12 30 percent mass flow.
13 I don't know if that's true. How did you 14 translate valve open fraction into mass flow rate for 15 your calculations?
16 CHAIR STETKAR: That's a MELCOR question, 17 right?
18 MEMBER BLEY: Well, I --
19 MR. FULLER: It's a --
20 MEMBER BLEY: -- don't know if it's an 21 input to MELCOR or if it's something done inside the 22 code. I don't know if that aligns with how these 23 particular valves are designed for flow rate versus 24 percent open.
25
123 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. FULLER: Yes.
1 CHAIR STETKAR: And in particular, take 2
that very, very unlikely thing in the center, where 3
it's stuck open 37 percent.
4 MEMBER BLEY: That could be a whole lot --
5 CHAIR STETKAR: That's --
6 MEMBER BLEY: That could be almost full 7
flow or it could be fairly low flow, depending on 8
exactly how that valve --
9 MEMBER MARCH-LEUBA: If it's high pressure 10 11 MEMBER BLEY: -- seeping surface is 12 designed.
13 MEMBER MARCH-LEUBA: If it's high pressure, 14 it's likely to choke flow and be linear with area.
15 And at this pressures, I would say it's choked flow, 16 right?
17 MR. WAGNER: That's a true statement.
18 Sometimes, you get what you're getting to, you get 19 stem position versus flow area. And so, it will 20 change and it can be nonlinear.
21 MEMBER BLEY: Well, no, actually, this 22 isn't one, but if you have a gate valve and it opens, 23 it only has to get a little bit open before you get 24 nearly full flow through it.
25
124 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 If you have a carefully designed throttle 1
valve, then the flow rate, ignoring the point you just 2
brought up, can be proportional or some other 3
relationship to the position of the valve stem. So, 4
there's a lot of detail here and I don't know if the 5
detail has been worked into the model, that's kind of 6
what I'm asking.
7 MR. WAGNER: No, we have a simple linear 8
representation of that.
9 MEMBER BLEY: Okay. Which would be sort of 10 appropriate, if it's a good throttle valve, if that's 11 the design.
12 MR. WAGNER: Yes, it --
13 MEMBER BLEY: I don't know that relief 14 valves are designed to do that. They're designed to 15 pop open and then, pop shut.
16 MR. WAGNER: Yes. Well, we're --
17 MEMBER BLEY: So, 30 percent open might be 18 damned near full flow, I don't know.
19 MR. WAGNER: We don't have many samples in 20 the middle there, we're kind of bimodal distribution 21 of weeping or full --
22 MEMBER BLEY: Well, you got a --
23 CHAIR STETKAR: But, again, to answer 24 Dennis's question, that's why --
25
125 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER BLEY: Something near ten percent 1
chance of 30 percent open.
2 CHAIR STETKAR: Yes. Well, but --
3 MEMBER BALLINGER: So, you're satisfied 4
that 30 percent is the right number? I mean, that's 5
an awful lot of really bad screw-ups in calibrating 6
those valves, because those valves have to be --
7 they're tested quite frequently. It just seems like 8
30 percent of them --
9 CHAIR STETKAR: No, this is 30 -- given the 10 fact that it's stuck open --
11 MEMBER BALLINGER: Oh, okay.
12 CHAIR STETKAR: -- what is the fraction of 13 the area that it's stuck open at?
14 MEMBER BALLINGER: Okay.
15 MR. FULLER: So, I think --
16 CHAIR STETKAR: It -- what I'm -- my 17 initial question was, how likely is it that it sticks 18 open? And that number, right now, is on the first 19 demand, 2.65 times ten to the minus two, I think, 20 based on 16 failures in -- I've forgotten the -- I 21 don't have the table in front of me here, but --
22 MEMBER BALLINGER: Okay.
23 CHAIR STETKAR: -- 16 failures in some 24 couple hundred demands, 600 demands or something like 25
126 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 that.
1 MR. FULLER: So, I think the one -- the 2
reason we gave that ten to 30 percent a little bit 3
more weight than the 30 to 90 percent is, again, we're 4
trying to squeeze as much as we can out of a very 5
small data set.
6 And in one of the events, we were able to 7
calculate, they had enough information that we could 8
calculate that there was 20 percent flow area that had 9
opened up. So, that kind of said, well, so it does 10 happen, maybe -- so, that was our -- we had that one 11 data point that --
12 MEMBER BLEY: Just an overview.
13 MR. FULLER: Yes.
14 MEMBER BLEY: What you've done is a lot 15 more physically pleasing than 2016 was, which wasn't 16 at all physically pleasing. So, in general, the idea 17 makes sense. I don't know how far to push this. I 18 didn't find much to give me, in what's written, to 19 give me confidence that where you ended up made a lot 20 of sense either. But it's at least in the right 21 direction, I would think.
22 CHAIR STETKAR: I think that, I mean, we've 23 talked about this a few times, you're clear on our 24 sort of concerns. My personal thought is that this 25
127 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 issue, for the purposes of this particular study, 1
ought to be met head-on.
2 Acknowledge the fact that perhaps you do 3
not have good justification for that failure rate or 4
perhaps even the specific probabilities that's shown 5
for the uncertainty distribution, although the shape 6
of the uncertainty distribution is -- makes a lot of 7
engineering sense.
8 And that indeed will have implications on 9
the pedigree of, again, I'll call them numbers, I will 10 not call them data, I refuse to do that if nobody can 11 tell me where it came from, on the numbers in 12 NUREG/CR-7037.
13 And I personally would have absolutely no 14 problem calling into question the fact that they are 15 suspect and that they have not been corroborated as 16 part of this effort. That doesn't say that the 17 overall study is fatally flawed, it just says, use my 18 data, get my results, and you used this data and you 19 got your results.
20 Going forward -- and we raised this 21 question back, I looked back through my notes, in the 22 Surry study about the valve failure rate. And at that 23 point, we said, well, it wasn't all that very much 24 important, so get over it, it's from 7037.
25
128 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 Anything you do from here on out in the 1
future, ought to, in my personal opinion, recreate a 2
defensible failure rate for those valves. Even if you 3
have to go back and start with a piece of paper that 4
looks like a blank piece of paper and create your new 5
failure rate, because it has to be justified somehow.
6 MR. FULLER: So, I think we will beef up 7
the discussion in the report. What I did try to say 8
in the executive summary and the conclusions was that, 9
we recognize that what you assume for the safety valve 10 modeling is very critical to the outcomes and that 11 there's still considerable uncertainty in our state of 12 knowledge.
13 And I think that's an accurate statement.
14 I mean, even if you --
15 CHAIR STETKAR: It is, Tina, but it -- if 16 I'm a reader of the executive summary, I get these 17 words uncertainty thrown in and I get confused, 18 because what you're talking about is, we have no 19 confidence, I'll be blunt, we have no confidence in 20 the numbers that come out of that NUREG.
21 We have uncertainty distributions about 22 those numbers that we use and that we talk a lot 23 about, because we do this parametric uncertainty 24 analysis and we went back and revisited the open area 25
129 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 fraction and we have an uncertainty distribution.
1 It gets muddied to a reader of the 2
executive summary, because all you say is, yes, 3
there's uncertainty about what the failure rate should 4
be, in the executive summary. And I say, yes, you 5
have an uncertainty distribution about the failure 6
rate as a parameter in your model.
7 So, all you're doing, to me, as kind of a 8
dispassionate reader of the executive summary, all 9
you're doing is reinforcing the fact that you looked 10 at uncertainty and that parameter. That's the way I 11 came across, reading the executive summary --
12 MR. FULLER: Okay.
13 CHAIR STETKAR: -- rather than saying, 14 look, we used the failure rate from this source, which 15 is the best failure rate that we had at the time. We 16 updated it with additional operating experience, but 17 we still have questions about what the number is.
18 And that a takeaway from the study is 19 somebody really needs to go back and reexamine the 20 technical bases and operational history for not only 21 the failure rate, but given a failure, what the 22 relative fraction of the stuck-open area is.
23 MR. FULLER: Yes, okay. Yes. I --
24 CHAIR STETKAR: That's kind of what -- in 25
130 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 the sense of a lessons learned. It isn't -- I'm 1
coming across as kind of condemning the study, and I 2
would condemn the study if it says, these are the data 3
that we are going to hang our hat on, as far as 4
drawing insights and conclusions about risks.
5 MR. FULLER: Okay. So, I think we are 6
taking your comments again, we'll try to do a better 7
job of framing the discussion. The only other thing I 8
can offer is that, we've had pretty wide review of 9
this at this point, because of all the questions.
10 We've presented the work at an ASME Pump 11 and Valve Symposium, we've talked to more valve 12 subject matter experts. Nobody has looked at the 13 approach and said, you're doing this totally wrong, 14 there's something a lot better out there. So, we 15 haven't -- I understand your point.
16 Maybe we have to reframe how we present 17 the work a little bit better, given that there is less 18 confidence in maybe one of the key parameters here.
19 So, again, as I said, every time I see the words 20 safety valve on the paper, I have a deep sense of 21 dissatisfaction.
22 CHAIR STETKAR: No, it's -- but honestly --
23 MR. FULLER: It's just kind of where we 24 are.
25
131 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: Having collected a ton of 1
data in a previous life, it is absolutely true that 2
there, in many cases, must -- in all cases, 3
engineering judgment is involved, because in some 4
cases, you can't even divine whether something was a 5
failure or not.
6 So, there's even engineering judgment in 7
how you throw the numerator into a box for the 8
numerator. And very often, very often, there's more 9
engineering judgment in calculating the denominator.
10 That being said, a responsible data 11 analyst will document the judgment that was used to 12 calculate the denominator. Today, I used the 13 following criteria, based on my knowledge, for these 14 types of events, will result in a safety valve demand.
15 So, that's the type of event.
16 And then, if the event occurs at Plant X, 17 this is the number of safety valves at Plant X that I 18 included. Now, why is the second part important?
19 Well, if you know something about the plants, for 20 example, Plant X might have five safety valves, Plant 21 Y might have three, typically not all of the safety 22 valves will be demanded.
23 So, out of a population of five, you might 24 get one or two. So, if I, today, assumed that all 25
132 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 five were demanded at Plant X every time I achieved 1
these criteria, I'm probably being very, very 2
optimistic in the denominator.
3 But it's incumbent on me as a data analyst 4
to document that engineering basis for that 5
denominator. And if I don't do that, nobody should 6
have any confidence in the numbers that I put forth, I 7
have failed as a data analyst. I'm not a valve 8
expert, I'm not somebody who attends a conference, I'm 9
somebody who analyzes experience.
10 And I may have uncertainty about that 11 denominator, there might be plus or minus a factor of 12 two or three on that, which is fine, I can handle 13 that. Okay?
14 MR. FULLER: Okay.
15 CHAIR STETKAR: We've beat that one up 16 enough --
17 MR. FULLER: Yes.
18 CHAIR STETKAR: -- we do have to be 19 cognizant of the time here.
20 MR. FULLER: The final bullet on that slide 21 was just to note that we did add the discussion for 22 why we didn't include the failure-to-open failure mode 23 and hopefully that was convincing. We showed that 24 there is a
couple orders of magnitude less 25
133 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 probability.
1 CHAIR STETKAR: I'll only make a statement 2
that I have about a page and a half written up on 3
that. I agree with you that the failure-to-open curve 4
that you show in the report will lie below the 5
failure-to-close curve.
6 I believe that the margin is considerably 7
less than is shown there, because even the day --
8 sorry, I almost used that word again, let the record 9
show I said day, not data -- even the numbers in 10 NUREG/CR-7037 for the failure-to-open failure mode 11 give you a higher curve than what you show.
12 And it's not clear at all to me how you 13 accounted for common-cause failures. Because I -- the 14 curve, I backed out a, if you want to call it an alpha 15 factor, or a beta gamma factor combined, for all three 16 valves of something on the order of couple times ten 17 to the minus five.
18 And if I look at NUREG, the latest version 19 of whatever the NUREG for common-cause failure 20 parameters for safety valves, this is a -- for that 21 combination of three spring-loaded safety valves 22 failing to open, I get a couple of orders of magnitude 23 higher than that.
24 I get kind of two or three or four, I've 25
134 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 got the numbers here, but it's not worth whining about 1
them right now, higher than that, so that if I apply 2
the individual failure rate from NUREG/CR-7037 and the 3
composite common-cause failure fraction from this 4
other NUREG, I get a couple of orders of magnitude on 5
the failure-to-open on the first demand, if you will.
6 MR. FULLER: Okay. No, that's a fair 7
point.
8 CHAIR STETKAR: And that's -- again, the 9
mark, it's still below --
10 MR. FULLER: Yes.
11 CHAIR STETKAR:
and I
- think, 12 qualitatively, you can still draw that conclusion.
13 MR. FULLER: Yes.
14 CHAIR STETKAR: But the curves that are 15 plotted, they're --
16 MR. FULLER: No, that's --
17 CHAIR STETKAR: -- a little snarky.
18 MR. FULLER: That's a fair point, I think 19 we have to discuss the common-cause aspect. So, we'll 20 do that.
21 CHAIR STETKAR: The only reason I hang up 22 on this is, I have absolute -- what happens, Casey 23 would know this, what happens if none of them open at 24 time -- on the first demand? Things get a lot more 25
135 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 interesting, don't they?
1 MR. WAGNER: Yes.
2 CHAIR STETKAR: Okay.
3 MR. FULLER: Yes, so we haven't modeled 4
that lower probability scenario. We did --
5 CHAIR STETKAR: Right.
6 MR. FULLER: -- do a sensitivity in Surry, 7
but --
8 CHAIR STETKAR: Yes.
9 MR. FULLER: -- we haven't done that 10 sensitivity here.
11 CHAIR STETKAR: Yes.
12 MR. FULLER: And you're right, it's a 13 completely different --
14 CHAIR STETKAR: It's things -- so, for the 15 purpose of this study, I think it's clear that you 16 haven't modeled it. I think the rationale for why 17 you've not modeled it is -- can be punched up a little 18 bit.
19 MR. FULLER: Okay.
20 CHAIR STETKAR: Especially because somebody 21 like me is going to go pull those numbers out and do 22 that calculation. Okay.
23 MR. WAGNER: Do you have any guidance on 24 time? We have about two slides left.
25
136 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: We're going to try to get 1
through as quickly as we can here. I'm going to make 2
an executive decision just because --
3 SPEAKER: Control yourself.
4 CHAIR STETKAR: I cannot do that. I've 5
tried in the past.
6 If we have to run a little bit long, we'll 7
run a little bit long, but not much. So, let's try to 8
get through these and I'll try to be good.
9 MR. WAGNER: The next area that we 10 addressed is time in cycle sampling. And we had a 11 lively discussion on that one, too.
12 We have a two-part approach to addressing 13 the concerns that were brought up. The first part is 14 we improved the discussion of the distribution between 15 BOC, MOC and EOC in the current documentation. No 16 changes were made, but we improved the documentation, 17 the justification why we went into the three areas.
18 The second approach is we took that to 19 heart and for the Surry UA we're going to do a much 20 more continuous approach of using many more ORIGEN 21 results across the full cycle.
22 CHAIR STETKAR: I saw that. I still, you 23 know, it is what it is. The discussion is much 24 improved. I understand the rationale here.
25
137 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 I still don't know why you picked 200 days 1
and then said there must be 50 percent of the 2
probabilities, you know, equally distributed on either 3
side of that 200.
4 Why you didn't pick 200 and whatever the 5
midpoint is, why you didn't pick 275 days, say, you 6
know, 250 percent, but it is now explained.
7 Somebody can read through there and say, 8
okay, I might take issue about your selection, but I 9
understand the rationale that you used.
10 MR. WAGNER: To briefly go over the 11 rationale, more ORIGEN work was done to develop the 12 rationale. We looked at -- we had three points when 13 we presented last time.
14 We did about 10 or 15 more ORIGEN 15 calculations to kind of put those points in context 16 and give us some insights on the buildup of the short-17 lived radionuclides.
18 And so, the -- try to get justification 19 for the BOC space is for that buildup to the secular 20 equilibrium of iodine-131, that was our surrogate for 21 short-lived, and how the decay heat changes over that 22 60 days.
23 And then the EOC, the justification on 24 that you can -- we had more data points on the Cesium-25
138 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 137. And so, that was a definition of why we put an 1
EOC boundary there.
2 And then the sampling looked -- the 3
particular value that was used, looked at load decay 4
heat. So --
5 CHAIR STETKAR: No, the story hangs 6
together much, much better.
7 MR. WAGNER: Next slide.
8 And a question was on the stability --
9 statistical stability analysis conducted for the 10 figures of merit. That was performed and that was 11 added to Appendix A.
12 We recognized the possible impact of 13 eutectic melt temperature on burnup. We didn't change 14 anything, but we have some --
15 MEMBER POWERS: Since there is no eutectic 16 in the system, I really wonder what that is. There's 17 a monotectic.
18 MR. WAGNER: We use eutectic as our 19 conglomerate melt and sort of dissociation temperature 20 for the field.
21 MEMBER BALLINGER: In that case, eutectic 22 should be put in quotes?
23 MR. WAGNER: Oh, sure. Yeah. Yeah.
24 MEMBER BALLINGER: Okay.
25
139 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER REMPE: Just to follow up on the 1
earlier discussion very briefly, I'm looking at a 2
paper from IRSN and Karlsruhe, Comparison of Core 3
Degradation Phenomena for Various Experiments.
4 There is a quote in there that clearly 5
it's referring to some PHEBUS data and it has 6
"irradiated fuel is dissolved more quickly than 7
unirradiated fuel." It cites the PHEBUS-FP tests.
8 And in that whole paragraph they're 9
comparing differences between irradiated fuel and 10 temperatures versus unirradiated fuel. And so, 11 again, it might be a good place to start looking for 12 data.
13 MR. WAGNER: Yeah, took some notes on 14 that. Thank you.
15 There was some added discussion on our 16 approach for failed computer runs. It is already 17 noted that we --
18 CHAIR STETKAR: Yes, there was. We've 19 discussed that. Next bullet.
20 MR. WAGNER: I wanted to go on. There was 21 some discussion of ice bed response during seismic 22 event. Improved vessel failure discussion to address 23 the cited issues.
24 MEMBER REMPE: So, again, a nitpick, but 25
140 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 it -- the text there needs -- first of all, what is 1
"gross failure"? It's better than variable failure, 2
but you have the word "gross failure."
3 To me, that implies it unzipped and the 4
whole head fell off. I don't think that's what you 5
assumed. You assumed a limited area, specific area, 6
right? And so, find some nicer words that are more 7
representative of what was done.
8 Also, there's another place where it used 9
to talk about the drain line and John harangued you to 10 get rid of that. And now it has something about that 11
-- about penetration -- instrument failures or other 12 failures. And you have in quotes, "instrument 13 failures."
14 And so, look at the words there and just 15 fix it, because it's -- you're having "instrument" 16 both times and it's just --
17 MR. WAGNER: Oh, I see. Okay.
18 MEMBER REMPE: It's an editing thing. But 19 I look for those things, because we have had some 20 comments about it.
21 MR. WAGNER: Yeah. Yeah. The Committee 22 pointed out that there was some confusing discussion 23 on the long-term station blackout on specific 24 generators versus symmetric and asymmetric responses.
25
141 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 And I tried to rewrite that to address that.
1 CHAIR STETKAR: That reads -- there's 2
still obviously a numbering difference, but it's okay 3
now.
4 MR. WAGNER: Great.
5 CHAIR STETKAR: As far as I'm concerned, 6
it states what's done.
7 MR. WAGNER: Okay. Great.
8 MEMBER REMPE: Could I go back to the very 9
beginning since it has the phrase "steam generator,"
10 or the acronym there, about our discussion earlier 11 about the limited scope and not considering 12 consequential steam generator tube rupture.
13 And you did consider countercurrent flows, 14 so I bet your MELCOR runs actually have temperatures 15 of the hot leg and temperatures of the tubes in it.
16 And so, again, my earlier point where I 17 was slapped down because I was saying, can you use 18 some insights based on other evaluations you've done 19 so that up front you -- you're basically looking at 20 different pump seal leakage rates and you're making 21 comments about releases into the containment and 22 things like that.
23 I bet you could even get those insights 24 that are plant-specific based on the runs you have on 25
142 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433
-- and infer that the temperatures of this tube versus 1
the temperature of the hot legs, you might be able to 2
come up with a conclusion about what would have been 3
more likely or not, even.
4 And it would be plant-specific; right?
5 MR. WAGNER: Uh-huh. Yeah. I've noticed 6
that when I've implemented the natural circulation 7
modeling, three loop versus four loop. There are some 8
differences.
9 MEMBER REMPE: Okay. Thank you.
10 MR. WAGNER: Oh, we looked at RCP leakage 11 as an ignition source, the gases that might be coming 12 out of the RCP and added a discussion.
13 CHAIR STETKAR: That's an interesting 14 discussion. And that's all I'm going to say. That's 15 an interesting discussion.
16 I have some questions about that, but we 17 don't have enough time for it right now. There were a 18 couple of curiosities that I --
19 MR. WAGNER: Okay.
20 CHAIR STETKAR: But it's an interesting 21 discussion.
22 MR. WAGNER: The very last slide is just 23 to point out that we added two things that we talked 24 to you about. Hossein's discussion from last meeting 25
143 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 got into the report, and then Trey's focused 1
pressurizer safety valve study is now fully documented 2
in the report.
3 On to MACCS.
4 MR. BIXLER: Okay. On the -- there were 5
several comments also in the MACCS, the consequence 6
analysis portion of the work from last time.
7 And I think I picked the more major things 8
here and tried to respond to those and explain what we 9
have done to address them.
10 The first thing here is a question on 11 emergency response and triggering by -- in our case, 12 we had some emergency response triggered by SAE siren.
13 And there was a question, what is that? Does that 14 really happen? Is that what TEMA would do?
15 And so, we went back and looked at that.
16 And I think we offered at the time of the last 17 meeting, that we would take a more careful look at 18 that.
19 And what we found, we found a couple of 20 things. One is that there's a documentation by FEMA 21 in 2004, of an exercise for the Sequoyah Nuclear Power 22 Plant that shows pretty clearly a time line of events.
23 And one of the earlier things on the time 24 line is the SAE declaration. And then following that 25
144 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 is a first siren.
1 And it notes there that that would trigger 2
school relocation, that buses would be -- bus drivers 3
would be notified to deploy, et cetera. A little bit 4
more is included on the first SAE -- or the SAE 5
message -- the EAS messaging.
6 And then following that, at a later time 7
is the GE declaration. And then following that is --
8 it clearly specifies a second siren and a second EAS 9
message.
10 So, that definitely corroborates what we 11 were modeling the way we had modeled schools and some 12 of the other cohorts that were also related to that.
13 MEMBER SKILLMAN: Nate.
14 MR. BIXLER: Yeah.
15 MEMBER SKILLMAN: This is Dick Skillman.
16 MR. BIXLER: Yeah.
17 MEMBER SKILLMAN: I asked that question.
18 MR. BIXLER: Yeah.
19 MEMBER SKILLMAN: And I asked that 20 question, because the gentleman from -- when 21 challenged, the gentleman from TVA said, "Yes, we 22 really do move the children at a site versus a 23 general," which is where, at least in my experience, 24 the bulk of the evacuations occur.
25
145 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. BIXLER: Uh-huh.
1 MEMBER SKILLMAN:
And this has 2
specifically to do with Cohort 2.
3 MR. BIXLER: Yes.
4 MEMBER SKILLMAN: And this is actually 20 5
percent of the population --
6 MR. BIXLER: Yes.
7 MEMBER SKILLMAN: -- of the area.
8 MR. BIXLER: Yeah.
9 MEMBER SKILLMAN: But what's interesting, 10 and I'm going to go back to what either John or Dennis 11 said, the study is a specific study for this site.
12 And the study for this site is tied very closely to 13 TEMA, to how Tennessee Emergency Management and TVA 14 have agreed in their emergency plan.
15 Because what they do is at an alert, they 16 dispatch the buses. So, the buses are staged before 17 the site occurs, if the timing is --
18 MR. BIXLER: Right.
19 MEMBER SKILLMAN:
So, from that 20 perspective, one must be aware that this is not, for 21 instance, applicable to McGuire, Catawba, TMI, Oyster 22 Creek --
23 MR. BIXLER: Okay.
24 MEMBER SKILLMAN: -- pick your plant.
25
146 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. BIXLER: Yeah. Okay.
1 MEMBER SKILLMAN: Because this is, at 2
least in my experience, unusual that an alert --
3 actually, the control room calls the school.
4 MR. BIXLER: Uh-huh.
5 MEMBER SKILLMAN:
That's what the 6
gentleman from TVA said.
7 MR. BIXLER: Okay.
8 MEMBER SKILLMAN: So, I mean, the analysis 9
is what it is if that agreement is as stated. I have 10 no reason to believe it's not --
11 MR. BIXLER: Right.
12 MEMBER SKILLMAN: -- but one must be clear 13 in one's own mind to recognize this is different.
14 MR. BIXLER: Okay. Yeah, and we -- I 15 don't think we tried to draw any conclusions for other 16 plants, just for this one.
17 MEMBER SKILLMAN: But this is 20 percent 18 of the population.
19 MR. BIXLER: That's right.
20 MEMBER SKILLMAN: So, this is not an 21 inconsequential --
22 MR. BIXLER: Yeah.
23 MEMBER SKILLMAN: -- uniqueness here.
24 MR. BIXLER: Yeah.
25
147 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER SKILLMAN: This is important.
1 MR. BIXLER: Yeah. Okay.
2 MEMBER SKILLMAN: Thank you.
3 MR. BIXLER; And I think I can probably 4
conclude with that. One additional thing I'd like to 5
say, though, is that even for this case where we --
6 and we believe faithfully triggered the beginning of 7
the emergency response by SAE, it's still fairly late.
8 It doesn't happen early on the overall 9
time line, because we assume that there'd be quite a 10 bit of delay for the buses to be able to get to the 11 schools.
12 And so, it didn't end up being a 13 particularly early evacuation of the school kids, it 14 was just triggered earlier. It started the chain of 15 events earlier, but it still took a while to 16 accomplish by our time line.
17 MEMBER SKILLMAN: And one probably should 18 recognize that there could be hours or minutes between 19 the site and the general.
20 MR. BIXLER: Yeah. Yeah.
21 MEMBER SKILLMAN: And so, that's really 22 the important piece here. When you tell the children 23 "Go" at a site, they may be in the next county by the 24 time their parents are told to leave.
25
148 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 So, there are some practical implications 1
associated with this assumption.
2 MR. BIXLER: Yeah.
3 MEMBER SKILLMAN: Thank you.
4 MEMBER BLEY: Nate.
5 MR. BIXLER: Yeah.
6 MEMBER BLEY: There was one thing I think 7
we talked about the last time, and I wonder if you 8
folks thought more about it.
9 You do have an assumption that bridges 10 within the ten miles are unusable --
11 MR. BIXLER: Right.
12 MEMBER BLEY: -- and you shoulder people 13 until you find a route out for them, but I think we 14 asked if -- what about the people who don't wait for 15 that and who jump out on the roads and then find 16 themselves cut off?
17 I know that area kind of well and there's 18 a lot of bridges in a lot of place nearby there.
19 MR. BIXLER: Yeah.
20 MEMBER BLEY: Do you think about people 21 who get stranded for a fair amount of time because 22 they can't find their way back out again when they 23 keep running into dead bridges?
24 MR. BIXLER: We did --
25
149 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MEMBER BLEY: Did you leave a fraction of 1
those who get caught outside rather than sitting 2
inside?
3 MR. BIXLER: I think we ended up doubling 4
the amount of evacuation time to account for the fact 5
that people might reach dead ends and then have to 6
retrace their steps and try something else.
7 MEMBER BLEY: Okay.
8 MR. BIXLER: So, we did try to account for 9
that in our evacuation time line, hopefully, in a 10 reasonable way.
11 MEMBER BLEY: Okay. Yeah, I guess that 12 makes sense.
13 MR. BIXLER: Yeah.
14 MEMBER BLEY: Thank you.
15 MR. BIXLER: Okay. I think -- next slide.
16 Okay. There's also a question about how 17 MACCS treats deposition under humid conditions, which 18 I imagine you're likely to get in the Tennessee Valley 19 area there. And that was a good question.
20 And at the time, we responded that that 21 would have an impact -- in principal, it could have an 22 impact, at least, if you have hygroscopic aerosols 23 being released that are not fully saturated with 24 water. They could grow to a greater size, deposit 25
150 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 faster, et cetera.
1 That's something we don't specifically 2
treat in MACCS at this point. There's no accounting 3
for relative humidity in the area when you're doing a 4
release.
5 But what we ended up doing in response to 6
that, was to do a bounding estimate of how much 7
difference that could make on the deposition velocity 8
of the aerosols.
9 And I think that's reasonably well-10 documented now in the new version of the document, but 11 what we ended up finding is that it wouldn't even be a 12 factor of 2 in deposition velocity.
13 And our uncertainty range covered a range 14 of ten plus or minus the square root of three -- or 15 square root of ten, roughly a factor of 3 up and down.
16 So, I think we covered that uncertainty 17 already in our uncertainty range, but it is -- it's 18 something that we thought was worth adding some 19 additional description/discussion in the document.
20 So, we went ahead and did that.
21 Okay. Next slide. And last comment -- or 22 set of comments was on the presentation of some of the 23 risks that were in the -- with the earlier version of 24 the document, both EF, early fatality, and LCF risks.
25
151 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 So, I basically agree that at the time 1
that we met last time in June, that we would go back 2
and reconsider that and add to the discussion. And 3
so, we've done that now.
4 There's a specific discussion on the 5
bimodal nature of the LCF risk and what that stems 6
from and we reduced the amount of presentation on EF 7
risk.
8 Basically, there used to be a table and a 9
figure there that we eliminated and now there's just a 10 minimal amount of discussion saying that it only 11 occurs in a few of the realizations that we had.
12 And even there it's trivial overall that 13 there's almost no chance of an early fatality. So, 14 those are the things we did to address the comments 15 from last time.
16 CHAIR STETKAR: I commented earlier on the 17 early fatality. There's still a tabulation in there 18 that shows the mean value and zeroes for all the other 19 percentiles.
20 I think -- again, I think a tutorial will 21 help.
22 MR. BIXLER: That might be in the 23 appendix, are you thinking, or --
24 CHAIR STETKAR: I don't -- no, there's one 25
152 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 in the appendix, but there's -- the one in the 1
appendix has mean values and 95th percentiles, as Trey 2
showed --
3 MR. BIXLER: Right.
4 CHAIR STETKAR: -- for four of the five 5
distances that indeed are higher than the mean value, 6
which is reassuring to a lot of people.
7 There is a table in the main body of the 8
report and I, you know, if you want me to --
9 MR. BIXLER: In Chapter 6?
10 CHAIR STETKAR: I don't remember whether 11 it's Chapter 6.
12 MR. BIXLER: Okay.
13 CHAIR STETKAR: It probably is. I think 14 that does show non-zero mean values and zero values 15 for the median, fifth and 95th percentiles.
16 MR. BIXLER: I think the intention was to 17 take that table out.
18 CHAIR STETKAR: Well, it's still in there 19 and it -- you can argue one way or the other about 20 whether it's useful. It might be useful.
21 If it's in there, as I said earlier, I 22 think that it would be very useful to have a bit of a 23 tutorial for folks --
24 MR. BIXLER: Okay.
25
153 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: -- about how that type of 1
behavior makes sense --
2 MR. BIXLER: Right.
3 CHAIR STETKAR: -- because most people 4
don't get it. And most people -- what I've found is 5
many people do not support the notion of quantitative 6
uncertainty analysis and are too easy to dismiss it as 7
simply an abstract mathematical, statistical, 8
nonphysical exercise that you push a button at the end 9
of a study and something comes out.
10 So, I think that table is actually useful.
11 It shows that the numbers are small --
12 MR. BIXLER: Okay.
13 CHAIR STETKAR: -- but it needs a bit of 14 discussion.
15 Now, regarding the bimodal nature of the 16 distribution, in the executive summary, the notion of 17 a bimodal distribution is introduced. You show the 18 cumulative curves. You still have this notion that 19 risk decreases as a function of distance from the 20 site, which is not entirely true.
21 The discussion in Chapter 6 is much, much 22 better. You essentially - now, remember an executive 23 summary has to be written not for people who did the 24 work. It has to be written for people who want to 25
154 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 understand what you did and what you're trying to 1
convey with the results so that when you say a bimodal 2
distribution, most people won't understand what the 3
heck that is even from the shape of the cumulative.
4 They just won't.
5 If you say, well, if we turn up the 6
microscope on this funny-looking thing and show what's 7
shown in Chapter 6, and then say, well, look, for the 8
80 percent -- 80, whatever the heck it is, seven 9
percent of the time when we get an early containment 10 failure; A, it's 87 percent of the time and here's 11 what the risk -- conditional risk profile looks at for 12 that 87 percent. And, look, this one's from the site, 13 doesn't make any difference.
14 But the other 13 percent when you 15 basically don't get a containment failure and have a 16 very small, protracted release, the risk from that is 17 really, really, really small. And in that case, 18 distance from the site makes a big difference.
19 MR. BIXLER: Right.
20 CHAIR STETKAR: And I think telling that 21 story that way in the executive summary, to most 22 people who read it, is a lot more beneficial than just 23 the words that are in there and the figures, because 24 people won't get those figures.
25
155 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 MR. BIXLER: Okay.
1 CHAIR STETKAR: And the same way as 2
publishing the results as a range from something on 3
the order of ten to the minus ninth to something on 4
the order of ten to the minus three is misleading 5
without that corroborating story.
6 MR. BIXLER: Okay.
7 CHAIR STETKAR:
Especially in the 8
executive summary.
9 MR. BIXLER: Okay. All right. We'll take 10 that comment to heart and go back and reword that a 11 little better.
12 CHAIR STETKAR: Because it's really -- I 13 think to a lot of people it's unexpected. That shape 14 is unexpected and the fact that that for in this 15 particular site, in this particular study, distance 16 from the site for the vast majority of the things that 17 you're looking at doesn't make any difference is also 18 counterintuitive to a lot of people.
19 MR. BIXLER: Okay.
20 MS. SANTIAGO: So, that brings us back to 21 talking perhaps about the topics that we should focus 22 on for the November 2nd full committee presentation.
23 MS. GHOSH: Yeah. We didn't have any 24 slides on this, because basically we're soliciting 25
156 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 your input on what we should prepare and focus on.
1 CHAIR STETKAR: Well, in my opinion, I 2
think recognize that there are some members of the 3
Committee who I believe have not heard any of this 4
story.
5 I've tried to go back and see who attended 6
what meetings, but let's go on the presumption that 7
there are at least one or more members of the 8
Committee who have heard nothing of this.
9 That being said, I think that you need to 10 provide an overview of the objectives of the study.
11 The fact that it was tailored, in a sense, to examine 12 hydrogen issues focusing primarily on early 13 containment failure, I would not dwell too much on the 14 long-term station blackout despite the fact that it's 15 part of the study.
16 I think you don't have enough time to draw 17 that distinction. I would keep focused on the short-18 term station blackout exclusively and said that's the 19
-- that's where we did all of our analysis, that's 20 where the integrated uncertainty analysis was done and 21 we just don't have enough time to kind of talk about 22 the others -- the other part of it.
23 I would show the -- certainly show the 24 results and explain the results, the bimodal effect of 25
157 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 the results, you know, where they're coming from so 1
that other members can understand, you know, 2
essentially the sources of risk and the kind of 3
phenomena that are contributing to them.
4 And then certainly, you know, in my 5
opinion, and I'll ask for other help here, I would 6
grapple -- I would grasp the demon by the horns and 7
say, you know, what are some of the lessons that we've 8
learned from doing this study?
9 I think one of the lessons that we learned 10 is that the -- paying careful attention to the 11 uncertainties as you do the study rather than as an 12 afterthought is; A, essential, and; B, can have a 13 measurable effect on the results.
14 You may want to show a couple of examples 15 of that in things like the open area fraction and the 16 fabric seal type of things, but only as an 17 illustration.
18 You don't want to get bogged down in 90 19 minutes in a full committee discussion with people who 20 don't understand how the machine works and don't 21 understand the study in terms of detail.
22 So, I' m not sure, you know, think about 23 that, but I think part of the message is that the 24 uncertainty analysis is not an afterthought and it's 25
158 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 important that it's not an afterthought.
1 I would -- this is -- address the issues 2
that we have discussed, the non-completion of MELCOR, 3
the concerns about the pedigree of the numbers for the 4
valve failure rate and, you know, you've already 5
established the fact that that's an important 6
parameter in the study. And perhaps one or two other 7
issues, you know, that we've discussed.
8 I think that -- here's one area, and think 9
about this, again, it's -- you're going to have to put 10 together a
lot of material in a
90-minute 11 presentation.
12 A lot of the really elegant stuff in this 13 study, both in terms of the modeling and how the 14 sampling algorithms are done, is really, really neat 15 stuff.
16 And all of the stuff that you did on 17 regression analyses to kind of ferret out important 18 contributors and how they may vary, are really, really 19 interesting.
20 I don't know how you jam that into 90 21 minutes or even whether you try. And that's where I'm 22 looking for -- oh, it's a lot of really neat stuff, 23 but I'm not sure whether it's worth trying to get at.
24 MEMBER BLEY: Well, what do you want from 25
159 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 the ACRS? I think that's what you ought to focus on.
1 And anything that you really like, but isn't related 2
to what you want from the Committee, you ought to put 3
aside.
4 For example, you've done a lot of work 5
recently on this focus safety valve study. I would 6
put most of that aside.
7 On something that John talked about, I 8
might go a little further, because when -- I like most 9
of the things you talked about, John, but I think you 10 ought to really narrow it down to the things that will 11
-- what did you learn, what's good?
12 Now, the one thing I hoped you learned was 13 when we started going with you through the uncertainty 14 analysis and said, gee, everything is pointing to this 15 open fraction of a safety valve, all the important 16 stuff was sitting there that led you to go back and 17 say, "Did we really do that right," and reevaluate it, 18 I think that's an important story to tell.
19 Great details about what you actually did, 20 I don't think, is that important, but that you learned 21 from the uncertainty analysis that some part of your 22 modeling was a lot more important than you probably 23 figured it would be when you started.
24 I might be wrong about that, but that 25
160 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 would be my guess. So, that would be the one thing I 1
would focus on a little.
2 MEMBER SKILLMAN: I'd like to offer a 3
comment, if I could. The major take-away for me is 4
the uniqueness of the applicability of this study to 5
this plant.
6 John said it very well a couple hours ago.
7 One should not paint all ice containments with this 8
result.
9 McGuire and Catawba or Watts Bar, because 10 of their site and because of their emergency plan, 11 might have a very different outcome.
12 And so, in the abstract and in your --
13 what will become your executive summary, I think it's 14 very important to communicate the unique applicability 15 of this specific study for this specific plant.
16 That doesn't take away from the potential 17 use of some of the results, but the final conclusions 18 are unique and the users should be advised don't be so 19 hasty just to apply this anywhere.
20 And if that caveat isn't in the Peach and 21 the Surry analyses, it probably ought to be because 22 these are very highly-specific, complicated analyses.
23 It's a unique piece of -- it's a huge piece of work, 24 excellent piece of work, but it's unique.
25
161 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 CHAIR STETKAR: Any other suggestions as 1
far as the full committee presentation from our 2
members?
3 MEMBER MARCH-LEUBA: Yes. For my benefit, 4
I would like to hear something about why we're doing 5
this and -- let me rephrase the question in a more 6
final way.
7 Have you guys worked ourselves out of a 8
job? the moment you write the report, we can fire you 9
because we're never going to use this again. Or what 10 are we going to use -- what benefit does the Agency 11 gain from this exercise? How are we going to apply 12 it?
13 Is it going to apply to licensing? Tell 14 me why we did this.
15 CHAIR STETKAR: Okay. Be careful about 16 that, because you haven't been here for the last 17 decade. The --
18 MEMBER MARCH-LEUBA: More reason to tell 19 me why we did this.
20 CHAIR STETKAR: Okay.
21 MEMBER CORRADINI: But I think Jose's 22 point is fair. Ten years ago, or thereabouts, this 23 was done for a particular reason. But now that you've 24 done it, how do you use it going forward? I think 25
162 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 that's what Jose is asking.
1 MEMBER BROWN: I would -- can I -- since 2
I'm uninitiated and I've been here for nine years and 3
I've sat in on several of these meetings, but not all 4
5 MEMBER BLEY: Started the year before you 6
joined us.
7 MEMBER BROWN: I know.
8 MEMBER BLEY:
You're no longer 9
uninitiated.
10 MEMBER BROWN: Well, after the first 11 couple of meetings, I decided I would be more 12 selective in the meetings that I attended.
13 (Laughter.)
14 MEMBER BROWN: I phrase that very 15 succinctly. And I agree with Jose in that I think 16 it's important since this is kind of the end product, 17 to say why we did this and what are the conclusions we 18 should gather from this in terms of how we use the 19 results.
20 The ones I've sat in on hydrogen 21 deflagration and stuff like that, how does it apply to 22 the safety posture of the plants? How can it be 23 generally applied, if at all, to plants other than the 24 ice, you know, this specific plant?
25
163 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 Because, otherwise, if you go through a 1
bunch of stuff where you show graphs and lots of ten 2
to the minus seventh, eighth and ninth, it's just --
3 it's a sleeper. You're going to fall asleep.
4 You'd really like -- at least I would, you 5
really want to know what conclusions do we draw out of 6
this in terms of is this a big, big problem, safety 7
posture, because it's just going to spread all over 8
the place, or did we get -- were we able to come look 9
at the results and say, look, yeah, this happens, but 10 here's the general configuration or the impact at the 11 site or the general areas, because that's -- it's the 12 safety posture that counts, not the fun and games of 13 playing with titivating models.
14 That's my personal opinion in terms of 15 what ought to come out of the final presentation at 16 the end. That's all I have, John.
17 CHAIR STETKAR: And something to keep in 18 mind kind of along these lines of why did you do it, 19 what are you going to do with it, etc.
20 The executive summary does touch on the 21 fact that some of this, a good fraction of it, is 22 developing methods, models and tools that can be used 23 throughout the Agency in many other applications. So, 24 it isn't just it, the study, and developing expertise 25
164 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 within house on those.
1 And I think that the full committee --
2 MEMBER MARCH-LEUBA: We need to answer the 3
question now.
4 CHAIR STETKAR: No, but, I mean, I wanted 5
to kind of alert them that it isn't just it, this 6
thing.
7 MS. SANTIAGO: Right. We've talked about 8
that.
9 CHAIR STETKAR: Yeah.
10 MS. SANTIAGO: And it's a good point.
11 CHAIR STETKAR: Anything else on stuff for 12 the full committee presentation? You have a 13 challenge, clearly.
14 If not, I will ask if there are any 15 members of the public in the room who would like to 16 make comments. Please come up to the microphone and 17 do so.
18 And not seeing a stampede, I'll ask if 19 there are any members of the public on the bridge line 20 who would like to make a comment. Please do so.
21 The bridge line should be open. Please 22 identify yourself and make a comment.
23 (Pause.)
24 CHAIR STETKAR: That's a good indication 25
165 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 the bridge line is open. So, not hearing anyone, if 1
there's anyone out there, again, please make a 2
comment. Okay. We'll get that reclosed again.
3 As we always do in a subcommittee meeting, 4
I'll go around the table and ask for final comments by 5
each of the members.
6 Ron.
7 MEMBER BALLINGER: Well, I think we've 8
pretty much beat the dead horse in understanding what 9
went wrong with the computer runs and things like 10 that. So, that would be my only comment.
11 CHAIR STETKAR: Matt.
12 MEMBER SUNSERI: I have no other comments.
13 Thank you.
14 CHAIR STETKAR: Dick.
15 MEMBER SKILLMAN: No further comments.
16 Thank you.
17 CHAIR STETKAR: Dana.
18 MEMBER POWERS: I have an organizational 19 conflict of interest here, so I don't make comments.
20 CHAIR STETKAR: I got to remember that.
21 MEMBER POWERS: I don't believe in 22 eutectics and the uranium-zirconium oxygen system 23 based on a huge number of experiments.
24 MEMBER CORRADINI: I just thank the staff.
25
166 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 I do think, though, I guess my only emphasis would be 1
I would treat the other four or five members that 2
aren't here as your first shot at the intelligent 3
public.
4 So, cast it in that regard for them, 5
because you'll get "why" questions and what-do-you-6 use-it-for questions from some of the other members.
7 And I think that's kind of how I would at least try to 8
-- at least customize the beginning of your couple of 9
hours in front of us in full committee.
10 Other than that, I appreciate all that 11 you're trying to do to educate us for those that 12 forget what you've done in the past since it has been 13 a few years.
14 CHAIR STETKAR: Dennis.
15 MEMBER BLEY: Nothing more to add.
16 Thanks.
17 CHAIR STETKAR: Jose.
18 MEMBER MARCH-LEUBA: Nothing.
19 CHAIR STETKAR: Charlie.
20 MEMBER BROWN: Nothing more.
21 CHAIR STETKAR: Joy.
22 MEMBER REMPE: I also want to thank you 23 for your presentations and your efforts. I guess the 24 only point I'd like to emphasize is as you finalize 25
167 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 the executive summary, please consider caveats where 1
needed when you have conclusions and statements.
2 And I think I've tried to emphasize a 3
couple of examples today. Thank you.
4 CHAIR STETKAR: Now, as part of a -- let 5
me also say thanks. Thanks for all of the work that 6
you did between May or June or whenever and now. A 7
tremendous amount of stuff has been done.
8 I think the report benefits from it. I 9
think the understanding benefits from it. Although we 10 probably won't discuss it, I like Appendix I. I think 11 it's a confidence builder. And thanks for all of the 12 work.
13 I was going to say -- oh, the ACRS will 14 write a letter on this. The ACRS letter will be 15 written against the material we have and today. So, 16 there will be no update to the executive summary to 17 effect an ACRS letter or anything.
18 If you want to send us a little side 19 report to bolster confidence in the numbers for the 20 valves, that's fine, but we're not going to entertain 21
-- we can't because of our scheduling. So, we're 22 going to write a report on what we have and today.
23 I would ask, and I don't know how we're 24 going to do this given our other constraints, I really 25
168 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W.
(202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 want some input from the individual members on what 1
might go into a draft of our letter. And let's 2
discuss that offline.
3 With that, the meeting is adjourned.
4 (Whereupon, the above-entitled matter went 5
off the record at 12:07 p.m.)
6 7
8 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23
State-of-the-Art Reactor Consequence Analyses (SOARCA) Project: Sequoyah Integrated Deterministic and Uncertainty Analyses ACRS Subcommittee Briefing October 18, 2017 Patricia A Santiago, Chief Accident Analysis Branch Division of Systems Analysis NRC Office of Nuclear Regulatory Research
Schedule ACRS Full Committee on SOARCA Sequoyah Analyses -
November 2, 2017 Transmit SOARCA Sequoyah NUREG/CR report to the Commission and NRC publications - November 30, 2017 Updating SOARCA Surry Uncertainty Analysis - ongoing -
June 29, 2018 2
Core Team Members 3
MELCOR and severe accident progression: Kyle Ross, Jeff Cardoni, Chris Faucett, Troy Haskin, Randy Gauntt (SNL);
Casey Wagner (dycoda); Hossein Esmaili, Trey Hathaway, Allen Notafrancesco, Salman Haq, Ed Fuller (NRC)
MelMACCS: Nathan Bixler, Doug Osborn** (SNL); Trey Hathaway (NRC)
MACCS, consequence analysis and emergency response:
Nathan Bixler, Matthew Dennis, Joe Jones, Doug Osborn**,
Fotini Walton (SNL); Trey Hathaway, Jonathan Barr, Keith Compton, Todd Smith, Edward Roach (NRC);
UA methodology: Dusty Brooks, Matthew Denman (SNL); Tina Ghosh**, Trey Hathaway (NRC)
Accident scenario development: Selim Sancaktar, Jose Pires (NRC)
- Co-leads
- Focused pressurizer safety valve study
- New Appendix I in draft report
- Updates in response to ACRS members comments from June 6, 2017 subcommittee meeting
- Discussion of topics for presentation to full committee Outline 4
Focused Pressurizer Safety Valve Study Trey Hathaway Accident Analysis Branch NRC Office of Nuclear Regulatory Research
Focused Safety Valve Study
- Updated Sequoyah model run with sampled safety valve cycles less than 65 (13% of distribution) and open area fraction greater than 0.3 (40% of distribution)
- Most parameter distributions identical to UA
- Safety valve distribution constructed from sampled values of of UA and sampled on range of 1 to 65
- Open area fraction distribution sampled with lower bound of 0.3
- Calculation initially limited to 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br />
- All early failures were less than 15 hr
- Identical model to UA
- Sampled fabric seal failure pressure input corrected 6
- NOTE: This data was taken from the draft 2016 Sequoyah SOARCA analysis (ADAMS Accession No ML16096A374)
Figure I-1 Cumulative pressurizer safety valve open area fraction versus total number of safety valve cycles, where early containment failures are shown in red**
Early Containment Failure BOC MOC EOC Mean 6.0 6.6 6.6 Median 6.0 6.9 6.8 5th-Percentile 5.5 4.3 4.5 95th-Percentile 6.3 10.8 8.8
- Approximately 17% of the realizations in focused study resulted in early containment failure
- BOC: ~15%
- MOC: ~16%
- EOC: ~19%
7 Table I-1 Statistics on time (hours) of early containment rupture for BOC, MOC, and EOC realizations.
In-vessel Hydrogen Generation and Transport to Dome Up to First Deflagration
Figure I-13 Difference between peak containment pressure and sampled containment fragility plotted against maximum hydrogen reaching the dome around the time of burn initiation in the dome.
Hydrogen Deflagration Peak Pressure 9
Figure I-9 Time to containment failure plotted against the containment rupture pressure.
Early Containment Rupture Late Containment Rupture Cs release I release Cs release I release Mean 0.022 0.063 0.004 0.021 Median 0.022 0.06 0.002 0.014 5th Percentile 0.01 0.028 0.0003 0.004 95th Percentile 0.042 0.122 0.015 0.084 Figure I-14 Cesium release fraction versus time.
Figure I-15 Iodine release fraction versus time.
Table I-10 Statistics on Cesium and Iodine release fraction at 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> for the early and late containment rupture realizations.
Cesium and Iodine Release Fraction 10
Early Containment Rupture Late Containment Rupture Emergency Intermediate and Long-Term Emergency Intermediate and Long-Term Mean 2.4E-04 4.6E-04 7.6E-08 8.3E-05 Median 1.1E-04 4.2E-04 5.3E-08 6.7E-05 5th Percentile 8.5E-06 1.7E-04 2.5E-10 3.0E-09 95th Percentile 1.0E-03 1.0E-03 2.4E-07 2.2E-04 Table I-12 Statistics for the emergency phase, and intermediate and long-term phase contributions to the 0-10 mile individual LCF risk assuming LNT and conditional on the occurrence of a STSBO.
Figure I-18 0-10 mile Individual LNT LCF risk conditional on the occurrence of a STSBO as a function of containment rupture time.
Latent Cancer Fatality Risk
- NOTE: Only 20% of the late containment failure realizations were extended to 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />.
Early Containment Rupture Late Containment Rupture Mean 7.0E-04 8.3E-05 Median 5.8E-04 6.7E-05 5th Percentile 2.1E-04 3.2E-09 95th Percentile 1.5E-03 2.2E-04 Table I-11 Statistics for the 0-10 mile individual LCF risk assuming LNT and conditional on the occurrence of a STSBO.
Statistics on Total Risk Statistics on Emergency and Intermediate/Long-Term Risk 11
0-1.0 mi 0-1.3 mi 0-2.0 mi 0-3.0 mi 0-4.0 mi Mean 2.6E-06 1.7E-06 1.1E-06 3.8E-07 2.2E-07 Median 1.9E-12 5th-Percentile 95th-Percentile 2.0E-05 1.3E-05 6.8E-06 2.0E-06 Table I-14. Statistics for mean, individual, early fatality risk conditional on the occurrence of a STSBO and on an early containment failure (within 15 hr of accident initiation). Tabulated statistics are means over weather variability and express epistemic uncertainties in the MELCOR and MACCS input parameters.
0-1.0 mi 0-1.3 mi 0-2.0 mi 0-3.0 mi 0-4.0 mi Mean 2.3E-08 1.5E-08 9.7E-09 3.4E-09 2.0E-09 Median 5th-Percentile 95th-Percentile Table I-15. Statistics for mean, individual, early fatality risk conditional on the occurrence of a STSBO. Tabulated statistics are means over weather variability and express epistemic uncertainties in the MELCOR and MACCS input parameters.
Early Fatality Risk 12
Updates in Response to ACRS Members MELCOR Comments Casey Wagner, dycoda, LLC Tina Ghosh, Accident Analysis Branch, NRC Office of Nuclear Regulatory Research
Fabric Seal Failure Criteria 14
- ACRS member comments concerned identified error, confusing description of the failure parameter, and seal failure area
- New documentation added on the impact on early containment
- Flow through the ice chest and other leakage pathways allow hydrogen transport to the dome prior to the first burn
- Focused valve study includes fabric seal correction
- Comparisons to draft 2016 UA results show continuous behavior
- New documentation added on impact on late containment failure
- Intact seal failed following the first hydrogen burn in the dome
Fabric Seal Failure Criteria 15
- Fabric seal parameter description was re-written to clarify UA parameter sampling versus the deterministic thermal-mechanical failure criteria
- Sampled UA parameter is for the cold failure differential pressure
- Deterministic non-sampled failure criteria for elevated temperature conditions
- ACRS member questioned whether seal failure area is different for mechanical over-pressure versus due to elevated temperature
- Range of failure areas allowed robust natural circulation
- Failure area was not important parameter in regression results
Ice Condenser Doors
- Sampled AJAR parameter controls lower ice chest door open area after fully opening
- Five separate flow paths (Only 46 Pa required to fully open the doors)
- AJAR flow paths are large and not the controlling flow resistance when
<5 flow paths are open
- New UA discussion illustrating range of door behaviors
- Analysis from the focused SV study and UA shows all 5 MELCOR flow paths usually satisfy AJAR criteria when there is a hot leg failure
- Partial (<5 flow paths) or no AJAR is the usual response when there is no hot leg failure 16
Hydrogen Ignition and Burn Direction
- Known ignition sources
- Hot gases from the hot leg and the PRT
- Ex-vessel debris
- Location of the ignition within the compartment has uncertainty
- May not be near the hot leg, PRT, or ex-vessel debris
- Dependent on local gas concentrations, mixing, and strength of the ignition source, etc.
- New discussion identifies flame directions as it propagates within the compartment as surrogates for uncertainties in the ignition location 17
Safety Valves
- Added discussion of how the number of demands (denominator) is calculated in NUREG/CR-7037 operating experience data
- Revisited publicly available information from Licensee Event Reports on the main steam safety valve failure-to-close events, and SMEs summary of these events, but did not add anything further to the report
- Added discussion on rationale for not including failure-to-open SV failure mode 18
Time in Cycle Sampling
- Two-part approach 1.
Improved the discussion of the distribution between BOC, MOC, and EOC in the Sequoyah documentation 2.
New Surry UA uses more continuous sampling of inventory &
decay heat throughout the fuel cycle
- New Sequoyah discussion includes
- New ORIGEN work to characterize selected Sequoyah values within the continuum of the fuel cycle
- BOC sample space defined with consideration of decay heat build-up and development of secular equilibrium inventory of I-131
- BOC investigates low decay heat response
- EOC sample space is developed with consideration of long-lived Cs-137 inventory
Other Additions
- Statistical stability analysis conducted for figures of merit and documentation added in Appendix A
- Recognized possible impact of eutectic melt temperature on burnup but also inadequate data to quantify
- Added discussion on approach for failed computer calculations
- Added discussion of ice bed response during seismic event
- Improved vessel failure discussion to address cited issues
- LTSBO discussion improved to address limitations on asymmetric SG response
- Added discussion/plots on RCP leakage as an ignition source 20
Other Additions
- Comparison of new UA results with draft 2016 results with different valve failure distributions
- Presented last ACRS meeting
- Focused pressurizer SV study results 21
Updates in Response to ACRS Members MACCS Comments Nathan Bixler, Sandia National Laboratories
Emergency Response Timing Question - Is emergency response normally triggered by GE?
Response - Verified that emergency response modeling is faithful to TEMA planning Resolution
- Final Exercise Report Sequoyah Nuclear Power Plant (FEMA, 2004) timeline lists sequentially
- SAE declaration
- GE declaration
- Notes from conversation with TEMA state that upon SAE, schools are evacuated to paired schools outside EPZ
- Discussion added to final document 23
Humidity
- Question - How does MACCS treat deposition under humid conditions?
- Response - authors indicated that humid conditions could cause hygroscopic aerosols to grow and deposit faster, but MACCS does not treat this mechanism
- Resolution
- Performed bounding estimate to show that deposition velocity of hygroscopic aerosols could increase by less than a factor of 2 and this is within the uncertainty range
- Expanded discussion in section on uncertainty of deposition velocity 24
Presentation of Risks
- Comment - Improve presentation of EF and LCF risks
- Response - Added discussion of bimodal nature of LCF risk, minimal dependence of risk on distance, and reconsider discussion of EF risk
- Resolution
- Discussion of LCF risk was expanded to address comments
- Figure and table showing EF risk was eliminated 25
DISCUSSION OF TOPICS FOR FULL COMMITTEE PRESENTATION 26
27 References SECY-12-0092, State-of-the-Art Reactor Consequence Analyses -
Recommendation for Limited Additional Analysis (July 2012)
NUREG-1935, State-of-the-Art Reactor Consequence Analyses (SOARCA)
Report (November 2012)
NUREG/CR-7110, Vol. 1, SOARCA Project Peach Bottom Integrated Analysis, Rev. 1, (May 2013)
NUREG/CR-7110, Vol. 2, SOARCA Project Surry Integrated Analysis, Rev. 1 (August 2013)
NUREG/CR-7008, MELCOR Best Practices as Applied in the SOARCA Project (August 2014)
NUREG/CR-7009, MACCS Best Practices as Applied in the SOARCA Project (August 2014)
NUREG/CR-7155, SOARCA Project Uncertainty Analysis of the Unmitigated Long-Term Station Blackout of the Peach Bottom Atomic Power Station (May 2016)
NUREG/BR-0359, Modeling Potential Reactor Accident Consequences, Rev. 1 (December 2012, update in progress)
Acronyms & Abbreviations AC Alternating Current BOC Beginning of Cycle CCDF Complementary Cumulative Distribution Function CCI Core Concrete Interactions CDF Core Damage Frequency CST Condensate Storage Tank DC Direct Current EOC End of Cycle EPZ Emergency Planning Zone EF Early Fatality GE General Emergency HL Hot Leg FLEX Diverse and Flexible Coping Strategies FTC Failure to Close FTO Failure to Open LCF Latent Cancer Fatality LNT Linear No Threshold LTSBO Long-Term Station Blackout MACCS MELCOR Accident Consequence Code System MCR Main Control Room MELCOR Not an acronym - accident progression code MelMACCS MELCOR to MACCS Source Term Converter MOC Middle of Cycle 28 MSIV Main Steam Isolation Valve NTTF Fukushima Near-Term Task Force PDF Probability Density Function PGA Peak Ground Acceleration PRA Probabilistic Risk Assessment PRT Pressurizer Relief Tank PZR Pressurizer RCP Reactor Coolant Pump RCS Reactor Coolant System RLZ Realization RPV Reactor Pressure Vessel RtePM Real Time Evacuation Planning Model SBO Station Blackout SG Steam Generator SAE Site Area Emergency SIP Shelter in Place SME Subject Matter Expert SNL Sandia National Laboratories SOARCA State-of-the-Art Reactor Consequence Analysis STSBO Short-Term Station Blackout SV Safety Valve TDAFW Turbine Driven Auxiliary Feedwater System TVA Tennessee Valley Authority UA Uncertainty Analysis