ML20134J322

Transcript of 961106 ACRS Joint Meeting of Matls & Metallury & Severe Accident Subcommittees in Rockville,Md.Pp 272-384

ML20134J322
Person / Time
Issue date:	11/06/1996
From:	Advisory Committee on Reactor Safeguards
To:
References
ACRS-T-2080, NUDOCS 9611150153
Download: ML20134J322 (149)

Text

{{#Wiki_filter:-- Officirl Trcnscript cf Praccadings NUCLEAR REGULATORY COMMISSION .O flCAS7-209c

Title:

Advisory Committee on Reactor Safeguards Joint Meeting: Materials and Metallurgy and Severe Accident Subcommittees -Docket Number: '(not applicable) TRQ^ E R INAL TO BJWHITE M/S T-2E26 415-7130 THANKS1 Location: Rocl~ "lle, Maryland.:.. i eO 1 Date: Wednesday, November 6,1996 1 g 11 g 3 961106 T-2000 pga 1 Work Order No.: NRC-900 Pages 272-384 n y ! p qj\\4 [g j 3,0001 di\\ !O! bA L NEAL R. GROSS AND CO., INC. Court Reporters and Transcribers 1323 Rhode Island Avenue, N.W. gk Wash!ngton, D.C. 20005 if-d pn24234-4433 hnl3 n .,Ce v0]y-lee _r, a u J0f:'e._?e Of:T0 307 itjee

...--__.._______.,.._._.._._-_.___m__.., W O DISCLAIMER t PUBLIC NOTICE BY THE UNITED STATES NUCLEAR REGULATORY COMMISSION'S i ADVISORY COMMITTEE ON REACTOR SAFEGUARDS NOVEMBER 6, 1996 The contents of this transcript of the a proceedings of the United States Nuclear Regulatory Commission's Advisory Committee on Reactor Safeguards on NOVEMBER 6, 1996, as reported herein, is a record of the discussions recorded at the meeting held on the above.3 ate. This transcript has not been reviewed, corrected 4 and edited and it may contain inaccuracies. i 4 O NEAL R. GROSS COURT REPORTERS ANDTRANSCRIBERS 1323 RHoDE ISLAND AVENUE, NW (202) 234-4433 WASHINGTON. D.C. 20005 (202) 2344433

272 1 UNITED STATES OF AMERICA 2 NUCLEAR REGULATORY COMMISSION ,s a 3 +++++ 1 4 ADVISORY COMMITTEE ON REACTOR SAFEGUARD (ACRS) l 5 +++++ 1 6 JOINT MEETING 7 MATERIALS AND METALLURGY 8 AND 9 SEVERE ACCIDENTS 10 SUBCOMMITTEES i 11 +++++ 12 WEDNESDAY 13 NOVEMBER 6, 1996 \\p G'r 14 +++++ 1 15 ROCKVILLE, MARYLAND I 16 +++++ 17 The Subcommittees met at the Nuclear 18 Regulatory Commission, Two White Flint North, Room T2B3, 19 11545 Rockville Pike, at 8:30 a.m., Robert L. Seale 20 (Chairman, Materials and Metallurgy Subcommittee) and 21 Mario H. Fontana (Chairman, Severe Acc.idents Subcommittee) 22 presiding. 23 COMMITTEE MEMBERS: 24 ROBERT L. SEALE, Chairman, Metals & Metallurgy A{} 25 MARIO H. FONTANA, Chairman, Severe Accidents l NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433

273 1 COMMITTEE MEMBERS (continued) : 2 ' '/ s "~ 3 THOMAS S. KRESS 4 DANA A. POWERS 5 WILLIAM J. SHACK 6 7 ACRS STAFF PRESENT: 8 NOEL F. DUDLEY 9 10 ALSO PRESENT: 11 JACK STROSNIDER 1 l 12 JOSEPH DONOGHUE 13 STEVE LONG [ _) 14 'JEFF GORMAN 15 ROBERT PALLA 16 RAY SCHNEIDER 17 CHARLIE TINKLER 18 ROBERT JONES 19 20 21 22 23 24 ,a S l NEAL R. GROSS COURT REPORTERS AND TR*NSCRIBERS i l 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433

274' 1 A-G-E-N-D-A 2 Acenda Item Pace ,s i 3 Introduction, Chairman Seale 275 i 4 Results of Risk Analyses, Joe Donoghue 276 5 Discussion 366 ti 7 8 9 10 11 12 13 f3 14 15 16 17 18 19 20 21 22 I 23 24 ,/ 3, Q_) 25 l NEAL R. GROSS I COURT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISLAND AVE., N.W. l (202) 234-4433 WASHINGTON, D C. 20005-3701 (202) 2344433 l

275 1 P-R-O-C-E-E-D-I-N-G-S 2 (8:31 a.m.) l ,,h / 3 MR. JONES: Joe Donoghue will be speaking for 4 the staff. 5 CHAIRMAN SEALE: The meeting will now come to 6 order. This is a meeting of the ACRS Joint Subcomm'ttee 7 on Materials and Metallurgy and Subcommittee on Severe 8 Accidents. I am Robert Seale, Acting Chairman of the 9 Subcommittee on Materials and Metallurgy. i 10 The ACRS members in attendance are Mario 11 Fontana, Chairman of the Subcommittee on Severe Accidents, 12 Tom Kress, Dana Powers and William Shack. 13 The purpose of this meeting is to continue (y (s) 14 discussions with representatives of the NRC staff, the j 15 Nuclear Energy Institute and EPRI to gather information 16 concerning the technical approach used in developing the 17 proposed risk-informed performance-based rule and 18 regulatory guide associated with steam generator tube 19 integrity. 20 The committee will gather information, analyze 21 relevant issues and facts, and formulate proposed 22 positions and actions as appropriate for deliberation by 23 the full committee. Noel Dudley is the cognizant ACRS 24 staff engineer for this meeting. A(,) 25 The rules for participation in today's meeting NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 2344432 WASHINGTON, D.C. 20005 3701 (202) 234-4433

-_._.__.m._.-.___._.--_. l 276 i l' have been announced as part of the notice of this meeting l l i ( 2 previously published in the Federal Reaister October 12. l '( ) 3 The transcript of the meeting is being kept and will be l

4. made available as stated.in the Federal Recister notice.

I l-l L -5 It is. requested that the speakers first identify -l 6: themselves and speak with sufficient clarity and volume so l i 7 that they can be readily heard. 8 We received no written comments or requests I 9 for time to make oral statements from members of the l l 10 public. j 11 During the June 12-14, 1996 ACRS meeting, the \\ l 12 committee heard presentations by representatives of the t i l l 13 staff in the Nuclear Energy Institute on this matter. i 14 Today the subcommittee will hear from the staff concerning 15 the technical basis for the proposed rule, j i l 16 We will begin the meeting. I guess Joe, j i 17 you'll start things off for the staff. Is that correct? i 18 MR. DONOGHUE: Yes, sir. l 19 CHAIRMAN SEALE: Mr. Donoghue. l l~ 20 MR. DONOGHUE: At this time, yes. I am coming 21 through okay? i 22 My name is Joe Donoghue. I am in Reactor l 23 Systems. What we wanted to do this morning was to step i l 24 through the example calculation that we've done to assess () 25 severe accident risk of tube failure. s I NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISt.AND AVE., N.W. (202) 234M33 WASHINGTON, D.C. 20005-3701 (202) 234-4433

277 1 To start with, let me show you the list of 2 things I'm going to talk about. I wanted to start off , ~s ( ) 3 just with a general discussion of what contributes to tube 4 rupture risk. Then our emphasis again, the severe 5 accident contribution. These bullets go through the 6 thought process and significant portions of the analysis, 7 the example analysis that we've done. 8 What we plan on doing is having the people 9 here that are key players in each of those areas, so that 10 if we get into a lot of detail discussion, they can p ;ch 11 in. In the interest of time, we weren't planning on 12 having separate presentations. I'll try to go through as 13 much of that material as I can. vl 14 Back in June, we basically said that there's 15 two ways to have tube ruptures, either spontaneously or 16 induced by some means. We said at the outset that 17 spontaneous failures, we didn't expect the frequency to 18 change under the rule. ) 19 Under induced failures, we had listed pressure 20 and thermally induced failures and contributors. There 21 was a question from the committee about the potential for 22 mechanically induced failure. I'll address that in the 1 23 next slide. 24 Just to repeat a couple of things from June, (_) 25 when we looked at pressure induced failure, there's a NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D.C. 20005 3701 (202) 234-4433

278 1 chance in an ATWS you get to a high pressure that could 2 challenge the tubes, but the frequency of that event is m 306 i 1 not confused. The R or the N corresponds with,_for j i !L -2 example 6N is NRR-6 up here. 3 This is.just, again, these are the scenarios 1 l 4-we had available to us to use over the summer when we were 5 trying to quantify the event tree and get the tube failure I 6 probability numbers generated. I l 17 I'll' talk later about some studies that l 8 research has-done to understand the sensitivity of the l ) 9. thermal hydraulic results. We understand that there's l 1 10 some changes to the results from some of'these. cases. 11 We'll have to -- we have taken a quick look.at some:of 12 that information, but that's going to have to be l13 incorporated into our risk study. 14 But as I say down here, jumping around on my. 15. slide already, the results appear -- the results that we 16 used for this risk study appear somewhat conservative 17 compared to what newer information may be telling us..But l 18 we just'needed to take a closer look at it to see what the L 19 implications really are. 20 When we do the tube' failure analysis, I'll -21 point this out in a little more detail. But what we did 22 was estimated tube. conditions,-but the failure 23 probabilities is calculated relative to the failure b .24 probability of other components, other major components in f' 25 the RCS,. namely the surge line and the hot leg. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 2344433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 -.i.

i 307 J i 1 SCDAP/RELAP, as you understand, I think l 2 calculates a creep damage index for those components, for 3 the tubes, to surge on a hot leg. What we did was use 4 Larson-Miller modeling for that creep failure to try to 5 understand the failure potential of those components i 6 relative to one of the -- 1 l 7 MEMBER POWERS: Is Larson-Miller an 8 appropriate thing to use for the large section? i 9 MR. DONOGHUE: For the hot leg? i l 10 MEMBER POWERS: Yes. 11 MR. DONOGHUE: The way it was done in SCDAP 12 arguably could be done better. There has been some 13 discussion of the staff on ways to do a creep failure 7m i (_/ 14 calculation using more than just one thickness of the 15 pipe. But for our purposes, since we're not considering 16 really exotic materials for the hot leg or the surge line, 17 and the information is well developed, we think that using 18 Larson-Miller is an adequate waf to do this. I don't 19 think we are going to gain a lot by -- 20 MEMBER POWERS: Is Larson-Miller really well 21 developed for failures? Are these relative to the high 22 temperatures for thick section steel that's not in a 23 uniaxial loading? 24 MR. DONOGHUE: Well, we had some information r~s ( ) 25 from previous studies at high temperatures that was used %d NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4432 WASHINGTON, D.C. 20005-3701 (202) 234-4433 i

~. 308 1 to -- and I think only show you later on the tube 2 unfortunately. But there were uncertainty bounds put on .O G 3 the Larson-Miller parameters for both the surge line and-4 the hot leg at these temperatures, I think over a 5 temperature range. It seemed like it was -- I can give 6-you some more information on that because we do have some 7 more detail studies that were done to characterize Larson-I .i 8 Miller for us. I don't have the material with me. i : MEMBER POWERS: I guess I would like to see it 10 because I know that in looking at the BWR dry well 11 failure, which is actually a less diffjeult case than-l 12 this, that elaborate effort was made to improve upon the 13 Larson-Miller formulation simply because they didn't have (h 14 a good data base at the temperatures where failure would R '15 occur. They worried a lot about it not being.in a 16 uniaxial load. 17 MR. DONOGHUE: That second point has 18 definitely been discussed, but just to talk about the l 19 Larson-Miller modeling by itself, I think we have some 20 information I could show you to at least tel)-you what 21 temperature range was considered. L 22 Now as far as what the loading on the 23 particular components was, the SCDAP/RELAP certainly j 24 doesn't get you that far. It is something we have '() 25 discussed in our NUREG that we are putting together to lay NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-370) (202) 234-4433 ~v

309 1 out all this analysis. 2 There is at least a qualitative discussion of .s i e ' \\,/ 3 that. I think we leave it open for a further analysis. 4 I'll talk later on about the sensitivity 5 studies that were done which are very recent studies on 6 the thermal hydraulic results. That just gives you an 7 outline of what was used to characterize the.end states of 8 the event tree. 9 Getting into some of the details that give us 10 the conditional tube failure probability itself, one of 11 the key parts of this is the modeling of tube failure. 12 This is part of the analysis which we feel the best about. 13 There's been a lot of work done here, very recent work, t \\> 14 but it seems like.it's quite reliable. l 15 It's based on high temperature testing of 16 machine flaw tubes under pressure. When I say here that 17 we're using the original thermal hydraulic analysis, as I 18 mentioned earlier, we have done thermal hydraulic analysis 19 and when questions came up, re-analyses were performed. 20 At the time that the tube testing was being done earlier I 21 this year, there was a question on what should be used. l 22 I think when Dr. Muscara was here in June, he 23 showed you two temperature ramp rates, one generated by 24 our contractor, another one from a report submitted by /' \\ (,) 25 EPRI. I think we can still say that the model can handle l NEAL R. GROSS i COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. l (202) 234-4433 WASHINGTON, D C. 20005-3701 (202) 234 4 33

'310 1 a wide range.of temperature and pressure variation. 2 The' latest'results on the thermal hydraulic 31 analysis show us a somewhat lower temperature, a slightly 4 different ramp rates and I think right now people are 5 assessing to see if there is any kind of significant 6 change to what the_ tube failure model would tell you, but ~ 7 I think the first indications are that the testing that 8 was done covered a wide enough range of temperatures and 9 heat-up rates that we should be able to cover even the 10 latest thermal hydraulic analysis. They weren't -11' drastically different, is'the bottom line, of what we had 12 before. 13 The test results that went into this model ll 14 that we're using for our study only considered axially-l 15 oriented cracks. There was a question earlier about did-16 we consider other degradation types. The tube failure 1 17 mo6 1 as it stands did not, although there is work under 18 way to characterize and model circumferential failure. 19 What was found was that under the temperature 20 ramp rates at pressure ramp rates that we saw, the creep 21 failure mechanism was dominant. Earlier on, when we first 22 started doing this work before testing was done, a flow -23 stress model was used. The evaluation and' test data 24 showed us that that model wasn't really valid under the 25 conditions we were seeing. i NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234 4 33 WASHINGTON, D.C. 20005-3701 (202) 2344433

311 1 Also some unflawed tubes were included in the 2 test matrix. That showed that they could withstand the i,,, \\ 3 kind of conditions that we saw from the thermal hydraulic 4 analysis. 5 I'll show you on the next page, the very 6 detail of what I am saying here, that the model includes 7 the uncertainty of the effects of the crack and the 8 material availability on creep failure predictions. 9 CHAIRMAN FONTANA: I don't know if I missed 10 something, but that thermal challenge from the thermal 11 hydraulic model, what sequence did you say that was? That 12 wasn't one of the cleared loop sealed? 13 MR. DONOGHUE: Oh, no, no, no. That's right. (.. i s (s/ 14 This was the station blackout case with one steam 15 generator depressurized. That was the ramp rate that was 16 used. 17 CHAIRMAN FONTANA: Okay. 18 MEMBER SHACK: And it only stands in the sense 19 that it doesn't fail first? 20 CHAIRMAN FONTANA: Yes. Right. Correct. 21 MR. DONOGHUE: If you hold it up at a long 1 22 enough temperature, it will eventually creep fail. 23 I think I tried to say that caveated with 24 understanding that we're considering the event of ps{,) 25 interest, not necessarily getting to that condition and NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. l (202) 234-4433 V!ASHINGTON D.C. 20005-3701 (202) 234-4433 l

312 1 holding.it for more than several hours. 2 I tried to put up enough information to O 3 explain the tube failure model, but not enough so we're -. 4 getting-in trouble I think. 5~ The basic prediction, this is in SCDAP/RELAP, 6 integrates the time of the event, ratios into'the time, to 7-predict the time of failure for the component which is a 8 function of the temperature, there's time of-rupture right '9 there. Function of the temperature and the Larson-Miller 10 parameter. 11-The Larson-Miller parameter in this equation i' 12 is given right here. It's slightly different from what's 13 in SCDAP/RELAP. This is what we from the latest C 14 assessment of pre-failure information fol. the tube 15 material was shown as -- it was compe, red to what =16 SCDAP/RELAP predictions would be, which are not to be too 17 different. 18 This includes, I'll point out right away here, 19 this includes variability. This is at the 95 percent ' 2 0' bound for the material variability. You will see that 21 includes a stress, a hoop stress for the tube right there. i 22 So that's where you see time to' rupture is a function of l-23 stress. It's also a function of this multiplication 24 factor, M sub P, which I show right there, crack

i f 14 from the offices that generated them, are different, are 15 quite different in that there are six separate 16 distributions characterizing a distribution under the 17 research model. 18 Under NRR's, several types of degradation were 19 folded into one distribution for axial cracks only. There 20 are two specific flaw lengths that were used, a half inch 21 and an inch. Under the research flaw distribution here, 22 the functions were given as continuous over flaw length 23 and depth. 24 Now I'll throw the -- ( -) 25 CHAIRMAN FONTANA: Was this deliberately done NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 - -.~. - 319 l 1 so you have two parallel activities or did it just happen? 2 MR. DONOGHUE: Well, the question about flaw-3 distribution is not easy to answer. The first example 1 I 4 that we tried,-that we started using left us with a lot of. 5 questions that'I will get-to in just'a second. But a j 6 parallel effort was undertaken in the Office of Research. l 1' 7 What we tried to do was not supplant one flaw 8-distribution with another, but trying to consider both of l 9 them because both have strengths. We are going to have to l 10 say at the.end of this analysis, that the best thing to do 11 is have a plant-specific flaw distribution if you can j 12 generate it. l 13 That's a tough question to answer though'. 14 Like-I say, the inspection capability doesn't necessarily 15' get you there. l 16 CHAIRMAN FONTANA: But you tried the NRR one 17 first, the one on the righthand side? 18 MR. DONOGHUE: Right. 19 CHAIRMAN FONTANA: I see. 20 MR. DONOGHUE: Right. That was the first one 21 we had. We knew the work was underway with the research, .I 22 as they call it, distribution. L r ~ We have shifted to use the RES' study, does use l - 23 24 the.results from -- I'll show you exactly which part of 25 the results from this research distribution, for reasons NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 2344433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 - i l i - -._.__..~ _ _ ~. __

320 1

that some people might dispute, but that's the one that 2 we're using at the moment, although we' haven't done away O 3 with the NRR distribution. We still want to understand '4 why we have the differences between them. 5 MR. LONG: This is Steve Long with the Risk 6 Assessment-staff. 7 Part of this was just trying to deal with the 8 information that we could get at the time. This is .9 essentially the fourth shot at'the distributions. There 10 were three that came from-NRR. It helpedLto have two 11 different thought processes going on, because this is a 12 very uncertain area and'it gives us a little bit of a 13 chance to see the sensitivity from difference in the 14 perspective of how you would generate a flaw distribution. 15 I think.he has some graphs that will give you an 16 indication that that was really worth doing. 17 MR. DONOGHUE: Since I mentioned we were 18 concentrating lately on the research flaw distribution, 19 I'll just show you what'the degradation types were. ) 20 One from circumferential cracks, TTS top of l 21 tube sheet, that's stress corrosion cracking on top of 22 tube sheet. ODSCC at tube support plate dents. Freespan i 23 ODSCC. IGA and stress corrosion cracking of the hot leg j i l 24 sludge pile. Axial ODSCC at tube support plates, and then 25 flaws from loose parts. 4 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE.. N W. (202) 2344433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 ) l l _. _ = _ _ _ _ _ _. _ _ _. _ ( 321 l l 1 For.a second there, I thought I indicated the I 2 wrong ones that we used in the study, but I think it's' 3 right. Again, we were able to characterize axial cracks l 4 so we don't include circumferential cracks. l i 5 Earlier on it-was pointed out that the ~ 6 freespan cracks are the largest part of the population of 7 flaws. Then we knew that the temperatures in the tube j 8 bundle for these situations was going to be highest, right 9 in the hot leg sludge pile area..So those:two parts of 10 this research distribution are-the ones that were actually 11 used for our RES study. 12 Now.I'll put up a couple of graphs here. It's i 13 maybe unfair to put this one up before I'put up the actual (~ \\. 14 example of'the research distribution, but so beLit. 15' This is a comparison of the two distributions. 16 What was done was the two of the six were force fit, just 17 represent flaws greater than one inch. That is what you l 18 see here. You see number of cracks versus the through-l 19 wall depth. l 1 20 One thing that's obvious is that there's some j ] 21' large disparity here for shallow cracks. The basic 22 discussion goes that with this distribution, the l '23 assumption is made that you really can't protect those i i 24 cracks below about 40 percent through the wall, and that l () 25 that number is conservative obviously. i NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS I 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON. D.C. 20005-3701 (202) 234-4433 [ ._..y y 322 l' ' Down here, this above 20 percent, the 2' probability of detection was somewhat more favorable for ty/ 3 this characterization, and we're able to see some smaller 4 than 40 percent through-wall cracks. But this is probably .5-not. realistic either. It seems like it's probably a lot 6 lower than it should be. i 7 But what I want to point out for our purposes 8 here of understanding thermally induced failure, what we 9 are really worried about are flaws between about 60 and 80 i 10 percent, in that. range. ~ 11 We don't see a big contribution from the'small 12 flaws. So we are not too worried about that difference 13 for our purposes. 1 14 You can see that there is a bigger population 15 of these flaws of concern for the research distribution, 16 which is one reason that I pushed us to use it. I will l: 17 show you some specific numbers for the tube failure i. 18 probabilities based on these later, and show you why they-i (~ 19 make sense. L l 20 It makes a little more sense right now to use L 21 that distribution, but the difference is in the shape I l 22 here. The gamma distribution versus this kind of i 23 exponential function. We have to understand the bases for 24 these distributions are somewhat different. We have to 25 look into more why those differences come about to really l NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS l 1 1323 RHODE ISLAND AVE., N.W. (202) 2344433 WASHINGTON, D.C. 20005-3701 (202) 2M33 i ....~ - i 323 l 1 decide on which is the best way to characterize the [ Og. 2' representative flaw distribution. 3 But what was comforting was that there wasn't i 4 a large disparity in the area that we were concerned 5

about, i

l I L 6 CHAIRMAN FONTANA: Roughly how many samples l 7-are we talking about here? I 8 MR. DONOGHUE: I would have to go back and [ l' .l 9 look at the report, but the author is back. He might-be .l 10 able to tell you. i l 11 CHAIRMAN FONTANA: On the order of thousands j. 12 or hundreds or tens? t 13 MR. DONOGHUE: Oh, okay. I think the.research j l l l l 14-distribution was based on a large thousands of tube l i l 15 samples or inspection data points. Is that right? i L I l-16 MR. GORMAN: This is Jeff Gorman, Dominion 17 Engineering. The freespan indications were based on large 1 18 scale rotating probe inspections of Palo Verde 2,.and the 19 numbers of tubes inspected ~I think were in the order of l 20 4',000 tubes'per steam generator. The number of flaw::; 21 detected, if my memory is correct, was in the order of 600 l' 22 to l,000. .23 The number of tubes that were pulled and 24-destructably examined were smaller to verify the NDE P 25 results. f i 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 i iu ~ - _ _ -. ~. ~. ~._..... - -.... - 324 1 But the data that were presented are based on f 2 the rotating probe inspection results, so in the order of \\ 3 600 to 1,000. 4 CHAIRMAN FONTANA: Thank you. t 5 MR. DONOGHUE: I just wanted to show you this 1 6 is straight out of the Dominion Engineering report. This 7 is just'a freespan defect distribution. You can see 8 again, it's a gamma distribution. I don't'want to spend a j l 9 lot of time on it. It's just to show you the raw data f j-10 that got folded into that last slide I showed.you. This i 11 whole report is going to be a reference in the NUREG that { 12 we are putting together. j l 13 Now what had to be done was to.take the i Q' l , D 14-continuous distribution of length and understand where the 15 break points were for certain lengths, because that's 16 where the tube failure probability calculation is done, in 'l J 17 different bins. I'll explain that in a minute. 18 But you can see above an inch, there's a 19 relatively small fraction of flaws in these distributions 20. that we use. We used bins that were from one inch and l 21 greater, between a quarter inch and one inch. Those are g 1. 22 the flaws of concern for these situations. This just 23 gives you an idea of how they would bend and what the 24 contributions were. i-25 Now using the flaw distributions and the NEAL R. GROSS - COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. 4 (202) 234 4433 WASHINGTON, D.C. 20005-3701 (202) 234 4433 I 1 325 1 thermal hydraulic analysis results that gave us an 2 understanding of the creep rupture potential, a program ' 's/ 3 was put together to calculate tube failure probability. 4 You already know how the event tree was put 5 together, and that we had to include information for 6 pressure and thermally induced failure. But what was done 7 with this model was to actually take the information -- 8 I'm trying to find a slide to show you what we start with 9 here. SCDAP/RELAP calculates that creep damage index. 10 This is just an example. This is just one of 11 the cases. This is not in your handout. Just a backup to j 12 help illustrate. Over the event time, Larson-Miller is -- 13 SCDAP/RELAP is calculating the Larson-Miller creep index, /y i ) 14 the integral that I put up earlier. 15 Eventually, the conditions are right that the 16 component is going to experience creep failure when you 17 get to one for that damage index. 18 What was done was the program calculates the 19 probability of failure over time. I'll show you another 20 slide in just a few minutes, but this is what we start 21 with. We start with the creep damage index. We have to 22 understand what's going to change it, whether the 23 variability of materials is going to change one way or the 24 other to change this prediction, the thermal hydraulic ,/ 3 (,) 25 analysis results have some kind of variability that could NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 2344433 l 326 l ) 1 change these, and so forth. i I 2 .So.that's in essence what the tube failure j Lg m,1 - 3 probability calculation is doing, is understanding how the i i l 4 relative -- how these creep predictions for the hot leg, t l. 5 the tube, or the surge line to the tube predictions would ') l 6' change over time. l c I i l 7 MEMBER POWERS: Could you explain a little i i 8 more about the legend on that figure? 9 MR. DONOGHUE: Okay. This particular -- l l 10 MEMBER POWERS: I mean in SM equals two. j l 11 MR. DONOGHUE: Right. This particular figure j 1 i 12 was showing us the effect of a crack that would have a l l 13 stress multiplier to keep changing our variables.: I'm .,e~\\ i k-l 14 sorry. But stress multiplier, M sub P that we had of two. I i 15-That represents I think a 60 percent through-wall half-16 inch crack, something on that order. -17 Then we have the creep ~index,.the creep damage 1 l 18 index that was produced by SCDAP/RELAP or R/S, somebody l 19 used RELAP/SCDAP as an abbreviation there. l 20 Sc you see what was actually generated by i L 121 SCDAP/RELAP in the dotted or dash lines, and then what 22 program that we used that calculated the effect of the l I 23 crack on that tubes is a solid line, ,v. 24 MEMBER POWERS: And what you conclude from all 25 of this is that almost no matter what you do, these things NEAL R. GROSS i I COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 327 1 are hitting their creep damage index over a period of 2 perhaps 20 minutes, 10 minutes? ,_s I ) \\ 3 MR. DONOGHUE: This is I think in this case it 4 was about 20 minutes. If I put up the -- this is a space 5 of 20 minutes, roughly. That was pretty consistent with 6 the other cases that we saw. Exactly. 7 If one of these was skewed off, and for 8 example, when vessel failure was considered, that was at 9 such a different disparate time from these other 10 components, that this kind of analysis really was not 11 warranted. 12 But uince these things are pulling around on 13 top of each other, this kind of calculation for tube /m i \\ (m/ 14 failure probability is really needed. 15 MEMBER POWERS: It's just very difficult to 16 believe that you can calculate something in a severe 17 accident to plus or minus 20 percent -- I mean 20 minutes. 18 MR. DONOGHUE: Well, I am just showing you 19 here is the effect of one variable. 20 MEMBER POWERS: It's a very illuminating 21 graph, yes. 22 MR. DONOGHUE: As I say, there are several 23 components. There's nothing about flaw distribution in i 24 here yet. This is just an assumed flaw on one tube. O* 25 MEMBER SHACK: But what Dana is pointing out iw] NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D C. 20005-3701 (202) 234 4433 328 1 is you go from the unflawed tube to a fairly severely 2 cracked tube, it's 20 minutes. ( ) 3 MR. DONOGHUE: Right. I am not including any 4 variability of thermal hydraulic analysis results. The uncertaintyanalysisthathavejustbhendoneisnotpart 5 6 of this. 7 MR. LONG: This is Steve Long again. If you 8 can think of the lines there that were originally 9 calculated by RELAP/SCDAP as fuzzy areas, as some sort of 10 distribution in probability, and then what we do is try to i 11 overlap those distributions and ask what's the i 12 probability, at least as far as we know, in the variation 13 of these various lines, that essentially the tubes would tm ( ) ks' 14 fail before any other part of the pressure boundary. l 15 That's the technique that we used. 16 Then we have to do that for tubes that have l 17 different sizes of cracks. Then we have to ask how 18 uncertain are we about the effect of those cracks on the 19 strengths of the tube. We are starting to get into that 20 kind of result now. 21 MEMBER POWERS: I understand. I think this is 22 an extremely illuminating graph on what the range of 23 uncertainty is here. I mean you find all the failures 24 occurring in roughly the same time as small variations in /j N 't 25 parameters move their relative positions around a lot it l NEAL R. GROSS l C00RT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISLAND AVE., N W. l (202) 234-4433 WASHINGTON, O C. 20005-3701 (202) 234-4433 i 329 1 secms. 2 MR. LONG: Well, for instance, the curve that r~s 1 l ( ) 3 goes up to one earliest, the one on the left for the surge 4 line, if you include the materials variability in there, 5 there's just a few minutes. I think it's three or four 6 minutes maximum, difference between the five percent 7 probability that you will have failed and a 95 percent 8 probability you would have failed. 9 So that the tubes and the surge line look 10 relatively distinct, even when you put the probability 11 distributions on them, as far as materials uncertainty is 12 concerned, materials variability. Now if you start adding things that you 13 14 brought up earlier about well, the hot leg may have a bend 15 lir it that its stress isn't calculated, that's true. 16 There's some systematic problems with this. 17 Dut just in terms of the materials 18 variability, they do look fairly distinct until you start 19 putting cracks in them. When you move the cracks, move 20 the lines with the cracks, you start getting different 21 amounts of overlap. Eventually the crack will, at least 22 with materials uncertainty, clearly fail before the surge 23 line if the crack is large enough. 24 MR. DONOGHUE: I just want to throw this up to n ( ) 25 illustrate something Steve just said. This was one of the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 2344433 WASHINGTON, D C. 20005-3701 (202) 234-4433 m. _ _ _ _._ _. _. ~._ _ -m..- 330 -1 earlier plots that was done once we had some understanding 2 of the variability in the Larson-Miller parameter. O 3 For the tubes, you can see when we~get up near 4 one, it varies over-several minutes where the surge-line 5l is fairly tight over just a couple of minutes. 6 Again, there's several ways of'doing the 7-variability study. This was done earlier on when we were 8 just considering the material and the crack variability. ') 9 But when we-tried to' include all the other things from the j I 10 model as I'll explain in a second, tube failure 11 probability gets quite complicated. ] l 12 I think I am on page 19, after a lot of 1 13 flopping around. !0 I -14 I think we have talked'a lot about the top j 15 part of this page. What I wanted to point out here is i' 16 that again,. flaw distributions were used from what we are 17 calling the research distributions, and the calculation is. L - 18 done at bins and length and depth of_ bin. I will explain 19 that in a minute. But understand that what we tried to do l L 20 here was leave the ability for other flaw distributions to 21 be considered at some point. I think the model is put 22 together so that we're not locked in to any particular i' ll 23 flaw distribution, but as information is updated, it can 24 be included. 25 At the end of the presentation, I will be L NEAL R. GROSS l' COURT REPORTERS AND TRANSCRIBERS [. 1323 RHODE ISLAND AVE., N.W. (202) 234 4 33 - WASHINGTON, D.C. 20005-3701 (202) 234-4~" i 331 1 talking a little bit about sensitivity studies that were 2 done using this probability calculating tool over a range O 3 of tube temperatures to understand.the effect of 4 variability uncertainty in the creep magnification factor.. 5 We did it at one different RCS pressure to see'what the 6 difference would be from a PORV.to a safety valve set i 7 point on the RCS. Then we did is using, as I said, both 8' flaw distributions. 9 Okay. We have talked a lot about these points 10 up here already. The point to me here is that the two 11 length bins were picked for particular reasons. The crack 12 length for -.the critical crack length for normal 13 temperature and normal differential -- I'm sorry, ,O V 14 differential pressure at the PORV setpoint was one break 15 point. l l i I 16 The next was for critical length for the i 17 temperatures associated with core damage. That's where l 18 the flaw distribution was broken into. greater than one 19 inch for this critical length, and between a quarter inch 1 20 and one inch for those lengths. So those are the length 21 bins that we are talking about. Then the through-wall i 22 depth bins were over five percent increments. l L 23 So the creep model that we talked about i 24 earlier is used at each flaw sized bin and it's done for 25 each thermal hydraulic case. Take the results from a NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234 4 33 WASHINGTON, D.C. 20005-3701 (202) 2344433 332 1 thermal hydraulic case, and you take those creep damage 2 indices. You do it at each of the length and depth bins. (,_ ) 3 CHAIRMAN SEALE: Joe, you are going to be 4 showing some detailed results here for the next few 5 slides. Is this a good point to take a little hiatus for 6 about 15 minutes? 7 MR. DONOGHUE: I think so. 8 CHAIRMAN SEALE: Okay. We'll come back in 15 9 minutes. 10 (Whereupon, the foregoing matter went off the 11 record at 9:58 a.m. and went back on the 12 record at 10:15 a.m.) 13 CHAIRMAN SEALE: All righty. Proceed. (_) 14 MR. DONOGHUE: Just before the break I had 15 finished mentioning how we had binned the flow 16 distributions for this calculation. And what's being done 17 -- and somewhat restating, I think maybe, what I've 18 already said -- was that the probability of another 19 component -- and the other components being a surge line 20 or hot leg -- probability that they would not fail prior 21 to the tubes at some time when the creep damage would 22 equal one -- integral I was discussing before -- that's 23 what's being calculated. \\ 24 Now, a little more detail on what's in this t'^N ( ) 25 calculator tool. Monte Carlo method is used to account NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. l (202) 234-4433 WASHINGTON, D C. 20005-3701 (202) 234-4433 333 1 for tue probability over a range of Larson-Miller 2 parameters as I mentioned, and not only for the tubes, / ) 3 also for the -- well, for the surge line and hot leg, I 4 think that's included as well. The tube wall thickness 5 and the tube diameters, the variation there is included, 6 and then the bins as I mentioned. 7 Now, on the event tree -- which I won't flip 8 to, I'll just mention that -- pressure-induced top event - 9 - there's a limit load analysis that's performed 10 independently, using the conditions for main steam line 11 break, the high differential pressure condition, but 12 without a temperature challenge. And that -- I think I 13 say down here later on, basically by doing that earlier in f ,\\_/ 14 the event tree we remove those flaws from any 15 consideration for temperature challenge. 16 Calculation -- figured out the probability of 17 failing a tube ahead of the other components, and again 18 for each bin -- I keep saying that -- but then, we have to 19 make sure that we don't use the flaw distributions for 20 cracks greater than 100 percent through wall. 21 As I was showing you earlier, that plot of 22 flaw distributions, the actual functions go beyond the 23 hundred percent point. And one way you could consider l 24 that is to take everything that's out beyond a hundred n( ,) 25 percent and stack it up at 100 percent. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 W.SHINGTON, D C. 20005-3701 (202) 234-4433 _. _. _. _. _ ~. 334 -l 1 And we don't'think that would be realistic. -) i 2 It would just tell you that you have a much higher-l 3 probability of failure then you really ought-to. So-i 4 basically, I think we're truncating at-the hundred percent i r 5 point,.those functions for use in these calculations. j l 6 As I said -- oh,.this is a different point-3 ? 7 the flaws expected to burst in normal service are removed i .8 from:the distribution. So I think greater than'89. percent 9-through wall,.I think that's the through wall for a one-i 10 inch crack. I could be wrong on-those numbers. But' flaws l i I l 11 that are going to fail spontaneously are removed. That 12 kind of makes sense. j 13 And as I said, anything that's going to fail L 14 under the pressure challenge is removed earlier on, so L 15 that we're only considering the' failure probability of i 16 temperature-induced rupture. L 17 Now, some of the results -- let me start by l-18 saying, before I get any barbs about the number'of figures 19 in.these numbers -- it does not indicate the accuracy of i 20 the calculation. They were simply carried to prevent us L 21 from having larger roundoff errors when we start combining 22 all these results. So I'll just -- I sFsuld put it up j I, 23 there in big red letters, I-guess. I 24 But you can see, first thing we did -- I'll 25 just lay out how this -- again, have the descriptors for L l l NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS [t . (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234 4 33 1323 RHODE ISLAND AVE., N.W. l -l , s,, _ ~ _. _ _ -.. _. _ _. _. _ - L 335 j l 'l the thermal hydraulic analysis that we used for the event 2 tree. Then, first that pressure-induced calculation was

fl.

3 performed and as you can see --and you have two pages that l 4 look almost the same here -- this one uses the research 5 flaw distribution that I have been discussing, and then l 6 that sensitivity study that I mentioned where we do two _7 different flaw distributions, came in_ handy. 8 I won't keep this up here too long; I just 9 wanted to show you the comparison of the -- you see over 10 there for pressure-induced under the research 11 distribution, somewhat higher values than you have here 12 for the-NRR distribution. And this corresponds -- well, 13 this is pretty close to what we're allowing under the Rule i O .tiV 14 for pressure-induced rupture. And it seems realistic to 15 keep that -- that's one reason why we wanted to keep the. 16 research distribution into consideration. 17 And then when you see, for temperature-induced 18 rupture you see quite a difference, but I just wanted to 19 point out that the pressure-induced difference kind of 20 drove us to consider -- one reason to consider using a 21 research distribution. 22 Now, a couple of things to point out here. 23-The bracketed "1.0" is not a calculated value. We've j 24 mentioned earlier that if we had a clear loop seal with a l-25 depressurized generator, that's going to lead to higher I i NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISI.AND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234 4433 f i 336 1 temperatures in the tubes. I think we've~had -- you've -2 heard that discussion before. i 3 If you have a one-directional flow rather than j 4 the counter-current flow, it's. going to lead to a greater 5 tube challenge. And in this case if you have that happen, 6 the temperatures get high enough where you're almost i 7 certain to have a tube failure if the steam generator is I 8 depressurized, l 9 'You can see the difference is much -- it's in-I 10 the weeds here, two percent -- if you have an intact steam 11 generator with a clear loop seal. So that's where I was t l_ 12 saying -- using the term earlier, Russian roulette -- if 13 you have only one steam generator depressurized, you have 14 that RCP seal failure, even a clear loop seal, but the 15 question is, where. It's going to be the depressurized 16 generator or not? 17 The other point to make is that, if all these '18 generators are depressurized for a station blackout case, 19-you have a relatively high, but not drastically 20 unacceptable, tube failure probability. And then the 21 other contributor is up here. I think is a -- well, 22 correct me if I'm wrong; this might be a -- I think this l 23 .17 is a combination of the numbers below it -- no, it's l [ 24 not. 25 This is -- oh, thank you, okay. This is just ? NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS i 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 + 1 337 1 a seal LOCA with a single steam generator depressurized. 2 And that's relatively high and discussion on that is -- -s s \\ t / 3 get's back to the thermal hydraulic results where you have 4 a -- not considering a loop seal clearing, you have 5 accumulator injection causing a pressure spike. It gives 6 you a high tube temperature, very momentary high tube 7 temperature that could lead to tube failures. 8 MR. LONG: Joe, this is Steve Long with the 9 staff. Let me just clarify what's going on under that 10 case 9 up there. It's essentially a combination of three 11 of those four probabilities listed there. We essentially, 12 in the thermal hydraulic case, had one steam generator 13 that was depressurized, that didn't have a clear loop seal c '3 14 -- that's the 17 percent. 15 We had another one that had a clear loop seal 16 but the steam generator was pressurized, and that gave us 17 something down in the noise, and when we averaged -- 18 excuse me, when we combined that -- and plus a third steam 19 generator with a zero failure probability because it was 20 pressurized and remained relatively cool because the loop 21 seal didn't clear. 22 So those are three steam generators we 23 effectively added up for none of them having a failure, 24 and then took the complement to get a total failure /~' (,)x 25 probability, or one tube somewhere in all -- any of the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASH'NGTON, D.C. 20005-3701 (202) 234-4433 . ~... ~ -- .. l -338 1 1 three stream generators. So the number you'll see for r .2 that case combined is about 19 percent, I think. I I.% 3 The thing in the brackets we didn't really 1 4 have in the thermal hydraulic case, and that was.the f 5 combination of the clear loop seal and the depressurized 6 generator, but the temperatures are extremely high. I l 7 think I saw 1308 K as a peak temperature in the l 8 pressurized generator with a clear loop seal on -- we just l 9 assume that's going to fail if it was depressurized 10 instead. 11 And when we've tried to put this in the event f I 12 tree we had to take all the possible combinations of where j i 13 that depressurized generator might be. So those are the i 14 four constituents that you add to different degrees in 15 different parts of the event tree for a seal LOCA case. t l 16' MR. DONOGHUE: Thanks. Just to show you again i i 17 the NRR average distribution -- the values for 18 temperature-induced failure are much lower than we saw 19 with the research distribution, and that reflects again, 20 on the difference in the area of concern here. You can 21 see again, the research distribution is somewhat higher, l 22 flawed tube population than the NRR. l t 23 I think I've laid out -- and I'll go back to, l 24 just to kind of recap how -- what we've gone through so L

(_)

25 far; in fact, one of my first slides. We discussed the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 l 339 1 event tree, at least laid it out for you and showed you 2 what went into quantifying it and how it was structured to ( ) ' ~ ' 3 some extent. 4 The quick discussion about thermal hydraulic 5 analysis results, the bacis for the tube failure model, 6 the flaw distributions that we considered, some of their 7 limitations, and a quick discussion on the tube failure 8 calculation -- that put these three pieces together to get 9 a number for tube failure that's going to go into the 10 event tree. 11 Now, you already saw a result from the event 12 tree for containment bypass; that low-to mid-10-' number, 13 for example. And the next thing I wanted to do is just ,c (./ 14 discuss some sensitivity studies we've done on these 15 different components of the calculation, and then discuss 16 what we consider to be some of the key issues that remain 17 and limitations for our work. 18 I'll start with several sensitivity studies 19 that were done using the flaw burst calculation 20 probability. We did a sensitivity over a range of tube 1 1 21 temperatures, and this was done earlier on when there were 22 a lot of discussions going on about the uncertainties 23 involved in the thermal hydraulic analysis. 24 And right now there's some work that's in-hand ,y(,) 25 that looks like it's going to resolve, or at least NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D C. 20005-3701 (202) 234-4433 340 1-address, a lot of those' issues. But what we did -- and i 2 this again, sorry is not in your handout -- what was done _o '3 was, in over a~ range of temperature this offset -- this is l 4 Kelvin temperature -- is an offset from what the '5 SCDAP/RELAP would get the tube temperature for hot tubes o i 6 would be. l 7' And we offset it 60 degrees below that, up to l i 8 90 degrees above that value for these different cases. 9 For the base' cane, station blackout, loss of' auxiliary l i i 10l feedwater -- with all the steam generators intact. And i i 11 then the next case with.the uteam generator depressurized, i i f 12 ~ and then we used another flaw distribution on that j I 13 particular case. And then the last one is the RCP: seal f 'b \\/ 14 LOCA scenario with the generator depressurized. i 15 And what you see is, for the no offset, you 16 see the basic results that -- the big contributor here is 17 the RCP seal LOCA, up above 10 percent tube' failure l ~ 18 probability. And then-the base case that we've used for l l 19 this study up till now -- the case 3-R we were discussing 20-there -- is between five and ten percent tube failure. 21 And you can see as the temperature increases l l 22 the failure probability is higher, but at some point we e i 23 reach -- at 70 degrees offset, when we-add 70 degrees to -24 the. temperature calculated -- you see that the smaller [ 25 flaws become bigger contributors. l NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS y 1323 RHODE ISLAND AVE., N.W.- l (202) 234-4433 WASHINGTON D.C. 20005-3701 (202) 234-4433 . -.,... -.. - ~. _... - -. =.,. - - 341 1 And then this green curve, using the NRR flaw 2 distribution, takes off. And that's where'I was showing-j O 3-you earlier, the flaw distribution, NRR flaw distribution j 4 is a flat-line below 40 percent at a high value -- that's .j 5 where we're not sure it's realistic. It counts--- when 6 you get to higher temperatures small flaws are going to be 7 more of a concern, and this becomes'such a large 8 contributor we're not sure that's realistic. ) I 9 But by the.same token, you saw.how-low the 10 research distribution-was for small flaws. So these-11 values may be somewhat different in reality. We think-12 that, at least for purposes of doing a sensitivity study 13 on our risk estimate, that we've picked originally a 70 l L 14 Kelvin degree temperature offset just for giving us some 15 idea on how sensitive these results would be. l l 16 I understand now -- and I'll talkLa~few pages l 17 -- about the recent thermal hydraulic sensitivity analysis -j 18 that show that the temperature with some certainly is 19 going to be offset by something much less than 70 degrees. 20 Somewhere in.the 40, I think the 50-degree, the Kelvin 21 degree range is where you can say the maximum offset ought 22 to be. So that makes us feel a lot more comfortable of 23 something at 70 or above. 1: i 24 So I think I've given you a picture of the f 25 temperature sensitivity for tube failure probability. And 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 342 1 then we did creep model -- the multiplier, the Mg 2 variability -- and again, another one that's not in your 1 ~s ) i 3 handout -- just to show you that, using the research flaw 4 distribution you can see that -- or what I'm saying here i 5 is that it's about equivalent to a ten degree, or added 6 ten degree error. 7 If I take, for the 95 percent error --or 8 variability in M, that's up around -- well, it's says.06 p 9 -- that gives you something less than.1 probability of 10 more failure. And if I go to the graph I just had up, 11 this is where the ten degrees come from. 12 For the base case, with a 3-R, it's the red i 13 line. If I look at something just below ten percent, 'w about eight to ten percent, that tells me it's something j 14 15 around ten degrees. That's where that statement comes i 16 from. So it looks like the flaw stress multiplier is not 17 a large contributor. 18 And the other sensitivity that was done was 19 for pressure sensitivity; just adding a hundred pounds to 20 the output from SCDAP. It was equiva]ent to about a 15 l 21 degree temperature difference and failure probability. 22 Now, there were further sensitivities done on 23 the event tree, or accident progression event tree, or 24 APET a lot of people call it. And these are -- this work p) ( 25 is quite recent; very recent, I'll say, and we still are NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234 4433 .._..,._._.7_.._.-__._.. ,(; j. 343 1 evaluating. But we knew --these were the kind of 2 sensitivity cases we wanted done. 3-And basically they fall into secondary, 4; integrity, primary, system integrity.-- including RCP seal 5: LOCAs -- and then a temperature offset is discussed. And r (- 6 we'can -- as I said, this was quite conservative, this 70K I [~ 7' offset; at least it appears it was quite conservative to l- [ 8 use that. And that's something we'11 have to update'with -I 9 more recent information as far as thermal hydraulic c10 uncertainties go. 11 But one-could argue whether or not you should '12 do a sensitivity considering perfect integrity or-( 13" absolutely no integrity under.these kind of assumptions,. 14 but this is not supposed to represent necessarily, the L j 15 effect'of the Rule, that the Rule provision would have. l' 16 It's just to give us some insight into the event tree 1: 17 sensitivity. -18 Questions on that page? There's a lot of text i 19 on there. 20 Now, this is probably quite exceeding the 21 limit for putting up slides that aren't in your handout, i 22-but I~do think you have this page separately in case you 23 wante'd to take notes on it. This is some, as I was l 24 saying, very recent work so it didn't even make it into j 25 your handout. But, results from some of those sensitivity { NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS j' 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005 3701 (202) 234-4433 l' _. - - --.- ~. 344 t 1 studies that I just listed, I'll discuss here. E 2 .Now, what we saw from the base case assumption i o 3 is that you see about a 20 percent chance of containment l t 4 bypass given the core damage -- the Hi/ Dry frequency. So I. 5 you're seeing about a --'.a reduction by.five. Now, the' 6 majority of the events that lead to bypass -- it's not the l 17 majority of the Hi/ Dry frequency but it's -- the majority L 8 of that bypass probability was in RC-1, that large release i 8 9 category. l 10 I think I mentioned that earlier. It's the p 11-direct release to the environment via some failed 12 secondary component when you have a tube rupture. One j 13 thing that's important to point out is that the event tree r l 14 here, the analysis that.we've done, doesn't make a P 15 distinction between a single tube rupture or multiple tube 16. ruptures. 17 There's been some discussion in the~ staff 18 about what the potential is for a failed or failing tube t 19 to lead the tube ruptures, and what we're saying is that 20 if you have a tube that has ruptured in this sequence, you 21 have containment bypass whether it's a single or a 22 multiple tube rupture. It's all binned together. t 23 MEMBER POWERS: When you assign things to RC-1 24 category, have you looked in detail at the transport from 25 the site of the rupture, to the release to the NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE.. N W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234 4433 m _ - _. _. _ - -.. - _. _ _ _ - _ _ _._ _ _.. _ _ l 345 1 environment? 2 MR. DONOGHUE: I know there has been analysis-3 done using the Melcor source term, and I believe it 4 account's for hold-up fractions in the steam generator and i i l 5 so.forth. So I think it's been done in some detail. 6' MEMBER POWERS: What Im interested in is 7 that, if you look on the secondary side, depending on l I' 8 where your rupture is, the fission products would have to l. i L l 9 flow through a forest of tubes and bins and'a variety of .j 10 things. It's not 100' percent ~ obvious that they could make i l 11 it through all that tortuous path; it's not obvious that 12 they wouldn't. And certainly in the older codes there was l l 13 just no attempt to model that, and I wondered if that was i d 14 still the case. 15 MR. DONOGHUE: Well -- I'm sure I'll be i 16 corrected if I'm wrong here -- but I believe that there 17 were assumptions made for hold-up and the position in the i 18 steam generator. I just don't recall what they'were. 19 That information we can get to you. But for purposes of 20 categorizing things on this event tree, it wasn't l; 21 necessarily on the value of the release; it was just i l l 22 basically, do you have the release -- certain release i i c 23 path. 24 MEMBER POWERS: I understand. But I mean, you 25 understand that these bypass events are not typically a j a NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS j 1323 RHODE ISLAND AVE., N.W. j (202) 234-4433' WASHINGTON, D.C. 20005-3701 (202) 234-4433 ) . ~. 346 1 frequency-dominant issue; they show up as a risk-dominant o^. 2 issue because of the magnitude of the source term ~ 3 associated with it. ) 4 MR. DONOGHUE: That's why -.yes, right. I l 5 mean, this number that I'm calling a result for this 6 example ane. lysis, this 10-' number, is,.yes. .It's.not 7 necessarily a stratosphere itself, but we understood that. ) 8 That is of concern because -- l 9 MEMBER POWERS: But if the source term is 10 wrong, they're overly conservative; then maybe we're going 11 through a huge amount of effort that's not merited. l 12 MR. DONOGHUE: Right, but you know,-my i. lO MEMBER POWERS: I understand that. I'was 13 presentation stops short of discussing that. 14 l 15 going to push _you to say, let's cg> on and look at the risk 1 16 now. i l-17 MR, DONOGHUE: Well, I.think you heard me say ) 18 yesterday, we're putting in words in the Reg Guide that L 19 are -- we're trying to make sure risk is consistent with l-l. 20 the subsidiary safety objectives, and the measure that j 21 we're using -- I guess, maybe to avoid arguments about 22 anything more than containment bypass frequency -- is the l 23 containment bypass frequency. 24 MEMBER POWERS: But what happens if a licensee () 25 comes in and says -- conservative values for the frequency l NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 347 l 1 of a bypass accident, but then I looked in depth at the i 2 consequences of that accident -- and this is not a 7_s i 1' '! 3 hypothetical example -- I happen to know that there are at 4 least contractors for some of the utilities that are in 5 fact, looking at this issues, and they came in and made it 6 a risk argument. Would you be in a position to evaluate a 7 risk argument? This is -- 8 MR. LONG: This is Steve -- 9 MR. DONOGHUE: I think I could say we'd be in 10 a position to evaluate it; however, what we've said all i 11 along here is that, this comparison to the subsidiary 12 safety objectives has to be done witn an understanding of ) l 13 the defense-in-depth effect -- the effect on the defense-(gl ~s 14 in-depth that you're having. 15 I mean, we're talking about containment 16 boundary essentially; talking about the tubes in this 17 situation. So that's why we focused on the potential -- 18 the containment bypass frequency -- the potential to fail 19 that boundary and less on the release consequences as far 20 as understanding what we had to have the Rule do for us. 21 MEMBER KRESS: But if a licensee came in with 22 that kind of analysis, you would entertain it as a -- if 23 you had a rule based on, strictly the frequency of the l 24 failure, and they came in with a subsequent analysis like (n) 25 Dana talked about -- even though they exceeded some NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D C. 20005-3701 (202) 234-4433 348 1 criteria in the rule on this frequency, you might say it's 2 okay, you guys are all right, you don't have to reduce 7-s 3 that frequency? 4 MR. LONG: This is Steve Long. Let me add a 5 couple of concerns here. We already have an analysis like 6 that from NEI that tried to draw a distinction between the 7 number of tubes that ruptured and whether or not if one 8 ruptured and then the reactor pressure boundary gave way 9 somewhere else, it would cut down the driving force and 10 drastically reduce the release. 11 The problem is, once we get to the point where 12 we think we have a significant failure of the steam 13 generator tube at very high temperatures, we're really not ) 'w/ 14 sure what that's going to do to adjacent tubes. And 15 trying to predict the progression of the accident sequence 16 after that is pretty tough. If you have impingement of 17 very hot gases on adjacent tubes, we're not sure what that 18 leads to. 19 So just what's getting into the steam 20 generator is already uncertain, and how much gets into the 21 steam generator, of course, will have something to do with 22 the temperatures of the steam generator's surfaces, 23 secondary side reach. I think some -- I'm not sure if 24 Victoria calculations have been done on the secondary side ( ) 25 -- it's in the back of the room either. v l NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISLAND AVE., N.W. l (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 _m_. ~ _. -.. _ _ _.... _ _ _ 349 1 .There~are codes that could address that, but l 2' you're getting into some very researchy areas to-try to ' %.) l 3 draw good solid conclusions; things that' we didn' t feel lee j l 4_ could do for the study, and things that we'didn't feel we l 5 could really credit from' industry right now. What we'd l 6 really be doing here would be saying, we don't really need I 7 containment for this kind of an accident'in the form of 8 physical barrier. It could be more or less in the form of j i I 9 a filter. l i 10 And we've had those arguments years ago, also, i i _ 11 but we were talking _about much more effective filters. j i 12 MEMBER KRESS: Would you comment on the 13 defense-in-depth arguments you're making? O 14 MR. LONG: Well, people have discussed things 15 like sand bed filters for venting containments in the l l. 16 past. And I think probably you can generate a good enough 1 17 filter that that might be a viable option, but we're not 18 so sure what kind of a filter we have here in a steam 19 generator secondary. Until we could really have some 20 confidence there I don't think we want to do away with the 21 requirement that the boundary stayed physically intact. c l l 22 MEMBER KRESS: So I interpret your answer to 23 my question being no, you'd probably -- l 24 MR. LONG: No. f b 25 MEMBER KRESS: -- give much credit for NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 4 ~,, _. 350 -1 attenuation calculation of fission products because of the [ '2 large uncertainties associated with that kind of I\\ 3 calculation? i 4 MR. LONG: I guess what I'm saying is, I know I 5 that the industry -- and I think we have been doing some l 6 work to try to investigate that and see what kind of l l 7 effects we see -- but it's very difficult to reach a 1 L i 8' conclusion where we think we could, in a short-period of ) 9 time, allow the industry to go ahead with the Rule that

f 10 we're talking about here.

l 1 11 It didn't seem to be a very quick way of j 12 solving _the problem, let me put it that way; however, if i i 13 we got into a cost benefit discussion on the benefit of l l 14 avoiding some accident where we did l i 15 MEMBER KRESS: You almost have to do that in a l l 16 cost benefit, I think. l 17 MR. LONG: We have to, but how conservative we 18 have to be is another matter. 19 MR. DONOGHUE: I'll just point out that I 20 guess, one way of answering your question is by pointing 21 to the outline of the section in the Reg Guide discussed i L 22-briefly yesterday. I mean, that lays out what we think is j 23 an acceptable way to approach a risk assessment here. 24 And the two options again were: do a PRA to I Lq ,Q. 25 understand what your. initiation frequency here would be, NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS j. 1323 RHODE ISLAND AVE., N.W. l (202) 234-4433 WASHINGTON, D.C. 20005 3701 (202) 234-4433 ll' ~ 351 1 and if that's low enough then you're okay because even if ' 2' you challenge the tubes you're not going to -- if.you're O_ 3 below that 10-' number to begin with, you' re not going to j 4 exceed the safety objectives. 5 The other option is to plow through a 6 calculation like this, including the tube failure kind of i 7 modeling,. understanding your flow of distribution, etc. i 8' So that you know, they'.re given that option but nowhere do l 9 we really say, show us how small your release is; get .10 below some certain number. We aren't able to that. 11 MEMBER POWERS: But I mean, it seems to me, 12 from a, personal point of view it seems a lot easier to 13 attack the source term than try to understand flaw 14 distributions in pipe that I can't see. 15 MR. DONOGHUE: It might be easier to attack ~ 16 flaw distributions but it's not easy for us'to say how 17 large a release is acceptable, necessarily, unless the PRA l 18 implementation guidelines end up spelling it out in 19 detail. But that's not a foregone conclusion from what I i 20 understand. ) 21 Something to say, Charlie? i l 22 DR. TINKLER: Yes. Charlie Tinkler from the i 23 Office of Research. I just wanted to clarify -- we did 24 determine the off-site consequence using Melcor which gave 25 some credit for deposition on the secondary side, but not 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' 352 1 a great deal. It did not take credit for the, all the 2 additional deposition that might be attributable to the

,8 3

presence of the tube bundles, separators, dryer, etc., 4 etc. We only took credit for it as a volume with a 5 residence time, and things like that. 6 There was some credit given, but greater 7 attenuation of the source term was not credited in the 8 Melcor calculations. 9 MEMBER KRESS: Do you use the standard dies, 10 1465? 11 DR. TINKLER: No, we actually -- 12 MEMBER KRESS: Or did you let Melcor -- 13 DR. TINKLER: We actually let Melcor calculate 14 the source term, and as you heard in other presentations, 15 we did calculations with Victoria and Victoria did produce 16 a lower source term than Melcor -- for a couple of 17 reasons. But for these off-site consequences -- 18 MEMBER KRESS: -- mostly. 19 DR. TINKLER: No 1/p dependency and things 20 like that. But we did use the Melcor calculated source 21 term and a limited attenuation of the off-site release due 22 to deposition on the secondary side. 23 MEMBER POWERS: At least when we have studied 24 dryers and separators in the upper internals of BWRs we ,a (C) 25 have found they can be a fairly significant attenuator. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. l (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 1 '353 1 DR. TINKER: I wouldn't dispute =that. pso 2 MEMBER KRESS: But I don't think I would st '3 describe a steam generator as-(indiscernible). 4; MEMBER POWERS: Well, I can't disagree with 5-you that they're different, but you have a significant l 6 attenuat' ion here and ---I mean, the potential of some 7 attenuation -- and I'm wondering why wouldnt an applicant 8 licensee in the face of an awful lot of work say gee, this 9 is an easier way to get around the skin-the-cat; just.to i 10 go after the consequences rather than trying to understand 11' flaw distribution. 12 MEMBER KRESS: I think he would if he thought 13 that staff would accept that as a reasonable argument, to i 14 not fix (indiscernible), talk about (indiscernible) fix. '15 MEMBER POWERS: It seems like we ought to be 16' in a position to evaluate the legitimacy of the arguments 17 that he puts forward. 18 MEMBER KRESS: Well, I think there ie some 19 credence though, to this concept of there.may be more 20 uncertainty there than in these calculations, and I think 21. that's -- I think the uncertainty drives whether or not 22 you treat this as a defense-in-depth concept or not-. 23 MEMBER POWERS: There clearly is a very big 24 defense-in-depth argument here, because in contrast to f~'% 25 many other parts of the RCS, we do not have another (,) - i l-NEAL R. GROSS j COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 _ _ _. _ _ _ _ _ _. _. _. - _ - _ -. _ _.. _. _. _.. ~. _... l: 354 l l 1 physical barrier beyond this one, and so there is a very j l i l The question then boils

-X 2

big defense-in-depth question. ! ( ) i 3 down to the one that Mr. Long articulated so well. ~Is-I l 4 that filter a barrier or not? I .5 MEMBER KRESS: Yes. Well, I think it is;.~it's r ~ 6 just how much confidence you can have in its ability to -- i i 7 MEMBER POWERS: Well, you won't have any 8: consequences at all if you don't look'at it, s t 9 CHAIRMAN SEALE: It's a bit of a barrier. l i 10 MEMBER KRESS: That's true. l 11 MR. DONOGHUE: Okay. I think I. lef t off at-i 12 this point. Yes, the balance of the events where we' talk } 13 about RC-2 on the event tree, were again,'due.to partially { i O i 14 or low -- partially depressurized or a low pressure RCS l i 15 with'an. intact secondary; however, we're not sure about i i 16 the MSIV integrity. We make an assumption-there that-l 17 we're going to depressurize the secondary and possibly l 18 have a release path. i I 19 I mentioned the RCP seal LOCA contribution -1 20 here, and we think it's a large contributor depending on - ) 21 - it does depend on again, which plant you're talking ) 22-about -- RCP seal leak model varies from designs, but for .i 23' our example, it's 70 percent again, of the containment l 24 bypass frequency. It's not -- I'm sorry, it's 50 percent 25 of the total of the bypass frequency. Am I saying that i NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 ..___.,_._....._._....-.-.__._.._._.._______._.__..._._m.-_._.-____ .___m._ _a.... - 355 1 right? ~ t j v 2 So we see it's a'large contributor and that's OU 3 why-we spent-a lot of time looking at.this, and I'll just 4 divert for a moment from the slide. When we first saw the j 5 high tube failure probabilities under this case, a lot t 6-more work went into looking at the RCP' seal LOCA branch on l 7 the event tree. 8 Because originally it was put in the event j 9 tree thinking it was going to give us a benefit, that the i 10 depressurization from the seal leak would actually help us

11 and avoid tube challenges.

But for the reasons I l 12 mentioned earlier, it does lead to a' challenge to the. l ' 13 tubes, somewhat higher ~than we thought it would be. i 14 So we're going to need to look a little more 15 closely at some things that are used to pick RCP seal i i 16 leak; its frequency and the magnitude that you're going to i i 17 see; how it progresses through the event -- how the leak f 18 actually may change during the event'. 19 The sensitivity studies I showed you just a l l 20 minute ago that list the sensitivity studies were done -- 21 and you can see a variation'for the conditional 22 containment bypass frequency changing from a very small 23 number, two percent, up to -- you can sc9 30 percent for l I t l-24 that 70K increase i f 25 Now again, I think this is conservative, so I NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 356 1 think we're seeing something from below ten percent up to-i 2 the order of maybe, if we use some maybe more realistic O 3 offset for the temperature, maybe in the 20 or so percent -l 4 range for containment bypass. l 5 I say right up there, digital sensitivity I r 6 studies are planned on the event tree -- I'm not going to f l 7 go into detail on what those might be. We have a list l I 8 that we're considering to augment what -- a list.I showed j i 9 you a minute ago. 'I 10 Now, another piece of recent work were j 11 sensitivity' studies done using the thermal hydraulic .12 modeling, and these results have not been factored into j f 13 the risk conclusions I've been talking about today. We ,.i( l 14 just haven't had the chance. Some of this stuff is still l 15 being documented, as a matter of fact. l 16 But I at least wanted to point out that some l 17 work has been done to address the questions that have come l l l 18 up before this committee, and we're going to first look ) 19 here at some of these results as telling us -- at-least I i 20 for looking at tk' revised base case that we have used ) 21 before. J 22 It tells us that what we used before might be 23 somewhat conservative; however, I'll point out here that l 24 the original base case we had, had a tube split -- this 25 35/65 is the hot upflow in the tube bundle versus the cold l NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS i 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234 4 33 i m_ l-i 357 l' 1 return flow to the inlet plenum. { 2 Well, if we decide that it's more O i 3 representative to usa what was seen in the transient tests l l 4 -- the mock-up, the Westinghouse mock-up, included steady i L 5 state and transient tests -- the transient tests showed a i 6 more even split between hot and cold tubes -- if we think i 7 that's more representative and we switch to a base case l l 8 that uses that modeling, not only do we have lower tube l 9 temperatures but we have more tubes being affected by l 10 whatever that hot tube temperature is. l So the first look at this tells us that it's. j 11 l 12 probably a wash as far as the tube failure probability. l 13 You might have lower temperatures but you have more tubes t 14 being affected. But we need to do a detailed look at i 15 that, at these conditions, going through the calculation 16 that I've been talking about, and really see what the l I 17 result is. j l 18 MEMBER POWERS: Is the sensitivity of tube 19 failure or tube temperature, linear? Because the number 20 of tubes is linear -- is a linear effect here on the-21 probability o!' failure. 22 MR. DONOGHUE: Well, just keeping the tube -- l .23 I believe in these cases the number of tubes being 24 affected is the same, in all these cases. 25 MEMBER POWERS: So what you can argue is that NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 358 1 it's essentially linear in the load differential 2 temperature region, and it may become non-linear at very ,~ !'~'! 3 high temperatures, or something like that. 4 MR. DONOGHUE: Well, unfortunately, the flow 5 distribution plays into this -- 6 MEMBER POWERS: Yes, I know. 7 MR. DONOGHUE: -- and mucks it up. But you 8 can see it's a relatively -- it's a straight enough line 9 over the temperature ranges we're talking about. The 10 temperature offset range we're talking about. It will 11 correct. 12 Just to give you some background on whee was 13 just completed of these sensitivity studies. The ) '/ 14 SCDAP/RELAP5 model was updated with a mixed convection 15 heat transfer correlation and that was used to rerun this 16 case 3. That came up with somewhat -- I'll put up a slide 17 in just a second that will show you some of those results. 18 And then also what was called case 6, which 19 was not used in our risk study originally because we 20 thought that this tube bundle split was more i 21 representative of a base case, but we are now considering 22 that this may be more representative, so we might be i 23 switching again as I mentioned a minute ago. l 24 So both of these cases were used in a (3 () 25 sensitivity study, but the emphasis was put on what might 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 1 359 1 be our next base case, and these are the different 2 sensitivity studies that were run, including changes in s / T \\,) 3 the heat transfer correlations, heat exchange in the hot 4 leg between the twc flows, and then applying the five 5 percent limits for those three mixing model parameters: 6 the mixing fraction, the flow ratios, and the percentage 7 of tube bundle. 8 I think you've heard the discussion about 9 mixing models to some extent, and this -- I'll show you 10 the next slide -- these are just -- I tried to put up both 11 civilized and barbarian units here. And you can see for 12 the base case, the hot tube if you recall, about 987 I 13 believe, was the number for the original case. (_) 14 You don't see too much of a change, but you 15 can see if we switch to another base case it does drop the 16 tube temperature somewhat and as I mentioned earlier, it's 17 a wash probability-wise, because of a larger number of j 18 tubes affected. 19 Then you can see, you don't see a large 20 difference until you apply all those mixing model i 21 parameters at their limits, which may not be -- is 22 probably not -- realistic, because although we're trying 23 to -- the purposes of doing this study were to understand 24 some synergistic effects, that might be -- or the effect 73 25 of synergistic effects -- it's not clear at all that you t,s-) l NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., NW. l (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 l t 360 1 would ever have a situation where you had all three of 2 those mixing model parameters at the same extreme. O 3-There's compensating effects there that are going to 4 prevent that from happening. 5 But you can see that the temperature offset ~is 6 going to be something less than 70 degrees. We were much. 7 too conservative when we picked that number, but we're 8 rather say we were too conservative than not conservative 9 when we did that. l 10 And we'll be looking at this information.to 11 decide what the real temperature sensitivity ought to be 12' that we'use in the sensitivity study we do on.the risk 13 study. ,p 14 Now, that was the synopsis I intended to give 15 you on the sensitivity studies that we have put together. 16 The next couple of pages, I just wanted to discuss where l. 17 we.think the soft spots are, and.what key issues we've 18 dealt with.or are still dealing with. 19 I'31 start with probably, the hardest part of l L l 20 this whole analysis to deal with, the flaw distribution. l 21 I've already mentioned several times the difficulty in- -l 22 specifying.a' flaw distribution for a certain plant. The 1 23 representative flaw. distributions themselves have, as you e f 24 saw when we contrasted the two distributions that we use 25 in our examples, have certain things that we need to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISt.AND AV5., N.W. j (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 i __i .-_-_.___..__.m ~. .m 361 1 understand. You know, what's.the effect for really small 2. flaws?- 3 But for that one piece of the flaw 4 distribution, thermal hydraulic analyses are showing'us 5 that if the temperatures don't get too high, which appears 6 to be the' case,'the small flaws are probably not a big-7 concern. So I think the emphasis is really going to be on 8 the shape of the distributions in that area, above 50 or 9 60 percent through sall where you really see the thermal 10 risk coming from. 11 The other parts of this analysis when we try 12 to understand, for example, the~ thermal hydraulic results 13-much better, helps us key in on the other parts of the s/ 14 analysis that we may need to understand. It's really 15 clear here, the flaw distribution questions are somewhat 16 more focused now than they were just a few months ago, but -) i 17 there is work to do there if we're going to try to get any l 18 particular flaw distribution from a plant. i ' 19 Event tree quantification. We mentioned 20 failure frequencies. Right now in there are essentially 21 guesses-from 1150 for primary / secondary components to some l 22 extent. Sensitivity studies that were done on a list 23 there try to account for either no integrity or perfect 24 integrity.on those components to see what the effects l( ) 25 would be, and they're not huge by -- well, I don't want to i NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 l 362 1 put up another slide right now, but -- but we saw the 2 overall changes from 2 to 30 percent, and it's probably 2 s l I '~J 3 to 20 or 25 percent for containment bypass, so even there 4 we don't see a large contribution. But it's still going 5 to have some effect. 6 Besides the quantification of the event tree, 7 the event tree itself could change for different plant 8 designs, and also different plants will probably have 9 different contributors to the events of concern for the 10 Hi/ Dry scenario. 11 I did mention that for Surry, besides station i 12 blackout, loss of feedwater transients and the DC bus 13 loss, are small components. Those things could change for ? t t \\/ 14 other plants and those might have to be things that would 15 modify the event tree relative to a station blackout if 16 they're really big enough contributors. 17 Another point that should be made is that, 18 everything I've talked about as far as RCP boundary i 19 failure goes, is hot leg surge lines and tubes. It's 20 apparent to a lot of people that there's other parts of 21 the reactor coolant system that could fail under these 22 high temperature, elevated pressure conditions. 23 Now, in a study that's been done, the staff 24 did take a look at that but not in any quantitative way. n () It's basically a discussion of how hot things will get and 25 NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISLAND AVE., NW. (202) 234 4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 i ..._-._m f 363 i 1 not' understanding the materials and their design, chances, j l whether or not they're going to fail or not 2

judgment, 3

under these conditions -- but that's not something that's 4 been-factored into the risk study. Again, it's a 1 5 discussion after you crank through the numbers. 6-And just incidently, things like Manway l 7 gaskets, I think some valve bonnets -- the seals on valve f 8 bonriets -- the things that are going to be of particular l L 9 interest if we're going to look at other places and the 10 reactor coolant pressure boundary may fail ahead of the i l 11 tubes under these kind of challenges. l 12' Okay, we've' talked a lot about thermal ? j 13' hydraulic results before today and a little bit today, and -j 14 basically I think we're feeling a lot more-comfortable i 15 than:we did just a few months ago as far as the. analysis i - 16 that we've used. l [ 17 I think the modeling issues with this latest i 18 work are going to be addressed, and it looks like there's 19 a pretty convincing case that's made for -- not too i-20 drastic offset that could be applied to the basic results. 21 But one thing that has to be understand is 22 that again, plant design specific features are going to 23 affect those thermal hydraulic results -- you know, change 24 some things relative to the one we use for the sample ] 25 analysis. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS r: 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON. D.C. 20005-3701 (202) 2344433 ,,e -e s m -n 364 1 Staff has done some other analyses with the CE 2 plant. There are other analyses out there people have -,y / ) 3 done for things like DCH studies and so forth for other U 4 tube designs -- other PWR designs -- and the risk study 5 would be expanded a lot by trying to factor in all those 6 different kinds of designs, but when you look at the 7 frequencies that we started with, the range of frequencies 8 from the IPEs that we looked at, we already know there's a 9 range that the results could fall into. 10 I think the best that could be done at this 11 point is if one were to try to do tnis on a plant-specific 12 basis, you not only have to understand of course, the flaw 13 distribution for that particular plant, but you're going k _) 14 to have to understand the thermal hydraulic response for 15 that particular plant under these kind of events. 16 Now, although I said the tube performance i 17 model itself is probably the component of the analysis 18 that's of the highest confidence, I will say that a couple i 19 of things -- a couple of questions are left. What we're 20 considering here is burst for these tests. That's the 21 failure model considering creep failure leading to tube l 22 burst. 23 The change of leakage, maybe high leakage or 24 leakage at these high temperatures impinging on other ( ) 25 tubes and causing other tube failures as I mentioned s., t NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234 4 433 ._._.m. _--_.m 365 1 before, is not considered here, and when we did the risk i p y. 2 analysis we basically said, anything that's going to burst t l 3' is going to be a containment bypass. But there could be I 4 other. failure modes that could be considered that could l 1 .5 lead to.the bypass situation. i o it's specific to axial cracks. -And '6

Also, l

7 again, I'll just repeat that the circumferential cracks -- l l j 8 we haven't included in the risk st.ady -- there'is a flaw l 9 distribution out there for them,. and there is work ongoing L i' l to understand response to cir cracks under these L 10 .11 situations, but it's just not matured enough to the point. i 12 that we could fold it into this kind of analysis yet. ] l l 13-And the last point on here is the creep i; l. 14 failure prediction is dependent on understanding the i 15 temperature and pressure histories for a particular event. j 16 I don't think this is a large concern, but we do have to-i [ 17 understand that when people do different analyses,.as I .j ~ mentioned early on today, we did see a different 18 l 3 l' 19 temperature history for the tubes from our' analysis that 20 our contractor did versus what was presented by an j l l 21 industry analysis. i 22 There is an effect on the creep potential for t [ 23 the tube, and you can argue that the failure model could 24 be altered somehow if you had a drastic change in the 25 temperature and pressure history under those conditions. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433- 366 ~ ~ 1. That's all I prepared for today. If there's '2 any more questions I have lots of backup material and lots O 3 of people'-- 4 CHAIRMAN FONTANA: Looks like a nice job to 5 me. One. thing that comes to mind. It may be worthwhile 6: estimating the fission product attenuation in the 7 secondary system, particularly if you have the break in a 8 steam line. The hottest part of the-steam generator 9 probably would be somewhere near the inlet plenum of the 10 steam generator, so if you get a leak there, if the tube 11 b'urst there, two things would happen. 12 You'd have a high velocity out of the tube 13 burst itself, but then it would impinge on the surrounding 14' tubes and that gives you a lot of surface area,cand since 15 most of the fission products will be as aerosols in 16 particulates, you get some attenuation there. And then 17' the flow as.it goes up through the tubes and' spreads out 18 will probably be kind of slow, so you'get more 19 attenuation. Then it has to go through the steam 20 separator region. i 21 So you know, it may be worth trying some back l l-22 of the envelope calculations and see what kind of 23' attenuation you might be getting. 24 MR. DONOGHUE: Well, I agree that it might be (/ 25 interesting to do that, to understand the release, but NEAL R. GROSS COURT REPORTL RS AND TRANSCR;3ERS 1323 RHODE 'SLAND AVE., N.W. ' (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 .__~_____..__._._.__.._..._.....____.__________m._.. ...y 367 .1 from the aspects of trying to put together a Rule that 2 addresses the risk, we wanted to make sure that we didn't O 3 lose defense-in-depth, and saw that having a focus on the 4 containment bypass potential, just the potential of having 5 a release path was what was really important to us. 6 CHAIRMAN FONTANA: Yes, I wasn't suggesting 7 changing what you had, it's just a little better 8 understanding of the fraction'of the release that wouldn't i 9 get to the environment would be interesting. 10 CHAIRMAN SEALE: Are there any other comments 11 that anyone -- any on the Committee wishes? Any 12 questions? 13 (No audible response.) l 14 Anyone want to make any comment? Go ahead. 15 MR. SCHNEIDER: Ray Schneider, ABB/CE. I 16 guess a lot of the work that you're doing, a lot of the 17 numbers in your APET are based on the NUREG 1150 logical 18 structure, which is our first look at severe accidents and 19 our first understanding as to how the plant may cope with 20 severe accidents. l 21 Over the past years plants have been -- or at l - 22 least the industry has been asked to deal with severe '23 accident management, looking at the issues. There's a i 24 general understanding that high pressure RCS. scenarios are 7 25 bad for a number of reasons, not just the steam generator NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS i 1323 RHODE ISLAND AVE., N.W. (202) 2344433 WASHINGTON, D.C. 20005 3701 (202) 234-4433 368 1 tube rupture issue. l 2 And the guidance in most of the severe ,s ( ) '~' 3 accident -- or maybe all the severe accident guidelines or 4 at least most of them, at least ours -- is to basically 5 do whatever steps you can to basically do two things: 6 one, get water into'che steam generator, and the other one 7 is to depressurize the primary side. 8 And I see all this discussion, all this 9 analysis, tremendous amounts of detail, tremendous amounts 10 of sensitivity, and absolutely no consideration for the 11 fact that the industry is aware of these issues and is 12 trying to implement strategies and procedures to mitigate 13 them, which should be included into the logic structure /; (m l 14 and so would basically eliminate probably 80 percent of 15 your tree. Or at least put the significant part of your 16 tree into the low probability range. 17 MR. DONOGHUE: Well, let me say that, you 18 know, the Reg Guide that we've written up gives people the 19 opportunity -- the licensees the opportunity to use 20 updated information in the PRA to tell us that the event 21 frequency itself was going to be low. 22 You know, we used information that we thought 23 was -- maybe is somewhat dated -- but we thought was 24 pretty reliable from past studies, and we understood from ,.m () 25 the onset that there was other information that people NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS l 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234 4433 --._ ~._ _. _ _ _ _ l-369 l l t 'l could use'to update,' based on operator' actions, like i I'fs 2 you're saying, EPGs.and so forth, to effect that f-k_/ 3 initiating frequency. I i 4 MR. SCHNEIDER: Actually, my concern is j l 5 probably.more.of, there are two NRC programs: one.to l f i i 6 implement accident management guidance to reduce the risk i i 7 of these events, and the other to demonstrate that the j 8 risk of these events is higher and we totally ignore the i 9 other piece of it. ) l 10 And shouldn't there be kind of a meshing like 11 -- you know, you have the same person kind of. working on 12 both ends of this task -- shouldn't there be a meshing of L 13 the' fact that you anticipate accident management's. p k-14 guidance to basically be beneficial to the industry and l p 15 reflect that in your Rule? i t L 16 MR. DONOGHUE: Well, as a matter of fact,jos l 17 had the people that were working'on -- and he's right l l 18 there. 19 MR. PALLA: Bob Palla with the staff. Let me 20 just clarify that. This analysis that you've just heard 21 about was based on 1150 results. We looked at the 22 informatio. coming out of the Level 1 plant damage status L 1 l b 23 and we basically backed out from that, you know, the j i 4 24 fraction of these events that occur with stuck-open relief j () 25 valves, seal LOCAs, depressurized steam generators, etc. l l NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 370 1 We recognized that implicit in those numb rs 2 is some credit for operator actions, but that the credit 73 ( ) ~ 3 at the time of 1150 would not have extended to severe 4 accident management guidance. Now, what one would do if, 5 you know, if we had more time we might have pulled a model 6 off -- IRRAS-based model and perhaps added additional 7 branches to the event tree to represent depressurization 8 of the primary or operator actions on the secondary side. 9 But we did not do that. We realized that 10 that's -- it's possible to improve the rigor of this J 11 analysis. One thing I want to just mention, though, is 12 that there would be some difficulty in assigning j 13 probabilities to the operator successfully carrying out ) m/ 14 these actions, because accident management guidance are 15 not as prescriptive as emergency operating procedures. 16 And frequently when one deals with issues 17 about quantifying the reliability of an operator action 18 they look to see just how explicit the guidance is to give 19 some sense as to how much can I count on that action. 20 And the guidelines are more general in nature 21 so there's, number one, a question about the human error 22 probabilities that one would assign, and then furthermore, 23 these are events -- at least for Surry -- we're looking at 24 events that are dominated by station blackout, and it's a 7 ( ) 25 combination of early station blackouts that involve loss v NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 371 1 of'feedwater from, you know, basically at the initiator, 2 as well as late station blackout. 'O .3-And these -- you get into some very plant-j 4 specific questions as to whether the support systems that ~ 5 you would need to depressurize.-- even'if you have an l l 6 accident management guideline thatLtells you to i l 7 depressurize, you may-not have the available support i 8 systems. I mean, your power operator relief valves will 9 require air plus electrical power. 10 So these guidelines may be fine, and these are .11. good' objectives to depressurize, but one would have to go 12 into the very details of the plant's capabilities to 13 actually use those depressurization features, and that's .t 14 one of the reasons we didn't try to do it, is'because we 15 wanted to get finished. 16 But licensees in a more complete analysis 17 would want to look at their specific support systems and 18 determine whether their guidelines, their accident 19 management guidelines would provide the competence that 20 they could depressurize. 21 MR. LONG: This is Steve Long. Let me remind 22 you of one more thing. We did do a thermal hydraulic case l-l.. 23 to see what would' happen if we opened a PRV and that 24 bottled at about the time that the EOPs would say -- or 25 rear accident management would say to open it. And it did NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 l 372 l 1 show in effect, it was beneficial and we did do -- or at 2 least we're doing a sensitivity study -- I've forgotten ,s ) '~' 3 which, if it's done yet -- to see what happens if that is 4 maximized for -- 5 MR. DONOGHUE: That was on -- yes, that was -- 6 MR. LONG: So we have looked at the effect. 7 The problem is what Bob was talking about. What's the 8 probability of being able to succeed in getting that 9 effect? 10 MR. SCHNEIDER: Well, I understand, but you 11 have on one hand you're trying to -- you're pretty 12 confident you can identify the flaw distribution to within 13 a few percent, but operator actions aren't quite as clear. Lk >) 14 MR. LONG: I wouldn't say that. 15 MR. SCHNEIDER: I know; I'm overstating the 16 position. But the point is that you really have an issue 17 that I think is equally as important as any other issue on 18 the table. And I haven't heard the words accident 19 management, you know, guidance, and recovery actions, and 20 stuff like that, significantly discussed in the issues. 21 And that's the one message you want to get 22 across to the industry I think, is that this is an 23 important issue and you want to make sure that we respond 24 to it, not just in calculations, but in tangible action (f")8 25 should an event occur. l NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 2344433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 373 1 1 MR. LONG: I understand. 1 2 CHAIRMAN SEALE: That is a good point. The 7 J 3 influence of emergency or accident management activities l 4 on the motivation of the operators to do that job 5 appropriately and so on, could very well influence the 6 human factors; that is, the success of actually doing it. 7 So the relationship needs to be drawn somewhere to help j 8 make that point fairly obvious. 9 Any other questions or comments? (.o audible response.) N 10 11 Well, the Committee is concerned about having 12 enough time and enough opportunity to really prepare -- 13 I'm sorry. Thank you very much, but don't leave yet, 77 ( 4 \\) 14 please, okay? About having enough time and an opportunity 15 to really do justice to the efforts that have been put 16 forth by both the staff and the industry in preparing your 17 presentation so that we really reflect the best we can in 18 the letters or whatever messages we may see fit to send, 19 either to the EDO or to the Chairman and the 20 commissioners. 21 For that reason, we would like to now try to 22 discuss any issues that -- the things that the members of 23 the Committee have identified as being important to them, 24 and use that then, as guidance for what it is you might n () 25 wish to say tomorrow, either the staff or the industry, l NEAL R. GROSS l COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 ..__-_.~.._,____---.=____-.-_____...___...4 _.~a (l 374 I when you meet with the: full Committee. And then from that 1: 1 2 we can also synthesize the kinds of things we might wish 0 3. to include in any formal communication. 4 Now, one other thing. The. presentation on j 5. this subject is scheduled from 8:45 to~10:45 tomorrow 6 morning.. We have received a request from.Dr.'Hopenfeld to l i 7 present the features of a differing professional opinion ~ and we estimate 1 L 8 he filed back in 1994 on this subject, l 20 minutes L 9 that that's going to take somewhere around, oh, l 10 or so. We would like to reserve some time for the l 11 Committee, . after we hear everyone. 12 So doing all that manipulation and all, it 13 looks like the staff should count on perhaps, 45 minutes iI l 14 tomorrow to do its discussions, and so we might keep that 15-in mind when we start trying to identify issues, both 16 members of the Committee and your response to it. ;And I 17 guess I think the industry might want to have perhaps, 15 18 minutes to summarize your position as well. 19 So with that background, I'd like to ask the 20 members of the staff -- or the members of the Committee, 21 to identify the things they might think would be 22 appropriate to present to the Committee as a whole. 23 Dr. Powers, do you have some comments?. 24 MEMBER POWERS: The Rule that's been put forth 25 here is extremely complicated. It has a lot of elements NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 1 375 1 in it, and it's diverse in its nature. And quite frankly, l 2 I still haven't gotten the length and the breadth to know 7_s ( ) 3 why each one of these major blocks is there. 4 But it strikes me that, the first and foremost 5 thing is to present to the Committee as a whole, the major 6 blocks. There's material on characterization of the 7 tubes, there's matcrial on monitoring, there's material on 8 risk analysis. And probably at no greater level of detail 9 than that, present those blocks and say, here's what we're 10 trying to accomplish with each one of these blocks. 11 I think that would be useful in an abbreviated 12 presentation of just 45 minutes, to kind of cast things in 13 that format. (_ 14 Then there are lots and lots of detailed 15 things about each one of those blocks that I have to 16 admit, I still haven't gotten my hands around, and I think 17 they just invite protracted discussions -- 18 CHAIRMAN SEALE: We've noticed. 19 MEMBER POWERS: -- on those things, because 20 you know, for the life of me this.05 business and.025 21 and 1 x 10 to the minus -- there are a lot of 22 probabilistic criteria that I don't understand. If they 23 can help me with a sentence that wouldn't get Dr. 24 Apostolakis launching off into epistemic analiatory -- ( ) 25 which I think is impossible -- I'd appreciate them helping NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON. D.C. 20005-3701 (202) 234-4433 - - -. ~. - 376 1 me'but I -- in defense of our schedule, maybe they should l 2 help me offline. O 3-CHAIRMAN SEALE: We're not sure we want to l i 4 provide that challenge. l 5 MEMBER POWERS: However, to help me, they must 6 avoid the words epistemic analiatory. 7 CHAIRMAN FONTANA: He's not here to defend 8 himself. l 9 MEMBER POWERS: And that's why I bring it up, i 10 because his defense is withering.as well. That would be j 11 the first bit of guidance that I would make. 12 Then I think, Mr. Sheron's introduction on l l i3 this subject where he 32id down -- here are the issues and i 14 the. areas that I'd like to get some guidance from the 15 Committee as a whole -- was a very helpful insight to this j 16 that get elaborated as we listened to some remarks made by f 17 representatives of the industry. 18 And to the extent that they could articulate i 19 those to issues they had, which I believe, if my notes are l 20 correct, one was that there were concerns that the Reg 21 Guide might be too prescriptive, 22 On the other hand they were concerned about 23 the amount of flexibility they.ought to grant to the 24 industry because they feel like they come in very much 25 after the fact on this. I think it's important to NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. l. (202) 234 4433-WASHINGTON, D.C. 20005-3701 (202) 234-4433 -p y .r v v J 377 'I articulate the area they'd like to have some advice on. a 2 And finally, I think-that it would be useful O i 3 for them to articulate very carefully, but at still a j 4 fairly broad-brush level, how it is that risk gets j (. 5 factored into this. And I say at a broad-brush level { 6 because there's' lots and lots of detail here, and each'one i 7 of those details invites lots of questions about what ) 8 particular probability distribution you used, or how you. "f r 9 broke down the event tree, but I would say more at a i 10 philosophical level how risk has been fa'ctored into-this, i 11 and whether it is really risk or it is really core damage j 12 frequency that's been factored into this. l 13 There was, at a point in the presentations l' 14' yesterday, a comment made that -- I wrote it down in the l 4 I 15 notes but quite frankly, it eluded me as I listened to the l-i t t j 16 presentation -- was that they had chosen to put their 17 conservatism in the source term-calculation rather than in l I i l 18 the failure calculations. .l 19 Well, quite frankly, it looks to me like l 20 there's lots of conservatisms in the failure calculations l-21 and I'm not terribly wild about putting unquantified L 22 conservatisms in on things in the midst of calculations. 23 But I'd like to understand that just a little bit better 24 because I think there's more meat than that comment 25 ' elicited from me at the time it was made. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234 4 433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 l \\ 378 1 Those are the bits of advice I have. 2 CHAIRMAN SEALE: It's rather comprehensive, 4 s I \\ \\,/ 3 I'd say. j 1 4 MEMBER POWERS: It is one of the most diverse j 5 Rules that we've ever had presented to us, and quite l 6 frankly, let me echo Dr. Fontana's comment that I was 7 quite impressed with today's presentation in the length j 8 and breadth of their attempts to try to understand the 9 phenomonology here in some sort of a probabilistic 10 context. Quite impressive the amount of work they've done 11 on this. 12 And equally well, I think it's quite 13 impressive the amount of work they've done in all parts of l I l A' 14 this Rule. This is a monumental effort that they've 15 undertaken, and we should spend some time on this. 16 CHAIRMAN SEALE: Yes, I think that it's -- 17 earlier the comment was made that they went so far in 18 their analysis and then they had to stop, or they knew 19 when to stop, or whatever -- or they had to make a 20 decision. And it's very clear that you had that problem 21 repeatedly, because you had such an incredible number of 22 alleys that you could start running up with your 23 calculations and never get back out of the mess if you 24 didn't do that. () 25 MEMBER POWERS: And that is an area that they NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (2n2) 234-4433 WASHINGTON, D C. 20005-3701 (202) 234-4433 379 1 should think about in formulating a presentation with the 2 Committee, is frequently we heard -- well, we just didn't ( ) v 3 have time to do this, or this is comething we're fixing to 4 do, or maybe one of these days we'll get around to doing 5 this -- is just how close are we? l 6 Are we really ready to get this thing out on 7 the board, are these leftovers that important or l 8 unimportant, or have we really got it sorted out before we 9 send out for public comments or can we do it after? Some 10 sort of a context to know where we are. 11 CHAIRMAN SEALE: Dr. Shack, do you have any ] 12 comments? 13 MEMBER SHACK: Yes, I guess Dana alluded to ,m i( 14 it, but let me just emphasize it again. I think, you 15 know, one of the things that's going to come up is this 16 question of, what level of review is going to be given to 17 what level of documents? And the staff has told us that 18 the current plan is not to review the detailed documents, 19 the methods in a sense. 20 And I guess I'd just like some words about. why 21 they feel that's the right way to -- you know, Jack 22 mentioned that, you know, they were drawing this box to 23 try to describe acceptable methods in the Reg Guide. What 24 are the advantages of doing that versus a simpler Reg ,m (w) 25 Guide and then review the procedures, and why you seem to 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 380 1 opt for one approach over the other. 2 Then of course, the -- and again, the question ,s l i ( / 3 of how the .05, you know -- or, how the performance 4 criteria, whatever number one happens to pick, is going to 5 be, you know, related to risk. Something that just sort 6 of has to be -- come up again. 7 CHAIRMAN SEALE: Anything else? Dr. Kress? 8 MEMBER KRESS: I think I'd like to have a 9 better -- on how one arrived at the acceptable 10 probabilities for one tube failure versus two tube i 11 failures versus three or more. 12 MEMBER POWERS: With or without epistemic 13 analiatory in it? U 14 MEMBER KRESS: Yes, without. Please. I 15 thought a little better -- more discussion on the factor 16 of 3, -- factor would be useful. The discussion we had 17 today I thought was very good. I would like to see a 18 repeat of a lot of that, some of that, plus -- 19 CHAIRMAN SEALE: At least a summary, a 20 summary. 21 MEMBER KRESS: A summary. But particularly, 22 I'd like to hear more about the reasoning behind not 23 looking at fission product transport and the site specific 24 qualitative health objectives, as opposed to conditional ,qQ 25 containment failure probability as an acceptance criteria. NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N.W. (202) 234-4433 WASHINGTON, D.C, 20005-3701 (202) 234-4433 -..-.. ~ _ _. 381 L 1-Why that'would not be also an acceptable approach. l L 2 And I think that's all I had on it. LO 3. CHAIRMAN FONTANA: Well, after three smart i 4 guys have' talked there's not a whole lot left to say. The 'I 5 thing is though, in trying to read the Rule, you see that L 6-you're trying-to patch on to a -- what's originally a 7 deterministic, if you want to call it that -- sort of l: 8' expo'sition, and trying to patch on some risk-based c l 9-thinking onto it'. 1 10 And the impression.I get is, you end up with 11' something like a 3-hump camel, which will probably get you 12 across the desert but probably would have pretty nasty l 13 disposition, so you don't know when he's going.to spit at f~d 14 you. 15 I would like to see that thing _ kind of reduced l 16 down to essentials'if it's at all possible, with as much j 17 application of risk-based thinking as possible, which -- 18. well, that goes tar enough. i - 19 CHAIRMAN SEALE: Okay.

Well, I,

as I went l 20 through here I've'tried to identify in my own mind the 21 issues that Dr. Apostolakis would have raised if he had l 22 been sitting there, and I won't use the magic words. But 23 certainly the question of how risk gets factored in, that l 24 - Dana raised; the question of the level of review of the 26 documents that Dr. Shack raised; and Dr. Kress's point i NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLANO *VE., N.W. - (202) 2344433 WASHINGTON, D.C. 20005-3701 (202) 234-4433 l: T p .,y pp 382 1 about the acceptable probabilities for single and multiple 2 tube failures. ,m 3 Those are all issues I think, that Dr. 4 Apostolakis would have added. I guess the other comment I 5 would make would, perhaps, be more addressed to the 6 industry and that is that -- and I certainly concur with 7 the comments that have been made here -- but I would think 8 that it would be worthwhile for the industry people to 9 briefly outline the approach that they took for the Rule, 10 because at least formally it seemed to me, there was a 11 little bit more of a risk-based initial approach to 12 setting the case up, if you will. I think you know what I 13 mean. ( 4 / 14 And I think that would be very useful to the 15 members of the Committee to see that somewhat different 16 approach. Other than that, I think we know what we want 17 to do as far as the presentations tomorrow are concerned. 18 Any questions? Going to be able to do it? 19 MR. STROSNIDER: No, I don't -- this is Jack 20 Strosnider from the staff. I don't think we have any 21 questions. The issues are clear; we'll have to see if we 22 can cover it in 45 minutes. 23 CHAIRMAN SEALE: Okay, fine. j 24 MR. STROSNIDER: We'll make an attempt. ) (,) 25 CHAIRMAN SEALE: I understand. It's a tough i NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHODE ISLAND AVE., N W. (202) 234-4433 WASHINGTON, D C. 20005-3701 (202) 2344 433 _. -.. _ _. _ _ ~ -1 383 1 1 1 one. Any questions or comments from the industry at this 'l 2 point? i O' -3 (No audible response.) j 4-I have one other thing to ask. Yesterday at '5 some point someone talked about the possibility of_this i 6 representing'a~backfit kind of activity. :And whether or l I 7 not that's the case, at some point -- and it may be 8 premature today or tomorrow -- but at some point it seems. l I 9-to me that it would be very worthwhile to hear if anyone l I 1 10-has done a risk benefit assessment of what this Rule does. l l 11 And that's now in the broader sense of, you '] 12 know, how many -- you know, what is the impact on tube l I 13 availability and what are the potentials for saving i 14 inspection times and all of that sort of thing? Or_what 15' are the costs of increased inspection times, and so forth. i .16 Did you want:to comment on that?. 17. MR.-STROSNIDER: Yes. This is Jack 18 Strosnider. Just a comment on -- that analysis is being j 19 performed. I think it's probably middle of December 20 before it will be complete. It has to be performed as l' L 21 part of our promulgation of the new Rule. We have to [ 22 determine under what parts of the backfit Rule you would 23 do these different things. l 24 So the work is in progress. I'm afraid it bp 25 probably is a little early for us to lay out all the logic Q 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 384 ) 1 there because it's still being done, j 2 CHAIRMAN SEALE: I can understand that. On I) l 3 the other hand, I think it would be very helpful to us in 4 assessing what's involved here and what the traits may be,. j i 5 for us to hear about that when it is available. 6 MR. STROSNIDER: Yes, I guess, as I indicated, t 7 we'll be looking at whether these things cnr backfit is j 8' under compliance, or safety enhancements or cost- .l l 9l beneficial -- that will be included in the package that -l l 10 comes around for office review and concurrence, which the j 11. Committee will also a get a copy of. And I guess my only i 12 comment there is, the' Committee would have to decide l c 13 whether reading that package is adequate or if you want to l 14 hear from us again.in our presentation. 15 CHAIRMAN SEALE: Very good. Are there any 16 other questions that we need for the record, at this l 17 point? 18 ONo audible response.) 19 Well, we'll dismiss this session. I j l i 12 0 understand there's another Subcommittee meeting this l 21 afternoon. 22 (Whereupon, the ACRS Joint Meeting was 23 concluded at 11:40 a.m.) 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 j i i C% V CERTIFICATE { This is to certify that the attached proceedings before the United States Nuclear Regulatory Commission in the matter of: Name of Proceeding: ACRS JOINT MEETING: MATERIALS AND METALLURGY AND SEVERE ACCIDENT SUBCOMMITTEES l Docket Number: N/A Place of Proceeding: ROCKVILLE, MARYLAND were held as herein appears, and that this is the original transcript thereof for the file of the United States Nuclear Regulatory Commission taken by me and, thereafter reduced to typewriting by me or under the direction of the court reporting company, and that the transcript is a true and accurate record of the foregoing proceedings. [ ll24 W / M ENE GRAY OfficialReporte[r Neal R. Gross and Co., Inc. O NEAL R. GROSS COURT REPORTERS AND TRANSCRIBERS 1323 RHoDE ISLAND AVENUE. NW (202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433 INTRODUCTORY STATEMENT BY THE CHAIRMAN OF THE MATERIALS & METALLURGY AND SEVERE ACCIDENTS JOINT SUBCOMMITTEE 11545 ROCKVILLE PIKE, ROOM T-2B3 ROCKVILLE, MARYLAND NOVEMBER 5-6, 1996 The meeting will now come to order. This is a meeting of the ACRS Joint Subcommittee on Materials & Metallurgy and Severe Accidents. I r hhert Seale, Chairman of the Subcommittee. The ACRS Members in attendance are: Ivan Catton, Mario Fontana, Thomas Kress, Dana Powers, and William Shack. The purpose of this meeting is to hold discussions with representatives of the NRC staff, the Nuclear Energy Institute (NEI), and the Electric Power Research Institute (EPRI) to gather () information concerning the technical approach used in developing the proposed risk-informed, performance-based rule and regulatory guide associated with steam generator tube integrity. The Subcommittee will gather information, analyze relevant issues and facts, and formulate proposed positions and actions as appropriate, for deliberation by the full Committee. Noel Dudley is the Cognizant ACRS Staff Engineer for this meeting. i The rules for participation in today's meeting have been announced as part of the notice of this meeting previously published in the Federal Register on October 21, 1996. A transcript of the meeting is being kept and will be made available as stated in the Federal Register Notice. It is requested that the speakers first identify themselves and speak with sufficient clarity and volume so that they can be readily heard. () We have received no written comments or requests for time to make oral statements from members of the public. (Chairman's Comments, if any) During the June 12-14, 1996 ACRS meeting, the Committee heard presentations by representatives of the staff and the Nuclear Energy Institute (NEI) on this matter. Today the Subcommittee will hear from the staff concerning the technical basis for the proposed rule. We will proceed with the meeting and I call upon Brian Sheron, NRR, to begin. O:v i l 4 w.- W 4 pa o RESULTS FROM APET ANALYSES Under base case assumptions, ~ l in 5 chance of containment bypass, given core damage o with high primary pressure and dry SGs Majority of bypass events (90%) assigned to large release category (RC-1 on APET) o Direct release to environment via open ADV or MSSV No delineation in APET between single and multiple tube ruptures due to inability to ~ assure integrity of adjacent tubes i Balance of events (10%) have lower releases due to partially depressurized RCS, intact (but o leaky) MSIVs, and secondary side holdup / deposition (RC-2 on APET) SBO with RCP seal LOCA is dominant contributor to thermally-induced SGTR frequency o (70%), and total containment bypass frequency (50%) l r Conditional containment bypass probability found to range from 0.1 to 0.4 in preliminary o sensitivity analyses (to be updated) minimum of 0.02 for optimal secondary side integrity maximum of 0.3 for 70K increase in temperature history o Additional sensitivity analyses planned - -.... m. m w v a w r u--- Proposed Steam Generator Rule Safety Analysis ACRS Materials and Metallurgy Subcommittee and Severe Accidents Subcommittee November 6,1996 Joseph Donoghue, DSSA/SRXB Office of Nuclear Reactor Regulation (301) 415-1131 AGENDA o Contributors to Induced SGTR Risk i i o Accident Progression Event Tree o Thermal-Hydraulic Analysis Results i o Flawed Tube Failure Model t o Flaw Distribution t O Tube Failure Probability Calculation i o Results/ Sensitivity Studies i ( o Key Issues / Limitations l i i 1 j i f O O O ~ CONTRIBUTORS TO INDUCED SGTR RISK o Spontaneous and induced tube failures contribute to SGTR risk Spontaneous failures often due to unknown mechanisms Assumption: No impact by rule on spontaneous failure frequency. Induced Failures: Pressure - ATWS (low frequency), Secondary Depressurization Thermal - Severe Accident Mechanical - Impact from spontaneously failed tube o Previous studies concluded that severe accidents would not challenge tube structural integrity (e.g, NUREG 1150, DCH studies) o Development of new degradation mechanisms and additional insights in high temperature tube performance o Significant @estions: Understanding of predicted RCS conditions and RCPB response during severe accidents 2 O O O ~ MECHANICALLY-INDUCED TUBE FAILURE o Potential for a circumferentially failed tube to cause subsequent failures o Two events involving circumferentially failed tubes: North Anna - 1 July 17,1987 Mihama - 2 February 9,1991 o Each failure was a complete circumferential break due to high cycle fatigue j o North Anna: Tubes adjacent to failure site were in service Eddy current examination, no reported damage to tubes adjacent to break O Mihama: Most tubes adjacent to failure were in service Extensive examination of adjacent tubes: contact had occurred with broken tube; contact traces on tube deposits without any denting of the tubes. o Maine Yankee Circumferential Cracking: Calculations submitted showing that impact and jet impingement forces would l be less than the shear load margin for a cracked tube. 3 ACCIDENT PROGRESSION EVENT TREE (APET) STRUCTURE o Treats subset of core damage events: "Hi/ Dry" events Elevated primary / secondary AP and dry SGs at core uncovery o "Hi/ Dry" events sorted by RCS and secondary system condition at core uncovery Primary system: Secondary system: Intact (high P) Intact (all SGs at pressure) Stuck-open PORV/SV Stuck-open ADV/MSSV RCP Seal LOCA Manual Depressurization o Pressure-induced rupture of flawed tubes at normal temperature (PI-SGTR) evaluated since not considered in baseline PRA model 4 APET STRUCTURE Additional mechanisms for late primary / secondary depressurization also evaluated O Stuck-open PORV/SV following core uncovery and heatup MSIV leakage following SG dryout (not considered in past PRAs but suggested f by anecdotal evidence) l l o Potential for thermally-induced tube rupture (TI-SGTR) prior to HL/SL failure [ evaluated as final event on each APET branch r i I L l i 5 i C \\ k j i 9 1 1 ~ C 8 Wj eI ,* j h, jE '9l., W;W h ! ,W! ra; ,l h,l et e r e h I .Wi .W T T!' W W g I .I -.I ri .I 6 i .. i i ~~ 9 l 1 l t 1 ? i l l 8 l W h ) 5 ai M W .E ' W 5 e e e e e ~ M' W w w W w w r .w w w w w w w w e 8 w **5am p meig g g g g gg g3 g g 3 g g 3 3 gg gg 3 3 g 3,3 33 3 3 3 W I i 1 H e!= sie se as a n!a a a min n,s a n!. M m, H 5awis M 8 g3 985 g,4 5 g g,8 g 5 g g 84 g 5g g;8 g g 8l8 38 g 2:8 gi g i T .i J .i .i g Ii i I Z l l l W l ll' I 1 i .rk ~l-M J - , 'e .i '. 4%.g .E.O..H2.n. H..tj n.;.n..I. .! H.!.!.*. .I,.f.s .I .X.;.E. :.E 7 .H..,,I,I ! I ) W Z i 1 i i~ I* ? l ~ i I O '. x' i g" ~ ~ > y' .g W g w w e 5 i. e .o u m v :. e., es.os, l 1) t. ..n. v.cns4-, g g g g g i :w c..ca..rs;4....cw: w M a. ,'<,.v.cn,c g g g g g g g , 3 a : -e :n una., s:e o Zw ~ a t< g e g g g g W i c g a .n.... x.... x4, wi y,, _UU< I I i g ) g g o ONO O'P $ Am0M g =g g g g g g g g g J . ensw e s danown 4 0) g une'd W g 2 W ] W W ] W d des %,e 4 wc0 d.0.880 0

I 4

3 I 1 3 2 1 E u.c. .a .>:..r m,. u I l $ 1 e.... :. us s. 3.. .V .N.. y l 1 i i I O O O ~ APET QUANTIFICATION o Top events quantified based on NUREG-1150 analyses for Surry and Sequoyah, and survey of IPE database o APET entry frequency== 2E-5/ reactor-year NUREG-1150 plants dominated by LOOP events, with balance from loss of feedwater transients and loss of DC bus IPE survey suggests leading contributors highly plant-specific o Primary system status (early): Intact in majority of events (70% for Surry, 50% for Sequoyah) RCP Seal LOCA contribution where RCS failure occurs with dry secondary (20% for Surry,50% for Sequoyah) PORV/SV failure a minor contributor (10% for Surry, 3% for Sequoyah) 7 O O O ~ APET QUANTIFICATION o Secondary system status (early): Probability that 1 or more SGs will be depressurized is significant in Surry-t 1150 (75 %), but minimal in Sequoyah-1150 (5%) Attributed to differences in ADV dependencies and procedures for SG depressurization during an SBO o Likelihood of late primary and secondary system depressurization assigned scoping values of 0.5 in base case and assessed via sensitivity analyses, due to lack of information on valve performance l Pressurizer PORV/SV reliability under repeated cycling and severe accident temperature conditions MSIV leak rates (MSIVs not included in Appendix J leak rate test program) l 4 8 O O O APET QUANTIFICATION o A single accident sequence was selected to represent the family of sequences defined for each branch 1 Limited variety of scenarios analyzed i o Probability of TI-SGTR quantified based on stand-alone calculations o Probabilities of PI-SGTR and TI-SGTR for each branch adjusted to account for number of intact /depressurized SGs (and loop seal clearing in RCP Seal LOCA events), and imported into APET o Final APET quantification provides the frequency of containment bypass due to induced SGTR resulting from core damage events t 9 b ~ O O O THERMAL-HYDRAULIC ANALYSES o Used SCDAP/RELAP5 results from Surry model o Cases used in safety assessment: Case Designation Case Description RES1 Base Case: SBO, loss of AFW RES 3 One SG depressurized (failed relief valve) RES 7 ALL SGs depressurized RES 9 RCP Seal leaks, one SG depressurized 1 NRR 6 PORV fails open, One SG depressurized o Calculations performed to estimate tube conditions at time of predicted RCS pressure boundary failure and relative times to failure for major RCPB components (hot leg, surge line, tubes) o Results appear somewhat conservative based on recently completed sensitivity studies i 10 i TUBE FAILURE MODEL o Model based on high temperature testing of machine-flawed tubes using original thermal-hydraulic analysis results o Only axially oriented cracks considered o Results indicate that creep failure is dominant mechanism ) i o Indications were that unflawed tubes would withstand thermal challenge o Includes uncertainty in effect of crack o Includes material variability effect on creep failure prediction 11 t TUBE FAILURE MODEL i i f8 f d' Creep Failure Prediction =1 g, t g,m,o) a Time to Rupture P'" 15 T 10 t = g I a Crack Magnification m, = mh (1 0.06) '-{ l [ Larson-Miller P,=(23.2i0.7 - 2.4 Ino) x 103 i Crack Magnification m = 0.614 + o.481x + 0.386exp(-i.25x) (through-wall) l Normalized crack length A=t12(i -v )$ c c = i.82 2 /R,h /R,h 12 m TUBE FAILURE MODEL Where: a crack depth c semi crack length hoop stress a P Larson-Miller parameter im R mean radius of tube m T temperature crack depth parameter a m magnification factor for through-wall cracks v Poisson's ratio h tube wall thickness 13 ~ O O O i FLAW DISTRIBUTIONS I o ' Representative' - based on operational data and inspection information l o Inspection methods not capable of defining a plant-specific crack size distribution O Plant-specific attributes o Two approaches developed: - Each uses three categories of distributions RES NRR 6 Degradation types - Axial cracks only Continuous function of - 2 specific flaw lengths length and depth o Creep failure calculations for axial cracks only { o Neither distribution realistic for shallow cracks ? 14 l t F ~ O O O FLAW DISTRIBUTIONS o RES distributions developed for six types of degradation Circumferential SCC at TTS Circumferential ODSCC at TSP dents = Freespan ODSCC = IGA / SCC in hot leg sludge pile Axial ODSCC at TSPs Flaws from loose parts

• Indicates used in risk study 15

~ o O O Comparison of NRR " Average" and RES " Moderate" Distributions 2.0 O q O o RES axial cracks > 1" O o NRR cracks = 1" 1.5 D m

• oe t3 1

0 1.0 5 's D i E \\ a C 1 (S 0.5 c. s \\ n-n 'B., ^ 0.0 0 0 " ' ' n -+ # 20.0 40.0 60.0 80.0 100.0 c depth (% through-wall) 16 i l RES DISTRIBUTION Free Span Defect Depths 0.0e00 ~ ~r --- 0.0500 Dietettpuelon Fitted to 0.0400 Alpha = 11.04 Bete = 3.99 Estimetod Actual g MeYt Variance = --- Fit to Observations 0.0300 ,/ _\\ Estimated Actual Sire ~ s l l Dietettsution g Alpha = 29.09 0.0200 --i Beta = 1.42 g / N t%meweb.nce= [ \\ 7.7 I 0.0100 - - l -- \\ l l Detection Etficiency = f i 66.1 % g' 's, gg * = 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 Depth (% TW) 17 9/6/96 O RES Flaw Lekyth Distribution 1.0 i 0.8 X 5cn C E 0.6 - I u Ge Ox W6 0.4 8 t ~B m* 0.2 - 15.7 % ~ 3.5% 0.0 ~ 0.0 1.0 2.0 3.0 4.0 t flaw length (inches) 18 i ~ O O O FLAWED TUBE FAILURE PROBABILITY o Event tree uses split fractions for pressure-and thermally-induced tube failure Split fractions change depending on predicted thermal-hydraulic conditions O Probabilistic models developed for flawed tubes, hot leg and surge line creep O failure o Calculation uses flaw distributions i i o Sensitivity studies: I Tube temperature Creep model magnification factor RCS pressure Different flaw distributions t 19 [ ~ O O O FLAWED TUBE FAILURE PROBABILITY o Failure probability of surge line or hot leg versus tubes as a function of time Used SCDAP/RELAP5 temperature / pressure results Included distribution of Larson-Miller creep characteristic for materials o Flaw distributions divided into 2 length bins j Exceeding critical length at normal temperature and AP at PORV setpoint Shorter flaws exceeding critical length at temperature associated with core damage o Flaw distributions divided into depth bins spanning 5% thickness o Creep model for axially cracked tubes used to calculated creep damage index for each of 32 flaw size bins for each thermal-hydraulic case o Probability that another RCPB component would not fail prior to the tubes based on time at which flawed tube creep damage index equals 1.0 j 20 m O O O FLAWED TUBE FAILURE PROBABILITY o Monte Carlo method used to average flaw bin. failure probability over range of: INCONEL 600 Larson-Miller Parameter Tube wall thickness Tube diameters Crack length and depth bins i o Limit-load analysis basis for pressure-induced tube failure calculation under normal temperatures i o Probability of tube failing first in each sequence found using results of failure i probabilities for each bin combined with population of flaws in each bin o Flaw distribution not used above 100% through-wall depth o Flaws expected to burst in normal service were removed from distribution l o Flaws expected to burst under elevated pressure at normal temperatures were j removed from creep calculation l 21 l O O O FLAWED TUBE BURST PROBABILITY - RESULTS RES " Moderate" Plant Flaw Distribution Sequence Designator / Description Pressure Temperature Induced Induced RES-1 No SGs Depressurized 0.0 0.0174 i 1 RES-9 Seal LOCAs - Depressurized SG 0.0549 0.1671 -Depress. SG, clear loop seal 0.0549 [1.0] Intact SG, loop seai intact 0.0 0.0 l Intact SG, cleared loop seal 0.0 0.0207 r RES-3 One SG depressurized 0.0549 0.0791 NRR-6 Pzr PORV open,1 SG depress. 0.0549 0.0184 RES-7 All SGs Depressurized 0.1646 0.1197 f I i i 22 } o o o ~ FLAWED TUBE BURST PROBABILITY - RESULTS NRR " Average"' Plant Flaw Distribution Sequence Designator / Description Pressure Temperature Induced Induced i RES-1 No SGs Depressurized 0.0 0.0092 f RES-9 Seal LOCAs - Depressurized SG 0.0059 0.0774 Depress. SG, clear loop seal 0.0059 [1.0] i Intact SG, loop seal intact 0.0 0.0029 Intact SG, cleared loop seal 0.0 0.0108 RES-3 One SG depressurized 0.0059 0.0390 NRR-6 Pzr PORV open,1 SG depress. 0.0059 0.0054 RES-7 All SGs Depressurized 0.0175 0.0547 i L l 23 FLAW BURST PROBABILITY - SENSITIVITY-o Temperature sensitivity: considered wide range due to earlier questions regarding thermal-hydraulic uncertainties. l o Distribution sensitivity RES distributions yield higher burst probabilities except for temperature errors 2 70K o Creep model flaw stress multiplier sensitivity: i 95% confidence level = +10K temperature error o Pressure sensitivity: I 100 psi = 15K temperature difference i. 24 I s O O U APET SENSITIVITY CASES Sensitivity Case Represents 1. No late SG depressurization due to Potential impact of confirming MSIV leakage MSIV leakage integrity 2. Lower early SG depressurization Plants without procedures / capability to manually probability (0.05 based on depressurize in SBOs. Better MSSV reliability as Sequoyah 1150) claimed by EPRI 3. Cases 1 and 2 combined Optimal secondary side performance 4. Probability of late primary Potential impact of providing AC-independent depressurization = 1.0 depressurization capability. Iligh probability of i PSV failure for liquid cycles as claimed by EPRI 5. Probability of late primary PORVs continue to rescat under repeated cycling depressurization = 0. or fail with small leak area. Operators fail to manually depressurize using PORVs. 6. No RCP seal LOCAs Reduced contribution of RCP seal LOCAs at plants with AC-independent seal cooling systems. Lower probability of RCP seal LOCA with Byron Jackson pumps (CE plants) 7. Increase temperature histories by Impact of upper bound temperature estimate on 70K (RES flaw distribution) TI-SGTR probabilities 25 O O O THERMAL-HYDRAULIC MODELING SENSITIVITY i o RES recently completed sensitivity calculations using the Surry SCDAP/RELAP5 model l 1 0 Cases used: f i RES 3 One SG depressurized (failed relief valve), 35/65 (hot / cold) tube bundle split i RES 6 One SG depressurized (failed relief valve), } 53/47 (hot / cold) tube bundle split i o Sensitivities: j RES 6A +20% heat transfer correlation (HL, SL, upper plenum, SG tubes) RES 6B -20% heat transfer correlation RES 6C +30% heat transfer correlation (Entrance to HL, SL, and SG tubes) RES 6D Increased heat transfer correlation (SG tubes only) RES 6E Radiation heat exchange in hot leg RES 6F Lower 5% limits on mixing model parameters 26 I i i m O O O THERMAL-HYDRAULIC MODELING SENSITIVITY Selected structure temperatures from Surry SCDAP/RELAP5 calculations Case Time difference Structure temperature (K/ F) Surge line to @ surge line failure time tube failure (min) surge line hot SG tube 3 17 1259/1806 973/1291 6 21 1254/1797 957/1263 6A 22 1260/1808 938/1228 6B 20 1258/1804 964/1275 6C 21 1259/1806 944/1239 6D 20 1255/1799 957/126* 6E 25 1256/1801 937/1227 6F 13 1257/1803 1007/1353 27 h t KEY ISSUES / LIMITATIONS o Representative Flaw Distribution Significant uncertainties remain in characterizing distributions i Inspection capabilities do not provide plant-specific distributions based on flaw size i o Event Tree Quantification Failure frequencies for pressure relief components'in RCS and secondary are not well known Pla:it-specific configurations and procedures coald affect structure and split fractions i o RCPB Weak Points Only qualitative treatment for potential of other component failures (other than { hot leg, surge line, tubes) t l' i i 28 i i t O O O- ~ KEY ISSUES / LIMITATIONS f i i t o Thermal-Hydraulic Results Plant-and design-specific factors could affect thermal-hydraulic response Resolution of modeling issues e o Tube Performance Model Based on high temperature tube burst tests Does not consider other failure modes (failure due to leakage) Specific to axial cracks Creep failure prediction dependent on understanding temperature and pressure time histories [ i r I I t i 29 i l}}