ML20236A977
| ML20236A977 | |
| Person / Time | |
|---|---|
| Issue date: | 03/01/1989 |
| From: | NRC |
| To: | |
| References | |
| NUDOCS 8903200249 | |
| Download: ML20236A977 (179) | |
Text
{{#Wiki_filter:r. , , l o o. L.
~
UNITED STATES NUCLEAR REGULATORY COMMISSION
. PUBLIC WORKSHOP ON THE , )
- INDIVIDUAL PLANT EXAMINATION )
~
.l Pages: 210 through 385 Place: Fort Worth, Texas Date: March 1, 1989 HERITAGE REPORTING CORPORATION' b OficialReportas 1220 L Street, N.W., Suite 600 890320o249 FDR 10CFR 4::902 cl n-D)
\ f'O O \
/ PT9.7 PDC ,
Washington, D.C. 20005 (202) 628-4888 t
- m. -
!y* . ., t o 210' I . UNITED. STATES'OF AMERICA 2 NUCLEAR REGULATORY COMMISSION ,- 3 _.. _ _ _ - - _ _ _._ _ _. _. x- -l
,' t '
.PUBLIC WORKSHOP ON THE :
4L
- VOLUME II ,
.5- ' INDIVIDUAL PLANT EXAMINATION :
i
~6 -_____-.-_______.x ,
7- The Gulf Room
-Worthington' Hotel 8 200 Main Street I Fort Worth, Texas 9
Wednesday, 10 March 1, 1989 11 The above-entitled meeting commenced, 12 pursuant to adjournment, at 9:00.a.m. k,7, '13 U.S. NUCLEAR REGULATORY COMMISSION PANEL MEMBERS: s.- 14 WILLIAM BECKNER, Chief Severe Accident Issues Branch 15 FRANK CONGEL 16 Office of Nuclear Reactor Regulation 17 R. WAINE HOUSTON,, Director. Division of Safety Issue Resolution . 18 Office of Nuclear Regulatory Research ,
, i l'
19 ERIC S. BECKJORD, Director Office of Nuclear Regulatory,Resea'rch , 20 i THEMIS P. SPEIS, Deputy Director 21 Office of Nuclear Regulatory Research j l 22 JOCELYN A. MITCHELL, Section Leader Vy Section B 23 Severe Accident Issues Branch , i 24 ASHOK THADANI l 25 JOSEPH MURPHY - I Heritage Reporting Corporation
- _ - _ _ _ _ _ _ _ _ _ _ . _ - nm
'j
~ .. .
r-n}
'\ \a
._) ~
211' 1 U.S. NUCLEAR REGULATORY. COMMISSION PANEL MEMBERS: (Continued) , l 1 2- ' BRIAN SHERON 3 FAROUK ELTAWILA
'4. JOHN H. FLACK .
1
'l 5 JOHN T. CHEN 6 ' FRANK D. COFFMAN 1
1 7 RICHARD J. BARRETT l l 8 THOMAS H. COX ;
-i 9 -__
l i 10 11 12 3 13' 14 15 j 16 17 18 19 20 i 2I . 22 23 . 24 . O 25 Heritage Reporting Corporation f L _ --
(-
?
i
)
/ -~ s 1 INDEX 2 PAGE 3 Session Chairman's Introduction William D. Beckner 213 4
Accident Management 214 5 Frank J. Congel 229 6 Question and Answer Period 7 Individual Plant Examination External Events John T. Chen 255 8 Guidance for Conducting Human Reliability Analyses 9 in an IPE Frank D. Coffman 268 to ) Question and Answer Period 276 1 11 Comments on Generic Letter 88-20 and'NUREG-1335 12 B. John Garrick, Pickard, Lowe, Garrick, Inc. 296 ("; 13 Yankee Atomic Severe Accident Response James R. Chapman, Yankee Atomic 315 14 Experience With the Big Rock Point PRA 15 Patrick Donnelly. Consumers Power Co. 327 16 Pennsylvania Power & Light's View of Severe Accident Issue Status 17 Paul Hill, Pennsylvania Power & Light 335 18 Interim Containment Sequence Dispositioning Criteria Casimir Kukielka, Pennsylvania Power & Light 346 19 Insights from PRA Studies 20 Wang Lau, Tennessee Valley Authority 353 21 Containment Event Tree Implementation Guidance Edward Burns, Tenera 365 22 I Question and Answer Period 376 23 24 25 Heritage Reporting Corporation Sell) 428-4400
v .. ; ,
! .? ,
213 P R'O C.E E.D I.N G S'
~
1 2
'DR. BECKNER: Good ~ morning.. >
.]
3 Can you please sit down and take your seats and-4 'we'll get started..
- 5 I'd like to welcome you back to the IPE Worksh'op ;
6 for the second day. l
'j 7 I have two brief announcements. First of all', ~l 8 yesterday we mentioned the Mark I containment performance
.]
9 improvement' recommendations that the. Staff had made to the= - 10 Commission.. i 11 It's my understanding that this commission paper 12 is now available in the public document room, . if anyone is-13 ' interested in it. e. k.g'- 14 The number is SECY-89-017. That's SECY-89-017,- f 15 the Mark I containment performance improvement program. 16 Also, you are reminded that if you asked questions 17 yesterday and have not 'given your correct name spelling to 18 the' person taking the' transcript, please do so this morning. 19 They're going to start typing the transcript this 20 morning, so if.you want your name spelled right, please niake 21 sure that they've got it. 22 This morning we have three presentations. I think 23 they should be relatively short, items related to the IPE 24 primarily. The first one will be on accident management. 25 Then we'll have a brief, discussion on external events, of
m f 4 __ 214 1 course, which is being delayed, but we'd like to say a 2 little bit about that. 3 The last thing will be a presentation on the 4 treatment of human factore in the IPE process. So with 5 that, I'd like to go ahead and get started. 6 I'll introduce Frank Congel of the NRR who will be 7 discussing accident management. 8 STATEMENT 9 OF 10 FRANK CONGEL 11 [ Applause.] 12 MR. CONGEL: Precisely the response that I was 13 hoping for.
. .s ,.- 14 [ Laughter.]
15 Just a little story about that. That was 16 intentionally put up just to get attention, because it got 17 mine. It's what you can get when you turn a staff. member 18 loose with chartmaster and a brand new color plotter. 19 What you'll see throughout the slides is that he 20 has clearly accessed all the options. 21 [ Laughter.] 22 Good morning. My name is Frank Congel of NRR. I 23 was very kindly invited to speak by my colleagues in 24 research today. 25 They set aside an hour, and I just heard them say 1 i '
. i
^ '
l v)- t
- fiO ~ l I
. 1215
[] 1 these are going to be relatively.brief. Exactly how long j 2 I'll take remains to be seen. - 3 I've written down a lot of notes, and I have all 4 the things down that I think 'I'd like to get across. How ') 5 long it will take, we'll see. 6 In any case, the accident management portion of 7 the IPE is relatively new. When I say new, we haven't -- l 8 Hold off the slides for a minute. We've dazzled them with I 9 that. j 10 The accident management portion of the IPE is now g 11 just under development. An important point t'o make en this: 12 It is clearly not as far down the road as tha other psrts of ! 13 the IPE are. T ud 14 You do know that the accident management is , 15 mentioned in the IPE letter as one of the subparts of the 16 whole program. But clearly we'll -- as you'll see in a l 17 moment and probably already know -> we're not as far down i 18 the road in developing the accidert management portion. , 19 What we're going to try to do this morning is give 20 you a status of where we are and where we hope to go. , 21 The formal interactions that we've had with you 22 people -- or really with NUMARC -- really began less than a 23 year ago. We had a kickoff meeting during the summer, and i 24 we've had several meetings in the fall and this past winter 25 discussing various aspects of what an accident management i
L. l . . 1 L.- v 4[ g,. . . Q. q 216 l l '1 1 program 1should be.and what it should contain. . i g in just January of this year,.a little.over'a month ago, we had a Commisnion meeting where we expressed to i ! 3 I 4 T the Commission what our plans were and what actions have 5 taken place over the past eight to ten months. l-
, 6 We're presently waiting for a forma.1 respouae from 7 the Commission as a result of the meeting that we had with . l 8 them. i i
9 We did have a quick meeting with NUMARC and,EPRI a couple of days after the Commission meeting, and we 10
]
11 discussed some of.the details of establishing a program and 12- describing how it would fit overall into.the IPE. 13 ' Before I go any further, let's see what we can 14 agree upon as to what an accident management program is. l ui The first slide, please. The second slide 16 actually. 17 I told you he optioned everything. j. l 18 Okay. What is accident management? The d 19 definition that we've been working with is it is actions 20 that can be taken by the plant staff -- and When I ,say the 21 plcnt staff, I mean everg.ibody from the operators to the 22 managers -- that would be taken during the course of an , 23 accident. ; k 24 The first thing, of course, would be to prevent 25 core damage. In the event that core damage is already [ i
S- u I
~
0; O i 7
- 217 1
underwaye.We want to terminate it, stop the accident, retain j 1 l 2 the core within the vessel. - 3 Simultaneous's wii.h all of this, wc clearly want to 4 maintain containment integrity and minimize to the extent j i 5 practicable offsite releases. 6 In' order to accomplish this, you don't do it on an 7 ad hoc basio. The analogy that I'm fond of making is tha 8 one about Red Adair. Red Adair is the famous oil platform 9 fire fighter. Any time they have a very stubborn fire, they, 10 call in Red Adair. 11 We may have a few Red Adairs in'our industry, but 12 generally we don't hcVe the time required to find a guy to 13 bring him- to Plant X and have him snuff out the problem.
,_i 34 So, in order for us to have an accident management 15 Program in place, we have to know in advance what could to happen, what we expect and how we respond to it.
one of the things: We have preplanning. Clearly, 17 3g with preplanning we have to learn about accident sequences. 39 We have a rather significant range of capabilities in the l 20 industry right now on how you can respond to that. 21 We have the Red Adairs. We .31so have typically 22 maybe smaller utilities that have had experience in working 23 with PRA's, for example, to understand sequences. 24 So what we have to do is get everyone up to a higher level in being able to rewpond. We can use hardware. 25
)
O O 218 I It may not necoscarily be designed f5r a particular action, 2 but it could be prossed into service in an emergency. I'll 3 come up with some examples of that in a few minutes. The other thing is to establish the thought 6 process, the attitude, perspective, that it's not a good 6 thing that every day, every year, throughout the life of , the 7 plant everything is going to be fine. When something isn't 8 fine, we just turn to the right page in the procedures, go 9 down to the right line and do that. 10 There are times when we've got to get beyond that. I So that's the basic definition of what we're 12 looking at with accident management. 13 Next slide, please. , Okay. So what are we doing? Accident management 15 really is extending the present depth principle and 16 i extending it beyond just the operators, but to the staff as I' a whole. 18 It's also establishing operating procedures well 19 beyond the design basis of severe accident regime -- or into 20 the severe accident regime, which is not typically dono 21 right now. 22 But the basic bottom line,is: Do we want to 23 enhance the staff operating capabilities should such an 24 unfortunate event occur? And we want to be able to do it 25 with existing plant staf f and with the existing equipment.
, e
4 O D f a . 219 l
~
-I Next' slide,.'please. i 2 Presently, we have accident management j i
'l
~
3 capabilities, as I ment 1oned-earlier, that. exist to'a wide 4 degree from virtually -- I'd'say almost nothing up to rather 5
- sophisticated capabilities. j 6 What we want is to improve everything, as I have ;
7 on.the slide, an incremental improvement in everything, 8 including the good programs. - We want te have a'hifa level q 9 capability across the' industry. - 10 In other words, what we're looking at is something 11 that goes beyond the ECP's, emergency opettating procedures. 12 Most of us are aware, the EOP's, for example, for BWR's 13 typically will fall into accident mana.gement regimes than :
.' 14 for PWR's.
15 We're looking for something across the board. 16 In putting together the program, we'rv committed e 17 to work closely with industry. It's not something that the i 18 Staff plans to do either cloistered by itself or just with 19 contractors have it pop up like a jack-in-the-box. 20 We want to examine and then adopt. This is gding 21 to be done up front in a cooperative straight-forward 22 arrangement. 23 We're very interested in this getting across 24 philosophically, as well as in reality. We're making some 25 commitments. 4 i !
V) - i v; 220 1 We think that this can be done without any major 2 hardware changes. I'm sure that you've heard this before. 3 But indeed we mean this, and in fact based on the 4 interactions we've had with our Commissioners, they're 5 committed to the same thing as well. 6 We're also not going to be looking for one of 7 l these things that says here's the procedure we want you to 8 implement, and then we charge ahead with our inspectors and 9 the checklist to see thau you have everything done. 10 We're changing our philosophy. I think you've 11 seen over the past few years a change even in our inspection 12 philosophy. It's performance rather than compliance 13 oriented. 14 We want to see results. If the results are 15 attained with something outside of a main stream of a 16 checklist, all the better. l 17 - Next slido, please. ! 18 The basic objective that we're looking for in 19 accident management is that we want each licensee to have a j 20 plan that provides & f ramework. The framework is the thing, i 21 the skeleton to which we're going to put: the muscle. I 22 We're going to have the capability to establish by 23 this frnmtwork thac will be -- for sure the capability, I J 24 shculd say -- for evaluating information on severe 25 accidente,. -_ __ ~
p - d 1 D .
- p. .
/ '
221-1 , :This'is.On'o continuing. basis. The informati'no we l
. ., d
,2 .have now is a database,'as;well as whatever we're going toi E 3 le'arn in the future.as a result of the IPE's, and:as my.
- 4. friends.in research will be doing, we_'re going lto continue 5 to ' study plant safety,inlierent'aspectsofhlantsafety,.and )
6 where we,can improve it over the years for the life of the- O
)
7 ' plant, we're going ts. 8 Were going to ultimately'one up preparing and' 9 implementing severe, accident and operating proc 6dures. ;The ) "- 10 operating procedures are going to apply.not only to the 11 operators, but they're also going to be applied to -- as ] 12 3 you'll see on the next slide -- other parts of the technical l 1 13 ) staff, those that make up the ?schnical support staff, as: 14 Well as the renior managements-l 15 We want to 6Ltublish ' a -line of command, a chain of 16 , command, so that important decisions can~be made of..a timely. -) f 17 basis. . .l 18 We were looking at things where they wanted'to
.19 bypass interlocks, change proced1res,'maybe sven-violct'e'a ']
1 20 tech speci .if conditions called for it. What we're looking i 21 for is the ultimate in response ehould the ' dreadful thing i I 22 occur. 23 The last slide is one that will most likely be jl 24 particularly applicable to tomorrow's session on accident l~ t 25 management where my staff people will be going into much -d 1 1 . l I i
-t j
.,. 1
'w'
i (~T p l q) G ; 222 1 more detail, z 2 But thesejare what we call the framework elemento. 3 There are five. 4 As was mentioned earlier, there's *.h3 accident 5 management procedures themselves. Those are the procedures 6 that are going to go beyond the design basis accident, that 7 are in place to accomplish what I just outlined under the a framework. . 9 The idea is we want to implement what was 10 described under the framework. To accomplish this, we're j j 11 going to have to have special separate training in the aveu j 12 of accident management that will go across the board. 13 These would also include the technical support
.m
. 14 people that would be analyzing, evaluating whatever accident 15 you happen to be in, mad come up with as creative and ]
t 16 innovative solutions to stopping the accidents as we 17 possibly con. l 18 The guidance and computational aids for acpident 19 response, those would be in manuals, computer codes, a *s well { 20 ac in the minds of the technigsl people who will b.e called , 1 1 21 upon should such a thing occur. I 22 You have to evaluate the instrumentation that you g hav An your plant for its capability for responding in i 24 cert ain severe accident situations or occurrences. What can ' 25 you expect from the instrumentation, given various
'~ip,f W Q ,T . . > j
?( .
'(f J i
.gm r. ,
7 223 !
"'2
- 1
. (. ,
1 .s eguence:s? j L 2 When is'it raliable?' When'is it not reliable?
^
.3 Are there any changes you want.to make? p 4 These are'the. kinds of. things that you'll~have;to f
5 evaluate and decide on your own. 4 6 And the last thing, of course, and what I
,s; -- mentioned earlier as well.-- with the misspelling, only the -
a second misspelling I've found, but the colors are otil,'.' ! 3 pretty -- it has the delineation of decisionmaking to , responsibilities.. ;
. .j 33 This sounds like perhaps on the. surface.it's not 1
. .]
12 that important, but'I wanted to emphasize it again,fbecause , 13 very of ten in the heat of a severe emergency, confusion can
,~s .,
3, ) 14 reign. 15 Unless you havs a clearly stated procedure on how
.16 to respond even in a.very critical situation,.this kind of i 17 thing.could cause you serious problems.
u 18, Let me give you an outline of.what our, timing is - t ! 19 ' ~ What otii1 expected timirg if over the next year o;; 6;s 20 We've put the slidec down now, and you cae ' cake 21 the sunglasses off. 22 As was mentioned earlier, we have the defi si, ion ) i g of what'we call an acceptable accident management proep sm 24 under development in house, and there is a parallel effort p under the sponsorship of NUMARCe f i l
~. ~
- p. -
;o
.o '
) .
) ,
224 3 The plan that we have is to meet with NUMARC! with
-- a version of their. draft sometime May/ June time frame ofL 2,
L } , 3 thi.s year , t'm sure with Bill Rasin here and Ray Anglin, they 4) will te L1 me if that's not correct, ban that's our 3 6 understanding. 7 In the May/ June tf.rae fre.me of this year, we will 8 be looking, at first., drafts .of what constitutes a framework. - g We're going to continue iterating these drafts
- s. ove.r a Period of about one year. Wo expect the final 10 11 ,
Package to be developed sometime perhaps lise next spri'ag , 12 about one year from now, At that time we would have to go back to the 33 $
,i 9 ] Coranission and rebrief them on the status of an accident
>( ;
\ management plan. That's the agreement that we have.
15 t When we get approval by the dommission, we'll'be i6 37 able to 'linalize the document and send .i t out in the form of 18 - a gene & Wer somedme mid to. late ned yen. In the shorter term, we have an intention to 39 distribute in the form of a supplement to t.he IPE Aetter a 20 list of generic accident management strategies -- what we 21 22 ca.13. " strategies." The strategies are really a list or a compilatlop p 23 of le.vn na learr.aa by un as it result of revicwing thirty-24 25 Plus PEA's. What these are are relatively quick fiv.es or 1 k 1 5 ! e _ _ J
_y 3 k
< 1 17 , ' ,
. 225" ,
3 changes in procedures:or minor changes in hardware that.we_ ') i 2 have,found can significantly reduce vulnerabj.11 ties. '
]
,a '
,' 3 How they apply.to'your. specific plant is dependent. L upon your individual review. But based on our long-term 4
5 review, ;we're g ing to be sending these out May or June of this. year.- 6 , Like I said, it's a form of a. supplement to the' 7 , s' letter for your consideration. g 1' g This is an example of a' number of things that. we found would significantly reduce poter.tial vulncrabilitios, 39
.y We have a procedure to enable emergency use of available pumps to accomplish safety,l' unctions. A diesel 2
! f 1
,3 etre syotem for injection into a swa-core under certain
) g circumstances. l
,, In other plants, that could be accomplished by addition of a spool piece or a cross-connect.
6 I g There are a number of these. I've got three pages je g
.of them typed up. ,
l But before.we send them out, it's our
,g understanding at this time that the Commission wants us to !
20 ) I ensure that there's no net reduction of safety by maybe 21 these cross-connects or connections available for any .of the other pIGeedures that we have 1isted here.
,23 Af tdr we go through this review -- and like 1 said, it should be done in a couple of months -- we're going g
9' I ____ __ _ . 1
i f , t p 226, s' ' f to send out'this list., Then it's going to be incumbent.on l 2 l both you, the industry, as well ,as us, to look it over and* i l 3 i1 determine what, if any, would be an enhancement to your 1 4 particular program. .; 5 L What do you do in.the meantime? In the meantime
" The IPE process, as I said,-is you continue with the IPE.
7 beginning; it'c starting; and the accident management is a 8 phase or two behind. 8 There was a question yesterday, they asked would l 10 we come up with a procedure, an accident management problem, Il that could reduce the vulnerability; do it. 12 Absolately nothing seys hold back until you have. 13 more details on an accident management program from us. I s 14 The thing is, we just want a program implemented 15 that ultimately gets to our bottom line goal of improving 16 plant safety. 17 The way I view the accident management program is I 18 the IPE provides the flesh or the mnscle on the framework 13
. tht!.t's being developed under this accident management 20 program.
21 The insights that are gained from the IPE are 22 i going to provide the details on the vulnerabilities that 23 have to be addressed. Artd once you address those 24 - vulnerabilities, the ideal goal, of course, is to reduce the U overall risk of the plant. __m _ . - . _ _ . _ _ . _ _ . _ _ _ _ _ . _ ._ e
,R ' '
J%, >Rnq a.p g - F y o;- .O . ' ' 4 227 1, --I " ~I can summ'arize-this;andftry to give'it to you in' y "
$2 the'three phases that I have'in:my mind,:in trying 'to keep
.3. .all' of these things straight.
'The'first-phase is this present time, where'you 5- .have a-parallel, but a' separate effort from'IPE. The 6 accident management is on one-track,:IPE on another.
7 We're ' going to examine and evaluate the lessons
, 8 learned from the.IPE's -- excuse me, from the'PRA's.- Those !
l
'9 will be listed 'in the ' table that I said would be coming:out '
to 'in a. couple of months.- l
~
11 You'will be getting the IPE. -The two of them will' r 12 mesh [in what . I call the second phase. The second phase is l 13 where you have the framework that's going to be developed by
.j f 14 both industry and NRC cooperative effort. ;
'15 7 .think it's . going to be available in the late is 1990's, as I said earlier, and you can merge the muscle on !
-17 the framework in about thar. time frame.
18 The third phase 'is to make sure that - the accident 19 management program.is what I call a living process. .All of-20 us are not going to stop learning about accidents after we ] l ' 21 finish this effort. 22 We have t,o build up to the capability of accident 1 23 management, the ability to fold in new informat. ion as : 1 24
, changes take place. ) 1 25 So it's not something that you get once, file 4.t j
. l L .
'~
,T
/%J
~1 j i
228 4 1 in a cabinet and'not look at again. . I 2 The degree of impor'tance that the Staff feels on l ) 3 this I don' t think can 'be stated strongly enough. We have 4 to come to conclusions about the overall safety of the 5 industry before we embark on new processes, new programs. ) 6 The one program that's coming up and is clearly on 7 the hori;on is license extensions. The way we see license 3 extensions now is it corvos as a logical continuation of l 9 this overall accident management IPE review effort. 4 l 10 When the Staff feels that there is a solid base of ) 11 understanding of severe accident phenomena and that the best 12 has been done by the industry to minimize that, we can 13 proceed to the next level.
\
14 But our iriitial discussions ca how we're going to 15 handle license extension issues always comes back to a i 16 performance and a probabilistic base of plant operatica. 17 With that, I invite you to listen tomorrow to my 18 colleagues talk on the' details of the accident management 19 program and I'll be happy to entertain any questions that ) 20 you may have. - 21 Thank you. 22 DR. BECKNER: As Frank said, there wil'. Es 23 presentations tomorrow on accident management, but I think 24 if you want to specifically talk to Frank, you've got to 25 catch him now because he's leaving later.
- i.
o
V_'
^
r
'u
_ 4 K - ;i e , f229-
~
%s
~ '
i ,Are there any questions at this point in' time?., 2 -MR.4FALLACARA:. My name is Vinnie Fallacara with
'j 1
. .3 . Boston Edison. I 4 I guess.I'mLa.little confueed. If'a BWR is ;
i 5 implemented EPG Rev.4, I believe.it meets the.four stated'-
-6 -definitions of accident management:' . preventing ' core - damage,:
7 keeping the core in the reactor, primary containment and l 8 radiological release. . ' r 4 9 Where do we.have to'go from there? l 10 MR. CONGEL: If you're- there, you may be home, if 11 you happen to have that level. 12 On the other hand, as I said, we haven't developed -l 13 all of the details associated as to what would constitute a q s
+ 14 good accident management program.
i f. ' But I recognize where you are and know where you ,
.16 are. You are one of the --
4 17 MR. FALLACARA: So that means that there.isJa , l 18 chance that those who just go home with'the upgrade to Rev 4 l 19 -may already be there, and th'ey'd have to do little else? - ; 20 MR. BARRETT: I'm Richard Barrett.with NRR. ; 21 I'd like to say a couple of thinga about that. I-22 think that one of the things that Frank said , earlier was - 23 that the intention .is'for the IPE to provide a great deal of
]
24 the meat that would go on in the accident management l program. ! 25 l
- \ \
1
y -
.g , :c .
W
^
n
. ;O D -
x :
;j . s .'
. 230 -
'E '
I' After yo ve performed.your IPE, the expectation.. 2 is that' there will be 'possibly procedures ~ that yow would l
- want to implement to try to eliminate any' vulnerabilities.
'3 1 4 that you've identified.
- 5. .And in addition, _ as Frank pointed out, there will 6 be a' list of generic' accident management strategies which 7 we're going to send out as a supplement to.the IPE guidance, a which we'd like you to evaluate as possible enhancements.to 9 your existing EPG's and FCP's. .j to I think that. it certainly is very useful to have 11 -implemented Rev 4 of the EPG's, but we'wouldn't'want to.say
. 12 that. that would be the end of it, as far as procedurec'for:
13 accident management are concerned. l
'I 14 The other aspect of the' program that Frank 15 mentioned was thS framework. itself, and the. implementation 3
' of the procedures is certainly one trac)S.of.what-we're 16 17 .looking for in accident management.
18 But the other aspects of the framcwork, such as' j 1 19 ensuring that there's-training for, severe accidents, for ] 20 .PeoPl e all up and down the organization, the evaluation of 21 the decisionmaking processes and the other aspects of the 1 l 22 accident management program are also quite important. 23 So I think the answer is that just implementing 24 EPG Rev 4 is not the whole answer. 25 MR. VIRGIL: Bill Virgil, Combustion Engineering. ; L a I 1 i
7 O j G (~)'s 231 > i T 1 A two-part question. On th6 EPG's WO have' ten 2 yehrs of history with the interaction with the Staff, ! i 3 particularly through the owners groups in developing EPG's . 4 and in' implementing EOP's or EOI.'s. 5 I wonder if y'ou see this process that'you've
- e. described as being similar to that, in other words, in'two 7 years will we enter into another extensive iteration of that 8 - type? .
g And then, secondly, the followup to that, which is 10 very important, 1slthe training of plant staff, not only the 11 operators, but, of course, all plant staff. INPO has 12 obviously played a very ke-j role in that. , 1 13 1 haven't heard you mention INPO at all in your i fm ( , J. ' 14 remarks, an d I wonder if there's thinking that has been-
]1
- 15. developed on that particular interface. J q
16 MR. CONGEL: To answer the second part first, my j l 17 choice of just talking with NUMARC was intentional . In .] 1 18 fact, I had to retrain myself because I spent quite a number ) l 19 of years interacting with INPO on health physics matters. 20 But NUMARC is our central contact. They will be l 21 involved with both INPO, as I understand it, as well an l a I 22 EPRI. 23 So there was nothing intended by -- 24 MR. VIRGIL: I guess in particular in the operator 25 training area, there was an actual formal Commission policy l
. 1
m,, , . , .e
. p, e n ; ,m ig , 4 ,
C.. . (3 , 232 (. N 3
. statement : that def erred to INPO. - I wonder if tha't has bean!
lLL . 2'
. thought about here..
C . 'MR.'CONGEL . It's something.I've thought about,-
;.3 4
but we haven't-decided on yet. '
.. l I want to emphasize --:and Rich also has: pointed .
- j. . 5 1 1
j ..out -- we,'re still at-the beginning.of this. There'are a: 7 1 lot of' things that.are just in the. thought- stage of how 8 we're g ing. to ultimate 3 y _ solve- them or come up with - them g has not been-established.
.10 But the first part of your question about the 1
I 33 details associated with a review process, I have to tell you j 1 12
.that'my' inclination is'to stay away from a detailed J 4
. iterative' process. .
]
13 I don't particularly care for that. It takes a 14 lot of ataff time, a' lot of -- well, staff time on both 1.5 gg parts,.the NRC.as well as the utilities.
-What I'd like to do is-come up with a program i 37
.that's clear enough so'that after the implementation la 33 started -- when it's near completion, we can integrate our 39 ,
g inspection process on a cycling -basis to look at the final' I 21 results. , It's what I have in mind right now, but not the 22 { detailed kind of looking at every line on your procedure ; 23 notebook and making a determination on that basis. ! 24 MR. VIRGILt If I could just make a remark in
, 23 l
c l
.1 j
O.. A- l
, 'V U ,j 1
233- -{
'~.s . !
3 ,
. response.
Certainly we both would like to see that happen,- ) to save staff time. The experience on the EPG's doesn't , indicate that that will likely occur. 5
.MR. CONGEL: Well, that may be what you say. . I'm 6
not convinced that that would be the case. 7 MR. VIRGIL: 'Thank you. I 8 DR. BECKNER: Are there any other questions for Frank? j 10 MR. TUNSTILL: Yes, sir, I've got one. II I'm Jack Tunstill with Florida Power Corporation. 12 Yesterdty there was a sign put up that said that the 13 vulnerabilities which are above the screening criteria don't
l# have to have justification given for not fixing a sequence.
15 'l I hear you say you're now going to put out a list, 16 which I gather has vulnerabilities in it. 17 My question is: This -- Did I understand that 18 right or wrong? 19 No, no, this won't have MR. CONGEL.: 20 vulnerabilities in it. 21 What it has 'is a result of looking at PRA's and , 22
, quick fixes to vulnerabilities that show up as potential .
23 main contributors to risk probability. 24 MR. TUNSTILL: How do they relate to this -- And I
- 25 But how do they relate t0 the .
again I'm not a PRA person. I - -
. y '
e . V L 4 ' l 1 234 I screening criteria? , ! 2 'i In other words, are you going to turn around in 3 the compliance side of the generic letter, look for us to 1
-4 'have had' documentation in our files saying how we responded !
5 to your list that you're providing as part of this generic 6 letter. . 7 Would we maybe haven't completed a PRA or haven't j 8 -- you know, the utilities that haven't gone that far yet? 9 Do you understand what I'm asking? 10 MR. CONGEL: I'm not sure I see what you're 11 getting'at. 12 Does anybody -- 13 MR. COFFMAN: Let me try to give you some f
;> 14 information about that.
15 What Frank referred to was a list of generic 16 accident management strategies that was developed in house l 17 in NRC by a group of people in house who have a lot of
.)
18 experience with past PRA's. l 4
- 19 They're strategies; they're not procedures. They.
20 point to general types of actions which you might want to l l 21 consider for inclusion in your procedures, either in your 22 emergency operating procedures or in special procedures that 23 might be available to the tech suppott center or the OSC or 24 wherever they might be appropriate. 25 These are strategies. The example that Frank gave
o (\ LJ V 235 - .m ~ 1 was the strategy of making alternative , injection sources 2 available to the core, or in the case of a PWR y to the steam , 3 generator, for instance. 3 4 The intention at this point is to put those 5 generic strategies out as an information supplement to the 6 IPE generic letter with the idea in mind that you should 7 consider them as you are doing your IPE and ae you are 8 learning about the risk aspects of your plant and the 9 potential risk vulnerabilities of your plant, so that you i 10 can do an integrated assesament of these potential 11 procedures. 12 The question you've asked, I don't thir we're in 13 a position right now to answer it; namely, what are the
) 14' t - status of these procedures in terms of ....
15 Let me just try and help a little historically. j i 16 The screening criteria about three or four years ago were 17 used by the staff to look at the existing PRA's and IDCOR 18 analyses in core test applications of the IDCOR method, to 19 look at the results and see what insights, what 20 vulnerabilities exceeded those screening criteria. 21 From those screening critcria the staf f then 22 selected insights that might be generally helpful, and those 23 were issued in NUREG/CR reports 4920, which have already 1 24 been issued. 25 - Those exist as maybe some insight nto what might i I 1
,, e. -
% /
236
^
1 be coming down later, as far as suggested strategies. But 2 .at least historically that'c how the screening criteria 3 relate to the insights. j 4 MR. TUNSTILL: The reason I asked this question -- 5 and it goes back to the comment I made yesterday regarding i l 6 instructions to your inspectors in the later years -- is if , I l' 7 you don't specify in some of your internal documentation, as : 8 a licensing engineer I'm going to be staring this inspector : 9 in the face, and he's going to.want to know, "What did you 10 do and where did you document how you handled this 11 supplement to the generic letter?" l 12 What yeu're telling me is these things are for 13 information for us to look at. I think you need to be very
,.3
,' 14 specific in your instructions to both your reviewers and 15 l your inspectors -- and I can't iterate this enough -- that f 10 l six, seven years from now, again we're not going to remember l )
17 f what you said at this meeting because the inspectors aren't 18 herd. 19 So you've got to document that very clearly as to , l 20 how we're to keep this documentation. ; 1 MR. BARRETT: And actually there are ' 21 I hear you. 22 a couple of inspectora here, but they're incognito. 23 - [ Laughter.] 24 MR. TUNSTILL: I know one of them is here. 25 MR. WALSH: Larry Walsh from the Public . Service
..7
, _ q 1 t if i lI '
s a o sv . d L - 237-
]
'1' -Company of New Hampshire.~
L+ u
. '2 ,
Looking at your slides, I see soms contradiction a. L i u . . ': H little' bit that I .think is ' there, and that is, are;you f
. 3 k' 4 > people going to look into the~other avenue'ofLemergency plan-L ~
i 1 5 . reactions 1by' plants offsite and onsite,.because right now I L 6 see' negative't' raining.onsite, because you always stop the-
- 7. operator.before he can solve the problem. -
a 'So the scenario will continue, and the offsite-y; 9 peoplencan do their thing. 10 MR. CONGEL: . That's a problem thatswe're'well ; 11 ' aware of.and are addressing 1outside of this context. I!know 12 what you're.getting at. , 13 We want to change some of the emphasis now from
,s(x f..M ' 14 : offaite more to onsite.
15 And just as an aside, I.tell you our-thought'is to 1
.16 _ require separate, in some: cases, training for both operators g 17 . responding to an emergency and then separate that from 18 offsite responses, because we all know that with offsite d q
, 19 responses you have to go to declaring a general emergency. '! 20 That does' stifle creativeness of the operators. 21 Ind6ed, that's a problem and we're working on that.
')
We know about that outside of this context. 22
]
I 23 MR. WAL5H: Thank you. I i i 24 MR. HOWARD: Edward Howard, Bosto'n Edison. I 1
. \
25 There has been a, lot of recognition in the 4 l --__---.u.-n_-L_.2__.._ - - -- -
- Y
- j t' ;
' '( .
O O I 238 l
'"*! 1 presentations about the elements of your various programs 2 and that they need.to be integrated. j I hear a sense in your mind that sometime'in the k 3 .
4 future you do. intend your actions to produce the integration 1 5 of .all these things. { l 6 For example, you've talked about modifications in 7 all three elements: accident management, generic 8 containment improvements, particularly in Mark I, and h f
'g certainly in your IPE's you're looking for modifications. i i
10 From the industry's point of view, and I think 11 from your point of view, one of the most difficult issues.to i 12 'make this effort effective in the.use of the resources -- j l 13 and it's going to be very resource-intensive, as was i 1
/' 34 discussed yesterday -- is to very much worry in the front 15 end of these activities that they will effectively merge at 16 the back end.
17 To set out to conduct each one of them on these 18 Parallel pcths with the expectation that at the back end 19 they'll integrate is risky for all of us. 20 We need to be attentive to on the front end' the 21 P l anning of the activities so that they will effectively 22 integrate in the back end. 23 Again, the example of modifications. I don't l 24 think it is helpful for any of us to -- in any program talk l 25 about modification individually and fail to recognize that i
, ~
.[ .
.z >
O 1 O l 239 1'V
.. ?
3 they;may be similarJrecommendations, but they're coming out 2 of -- at least'as I. perceive it, three different program I 3 plants.
~#
- I have a question, I guess, for the Staff in terms l of perhaps helping us understand your plans internally to
.6-
; integrate.those activities.
l 7 f I.mean, I've heard the intent to. But how are you 8 really going. to go about doing that and help assure that as
.the' industry proceeds with implementation,.that we don't f 10 redo these' activities, as the lessons from the ~ front end get f
Il integrated on the back end and we wind up having to recycle ; 12 through some of these. f 13 A modification is an example, and I think the idea - l#
~ of procedure improvements is.another example. l We look at the list of strategies that have been 16 identified by your group of people who are familiar with I7 PRA's . and have looked at useful strategies.
18 When I look at that list against the strategies 19 already embodied within 'he t boiling water reactor emerg4ncy 20 procedure guidelines, Rev 4, the bulk of those strategies 21 are already in there. 5 22 I don't know whether you've yet had time to take < 23 the group that has looked at those and now go back to your 24 experts in the staff who have. reviewed Rev 4 EPG's in BWR's i 25 and ask them to look at them and see how many of those are ;
N] ) 240 1 already there and what's new. 2 I understand you're asking us to think about it, 3 but I think equally it's important, and the NRC needs to 4 worry about its internal activities towards helping 5 integrate it because your resource is constrained, as is the 6 industry. - 7 That is a real issue for all of us. Go if you 8 could help enlighten that, in terms of the plans to reach 1 9 these intended end results you're talking about, I think the 10 schedule on which that gets done is an important 11 consideration t'nat all of us need to pay attention to, l 12 including the staff. 13 MR. SHERON: I think you've raised a very 14 important consideration. One of the things that we are very 15 sensitive to right now, I thir.k, is the fact -- as I 18 mentioned yesterday -- the Staf f is sitting on a wealth of 17 information that we received through the revieses of various j l 18 PM's that the industry has done and submitted, PRA's we l 19 have done on our own. 20 The objective cf sending out this information, 21 what we'll call the strategies, is nothing more t'han to 22 share what we've learned with you right now. 23 We would expect that even if we were not going 24 through this process today of an IPE or the like, that if we , 25 became aware of information that might be potentially useful l
6?@U ,so _} , 6 l. .
~
" i- ,.241 fh[ 1 2
to the. industry, we would share it with y'u. o We're not saying that you have to address it; you s..
'3 have to document how you dispositioned everything there, but
.I 4 we want to make it available to you right now.
,p.
.5 If you think.there's something that's good in
- I 6 there, it's your judgment if you would implement it now or 7 you may want to defer it.
However you want to handle it is 8 your business. 9 But the fact is, the information is there-and 4 10 available to you. 11 We would expect that as you get into your.IPE's, 12 you would most likely come across areas where relatively 13 practical Improvements could be.made to the plant through l
~ N .'.
.s 14 the use of either improving procedures, or adding to them or 15 the like, as opposed to very expensive hardware fixes. s 1 16 We would think that even a natural outcome of your { 17 IPE would be an accident management program that you would s 18 put.in place. 19 So we would expect that as a result of doing the x i 20 ,IPE, you'would take the information you learn from that and )
.i 21 factor it into some sort of a structured program.
22 But the intent is not to confuse you by saying, L 23 "Well, you have to put some sporadic procedures in place-24 today, and then you have to do some other proc later, and ] i 25 then there will be more stuff coming later on." 1 I 1 M* 1 l e i J ? 1...' Ic) . .
)
1 I 242
' We're saying, "You need to i eally think about, 1
2 when you sit down to do this work, how you best want to 3 integrate it yourself.'" l 4 MR. HOWARD: I fully agree with that. Our 5 experience would suggest that that is very important and
. 6 that is a major challenge for senior management at any 7 utility to undertake its planning so as to achieve that .
8 integration. j 9 But 1 think it's equally :f air for industry to l 10 expect of the MRC a similar senior management attention to 11 the integration of these program elements and that you not 12 elect to come to the industry from three diff erent program-13 perspectives individually communicating those to us without
'.I 14 senior management at the NRC doing it:s role in helping 15 achieve integration of those.
16 I think the activities are sound, as I hear about 17 them, and that they are going to be helpful. But I think 18 the senior NRC management needs to establish an approach 19 towards communicating -- working with the industry 20 internally on an integration effort as well.
'21 I hear -- maybe I'm misunderstanding, but I hear 22 - modification proposals for our review coming out of accident 23 management, generic containment improvements coming cut of a 24 program there and an expectation for modifications out of ;
25 IPE's, both modifications and procedure improvements.
l. m .i. A(j ("J 243 -! I'm just sugge, sting that you need to help-in.your []. 1 2 management of~all those programs, because it's a tough: test 3 for all of us. i 4 We would all, I think, be seeking the right 5 objective. It's easy to articulate the objectives, I think,. 6 and is very-difficult to define the implementation process 7 yet because we're still struggling with what it means'and 8 how to deal with it. 9 MR. SHERON: You might be interested that the - 10 Staff did produce a document called the Severe Accident-11 Integration Plan, which I hope you have seen. 12 MR. HOWARD: I have. It's the implementation of 13 all those arrows between the boxes that I'm talking about. i" 14 The diagrams are nice, but how do you implement ] 1 15 the arrows among the various program elements? That is not ) I 16 a. simple task for senior management. 17 MR. SHERON: That's one of the reasons we're - l 18 trying to work closely with NUMARC on this. 19 I think Rich wants to add something. 20 DR. SPEIS: We are mindful of these. You 21 mentioned whether senior NRC management is involved in this, 22 and I can assure you that Tom Early, the Director of NRR, as-23 well as Eric, the Director of Research, are spending a 24 considerable amount of time to make sure that these programs j 25 and this whole process goes the way you are talking about -- Il
---_-_--------_____----__w-___-------a-----__,n--
_ - _ . M
l y '> y' p) s. (m
\-]
. 244 ]
' 'S . 1 okay -- that there'is an integration, that there is an
{
. . .t .
2 overlap. 1' 3 It's an almost daily involvement, especially for 4 the last few weeks or few months; and it's going.to continue
$ on.
4 6 I understand that Tom is going to telk to you 7 people in April when he invites the utilities to -- share 8 with NRR, you know, scme of these things that are going on I g right now. 10 But we hear you, and we have to make it happen. I 31 agree with you that sornetimes boxes and arrows, you know, i 12 and all these things, they look good, but you have to get to 13 th'at' next stage, which is the implementation of them.
,.) 14 MR. CLEMONS: Good morning. Dan Clemons, ;
15 Rochester Gas & Electric. 16 Presently the lines of authority in the control 1 17 rooms are very clearly defined. Do you envision that the j 18 technical support staff and other managers that are going to 1 39 be trained in accident management, are they going to have to 20 have some sort of a qualification sticker that says, Okay , ] 23 you're qualified as a severe accident manager," and during a 22 severe accident, is there some point in a severe accident 23 that they would usurp the shift supervisor's authority and j I 24 take that away from him? 25 MR. CONGEL: I don't want to make any statements
1 i /3 v) l v i 245 l 3 about the usurping of authority or changing things in a way j 2 that could possibly be detrimental. 3 But what we want to do in change the thinking and 4 the planning process so that you are capable of going beyond t 5 those procedures that. address accident response up a certain 6 level. 7 Whatever it takes to respond to accidents beyond 8 that level is what we want to have examined and then g evaluated before it's put into place. 10 We're getting into a realm where one has to be 33 very careful because, as you'll see if we distribute our 12 table in the form that.we presently have it, procedures 13 where you defeat interlocks, violate procedures, violate w) 14 tech specs. 15 We're talking some very, very serious matters, not 16 to be taken lightly. 37 So before I pass any judgment, it requires this 18 evaluation -- very careful evaluation as to what you want to i e able to do, given the situation. 19 1 MR. CLEMONS : Thank you. 20 MR. CRANE: George Crane is my name. 21 I'd like to ask a question following up the prior 2 discussion on the Commission's integration of these things 23 int ne pattern. 24 We are n t really talking about risk because we 25 f
0 O . l 246 I haven't gone out to the consequence analysis. In the 2 consequence analysis there'is wide var'iability in' broad 3 uncertainties associated with how you get an acceptable 4 answer er one that's acceptable to a large, diverse 5 ' audience. 6 In looking at accident management, if you haven't 7 got to the . risk calculation through consequence yet, and 8 you're still talking about factors of risk release 8 probabilities and possible. pathways. ! 10 ~ You haven't yet addressed the dispersal and Il diffusion of those fission products somehow. You haven't o 12 introduced the biggest variable in the problem. 13 The calculational technique of using the average
.J 14 metro or representative metro, representative dispersal f 15 diffusion calculations, is like taking a three-foot average 16 depth to cross the river with a five-foot man, but the 17 middle is seven feet' deep. Walking on the bottom he'll 1
18 never make it. ] IS I suggest there is.a sense of responsibility in 20 the Staff -- or ought to be -- that looks through the risk ) i 21 calculation and the consequence and asks as many potential ) l 22 variables for management improvement, from that viewpoint as
]
23 you have in the other. facets of the problem. 24 If you'll look at operations that have been done 25 in the past with meteorological control on them, and efforts i l
/R R V- (): j P
t' 247' ] x 1 made to distinguish the meteorological categories, ycu can-2 find'in a good number of sites in the United States where 9 there's seasonal weather, that nine-tenths of the time you 4 can operate two or three decades down on your risk -- pure
,- 5 risk consequence calculation from what it is at the most I
6 adverse time. 7 To take the strictly most adverse time window for 8 your weather and mske your consequence calculation that way, i 9l you obtain an answer that's unrealistic for a goodly part of 10 the clock of the year. 11 Until' you get to the point in the Staff where 12 you've analyzed the value of' incrementing your. titae tin your R 13 consequence cal.cu la tion, I don' t believe that it wou'id be 1 14 prudent to put forth accinent mtnagement strategies atrictly 1 a 15 on the most adverse time window. 16 My question is: What's the sense of the NRC Staff 17 and responsibility for including the. consequences in 0; 18 determining the priorities and urgency of your accident l l 19 management strategy? j 20 MR. MURPHY: George, let rie respond to that two 21 ways. 22 First, in the way the NRC looks at risk, in codes 23 like MAX, typically What 's done is we take a year's weather j 24 pattern and start the' calculation randomly by rendomly 25 selecting in thac year's weather pattern a num.,ex of tiines I L _ _ _ ___ _
(O (e-)\ 248 2
,, .3 1 and a number of Monte Carlo runs.
2 So we represent the actual meteorological 3 distribution of the plant, rather than looking for a worst 4 case situation. 5 In calculating the risks, we do exactly what 6 you've suggested, in terms of sampling from the entire 7 weather distribution over the year. - 8 So-we're not focusing on a worst case situation. 9 I think in terms of accident management -- you look at it 10 twofold. 11 There are a number of things.that you can do ( 12 aiming at prevention of core damage, and you want to do 13 those without getting down :-Ito the nitty-gritty details of
.~.,
14 consequence modeling. <
I 15 It's a hierarchy of events. Obviously, then you i 16 look .at the mitigative functions. One of the things you 17 look at in that area is the effect on consequences. l 10 But I personally think you can get more bite for 19 the buck out of focusing on the prevention end, rather than j i 20 on the nitigation end. 21 MR. CRANE: I fully endorse that. That's the l 22 approach the industry has had for years. But accident ) 23 management was defined as the period during the accident. 24 That's the di fficulty. j i l 25 I merely wante.d to get a sense of what the -- ] l l l i - - _ - - - - __ _.8.
,y ,c V . V 249 1 MR. HOUSTONt I would like to add some comment on 2 that.
3 The accident management program, it seems to me, 4 has to be viewed as that combination of strategies which 5 could be developed into procedures for which the plant staff 6 still has an element of control with respuch to the outcome. 7 Now, there's an interface with the plans and the 8 participation of the utility in offsite amergency plans. - 9 There's no question about that. 10 But I think the real focus is on the things that
- 11. you can still control and do something about within the l'
12 plant that can perhaps lower the likelihood of large 13 releases. 14 But I don't understand that, you know, carrying, 16 for example, the IPE process into the Level III PRA category 16 and trying to do some -- sort of bottom line risk 17 calculations would contribute to that significantly. 18 MR. C RAITE.t Well, there's a cense in the emergency 19 procedures on boileIJ for venting through the vet well, 20 there's a cense of monitoring inventory-limited water, if 21 you're going to use the sprays in the dry well. 22 There's a sense of competition of cooling the core 23 or providing scrubbing. Those things are already accounted 24 for in the procedures that the operators use. 25 You might also decide that it's better to veut in
* .1 ' 1 1
O; O - (
-\\ - q 250 q
.1 t.he af terncou than it is . in the early morning, if you have
'j l
tj 1 a, an inversion process. Those things are in the' procedures, , 3 j and therE's a sense of 'that amongst the' operators. 4 So I don't know what you could put in the list l
, , . i 5 that if you haven' t included the risk factors out there and j l
6 gone that far, that you could really significantly tmpact , 1 7 'what has Mroady been thought of. j d a i We can await your list and see, l j
, 9 MR. TUNSTILL: Jac}$ Tunstill from Florida Pcwer !
]
10 Corporation.
}
11 As I read your documentation that i yon sept'out, .-
.2 12 the generic letters, an-d listened to the focus here on the j s
13 accident mandgement, I have to stop and think about its Now' 14 how am I going to de this as a licensing engineer and as an .! 15 advisor to my management on how to implement some of this .] I 16 stuff? 17 The question comes back to me that in my company's
],
~.
is precedures, whether we make pcocedure changes or hardware I
- 19 changes, the grunts (the guys that actually do the work) l l
20 have to sit there and do a 5059 evaluation on these changes. i 1 l 21 of course, the question that'the 5059 reg asks y'ou ) ! 22 to look at is how you evaluate this change against the 23 accidonte as described in your FSAR. . 24 ) Ok&y. Now, what I'm asking you is -- I'm not 25 asking. The statement is:, The FSAR doesn't consider severe I i l l i - . 4 ,
q J V v l 251
- 1 accidents, you know, because there's a loss of cool &nt 2 accident, but not going into the realm that you' re talkir.g ,
3 about. here. 4 Are you expecting us now to have two separate e 5 procedure change mechanisms? For example, if you're gaing 6 to make chsnges which offect the plant as evaluated for LOCA 7 ( analysis, you still would do it under 5059. 8 However, if you Pant to start making procedure 1 9 changes or hcrdwars changes that relate to the severe l t 10 accident, how do you satisfy the 5059 criteria? You're l
) 1 33
+
going to be evaluating against something that 's not 2 presented in the document which is your license. Have you thought about that7 13 l !
/"s i..' 14 MR. BARRETT: I'd like to -- Let me just say a g3 7 word or two about that.
33 I don' t know the specific answer to your question, i 17 but I do know that in the past when we've had PRA's 18 submitted to the Staff, in most cases the utility that has 39 performed the F2A has idsatified a vulnerability or two 20 which have been addressed by means of new procedures. 21 I don't know exactly how those procedures wery 22 implemented. I belieVe they were implemented as part of tlia 23 5059 proceas. 24 So cithough I don't know the detailed answer to j
)
25 y ur question, I know it is doabl,q. 1
' ;./c i-4 O
yr (V^Y
/
252 I Let me promise to final out, ?/ Jake sure that we get I 2 a clarification on "a" th'at proc'ss e would work. D
?
a r
!!R: TUNSTILL: ihat has to<be a definite 4
clarification h%'ause our problem becomes how do we satisfy 5 ' thuso requirements . j 6 It's well- and good to present a program that ip , 7 r your collective thoughts, but you've.got to step down one or. , s 8 I two levele and figure out what it is you're tellir.g the guy 0 to'do and how is he supposed to comply with it frcm your 10 regulatory compli9rAe side of the house, because you'll I' hammer us to death on -- you know, "You dLE '.t do rhic 12 -
- 3. properly-in accordalde with 5059
13 I'm going to sa*. "This stuff was not. considerni l' under that " 15 Of e.ourse, then we'll get into a r,1,:e dialogue j 16 wi.th each other. > ! 17 ' [ Laughter *] . i 18 MR. SHERON: I would just add that' I would imagitttL L 19 it would proceed the same way that the Rev 4 revision to the 20 BWR owners group procedures -- 21 BWR, I dors't know anything about MR. TUNSTILL: 22 the BNR's. , 23 MR. SHERON: I'm just saying it was done. 24 I moan, they have procedures thich go intG the EE severe accident realm, and they were ;submittad; they were I
)
/
n ., ?- - - -
,1 cy a p ,,\ , ,
o
'i t 'i l{ .<,
Q th. .
)
f: .. ,
- l. .c b i' 253 ]
S. 1 m/ > reviewed Pnd. approved by the Staff, as I understand it. Ls 2 I was not. involved personally in that, but it has ;
> / '
1 3 i j baen done. 4 And I th$.nk, you know, it's something -- you know, ! l
' 3 } you may want.to talk with someone'that has-a BWR. But it's i l<
b 6 l a good-question, and we can certainly get back to you on it. MR. TUNSTILL: Thank you.
; 8 , DR. BECKNSK Okay. Let* s take on62 more . question, s 5
9 and thea we'll have to .nove on. , 10
-MR. Ki/WZ: My name is George Klenz of I 11
' Commonwealth Edison. i e
i 12 I have.really two observations'to offer. First of 13 all, with respect to the question of meteorology that just i
- 14 arose, 1" work botn in the PRA are,a and as one of the.Edisors- t 15 , designated accident manegers.
16 I think it's lvir to say that neither I nor any of 17 my contemporaries in the ccapany would want to do anything 18 leso, given meteorology, than we would given unfavorable 19 meteorology. 4 20 I think we view the accident management program as 21 largely . independent of current meteoro:Mr , although we 22 obviously have to be cognizant of what that meteorology 23 might bt, and its effects on what we're doing. 24 The main reason I dame up here was to offer on 25 observation in support of what Mr. Hcward brought up a j 4
I ., t ' 354
- l. .
1
'minste ago relative to integration of everything that we're-2 doleg' programmatically here, and to;1 eave you with one other
?; r
- 3 '
thought.
'4 We not only need to integrate.these things for e 5 each plant, but fer thoc9 utilities like Commonwealth Edison 6 that-hrve twelve units and sin sites. and a diversity of 7 types and ages oi units, we need to eneare that we're 8 programmatically consistent across the boal'd, boiling- Water j 9 to pressurized water, and within the various generations of
- 10 these plants as well.
11 In the past, as I think we all recognize, we have }
-)
12 far too of ten found ourselves with different standards for 13 the boilers than we have for the pressurized,.different
.s
/ 14 ( stanJards for the older boiling water plants than we have f5 ffor t'ue newer boiling water plants. '
16 ) Not intentionally obviously; it's just the'way i 17 various organizations have fit or fhiled to fit. But I & 18 think in this particular cree when wn': e talkins about 19 integrating these things within a company the size of nine, 20 across tha diversity and the spectrum that we have in our
- 21 company and others similar in complOxity, we need to ensure 22 both . from our side and from yours that we' re not only l 23 integrated, but we're consistent 14 that integration.
24 , Thank you.
<s l ,
. 25 DR. BECKNER: Okay. Thank you.
j ; I l i (
)za m- .
mN
~mp , JL \ 1 o O ; 1 255, J'
~- . .. ,
l d<s 4 1 I think we had better' move on. We're going to' 2 ; have plenty of time I think after the break for more 3 questions ~. l 4' Let ne'go ahead and introduce John Chen, who -- I i 5 'thinkLthis will be brief'-- will briefly :alk about external.
] ?' 6 evknts and what we are doing as far as looking ahead:as far H
7 as What might happen iri that area. 8 STATEMENT - 9 OF 10 JOHN CHEN 11 MR. CHEN: ' Good morning. i 12 I think my talk will be brief, but I also probably 13 will be able t6 give you some 'goed things about integration: , 1
~s
'# 14 f.n antexternal event. . l 1
15 However, I don't think I will bi. abl'e to give you
\
16 such a colorful talk like what Frank just.-did. 17 Last Thursday when Mrse Bush was in Tokyo,-she 18 asked Mrs. Taganaki, "What is the foremost thing you
'4' 19 , Japanes& fear the most?"
20- The answer ist- the earthquake, t'ne' fire, the-4 21 lightning and the old man. Fr
/C I don't know what Japaneae fear about old men. My 23 ' hunch is _ probably because they're wise; they're probably 24 more experienced. They muy be able to give you a much more t
S colorful talk. l I i 1 . .
t ; & i 256 D 1 [ Laughter.] 2 But I do know one thing for sere. The earthquake, 3 the fire and the lightning is part of the external event. 4 What do we consider as external events? We have a 5 definition, I think a lot of people in the PRA arena 6 probably knows very well. 7 In the NUREG/CR-2300 we have defined the external f 8 event as all event initiators which are outside of the j 9 system, either under normal operating conditions or the 10 safety operations, safety equipment, these type of systems. 11 There's two special species. One is internal 12 event; the other one is fire. 13 They are kind of strange animals. They can be 14 induced within the systGm, but on the other hand, they also 15 can be produced outside of the system. 16 The next slide, 17 In 1985 NRC issued a safety -- severe accident 18 policy statement. In this stat <tuent we concluded we don't i 19 need at that tinte to make specific rulemaking against eevere l 1
~
20 accidents. -
}
}
21 We feel there's to undue risk to the public. I 22 However, we do visualize there is a lot of PRA which j t i 23 indice,ted a significant risk. ] 24 Therefore, we need a systematic examination to 25 idontify plant.-specific vulnerabilitiso.
m ' , l 1 O v /] U j I I ( 257 j 1 The severe accident policy statement did not 2 differentiate what accident -- whether it's initiated from {
)
3 external events or from internal events. 4 4 However, the past PPA's, as you can see from this ) 5 chart -- I tried to use something which we just borrowed , i from Pickard,'Lowe & Garrick, because they have this /
- 6. I 7 colorful chart, and I think it's very easy for me to use, so
_ ._ 8 1 just reproduced this thing which shcws the externa 1 event g can contribute significant risk compara' ole probably to the j 1 10 internal event. j 11 However, the perceptions may be quite different i 12 because external event, as everybody knows, we may introduce ! l 13 evidence of it ist our analyses. ('
') , ..- ;g Next slido.
15 For all those things, this is -- In our entirs { l 18 Program this ir. formation gives you a lot of background l l 17 information. If you want to take all the detailed l 18 information, you probably will be able to go through all ig those lists, and you'll get a lot of information out of 1 20 these. 23 SECY-86-162, we assess what we consider as 22 important things for that. l 23 We decided to def et the external event. We're not l l 24 going to have the same schedule as the internal event. 25 I'll discuss a little more later on this SECY I _ _ _ _ _ _ _ _ _ _ _ .. I}
O O a:.
. _( ; ,
-u,
-253
.1 -
u g .g , I JT 162. . 2 t We initiated a severe accident external event .j
, ' i 3 workshop in August '87. We had the first dialogue between the NRC and the ' industry. ,
5 At that time NUMARC was just in the transition' 6 staga. We discussed this issue with them, and we discus, sed
-7 all these things what were our. findings for our Phase I 8
program, try 'to find out what's important in the external { event, where we needed these -- what kind of program we 10 ' needed for the external event. i Il following that.in 1987,. NRC felt there was-a lot i 12 of integration that would have to.take place. This will j 13 y probably answer a lot of your concern about what NRC's
'4 management is considering in this. area. .
15 We do recognizo that there are a lot of programs-16 within NRC that have a lot of overlapping ~ about each other. lY ' ! So we are thinking about how best we can provide these 18 things, integrate all the plan.
' ' I9 NRC senior management established an external 20 'levent steering group which consists of quite a few 21 !c management people. Wayne Houston is one of the members of i 1 j 22 that, and we have Carl Otto, Larry Shaw and Tom Novak. ]a 23 Those are senior management people. They're going l l
24 to recommend how we can best deal with external event and j
)
25 how we can coordinate all the effort and integrate e::isting ) 1
.i ..
L 0 .__-____.--__-_-_---_-_2.-__--.2---_
)
7 , l O O
+ -
s . 259 I y% i I U NRC programs into a coherent program. 2 Then the next one,;I think there's.a 1ot of , 3 You probably also see there's a lot of inforniation . 4 beautiful flow charts that talk about how the IPE's ! 5 internal, external, how do they merge into identified 6 probabilities, and there you'can go into the' severe accident 7 management, and also the closure of these things. 8 That's in the SECY-88-147. ! 9 In that paper -- Commission paper, we also have a 10 fairly detailed description about what's going on in the NRC 11 '
'related to external events. ;
12 The next two things, I think you're all familiar 13 - .with 88-20. The Generic Letter itself. The Staff decided ,
... a
# 14 we're going to go for internal event first, and then however 15 we leave ourself open.
16 If you will be interested to go through the . i
-17 external event, we would be more than happy to review.and j 18 discuss with you.
19 In the externa'l event and the SECY-88-20 we stated ] 4 20 licensees at this time are only requestod to perform an IPE ) 21 internal event. - l 22 However, you're going to also deal with internal 23 flooding,'and there are some additional things you probably , l
, 24 have to watch out if you want to do internal flooding, what 25 you should watch for,'some external event related. I'll be i
1 i_ ___ _ _ . _ _ _ t !
s _ v 260 1 talki.ng about that at a later stage. 2 The deferment for external event'has three 3 purposes. The first idea is trying to permit an 4 identification of what's important in the external event 5 needs systematic examination. 6 If you review the NUREG-2300, you'll find out in j 4 7 that NUREG as far as the external event is concerned, we 8 have about two pages long of items. 1 1 9 Here at NRC we don' t feel every one of them is 10 important to be looked at. However, we do want to know a 11 few, which are important and need to be taken a systematic k 12 examination. j l 13 Another thing -- at the same time we also want to f 14 know some of our design criteria in external event is really 15 quite conservative. We want to know whether existing l 1 16 regulation would provide sufficient protection for the l 17 nuclear power plants fn regard -- against severe accidents. 18 That's our Phase I program for the external event. 19 We also want to provide some kind of guidance for 20 you to.utiliza. 21 Also we want to identify, if there is some . 22 available methodology which can save your efforts in 23 carrying out -- identify what's vulnerabilities in this { 24 purpose. l 25 Just to give you an example of what we feel as a l l l l l l l l l
O O , 261 1 sim'plified methodology itself. If you are dealing with l 2 seismic, you prbbably are' aware there is a program called 3 Seismic D'esign Modeling Program. 4 We collect all of the information from past PRA's j 5 -- seismic PRA's. We try.to simplify.the methodology. We l 6 try to give you a much direct approach to-utilize this 3 7 methodology. 8 And yet you can identify what the potential 9 vulnerabilities are in dealing with seismic hazards. That l to program has been tried out on three plants now. l 11 Maine Yankee was the first one, and Canaba was the ) 12 one that developed aJaimilar type of method. 13 We have a joint plan on the Havish, which the
. ,r . -. "< 14 result will come out sometime in August.
15 We also would like to develop-some kind of 16 guidance to provide you for IPEEE's. We also are going to 17 integrate other ongoing programs. 18 What we consider as an ongoing NRC programs, I'll 19 discuss in the later slides. 20 This will be' our Phase II program, which is 21 ongoing and probably will be done by September. We are 22 hoping by the end of the year we'll be able to tell you what j
\
23 the guidance is going to be and how you can carry on these 24 plans. 25 Next slide. , t is. . iii- isi --i i i, ir i .- - i-M.
O O
.l 262 I
As tar as the NRC Ongoing external event, we have 2 a whole list.of programs. Actually this list probably can 3 only expand for, those other things we realize at this. stage. 4 The seismic design margin program, I discussed a 5 little' bit before. But this program itself is a result of 6 the API's concern. 7 They are wondering whether all of our seismic 8 That designs in the past have the clip-type of things. 8 means you may not have.a significant margin beyond your 10 design basis, say for SSE situation. 11 So we went through a very long seismic PRA type of 12 We can development, and then we simplified that method. 13 utilize that to address that problem. Another is the seismic hazard calculation that's 14 15 used in the United, States. This program is a result of the 16 USGS changing their perception about the Ch'arleston 17 earthquake. 18 In the past we always say Charleston earthquake is i 18 associated with local geological structures. But in 1982 l 20 USGS changed their mind. 21 They said they cannot positively identify any J 22 specific geological structure associated with Charleston j 23 earthquake. 24 And another thing, you can interpret that. The 25 . Charleston earthquake can be occurring anyplace along the l
- . _ a
N <*7
) >
263 1 Eastern Seaboard. 2 That's one of the programs NRC has undertaken 3 about six years ago to try to identify what the seismic 4 hazdrd is along the eastern United States. 5 The report,is just about that and will be issued 6 probably in April of this year. 7 There is a followup program also looking to the 8 eastern seismic hazard in the eastern United States. We are 9 trying to effectively integrate these progrataa into our 10 thinking. 11 There is a NUREJ-1150 that also deals with 12 external event. A lot of things were learned from there; 13 we'll try to see her we'll be able to ef fectively utilize "' 14 some of those things in there. 15 We also have the program, fire risk sceping 16 studies. The report involved is NUREG/CR-5048. 17 This essentially is assessing whether the PRA 18 dealing with fire and fire risks and the current 19 regulations, Appendix R, whether that's sufficient or not. 20 We will also try to find if there are any 21 " additional issues which have not been issued in the past are 22 significant enough as a fire risk. 23 4 In the fire risk scoping study they have 24 identified six issues which are considered significant. We 25 are considering to incorporate some of those into IPEEE's.
h , [j c . s 264-i 1 .There is-also a' study in the past in the NRC, we
~2 are dealing with the flood itself,1 what's'the probability of
, 3 a flood.
4 The Academy.of Science, they have.a report on-5 that. . We are going to essentially utilize that in 6 addressing the flood issues. . . 7 There is another one -- all related'to seismic, l' l 8 USIA-46, th'e seismic qualification of the equipment in j 9 operating plants. We have examined that, and there's a lot 10 of things --'it's similar to what we can utilize. 11 There's A'-45, that has shutdown decay heats that 12 has been subsumed in the IPE's. The external event'will be 13 taken care of here in the IPEEE's.
. gn.
f 14 The USAI-40, the part they identified, there was _ is no safety issue primarily associated with ten. That has 16 been merged into A-46 and it will be very easy.for us just 17 to take care of in this IPEEE in the seismic examination 18 area. 19 In the USAI-17, it's internal, but primarily what 20 we are thinking about is the spatial interaction aspect, 21 because in a seismic event you have the spatial 22 interactions; and it's very easy to be taken care of through i 23 these IPEEE's. 24 There are certain things -- less important -- like , 1 25 B-103. That's a design for probable maximum precipitation, d _ m___ _ _ - _ _________m_ .______.______d_____.___..___. . _ _ . _ _ _ . _ _ __ _ _ _ _ . __ _ _ _ _ _ _ _ . _ _ _ . _ _ _ _ _ _.
O , D
- 265 -i em
'/
1 Which have been identified in the past as a' generic issue.- L. 2 But it's not -- We can't examine ~that and it' 3 would not be as significant. , 4 There is another.one I did not put on here. It's i '5 l 4 5 GIA-57'. 'Miat's actuat' ion of suppression systems. _f
- (
6 In our proposal we are considering-whether we'll. B 7 be able to take that into consideration,.Whether IPEEE will 8 be sufficient to take care of that concern. J 9 EPRI' has a program collecting . the . impact . of these 10 suppressants on the safety system. / 11 Essentially those are - the ' things we -are doing, and 12 I think our management are aware what's -- If you examine-
'13 the SECY-88-147, then you probably will know'what kind of 3
(.ar' ' '14 integration we are thinking about, and we do take our best. 15 If we can call on others among ourselves, as well is as .the industry, we probably will get things done in a much 17 more rapid fashion. l 18 This concludes my remarks. 19
- DR. BECKNER: ' Thank you, John. ;
20 I think we'll hold questions until after the 21 break. 22 One thing I do want to say, though, before you l 23 leave is this afternoon we have scheduled prepared public 24 comments. There is a possibility that we might be able to 25 get some of those started this morning. _ _ _ _ - . . _ _m_ ._ _ .__m.___.E-._.____m_m.-.__A.__._- #
~
~ l' -
, () '( ) -
v 266 !. 1 So anyone who has reserved a time to speak this 1 1
-2 ' afternoon, if you're early on the list, why you might want (
3 to be prepared to do it before lunch just in case we can'get 4 -to that point. 5 Okay. Thank you. 6 Let's get oack together about 11:00. 7 [ Recess from 10:25 a.m. to 11:00 a.m.]
.8 9 ,
i 10 11 12 13 , f.'4' ~
.. 14 15 16 17 18 .,
l 19 i 1 20 21 22 23 24 l
; 25
.l .
__ d1_.m . __1_.__ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ i
D- o' 267
.DR.-BECKNER: I think I'd like to' start 2 out by talking briefly a little bit about schedule. We've
~
. i
- 3. been ahead of schedule, I think, all week, primarily because j
~ 1 I
*-~ 4 people have been very. cooperative about getting back into i 5 the room, at least until now. You must be getting. tired. ,
6 We've had a lot more time for questions , i- i
. 7 and comments than what we anticipated. So what I'm looking 8 forward to is I think we may be able to close things down l
9 early tomorrow. 10 Tomorrow afternoon was scheduled simply 11 for questions and answers and comments. Given the fact 12 that we've had so much extra time, it's likely that we'll {} 13 be able to conclude the meeting by lunchtime tomorrow, 14 and I think we'll try to shoot for that. 15 So if you want to make any travel chenges 16 appropriate to that, we'll give you a fair warning on that. 17 The other thing, we've had a request from 18 the recorder for the people who are making presentations, 19 if you do have a prepared text' written out and have an' l 20 extra copy, she would like it. It just helps her make 21 the transcript a little bit more. J 22 With that, what I'm going to do is I'm going 23 to have Frank Coffman go ahead and give his presentation 24 and then we'll open it up for questions for both John Chen 25 and Frank. 1 Heritage Reporting Corporation j w _ _________ . _ - _ _ _ - _ _ - _ - _ . n wra _ __ _1
i m V 268 1 Frank Coffman from the Office of Research. 2 MR. COFFMAN: There's a provers that states 3 that a wise toolmaker knows at le'ast three potential limitations l t i 4 of the tool. i 5 If risk assessments are tools, then three 6 limitations are the ability to account for common cause 7 failures, the availability of. credible failure rate data l l 8 and the~ elementary state of human reliability analyses. I 9 I think I held in reserve sequence binning. 10 Within the limitations of the current state 1 1 11 of the art, my purpose this morning is to provide some 12 interim and general guidance for conducting human reliability 13 analyses in an IPE. 14 If we could go to the first viewgraph, please. 15 The objectives of human reliability analyses 16 are to discover and identify and model the potentially 4 17 important error-likely tasks and, also, to provide consistent i 18 and valid estimates of human error probabilities for inclusion i 19 in the IPE. 4 20 Consistency within the IPE will facilitate 21 the discovery of the leading human and induced human 22 contributors to the plant specific vulnerability. 23 Validation of the estimates will facilitate 4 24 keeping the human contributors comparable with the hardware 25 contributors and will keep the estimates indexed to the l Heritage Reporting Corporation W _ _ - _ _ _ _ -
i E ,
~
. er p ,
l 0 D j i'> > 269 I individual' plant operating and maintenance. personnel, j 2 < I'd like to go to the next viewgraph. 3: This is simply a list of topics that-are
]
, t
' 4 typically included in human reliability analyses.
5 There are these five: The task analysis; i 6 perfonnance shaping factors; the conversion to probabilities; , 7 consideration of the sources of the input data; and then 8 some consideration of uncertainties and sensitivities. ) 9 As far as the scope of tas'k analysis, this 10 is simply the decomposition of the important accident 11 sequences and the cues to which the teams re'spond. 12 At this point in the IPE, in the human I,]) 13 reliability analytes part of the IPE, it's helpful to have l 14 operating personnel participate in the task analyses to 15 identify these cues that the response teams will be cued 16 on. 17- Also part of the task analysis is to identify f 18 the actions and the knowledge necessary, maybe any decisions 19 psrmissives that have to be identified in order to success-20 fully prevent or mitigate the consequences of the sequence.
. 21 As far as the scope of the performance shaping 22 factors -- By performance shaping factors, those are the u./
M items which influence or shape the performance of the 24 individuals during the accident sequence. 25 These are such things as -- there are over 60 l Heritage Reporting Corporation __ caccra
u
.s s n U V 3 270 l l
't of them, but they are such things as the time available to i 2 perform the tasks- the training, the adequacy of the training 3 received proceduros, whether or not they're available," !
"' 4 whether or not there's a policy to rigorously adhere to l
5 procedures; whether or not there's a practice to tend to 6- do what's needed, regardles's of whether the procedures k l
'7 are uptodate; things like instructions, oral instructions j 8 and policier that the company has.
9 ~These factors shape the performance of the 10 team and need to be -- The scope of the factors to be i 11 looked at for each task that is important to an accident s 12 sequence have to be identified and have to be examined. [] 13 And then included in this would be scaling 14 of these performance-shaping factors. Scaling may be jargon. 15 It essentially means the quantification, the means for 16 quantifying the influence of the performance-shaping factors. l 17 Another topic that is typically included 18 are the methods used to do any screening analysis, if that's i 19 done, and the methods used to actually develop the human 20 error probabilities, and then also how the human reliability l
. )
21 analysis is integrated into the IPE.
.. 22 Some of these methods that are available .
l
'es . j 23 to establish the human error probabilities, I was going ;4 24 to list them for you, but I think I'll ask, John, if you'll {
25 put a backup viewgraph on, it might be easier than you l l Heritage Reporting Corporation *
- _ _ _ cacoa
t r) t p g) . 271 1 listening'to me identify-some of these references.
]i l
-2 If you look there in the center, these are' l i
3 different methods for quantifying human error probabilities. i
- 4 The IDCOR method in their Technical Report
{ e 5 85.3. Then there 's the THIRP _ method technique for human ; i 6 error rate prediction. l 7 SLIM-MAUD, which is Success Likelihood Index 8 Method for Multi-Attribute Utility Decomposition. d
^
9 Then the OAT, Operator Action' Tree, Time - 10 Reliability Curve method. 11 STAHR, which is, if I remember that correctly, 12 . Socio-Technical Assessment of Human Reliability. {}, 13 And there are techniques which are just . J 14 comparison techniques or direct estimates, and there are { l 15 a couple of NUREG's there. 2743 is one. ! l 16 MAPPS, the different MAPPS from the IDCOR l 17 MAPPS, and this is Maintenance Personnel Performance l 18 Simulation. It is a means for addressing primarily changes 19 to maintenance practices. 20 Then there's the Human Cognitive Reliability 21 work and there's SHARP. Now, SHARP is not a human reliability 22 analysis technique itself, but it does provide -- SHARP l 23 means Systematic Human Action Reliability Procedures. . 24 SHARP is, an EPRI document, NP-3583, which l 25 is a means of selecting human -- gives guidance on means Heritage Reporting Corporation E_ _=m ---- _ _ _ _ - -
4
~*'
O O ! V V
-272.
I~ of. selecting human reliability analysis procedures. 2 John,-if you'd go back to the.... Thank ; 3 you". ' r . ! 4 Another topic is the sources of the human i 5 error rate data which is to be used as input. Certainly; 6 one place to start in identifying available data, particul'arly-7 generic data, are past PRA's. 8 The data concerns are both for the generic 9 data -- I shouldn't say " concerns," plura*1., .You have ( 10 consideration of both generic and plant specific data; 11 The concern or focus of the attention in j 12 doing the human reliability analysis at this point'should- I {'; 13 be simply: Are the data valid for application at.that l 14 individual plant? 15- Then there are the considerations of.
.l 16 uncertainties and sensitivities. An important one l's the (
l 17 range of crew performance within an individual plant. d 18 If we could go to the last viewgraph, please. l 19 This is in your handout and it's intended i I 20 to provide some key references for conducting the huTian 21 reliability analyses, and it's an indication of - 'it's j
. 22 intended to be an indication of the state of the art.
- s. ) .
23 It generally reflects the. state of the l I 24 technology about 1985 or maybe a little more recent than 1
~
25 that. It's only coincidental that it's about the time l 1 Heritage Reporting Corporation I
,, - 1 , 1 i
f .- t e (
-.. ; 273 I that.'the-Severe Accident Policy was issued. There are. 1 L
j l ' : c. 2 newer techniques in these references and updates on techniques, f 3 like the Success Likelihood Index method. i 1 4 of course, 'if you reference the guidance l
' which you've already beard extracts f$om it on the IPE i 5 6 . submittal document, 1335, and there is the PRA Procedures 7 Guide, NUREG/CR-2300', which intends to be a compendium i
i- 8 of PRA techniques. ' 9 -
-Then there'c the Probalistic Safety Analysis 10 Procedures Guide, NUREG/CR-2815. You'll see in there listed 11 pretty much what I have given you.on the backup viewgraph.
12 Then there's the PRA~ Review Manual, NUREG/CR- -[
)- 13 3485,.particularly Section 3.6 in that manual. ,
14 Then the Accident Sequence Evaluation Program, 15 Human Reliability Analysis Procedure, NUREG/CR-4772. This 16 - is'the document that describes the human reliability analysis I 17 method used in 1150. i 18 The next one deals with Recovery Actions 19 in PRA for RMIEP. RMIEP means Risk Method Integration 20 and Evaluation Program. That's NUREG/CR-4834. 21 I'd like to just dwell on this one a minute 22 because there was a NUREG cominent on this one, and hopefully 23 I can provi.de some clarification or guidance. 24 The intent for referencing 4834 is because i-25 it describes a method to estimate human error rates for l Heritage Reporting Corporation
. () .-
274 i
'l recovery actions.that'is better than that containe'd in i
2 .IDCOR IPEM's because the method described in'4834 is based 3 upon-observable data rather than expert judgment and, thus, i 4 i's more closely associated with a realistic setting. 5 And the:NUREG-4834.. method treats both human , l 6 errors of omission and those errors of commission involving 7 cn incorrect performance of the tasks required for successful 8 mitigation or prevention of the sequence. 9 It's not.a broad coverage of errors of 10 commission but simply those that are involved in successful 11 completion of specific tasks. 12 The NUREG/CR-4834 method has already provided 13 guidance in the sense that the method was pioneering work (]) 14 and it foreshadowed research being performed by the commercial ! 15 nuclear power industry under EPRI RP-2847, the. Human Reliability _ 16 Experiment that's being ably directed by Dr. Wurlidge. 17 Also NUREG/CR-4834 method was applied to 4 13 selected recovery actions, the report provides guidance , i 19 on the application of the method to other hnman actions 20 that could occur in accident sequences subsequent to those 21 actions that'were identified and were treated in the ) 1 22 development of the method'. I r A :. l 23 And the report explicitly identifies those j 24 steps which are n'eeded to extend the method to these other l 25 applications. i 1 l L Heritage Reporting Corporation l ! gma - _. J
,. ex p . Y L , L) 'V . 275=
]
1 The data'in the report, 4834, might appear 4 1 2 likely to cause a lowering of the total core damage frequency, 3 if it were~ applied generically without regard to the guidance 8 4 given. 1 5 But explicitly, it's the method, not the 6 data, that should be applied in compliance with the intent , q 7 of the IPE process itself. That is, plant specific 'applica - l 8 tions to discover individual plant vulnerabilly. 9 It was recommended that the applicat3on j j 10 . of simulator data be used to calibrate the method to the j i 11 realities at the individual plant. j 12- The next-to-the-last reference is a Topal 1
-l
{} 13 survey of available methods that ampli fies material that's l 14 provided in the reference which is the third bullet, 2815. l 15 This provides a comparison and application 16 of quantitative human reliability analysis methods for
.)
17 RMIEP. i 18 I think I mentioned already the SHARP document. 19 It provides a helpful approach to selecting human reliability 20 analyses. 21 This has been a brief summary of the state 22 of the applied art, which is in a sense rather elementary s .. 23 and has its limitations. { 24 The guidance has been purposefully intro- ! 25 ductory. However, I hope it's helpful to utilities as-Heritage Reporting Corporation ___ _ - -_ --- I u,-
y ,h. a ,
<3
. n. J 1
>; p, e ?E, > <
- () .
a d ' 276 m (;9 7, I' they try to select a human reliability analysis. tool to. ! 2 initiate the IPE. j 3 DR. BECKNER: Tnank'you Frank. ! 4 I'd like to open up the.questiens now for 5 either. Frank or John Chen. John, do you want to come,back i . 6 up to:the table. -' 7 MR. TORRI: My name in Fred Torri from PLG. 8 ' i I am' confused about the external events j i 9 approach that the Staff has recommended nr is specifying. l I 10 On the one hand, external events have been analyzed com-l 11 ! prehensively in some 50 percent of ithe utility-sponsored 12 PRA's, and that to my estimate counts to something on the i 13 order of ten or more. l ) - (s ".."i 14 The vulnerabilities and insights that have a I j 1 15 been derived from these external events analyses, in my 1 16 opinion, are as striking as the ones that were derived j l 17 froni internal, event.s analyses. ]
. j 18 And the utilitier, in my opinion, my personal j i
19 opinion, have hal.en those vulnera' ailit'ie.s as serious as I l 20 those derived from the internal esrents analyses.
)
1 21 On the other hand, the Statf's recommendation 1 22 to defer these analyses and the presentation this morning, j 23 which did not, at least as far as I could catch, reference i i 24 the fact that a number of these analyces have been done, j 25 leads me to wonder what the basis ic for the Staff's decision i Heritage Reporting Corporation ]
==== ,
4 p e
; 7j E: ,
. 4 i
- e. ..
y w Q ,
.277 i I not to! proceed with external events. ( c.
'2 ' ' .I'm not trying;to convince'you that you
;5 n 3 should: change.your opinion, even though I believe'we are N-
) 4 ' ready for'it. -!
! \
a
-5 What I do'like to' understand is what - ' !
, 'f L
( '.( 6 to ask'for clarification of what.the basis for your deferment. l l , 7 Particularly, I would like to know whether you perceive 8 serious shortcomings with the methodology that'has been 1 k 9' applied in the past, which we would like to know because I 10 we nse(1 to get ready l for whatever -- correct whatever short -
!! ' comings you know cf; or'whether the reasons are different.
12 'Is,there insufficient stafi experi'ence with i r 13 external events analyses an'd reviews of PRA's that have {o. i ,
)
14 Jecluded external events? Is there insufficient industry i
.: \
15 hases for performing these' analyses and that may be the 16 reason for deferment. l'1 I'm really just trying to get insight as 18 to whether we'need to take some action with respect to
'19 methodologies we have used in the past or whether there 1 20 are other reasonis for the deferment. I 21 DR. BECKNER: Thar.k you. If you'd let me l
u ,
'22 try to address it. We gave three reasons for deferral j L. q
.23 [ and I'll try to expand upon those three reasons. j 1
, 24 First of all, is to decide just what external l l 25 events need to be considered. There are certainly some
)
Heritage Reporting Corporation 1 E __2_ _ _. - _ sacoa )
$i 41gs i., 'a r l cy , ., O?'. k., c i? ,
+ .,;
' O.
1g e 2 ,. 278 IL -external events that everyone would say we<need to.eddress-
.I t 2 them, based on the PRA's that you talksd about.. Bdt there's 9
- , Lf
'3 many others, alsa,.all the-way down to meteorites.and. things !
4 like'that. Do we haveito consider them? .' G So that's the first reason. :) 6 The second reason is to davelop the' methods. ,
\
7 There are a-lot of methods out there. There h' ave been 8 seisuic PRA's, fire PRA's and so forth; but there have 9 been questions. Can we develop more simplified methods? 10 The third reason, and I think thah the ~ area 11 of the seismic is the best example, is that there are a 12 number of ongoing seismic programs at NRC and also in industry . ; I 2 (~.u'; 13 Th[swouldbeonemoreseismicprogram,potentially,and' 14 so we wanted to go ahead and make cure that whatever we l 15 did was coordinated. ! 16 We weren't asking people to do different 17 things or asking them to do the same thing twice.- { 18 So at this point we're nou saying that we're -
- i .
19 not going to do external events.- It says that we're going
. (
20 , to delay it uncil we can come up w!.th firmer plans and ! 24 we are in that process right now. 21 MR. TORRI: lat I to understand from that a.aj $ 23 < that you really have not identified any serious shortcomings 24 in methods that have been used in the past? You're just l 25 saying there is enough going on to defer it to see what lieritage Reporting Corporation j _ _ - _ _ - - - - - - __ ___ - um j
. . p< -
, , _ , _ _ _ _ _ _}
\, i1 O <
O' 279 1 J
. e
'.f' .
'I elseJean-be gained from the ongoing programs?
I p; , ,
< 1 l
2 MR. CHEN: Fred, I think wha't you are thinking i
" i 3= J is'right. -We do not see any.currcut methodology as far
]
-r 4- ashRA'sare'concernedasdeficient.
) 1 1
5 However, we are saying we are_trying to 6 identify'certain shortcuts you can utilize and use that y 7 availabla methodology'to carry out your. work. c s 8 We are really trying to say, li.he I point ; i 9 out in my prescatation, if you read NUREG/CR-2300, there's 10 about tWo pages lictiry of all exEernal events. 11 We are now trying_to minimize, because some i 12 7 of those things are not'significant for the site and we (j 13 don't.want utilities to waste their resources to deal with 14 those extertial events. 4 15 So in all the programs that you see on this 10 -list, this gives you three reasons why we want to defer
'i 17 the external events. l t
h 18 Primarily, we are identifying what's'needed : 19 to be done and, also, we want to know some of those things l
' 20 which we believe are current criteria. j
'Il To give you.an-example, for wind, we are 22 designing -- one is 10-7 of the wind. We feel that's )
23 sufficient enough and there's no severe accident would ; 24 be taking place. , . 25 So for those sites which design against [ .c Heritage Reporting Corporation _._ _ _ _ - - _ _ - - . - _ _ - - - .cm m
(y ,
< s' '
L,) l 280 k
~1 the wind for current NRC criterf a, we don't seu there- is 2' a need for them to examine.
n
-3 But on the other hand, we also identified ,
N
- there are certain sites which.are not designed against-I 1 5 ,
current cr5teria. They may give you some kind of ris,k j l I 6 contributors there. q 7 So we are doing some kind of screening cr'iteria , 8 tryirC '.o develop some kind of scret:ning criteria ao you 9 will be able to utilize thes'e thinga and reduce the resources 10 needed.for examining these external events. 11 MR. TCRRI: It just'seems to me that in 12 the Level 2 area we have exactly the same kind of situation, 13 where there are still lots of uncertainties in sor.<e people's (. } -
- 14. mind about phenomena, which ones should be included. Yet IS y,e are proceeding with Level 2 enalyses at this time-and 16 i don't really ese a fundamental differer.ce between that 17 situation and'the external events eituation.
i 13 DR. BECKNER: I think ws agree with you. 19 , I'n fact, in our review we are making use of a lot of these 20 external -- in fact, all that we know of, all the external I 21 event PRA's and we are making use of them to get insight
,. s 22 from them.
G' 23 4 MR. CHEN: Let me add a few things on that 24 line. Regard to the back-end analysis related to external 26 events, we currently have a program to try to examine what Heritage Reporting Corpo ation
. pana
'G \) '
281 ) a I l' advantage can we take to utilize the internal events analysis. 2 We want to identify whether there is anything 3 which as a result of internal event, the internal event 4 result-will not be able to utilize. S In other words, we want to identify whether , 6 there is any situation -- for instance, the early containment 7 bypass or containment fail to isolate. These kind of scenarios 8 can be introduced in external event arena, nnd then we 9 can pinpoint those things for you to utilized. 10 But other than that, we feel we would like li to take advantage to the maximum extent possible, use internal l 12 event analysis. l () 13 1;R . TORRI: 'Thank you. l 14 MR. ECLLEN: Steve Follen from Yankee Atomic. l l 15 I guess a comment and a question. Given 16 that we are going to delay external events -- and I think 17 someone made some other comment yesterday. It's not clear 18 to me -- I guess it it clear to me. It just seems more 19 efficient that we address internal flooding with the rest 20 of the externals rather than now. 21 Similar techniques in spatial events and 22 lthingslikethat, and especially in view of the fact that,
.,j'#
23 as Mr. Chen pointed out, some of your external event pr6 grams 24 include things like maximum external floods, maximum 25 precipitation, inadvertent suppress!.on systems, which are He;!toge Reporting Corporation
***** . ]
n . /l Q v i 282-I all, at a' minimum, r' elated topics. ,; 2 It1might seem appropriate to deley internal 4 3- flood with the rest of that. Just a comfaent, I guesa.
' d DR. SECKNER: I'm going to respond. We've t
5 gotten that comment and:I think we're going to have to 6 go back and consider it, j 7 MR. FOLLEN: A question, again related, 8 I think, on A-45. It was listed on the external events 9 programs and per our Generic Letter and related guidance, 10 suggesting that A-45 is enveloped within the IPE to be 11 resolved, and at the same time, further programs going 12 on. , 13 I'm just confused about that, what you are ] ({)o i ' 14 . expecting now or later. 15 DR. BECKNER: A-45 has'been officially resolved , 16 I believe that's the case. l 17 MR. FOLLEN: By incorporation into IPE? j i 18 DR. BECKNER: I think Brian wants to say 19 something. 20 MR. SHERON: The approach was, is that when 21 we looked at the existing programs, the IPE programs, we
,7 , 22 felt that the IPE would, when it was done properly, would V
23 have done everything that we would have really gone forward 24 with on A-45. I 1 25 What we learned from A-45 is that if there Heritage Reporting Corporation
<= -
7
- r .t -
>g
.t t
r
.c 1
'g bc ,
283
~
4'
,,y db; ,
1 t
! 'are any vulnerabilitiesLAn the decay, heat removal system, 1
2 'they weta most J.ikely plant specific and there was no one-3- universal fix to the plant thkt was cost effective. v, t >
't' 4 What we said is that if, . for examplc. the 1
5 Commission had not cecided to go forward utth IPE Program, s I 6 we most likely would have recommended to the'ICommission 7 that each plant to a vulnerability search of theli: decay { f 8 heat removal system. ~
< 9 Because the IPE was 2.n place and woold l I
O accomplish this,-we se.id there was no need to do this; 1 4 1 11 CDerefore, we subsumed'the A-45 issue into the IPE Program. ' L 12 . Whatever you do for external events, whatever
^
13, s = t;b'? itaff Eltimately comes out with as guj. dance would in
- f. .e'.
4 14 I fact satisfy that aspect *of A-45 as well. 15 As far as you're concerned, A-45 is solved. , a6 You don't need to submit it, or address it as a ppecifib.: 17 7%uscic og U5I . !
]
18 The only thing we've asked is that you'look 19 at it specifically when you do your IPE. , 20 MR. FOLLEM: Okay. In looking at the internal 21 I events, sort of as part of that topic now and external 22 whenever tnat- comes up. I v ,' 23 DR. SPEIE: Yas. 14 MP. FOLLEN: Thank you. 25 MR. SHERON: I w% ld just point out on the Heritage Reporting Corporation . i -- . _
L# ' 's , c < 284
'o . ..
I external events, there's nothing wrong with1theLmethodology 1 tha't exists today, and if'the' utility wants to come in j
. i 3 and do an external evente PRA using state-of-the-art methods, I a
l
- 4 that's fine. L.i i
1 5 The whole objective was that we fGlt that , j
\
6 there were a lot of external events .that'didn,'t need to l 7' .be addressed and we could save a lot of money for'you all',; l l 8 und time, and let ycu put your resources to better effort, ) l 9 if we could better define exactly what needed to be addressed ) to ~in external evedts. ! l 1 11 That's the whole reason it's deferred. It's
]
I
- f. 12 not like tnero's no methodology or anything. Your current 1
(')- 13 'PIU4 methodology is acceptable, but it's expensive and it's 3 1 14 long. It~ takes a lot of time and we're just trying to i 1 [ 15 . find easier ways for you all to do it. ) I 16 MR, CHAPMAN: dim Chapman from Yankee. j d I 17 1 wanted to thank you for that comment. j l 18 We believe that the external events should be delayed, 19 unlers there's an issue that elevates the use of DRA to )
- 1 1
20 address the external events. l 4 I 21 At Yankee, we have done extensive external l 22 event analysis on Seabrook and we have also on Yankee Rowe,
.e \
23 but at the same time we don't want to devote resources 24 right now to going after external events. 25 We want to get internal events out of the l Heritage Reporting Corporation na) **aa _ t
- ~"'
m _7 : . a s, , ,
, ,.c 1 F <. '
' jl i
{ J yl: 1-A.m ' i
, 285:
3 y,y, p 2-
. \
?lat includes internal flooding. We think 1 I
-3 'that should be' delayed. g 4
4 Thank you.. m , ,e , 5 Mk. LEVINSON: Stanley Levinsdn'froml Babcock l o 3 j
.. )
. 6 & Wilcox. J l
'? -
I have a question concerning the HRA j 8 ' presentation.- You had a list of quito'a few methods up 9 there. .Are we to imply that a?,1 those methods are acceptable ' 1 4 { 10 fo: use in the IPE? [ 11 MR. COFFMAN: Ee haven't' established the 12 acceptance criteria on the methods. As Jr. Houston mentioned i
.,....[m )- 13. I yesterday, that aspect of the guidance-isn't mature enough i
14 yet for us to say what are the acceptance criteria or. the ! t 4
'15 . methods.
14 The best guidant,e available will be throught j 17 what has come out of the review of past 12A's that the-
)
18 Staff has reviewed. L ~! 19 MR. LEVINSON: Will there be ultimate guidance , j 20 in the final form of 1335? 21 DR. SPEIS: We do not plan to have guidance .
. 22 about human factors. I tilink Frank went through describing 23 . the state of technology and it's up to you to use your 24 judgment of what makes sense to use.
25 I don't think we can -- We would probably
. Heritage Reporting Corporation l raca 1
' ,4 ' ' E <
n O .. L k)' O 'l ( , .i pl , 286 ;l
- s. .
4, J, I ':hace'to work the next five years to come up with guidance , 2' 'that.we all agree is right. - H 3 MR. LEVINSON: Thank you. ) .':
'lo * ' ' ,
4 DR. SPEIS: PRA's have been done in the i 3 past. Some of them have used commonsense, good methods, .
]
6 good approaches; some of them maybe not'as robust. 7 The technology is evolving and Franh'more 8 or less gave you the status of that technology and you 9 guys shouldn't ignore this area. It's an important area, l 10 human factors, procedures, training. 11 They have to be integrated with sequences, I 12 with timing, and I think you know how to do it better than j ( .,_ 13 We do. i 14 MR. LEVINSON: I don't think any utility . 1$ intends to ignore this area. 16 MR. COFFMAN: I think it would be helpful 17 if tie keep the objectiva in mind. The objective is not j j 18 to establish the process of doing a human reliability analysis , 19 but to make the plant safer and it is thz- A ee human
]
20 reliabilittf analysis that you, by whatever variation of 21' methods is taken, that you will be identifying those things . 22 which will constit:ute an accident management program. l
%I l
.M MR. HANNAMAN: I'm Harve Hannaman from Wiscon.si a '
24 Elsctric.
)
)
25 Our Point Beach plant was selected as Heritage Reponing Corporation l O msm
v ,
-o -
pmg ;c , NF < 287 I I one of the case studies for the A-45 studies that Sandia k-L, 2 did, and EPRI sponsore'd a requantification of that limited l 3 scope PRA that was done on' Point Beach.
\
~
4 In discussions with the Staff on those.two 5 studies, there wore a number of differences of opinion, 6 a number of areas of the analysis, and the human recovery, 7 human reliability analysis was one area where there were 8 soveral differences. l 9 I'm wondering if there's some mechanism 10 for us to submit our initial results of our IPE studies 11 for review by the Staff before we spend a lot of time identify <- l 12 ing vulnerabilities and developing.the proposed fixes for (' ...], , 1.3 these vulnerabilities so that we can negotiate and discuss i 14 any differences of opinion about'how we performed our IPE's 15 and resolve any differences. 16 Could there he a two-step review process 17 or some way tio submit some initial results and get those , 18 reviewed before we spend a lot of resources on trying,to 19 resolve the vulnerabilities? 20 MR. HOUSTON: Let me respond to that, if
. 21 I may. We are not anxious to go through a two-step review 22 process on IPC results.
g
, 23 l The information that was developed related 24 to the A-45 program, the EPRI effort on A-45 is a document 25 that is in the Public Document' Room, along with the evaluation
)
I Heritagw hporting Corporation __-- ._ .- _ _ _ - eacxo .
) i O
l O 288 4 1
~
1 made by the.S'taff on that.- l 2 . If I recall' correctly, one of the comments 3 that'we got in.from'NUMARC is we did not list that as a j 1 I 4 ' reference, and we certainly should consider doing that. 5 To a very large extent, my understanding 6 is that that was a very well done PRA. It may be that .) 1 7 Point Beach may choose to use that as a significant basis 8 for its. IPE and 1.t may come in as part' of your final report. . I 9 But right now I woald have to say that we're .li 10 not disposed, I think, at this time to do an interim report 11 review, as it were, and looking forward then some year 12 later, year and a half later, wnatever it might be, to [ 13 doing a final review. (]F 14 I think.you should submit your best effort 15 as a' final IPE document. 16 .MR. SHERON: Let me just add.to that, if 17 I could. The difference, I think, came out to about a 18 factor of 31 at risk, frequency between the EPRI work and 19 the Staff work. 20 My recollection is there was a big meeting 21 between the Staff and EPRI and Wisconsin Electric and my c 22 people -- I wasn't there, but they told me that we agreed v...) 23 to accept some of your assumptions, there were some we ) 24 'didn't agree to, and as usu'al it came down to cplit the 25 dif ference between the two. Herifuge Reporting Corporation
- ____] .
tmum .
,, ', e i
/
(a) . 289 I I would say that now.that you have this 1 s 2 unique position of having the Staff having looked at that, 3 I would personally say you should consider would you do I 4 .anything different to your plant based on those two results. 5 In other words, if you followed the Staff's 6 approach, would you make any different decisions on what
]
7 youwould #ix or improve in your plant, as opposed to what
)
8 you would do if you followed your results. 1
)
9 MR. HANNAMAN: I guess the focus of my question I 10 was mee.nt to concentrate on how the review is going to 11 be done. If we submit our IPE results, how is that review ]
. i 12 going to be done and how is that going to be documented !
I f' 13 back to the utility? l 1 1 14 Are we going to receive a Safety Evaluation l 15 Report that then may question certain aspects of our analyses,, ! l 16 and how do we eventually resolve those things? l l 17 DR. BECKNER: I think Wayne addressed that i 18 a little bit. So why don't you just go ahead and repeat 19 what you said yesterday. 20 MR. HOUSTON: It is, of course, possible 21 - that in the revie'w process the Statf may have nome questions. 22 It's possible that it would be directed to the utility i 23 to clarify certain points, perhaps some assumptions made, 24 the sources of information used in the review process itself. 25 ' As I indicated yesterday, (>ur expectation l Heritage Reporting Corporation L_________ _ _ _ _ _ _ acoa
.1 eg -
(Q -tj >
~
290 l would be that for'each submittal of IPE results, preferably 2' in the format that we've requested,'that there would be ! l 3 a written evaluation. The conclusion of it would be a l
. I % 4 written evaluation of the IPE ' submittal, and the measure l i
5 of the evaluation.is how well the Staff thinks that the ! I 6 stated objectives of the IPE process are reflected in that i 7 submittal. ! 8 So yes, it would be available to the. utility. j y 9 It will be part of the docket record. ! 10 MR. HANNAMAN: One other question onLthe - 11 reviews. Most of the presentations have been given by 12 research folks. (]) 13 Does that mean that the reviews will-be ! 14 conducted primarily in the research area or is this going 15 to be done.by NRR people who aren't here'and wo don't nave 16 the benefit of their opinion on how they.are' going to do 17 the review? ! 18 MR. HOUSTON: I think I can respond to that, 19 also. While it is true that the Office of Research has 20 the word "Research" in its name, it also has a number of 21 personnel in the office who deal with issues which, strictly
.f-..,
22 speaking, are not research, such as resolution of issues. (;/ 23 A substantial portion of'us have had a good 24 deal of experience being in the Office of Nuclear Reactor l 25 Regulation, also. Heritage Reporting Corporation l E._ -- - - - -wm -- - - - a
U-Q G l 291
~
1 Bht to directly answer your. question, at 2 the present time the review effort will be a ' combined effort 3' between NRR Staff and RES Staf f, although the submittals
"# 4 will be made primarily to the NRR Staff and the view would ]
1
-5 be'that the NRR Staff really has the lead position in this, J I
6 but we expect to assist them and participate. ! 4 7 MR. HANNAMAN: Thank you. 8 MR. RUBEN: Let me add 6 also -- Mark Ruben, I 9 NRR, is here and both the systems people, Mr. Barrett from 10 the Risk Applications Group will be involved. j 11 DR. BECKNER: Could everyone here that? 12 The point was made that we do.have significant (}}; 13 representation from NRR and we have had all week. Mr. 14 Barrett is at the table and we've had Mr. Thadani here 15 earlier in the week. 16 MR. DUBE: Don Dube, Northeast Utilities. 17 Yesterday it was mentioned in the documentation 18 , portion that we were to list sequences where human error 19 is less than one in ten. 20 I can assure you that there are hundreds, 21 if not thousands, of.such human errors that are less than 22 one in ten, and when we do a PRA we implicitly exclude w J 23 certain human errors that we know are very unlikely or
)
24 will not be in the dominant accident sequences, for example. 25 So I guess I'm looking for a little more l Heritage Reporting Corporation - _ _ _ _ _ _ _ _ _ - ._ encaa
A bS U 292- I l' clarification in terms of'what exactly is expected of us 2 in that regard. 3 MR. COFFMAN: The one in ten limit I wouldn't 4' take as a target or a threshold to really focus around,-
. 5 but certainly I think we all'need to recognize that PRA's 6 are summary accounting and d'o not account for every single 7 way in which a component can fail or system can be unavailable.
8 They are really summary accountings, and that summary _ l 9 accounting that applies to the hardware equally applies 10 to the human actions. i 11 So it's more to be looked at in the summary i 12 of the human actions. 13 The screening of human actions by putting '{. ) . 1, 14 in high' failure rates for the human action on an initial 15 evaluation and what are the leading sequences is a process 16 that should cull out many of the unimportant ones, so that 17 the. rest of the analysis can focus on more important failures. 18 Maybe I should stop here and ask am I 19 addressing what you -- 20 MR. DUBE: I can suggest that what we've 21 done is identify the major cognitive operator actions and
,. 22 there may be 20 or 25 or 30, what have you, and then perform l
- i. '
l 23 a sensitivity study, or change the human error by a factor 24 of ten or some kind of importance calculation. 25 Would that be an acceptable approach? Heritage Reporting Corporation _ _ _ _ _ _ -_ _ _ caaxa -
.y .
Y 293 I' MR. COFFMAN: I'm not sure I could agree H
. l 2 to just 20 actions, but certainly, screening and then 3 sensitivity is, I think, a very sound approach. But I 4 would expect something more than just 20 actions. ]
5 MR. DUBE: Wall, I'm talking about major 6 cognitive actions, not the procedural type, errors of omission 7 and commission. ;l 8 Thank you. 9 DR. BECKNER: Do we have any'other questions? 10 MR. WIDEMAN: Steve Wideman with Wolfe Creek l' il Nuclear Operating Corporation. 17 Recently, the NRC has asked several utilities 13 to participate in the development of a risk-based inspection
?[)
14 guide by providing system descriptions, emergency procedures, 15 system procedures. 16 In its request, the NRC indicated that its 17 risk-based inspection guide wz.s not associated with the
.38. IPE; however, the resulta of the IPE would be factored,
- i 19 .into'the inspection guide.
20 The timing of this request and the apparent 21 development of this risk-based inspection guide would appear 22 to be definitely tied to the IPE. 23 I have two questions: What is the intent 24 of the risk-based inspection guide and are the NRC regional 25 offices going to use this inspection guide to follow the i Heritage Reporting Corporation _ _ _ _ - - - - - - - - - - - - - - - - acra
- l. a .
l t: . , . O O-294 l j. [
- k. l-p I development and. implementation of the IPE?
~
1: 2 MR. BARRETT: 'I would like to respond to; l
~
3 that. j e' l 4 First 6f all, the intent of the risk-based I ! 5 in:spection guide is to give the re'ident inspectors some '] l l 6 guidance concerning the most risk-significant aspects of 7 their plants and to try to make that risk-significant i 8 information as plant specific as possible. 9 The request that you're referring to is 10 an entirely voluntary request. You may or may not: decide 11 to participate. i f2 If you do participate, our understanding i i {} 13 is that basically what you would supply would be basically 14 ,some Xeroxing of some existing information.
; 15 The development of the risk-based inspection 16 guide is not in any way tied to the severe accident resolution 17 process. It's not in any way tied to the IPE Program.
i 18 The so-called rigs are purely for the use i 19 ' of the regional personnel so that they can conduct their. 20 inspections in ways that give us the most bang for the 21 buck in terms of looking at the things that are most risk
, 22 significant.
23 The reference to the IPE and the fact that 24 perhaps IPE-generated information.would be used in the 25 future to make these risk-based inspection guides more Heritage Reporting Corporation .--- _ _ _ _ _ _ _ - _ _ .___ _ _ _ckcaa
J o' O - 1 295 ) i i plant specific simply refers to the' fact that we would 2 want'to use the very best possible information available 3 as it becomes available.
~
' 4 The answer to your question is this is a
'5 voluntary request to you. You may or may not choose to a
6 participate, but it really should not be confused with , 1 7 the IPE process. 8 MR. WIDEMAN: Thank you. 9 MR. HOWARD: Ed Howard, Boston Edison. l 10 I think that's exactly the kind of issue j 11 that caused my concern about coordination and integration. i 12 These things all have to tie together when you're done, I 13 including the inspection programs of the Commission. 14 The success of accident management processes 15 is not going to be solved only by utility action. The ! 16 NRC is on the other end of those tel'ephone lines when we're > 17 in the midst of responding to an emergency and we're all 18 part of that over-all response team and we all have to 19 understand what's going to happen in similar ways and work 20 together on it. 21 I think that area does need more attention.
. 22 DR. BECKNER: Okay. Are there any other 23 questions? There doesn't appear to be.
24 What I would like to do is, first of all, 25 turn the chair over to John Flack and what we'll do is Heritage Reporting Corporation l .. _ _ - - _ - _ - _ _ _ _ _ _ n a~m a
O. O ! 296 j i start-the prepared public comment section and probably 2' get in one.or two this morning. 3 ' MR. FLACK':' We're in that part of the program 4 where we have received some prepared comments. 5 I'd like to introduce our first speaker, ! 6 John Garrick from Pickard, Lowe & Garrick, who wil.1 provide 7 comments on Generic Letter 88-20 and NUREG-1335. 8 Let me add that there will be handouts 9 available at the table. There are soma at the table'now 10 as you walk'in.- We are having others made and they will 11 be available tomorrow morni g. 12 MR. GARRICK: Thank you. { 13 Yesterday I heard a member of the industiy 14 make the comment that in reviewing how industry is doing 15 and what they're doing in the way of technology, it was 16 kind of nice to hear some complimentary things about what 17 they're doing, rather than just being.a critic. 18 I guess I'd like to return the favor. I 19 thi.nk that industry, on the other hand, might well say 20 some positive things about the IPE process, rather than 21 dwelling, unduly perhaps, on the negatives.
, 22 I happen to believe that this represents 'i A.) .
23 a very important. milestone and wha,t I might even call a 24 cultural change in the way in which we think about safety. 25 I think it also represents a change in the . 1 Heritage Reporting Corporation
< =a *= _
j
t . 8
.- j A '
- (,)
V 297
'l .way in which we think about risk management, which is a
. i 2 broader issue, because it brings into the arena the matter 3 of decision-making, the matter of costs and benefits.
4 And finally, I kind of think it even goes 9 5 deeper than that. I think that what we i re really talking
-6 about here is a beginning of a way, a cultural change with 7 respect to how we engineer and' operate technical systems.
~
8 So I think that it's a long time coming 9 and that the principle is extremely beneficial and in the 10 right direction. All we have to do is work.out the details 11 of implementation, and I am frankly optimistic about that.. 12 Okay. What I'd like to do in'the next few {]; 13 minutos is offer some comments on the IPE docueents and, 14 to do a little reflecting on the experience base of doing. 15 PRA's for a group of utilities that represeht approximat'ely 16 one-third of the plants in the U.S. 17 I also would like to say.that some of my i 18 remarks are based on input from those utilities and I guees i 19 in order to make sure that you understand from whence I 20 speak, I will put it this way. 21 Nhen I ma making observations that all of 22 our clients agree with, I'm basing it on the' input from 23 the utilities. 24 When I am making other observations that 2f represent PLG, then I'm representing PLG. Heritage Reporting Corporation .l o
A , j O - u .. 298 1 Finally, whenII'm> disagreeing *with so'me
' ~
I 2-~ < of my colleagues at PLG,.I'm representing John Garrick.
- 3. So free at last, I'll' move ahead.
t
- 4 I will say that while I won't mention all 5 the utilities that are involved here -- there are some dozen 6 of. them 'ths.t we got ' input, and I'm not going to name them i :
7 except to say in view of the fact that we're talking in' L
' 8' Texas, that one of those is Houston Lighting &-Power.
9 I think.that firstoff we want to make a i 10 few general comments on the variods documents.that we have-11 . received'from the NRC, leading to and giving definition ~ ]
. 2 12 to t'he whole individual plant examination process.
h- 33 First and foremost, we agree with the t'ame h 14- that's in those documents'and that was1 pronounced so repestedi r-15 yesterday, that this should involve very Strong utility 16 participation. j 17 When things go wrong, we're not really 18 depending on the consultants. We'ra not really depending .)
)
19 on the NRC to take suitable corrective action, suitable ') 20 recovery action. l 21 'We're depending on the utility and its operators
.,..t 22 for the most part.
- i. >
. 23 - .We also are very much of the 6 pinion that L 24 the ultimate benefi'ts from these investigations, these 25 examinations can be realized with plant sp.ecific. data.
Heritage Reporting Corporation ! in___ Y _ ___ _ _ _ _ _ .cca n a l
~
) ! i l x./
299 1 I was surprised at some of the observations 2 that w3re made yesterday on data. I think that the data 3 handling techniques, the data handling experience is much l 4 further along than was kind of reflected by some of those l 5 observations. 6 On the matter of prescriptive judgmental l I 7 databases, we think that you only depend upon those when l i 8 you have nothing else to go on. We can't think of many : 1 9 exceptions where that would be the case. f I 10 I think the data situation is much more ! 11 solid than we have a tendency sometimes to realize. 12 As has been stated so many times, I would , t 13 also agree that the internal flooding issue, if we're talking ({} 14 about deferring other issues having to do with spatial 15 interactf.on phenomena like the so-called external events, 16 that we have difficulty discriminating out the internal 17 flooding, especially on the basi's of having done a large 18 number of external event analyses and finding internal 19 flooding perhaps to be an. issue, an early issue in only 1 20 one of those PRA's, and on the other hu.d fires and seismic 21 events have been issues in many of them. 22 So it's not a matter of saying that we 23 shouldn't do internal flooding. It's only a matter of 24 questioning why they are singled out. 25 There is a methodology for external events Herifuge Reporting Corporation l cmn.
- n -
l 1 ( ) i . ! ,_ / ! 300 i l 1 available. I think that Fred Torri made the comment that. 1 j 2 I would otherwise make there, the point here being that 3 we're not as bad off there as perhaps we might think. l
4 There is often heard the observation that l 5 we're much further along with respect to internal event 6 analysis than we are external event analysia.
7 My answer to that, it depends upon which 8 ene you're talking about. There are certain internal events, 9 such as vessel ruptures and containment bypass scenarios, 10 for which our technology is probably no better and in some 11 cases less than for the treatment of some of the external i 12 events. 13 So it's really a matter of putting it into ({) 14 context of the risk question. 15 I'm sure that most of you would think that 16 this next item would be in here, the treatment of uncertainties. 17 This is another area where there seems to be a tremendous 18 amount of confusion about what is meant. 19 Th6re is often the thought thrown out that 20 it's either/or. That is to say, either you do detenninistic 21 analysis or probabilistic analysis, and that's not the
,. 22 point at all.
23 The point is that uncertainty analysis is 24 that added dimension of taking the best analysis you can 25 possibly do and going out on a limb and saying what you Heritage fleporting Corporation zanum
C V 301
]
I I can say about how much confidence you have in it. I l 2 In my opinion, th,at is the central issue 3 in risk. The uncertainty is what risk is all about. l l l 4 The other thing that I think led to a lot 5 of confusion about uncertainty is what we mean by it. What 6 we mean by it is not the question of does a pump know when s 7 it's going to fail. A pump never knows when it's going 8 to fail. We're not depending upon that pump to tell us 9 when it's going to fail. 10 We're depending on somebody's state of knowledge f 11 about that pump and its failure characteristics, and if 12 there's one thing we can be quantitative about, it ought {s 13 to be what we don't know. 14 So that's the concept that I think makes 15 it clear that the lesc we know, the more important it is l 16 to quantify our uncertainties, and it's interesting to l 17 observe a kind of opposite view from some of the observers. 18 From a scientific process standpoint, I can't follow that 19 logic. 20 Additional requirements for USI/GSI i 21 resolutions undefined. I think that's a point that's been 22 made several times. 23 I think that most of us would like to know 24 as much as we can about what constitutes resolution and 25 I think that's something we should continue to work on. Heritage Reporting Corporation om m.a
I Let me talk a little bit about the reporting .
. l 21 requirements and get to some specifics.. There is a table 3 of contents that's suggested for an IPE. I say " suggested" l
4 because I hope that's.the context. 5 The other point that we want to make about i 6 that is that in protecting the interests of those of you I I 7 who have made substantial investments in risk assessment, l 8 I would hope that what is really meant there is the summarizing 9 of the information in a form that fits that table of contents, 10 rather than something more deep than that of perhaps 11 suggesting a redo of the analysis. I'm sure it's probably 12 the latter. () 13 There's been a lot of discussion about the i 14 sequence selection criteria and its arbitrariness, the ! 15 functional sequences versus systemic sequences. Y 16 I think this is an area that probably we 17 need to be extremely careful about. I think that one of 18 the things that bothers me a little bit is cut-off criteria, 19 just as one of the things that bothers me a little bit l 20 is the suggestion that we've got to wait a little while 21 to decide which external events are important.
, 22 The question that we're trying to answer 23 is which external events are important for that plant and 24 that is something that's very different from trying to 25 resolve the issue at a generic level, because at some Harltoge Reporting Corporation .
_ ___ _ _ - cacoca .
i O O i G a 303 1 I plants fires and floods are going to be important; in others,
~
2 it's going to be storms and earthquakes. 3 So I don't think we can really approach l c l 4 it that way. I 5 On the sequence selection criteria, enough. 6 has been said about this business of being able to decompose 7 a sequence category into smaller and smaller numbers in. l 8- order to meet criteria. I think that's well understood, j
'9 but I also think that in view of the principle here of to trying to be very plant specific and trying to adopt a 11 notion of individual plant examination, that the sequences 12 ought to be driven, again, more by plant specific charac-(]) S 13 teristics than anything else.
14 I don't see much gained by trying to define 15 functional levels, because they're not relevant at the 16 plant specific level, and what this is all about is trying 17 to be relevant at the plant specific level. 18 What I mean by that, also, is you have to ! I 19 be very careful about what is important with respect to 20 what. In a typical Level 3 PRA, for example, you may be 21 talking about nine or ten different plant damage states.
.. 22 One of the things that was discovered in
,j 23 the early large FRA's is that the ranking is very plant-24 damage-state dependent. In other words, the scenarios
. 25 that are most important for core melt are not necessarily Heritage Reporting Corporation maa )
E.
, j o e- 304 1
l 1 the scenarios that are most'important for off-site con-2 sequences, which is something that we have to continuously 3 keep in our minds. 4 As you impose an arbitrary criteria on Level 1 5 scenarios, you may as a matter.of fact cut off the most, 6 important scenarios in a Level 3 response, which indeed you 7 could very well do because most of the Level 3 scenarios 8 that are important are containment bypass, large earthquakes, 9 et cetera. ' 10 I think if you look at the big PRA's, you'll 11 find several scenarios that at least from the initiator 12 through the Level 1 model, under this criteria would probably {y 13 never find its way.into the importance ranking list, and i 14 in fact they constitute very important scenarios at the 15 consequence level. 16 On the matter of the back end, I think that 17 what we did observe in analyzing the documents is that 18 most of what is said is comparable to established Level 2 i 19 PRA practice, at least in the context of a limited Level 2 20 analysis. l 21 I guess I would make the same observation l l 22 thero, I think, as I make about the external events. I 23 don't think there's quite as much mystery there as sometime.s l 24 we tenEl to suggest, as long as we accept the notion of 25 trying to address uncertainties. Heritage Reporting Corporation _z__--.__ --._ - - _ ca m = 1
( -- - - - - - - - - - - - - - -
, 3 ,
i -
;)- .., .
c l , t i s ,* 305 q :i 1[ If,you try to do i without uncertainty,
.I a ,
'2 then of course it becomes hopeless, because you cannot '
.l i
3 l communicate your confidence or your' state of knowledge > l
!- 4 about what you're talking about.
5 If we don't understand something and yet j
.t 6' it appears to be very important in the conclusions of the 7 Level 2 analysis, we're obligated to do as good a job as 8 we can to quantify that lack of knowledge. {
9 On the matter of accident management, to
-l 10 be sure we agree that you need to address accident management,
!! which'for'the most part here seems to be a response of j 12 operators to accident progression.
I 13 I think this.is one of the powerful outputs
' ha - i 14 that we get from_a Level 2 type of analysis and, clearly, i
15 it should be an integral part of the process and it is ) i 16 part of what-is meant by risk management. 17 I think that one of the things that's j I 18 important is when you stage your analysis, be'it a Level l 19 1, a Level 2, that you do it,with the idea of the subsequent 20 analyses to be performed in due course. i 21 In other words, the key to a clean interface 22 between a Level 1 and a Level 2 analysis is the establishment I s I 23 of a logical set of damage states or bins that permit you > 24 to go directly into the Level 2 at the appropriate time. i 25 The approach to construction and Heritage Reporting Corporation N* *
, O. O .
30. m '
; 4 1 quantification' of containment event trees is, again, I 2 think, much better established than some of the observations ]
> 1 3 have noted. !
j 4 To be-sure, there are phenomena, physics 5 que,stions about which there is great uncertainty and it's 6 veiy important for us in taking our responsibilities as l 1 i 7 risk nianagers .to put forth that fact in some way. I 8 It's been said that there are three. ways 9" to deal with uncertainty. One is to ignore it; the other 10 is to do a kind of waving of the wand and deal with it; il and the third and the one that certainly I advocate.is 12 to treat it like'you would treat any other science, to ('y 13 ' be systematic, to be deliberate, te be visible in displaying-14 your 'c:onfidence ~1n your utate of knowledge. 15 So I think that's all I need to say about 16 that, i 17 Desired attributes. I think that one of I
.18 the things w3're talking about here is being plant specific 19 and so we want to get away .from generalizations that don't 20 give us an opportunity to understand in some level of detail 21 what is going on.
22 So what that means is that the analysis
,-,i s,,
23 must desbribe the initiators, the systems, human action 24 contributing to the total core damage frequency and 1 25 significant release categories. Herilogo Reporting Corporation
.-_ ._-_----__L . _ _ . -
* * * . _h
, e
. . i
,p g i d b 307 1
! Obviously, I think that one of the things
' l 2 'th'at we've heard a lot about here is that why haven't the 3 regulators been more specific and more prescriptive and.
4' what'have you. 5 I guess my thought on that is that what 6 they are doing is suggesting a. yardstick here and that 7 the most logical progression towards dealing with that
' 8 yardstkek is to apply some analyses, do some. work, and 9' let that evidence be the basis for calibration.
to I think for us to press the point too far 11 to be prescriptive, to be cookbookish, would be a serious 12 mistake. It would do two things. It would force.us into
'(]; 13 , patterns where we might miss some' thing very important, 14 possibly; and secondly, it'might do the thing that many 15 of the old-timers have worried about for a good long time 9
16 as the probability era matures, it will make this process )- t 17 too mechanical and we'll stop thinking. When we stop thinking 18 about the business, we're in deep trouble. i 19 Okay, let's go to the next one. 20 We've already talked about, so far as the 1 21- proposed criteria are concerned, that there ought to be l 22 some definition of functional and systemic sequences. I l
"(. - , ! , , 23 think here we want to be very careful about how far we i l
24 push the issue. ; I j ' L 15 We need some guidance perhaps on if we Heritage Reporting Corporation . i__ _-- -J- N=
- s , a , .
308 I are' going to be directed on functional sequences.e how to ; 2 make the transformation from those functional. sequences 3 to more systemic. sequences.- 4 As I've said, important contributors could 5 go unreported if individual sequence frequencies remain 6 below the arbitrary threshold, the point again being that 7 sequences and their importance are very damage-state 8 dependenz. .
'9 Suggested reported criteria. The suggest~ ion 10 'is that we report the highest frequency scene.rios 11 constituting a significant fraction of the total core damage.
12 In other words, we'd like to be in a position
/( },, 13 that when we answer the quest: ion about core damage, we've 14 got a pretty good list of scenarios that contribute the 15 most of that core damage frequency.
16 Similarly, we'd like to be in a position 17 that when we've*got a source term, that we can rank the I 18 contributors to that source term in such a way that they 19 are with respect to the source term. l l 20 And finally, when we rank the contributors' , l 21 to off-site consequence, we indeed have the proper ranking 22 there. 23 So the approach here that works the best,. 24 we believe, is to think in terms of reporting the highest 25 frequency scenarios, rather than necessarily thinking about L
> 1 Huritage Reporting Corporation j
* * "
- A
G 309 , 4
' I an arbitrary cutoff.
2 I think that's enough of that. 3 Now, having sort of addressed some of the 4 comments and highlight 6d some of the obs4 vations that i 5 we collectively have received. I guess I'd like to just 6 for a couple of minutes now highlight some of the lessons l 7 learned, and I have an ulterior motive in this and it'9111 , l t 8 be obvious. 9 l First of all, among the lessons we have 1 10 learned in doing and applying the PRA process is the relative 11 role of small and large LOCA's. Before WASH-1400 there i 12 was a total preoccupation with the large LOCA and it's (}l - 13 a good thing to some extent that there was, because on 14 investigation it turned out that that was not an important I 15 contributor to risk, that there were other things that 16 happened such as transients and small LOCA's that were 17 more important. j I 18 Secondly, the quantification of the role J 19 of containment and the discrimination of core damage and j 20 the containment failure contributors. "Until Indian Point 21 there was not much that we could say quantitatively about , i 22 the role of containment. t;. ) 23 In my opinion, that's a very.important step 24 forward. It came directly from a large-scope analysis. 1 25 Thirdly, outliers are not necessarily revealed
. Heritage Reporting Corporation 3_ m uum
O Q 310 ; i L .l. by limited scope assessments. 2 We certainly would not have 'found the outlier, 3 if I can call it that, in Indian Point oi* the outlier at
> 4 Midland'had we implemented the limited scope PRA.. A point L l
.5 to think about. )
i
,6 offsite consequences were generally nuch 7 lower than expected. We wouldn't have known that had there 4 d
1 J 8 not been comprehensive investigations. That'e one of the s l 4 9 reasons, perhap.s, there's been a little bit of a back off . I' , 10 - of emphasis on it, but it's very reassuring to see it done i d 1: and to see what in fact is the rick, i 12 The role of support systems turned out to l h~ 13 be much greater than had been anticipated. Most of the 14 early analyses were very much mainline system oriented, ) { 15 which was not where most of the risk was coming from, which 1 16 was where most of the regulatory e.mphasis was on, and it I 17 turns out that because the support systems have much greater i 0 18 variability, even between the same plants,.from plant to ) 19 plant, than do the mainline systems, that that'u where I 20 a lot of the action was, emergency planning zones are much 21 better defined in terms of their technical basis.
., 22 I'm not saying here that the ten-mile radius 3 . % ,/
23 is better defined. What I'm really saying here is there 24 have been some analyses done now that does provide the , I 25- basis for technical defense of evacuation zones on a plant Heritage Reporting Corporation j ___z______ _ _ _ _ _ _ _ _ . _ ('*"" t l
O - O 311 { 1 L I specific basis', and there is much to be gained, in my opinion, 2 by further exposure of th ? wa kinds of analyses. i 3 Data handling practices that a're'more 4 defensible. Yesterday we heard the observation that.they i 3 had'had a lot of experience but no failure, an implication 4 6 being that therefore that was no data. 7 Of course, that'C'nonsensa, That's very 4 8- valuab'le data, very impc2 tant date. It's the very best . 1 y 9 data you could have, is to have a hundred thoucand years to of ' operating experience and no failures. 11 We really have made a lot of progress on. 12 data, on how to combine generic data uith plant specific 13 data, and that e.Nperience and technology ought to be passed ({} 14- by everybody on that issue. 75 Data is something that's inherently an issue 16 in the minds of peopl'e that are exposed to this kind of ; i 17 thinking. It's something that, you know, I always say that 18 by the year 2000 t'ho two things we'll be complaining about 19 are no data and nobody.knows what we mean by reliability. 20 - Now, recognizing that that's the case, let's 1
)
. 21 go ahead and do the best we can. The best we can, I thir.k i 22 you'll find, is pretty damn good.
%) .
23 I've talked enough about the next two items. 24 So let me just kind of wrap up by saying 25 that the drivers of risk are dependencies and human action, Heritage Reporting Corporation
ol o
- 3. lF n
. o' 312 j i
, 8
~
q- ,
'l especially support systems.
v 1 lcy 3 2 - ' The great thing about PRA is'that it asks ] 3 a very fundamental question, and that is what is a risk. i 4 It does not ask if you are in compliance. It.does not t 5 -ask if that's a safety-grade piece.of equipment or a seismic ., 6 category piece of equipment.- ! k 7 It asks a'more fundamental question of what { l
& is a risk.
' I' 9' Given that question, we've madelin the-indrstry 10 and in the PRA community some very important discoveries. )
11 One of the discoveries we made was the mainline systems I 12 of the plants are pretty darn well designed, an;l that where ' j (j) 13 we're finding most of the action is with respect to the 14 way in Which support systems, mostly electric power-and l 15 service water and the way.in which human actions-are treated 16 in the model. 17 Risk management is much more than anything i 38 else an understanding of the plant, and I think that here 19 I'll.use the podium to say that we must be careful about 20 becoming too preoccupied with the use of the modCl or the l Il computer, tha.t if you can't send the guy to the blackboard
- 22 and h',ve him tell you a little bit about how the plant 4
U works under &c accident condition, a little bit about !
~
24 thermohydraulics, a little bit about the physics of what's 25 goihg on, I would have no confidence in the model or q Heritage Reporting Corporation 4 l
- _ - - - _ _ _ _ = - - _ _ _ - M '*** \
~
. , ']
(). h
- 313 i
, 1
,ihc ' 1 the risk anal'yst. . l l 2 i So understanding plant is really the first 3- and foremost requirement. And making good decisions.. 4 So t'nat, in my opinion, is in a nutshell 5 what it!s.all about now. l 6 My conclusion. I think we're making progress. 1 h 7 I think'that the IPE is long coming. I would' caution tha't i l l 8 now having taken that giant step, we don't take backwards j 9 steps of trying to make it more generic than the spirit- , j 10 of the individual plant examination idea is' intended to I 11 be. 1 i II I would allow what's important and not importan-: ) i t 33 to be driven'by individual plant investigations. (]' ). 14 As far as the other items, we will again 1 15 say that we see no technical basis for. internal flooding 1 16 to be separated at this time from the consideration of 17 other external events. At least there is nothing that I l 18 has come on the' horizon to suggest that that's particularly ! 19 wise, as an event, to isolate. 20 I'll say that the proper characterization 21 of uncertainty is ke'y to understanding the results. If l 1 s 22 we do not deal with uncertainty, then I agree, it is almost 'l, i ,f l 23 impossible to address something so complicated as the ;
)
l ,
.24 progression of core event in the Level'2 arena. l l
l 25 But if we do recognize that there's a lot ! L .l ; i l britage Reporting Corporation !
- i -_-_ _ . . _ _ _
== I
O: O 314 1
-1 we.do know and that what we don't know we're going to be 2 open and admit it, I think it's amazini how much order 3 and how much structure to the problems we can develop.
4 Finally, as a kind of over-all comment, 5 I think we need to be very careful about the direction 6 5 that we're going with respect to sequences. I'm very 7 encouraged by the fact that we are indeed moving in t;he 8 direction of a scenario-based approach to risk.
9 That's very key because it keeps the human 10 actions, it keeps the hardware in perspective in the thought 11 process about riek. management; whereas if we deal strictly 12 at the hardware 3evel, we sometimes lose the context of { 13 the hardware with respect to the applicable bcundary 14 conditions,. 15 In some scentrios this depends on a cooling 16 system and auxiliary power supply; in others it does not. 17 At the scenario level we never lose track of that. 18 The emphasis on sequence, in my opinion, p 19 is very constructive in that regard, but I caution that 20 we not back off too far with respect to trying to omr-9 21 generalize. the scenarios.
,., 22 Thank you.
23 MR. FLACK: Thank you very much. 14 This would be a good time to take lunch. 25 We'll do that now and reconvene at 1:30. Heritage Reporting CorporoAon _- __ ____- __ - _ __ _ _ _ _ _JW~
O-O 1 315 1 i l AFTERNOON SESSION, j 2 MR. FLACK: We are in that part of the program 3 again where there have been prepared comments. 4 The next speaker I would like to introduce 5 l's James Chapman from Yankee Atomic. His. presentation 6 is entitled Yankee Atomic Severe Accident Response. l 7 MR. CHAPMAN: Good afternoon. I'm Jim Chapman 8 from Yankee Atomic Electric Company, and I want to thank 9 .the NRC for allowing me a few minutes. I think it's been io a really good two days so far. ;
!! Out topic is Yankee Severe Accident Policy 12 Reso!.ution Response, focusing on IPE, not on severe' accident j
(]) 13 manageuent or containment performance. 14 The agenda. I want to.give you a quick 35 background on what Yankee is and who I.am and talk about 16 the Yankee program and give come comment on. specific IPE ! 17 issues affecting, quote, unqu6te, closure, and offer a 18 ' rathei flexible template for IPE closure. -) 19 I'm not going to get"into detail in commenting o 20 on the IPE Generic Letter and other supportirg documentation. 21 We intend to provide further comments formally both to
. 22 JJRC and NUMARC. ,,, e 23 First, Yankee has been around for quf.te 1
24 a while. We were founded in 1954 to design, construct 25 and operate the firist commercial nuclear powerplant in l Her}Nge Reporting Corporation e- -_ }
x v '- ' c , , , e
. , ,' cm
,y l'O '
~
< ,]
9
/ ' .c : , .
~.g s 4 316 ;
, . i
'l New-England.-
(' '2 We have operated Rowe since 1960. We expect,, t 4 3 ,to operate.it until at least the year '2020. . We expect-
- - 4 to extend the current license.
5 . We also support Maine Yankee, V3rmont Yankee, 6 Seabrook and other plants. 7 Personally, I managa a group called Safety
.l 8 Assessment ~It has 15 people involved and it' involves 9 PRA reliability a'dn design and operational review.
10 I consider us to have very good Level 1
~11: PRA capability; very limited Level 2 PRA capability, although 12 g we 'do have design basis, LOCA, transient and some hydraulic - ,
4 s
+
< 'A
,'. { ' 13 capabilities; and we have excellent Level 3 capabilities.
14 In addition, we have done both internal 15 and external events analysis, Just to give you an ides 16 of where we're coming from'. 17 Yankee charted in the PRA businesa just We have c It about ten years egc, back in February of 1979. i;I 19 since that time completed some major work. The Yankee 20 PRA; Seebrcok PRA, which was in essence a contractor job 21 and~was done back in the eart.y eighties, '82 '84. f fg. 22 L We have done external events analysis. Though
.,s
]
J 23 I proponc that we don't do external event analysis now 24 for IPE, I want,to point out that we have done it and resolved 25 specific issues and found it very effectiva to both identify
' j i
1 Heritage Reporting Corporation l _ __ _ 1 -_ .__ _ 1. m- -
-)
e . 7 q e > n ! l
-l
() e I more appropriate changes.and also more cost effective ener. L i We've done some containment evaluations
- 3 on Vermont Yankee back in L986 when the five issues came j l
4 out and we got hit very hard by our state.' 4 5 Wo started a Maine Yankee program in 1987, 6 which I'll talk about shortly, and plant life extension 7 at Rowe. 8 I'm not going to go over this. This is 9 a' sample of s~ e of the uses of PRA at Y.e.nkee. It's one p to . page of probably a half a dozen pages and does notsinclude t n ? 11 , opr routine design and operational evaluations; , s o
, 12 To give you an example, tornado /high wind; ,
.l
_ (?. i 13 seismic, battery room ventilation, ATWS, to evaluation 'j
- , g 14 of con'cainment effectiveness, tech spec changes, et cetera.
i 15 We UGa it very extensively. 16 Our IPE approach in pretty straightforward. t 11 In 1982 we completed a Level 3 PRA on Yankee Ecwo. That 18 .was"done about 50 percent by contractor and 50 percent IP by. Yank 2e personnel. 20 Right now W9 are updating that PRA, a Level 21 3 PRA. We expect to have it complete lata 1!90 or mid-1991. L . l
, s. 23 It may include selected e'xternal events. It may not. l
,j 23 It's essentially being done completely.in
- 24. house, understand how much work it is to do these studies.
25 When we first did the job back in 1481, I was on it full l Herunge Reporting Corporation _ _ _ _ _ _--- _ _ __ _ _ _ .. mee- ,
~ > '
(~
^
l . L' 318-i Eime. .I no lorger can do that, and we are essentially 2 training an entirely new team. 3 on the Maine Yankee job, we started that' 4 in 1987, about mid '87.' We just completed a plant model 5 phase one of internal events, plant damage states. l 6 It's essentially completely in house and 7 we have about six person months of support from contractors. 8 It's a. fine job. We. concentrated on identifying versus O characterizing.the plant based'on other PRA findings. 10 In other words, we J. investigated.. We didn't I 11 take somebody else's PRA model and requantified them. I 12 think that's especially important. {[,)f 13 + Right now in Maine Yankee, and it's not l 14 on the slide, we're into a limited phece two which will 15 involve Level ? containment analysis, and enhancing what i 16 we found in our phase one program.
!? Seabrook was your classic courtroom case.
18 Done in the early eighties. It has been extensively reviewed, l l 19 both externally Yankee and internally Yankee, We h&Ve 20 . updated it to reflect the plant as of about 1906. There 21 have not been significant changes to the plant. 22 We will probably Lepackage it for IPE l
^ ;/. -
l 23 submittal. We're considering submitting all of it, including 24 cxternals. We have not made the dec.ision yet. 25 The other interesting thing, and I keep Heritage Reporting Corporation A mc a -
~
O' O
~
q 319 i i saying this. We're training a new team. People that. worked f2 ' on that"have either since left or have new positions. < j r 3 At Vermont we're not sure what we're going 4 to do. We're probably going to do something similar to 5 Maine, which is what I call a-focused PRA, Level 1,' Level- . 6 2, a little bit of Level 3. 7' In terms of resource. split, the majority 8 in house and a little bit with contractors. That's the 9 key to doing these things. If you give them to contractors, 10 you're in tEouble. i l 11 ' This one is our view of the keys to success, 12 which is the technical tasks. I won't go over all of them. , First of all,, I'm a big believer in looking ll 13 14 for initiating events, uomething I think that oftentimes ($ fgetsoverlooked. I'm rather upset when I see people list j 16 five initiating events and that's all they evaluate. I 17 don't think that accomplishes t:he job. 18 We feel it very important to do Failure 19 Mode Effect. Analysis on control systems such as D.C., A.C. 20 l and air. That's where the action is. l 21 If scrieone wants to have an example of what 22 we find when we do it, and certain other companies I'm 23 aware of, we can talk separately. 24 It's critical to look at cross-connect 25 capability, especially in the older plants. Heritage Reporting Corporation , j _ _ _ ___ ____.---_________ ___ __ _ acacca _1
7s . )). r' y 10 ' , '320 l , l% , a . < . Ly 1 Inadvertent' operations are critical. And L 1 2 Tee think instrumentation' buses are important. l 3 We also think you should think very carefully l
~
i' 4 about front-line systems. 1 5' In looking at LOCA's, you should be looking I 6 Lat inadvertent PORV operation, not just putting a number-7 in. Look.at the* control circuit and see what happens.
~~
E We also think it's important to-model non- ! 9 safety systems, such as turbine trip. l 10 We think it's important to do a good event 11 sequence analysis, not take somebody's out of'a cookbook, , 12 and find out.how the plant really works. Also, as an example, l( ) 13 you'have~to look at'your procedures. 14 'A quick example, at Maine Yankee, the 15 amergency feedwater control valves are normally open. A 16 closed trip, the preferred means of providing feedwater 17 is working properly, they close the control valves. 18 'Now you get into failure mode. If main 19 feedwater fails, in five minutes you've got to open these-20 valves. You won't catch that by looking at models. 21 In terms of function versus systems analysis, 22 Yankee believes in function analysis. When I give an analyst 23 a job, I say, "I want the second heat coolant. That includes 24 steam, feedwater; make sure we don't have short-circuits." 25 We look A to Z; we model A to P. We don't Heritage Reporting Corporation nn
/M-
[ \ l,_)
/
321 1 think it's necessary to model closed circuits, relays and 2 contacts, unless it's some, thing in common. Then you should - 3 look. 4 Check valves are important. I won't get 5 into all of this. Train swapping. We do do parametric 6 common cause. We try not to spend too, too much time on 7 it. We spend quite a lot of time, also, explicitly trying 8 to identify potential for common cause. 9 The data, we do it for relevance, we don't ) 10 go crazy on it; explicitly watch out for maintenance. 11 Human actions bothers me a lot. I listened j 12 to a nice presentation today. I still think there's no (])~ 13 clear-cut methods for doing it. It's still very troublesome 14 to us. l 15 Quantification. One of the problems that 16 Yankee has with fault tree linkage is that you can truncate , l 17 out support systems. It becomes very important in initiating 18 events. They get truncated out and you never see them. 19 Uncertainty we think is very important. 1 l 20 We think it's important to document areas of uncertainty ! 21 as the study progresses. Concentrate on qualitative } 4 22 evaluation, high, medium, low. j N.) j 23 That does not mean that you don't understand l l 24 quantitatively the potential uncertainty in different l 1 25 parameters, but we don't think it's important and critical l Heritage Reporting Corporation l mm- . __ ._ i
, y l
O O I: ! C) d 322 j l 1 to go through a formal Monte Carlo type of propagation. 2 Documentation issues directly related to 3 IPE's. Thick reports are expensive and their usefulne.ss" ! 4 'is very. suspect. We have a Seabrook study this thick. 5 It's a heck of a nice job. It's a big, thick study. 6 We have a Maine Yankee report yea thick 7 and calc files that take up a couple of bookcases. It's 8 actually a more useful study for us internally. 9 Concentrate on structured calc files. It's 10 critical on all people in my group to ensure all key 11 assumptions are documented and their potential implications , , 12 assessed. Not in every formal report necessarily; in the 13 calc files. j 14 Documentation should meet good engineering 15 practices. It should not and cannot, in our opinion, meet 16 Appendix B requirements. 17 In essence, Yankee uses a structured i 18 documentation process for every task. 19 I'm being a little. cocky here but this is 20 the kind of template for closure that we see at Yankee. 21 - Aasources. A couple of dedicated PRA types. 22 On the Maine Yankee job I have two-and-a-half. One design
.) .
23 type; on the Maine Yankee job I only had a quarter. We're 24 a flexible company. 25 A quarter of an operations type. We were Heritage Reporting Corporation _ -- _ -_____ ocxa j
. . \
/]~ Q.
323 1 blessed with the people on our team, but I tilink you can
~
2 do these jobs without overloading operat' ions. -In the 3 case of Yankee plants, we're not overloaded with operations 4 people sitting around. 5 The Maine Yankee job, we did it with a half 6 a contractor for a year. In general, it might be more ! 7 reasonable to assume one or two. At Vermont Yankee we're i i 8 going to one contractor year. 9 We think it's important to go in phases ! l 10 so you get something useful early so you can learn, which 11 we think you can do within two years, which includes a 12 detailed dependency investigation. I cannot emphasize 13 that enough, especially in initiating events. (}b 14 I think it's something, quite honestly, 15 if I can comment on 1150, I think it was overlooked. I 16 think it's been overlooked in a lot of work that's been 17 done. I think that's really where the action is. 18 A reasonable report I would characterize 19 as about 500 pages, quote, unquote; and detailed calc files. 20 Phase two, because you can never do these
. 1 21 PRA's the first time and come up with a perfect product.
22 Figure out what you couldn't do right, fine out what you 23 did wrong and fix it in phase two. 24 Do a moderate Level 2. I made comments 25 earlier that I thought'the Level 2 should be delayed. I Heritage Reporting Corporation
\
i
' / %
324 ! I I do believe it should be delayed. I think it should be l l 2 a moderate Level 2. We think of that as phase two. 1 3 When I say "two years," this could be a j 1 4 year, a year and a half, two years, depending upon your - 1 5 resources and other commitments and uses you have in mind. 6 Phase three, based on wh'at you find, be I 7 it external events or more Level 2 analysis. 8 You'll find we're not using this template 9 at all at Yankee, but we think it's good. 10 A template for closure, regulatory reaction. 11 I think it's critical that we have interactive review and 12 comment. I expect recommendations. I don't want to wait
/% ,. 13 till the end.
14 Concentrate of conclusions, not statistics. 15 And it's been our experience if we don't 16 get experienced reviewers, we spend more time explaining 17 it to the reviewer than we did doing the work. So I think 18 that's a challenge to NRC; we need experienced reviewers. 19 We don't have time to teach people fault tree analysis, 20 and of course, we have the same challenge. 21 I think it's important to avoid debate on 22 unimportant topics. 23 Quite honestly, let's learn how our plants 24 are designed and operated from this different perspective 25 together. Let's not wait till the end for you to take Heritage Reporting Corporation I emum.
, l i
k i V L 325 1 our stuff and give it a review. That's my opinion. 2 Finally, I heard yesterday an offer for 3 some future meetings to talk about specific issues. I. !
- 4 personally don't think that we have a consen' sus on how 5- to do a Level 1 PRA. That does not bother me but I think l
6 there are some critical issues that deserve some more 7 discussion. 8 I'think we need to think about an integrated 9 effort., maybe some meetings to exchange ideas. 10 Ensure industry participation; and strive 11 for practical versus mathematical eloquence. 12 Again, let's ta'lk. f([.'je . ' 13 In summary, and I have talked pretty fast, 14 I think that Yankee has reasonable PRA experience. I think 15 we have exceptionally strony risk management experience. 16 We've been doing risk management since day one and I think 4 17 every plant owner / operator has. We call it different things. 18 Specific IPE approach and product varies 19 amongst our plants, because the plants vary. Documentation l I 20 varies. The resources available to us vary. The complexities l 21 of the plants vary. ] 22 If we look at our plants, Seabrook is the
- s. !
23 easiest to analyze. Maine is tough. Rowe is real ba'd j i j 24 to analyze, because it's got so much redundancy and that i l 25 is very difficul.t to analyze. Heritage Reporting Corporation ) ______o_____________________.___. ___ _ _ _ _ Javaa= ___ _ i
^
~
O O 326
- l. Prescriptive qu'idance, standardized product 2 inhibits learning, product usefulness, product quality.-
3 We understand it's important for the NRC to indicate what 4 they'd like to see and we think that our product will meet
, j 5- the intent of that information, but it will not be in that 6 structure for these reasons:
7 We are doing PRA. The specific scope varies 8 from plant to plant, depending on the issues. We are focusing .; 9 on internal events, although as I indicated earlier on 10_ y at least one, Seabrook, we have a comprehensive external l
!! I event analysis.
[ 1 C On the Yankee plant we have dcne limited F t.. j . 13 external event analysis for a specific issue where a PRA f 14 seemed to be the optimum wa6y to resolve the issue. j i 15 We are focused on plant damage, but again, 16 we are doing some Level 2 and Level 3 work, but not at 17 the same level of detail as the Level 1. 18 Again, I want to emphasize dependencies. ; 19 And our documentation is structured for 20 our use and we could provide a NUREG-1335 cross-reference. 21 We feel confident that we will provide what you want. It
., 22 will be different sections.
t j 23 That's all I have to say. Thank you. 24 MR. FLACK: There will be a time in which 25 questions and answers can be raised, a question-and-answer Heritege Reporting Corporation' ! _ _ ___ __ 9 ma~
- O O j 327 l 1 period after the break. So why don't Sni just go to the 2* next speaker.
. I 3 Patrick Donnelly,. Consumers Power. His !
4 presentation is experience with Big Rock Point PRA.
.i 5 ,MR. DONNELLY: First of all, I guess I would- l 6 like-to thank the NRC for the invitation for us to speak j
7 on the successes of the Big Rock plant PRA. ' ' 8 Many of you might not know but we've been i i 9 around since about 1960 and my presentation will tend to 1 10 be' divided into three areas. One will be the history of l 11 Big Rock up until today. 12 Two will be the PRA study, why we did it f'T J 13 and some of the applications that we've used it for.- 14 And third, I'd like to talk a little bit is about today's application and the way we use Big Rock in 16 what we believe is a little bit of accident management, 17 the current subject today. 18 First of all, the history. Big Rock was i 19 announced in 1960 and its construction started with part 20 of GE prototype end of the business. There were three j j 21 plants. It was called Operation Sunrise, j
,~, , 22 It's basically a boiling water reactor with 's l )
23 'a steam drum and a PWR-type containment, large steel ] l 24 containment with about a million cubic feet. ) l 25 In the early sixties, we used the plant i l Heritage Reporting Corporation _ __ ._ enca - 1
,/ %_,
328 1 and we had a lot of the startup problems that a lot of 2 people have experienced, and we ironed our way through - l 3 those problems. And in the late sixties, early seventies, 4 began a real operation stage where we were producing a l 5 lot of power and making long, good, steady runs. l l J 6 In 1974 we had a 343-day run, following 7 by a 319-day run, I believe, which was real good, and we 8 were running real well with about a staff of 60 or 70 people 9 in March of 1979 when everybody knows what happened. Along 10 came -Three-Mile Island. ] 11 Three-Mile Island pretty much motivated 12 us to.do something. The preliminary cost scopes for l []' 13 modifications resulting from Three-Mile Island were in 14 the neighborhood of $200 million. The plant only cost 15 $22 million to build. I l 1 16 We did some simple arithmetic that we had j 17 to do something. So we proposed that the NRC delay the 18 implementation of the 0737, 0654, 0578 type regulations 19 for a year while we did a risk study. l 20 We employed a contractor. We tried to do l 21 a Level'1,'2 and 3 PRA in a year. It took us 12 man years 22 worth of effort. We had key people on our staff dedicated l s 23 to this study because we wanted the answer to be right. i 1 24 The reactor engineer, the INC supervisor, 25 shift supervisor, the input from SRO's and RO's alike. Heritage Reporting Corporation (2er) 42s.4sse
f3 (~ v \ . t 329 t 1 As you can see by the slides, this first 2 slide -- I apologize for not having enough for the audience 3' and they are a little hard to read and I realize it's busy,
' 4 but there's a lot of information.
5 Post-accident shielding was in the neighborhood )
' 1 6 of forty million, because we have just a large steel J 7 containment.,
8 Post-accident sampling, $300k. 'nstruments
. l 9 to detect inadequato core cooling, $1 million.
10 All of these items were proposed by 0737, ) 11 ' I believe. 12 After we got done with the PRA, we came- J i up with some proposed modifications for Big Rock that we '
-() 13 14 thought did more for' safety probably than the other ones I 15 did and we were able, by the last column, PRA methods used, i
16 to eliminate many of those required items. j 17 The thing that we did find out was that 18 in a small LOCA, for example, the temperatures inside the 19 containment exceeded the fuel in some areas, so we proposed l 20 an earlier enclosure spray that cost us about $200,000 21 to put it in. We put it in, of course, immediately. 22 PIS position locks, that's a post-incident ! l 23 system. We have a recycle system similar to most containments. f l 24 Water builds up in the bottom of the containmen.t and there's ; 1 25 seven or eight key valves, manual valves inside the I l Heritage Reporting Corporation
- om m _
4 D O
- r:
(%.) 330 I containment'that-if any of them are in the wrong position, 2 e will' fail the syst'm. 3 Even though these valves had been double-4 ' verified for years and. years, it still turned out to be i 5 a pretty good risk. So we developed a nethod to lock those 6 - valves;with locks that will only lock when they are in j 7 the correct position and that eliminated some of those i 8 type failures.
.[
- 9 Remote fire protection system makeup to
.i.
10 emergency condenser system. The emergency condenser at ; 11' Big Rock, if'the rods go in an'd the containment isolates, i II it's relatively' easy to remove decay heat with the emergency ( }; 13 condenser. 14 One of the problems that we had, it was 15 dependent on AC power. The makeup of the shell-size emergency
- 16. condenser was depe'n dent on AC power. So we proposed 17 modification to make remote makeup to emergency condenser 18 ehell using the diesel fire pump or the electric driven 19 fire pump, which is e.lso supplied by the emergency generator.
20 That eliminated a lot of sequences. 2! PCS isolation valves. We found several j 22 eituations in our primary system where the primary system
- s ..'
23 had single isolation valves. For the interfacing system 24 LOCA the probability was higher than what we'd like. In 25 all instances we added double isolation valves and put Heritage Reporting Corporation ! wm l
~ !
O- O. 331 1 them on a checksheet. 2 Feedwater instabilities is a problem that
. i 3 we're still working on s'ince 1981, I.ts much more complex !
4- than we had originally anticipated and it had to do with 5 instabf.lities in the condenser and hot well when we get i i 6 a full load rejection, transients. , 7 Hopefully, we'll have that solved by the 8 end of this year. f 9 Again, alternate shutdown panel. Fires i 10 turned out to be extremely important at Big Rock Point il in many PRA's. Originally, we planned on spending about i 12 $2.6 million to develop an alternate shutdown building. 13 We had a control point outside the control
-(]) .
14 room and wires running to different locations. One of 15 the problems that we had in an earlier design plan was a 16 that all of the electrical penetrations went in through i 17 one room and, of course, if you had a fire in that room, 18 it was not good. 19 So we proposed an alternate shutdown panel 20 to answer the fire situation. In later analyses, as you'll
. 21 see on the next slide, it became apparent that winds and l
22 tornadoes and missiles and other things, external events,
. 23 were important to. ,
24 So we changed the. location of the alternate l l' 25 shutdown building, made a seismic and missile-proof building Heritage Reporting Corporation E--------___ ._ r,nm
,n m , ; c.
i k 4 >
/ c- 332
% I j 1 there. ,So the' cost escalated from'$2.6 million to-about J 2 $4 million, but we:got more bang for our-buck. , 3.- Since Three-Mile Island and since the SEP 44 Program, we.'ve answered several it. sues 'using t' e PRA nodel.- , 1 5 You can'see here a list of a lot of them. 6 shift staffing, for example. Almost everybody
'l~. l 7 has a shift i -- 'I believe everybody his a' shift engineer.
8 ~At Big Rock we wer6 able to go through all l l 9 the sequences and look at operation action event trees 1 i 10 and develop operator action event trees, because the personnel 11 W requirement to analyze where we were at in a situation 12 and who we needed there, and we were able to put the shift j j, 13 , engineers on a thirty-minute or one-hour callup. I 14 So we don't have shift engineers on site, and I 15 we have them on beeper and they are demonstrated to be 16 ' in the plant in an hour, which turns out to be very ccst ! i 17 baneficial for a 72-megawatt electrical plant. I 18 Recirc pump trip. Again, one of the fixes 19 for ATWS and boilers was a mandatory recire pump trip feature. j 20 Recirc pump trip turns out not to be all that important. i 21 At Big Rock our Aiguid system is similar i 22 to a slick system and a big blower is different in that 23 it goes in at 132 gallons a. minute versus 38 or 43, which 24 allows us to get a shutdown path within about 40 seconds. I i 25 So that significantly reduced the need Heritage Reporting Corporation z___._--_ u m*=
-~
. g.p
- i. y '
3a: 3.,- 7 , 3 v . g, g
, h. ~+
333' i
- . M for recirc pump; trip e.t Bdg Rock..
ip 6 t 1
,l
. 2 On down the list there'u'several other things that we've been able to do 'at Big Rock Point ' to make the 4 3]
y ., 4- plant cost effective and yet. still provide' a safety margin
~
5'
~
that everybody is looking for.
"! > 6 If I could have the: final-slide,-I'd like ,, S 7 to' talk a little bit about what we're doing now. Everybody 8 .seems to be using the risk management tunbrella.
9~ Some of the things that we do wi.th our PRA
'30 that may be unique to'anybody else is that we use it in
!! . the training function. What I mean by that is that every 1
12 RO, every SEO, every shift engineer or technical advisor,- hV 13 _as we call them, goes through~ accident training and analysis j 1 1
-14 training.
I 15 A part of that training is there's lesson 16 plans developed in the accident sequences and.they are 1 17 spelled out in the event tree' form. The training instructor 18 goes through all the accident sequences with the operators ' 1 ! 19 and the STA's. .j i, 20 Then at the end of that session someone
]
21 from the risk c)roup goes in a'nd we talk about support systems 4 i 22 and what the key parts of our plan are as far as safety ; 23 is concerned. I 24 In emergency planning, I think everybody 25 that has a PRA finds some usage for it in emergency planning, Heritage Reporting Corporation a :_ -_ - nn
Ng7 >v i >
.L m.
- V i x^ ..
.m t
.;c a
n , 1 d * @W . t; .j ! (*Q : y,3
' 0 c.' .
334 j l ti I as far as Ldevelopment .of exercise scenarios and . data that 2- is used for exercise scenarios. 3- Control room design, review. What we did 4 in the control room design review area is put. dominant accident 5 sequences and we looked at all the headings in the dominant-6 accident sequences and the equipments that are used in 7 it and we used that for our function and task analysis , 8 for the CRDR. 9 We've told the operator, "Yes,-I can use 10 this switch and here it is." We made some changes to our 11 control room based on our dominant accident sequence. 12 Emergency operating procedures is another 13 .. application that we used the Big Rock PRA for.
<l}
14 Even though we are a boiler, we're a whole 15 lot different thari any other boiler that's out there and l 16 although we've sat on the EOP subcommittee that developed 17 the emergency operating procedures, we weren't'real 18 comfortable that everything that was in there applied to- i 19 Big Rock. f 20 So after we took the plant specific technical i 21 guidelines and developed our guidelines from the generic i a 1 22 guidelines, we took all the headings in the PRA and developed l 23 symptom matching sets. i I 1 24 In other words, the equipment that led to l 25 the heading failure had symptoms that caused it to fail, I Heritage Reporting Corporation ec:a - - j
O
- l. ]: .
(~ (~r , V) l
. .J
- l 335 i l' and we bounced those same symptoms off the emergency procedureo !
2 and added things that were plant specific. l l 3' I think we got a little bang for our buck' ! l 1 9 ' '/ - 4 in that. At least we felt comfortable that the EOP's as
- l. . :
5 built certainly covered all the dominant sequences. 6 That who.'.d umbrella, that's kind of what 7 we call at:cident managemsnt. All the operators are trained 8 in the sequence 6s Contm i room design review went through 1 9 the sequences to make sure that all the equipment-that 10 we're relying on was there. 11 The emergency planning,uses it to develop 12 their scenarios. We do allow some success for the operators, () 13 that they can restore the plant, put it back in order, 14 which is allowed in the years when the NRC isn't fully 1 j 15 . involved. , l 16 And we use it in EOP's. 1 17 And I thank you. I
. :l 18 MR. FLACK: Thank you, Pat. We will have j 1
1 ( 19 the slides available for you tomorrow morning us will v.he l 20 other speakers' slides be made available to you, j l 21 Our next speaker is Paul Hill from I
... 22 Pennsylvania Power & Light. The title of his presentation l
>)
i 23 is Pennsylvania Power & Light's View of Severe Accident !
. 1
-24 Issue Status.
25 MR. HILL: My name is Paul Hill and Heritage Reporting Corporation _ _ _ _ _ _ _ _ _ _ - . - _ - - - - . - _ _ _ r,u~
r___--- .,
.q
'. H 336 I
1 in this presentation I will represent the views of ! I q 2 Pennsylvania Power & Light company on the seve.re accident
, 1 1
3 issue and the role that tho.IPE process will play in the , I "Y ' 4 resolution of this' issue. . 5 VOICES: We can't hear you, Paul. ' 6 MR. HILL: .Shall I speak louder? . 1 7 PP&L operates a two-unit BUR-4 plant with 8 a Mark II containment in Pennsylvania. The units are rates 1 9 -at'.3293 megawatt thermsl each. o 10 We have completed an IDCOR, a modified IDCOR I Al IPE on Su!3quehanna for a single unit at Susquehanna at i
)
- \
12 the end of 1985. 1 7- , 13 _ In the intervening period of time we've j I
' l4 reviewed the results of that IPE that we performed and -j l
15 from it learned quite a bit about tae issue of severe accident s l l t , 16 and how the frequency of more serious consequences from l i i 17 such severe accidents could be reduced. 18 The first vieugraph. I'd like to give some 19 views that we have on the status of the severe accident I 20 issues at the present time. i 21 The first is that in our opinion the current 22 focus on off-site consequence, which really has dominated )
,)
- i 23 probabilistic risk assessment from the time of the reactor )
i 24 safety study, has led to the use of inappropriate ) { l 25 conservatismse l Heritage Reporting Corporation , j l' <mni a
)
- l. - . _
n p,
. %_h wl 337 1 The reason for this was to enhance the credibility 2 of the results of offsite consequence calculations; that is,-
3 the desire to be conservative so there could be no 4 arguments, but that offsite consequence calculations were in i 5 fact founded. 6 The result of that in our view is that the J 7 Characterization of human performance that has come out of l 8 such studies would lead you to believe that operator 9 performance is from our viewpoint once again relatively 10 poor. l i 11 Their error. rate is h!.gh, and they err at the most ) l 12 critical times. 13 We think that the treatment of common cause ] P '% M I 14 failure has been, in general, unrealistic - The numbers used l C is typically have a major impact on the results of the risk is assessment, but at the same time they have virtually no , 17 basis in observation, or only limited basis -- to an extent 18 that has much less of an impact than the assuraptions it ; 19 would lead to. 20 We believe there has'been a use of overly 21 conservative and unrealistic phenomenological modals in the 22 assecaments, and that there has been an Overly conservative 1 23 choice of success criteria. j t p4 We think that the results of these conservatism 25 which may be laudable from the point of view of the ; 1 i ^ _ . - _. - - - a
;, ,r\
%, N) 338 .
1 writer -- upper bound to the estimate of offsice consequence , t 2 -- has been that it completely masks and conceals the 3 potential benefits of optimizing the emergency operating ] 4 procedures for the plant. l 5 We believe these emergency operating procedures l
\
6 are the true key to reducing risk, as opposed to hardware ) 4 7 modifications. ) 8 We think the approach has resulted in an g appropriate and negative assessment of BWR containment 10 Performance, in particular. This is something we find 11 Particularly troublesome. j 12 That negative assessment has resulted in pressures 13 for potentially inappropriate prescriptive hardware fixes
. ..c ( 14 for the Mark I containments.
15 And, of course, our concern is that the Mark II 10 containment is functionally very, very similar to the Mark I
,7 containment. So we see ourselves facing these same 39 prescriptive fixes.
39 We believe the most effective course for risk ] l 20 reduction the BWR plant is developtaent or optimized 21 I emergency operating procedures. And a very important
'i 22 concept, a demonstration -- a demonstration -- that they'll 23 , be effective. .
24 The demonstration means that you do transient as ' analysis to show that those precedures have the effect that ) l
7 7 9.. . , , g ,
- p. O O V ,
m -' 339
.f5
, you want them to have. They, in' fact, work.
p., '
, 3 L You do.transf.ent calculations to see if there's
- 2, 3
suff'icient time for the procedural actions that.are defined. 4 at a reasonably high confidence level for success. 5 Finall'y, the last item on this viewgraph,.we 6 believe that the currently existing models for BWR plant. , 7 response to reactor vessel . failure -- that is, cora melt j 8 with vessel failure, followed by a core degrading into.the g dry well -+- really can't be considered to be a valid basis 10 at the present time for prescribing hardware modification. 11 We'll be saying more about this point later in my 12 Presentation and in the next presentation by casimir, 13 Kukielka. r) ' I'm going to have trouble going through this
..ej 34 15 material in the time allotted, so I'm going to skip over a 16 . number of things.
37 For anyone who is interested in the content of 18 this presentation, I would suggest that you read the printed i 39 material that we have handed out with copies of this 20 presentation.- y I 21 There are six items that we believe have to be 22 addressed if an IPE is to be useful in severe accident t 1 23 management, or as we call it at PP&L, the broader issue of l { 24 risk management. 25 The first item on this list involven developing a i tsO 1 l
= _ . _ _ -
- k. -
g . - , -
~
O O ' ;1 340 l
.."'N logical and realistic and verifiable model for operator j 3
2 . Performance in response ~to initiating events. 3 We make a distinction in the human error arena between human error that causes an initiator, human error 4 5 that causes a system to be unavailable, and those human 6 errors that have to do once an initiator haa occurred. i We think this distinction is-very'important. 7 8 Thesa first two we view as being embedded in the plant-9 records, and when we go to quantify our initiator frequency. 3o ] and our equipment a%ilabiliti=as, ye use p), ant data for 11 that. So whatever the humar. error contribution is, it ' is 12 13 embedded in the records. On the other hand, in response.to initiators, ['- 34 there's another categorization that's extremely important;- 15 16 and.that is, you have to make a distinction between errors 37 in logic and errors in taking actions that are dictated ~by the logic. 33 39 The way we do that at PF&L, we have logical, flow j 20 diagrama to represent our EOP's, and they represent the , logic for the operator's response to any initiated event. 21 g Any additional equipment failures would either 23
. result from the in5.tiair.or or would come independently of the initiator.
24 { 1t.a very important that a very high retiabiitty
]
4 l b
* -. . l.
r = . r
<8
,, -m:c
't, .,q i ,
,e h' .
.341 "F-1
~
for operator action in following those procedures be- l 2 -
.We've made measurements at the'Susquehanna achieved.
3 .. . simulator to attempt to de.monstrate'that our expectation of 4 . I this error rate is low; and'to date w.e believe we have' 5 verification,, and"we have quite' a low error rate. 6 We haven't shown it's as. low as we think it needs 1 7 3 bo-be, but we are ? working on the problem. l We have a .new set of' measurements that we will' bn 9 making in the future that have. to do with the actual' actions 10 .
] x that are called for by the EOP logic flow diagrams. But for. ]
'l 11 . .
these the error rate that's tolerable is'much, much highers l 3 12 - I
.that is, as long as the error rate is co.mparable to or lesr j I 13 , . -3 ,
than the unavailability of the equipment involved, then j J( 14 there's a relatively modest impact on the consequencesEthat' , 15 , result from that. ,
'l 16 Typically, the error rates are in the range of' ten 17 to the minus two. .
18 We believe that it's very .important to make these-19 distinctions,'and we are attempting to establish measurement 20 .s programs that will enable us to demonstrate that we achieve j 21 the noccasary rates. ) :
, .t 12 i
'he next item -- and this is another extremely T
i 23 J important item -- is there's a need to develop a q " i
. 24 comprehensive; realistic and economica) accident transient 25 .
I analysta program. ; I ( I s
77 p, . . .
+ ,
g-I
' (( v '
. 1 1
342
-V- q 3 ~
s - Our initialJtransient -- or IPE work,was based ' 2 i entirely on that. - In - recent tilnes we have chosen 'to adopt' 3 ' the Oak Ridge National Laboratory ' developed code, BWRSAR; ' I 4 and'we intend to link the Sandia-based contained code to ;
=l n S >
BWRSAR to give us an $ntegrated. accident' transient. analysis. o 6 - - capability. , , t i
'7 ~
Now,' cur interest in this is'not source terms. j s , car inter'est is in determining the proper piocedures to i g .
. l accomplish a'sufficiently low rate of failuree in equipment 1 4
t 10; 'and. compensation for' failures of equipment so that our risk ] 11 of a severe accident event'is sufficiently' low. 1 3 12' " ( Wo need codes that have goad physical models for 13 . the cystems -and that are able. to represent properly the J.
.- - 14
, operator'st actions that are taken to interfere or. interrupt
-15 the accidenE progression.
<a h The third. item that we think is important is the
. , 17 [ development of what we call' optimized emergency operating 10 ,
l procedures. 19 What we mean by this, these'are procedures that 20 have to be based o.1 plant-opecific desiga features. That l is, they.have to take into account the equipment that you
'2 have in the plant.
23 The way we state this is our goal is to make sure U i l' 24 that we have fully utilized the plant's capabilities to 25' cont;ol the course of an accident in an optimum fashion. We i a 2 1.- .
~, -. -
-y,'._. ,
i;%. ( ' i , 3 ,, j n y gH g, m - '
;n \ q sy !~ t . .
~'.
K . n;
,Q< a
'( ,
g m :< t
< i 343 -
'i v.
I want to make sure that our operators will use everything. a, . 2 that's available t.o them and that they will -do it corrsetly. 3 We believe:these procedures hava to be based on. 4
' unambiguous, prioritized,, symptom-based icgical' flow 1
1 5 diagrams. The logic is crucial in making sure that you have l .) 6 a good set of procedures. - 7 The actual actions that have to be taken, thoso - 8 procedures need to be good as well. But they are 1sss 9 critical, as I said earlier. ; 10 And, finally, we have to demonstrate, using our II i transient analysis programs, that those procedurss are 12 ' effective on all of the initiators and all of the equipment-
'8 failure combinations that your IPE identifies at your plant. !
* '4 The next item is the development of a j i
15
. comprehensive, realistic computer-based method'for risk ;
I6 analysis. I7 ' At FP&L, based on our epiginal' IPE work, we 18 believe that we have found a wcy to formulate such a l 18 computer-base d program. - That program is now in existence, 20 and in in, I hope, the final stages of debugging. 21 Casimir is getting answers out of it. 22 Unfortunately, they're not yet quite the right answers. 23 This is a very important thing because du this' 24 viewgraph I've listed the things that have to characterize 25 that c!tlculation. J l . t t' mn.e ---a____ - - - - - - - _ _ _ - - - _ - - - - - - - - - - - - - - -
,, yy, b; , . ;, - ,.
- (, I
- {j ,.
' u!pw n '-
,j
/
i. ( n
.p " y m n i
3 g. a , < yq, l' l 1 < 3'44. - yy I .An"2 Wile I don' t have time do go into the j r;q 2 details, ~ these- thirigs 'are extremely important in gettirg a j k 3 realistic representation of' the ' dominant risks or threats' to . j 4 4 j
. the . plant .- ,
5 The fif th item !s development of a performance-6 based acceptance criteria for demonstrating the adequacy of I the plant from the point of view of severe accident risk.,'( f
~ 8 l 'At PP&L we use a -set of what we call defense-in- l 9 depth criterie. Our definition for d'e?ense-in-depth is l
'10 somewhat different than most people tend to think of.-
E "' 1 If anyone is interested.in this, we have a few 1 12 copies of a document that we developed for an IPE Workshop 13' for the BWR owners group, in which all of ' these, terms that l 14 I've used here in this presentation are defined. We have a
! ,15 ' few copies that we can hand out to anyone that's interested, j
.16 If there are too many requests, we'11 be happy to mail a 17 copy to you..
18 And, finally, the last item is. we need measurement 19 programs to demon, strate an adequate level of equipment and l
- j. 20 operar.or performance.
21 There are six items that we think are very 22 important in this. That is,, equipment unavailability, 23 . equipment demand failures. We believe this will come fro'm i I 24 L our plant records. ') l 25 Initiating events also from our plant records. - ].
)
l
\
- - + - , (
) , e\
V V 345 :j i
^) 1 Precuracy events are the way we will deal with the 2 very rare equipment failure combinations, which most people 3 call common mode or common cause failures, because we just 4 simply wall never have enough data to get a reliable 5 estimate from our plant for these typa of events. h 6 . But in uur studics of the reactor scram system and 7 the RHR suppression pool cooling system, which are very 8 highly regarded throughout the system, we've used these 9 techniques and we believe derived pretty credible results.
10 Operatcr adherence'to the emergency operating 11 procedures, this is a crucial test. 12 And then finally operator execution of the EOP ; 13 actions that are called for. 14 In conclusion, we think that the current basis for 15 prescriptive fixes is really not appropriate for the BWR's 16 , based on some of the phenomenological problems that we see j 17 with the accident progression codes that are in existence 18 today. ) 19 We have a cooperative effort with EPRI to work on , i 20 developing better core concrete interaction models, Which 1 i 21 are what is currently needed to get reliable and credible ] 22 results for the accident progression. i 23 In the next presentation by casimir Kukielka will 24 tell you our views on this subject and how we intend to 6 25 handle it. j l ! 1 _ _ _ _ - . _ _ _ _ _ _ _ _ _ _ _ _ _ . {
' ~
- o. ,.
346
. f .s },- 1 MR. FLACK: Thank you, Paul.
2 Our next speaker is Casimir Kukielka, and I must- . 1 3 apologize for~ spelling his name wrong, although he told'me 4 he spells the name wrong himself, so I don't feel so. bad. l 5 Casimir is from Pennsylvania Power & Light. The- ; 6 title'of the presentation is " Interim Containment Sequence 7 Dispositioning Criteria." 8 STATEMENT : 1 9 OF f 10 CASIMIR KUKIELKA 11 M2. KUKIELKA: My name is Casimir Kukielka. I'm j l 12 from Pennsylvania Power & Light. 13 I wish to talk about some interim containment s '
.. 14 sequence disposition criteria that we plan to use in our 15 IPE.
h3 Before I do that, I'd like to go over some i 17 -background information as to why we think these' interim f 18 criteria are.necessary. 19 First of'all, when we looked at the generic letter 20 that came out, we didn't find that there was speci'fic' 21 guidance on how to perform the containment evaluation. l 22 The way we read it was that the generic letter 23 defe'rred this issue to this containment performance and 24 improvement program. 25 , Secondly, from our view of existing computer codes 4 m_.----_--a-_a.-_a-- a-------t
' .y 'a
-~ ~
~ - -~ '
T gjn :
.j M- -
[
$_ 3. / Ji ;
( '
;h 4 -
.1 rT ;
j![ra .. 347 1 1 .andLmodels -used by..the - codes industry and NRC contractors, . 1
)
K 2 km'do-not.balieve-that' credible models existed for ) 3 evaluating the damage for progression beyond . reactor vessel
' 1
!. 4 failure.- a c As an example, the contain code which is used in { 7 's the NRS containment sequences utilizes a subroutine. This-7 code is developed through rapid dump to apply super _ heat in i~ c- a. core melt simulator to a deep concrete crucible. , l 9- The BWR containment isn't.a deep concrete to crucible. It's a flat floor, and in the BWR is usually very 11 .long in nature, and at first have low super heat materials
- 12. in' the bottom of the vessel.
13 ' Secondly, our particular containment we have aL j
/%.. steel liner on the floor, and we can put as much as 18 1 %1./ 14
'is inches of water also on the floor.
1 te Therefore,-we think it's really unrealistic to
\
17 expect realistic results from a code such a core-con, based. y 18' on the total physical arrangement. !
-i.
19 Next slide. i 20 However, we still believe that IPE processes go t 21 forth, and it's very important. We base this view on these
, l 22 three items.
- j 23 -First, we believe that the majority of the 24 accident sequence is to be terminated inside the vessel. l i
25 With these type of sequences, such core concrete interaction O e
L l p f) . [ L/ L' ) ! '348 i models are unnecessary. 2 . Secondly, as Paul told you, we are working with 3 EPRI on an analysis and also some experiments. We believe 4 that we can put into place good models to analyse ,the core 5 debris and interaction in the next few years. 6 Furthermore, we believe that at'this time we can 7 Put together some criteria that these models that we hop'e to 8 develop in the next few years will support. 9 We' based this belief on our review of calculations 10 Performed to date, our review of the experimental data to 11 date and also some scoping calculations that we performed in 1 12 house with core debris on the BWRSAR program. 13 With this background, I'd like to go over quickly ) i
, ' I 34 the criteria that we're going to use. For the case where ]
15 there's no core damage, it's basically just the ultimate 16 strength of the containment, when the containment fails. 17 We've identified two types of core damage or 18 interest. The first part is the loss of core coolant that 19 you recover cooling prior to vessel he'd a failure. 20 In this situation again it's simply the 21 containment will fail on over pressure when you reach the 22 containment wall strength. 23 However, in this case we have to conci' der the 24 generation of noncondensible gases and to get the partial 25 Pressure of a noncondensible gas in this situation, we plan
, , , y ,
O 1 O
~349 MB' -
3 to use BWRSAR again as progression modeling.
..+ .y It.was earlier said that 1EPRI is' putting hose ']
~2
~
l models into MAPP, so they would be available through -the. j 3
. l
, .. 4 BWRSAR or through MAPP. { w J
'5 The other type of damage we looked at is from .)
6 limited site. operation or excessive. reactivity occurring 7 during ATWS events in either pressurized reactor- with a-8 significant amount of; fission products. g We believe this transient is potent'ially severe _ 1 10 enough to cause loss of the chemical integrity of the clad I 11 and, therefore, you'd have noble. gas to contend?with.
'12 . Lastly, we think the last two criteria deal with 13 accident sequences in which the reactor vessel overhead 'has , )~ 34 failed.
15 We have two criteria there. One is for over .I 16 temperature failure. 17
% 0 way we plan.to disposition that is if the l 1
18 fission-product bearing debris exceeds the calculated depth, gg we'll see the containment fails. 20 If it's less than that depth, we say we can 21 stabilize the debria. This depth.will depend upon the , i 22 presence of water. 23 In the case where we've managed to avoid over 24 temperature failures, we still have to contend with over 25 Pressure failure challenges. Again, that challenge wil1 be F y l' l
.j c - - b
- r. .
U '
'(}
350' l i
p - 3
. when the pressure exceeds the containment's ultimate ;
I' 2 strength, the noncondensible gas generation again'comes from 3 Oak Ridge's BWRSAR code, and also.the noncondensible gas l 1 4 generated during the coring process or the reactor vessel. l 5 depressurization process once the head has failed. .q 1 Plus the case of water on the floor, we believe j 6 7 that t.en percent of that additional amount of hydroger has 8 to be accounted for. g We also believe though there are uncertainties in 10 this approach, and we plan to treat parametric' uncertainties in the modeling by parametric studies. We will look at the 11 effect' these criteria have on our end results, in 12 sensitivity studies rather than uncertainty. studies. 13 [)~~ g - Next slide, please. l 16
- 9"*** Y **** * "A ** * * *' ** * *'*
these criteria ale neceasary if you wish to establish 16 defense-in-depth in our plant. And as ' Paul-alluded just a ' 37 minute earlier, certainly we intend that this will be okay, 18 39 because we believe that EPRI will lend their support,to 20 these criteria. g Finally, should this be untrue, we think the 22 impact on plant risk will be' minor because our risk analysis has demonstrated that we can terminate the majority c5 1 23 accident sequences in a vessel. 24 That concitides ' my r6 marks. t,qe emum,s--,---_a-au--.-a-------._-_---x- -- _. a O Q $
q) f) v 351 1 MR. FLACK: Thank you, Casimir. 2 At this moment I would like to turn the floor over 3 to the Director of Research, Eric Beckjord-4 MR. BECKJORD: Thank you, John. 5 I wanted to say a few words before I leave. I 6 will be leaving in a few minutes. 7 I'm not leaving because I'm not interes.ted in this i a subject. I 'tish I could stay through the entire meeting, 9 but I have to be back for some budget discussions tomorrow 10 morning relating to the research, and also I might add, the 11 IPE budget. 12 So it's important that I follow through on that. 13 I'm very pleased -- I wanted to tell you that s/! 14 I'm very pleased with the courso of this meeting. I think 15 there's a very construct'ive spirit here. 16 I think that we've gotten some very frank and very 17 good comments, and we are going to work on that. 18 Excellent comments. Some important questions have 19 been raised, and we're going to go over the transcript and j 20 ' answer those questions, or we will do that in writing and we 21 will get copies out to the utilities, the attendees. 22 Those will be carefully considered comments, in 23 addition to the spoken answers that you've gotten here. : 24 We'll also put that in the public docanent rooms. 25 Further, we are going to revise NUREG-1335, as l l l mum.__. ..____'.____m__
j . LJ 352
- 1 we've already stated. But we will revise it to reflect the 2 answers -- to reflect our answers to the questions that 3 you've raised.
4 There are some points that I think are especially 5 important that we' want to give more careful thought to and 6 give you our considered response. 7 We're going to heed the advice that we've gotten 8 on several important points relating to the integration of 9 the entire IPE program, and also give careful consideration 10 to the points that were made regarding matters that will not 11 be subject to compliance in all of this. 12 With respect to accident management, as we've 13 stated, 'we will be giving you more of our thoughts on that
- 14 subject.
15 I think that's very important, and we also need to 16 be very clear about that -- about what we ' re looking for, j 17 and we want to take into account what we've heard today. 18 A. general comment on the guidance that we're going 19 to be giving you. We're going to attempt to strike a 20 balance. We're going to be as clear as we possibip' can on 21 it in regsrd to the IP5 requirements, but we do not intend l 22 to write any cookbook recipes. 1 23 We want to encourage careful thought on your part 24 about severe accidents and also to encourage creative 25 approaches to identifying plant vulnerabilities and to i
s' o i 353 'i
' 1 .taking. corrective actions.
i
'2 You yourselves can do a better job on th. tis if we 3 avoid being pres criptive.
4 That concludes what I have to say. As I said at i 5 ;the beginning, I'm very pleased with the course of the 6 meeting, and I wish I could stay for the rest of it. I hope 7 that the remcinder goes as well as the part that. I've been 8 able to attend. 9, Thank you. s i 10 MR. FLACK: Tnank you. 11 Moving right along, the next presentation is by 12 Wang Lau, Tennessee Valley Authority. .; 15 His presentation is insights from PRA Studies." [ 14 STATEMENT j 1 , 15 OF l 16 WANG LAU l 17 MR. LAU: Jim Chapman gave a very good talk 18 earlier. We prepared our viewgraphs separately, but af ter 19 seeing Pis viewgraphs in relation to his speech, ? feel like 20 1 can come up here.
- 21. - Of course, I have a Tennessee Valley accent ---
22 [ Laughter] , 23 -- and Jim has a Yankee uccent'. 24 Anyway, I would like to share with you en 25 experience we had in the last ten years or so. This PRA i (? _ f ..
.c ,
f g q y: L . L s . . i ' J O .. O-J u t- '
'1 0<
r 354 I 1 " business, you realize, of course, that we don't have many 2 experts in PRA. ,
- 1. >
3 Years agG we traveled down through alleys and! l 4 learned quite a1few lessons. Some of them were pretty 1 5 painful; some of them were pretty useful and .I would.like to ) L 's , share some of them with you today. l l 7 of course, three years from now we'll have so.many ? 8 experts you won't know what to do with them. ' 9 . .When.we' start doing a task, we keep objectives in , 10 mind. -The' objective is not to comply with regulations or l I 11 w
.hatever. The important thing is that we want safer plants.
12 We do not want another Chernobyl; we do not want-13 THI; we don't.want'too many precursors. Not zero, but not 14 too many. That's the objective. I 15 We have certain beliefs. We think that you share is the same belief with us, too; and that is the present
- 17. regulations do not optimize resources. The plant-specific ;
18 PRA la the best tool to determine the public health risk. 19 - The plants need safety goals. Why are we doing 20 anything? I do not want to be in the middle of that debate. 21 Okay~. But let's.just move on. 22 Even though we set safety goals, we've got 23' problems, right? That's why we do. things.- 24 I'm glad you've got problems. Otheryise, you
. 25 wouldn' t have anything to do.
w___ __ __ _ _ _ _ _ u __ . . . ]
. t q
b - U g 355 1 Increasing costs, perscnal dollars, availability,- 2 increasing outages, NRC regulations, these are some of the ; I 3 elements that Tom Potts is going to talk about that. He's 4 on the program tomorrow. l 5 I am sure that will get into his speech somewhere, ! 6 so I'm not going to go any further. l 7 Next one, please. l 8 Let's just say that TVA has done quite a few 9 PRA's. This is a summary. We have Brown's Ferry, Sequoia, 10 Watts Barn. 11 Drown's Ferry is General Electric, three units, i Mark I, restart. 12 , l 13 These are all in the 3400-megawatt, thermal.
'. 34 One is down for refueling and will be up sometime 15 this month ago.
~
16 . Brown's Ferry Unit 2, fuel reloaded, we hope to j l 17 get up soon. 18 We have done quite a few PRA's. Brown's Ferry,.we 39 have full scope Level 3, eight volumes of it have internal l 20 and external events. 21 Sequoia, Watts Barn, Bellafonte, we have done a , 22 Phase 1, a scope-type study of internal events only. 23 The TVA contractor -- These are approximate 24 numbers. h $ .
\
,_s , ,o i ; %)
l l l 356 j L .S 1 started off something like 90 percent contractor and ended ,
- 1 2 up 90 percent TVA. .
3 We are in a learning curve somewhere, and in the j
. )
4 rest of them we did some ourselves and some of them we j l L 5 contracted out. , ! 6 This here is maybe a little misleading. Sometime 7 it's very difficult, having the utilization aspect of it. 8 Other studies are the NRC studies, the WR study. 9 Some of the studies are very good studies. 10 NU REG-1150, there's no point in reinventing it. 11 It's good stuff and people should use it. 12 Next. . 93 This is not intended for commercialization here.
,.,~ 14 We just wanted to introduce to you what we have today. This 15 is what we have today, not what we had five years ago.
16 We have a staff of four supervisors who supervise 37 those people at the site, basically in the implementation 18 role. They are not there to cram our PRA. They have to 19 take the PRA. They can use it. 20 We are real heavy on the using aspect that they 21 are the ones who look at the modifications to see if they 22 fit.into the picture, that kind of thing. 23 These are some things which,are going to be 24 covered a little later on. 25 The key assignment for these people to do PRA's
p * (), Q - 357 and so'forth, maintenance, human reliability, freqtiency and
-) 3 2
things of that nature. We do performance monitoring, such 3. as critical components. 4 You say what do you mean by critical components? 5 Once you do the PRA, you can crank out, take the top -- say 500 sequences. You don't always take the top ten sequences, j 6 7 because if you modified them, the top ten,may change a 8 little here and there, but you take the top 500 or g something. I don't care how you shake them around; you still 10 come out. They go up and down. 33 I don't care absolutely, but it's pretty well 12 . inclusive of all the major components. You can run an 13 34 availability model, or you can run a risk model. You can come out with a few hundred components per 15 unit that cover -- you know, give it tender, loving care. 16 You can guarantee a good availability. Those are-what I i 37 call.a critical component. 18 You can generate that list from your model. It's 39
.very g d for my people in operations, people in procedure, 20 i writing people and so on and so forth, so that they can use 21 1
it to make the decisions. 22 The failure cause -- you know how often it fails 23 and why it fails and compare it to the industry data. That ] 24 2 int rmati n is very useful for the maintenance and operation _.______.____m_ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ . _ , . _ _ _ . _ _ _ _ _ _ _
g o . . 358.
, . plant people to use.
.These are the kind of. lists that my plant people ave, the lead engineers who work with the rest of the 3
people. 4 These people spend about half of their lifetime 5 s mewhere along the line, too. 6 Economic risks and human reliability performance, j 7 this -is an area that is hard for me. -I haven't been able to g live up.to expectation. I haven't done much yet. We're 9 10 "9 * **
- You learn it; you do some. But we are on the ;
g limitation and utilization phase. Okay. PRA. First of all, we've got to recognize-g what it is, what it does. These four goals'should always -- g Ten years ago we set these four goals. It's not just safety issues that we came out with. g We are not there just to identify'cutliers or vulnerabilities. I think. ten years ago, that was before the 8
,g word " outlier" was invented.
But.that was not the only goal. We stepped higher. i than that. We have allocation of resources. That means ! reducing of costs. Availability, resolution'of licensing issues. So } you see the framework is,there. 1 I think Tom maybe is going to speak about it i i I
i
. i r*~g A V' U ,
f 359 i tomorrow. j 1
]
The first three balancing costs, generatiori and f safety are the most important. You have to work within the licensing framework, but not totally dominated by it. That's the idea. l 5 : That's the idea we had ten years ago, and it's still true today. Next, please. l PRA is -- or is not merely for those people who have not done a PRA before. So for those of you, be patient with me, please. The PRA is for information. It is an information 12 j source, knowledge; it is a living document. . It is never N,f v' 14 done -- The data changes. You cannot control that. The data will change; therefore, your model will. change. Your number will change. It's a management decision tool; that's what it is, not the only decision tool. ) It is not a cureall. Sometimes you wish to use it for excuse engineering, but it doesn't do it all. It's not a one-shot deal. 21 Next, please. Lessons learned, always learning something. It is not a simple analysis to perform. You can see tears in my eyes as I say that.
rm r, V Q , 360 j i The payoffs are short term and long term depending 2 on the ingenuity of the user. So you do not have to wait 3 until two years, three years or until you finish the PRA to 4 use it. 5 Actually the moment you get something, you already l 6 get the benefit because if you're smart, you're at the site. 7 As the operator you are learning; you are already on the { 8 positive side of it. 9 Large initial outlay of money,-that's true. When 10 you talk about nine, ten man years, ths.'s a lot of bucks, 11 but it's worth every dollar of it. Very cost beneficial. , i 12 Difficult to free operations.and maintenance personnel to 13 provide critical input.
~s iN 14 Man, I'll tell you, you need to take off your tie 15 and get them to talk to.you, earn their trust. They can is talk to you, and then they can talk until you learn to 17 really communicate with them.
18 You have to pay your dues. But that job is very, 19 very expensive, very useful. Those are the people you want 20 to talk to is TA's, you know, the maintenance guy. Gain 21 their trust; work with them. -It's highly interactive. 22 ' It creates high demands on overall organizations, 23 crose g all bnunds: engineering, operation, maintenance. 24 You go around. Technology continues to, evolve. How true. 25 Okay. We have a few recommendations. Minimize 1 4 _ ________ _ _ ___]
1 , t 1 y o
, 9 1 361
- ) 1 contractor involvement', especially on the detailed system,.
]i
'2 and event analysis. l I
3 I don't mean to take money out of the contractor's j i 4 mouth; that's not what I meant. i 5 I mean, you learn from them that you have got to 6 do your own work. You cannot have those guys turning the , j 7 system analysis;and all success criteria and say, "That 8 doesn't work." ! 9 You've got to put in your six man months of work. ) 10 or whatever, or you can do a limited scope, maybe six weeks. 11 But you've got to put in yourself. You're going to get out-12 what you put in. So you have got to have the pain. 13 Begin with the project with a reasonable plan; , s , 14 define' reasonable schedules and deliverables, but remain 15 flexible. 16 If you're so doggone rigid that you already know 17 everything, why do it. You're supposed to come up with a 18 few good loose ends. 19 - Have one individual responsible for the overall 20 project. Without such a person, your PRA can get out of 21 contro'1, so get a centralized guy: recognize that guy and 22 he's in charge. 9 23 Talk to other utilities who have done PRA's and 24 take advantage of lessons learned as much as possible. In 25 this business we have got to be in tune. D
--___L_-_____--__D_ - _ - _ _ _ _ _ - - _ a, b
(~ _ v + ! i 362 1 For example,.you might consider sending one of , l 2 your engineers to some college, check out a book with them, 3 kind of on-the-job training. It will benefit both 4 organizations. . 5 We started -- We sent an engineer to EPRI and ho l 6 came.back pretty nice. It's that kind of thing. Use some ! l
.7 ingenuity. j 8 You much involve plant personnel as much as 9 possible. .If possible, give your. team basic training first.
10 :Where you go to school is.your job. There are 11 different schools, you know. Keep upper management involved 12 as much as possible, but remember the end use of the PRA is
~
13 as a management decision tool.
"5 m -!i 14 I learned that the hard way. Sometimes it's very 15 difficult to get upper management truly involved. They sing 16 many songs, but I guess they maybe get burned or something 17 else, and they think something more so than the rest of us.
18 To be good PRA practitioners you have to havs the 19 ability to lose your memory. By that I mean when we got out 20 of college, we were good engineers, and then we learned 21 rules and regulations. 22 Then when we go back to there, we've got to be a 23 good engineer again. You've got to learn to forget some 24 rules and regulations or it just messes you up. You've got 25 to learn to be a good engineer again. L _ . - _ - _ _ . - - _ - _ _ - _ - - - . - - - _ _ _ . - . - - . . _ _ - _ . - . . ~ *O
,y 2
'i 1
i .
- n. , ,
v U m ' V - '
. 363
;D , 1
'1 Theres a. lesson to be learned, .and one of these~~
(j
'v .
2 days you'll find'out. 3 e one more slide. 1
~
'4 The changing of the guard. I sense a signal being 5 'sent, a series .of signals, . starting from thelsaf ety goals to q 6 the rules'of the reaction policy statements, and now'the' "
7 train is coming, you know. l 8 You. go from design basis - to operation ' basis, going. ! g from, say, qualitative analysis to quantitative analysis, 10 changing from safety, nonsafety concept tua overall total 13 plant. 12 . That is the total concept. You go from FSAR.to 13 IPE. There"are many questions on that,~and it's still
^
14 sensitive. 15 ' Appendix B, QA, now goes to the graded QA thing,
. 16 which is kind of nice, you know. Now, I don't quite agree.
17 with this thing about extending def.ense-in-depth. I think 18 more focus -- You've'got to let go of some things, so that ig you can get on with something better. 20 The total effect could.be, instead of spending 21
$500 here and $100 Lover here, for a total of $600, you might 22 want'to spend $300 over here, and do this $100 for a total 23 of maybe three or four hundred, which is even more defective 24 .than before. That is the idea I like better.
t 25 So I don't like the word " extend," I like " focus," D b , e s.
.. -- - --', e;-
3;:- 3 .
.,p,
, 't h W. l'- ,. '
"; y
,, L*
x . T '-. 8
<O -
O '
~
V ,< o: , , ,
,. ., 364
' ' ' i' 3
shi f ting , " or sornothing ._
, , 1 2 We're definitely going to graded QA., The_ safety-3 yrelated. inspection,'we' talked about that. And'the risk ' i
-4. based performance basis versus compliance base, type thing, 5 which is good. !
6 Okay. So there should be'more effective use of 7 our resources, a-8 Now, you see. , g , [ Laughter.]- 10 If you do it right, you don't blow the 33 opportunity. Thank you. 12 [ APPl ause.]- 13 MR. FLACK: 'Thank you.
.i,_ -:3 4 Our last presentation'will be given by Ed Burns 15 from Tenera. The title is " Containment Entry. Implementation 16 Guide." ,
'\
18 1 :- 19 20 . 21 22
.o 24 25 4
- J O O 365 1 MR. BURNS: Thank you. I appreciate this l l
2 opportunity to speak. I have some things that I'd like ] I 3 to share with you, hopefully get your. feedback and get l l 4 the Staff's feedcack on these things that have been bothering 1 5 me a little bit about documentation that's been presented 6 by the NRC Staff and where we stand technologically trying 7 to address the issues that are in front of us from 8 containment response standpoint. 9 The first thing, the outline. What I would 10 like to talk about is just briefly discuss what the purpose 11 is, define some of the bases for a background that I think 12 .I'd like to speak from, identify what kind of analyses 13 . methods I think are out there and identify to the Staff {}\ 14 some open items that I would like their response on, sort 15 of interactive here and I have another set of questions 16 that are attached to the handout that probably would be 17 useful to study outside this ceeting. 18 The purpose of the presentation is threefold. , 19 The presentation really is meant to stimulate discussion 20 both here at the Workshop and in the future on the topic
'21 of containment vulnerability analysis techniques.
22 By way of a backdrop to this complex topic,
.. )
23 I would like.to express my own personal opinion. While 24 as engineers we find it really desirable to strive for 25 completeness and substantial detail, nevertheless we are Heritage Reporting Corporation mu
O O 366 { q t I always limited by both budgets and resources. 7 Historically, one of the driving forces 3 of the IDCOR IPE, both the front end and the back end,
/
- 4 was to find plant unique vulnerabilities commensurate with 5 the Severe Accident Po' icy Statement.using a limited scope 6 analysis.
7 The goal was to optimize the expenditure 8 of resources while addressing those central issues as a 9 policy statemert. 10 This rationale was presented early on by 11 the IDCOR Steering Group to the NRC Staff and formed the 12 underlying basis for mtch of the follow-on IDCOR effort
']' j 13 developing methodology which was to maximize the benefit 14 per unit resourco expended.
15 What I'd like to do today is to pose in 16 this presentation whether there is an a?.ternative approach 17 to containment vulnerability assessment, which would meet 16 the latest NRC requests for additional detail and might 19 1 be acceptable to utilities from a resource expenditure 20 point of view. 21 The next slido is sort of the background 22 from which I'm working to try to tell you what information 23 base that I'm using. 24 There have been some BWR CET'a, containment 25 event trees, developed, although they haven't been -- not Heritogo Reporting Corporation
-___-__l_-_____-____-______. _ _ _ _ _ _ . __ _
( "*) * *"~
'c
, .,.m- ,
1 . 3 p pd - J .1 , V -
. }
E:g, 367 l' all of'them have'been published and reviewed. 2 The ones that I'm familiar with are three
+ :3 Mark I's, two Mark II's and a Mark III. 'I 4 There exists 'some experience at the utilities .]
. 5 to implement these and use them. In addition, there are, 6 detailed CET's that have been used by the NRC Staff or ,
7 their contractors in developing NUREG-1150. 8 Although they may not be pictures, they 9 'are in. fact containment event trees, at least conceptually. 10 Containment event trees have also been used 11 in looking at -- for BWR's anyway, for backfits, including 12 containment ventings and EPG evaluation, for extension
}< '13 . into the severe accident analysis range.
3 H' In addition, the CET's on the BWR's I have 15 -identified have included both IDCOR and NUREG-1150 draft 4 16 ideas that have been formulated in both research and analysis. 17 The approach that I think that is viable' I 18 and I think it's consistent with what the NRC Staff has 19 . identified includes the use of a containment event tree, 20 which to me is a picture of something that you can communicate
;l 21 with with a sheet o'f paper. 3 22 It incorporates all the relevant phenomena Q.)
23 and operator recovery actions and containment failure modes. 74 It's unique to each p2. ant. Each plant is different from 25 What I can tell. Every analysis that I have done says l i b ___ __ - ___ _ __ ___ _________ _ _ _ _ ___.__
-_ _ _ Sa _ Heritage Reporting Corporation cca
yogg 4:
,I '
L f
'(3 Q
- y. ; 368 y,; & ,
1 that, and11t allows you to perform plant specific calculations ; ,s :J 2 consistent with that structure. > 3 'It allows you to optimize the level ot' detail !
~!
~
4 both suffici6nt to capture the key events, but yet brief ' t-5 enough to provide e vehicle for communication. ]
. 6 Some of the things that one would like to 7 do is integrate.all of the knowledge, both from the GTCP, ,
- 8 MAPP and any other code for PWR that are out there. This 9 provides a vehicle to do that. !
10 It allows one to quantify the probabilities
!! of the functional events and address the uncertainties 12 through sensitivity analysis. 'us)-
13 . I think those are the features that have
)
14 been emphas1. zed by the Staff. 15 The next slide I'll just talk from briefly. l 14 It is meant to show a way that the front end gets married 17 with the back end, which is a key point that's been brought 18 up several times during this meeting, and that is the 1 19 communication be' tween the containment event tree and the 20 . front-end systems analyses. 21 It's dependent both on the binning scheme - 22 that's used to transfer the information and on the
-l 23 deterministic analysis that feeds both. -
. l 24 The front end is dependent on how you set j t
25 your timing and what codes you're using and what systems l J
, Heritage Reporting Corporation l
_______o____ __._ m"" __ , j
c ; ll ~~f;; ;
.,2 ,.
K
.D. o
)
1
.. 369 1
>. {
N, <l' you're using to mitigate and what dependencies that cause y 2 failures. j l 3
- j ,
All that information needs to be transferred j
'N j
'j - '4 to the containment event tree analysis.
1 5 Finally, what I think is visualized as the- l L, f and states for that containment' event tree anailysis,are 6 { i 7 7' wh;st the status of the containment is, what the sequence 1 8 titning is that got you to each state within the containment i 9 event treo and what the radionuclides' release. characteristics j
, 10 are, which there siaems to be some misinterpretation.nbout-83 whether it's just tiaing or timing and magnitude.
'- 12 The next t.;11de sh6ws what I think would ;
13 , be'a streamlined containment event tree, which I think 14 can accomplish many of these goals that the Staff has and-15 is' consistent with the IDCOR methodology, j 16 It allows ease of communication in' that 17 we're focus 6d on a few key issues and it can be both i 18 phenomenological and operator response. 19 I've seen tools like this used effectively. i
. 20 For example, in discussing the behefits associated with 21 venting, it can be treated in this sort of a framework.
22 The next slide emphasizes that while I
\. .) .
13 communication is a key, if the objectivos become more broad, l 1 24 if we seek to encompass more within our intent, if we try j I 25 to expand to encompass things like a containment performance i Hor!toge Repo;1ing Corporation j A m* _ x i
% :p ' ,
. ;J 9 ; 9 ,
p g 370 a ,; .,. 1 improvement program, if we seek to have enough flexibility ;
~to eventually encompass accident management and all.of 2 I 1
3 its shapes, manners and forms, and if we. seek to involve 4- generica unresolved safety issues, then it may be more 5 difficult to use a simplified model like that.. 6 So you can make the judgment that the 7 simplified model isn't adequate, if you expand the scope. j 8
'fhe next slide is something that I found 9 can be used to address all of those issues. It does have
, 10 a lot. It has, a problem associated with it that it'n more 11 expensive.
II It has the problem associated with it that j'{y 13 it's more difficult to look at at first blush, but it does j 14 ' have the capability to deal with the issues that have been - 15 raised. > I A6 It is quantifiable. The information exists. j i 1" For those issues that we can't quantify 18' precisely -- precisely I'm using very loosely.now -- wo- . 19 .can do, as the Staff has suggested, sensitivity analysis 20 to bury those noder. 21 . I should mention at this point both in the 22 simplified tree and in th'e more complex tree, for each 23 one.of these nodes we have to have a fault tree that describes 24 what' failure modes are effective in each of these cases. 2T Go for containment heat removal there may Heritage Reporting Corporation m- ,__
.. .1 7<
O O 371 l m
.c 1 be a number of ways'for which we can effect conta.Inment j u j
.) 2 heat removal, and we would have all of those failure modes I
.I 3 identified within a sub-fault tree that feeds into this 4 event tree, similar to ths front-end analysis, j
5 The kinds of things that are treated here 6 are both in-vessel mitigation, which I'd like to thank- ; 7 Paul Hill, because he has been an advocate for a long time, . 8 identifying that as one of the primary things that needs 9 to be considered in PRA's and has in the past been a disconnect l 10 between the front end and the back end. Where do you consider i 11- that. 12 He has done a lot in the industry to support
^
g 13 that information generation. 14 Combustible gas control, which is very importan 15 in BWR's, both in the Mark I's and II's, and in the III's 16 for diff6 rent reasons. 17 Water availability in terms of debris 18 coolability and containment failure modes, what kind of 19 failure modec and where they're located and what the size 20 is, 21 Those are key things that are all in this 22 more complex, more expensive kind of tree. 23 The last slide I'll put up is just sort 24 of summary conclusions which reiterate some of the things
- 25 that I tried to set out in my purpose.
Heritage Reporting Corporation '
.. . d c= *= . _ . l
O
~
O , 3,2 1 1 My opinion is 'that we can perform the analys.4.s 1 7 . 2 that : reasonable people would support as adequate to address 3 many'ini'tiators on a plant specific basis. l 4 In other words, what can you do to your 1 5 plant or what can you in real terms support as something 6 that you would do procedural-wise or harducre-wise to make 7 your plant more safe without compromising other issuca 8 that have to be addressed by the operator? 9 This can be accomplished either with the 10 streamlined event tree, I believe, or with the more detailed J 11 tree. 12 Finally, reporting criteria -- this is a
]: 13 . little bit of an aside and I think John Garrick mentioned 14 it earlier -- may be a little restrictive in that they 15 may delete some things that we would like to learn, although 16 most utilitias will carry along that information in terms 17 of numbers so that the information remains accessible to 18 them. This is whether the reporting criteria is anonymous.
19 I am a little apprehensive that the guidance 20 puts the utilities in a vulnerability position that this 21 coulu be a very resource-intensive thing if one werit to 22 the more complex event tree. 23 In terms of. questions, I have three questions , 24 that I'd like to go over right now. 25' For 1335, a question arose yesterday -- Heritage Reporting Corporation _ _ _ _ _ _- _ _ L** ! _ _ _ _ _
,- , q
, )
r w9 9~ 373
! I I' Well, let's do this one first.' The question that I have- t P. .. .
.. .. l 2 ' relative to 1335 is the reference made~to the issue resolution
'3 papers that.were performed' interactively between the NRC- l 1
4' 6taff and IDCOR, published in 1986. ) j 5 That information presents some question . q l 6 'and atobiguity about what is to be done today in 1989, given f
, i 7 the status of research that has occurred between now and'
]
8 then.
-t .
9' i I was wondering if there was a way to update 10 < that information;and provide more specific guidance in I I 11 three or four specific' issues for at least BWR's. I 1 12 DR. SPEIS: Do you want a response now? )
"' ~ , 13
. Go ahead and finish. .
14 MR. BURNS: Okay. That one I can leave 15 as an open question.
\
16 The other question revolves around a question , 17 that came up yesterday, and that was identifying ~ source
'18 term and' magnitude for those plants that may'not have the 15 ability or the resources to do plant specific analysis l 20 using available codes. ;
I 21 You suggested that there were references ! 22 available and I know that in 1335 there are references i (,) . 23 to 2104 and 2139, but-I don't believe -- our investigation's , 24 are that wb can't fill up the matrix. l q 23 In other words, if we follow the Staff I Heritage Reporting Corporailon L-. i __ _ _ - 13 coa s. _-. .J
o
- o. .
374 1 I 1 gaidance and do the more detailed containment event tree,
> 1 2 we can't fill in all of the end state source terms that 3 we would like to fill in for a comprehensive analysis.
4 I was wondering if you have other sources 5 that you are referring to that are available., publicly 6 t available., that we would have access to? 7 The lart item th.at I would like to discuss i 8 is one that was broughu up by Bill Parkinson yesterday 9 and that was with regard to Criterion No. 3 from Apper. dix 10 2 of the Generic Letter that any functional sequence that 11 has a core damage frequency greater than 10-6 and leads 12 to containment failure can result in a radioactive release 13 magnitude equal to BWR-3 or PWR-4. 14 If I put up the handwritten slide, the front 15 end yields a set of accident sequences that I can understand 16 lead to frequencies of below 10-6, 17 Whdn we enter the containment event tree, 18 as I understood the response to Bill yesterday was that 19 whenever we pass thrcugh a node in the containment event 20 tree that drops us below a frequency of 10-6, then we no ll longer have to consider the source term, either timing 22 or release magnitude for that sequence. 23 I was looking for clarification of whether 24 my' interpretation of that response is true or whether it's 25 not true. Heritage Reporting Corporation
- - _ _ _ _ _ - _ - . _ _ _ _ xa eca
%r+- %
375 , 1 The next slide is a sort of summary of what 2 I' concluded from the guidance, which was that from the 3 standpoint of deportability, one would be reporting th'ngs i 4' like sequences one, two and three which have core damage 5 frequencies above 10-6 and wouldn't necessarily be reporting ) . I 6 source term magnitudes for those, despite the fact -- well, l 7 I guess it's ambiguous to me whether you report the. source 1 8 terms or not, even though the~~ frequencies are very low. 9 The fourth case is the core damage frequency l 10 would be below the threshold number, even though the release l 11 may be large and that would not be a reportable sequence ) 12 under the criteria that I see. i 23 Those are the three questions that I thought 14 sort of could be addressed ~today. The other ones are included ] i 15 in the handout. l I 16 Thank you very.much. I 17 MR. FLACK: Thank you. That concludes this 18 portion of the program. Why don't we take a break, half i 19 an hour break, and we will be back and reconver.e at 3:35. 1 20 (Recess taken.] 21 DR. BECKNER: If you would take your seats, ' 22 it looks like we have what I guess looks like a quorum. 23 I want to remind everyone that it is our 24 plan and I think we've had a lot of people agreeing with 25 it. They would like to do it. We are going to try to Heritage Reporting Corporation mr m 4.
p- . .
- m L.) .
.G,-s 376 1 get finished by noon tomorrow.
2 Again, we've had a lot of time for questions 3 already and I think it will be possible to finish the meeting 4 by noon or shortly thereafter; anyway, before lunch. 5 The second thing, the last presenter, Mr. 6 Burns, is he here? Is he in the room? 7 [No response.] 8 DR. BECKNER: It's our intent to try to 9 summarize the response to his questions tomorrow morning. 10 The first thing, we do have a session where we are going 11 to try to summarize the answers to some of the questinos, 12 summarize what we heard today. 13 . That will happen first thing tomorrow morning. i 14 With that, basically the only thing that 15 we have scheduled for the remainder of today is again the 16 question-and-answer period. I think we're open now for q 17 questions about things that have been discussed today, 18 and I believe if you have any questions or comments about 19 some of the prepared talks today, that will be fair game, 20 too, I think. 21 S'o do we have any questions or comments, i 22 either one?
.23 MR. POWELL: Mike Powell, Houston Lighting
# 24 & Power.
25 I didn't have time to get in touch with Heritage' Reporting Corporation
<=n=== ,
O O 377 1 you ahead of time for a prepared speech, but I'd like to 2 pass along a.few comments, if I may. 3 We made a commitment several years ago, 4 proactive decision to start a Level 1 PRA. We're due to 5 complete that this month. It's actually a Level 1-plus. 6 I'll define that being we're identifying the containment 7 damage states and then proceeding with the containment 8 analysis. 9 We made a significant commitment to resources 10 and personnel. We currently have four personnel dedicated 11 in 'he t PRA group working on this. 12 The focus of my comments is really geared ((j(~ 13 towards what I consider an additional expenditure of resources , 14 As far as HL&P.is concerned, there's not a commensurate 15 gain in understanding of severe accidents. 16 We're essentially doing a Level 1. We're 17 doing the systemic approach that Dr. Garrick talked earlier. 18 We used a PL&G technology. We had a technology transfer. 19 We've had our people in their shop; also, 20 full time in our house working on it. We've been heavily 21 involved in the analysis. - 22 We've currently written the report to the
. :),
23 format that you've seen on other PRA's. It doesn't meet 24 the guidance of NUREG-1335. 25 My comment is to map our report into that Heritage Reporting Corporation w~ _ _ _ _ _ _ __
ry (m V \f 378
~
I format will be an extensive amount of administrative resources. 2 We feel.that those resources are better
\
3 spent on tech spec optimization studies, looking at other J 4 hardware fixes that we have in mind, and would like you 5 to' consider some other approaches for'those of us who have. 6 already started.. I 7 Dr. Garrick left an item on the table earlier l l 8 today about this systemic versus functional sequences and
) i 9 screening criteria. We'really need to get some further 10 clarification on those issues.
4 11 Most important of all, I think we would i 12 whole-heartedly endorse Dr. Garrick's comments that he 4
'Il
us 13
. made earlier.
^
14 Lastly, we have already included internal 15 flooding and external events in our PRA. We would encourage 16 the Staff to allow some flexibility. i 17 I'm not sure where it's going after this L 18 Workshop. We've heard a lot of pros and cons of where 19 it should take us.. . 20 We would prefer not to go back and redo } 21 any work in that area. We 've used techniques. and methodologies s 22 that you've already seen in the past and we would propose 23 that the final resolution include the option that those 24 of us that have already started, as long as it's consistent 25 with what you've seen in the past, that that's acceptable, f Heritage Reporting Corporation _ _ _ _ _ _ _ _ _ _ _ - - _ _ < ma" -_
- a. ,
?"
] ,
t] -
.379
.\
l That's all I'have.
- 2. 'DR. BECKNER: Thank you for your comments. -l 3' MR. BONNEAU: My ncme is Clif f- Boiineau from 4- Northern States Power Company.
5- I basically have three comments that I'd 6' like to make. The first one deals with the=8100' hours l 7- listed'in the Generic Letter. 8 We just completed the IPEM methodology 9 from Monticello on' Mark I containment and it required about to 10,000' hours of utiliby time and about $600,000 consultant ! 11- time to complete the IPEM with the IDCOR methodology within 12- the enhancements that are listed in the SER. ( }, 13 _ To includa the enhancements and to complete l 14 the process, I would say.it would take anywhere from 16,000 15 to 20,000 hours of utility time or total time to complete 16 that process. 17 The second comment I've got deals with the 1 18 IPEM itself. We completed the IPEM using the Appendix 19 D's and we feel that it's really an acceptable method to 20 respond to the Generic Letter concerns and would meet all ,
. - 21 of the objectives listed in the Generic Letter with that 22 methodology.
~.. ) .
. 23 To go on and to expand that methodology !
24 w
,ith the enhancements, particularly with increasing all t
25 the fault trees really isn't warranted. I think we get Heritage Reporting Corporation
,, m
l
,y , <
) l i 380
'I all of the vulnerabilities, all of the insights coming j 2 out of the current IDCOR methodology.
3 We feel that it's a very important tool 1 4 that we have now. We'll be going inte a full Le' vel 1 PRA, l l 5 but still to respond to the Generic Let,ter ob'jectives, f i 6 it seems to be an appropriate tool to use that. l l 7 The third comment l've got has to do with i i 8 the CPI, CPI's with the Mark I containment issues. 9 I do not believe that those are generic to issues and it's best to be handling those as part of the j 1 1 11 IPE response to say, "Do these really apply to your plant," j i 12 because I don't feel that they are generic. l I 13 It would be more appropriate to address
])
14 those in that forum. 15 Thank you. 16 DR. BECKNER: Thank you. Do we have any f 17 other people wanting to make comments or ast questions? 18 MR. TORRI: My nane is Fred Torri from PLG. 19 I wanted to address a couple of points that j 20 were brought out by the presentation on containment event 21 tree methods before the break, because I think that that 22 discussion has brought out a couple of issues that have n.} 23 traditionally plagued and limited the usefulness of Level 24 2 PRA's. ! 25 The issue I'm trying to address is that Heritogo Reporting Corporation f?try 4744888
y j ,
~ 'n' p ,
O, ' O l e+ . 381
- l. 1 'I-believe th'at until we bite the bullet and put ai.1 l t.
i 2 the containment systems up into the plant model, whether l l 3 that's a linked event tree approach or a linked fault tree l j'
- l 4 ' approach, we will really not have the tools by which we ,
[ ( 5 can correctly trace the dependencies, the shared dependencies j
.i, 6 between containment systems and plant systems. 'I 7 In other words, they depend or the same -
8 electric power support systems and on the same fluid cooling, ) 9 ' secondary cooling or component cooling support systems. 10 Coming out of the plant model without knowing I 11 the. status of'the containment systems, we will not know 'i l 12 whether those containment systems have those functions 13 available. f..s) <
-14 Once you bite that bullet and put all there 15 containment systems up into the plant model as part of 16' the quantification of the plant model, the containment 17 event tree becomes simply a question of addressing j 18 phenomenological issues and containment recovery actions.
19 The containment event tree then becomes 20 quite simple and straightforward and the split fractions 21 become a matter of estimating our ' decree of confidence, ) 22 whether the calculations we use to support those split , l 23 fractions, whether they be generic or plant specific, are j
- i 24 predicting the correct outcome for the' containment event
-25 tree branch that we are considering under the plant damage l
I Heritage Reporting Corporation _ _ _ _ - _ - - _ _ _ _ _ - - - _ - - _ - - - ccaa x a
( 382 j
- c 1 state condition'which we are addressing.
C 1 2 The impression that what's left of a
,, a 3 containment analysis is an expensive and time-consuming a
t1 '
"4 effort', under the boundary condition that the Staff has 5 defined,'namely you don't need to do plant specific analysis g
6 unless you find it necessary, it's not correct. That's Y not a very large effort. That's typically a small fraction
~
8 of the Level 1 effort, in our experience, and that's been o i 9 more than one application that we have had that experience. 10 The second point that I would like to address i 11 is that if there are specific plant specific analyses that 12 you do consider, we have found the usefulness of a plant ; i 13 specific containment structural analysis to be quite [}): ^ 14- significant, if you are trying to address leak before break 15 issues. 16 I do not believe that. leak before break 17 issues cansbeladdress with generic calculations for'another 18 / containment because'they depend oli large strains, large 19 deformations and interferences that result from those l 20 deformations. 21 If you do not wish to address leak before l 22 break issues, the use of' generic calculations might be 1
- i 23 quite appropriate.
24 Thank you. 25 DR. BECKNER: Thank you. Heritage Reporting Corporation _ ____ . - - - acD
l p) q ^ . Q) ' 383 I MR.'DONNELLY: My'name is Pat Donnelly from -l l' ' 2 Consumers: Power Company and I would like to reinforce the 3 statement that the gentleman from Florida Power _made this 4 morning in regards to 50.59 reviews of the severe. accident i j 5 management program, the new Generic Letter that comes out. 6 I have a little experience at Big Rock'in
- 7 dealing with plants that aren't necessarily like everybody 1
8 else's plant, and when the EOP's came out, it was very f 1 9 .hard for me to get our PRC to do 50.59 type reviews without f
. i 10 allot of discussion and even meetings in Washington with !
i 11 the Staff, i 12 So any guidance that you can give plant ! i 13 (]) members, the people that are making those lower level 50.59 14 review Staff decisions will be beneficial. 15 Also, under the same severe accident management 16 type-procedure, any time that you would use instrumentation I 17 that's outside the design basis, the EEQ question'comes 1 4
. . 1 18 staring at you. ]
19 I would appreciate it if you would give. 20 that some consideration from the enforcement standpoint. ; I l 21 Thank you. j s
, . ,, 22 DR. BECKNER: Thank you. Do we have anybody i 23 else who wants to ask questions or make a comment? 1 1
24 MR. HOUSTON: I would like to try a brief ) l 25 responne to a part of a qu'estion or a comment raised a l l l Heritage Reporting Corporation j a-
.x ,
q ("T t) , (7.:
%)
j 384
\J j l few moments ago with reference to what we have. called the 2 Containment Performance Improvement Program. j i
3 At least in part,.if I understood the comment
' l
< 4 correctly, it may have to do with an interpretation of j il 5' the use of the word " generic." 4 f
1 6 The first thing,I would like to point out . 7 about the CPI Program is the word generic applies to the l 8 term vulnerabilities. i 9 We recognize that if there are what we sometimer i 10 call fixes or modif tcations t hat are proposed or recohanended, 11 they have to be done in a very plant' specific manner, but 12 the vulnerabilities that we're looking at in the CPI Program ( ;i 13 are generic. 4 14 Now, let me explain what the word " generic"~ j 15 means in that context. Formally within the proc.edures 16 that we use within the Staff of the NRC defines generic 17 as something that is common to two or more plants. 18 That is the criterion for our famous CRGR 19 review process. That is, if we deal with just one plant-20 specific unique problem and propose a plant specific backfit, 21 it does not have to go to our Committee for the Review 22 of Generic Requirements. r - 23 That's an internal formalism. l 24 In the context of the CPI Program, the word j 25 " generic" is taken to mean a number of plants that have Heritage Reporiing Corporation r,- .
sg:, ga . o; - co . S' - 385e l j p, ; Wp 1< - common characteristics. { f;p . 7
;2 So the-wordL" generic,"'for example, with' '
lu 3 ' respect to'the Mark I, or the vulnerabilities that we j' , 4' .have' analyzed an'd dealt with for the MarkLI plants..are-
~
.1 i
.in fact generic vulnerabilities to all'of.'the plants. .The1 5
6 fixes, if any, may be quite.Ldifferent forfdifferent' plants. L
~
7 The same thing is true-for the ice condenser 8 plants. That is, there's a commonality.:of' basic design .: 9 features. 10 So we do use-the word " generic" in.a sense 11' other than generic as common to all, plants.
, 12 It just seemed to'me that the comment or-GA.s 13 the question may have understood a somewhat different
, 14 .- definition of the term.
j 15 MR..BECKNER: I guess if there'.s no further -i q 16 comments'or questions, I.think we'll adjourn earlf today. I 17' If Mr. Burns is here, I think we should talk to him before 18 he.I. eaves. j 1 19 - -Thank you. -
.i 20 AWhereupon, at 4:00 p.m., the Workshop was i
2'l adjourned, to reconvene at 9:00 a.m., Thursday, March 2, 22 1989, in the same place./
,. s3 l 'Q/
23- --- 24 25 - Heritage Reporting Corporation ! a _ _ :____ __ _ _ _ = __ - rw~m
. r.
, i i
O) ,
\ . v 4
1 .C E R T I F I C A T E, 2 + 3 I hereby certify that this is a 'transciript of the
. . 4 4 Independent Plant Examination Workshop held by the !
5 U.S. NUCLEAR REGULATORY COMMISSION on March 1, 1989, 6 and that this is a full and correct transcription of the i 7 proceedings. 8 Place: Forth Morth,, Texas 9 10 q ui. p G,,
/
y? /_ n o l l GayE.pentoft, Reporter 12 This is the original transcript for the file of the United 13 States Nuclear Regulator Commission taken stenographically j v i 14 by me and, thereafter reduced to typewriting by me or under 15 the direction of the court reporting company. l 16 l 17 18 4 l 19 20 21 , f
, 22
. l 23 24 25 .
Heritage Reporting Corporation _ _ _ _ _ _. _---__ _-_.--_ _ - _ _ _ ("'""*"_ _ _ .. . . . j
- e :
.. m Ii> j 3 0 j o 3 ]'
4 1 ( '
,+ ,q l
f OGLY % . ( y\ \
% s" 6"" ' d~'2 l ,
k Twhbd 1 E \hk ' E%4{whi% 4 McAcshop M u sex (p h mb b 7t%. i m 1 i k s h d 'ot . N s\ 3ere . \ gA%ud (S execAd. t
% n sq j h.$9ad 1 N bb a
m-mmu____ __._m_.. _ _ m _}}