ML20148G923
| ML20148G923 | |
| Person / Time | |
|---|---|
| Issue date: | 08/25/1978 |
| From: | Advisory Committee on Reactor Safeguards |
| To: | NRC COMMISSION (OCM) |
| References | |
| ACRS-218, NUDOCS 7811130298 | |
| Download: ML20148G923 (31) | |
Text
_ _ _ _ _ _ _ - _
. o In its report to Chairman Hendrie on ACRS Activities:
May-August 1978; dated August 25, 1978 the Committee stated:
Dynamic Loading Combinations The Committee has attempted on several recent occasions, including its 218th meeting, June 1-2, 1978, to encourage the Office of Nuclear Reactor Regulation to reconsider the rationale for establishing design basis loadings and loading combinations in performing safety analyses. The ACRS recommends that such a reevaluation be under-taken as soon as possible.
l 7811130 AN
i ;
. - a W J, atB"' AC. ftS Idecling _
1 35 R . 10 0::*. "ill 'hc f c- u. . , ~ ;. . .
.. 'M-r^ ?
rp/fm1 2
CR7631 Will that be an acceptable solution or not?
3 MR. RUSSELL: The venting is a asible solution.
4 If tt.ey demonstrate the material comp bility of the B,C 5
in the spent fuel pool environme . The canning 6
was initially put there to p tect the B4C material. Even 7 thouch there has been s
/
surveillance ongoing for the last year 8 or se and some tes g, which does not at this time indicate 9 a problem, th e- aff has not completed its review of the 10
/
,N acceptabi .ty of B4 C material in a vented condition in de spe.n f 11 fuel ol.
We have that under review. There are other designs 12 4^
- - ewpos. ., ti . L .~ . waw om ../. . c t :pywveu.
13 DR. LAWROSKI: Next topic is that by Mr. Mattson on .hs 14 basis of combining seismic and other dynamic loads. How long 15 do you estimate your presentation will take?
16 MR. MATTSON: I hope about 10 minutes, Mr. Chairman.
17 Your agenda, I think, shows something like an hour 18 for this discussion.
19 I am not prepared to talk for an hour, so I hope 20 your questions don't run for that length of time.
21 Back in January the Committee expressed an interest 22 in hearing generally about the question of load t
23 combination. I appeared here in February and said that we 2d tefedered Reporters, Inc.
expected the best way to come at that rather broad question 25 was to try to write a white paper about where, how and why 6
,1
- , )
ka,J.
- 312 fm2 1 we combined loads. ,
}
2 I came down to talk to you in May to give you a 3
, status report on where we were on that endeavor. Time was k) 4 short, so we didn't get that.
5 I am here to give you essentially the same thing 4
6 today. The message I have, I am not sure we. are coming at the 7 problem in the most efficient manner. I am a little short on 8 resources at this point in time. I have got a lot of activities 9 going on in which iced chinations are an important element.
10 I wouln like to pose for your consideration today Il that we treat those particular pieces of our work as they bear i
12 on the load combination question, rather than restructuring 13 our work to answer your general. question on load combination.
' ~
Id Don't expect an answer now.
15 What I want to do is walk through the various things 16 going on in the Staff in the generic technical activities 17 program, current licensing activities, topical report reviews, 18 things of that nature, to show you where you are interacting 19 with us now on this question, and then .ptuse and see if we 20 can't reach a collegial:, judgment.that may be the way we 21 are handling it piece by piece is equivalent, that will give 22 us the same ansvers you were looking for broadly.
23 Before I list those pieces, I would like to respond 24 in an oral way to the questions you asked, rather than the Ace Federal , Reporters, Inc.
25 written way I promised a couple of months ago, to at least give l - . . . . . . . . . .
- - ~
' '[ V. -
e i ,
, ,-L ,6
l 'i-l am 313 I
- fm3 1 us the same basis in terminology and hopefully the same philo-l 2 sophical understanding of the problem we are addressing.
3 The first question you as'ed k was for us to describe
- (,)
4 the rationale by which various dynamic loads are or are not 5 combined.
6 Let me speak to the policy or philosophical reasons 7 for combining loads. As the people on the Staff talk among 8 themselves and read documents from the past, and interpretatien 9 of general design criterion 2 as they exist today, there seen 10 to be two reasons for combining loads. -
11 One I will characterize as the need to provide some 12 measure of margin to failure of equipment and structures. Thet k_) 13 is, the requirement in general design criterion 2 to combine i 14 loads appropriately, in the words of the general design cri-15 terion. i 16 Some people say and some document support were pro-17 vided to provide a qualitative measure of margin to failure.
18 There is another school of though that says, the 19 second reason for combining loads. The second reason I will 20 characterize with some shorthand words by saying people were 21 concerned in the days that general design criterion 2 was
)
22 issued about the undetected flaw causing a failure during the 23 design earthquake, leading to the loss-of-coolant accident.
t 24 Am.m. eda.oo,wn, inc.
That is, a concern with the possible simultaneity of events.
25 I would like to pause and note, simultaneity of events A______-_._____
- o
,, , vd [
, 314 fm4 1 is another interesting way to come at the combined loads i
2 question. AF we have looked at your question and talked 3
among ourselves, we realize that we combine loads by combining b--
4 events and sometimes we combine events by combining loads.
5 An example of the former, we combine events which 6
leads us to combine at least conditions. That example would 7
be the requirement that the equipment designed to mitigate the 8
fuel handling accident in the auxiliary building must be 9
safety grade equipment and the definition of safety grade ceans to in part that it has to have seismic qualifications. So, we have 11 combined events in that case by a;rather, indirect route, which
,12. leads to a combination of severe -conditions for equipment.
U 13 That is the design of that equipment.
14 On the other hand, the example where we combine 15 loads, which, in effect, leads to a combination of events, is 16 the LOCA, plus SSE. One might argue that the reason you combine 17 LOCA and SSE loads is to provide margin- to failure for the
, 18 safety grade equipment, but as I explained earlier that leads 19 to a presumption that, in fact, what you are doing is com-20 bining events, the seismic events and the LOCA which some peop1.e 21 say, yes, that is the reason we required them to be combined in Q 22 the beginning, in the undetected flaw argument.
23 So, with those two philosophical policy level reasons, 24 a r.s.ru n.oomr., inc.
we combine loads and we combine events. When we combine them, 25 we have done various, we have accepted various ways of combining 4 ,
e l
i
,,,,..-----m
, a 1
a tL 315 fm5 I loads. Combining loads is a rather long way to say it, too, 2 but we do, when we are looking at the structural capability e 3 of a piece of equipment, we are comb'ining responses. What we
, 4.)
4 have arrived at today is a position that says, when you can 5 show a time-phasing relationship between multiple dynamic loads, l
6 you can use that time-phasing reistionship in the combination 7 of those loads, which is essentially the same thing as saying, 8 if you have a basis for separating them you need to combine I
j 9 them. If you have no basis for separating the effects, and to . that is usually the case with rapidly varying dynamic' loads
, 11 when you combine the SSE and the LOCA, then you must super -
12 impose the loads in time and decide upon a mechanism for adding O 13 the peak. One way that is clea'rly acceptable to us is the 14 absolute or linear summation of the peaks.
15 That is, the peak responses of the equipment.
16 The people that proposed, and those people are 17 numerous, most of the industry, that other methods of combinatic:
, 18 provide appropriate margin, borrowing on the words of the 19 general design criterion 2, one such method is sum root sum 20 of the square, SRSS.
21 The only proposal for the use of SRSS are those O 22 instances where a time-phasing relationship cannot be shown 23 between the loads, or among, if they are multiple in nature,
- 24 and'when the loads are rapidly varyin9 in nature, wasas neoo,m. inc.
25 Now, General Electric, I guess, is about one step
~~ ~~ . . . . . . . . . . . . . . . . . - - - - -- . -
j-1 L 4
- ~ - _, j 316 fm6 1 ahead of everybody else in arguing the case for SRSS.
2
~
~ - I"think the last time I was here, we noted that a 3 number of you hive' heard the General Electric presentations O 4 on SRSS . I will get back to that in a moment.
5 ' "But it is safe to say that the arguments are not 6 physical in nature. They are highly mathematical in nature.
7 I think if you do as I have done, which is, go back to my some--
8 what advance textbooks on dynamic loads you will find that i 9 it's been traditional engineering practice for some time to 10 say that if you have a test of how rapidly varying superimposed 11 dynamic loads actually add on one another, then you must either 12 use absolute summation of the peaks or make some space argumeat, O is thet is, how certain do you ant to de that they den t ebse1ute .,
14 That is the essence of the General Electric presenta-15 tion as it stands today. That is, if you combine rapidly 16 varying dynamie loads of the following nature by SRSS then 17 there is some probability, which is the minimal analysis, which 18 is amenable to analysis, that you will exceed the combined 19 ievel by a certain amount, and it is possible to relate
(
1 2') the exceedance to the characterized margin in structures, and 21 derive a probability of exceeding the capability,~ structural
~ '- -
22 capability of the equipment.
-73 + - '" It is the essence of the CE argument. _.
9 bd ~~J ' ' ~ ~
0*;Others liave made it. Westin9 house ~ har taado it .
dNes W.o mei, in c.
We are not'as f5r'afong in the review. De ct .
- y,s. ta-1 king 25
)
a v
( )
_o kJ J
. 317 fm7 1 on the subject but not quite makin'g the same argument. I 2 Vill drop that discussion and come back to it in a minute.
' 3 DR. SIESS: Is that a different argument than the one
/) 4 for using SRSS in combining modal responses and things we do 5 elsewhere in the process?
6 MR. MATTSON: I have looked at it more costly and 7 become more sophisticated in understanding exactly what we 8 get when we combine things that way. It is essentially the 9 same na:hematical argument. That is, randomly varying dynamic to loads which you can't physically relate in time to one another.
- 11 You know, they are occurring at the same time, so 12 what's the likelihood of exceedance of various loads, arrived
(). 13 at by various ways of combining them, but we do use SRSS in tae 14 way you describe.
15 I don't think, as I understand the history of 16 how we got to that point for combining the modal responses 17 that we looked at it in as much detail as we have looked at end 18 18 this.
19 20 21 0 -
23 24 MFederal Reporters. Irm.
25
~
3 l
L
y ,
!.j 'h e
nj [
' J18 cR' 7631 FR,\NK 19 g DR. SIESS : Well, it has been used in tho
- h 1 2
structural field for many years now.
3 MR. IRTTSON: Yes. We in regulation, not we in 4
engineering.
5 I want to mention, there were three other questions 6
you c.sked in a subsequent letter in February, having to do 7
with the single-failure criterion and its relationship to the 8
combination of multiple dynamic 1 cads. We are not quite sure 9
we understand that question, but we think you must be referring 10 to the somewhat novel approach to the use of non-safety grade 11 equipment to meet the single-f ailure terion in the event 12 of a main team line break outside ainment. Where under I ) 13 i the traditional approaches i .- atainment one would say 1
- 14 all mitigating equipment fc: ' ass of coolant accident 15 must be seismically qualified, whereas outside the containment 16 for the main steam line break, we have said the same thing .
But when you add the single-f ailure criterion, if 18
', the safety grade equipment is not suf ficient, then you may 19 rely upon non-safety grade equipment to perform its normal
! 20 j funcation. That is, assume it han survived.
}
That subject has been discussed at some length with
~
22 j d you, as I understand it, a year or so ago in connection with 23 the safety issues raised by members of the staff.
24
- 4. sde non,,,, %, Basically, the single-f ailure criterion is a 25 measure of reliability of equipment, and the combined loadings i
IlI
J Ub 319 ch 2 ,
1 requirement is a requirement which speaks to the overall 2
goodness of that equipment. So one says, this is my design 3
basis; and the single-failure criterion, on the other hand, 4
says, even if it is my design basis, I know that there is some 5
unreliability of equipment, no matter what its design, and I 6
will account for that unreliability for the single-failure 7
critorion.
8 If that answers your question, fine. If not, I need some further elucidation.
10 Finally, the use of SRSS instead of absolute sum.
11 I will talk in a minute about a more definitive proposal for 12 speaking to that question.
O n A related question is why we assume that the pri. wry
, system fails during a seismic event. That is, we combino LOCA 15 and seismic, but that other category 1 equipment functions. I 16 think the best answer I can give to that I have just spoken to.
17 That is, that dichotomy in the logic people say has been 18 recognized from the beginning, but it is the very dichotomy that is the reason for combining in some people's minds.
O That is the undetected flaw in the otherwise seismically 21 designed and capable equipment.
22 Q And, of course, the weakness to that is, if there is o
an undetected flaw in one piece of equipment, why isn't there w%i n.pormi, , , in multiple pieces of equipment? And tho answer from a deterministic standpoint is, one flaw is f ar enough, although i
f
'4 g
.T M
di 3 320 I
you won't see that written down anywhere.
2 MR. BENDER: If it is all right to deal with it in 3
[]) one nystem but no in another -- you are arguing you are 4
going to be concerned about it in the mechanical systems but 5
not in the electrical systems?
6 DR. SIESS : Deterministic rather than a random- flaw.
7 MR. MATTSON: I was thinking of it differently.
8 I was thinking of it mechanical versus mechanical.
9 MR. BENDER: If you qualify the electrical systems 10 seismically, you don' t assume that it fails. You don't hold 11 .
the same argument when you do it mechanically.
12 MR. MATTSON: Yes. The basic reason underlying that is that the failure of the mechar.ical equipment in the 14 primary system leads directly to the loss of coolant accident, 15 whereas the failure of the electrical equipmcnt because of 16 the undetected flaw does not lead directly to the loss of 17 coolant accident.
18 MR. BENDER: I can hardly accept --
19 MR. MATTSON: That is the traditional argument for 20 that dichotomy. Whe ther or not it s tands the test in today's 21 light, that is another question.
(h You asked the rationale. The rationale is, we 23 have been able to restructure, as we have been able to -- is 24 ,
, Aa m .inswe.,. w. along those lines.
25 DR. SIESS : You put the flaw where you think it will I
e e
t
. - ,[
- ch 4 321 1
do the most harm in the system that will cause the LOCA.
2 MR. MATTSON: Yes, -- --- -- -- -
Q 4
'~
DR. SIESS: But you really haven't examined the whole system to see whether that is truly the place it will
~
5 do the most harm. r 6
MR. MATTSON: That is true also. Judgment, made 7
some years ago and carried through, through the years without 8
testing. - - ---
9 Let me briefly list some of the related staff 10 activities in which the question of load combinations is an 11 important part and tell you what is going on and when you will 12 begin to see that work. -
Probably the most important and near-term related 14 areas is generica technical activity A(2) , which has to do with 15 asymmetric loads for pressurized water reactors. You have 16 seen that task action plan some months ago. To refresh your 17 memory, it is the North Anna asymmetric loads problem that 18 you have seen addressed, or will soon, on Diablo Canyon, or.
19 Hatch, 'on other pressurized water reactors. -
20
' Hatch is not a pressurized water reactor . It is a 21 hoiling water reactor. In the Hatch revlaw within the last several months we have shown that there can be an external 23 asymmetric load question due to a break in the cavity, causing 24 Lask.JA. pod.n,ik. an 'oveiturning' moment on the vessel. That question was 25 addressed ori'thel hatch docket. '~ ': - ~C~ M --
. <t , ,
e
- 0 e n .. _ _ _ _ _ _ _ _ _ _ _ - _ _ _ - - - - - - - - - - - - - - - - - - - - - - - - -
_2 ;
322 ch 5 1
It has not reached you yet, because the SER is 2
just being printed. I will cal'1 your attention to that when 3
it comes down here.
(I a A(2), treating the asymmetric loads question, has 5
as a part of it the question of, how do you combine the SSE 6
loads and the asymmetric blowdown load within the cavity?
7 We have applied our interim requirements in that required for 8
plancs now in the licensing process. That is, absolute 9
summation of the peaks for those pieces of equipment that 10 don': pass that test, or meet that criterion.
11 The appli. cant can do an SRSS calculation and then 12 demonstrate margins of failures for each component, component
()I 13 by component, essentially what we have been doing over the 14 last year.
15 Now, the generic task action plan is scheduled to
< 16 be completed in about two years. That includes a lot of 17 reanalysis for operating plants and the approval of a number 18 of generic models for use in the .CP and OL licensing process.
19 one of the important first decisions to be made 20 in that task action plan ir, what methods will we use to com-21 bine rapidly varying dynamic load, SSE plus asymmetric cavity
()) loads? There has been a joint task force formed between the 23 Division of Operating Reactors and the Division of Reactor D
24 Ac. r.s.,w pm,,.n, ine, Safety, a task force of about five people, representing 25 structural design expertise. and equipment design expertise.
g S 1
8 f
- ch.6 1
They are due to finish their work around mid-July, and one 2
would expect it to be reported and available for discussion 3
and consideration by management and you all perhaps in August.
. O a Now, I want to move from that generic piece of 5
work to another --
6 DR. ISBIN: Were you inferring these were in-house?
7 MR. MATTSON: All in-house . General Electric 8
came to us in a related but separate generic activity for which 9
we have no category A number at about midwinter, to tell us of 10 a major reanalysis program at the General Electric Company 11 for all of the Mark II and Mark III facilities.
12 This reanalysis was designed to demonstrate that al-( thocgh the equipment in the Mark IIs and Mark IIIs was 14 within the scope of GE, supply had not been designed to 15 accommodate certain feedback loads from the pressure suppression 16 pool through the rest of the structure in the facility, 17 feedback loads arriving with SRV loads or with LOCA loads, 18 that even when those feedback loads were concerned, the 19 ,
structural capability met appropriate regulatory criteria.
20 It's a major, thorough-going reanalysis by General 21 Electric which is under way now for the lead Mark II plant, 22
[g Shoreham, Zimmer, LaSalle, NPPSS, and will eventually turn 23 to the Mark III plants as they come back in for the operating 24
- Aa1.o.,w n.m,,,,,, inc.
license review.
25 GE's proposal -- and we have essentially agreed with I
{'
, b0
.. 324 ch'7 1
the proposal E- is that we not look simply at the method of
~
2 combining loads to adjudge the appropriateness'of' the con-3 That is, servatism in the dynamic capability of the f acility.
)
we ought to look at some of the other assumptions that go 5
into the 'overa11' analysis .
6 That gives us a way of putting in context the 7
conservatism required for the load combination method with 8
some other important parameters : the method for calculating 9
a mass and energy release; the method for calculating the I 10
' cavity pressure, caviry pressure distribution, the load 11
' acceptance criteria, that is f aulted or emergency criteria 12 under the ASME code , and the definition of SRV loads.
(3) is That one is worth pausing on. GE has proposed, 14 and we are proceeding with, a review of a method for 15 demonstrating the time phase relationship of the dynamic 16 loads occurring within the pressure suppression pool. That 17 is something that will be occurring over a year or two-year
, 18 period consistent with the Shoreham, LaSalle, WPPSS reviews.
19 It is important to know it is going on. It is also 20 important to recognize, if you try to adjudge conservatisc overall, there's some of those pieces of conservatism that 21 are f ar out in time' relative to positions people would -like to
() 23 ~ ' '
~
~" ' u = ; '" '
make on ' S RSS . ' ~~~ '
' '24 '-: ~^-a.
'For example, if you wait for CE to decide the Aasween.porm ,im.
35 question.f6r Mahk IIh and'IIIs, than ! don t have~ an ' answer
-- ' 8
~~
P9 7
A s
325 -
ch 8 i for the asymmetric loads question, which has some higher urgency to it because of its effect on operating pl' ants.
Q ---
- An added dimension to the GE program for reanalysis 4
of the Mark IIs and Mark IIIs, we usually think combining of 5
loads is an important consideration for the SSE and the big 6
LOCA and that we are dealing with very low-probability, low-7 risk phenomena. In the CE reanalysis we have a little bit 8
different wrinkle. That wrinkle is, the SRV loads plus 9
the OBE loads, safety relief valvo, plus the operating basis 10 earthquake when combined have difficulty meeting the upset 11 conditions for the ASME code. So there 12 .
So there is a considerable bone of contention between the staff now and CE on the method of combinati;on 14 of SRV and OBE loads and the acceptance criteria for those loads.
15 Another related generic activity is the topical 16 report --
17 DR. ISBIN : You mean these arguments carry over 18 to the Mark I?
19 MR. MATTSON: The answer is, in most people's 20 judgment at this ti.me, no. In other wrods, the couplings 21 between the Mark I pressure suppression design and the rest of 0 the system 1s not as eight a coup 11ng. It is not as rigid 23 a: sys tem.
2d 4.% nm.,,, we, Jir6; d6'you want to say more about that question?
'" A W 4 .
.; - MR. ISRAE L: No.
+py
. i L__- _ _ _ __ _ -
a 326 CR'7631 Frank #20 1 DR. ISBIN: This was a point Bridenbough made dag - 1 2 some time ago. I thought it was answered in the negative.
3 MR. MATTSON: I am answered it negative, too. We O 4 have a little information on the nature-appeared cause of the 5 feedback loads and a little better information frorc GE than 6 we had in the past about the hangars, snubbers, the syetems 7 that are of concern in the Mark 2, Mark 3 system and thare are 8 good arguments for why the coupling, being fought as direct 9 and rigid in the Mark 1 System, isn't lead to the sara difficult.
,' 10 DR. LAWROSKI: Any comment relative to the Diablo 11 Canyon Plant?
12 MR. MATTSON: What we have done there is to apply 13 the interim approach which we feel we have a very defensible 14 basis of safety for. We might have some difficulty defend it's 15 the mcst efficient and cost effective way to proceed, but 16 absolute summation was required in the case of Diablo Canyon for 17 the combining of the SSE and LOCA lode.
18 Asymmetric lodes were treated in s'eme detail, a la 19 North Anna, with approved Westinghouse evaluation models.and 20 indep2ndent audit calculations using our own computer codes 21 within the staff.
ggg 22 MR. BEND: Do you suppose you will ever write down 23 these bases so someone can look at them in a collective way 24 and try to make some sense out of them? Because most of what Aa%=w nepon.,,, inc.
25 I heard today is just verbage. It didn't really address the
.m
. I I
w-______ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
wYA a
327 CR 7631 i dag - 2 1 question. I suggest there is no logic to what. you have done 2 up to now, you are inventing logic. -- -
~~. 8 3 MR. MATTSON: No. I think it says there was logic, O 4 that it was written down then. It's a little bit difficult
$ because it wasn't written down to resurrect it. We think we can 6 resurrect it. As we do it, some of it appears to be overly 7 conserv,ative and arbitrary and there are efforts ongoing to 8 put reason and rationale into what we do. ..
9 My message is, thero are so many activities ongoing 10 that if I try to overlay yet ancther one, I haven't the resources 1
11 to handle it. I would prefer to go on with the current approache:
12 under the Generic Technical Activities Program, related matters (h 13 in the Generic Task A24 on environmental qualification on 14 Class 1E Electrical Equipment, where we are looking at sequential 15 versus simultaneous testing of electrical equipment. Are there 16 differences, are there failure mechanisms that we don't see with 17 our sequential testing, to be concerned about the combined 18 events?
19 MR. BEND: I think you are hiding a sentimental 20 question behind a maze of detail. You ought to go back and 21 start thinking about the logic more, rather than doing this 22 infinitum perfect of detailed computatiora than you still.can't 23 match up the logic because they attack the problems in umpteen t
24 different ways, with upmteen different people and-a; total bill AceJed al n m ,ws,ina, 25 of. system. evaluation would encompass a lot, ,___... ,,
5
( I
a 328 CR 7631 ,
dag - 3 '
.) MR. MATTSON: I hear what you are saying, but I am 2 not so sure I understand exactly what you are proposing. Can't l
l 3 you turn that into a proposal, how w6uld you do it?'
i .
o 4 MR. BEND: I am suggesting you go back and look at 5 the basis on which you are deciding what things can cause a 6 system to fail and the logic associated with arguments that say 7 in one case the earthquakes will not, for example, cause 8 something to fail. In another case, causing the item to fai1 9 when both are seismically qualified.
10 There is just no logic to that. Unless you are 11 going to assume that the earthquake is going to fai1 everything 12 whether .it's qualified or not. or you are going to assume if O i3 the item had eeen que11fied for an earehauexe, te wou1dn't f a11 14 under those conditions.
15 I just can't see dealing with it in two different 16 ways. I don't care what the piece of hardware'is. It seems to 17 me that that is what you are doing. You are going through all 18 this effort to qualify equipment seismically, then you are 19 sayi ; there are some equipment will still fail seismically.
20 If that is the case, a-1 of it will and you might as 21 well eliminate the qualifications. I am only using that as an 22 example. I think there are dozens of other cases that are 23 similar that you are following and consequently we are getting
, 24 a lot of computations with numbers in them that we can't relate w=w narmi, iae. .
25 to any kind of behavior situation.
me ==-
- ' * * ' ~
e l
ir
!h
. _ . _ _ _ JJ
' 329 CR 7631 dag - 4 1 but, ethwEIL: The same argument could be used for ;
i 2 the primary system piping. It's again designed for temperature; i
3 and pressure and designed not to :st'1, but we postulate failure.'
l
(:) 4 T'his is a defense in depth philosophy. It's not ;
I
$ mechanism. It just provides some additional margin. {
6 MR. BEND: If that is the logic, it's all right, but 7 I don't believe it is. !
8 DR. SIESS: It seems to me the problem has to be 9 broken down into two parts.
10 The first question is when or whether you combine
, 11 lodes or events or phenomena. That is the one we got into, 12 combining LOCA and earthquakes, and it's philosophical in some
() 13 degrae, judgement in others, it may not be logical and it may 14 be approachable on a probabilistic basis.
15 The other problem, onco you have decided to combine 16 things, what procedures, what esiculations do you make to do it.
17 That is your combining dynamic lodos et cetera. And what stresses do you allow. Again, that is probabilistic. If you 18 19 are going to combine two things, you usually allow a higher 20 stress level, and so forth. 7, ,
21 There are two separato problera s when do you do it, 22 whether you do it, and if you do it. ,;
O -
'- - Now, I would like to a t,1 or.o more comment. :One, 22 to Irovido a qualitative 0 . . . . , _ . ~24 ' reaso~n' y'o'd 'fiv~o~r combining lode n wa s 48edersi neporwi, ine.
"25 mea'sure'of'm'ar'gi'n' to failure, or to r80v1.to ear 91n to failure.
~
m u- - - ~ - _ _ _ - _- _ - _ _ _ _ _
l
i Mu 330 CR 7631 1 That is probably the peorest reason you can come up dag - 5 2 with because it will inevitably, lead to widely varying margins I
3 of failure.
4 MR. MATTSON: Within a plant or between plants?
1 3 DR. SIESS: Within a plant, because you have some ,
6 portions of some elements of the system that may not be y stressed at all by one of the two phenomena,'in which case, you g have no margin by combining them. -
9 You have other elemente that are stressed equally 10 by the two phenomena , in which you have a significant margin 11 and the margin will vary from place to place, just because one 12 element is affected by LOCA, but not by earthquake or vice
() 13 versa.
14 MR. MATTSON: I agree with you. I understand'that.
15 It is in today's life, a mere way of producing margin.
16 DR. SIESS: You said qualitative, and it is. There 17 is a margin ranging from say =ero to two or something of that 18 odds that you don't -- you know what it is.
19 There are better ways of doing that to evaluate it.
20 MR. MATTSON: You suggest there should be a ,
21 probabilistic approach as to whether or not to combine.
22 We have some elements of that approach in the 22 reactor safety study. -
. 24 DR. SIESS: This is also a probabilistic approach.
A*%ro Moomri, Inc.
25 ASME Section 3, Division 2, the more areas you combine, the
-- .i 1
l l
i
. . y 331 CR 7631 dag - 6 ) more the icdc cycle.
2 MR. SIHWEIL: It's ndt qualitative.
MR. MATTSON: That is the seennd question. I thought 3
O 4 you were suggesting we ought to go to the first question, that 5
is whether and where and the probabilistic approach here.
6 Is there some basis using reactor safety methods, 7
perhaps, or say what you ge,t from the combined event is g
sufficiently small that they needn't be combined, make that 9
decision, then go to the next question. j 10 How do you provide margin to f ailure and to come to 11 that margin by a more rational, uniform method, that is, 12 a place to all
() 13 DR. SIESS: I do: 't think some of the things are 14 quantifiable. If you consider LOCA and earthquakes as 15 independent variables, independent phenomena, it's still not 16 zero probability. They would cceur simultaneously, but I think 17 it's also clear that a LOCA is somewhat more likely to occur 18 in the presence of an earthquake than it is in the absence of 19 an earthquake.
20 If we assume that we assume higher allowable stresses, it's 21 if you wish, because we are trying to balance the risk, 22 done very crudely now.
g MR. BENDER:
I am concerned about the emphasis put 23 Q
24 on the earthquake causing event associated with the LOCA as A*N'ederal Repomes, W.
, 25 they apply to pressurized systems, and excluding that phenomenor.
F1 J
- - - - _ ~ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _
332 ca ~'?I .
dag - 7 '
1 from scourzin9 in anything else which is the way you do it now.
2 ' You say the pras.urizoa system is the most vulnerable.
'3 That is not likely to be the case. I think you really ought to
~
O 4 go and do withoue suggestin,, consider the whether and when and 5 then work out methods.
~
6 'It seems to me you are doing the reverse.
7 DR. SIESS: Your point, mine, is they are combining 8 lodes in some parts and not in others.
~
9 Fui. BENDER: Without any reason.
10 DR. SIESS: This is illogical by any argument. I 11 don't think it can be justified probabilistically.
12 MR. MATTSON: I am not sure yet that I understand
(]) 13 what your problem is there.
14 MR. BENDER: If you assume that an earthquake causes 15 LOCA's, then it's all right to combine the earthquake and the 16 LOCA lode, but that says, for example, that the piping must fail 17 because of the earthquake in order to have the LOCA.
18 But since you qualified the pipe and not so it o
19 wouldn't fail in earthquakes, you have to argue that that is 20 not going to happen. But let me accept the argument that the 21 earthquake --
22 DR. SIESS: Not that, but lower the probability.
O MR. BENDER: All right. Let's accept the argument 23 t ' 25 that the earthquake will cause the LOCA in spite of the fact o
^- ' - .' n n-n-.. n . . : .. : . -
.. .. ....~
Why don' t I say that about all of the sw.e., s ex., .
25' that you ' uq' s1'ify it. .
--er- - -
l e .
l l
%-____m____i_______ _ _ _ _ _ _ _ _
1
.h_
333 CR 7631 . ..
dag - 8 electrical system which is equally qualifiable. There is a lot i
2 more of it, and I know a lot less about it. .
3 DR. SIESS: Of course, it also has single failure 4
so it's reliability may be sufficient MR. BENDER: If thoro is anything that lead to
, 3 6
failures, it's oarthquakes. Whatever causes one, will cause 7 them all.
$pD of 00 8
9 10 s 11 12
,a O
14 15 16 17 18 19 i
' 20 21 ~
l 22 1
0 23
- h 24 i
% a.o- w c o o.,
25 .
~
L. - . . . . .
>i
1
't
- k 334 i
i HR. ETHERINGTON: How much effort is being given, ,
., 8 e
il
- i t '
how much attention is being given to criteria for failure? j 2 l Are you considering those low cycles? When you add ex- :
3 j
- Ek g
tremo forces do you consider restricting any beha'rior or do you allow for stretches beyond the clastic limit and the effect 5 !
on the frequency which would again tend to diminish the 6) peaks?. !
7 MR. MATTSON: We allow for demonstration beyond the 8
elastic range if it can be allowed component by component.
9
' MR. ENIGHT: If I' phase out suddenly, my apologies.
to i
31 I am good for about 10 or 15 words with my present throat !
l 12 condition.
j 1) If I understand the thrust of the question, it 33 would be how much consideration do you give to physical 34 1
15 realities when postulating these breaks, and I guess my direct 16 answer is none.
MR. ETHERINGTON: Then you are extremely conserva-37 18 .tive.
I would agroo with that entirely, 19 MR. KNIGHT:
i MR. ETHERINGTON :
Why is this?
20
-4 21 MR. KNIGHTJhink,wo are ,just now getting into Wo have perhaps the best 22 the realm of reality, if you will. l
$k i
23 document in house'right now from Westinghouse -- that --
r think 24 .ingipaing assentially a fract_rfu , echsn.icaapproach, I
"* %mn, k. try to get at this
, 4 25 that is probably the way most peoplo . - _ . -
, - . f
'1 '
, 7 i 4
c '
335 vm2. j question, looking at, again, once again, this is not unique, 2 _it was done.in the past but not done too'well. Looking at the 2
.siz$oftheflaw, that would'be necessary, the size of unde-() 4 tected-flaw'that would be necessary to give you failure with 5 a[nyreaconableprobability,giventheloadingsthatyouwould 6 see under-an'SSE, we just received that about a month ago.
, 7 'Wp are just starting to look at it. We have asked the people g at Brockhaven'to look at it also. We feel'it would be a
~
19 relatively' sensitive issue, were we to hang our hat on 10 that, we would like to have as broad a base of consensus as ,
11 po s sib l e ... ,
12 MR. ETHERINGTON: I am glad to hear'that,'because ,
I think this is as important as a question of combining loads.
- (_) 13 14 DR. SEISS: Am I correct that in the-Diablo Canyon 15 probablistic study they looked at the probability of earth-16 quake LOCA combination, that the earthquake could cause a 17 LOCA, and also looked at the probabilit es that the earth-18 quake could' disable other parts of the system including s
19 electrical?- Did they cover that?
20 MR. SIEWEIL: You know, the single failure criterion 21 requires --
22 DR. SEISS: I am talking about the Diablo Canyon 0 -23 probablistic study that didn't postulate any single failures.
, 24 MR. SIHWEIL: I am not aware of that.
Am-r.o., i neponers, W . .n:--: '
..%,4,nr . . 25 MR. KNIGHT: I am not certain about the electrical
- .i- .-
i i
. 1
p b ,
r Yi t
336 vm3 j portion. I just finished mentioned this Westinghouse --
DR. SEISS: It took the scenarios all the way 2
3 through. It didn't stop with LOCA and assume probalistic failur=
() 4 They did a complete study right down to contaiment isolation, et cetera. It must have involved electrical systems.
5 MR. KNIGHT: It must have.
6 7
DR. SEISSL: Did WASH 1400 do that? Did it factor 8
the <aarthquake into every step?
9 MR. MATTSON: The way it was cashed out, it showed 10 it was not a significant contributor to the risk and was dis-11 missed. Them was a considerable amount of work since then.
12 DR. SEISS: Am I not correct that the research program to quantify, seismic margins will be looking at all
_() 13 14 aspects of this, or will it simply look at seismic margins and 15 so-called structure limits?
16 MR. MATTSON: I think there is a debate at the 17 moment between the NRR staf f and research staff as to exactly 18 what that perhaps is looking at.
MR. KNIGHT: There is a debate. I met with a 19 group yesterday. The senior advisory group on that program 20 meets again the middle of June, I guess. As currently pro-21 22 posed I think we will probably follow through -- well, it is O 23 a very broad program. I believe the intent here would be to o
24 say bridge the gap left in the WASH 1400 studies, to look Acewe nepon.n, inc.
following 25 explicitly at the earthquake from a risk standpoint, 8
. 4
i1 1
. - . _j d j 337 vm4 1 it all the way through to radiological releases.
2 DR. SEISS: All the mitigating factors and effects 3 on them. .
4 MR. MATTSON: I think I can characterize the debate 5 a bit better. Reactor safety study folks in research want 6 to come at it from the risk assessment view entirely. The 7 proposal we made was to include that view, but to go beyond 8 that and look at the dynamic capability, if you will, of the 9 equipment, the energy absorption capability, not just -- keep 10 it within the clastic range and everybody is okay.
11 DR. SEISS: This could well answer Mr. Bender's 12 question. Would the increased unreliability of electrical component due to the existence of an earthquake contribute
([) 13 14 significantly to risk as compared to decreased reliability of 15 mechanical components?
16 MR. BENDER: In the WASH 1400 study in assessing re-17 liability, they gave -- they showed no loss of reliability due 18 to earthquakes if the equipment had been seismically qualified.
s 19 That was the fundamental premise in it. What I am saying is,
. 20 if that amount holds in one place it holds everywhere else.
21 DR. SEISS: In Diablo Canyon,for example, you could 22 assume there was no failure if the earthquake did not exceed I k) 23 that for which it was qualified and it was a complete failure l
24 if it did exceed it. Or'you could put a tail on the lower end
!n wr.d w neoon.n, inc.
25 and say that there are going to be some defects in the equipment i
c
. .M 338
. . I vmS 1 and so forth so an earthquake even less than the one qualified 2 could cause a failure. It's sort of a gopher proposition.
3 MR. KNIGHT: It gets quickly into a difficult
( 4 area, that is the so-called fergility testing, how much above 5 some qualification level can you take a typical piece of 6 equ'ipment or classes of equipment and still not have failure.
7 MR. BENDER: All I am saying is, an elctrical circui-8 breaker has the same structural problems of any other structure, 9 and if you are going to qualify seismically to survive an .
10 earthquake you are making the same argument you are making when 11 you say the support structure for a piece of pipe survives.
12 MR. MATTSON: I think there is a difference in those
-(]) 13 two arguments. One is an interr.a1 pressure containing 14 component, rigidly constrained, and the other is a piece of 15 electrical equipment hung on a wall. :
16 MR. BENDER: Look at the history of the recent 17 earthquake in Los Angeles and see what failed. You will 18 find out the electrical equipment was the thing that was the 19 real problem.
20 MR. MATTSON: But non-electrical equipment designed, 21 review, tested to survive that earthquake. The pressure 22 component has the same character. We know the probability of O 23 its failure and the consequences of its failure, the conse-quences are serious, probability is raised by the earthquake, 24 O wie,w nenort.ri, w.
25 therefore we go the extra step, defense in depth, because "N
, ,a
i h
AJ 339 vm6 1 it's a direct cause of the loss of collant accident. Where 2 the electrical equipment, designed not to fail, if it fails, 3 doesn't cause a LOCA. .
4 MR. BENDER: If I have to rely on the diesel 5 generators and on the circulating equipment to remove the 6 heat, then I have to deal with the matter of whether the
-7 power supplied to them are there.
8 MR. MATTSON: But the diesel generator is outside 9 of this question because it sees no LOCA load.
10 MR. BENDER: But it sees the earthquake load.
11 The piping wouldn't fail under the earthquake either if it's 12 seismically qualified, pu tell me, but still you are adding the tx ..
U
'~
13 loads as if you were.
14 DR. LAWROSKI: The Chair would appreciate'it if"you 15 let Mr. Bender have the last word, Mr. Mattson.
16 MR. MATTSON: One pos'sible alternative way to come 17 at this problem is for there to be a subcommittee or perhaps 18 a full committee interest in that rescarch program.
19 DR. SEISS: Which one?
20 MR. MATTSON: To which you referred.
21 DR. SEISS: There is a great deal of interest.
22 DR. LAWROSKI: Gentlemen, we have a couple more
- 23 items to be taken up yet before we recess for lunch.
l
, 24 DR. SEISS: We will be discussing that research w d.,w neponen. Inc.
25 program.
f
~
4,,
. )
- d. -
y
. .m Y 340 vm7 j MR. MATTSOti: I am caught on how to respond to you 2 most specifically. I sympathi::e with your concern, I, understand 3 the margin, I understand the uniformity of the margin it 4 provides, but it's generally conservative and it can be 5 generally met and I have the resources to license casos and to 6 solve other safety problems, and I don't have the resources 7
to overlay this project on top of them.
end 21 g .
9 10 11 .
12 O n y .-
15 16 17 18 19 20 ,
21 O 23 ,
, 24 WFede si Reporters, lac.
2' e
f'
.* s p
J 0 341 (22 1 MR. BENDER: I think I am not sure the added con-yp/imi 37631 2 servatism you are providing does much.
3 DR. LAWROSKI: It gives you one more opportunity --
h 4 MR. MATTSON: I am looking for constructive suggestic .2 5 from you all as to how to go about it better.
6 DR. LAWROSKI: Thank you, Roger.
7 We have two :rcre items. One, could we --
8 Is Mr. Woodruff here row?
9 MR. WOODRUFF: Yes . -
10 DR. LAWROSKI: Can we have a few minutes su=ary 11 of the incident relative to the M5t tower at Skagit?
4 12 MR. WOODRUFF: Roger Woodruff from Inspection and
, ) 13 Enforcemen t.
, 14 Last Monday evening about 7:30 the meteorological 15 at Skagit collapsed. It damaged the instrument shack and took 16 out a power line.
17 The Applicant advised us that the cause of collapse 18 was due to loosening of 4 turnbuckles in the guy wire, the 19 4 guy wires on one side of the tower.
20 As you know, an LWA has not been issued for that 21 reactor. There are no security measures at the site 22 other than a caretaker ' who resides, maintains the residence
)
23 there. The caretaker was not there at the time of the inci-24 dent. He is under investigation by the Skagit County sheriff.
D
%. ine.
25 The FBI has been notified, the City of Seattic, t
. V, 9-3 I
, i i
. .