ML20147B343
| ML20147B343 | |
| Person / Time | |
|---|---|
| Site: | Diablo Canyon  |
| Issue date: | 02/24/1988 |
| From: | Advisory Committee on Reactor Safeguards |
| To: | |
| References | |
| ACRS-T-1647, NUDOCS 8803020070 | |
| Download: ML20147B343 (261) | |
Text
. -
, ) .
+' ' I i i y j j ] /3 O UNITED STATES NUCLEAR REGULATORY COMMISSION ADVISORY COMMITTEE ON REACTOR SAFEGUARDS IN THE MATTER OF: )
)
DIABLO CANYON LONG TERM SEISMIC PROGRAM )
)
l /
Q i Date: February 24, 1988 l
Pages: 260 through 416 I
place: Burlingame, california
.........................r w i 240mCEfRY----"------"
i}0 $0".'Removggm ACRS or. c .e HERITAGE REPORTING CORPORATION omaat Reporters O 1220 L Street, N.W., SuP.e 600 Washington, D.C. 2rA05 (202) 628-48&f,-
30 0 880224 l T-1647 DCD
a.
7s 260
%. ) -
-5C 1 UNITED-STATES NUCLEAR REGULATORY COMMISSION 2 ' ADVISORY COMMITTEE ON REACTOR SAFEGUARDS.
~3 4 In-the-Matter oft- )
)
5- DIABLO CANYON LONG TERM SEISMIC PROGRAM )
)
.g
'7~ . Wednesday
' February 24, 1988.
81 -
The above-entitled matter came-on for hearing, 9
pursuant to notice, at 8:00 a.m.
10 BEFORE: DR..CHESTER-P.SIESS 11 Chairman Professor Emeritus of Civil Engineering 12 University of Illinois .
Urbana, Illinois
- ' 33 13 4' ACRS MEMBERS PRESENT:
14 DR. WILLIAM KERR 15 Professor of Nuclear Engineering Director, Office of Energy Research 16 University of Michigan Ann Arbor, Michigan 17 MR.-JESSE C. EBERSOLE 18 Retired Head Nuclear Engineer-
' Division of Engineering Design 19 Tennessee-Valley Authority Knoxville, Tennessee 20 DR. DADE W. MOELLER 21 Professor of Engineering in Environmental Health Associate Dean for Continuing Education 22 School of Public Health Harvard University
). Boston, Massachusetts 23 24 25-(
Heritage Reporting Corporation (202) 628-4888 ;
-m- ,, ,, y e y - -- y-..- e.. ,,2-
-. . -p. . . . .- - . . . ,- -- ~ - . .- --- . . -
--~v..
'i ab, , ~
. fii;;' t ,
- -a ,,t ,
, . 4 ., -
- ', L , + ,,
- 1. .1 - ,
" 't6 : -
1261' !
pf'.
- .A
.y- #
.{,'
R lACRS COGNIZ' ANT STAFF ~ MEMBER .
m .p22 -
p Q... - :'
AliIgne .
t 4
. :3 +
3- ,
- 1. , '
NRC STAFF PRESENTERS '. .
^
i
('
I4 - , .
t Bob Rothman !
i: G 5-
- i. ,
9 - :.f a
t
- ... 7' ,
- 1. - :
t j .- ~ h. , i 2 3
=l
, f 5 ' - ,9: '
4-
- 1. L p -
10 !
4 :
i.~ 11 f
- j. 8 i:i :. ol2 a rg
} l
)..
~
~13' f ;- l - .
i;' 14 i e
15.- -l 16 .j 17 [;
18~ !
'19 !
i 20 'i ,
.s 4
'21 !
1, 22: :-
5
.t
'23 i, i
24- !
t t
25 :
01 :
Heritage Reporting Corporation [
(202) 628-4888 j t
I f
6
.w- A,-,-m -,,__.,,--n----~v-._.----a,.,c--,~.- .-l.---- ----------.----,-------!
,-s 262 7
') 1 PROCEEDI'NGS 2 DR. SIESS - Good morning.
3 when we recessed yesterday afternoon, we had heard 4 the presentation on soil / structure interaction and I think 5~ .there had been some questions, but I would like to ask-now if 6 there are any members or consultants that have any additional 7 questions they would like to ask of the speaker. The speaker 8 is here, isn't he?
9 Then, if there are no questions from the subcommittee 10 members or the consultants, I would like to ask the Staff for 11 their comments on the soil / structure interaction.
12 .MR. CHOKSHI: We have had about four meetings or so
.s 13 with the licensee. The last meeting was in November when J
14 basically the same material which was presented here was 15 discussed.
16 In' addition, there have been conducted some small 17 simplified studies of the containment structure with objective 18 .to assess the sensitivity.of responses to radiation in problem 19 parameters. These results were also discussed at the November 20 meeting.
21 For containment structure, we came to a similar 22 conclusion as Bechtel. In a nutshell, interaction effects were 23 found to be much more significant than the kinematic 24 interaction effects.
25 DR. SIESS: Which ones? I'm sorry.
O
! Heritage Reporting Corporation (202) 628-4888
r-( .g1 t
. .y 263
'1 ~
?: J.L 11 ,
MRlf CHOKSHIt' InertialLinteraction were much more R
-2 significant thah thelk'inematic interaction. 'Some additional 3 ~s tudies'areLplanned to evaluate effects of.different ground 4- motion'.- "
5 In trying to characterize comments made by various 6 panel. members, they generally concern more of the interphase 7 between'the ground motion and SSI and fragility studies. Most 8 of the comments pertained to.the consistency of assumptions 9 'between the various s'.adies.
10 . There has beentsome comments about meterologic, but i 11- most of the~ comments have been addressed-by PG&E. There are ,
t 12 some-minor comments, but I think the most critical comments are 13 within interface.
.o 14 We are planning an interpanel meeting. I believe it 15- is the-lith land 12th of May between various panels: fragility 16 panels,-ground motion panels and SSI panels.
4 17 We have also had a member of the SSI panel-attending
.18 ground motion workshop and vice versa. At this point in-time, 19 .most of.the concerns the panel has are more of that nature.
20- That is the brief status.
21J DR. SIESS: Now, where do you stand in the future on 22 the SSI? Do you fell that you have your methodology developed?
23 I am talking to PG&E now.
24 MR. WHITE: Yes.
25 DR. SIESS: But the inputs from the seismology, O
lieritage Reporting Corporation (202) 628-4888
e- ,
/r l - / A,\;'
- t. l'> w 264
[b I~
1- geology, et cetera,.are not final?
- f. 2 - MR.; WHITE: That's correct.
'3 DR. SIESS: Once those inputs and. ground. motion j:
4 inputs-are final, you will run them through'the same 5 methodology and turn them over to the fragility PRA people.
\?
I! 6 MR. WHITE: Yes, sir.
7 DR. SIESS: And the Staff seems to be satisfied more 8 or less with the methodology.
9 MR. CHOKSHI: Well, there are some minor issues, but 10 on the whole, I would say, yes.
11 DR. SIESS: Any other questions or comments?
12 Okay. Let's go ahead to the question of fragility.
13 MR. CLUFF: Bob Kennedy is going to present a brief
,, 14 summary of where we are in our fragility program. Bob?
15 MR. KENNEDY: I will be presenting a brief summary as 16 indicated of the Diablo Canyon fragility program. Since I have 17 already presented the basic methodology a number of times 18 previously to ACRS including for Diablo Canyon twice before, I 19 am not going to really go into the methodology to any great 20' extent other than a couple of points to refresh your memory on 21 points that are important to understanding some of the results.
22 (Slides shown.)
12 3 Basically, our methodology for making structures and 24 component fragility estimates is to express the fragilities in 25 terms of the equation shown here. Basically, to express both
. O Heritage Reporting Corporation (202) 628-4888 2
, ..,-.,c - - , - - , - . , , - - , . ,
265 i/
1 structures and component-- f ragilities . in terms of a _ ground 2 mot'lon capacity which is a random variable given by a median 3 ground motion capacity, one at which we expect a 50' percent 4 probability-of failure and two random variables that are
- 5. _lognormally distributed. The first of which is a random 6 variable, Epsilon R, that describes the overall randomness of 7 our fragility estimate and the randomness is primarily due to 8 variability of ground motion characteristics and the second 9 which express our uncertainty or ignorance concerning our-10 fragility estimate.
11 Using this form, our fragility estimates take a form 12- that is shown here, basically a median estimate or 50 percent 13' probability of failure point. Our ignorance or uncertainty
.O 14 bands on that estimate expressed by confidence bands.
15 Randomness in the estimate associated with variability of the 16 ground motion and, more recently, we have been concentrating 17 more and more on also' making estimates and checking the 18 reasonableness of our estimates of what are called a high 19 confidence, low probability of failure point essentially taken l 20 as 95 percent confidence /5 percent frequency of failure.
21 Now, in Diablo Canyon, the part of this fragility 1
22 curve that is most important in the risk estimates is the part 23 down in the vicinity of this HCLPF point or high confidence / low 24 probability of failure point, up to about 15 percent frequency 25 of failure.
Heritage Reporting Corporation (202) 628-4888
/
i:
. 266 d' 1 .The part up around the median is not very-important 2 to the risk estimates of Diablo Canyon.- That is somewhat 3- different than for previous PRAs and has resulted.in us having 4 -to. concentrate more on making sure of the reasonablenesses of 5 our estimates lin this portion of the fragility curve.
6 The fragility work at Diablo Canyon has been a phased 7 approach. Basically, it has been interactive in nature and the 8 first step of that phased approach was called Phase II and was 9 several years ago, I guess it was called Phase II because it 10 was the second phase of the PRA work. But it was our first ,
11- fragility effort. And in that effort, it was done fairly 12 similarly to the other commercial PRAs and, basically, the
, 13 approach that was used is to take existing calculations at 14 Diablo Canyon and estimate factors of conservatisms to apply to 15 each of the parameters in existing calculations but to do no 16 new calculations.
17 There were a couple of differences in this early 18 Phase II work. The differences that did exist in this earlier 19 work for Diablo Canyon from other commercial PRAs is that we 20 chose to anchor our fragilities to a different parameter than 21 previously.
22 We have chosen on Diablo Canyon to anchor our 23 fragilities to an average spectra acceleration in the 3 to 24 8.5 hertz range, as opposed to anchoring to peak ground 25 acceleration. And you need to remember that when you look at
)
l fleritage Reporting Corporation (202) 628-4888
b 267
( '
k l' 1 ' numbers, these numbers are average spectra accelerations in the
.2 3.toL8.5 hertz range for a 5 percent damped spectra 3 acceleration.,
4 We believe that this is somewhat more accurate.
5 because it-is.really in that frequency range that dominates 6 structures response and component response and damage, more so 7 than the' peak ground acceleration does. And, thus, we believe 8 -it reduces our uncertainty.
9 Second difference at Diablo Canyon is that Diablo 10 Canyon, we have treated the ground motions in the early work as 11 all having long durations associated with fairly large 12 magnitude carthquakes. As a result of that, we have taken less
.- 13 benefit, less credit for interlastic energy absorption at
'" 14 Diablo Canyon than in other commercial PRAs which were 15 primarily for East Coast sites.
16 After we finished Phase II, several years ago, we
. 17 went on to a second phase of the fragility effort which is 18 called Phase IIIA in which we improved the capacity estimates 19 on the dominate contributors that showed up in Phase II. The 20 dominate contributors to seismic risk. We get a substantial 21 amount of improv ~;nt of the capacity estimates as part of 22 Phase IIIA and I will briefly show those results.
23 At this time, we are in the third phase, fragility 24 phase which is being called Phase IIIB in which we are now 25 cranking in the improved site-specific spectra estimates, the Heritage Reporting Corporation (202) 628-4888
1;;
268 5"' 1 improved: incoherence estimates, the fact that now we have 2 reanalyses of-all of the-buildings to get more realistic median 3 -centered floor spectra rather than having to scale results from-4 the'Hosgri evaluation earthquake.
5 This phase that we are~in right now is making 6 substantial differences in our fragility estimates. I will 7 show the dominate contributors from the early phase because 8 -when you hear the PIUL results, it will be based upon our Phase 9 IIIA fragilities. They are going through substantial changes 10 as a result of the soil / structure interaction work that you-11 heard yesterday.
12 DR. SIESS: I am not sure that I know what you mean
-3 13 by "more accurate" simply from keying it to a different part of
'Qi 14 the spectra range.
i 15 MR. KENNEDY: Well, damage to structures and 16 equipment, at least by our estimates, are really anchored to a 17 spectra acceleration, not a PGA. The spectra acceleration at 18 some specific frequency like, let's say, 8 hertz, which is an 19 important frequency for the auxiliary building, can be i
20 estimated with less uncertainty as a function of the average 21 spectra acceleration in this frequency range from 3 to 8.5 22 hertz than it can be estimated in terms of the peak ground 23 acceleration.
24 So, I am not sure "accurate" is the right 'ord, but I
25 it reduces the variability in our estimate of spectra O
Heritage Reporting Corporation (202) 628-4888 ,
i I
7 . , _.
pT
, c h 269.'
?(~%
'/
'- 11 accelerations.at: specific frequencies of interest.to us.-
2 -- DR..SIESS 1 :It ~ essentially eliminates ' one step.:_
13 MR. KENNEDY: It eliminates.one source of
.4- variability,-yes.
5' .MR.'SEAVUZZO: Bob, could I ask a couple of-
-6 questions? Would you.use the same criteria-for a piece of ,
t 7 internal equipment. Most machinery would have frequencies well i
- g. .
i 8 above 8 hertz with-respect to the building. Is that correct?
.9 MR.' KENNEDY: Well, we need to anchor.all the '
O 10- fragilities to the same ground-motion parameter-for ease of W
11 transfer of fragility efforts into the PRA efforts. If we .{.
12: anchor-each fragility to a different characteristic.of-the- l r
13 ground motion, then we also have to provide cross-correlation :
~(:)
14_ between ground motion characteristics.- And, so, all of the [
i 15 fragilities are anchored to a single characteristic.of the -t 16 ground motion. And then we use an average spectra shade.to. I 17 convert to other frequencies.- .:
t 18 These were the frequencies that we did use'in the 19 early portions of the PRA. -Currently, we are using a somewhat i
20 higher frequency range. We will then report our results both 21 in terms of fragilities for that higher frequency range and for 22 this frequency range here. [
.23 We do have to make conversions to other frequencies. ;
24 Now, equipment in a building often may have a 25 frequency of maybe like 15 hertz or so. But its response, !
l Heritage Reporting Corporation [
(202) 628-4888 ;
i b
I
(-
[
s.
270
~
't
^(}--. .
1- 'because.~1t is up in the. building is.still being very heavily.
2 influence'd by the natural frequency of the building.; :Its'.
3; response is both a function of~the. natural frequency of th's l 4- building.and of the equipment. So,~even equipment with. higher 5 frequencies is heavily influenced by the spectra shape in~this
~6 frequency range.
7- MR. SEAVUZZO: May I ask one more question. :Let's ;
8 say you take a catalever beam at a weight -- all steel.~ What.
~9 is the allowable stress that you.would use to say the thing 10 failed? What number are you using? Yield? Code allowable ;
- 11. values? 3 SMM or what number are you using? ,
12 MR. KENNEDY: Basically, for strength, if I took'a !
r 4
- 13 beam, this hypothetical beam you described, we would tend to
- .O '
14 use the median estimate of its yield strength as a capacity t 15 estimate and then we would use for failure a drift criteria and s .
16 estimate'an interlastic energy absorption factor. In other.
17 words, we would use a effective elasto-plastic force. deflection 18 diagram for that beam and take it up to a drift criterion.
19 MR. SEAVUZZO Well, what do you mean by a drift? ,
20 MR. KENNEDY: How much deformation.
- 21 MR. SEAVUZZO: Based on the work that is being done 22 at EPRI, do you feel that the present criteria being used here 23 is conservative, unconservative? I know you are part of that 24 program, t 25 MR. KENNEDY: I think that there is still lO Heritage Reporting Corporation (202) 628-4888 ;
I
?
. , , - _. . , . . - _. . . . _ , . . . . _ . ._.~.
271
(. -
bl 1- conservatism in our estimates in that we have a tendency to 2 somewhat conservatively estimate the drift =that we are 3 permitted to go to. That when we use these same estimating 4 procedures for buildings that have gone through earthquakes, we 5 tend to overestimate the damage to some extent. I think for 6 equipment, we tend to have a conservative estimate because we 7 tend to base our fragilities for equipment on the highest 8 qualification test data available for that equipment or the
-9 ' equipment survived that test, but we have no legitimate basis
! 10 to raise our frag 111 ties well beyond the highest test data that
! 11 we have available. So, there are several sources of I
c12 underestimating of our median capacities. I think, at the same 13 time, we tend to underestimate our uncertainties. And these O 14 two effects tend to counterbalante each other.
15 So, I do think our high confidence / low probability of 16 failure capacities are in a reasonable range, that our medians 17 tend to be somewhat underestimated and our uncertainties tend 18 to be somewhat underestimated.
19 The effect of all of this is to slight overestimate i
20 the risk because it is not just the HCLPF level or the median 21 that governs the risk. It is the entire shape of. the fragility 22 curve.
23 DR. SIESS: Is there a question down there? Ben?
24 MR. PAGE: I think you mentioned or implied that i .
25 these fragility studies have been compared with actual damage I (
Heritage Reporting Corporation (202) 628-4888 1
P n 272
. 74
\-;} - 1 in buildings that have: suffered-from actual earthquakes.' And I-
~
2~ just wondered whether the comparison was made between fragility-3 studies carried out prior to the actual earthquakes or whether 4 it was kind of a post-mortem.
5 MR'., KENNEDY: They have all been post-mortem. They 6 are all after-the-fact comparisons to try to benchmark the
- 7. method used for fragility evaluations of nuclear power plants.
4
-' 8 But, no, none of them have been carried out before you knew the 9 -results of the earthquake.
10 -
DR. SIESS: Mike?
i 11 MR. TRIFUNAC: I just wanted to comment a little on 12 your question of somewhat accurate or perhaps I would say 13 somewhat less accurate.
U 14 The assumption that you can use 3 to 8.5 hertz range 15 dwells on the principal contribution, the response of the floor
-16 from say the dominate --
17 -It also dwells on the assumption that relative 18 contribution of different modes of vibration is relatively 19 fixed. Depending on where the predominant mode of vibration 20 falls in 3 to 8 hertz frequency range, you can get higher or 21 you can lower estimates. I don't see how you include the L
22 -- contributing response if you take this average. That is one 23 comment.
24 The other comment is that it is totally wrong to say 25 that the high frequencies are influenced at the site. That the i
) -
lieritage Reporting Corporation .
(202) 628-4888 f
, .273
/ \-
k - 1 high frequency motions in the building are governed-by the.
2 . contribution of the average shape of-spectrum and, say.the 3 first or second mode of vibration. .This would be the case if 4 the band'was relatively broad. That is if the damping.was 5 tremendous. But it isn't. We know the damping in these 6 structures is like 3, 4, 5, 6 percent. In that range. And so 7 the spectral peaks are really lateral. And as long as there is-8 a high acceleration or maybe a small earthquake that happens
.9 that-has, say, spectral amplitudes in 20-25-30 hertz range, it 10 is really not going to make any difference whatsoever what is 11 the average in 3 to 8.5. It is going to go right through 12 because the transfer function of the beam is going to basically gs 13 1. So, in some sense or some applications this may be more 14 convenient. But I am seriously concerned whether it is more 15 accurate or what is really something that should be permissible 16 for a broad spectrum of frequencies and floor response spectra.
17 MR. KENNEDY: We have done analyses. I am absolutely 18 convinced that it is more accurate than using the peak ground 19 acceleration. I do agree with you that it is multiple 20 frequencies that influence the response of floors and of 21 equipment and of the structure, itself, so that we have to use 22 an entire spectra shape, we can't just use values at one 23- frequency.
24 For the PRA purposes, we anchor or key that spectra 25 shape to the values at one frequency and we find that it is Heritage Reporting Corporation (202) 628-4888
. g;-
['
4 4
- '274.-
/
h V(~ 1 1 .more;accuratelto key that to the. frequencies, to spectra
- y. ~; ft l2; accelerations, toLaverage spectra accelerations over a
-3 frequency rangeithan a peak ground 1 acceleration.
I
~4 On response at high frequencies up in buildings,'lthe 5: zero period' acceleration of floors up in buildings is primarily-6 governed'by the fundamental frequency of the building for most
~.
7 earthquake records. For earthquake records that have' energy at 8 the fundamental frequency,sthe energy at'the fundamental 9 . frequency of the building is -- not the energy, but the 10 response is~ highly amplified as you go up through the building.
11 The higher frequencies tend to not be arplified as you go up
-12 through the building. Yes. They go on up through the 13 building, but they are not amplified. So, the ZPA at floors O' 14 high up in the building still is very heavily influenced by the 15 lower frequency modes of the building.
i' 16 I do feel that the lower; frequency modes of the 17 building and the spectra content at those modes. dominates the
[. 18 response of' equipment in buildings..
19 DR. SIESS: Bob, for the spectra you have been given, L
- j. 20 over what range does the ratio of zero period acceleration to 21 the 3.5 to 8 or 3 to 8.57
. 22 MR. KENNEDY: What is the median range? Basically 23: the 3 to 8.5 hertz frequencies for the median amplification 24 between the 3 to 8.5 hertz versus the zero period acceleration 25 is at 5 percent damping is approximately 2.35.
i () Heritage Reporting Corporation l
(202) 628-4888 i
275
^-').
~
l' .DR. SIESSI. And how much does that. vary for the 2 different spectra you have?
3- MR. KENNEDY: What we have found is a. ratio -- that 4 when we anchor to_an average spectra acceleration, now, the 5 -final work that we have done is not anchored to the 3 to 8.5 6 hertz. It's been anchored to 4.8 to 14.7 hertz. We are alming 7 at something in the 5 to 14.5 hertz' range. I mean that specific 8 frequencies of these other frequencies. But 4.8 to 14.7 hertz 9 range, when we anchor into that frequency range, what we find 10 is that all the way for 4 hertz cn1 up to the zero period 11 acceleration that-the peak and valley variability-of the 12 records that we're using and the directional variability of the s 13 records we are using lead to an 84 percent non-exceedence 14 probability spectra divided by a 50 percent non-exceedence .
15 probability spectra, leads to a ratio of 1.25. So, the 84 16 percent spectra is 1.25 times the 50 percent spectra. That 17 ratio is pretty consistent with the average ratio of 18 amplification factors that Newmark showed in his work. The 19 amplification factors that Newmark had in his work varied with 20 frequency and we find that when we anchor to an average spectra l
21 acceleration, we can keep them constant across the frequency 22 range from 4 hertz on up to the ZPA, that we get increased 23 ratios as we drop below about 4 hertz. But we have less 24 interest in the spectra below about 4 hertz.
25 The next point I want to make is in the Phase II and
( L Heritage Reporting Corporation (202) 628-4888 L
I
276
< ~x
\I l' IIIA PRA results, specifically, but to some extent even in our 2 finals, we-do make use of the separation of variables concept 3 that we.have used in past PRAs in which we are basically taking 4 an evaluation cf .he plant for some previous design earthquake.
5 In the case of Diablo Canyon, it is che Hosgri evaluation 6 earthquake. As our basic response analysis for the plant. And 7 we then scale that by factors of conservatism that we astimate 8 for each of the variables that are entered into the overall 9 structure and component evaluation. '
10 So, we are really effectively estimating factors of 11 conservatisms on each of the parameters and then we estimate 12 median values of those and variability of those and then w'e
,, 13 combine them together as independent random variables. A 14 strength factor to describe the capacity of a structure divided 15 by the loads applied to that structure for the evaluation 16 earthquake, the Hosgri earthquake in this case, interlastic 17 energy absorption factor to account for the fact that for a 18 limited duration earthquakes, we can undergo responses beyond 19- the yield or capacity limit. A structure's response factor to 20 account for conservatisms in the Hosgri evaluation of 21 structures response, equipment response factor to account for 22 factors of conservatism in equipment response relative to the 23 structures response.
24 DR. SIESS: Wait a minute. Are those factors --
25 would those factors be different for an original design than O
Heritage Reporting Corporation (202) 628-4888 ,
1
'l 277 l f x-C C) : 1= theyLare for the Hosgri redesign?
2 MR.LF.ENNEDY:- Oh, definitely.
-3 DR. SIESS: Tnese take into account --
~4 'HR. KENNEDY: Definitely. Because a number of the 5 factors of conservatism haveLa1 ready been removed.in the Hosgri 6 evaluation ~ .
7 DR. SIESS: I thought that is what you were saying.
8 'I just wanted to be sure.
9 MR. KENNEDY: Definitely, we have to look at how the 10 evaluation was dene for the design earthquake or evaluation 11 earthquake before estimating these factors to get up.to median 12 levels.
s 13 In the Hosgri evaluation, we did -- in Phase II and 14' Phase IIIA, we did estimate medial site' specific response 15 spectra. We did not use the Hosgri evaluation site response n 16 spectra and we scaled all the results by the ratio of the 17 median, our estimate of the median site specific response 18 spectra shape versus the Hosgri evaluation shape.
19 Now, in Phase II and IIIA, this is the response 20 spectra shape that we did use. It has got constant spectra 21 acceleration from 3.5 up to 8 hertz. I'm sorry. From 22 3 up to 8.5 hertz. It has an amplification of 2.35 for the 5 23 percent damp spectra. This was purely a judgmentally arrived 24 at median site specific spectra. It was based upon Professor 25 Tsai's judgment and my judgments.
O Heritage Reporting Corporation (202) 628-4888
9 j_ s 278-k-[ 1 Our final fragility efforts which are going on now;we.
2 are using a median site specific spectra' shape shown to y.nu 3' yesterday by. Wen _Tsing. That median shape looks very similar 4 to this one that we used in Phase II and IIIA so that our 5 judgments --
16 MR. TRIFUNAC: May I ask a question?
'7 MR. KENNEDY: Yes.
8 MR. TRIFUNAC: This is basically one earthquake 9 representative of a typical spectra. Now, from the point of 10 view of fragilities, I would like to know what if at all or in ,
11 what way you can consider the following hypothetical situation. ,
12 !
13 There-is a major earthquake. Say, on Hosgri, to be
()
7 14 ~ specific. And it results in a particular spectrum. Now, 15 immediately following that earthquake, we-know that we can.
- 16. expect some aftershocks. Some aftershocks may be far away, now 17 may be nearby, but the complete picture may be that during an 18 hour, maybe two hours, we have a large earthquake which maybe-19 has a spectra of something like this and we have also some 20 smaller earthquakes possibly with much high stress aftershocks, 21 which is not uncommon for aftershocks, also occurring nearby 22 which contribute to other spectra in time. How does that come l 23 into our consideration of fragilities? I suppose the l
24 components or equipment that we are trying to evaluate are all 25 ignnr:nt of whether this is an aftershock or a main event.
() Heritage Reporting Corporation (202) 628-4888 ;
e V
g -
279 I ')
5/- They all see that vibration in accumulative fashion. How-do we n-
-2 factor.that in?
3 MR. KENNEDY: I think there-is maybe two parts to 4 that' question. The first part is: Does damage accumulate such 5 that If you.go through a main shock and then several' 6'
aftershocks are you going to have substantially more damage 7 than if you just went through the main shock alone.
8 And the answer to that part, at least from within my 9 experience of looking at ea'rthquake damage -- both damage in 10 main shocks and damage after aftershocks, it has been my 11 experience structures and equipment which are damaged in 12 aftershocks that have spectra less than the main shock or less
_ 13. than or equal to the main shock,; structures and equipment can ,
14 be damaged in the aftershock even though the aftershocks do
- 15 have spectra less than or equal to the main shock, but the only 16 . times that structures and equipment have that damage in such 17 aftershocks is when they were somewhat damaged or on the verge 18 of damage from the main shock.
19 In other words, there was distress, observable 20 distress in the main shock and then they came down in the 21 aftershock.
22 We believe that's accounted for in our fragility 23 estimates by the large uncertainty ranges on our fragility 24 estimates and by the slope of the fragility estimates. Both 25 the uncertainties in the slopes say that you are going to be k
Heritage Reporting Corporation (202) 628-4888
280
) 1 estimating a probability of failure for that component. A 2 significant probability of failure for that component if it is 3 on the verge of failure. So that we basically feel that we 4 have covered that aspect of aftershvcks in our fragility 5 estimates.
6 The second part of your question is the aftershock 7 might have di'ferent frequency content than the main shock.
8 And such that in certain frequencies the aftershock in some 9 frequency ranges might actually have spectra exceeding the main 10 shock.
11 We believe that is being covered -- this represents 12 our median ground motion spectra shape. We do put on top of 13 this variability to account for peak and valley variability or
/T
14 frequency variability of different earthquake records. And we l
l 15 put on top of this variability to account for directional 16 effects.
17 We believe that variability covers a wide range of 18 earthquake ground motion characteristics. In fact, in i
19 estimating that variability, we took 38 ground motion records 20 and averaged them together and they do cover -- our variability 21 does cover the variability from those 38 records.
22 Now, none of those 38 records -- well, some of those 23 38 records have rather high frequency content. The probability 24 of getting high ground motion from either the main shock or 25 aftershock is already covered in the hazard period. So, I think g
\J Heritage Reporting Corporation (202) 628-4888
c: s V3lf' .
1 .that through.the variability that is covered in the hazard 2 acurves, the variability that we have in"uur spectra shape and 3 'the uncertainty bands that'we place on our fragilities, we-4~ believe~we_are covering the potential ~ damage from aftershocks..
5 .(Continued on next pago.)
6 7
8
'9 x 10 11 12 13 0 -14 15 16 17 18 19 20 21 22 23
-24 25 O Heritage Reporting Corporation (202) 628 A888 a
r L
282
+1
's' 1. MR. KERR: Mr. Kennedy, I'm not quite.sure how 2 uncertainty bands cover anything if they really are uncertain -
3 bands.. I mean --
'4 MR. KENNEDY: Well, they're wide bands and basically 5 in my opinion, how they cover, unless you have -- ground motion 6 that is on the verge of failing the component in the main 7 shock, the component will not fail in the aftershock.
8 Your ground motion in the main shock has to.be 10 to 9 20 percent of the ground motion that would have resulted in '
10 failure of the component so that it will at least have damaged .
11 the component significantly if that component is going to come 12 down in the aftershock.
- 13 MR. KERR I think I understood that. My question 11-)
14 had to do with the significance of an uncertainty band. Now, 15 if you really believe that that's not an uncertainty band, but 16 is a band of conservatism, that's one thing.
17 But if the uncertainty really is that large, to me,-
18 that says-that there is some reasonable possibility or 19 probability that the acceleration or motion or whatever, will 20 be there.
21 MR. KENNEDY: Okay, I think the answer that I can 22' give to that, our uncertainty bands tend to be about a factor 23 of four wide, between the 90 percent bands.
24 In other words, our 95 percent confidence band tends 25 to be like half of our median estimate, and so our 90 percent A
U Heritage Reporting Corporation (202) 628-4888
-m -
, n -
p, lY o -
'283 1,
- esl.
, ~('N J i =1:
. confidence band centered'around ou.r median estimate, tends to-4 2' be about'aJfactor of'about a 4Y.. So we've gotivery large--
~
!3' uncertainties. ,
4 'if we then added in.another ten to twenty percent'for
^
~
5 aftershock1 effects, this would be a trivlal' increase.in our; 6 already existing' uncertainty' band. ;
7 I think, you know, if our uncertainty.. bands'were very f 8' narrow, then I would agree,.that we would need to add in.an 9 additional factor-for aftershocks. The aftershock effects are !
10 small compared with the' uncertainty that we have to start.with.; ;;
11 MR. KERR That's true of a lot of things. Including [
H ,
12 .that.. ;
13 DR. !!OELLER: To follow up on Dr. Kerr, I gather '!
.14 it's a band that you apply to account for your assumption i 11 ~
15 regarding the uncertainty, is that a correct statement?
16 Mk. KENNEDY: .It's a band that we apply to account- l 17 for uncertaint'les for each.of the parameters that go-into 18 making up our fragility estimate. And in this field, there are !
-19 very large uncertainties, even when doing rather sophisticated 1 t
I 20 fragility estimates that have been done on Diablo Canyon, we 21 still have large uncertainties. There still is just an awful l l
22 lot of uncertainty concerning the ground motion capability of l 23 structural equipment to a wide variety of earthquakes. :
24 But it is an estimated uncertainty band. j 25 The next point I would like to cover so that you're {
(
Heritage Reporting Corporation :
(202) 628-4088 ;
l l
I
f ; ;
284
<\
(_/ 1 all aware of it; ~in all of our fragility estimates, phase 2.and 2 phase,3B and the_ current ones, we have not been using the soil
~
3 structure interact 1on effect or TAU effects that were used in 4 the last re-evaluation.
- 5. In fact, in all of our fragility estimates today,k 6 we've been taking very little credit for either soil structure 7 Interaction effects or other effects that have sometimes been 8 lumped in and called TAU effects.
9 We have a phase 2 and phase 3A fragilities included 10 on the statistical incoherence estimates of the effects of the.
11 statistical incoherence of the ground motion. These estimates 12 were based upon work that was done by Woodward Glide for the 13 NRC in their effective ground motion study.
O'_ 14 The statistical incoherence estimates are directly 15 applicable to ground motion on the El Centro differential 16 array, which is a very statistically coherent material, but is 17 soil material, as indicated by Dr. Trifunac yesterday, and so 18 there are uncertainties in the estimates that wo usu.
19 These are the estimates that we did use in the phase 20 2 and phase 3A fragLlities. No reduction to response in 21 frequencies of 5 hostz and less; reduction of .9 at 10 hertz, a 22 reduction of .8 at 25 hertz, for a typical 150 foot plant 23 dimension, and that's scaled with plant dimensions.
24 Those estimates are currently being revised as a 25 result of the coherence work that you saw yesterday. It turns Heritage . Reporting Corporation (202) 628-4888
285
- 1 out that the incoherence work that you saw yesterday agrees 2 rather closely with these entimates, but would support somewhat 3 greater reduction at the low frequencies, and somewhat less 4_ reduction at the high frequencies than what we included in the 5 fragility studies that tha current PRA work is based on.
6 So there will be a tendency to reduce the effect at 7 high frequencies, and increase the effect at low frequencies.
8 DR. MOELLER: What is the wave length of a 5 hertz 9 wave? I 10 MR. KENNEDY: Maybe Wen, can you -- well, for this 11 site, yes.
12 MR. TSAI: For this site it average of, let's see --
,_ 13 yes, 3,000 feet, so divided by five would be 600 -- divided by
(
)
14 15, 200 feet.
15 MR. KENNEDY: lot of this reduction effect come out of 16 --
17 DR. SEISS: I'm hearing a lot of incoherence. Dr.
18 Tsai, the wavelength will be 3,000 feet divided by the 19 frequency. So 3,000 divided by 5 would be 600 feet.
20 MR. KENNEDY: For 5 hertz.
21 DR. SEISS: For 5 hertz, thank you.
22 MR. KENNEDY: A lot of this reduction that came about 23 in the statistical incoherence study because of statistical 24 incoherence of the ground motion and not travelling wave 25 offects. Both effects are in that study.
(~)l Heritage Reporting Corporation (202) 628-4888
m l-.
286
,(~)-
'\ '
JL And that's why there's.some reduction at 5 hertz',
- 2 when the travelling wave effect wouldn't get very much of that 3 reduction at 5' hertz.
4' MR. TRIFUNAC: Can I ask a question then? In-your 5- calculations;you are, or you are not using the soll. structure 6 interaction results?
7 MR. KENNEDY:- In phase 2 and phase 3A fragilities, 8 being'those that you will get the PRA.results in after I i
9 finish, we did not_take any credit whatsoever for soil 10 structure interaction effects.
11 In the current fragility effort, which is about half 12 completed, we are taking full credit for the soft structure
- s 13 interaction effect that you heard about yesterday. We are 0 14 'using the soil structure interaction effect you heard 15 yesterday.
16 MR. TRIFUNAC: Well, in that sense then, I would like 17 to make a comment that, if you are not using soil structure 18 interaction effects in your calculations, which is fine, and if 19 you do want to take advantage of the reductions due to 20 incoherence of ground motion, you must add to your response 21 calculations effects of rock on it.
22 Because the incoherence in the way that it is 23 defined, brings about a very strong component of rock, and that 24 component of rock is going to be significantly increasing the 25 frequency, so you might say it is in proportion to frequency,
() Heritage Reporting Corporation l
(202) 628-4888 I
l' 287
- 'i 1 so that for the high-frequency treatment, whatever it is, the 2 effect may be even larger than that deduction you get from the 3 quartz and quartz TAU equivalent for incoherence, whichever way 4 you want to call it, so that taking advantage of one thing and 5 not considering the other would not be permissible, but if you 6 think that this is advantageous to take the reduction, then you 7 should take the other thing into your calculations.
8 MR. KENNEDY: The effects that I showed a few minutes 9 ago for statistical incoherence were based upon studies from 10 Woodward Clyde in which the detriment from rocking effects was 11 incorporated.
12 In other words, there were done for typical
~s 13 structures and these were the reductions of what were found, i' i 14 including the detriment from rocking and torsional effects.
15 When applied to Diablo Canyon, the current 16 incoherence studies that you heard about yesterday, have more 17 rocking effects than this work that was done using the El 18 Centro differential array, and that is one of the reasons why, 19 with the incoherence work that you heard about yesterday, the 20 reduction at the high frequencies tends to be somewhat less 21 reduction than shown here. There is more rocking effect 22 increase than was considered in the phase 2 and phase 3A 23 fragilities.
24 This .8 factor here would tend to be more like .85 or 25 .86. It's not a big increase, but we are not getting in the
'U Heritage Reporting Corporation (202) 628-4888
288 I statistical incoherence word that is being currently factored 2 into the current fragilities. We're not getting as big of an 3 effect.
4 The rocking effects are being included in the current 5 fragility offort, because they are in the statistical 6 incoherence work that Wen Tsin presented yesterday.
7 MR. TRIFUNAC: Could you then define what you mean by 8 the rocking effects, so I am sure that we are talking about the 9 same thing. What is "rocking?'
10 MR. KENNEDY: There is both rocking in the vertical 11 plane introduced by non-vertically incident waves, and there is 12 torsional effects, rocking in a horizontal plana, or a rotation 13 in the horizontal plane included by the statistical incoherence
\
)
14 of the ground motion. Both the rocking effects and the 15 vertical plane and the torsional effects in the horizontal 16 plane are being considered.
17 MR. TRIFUNAC: As a result of structural response to 18 excitation or as a result of forcing function of the ground 19 motion.
20 MR. KENNEDY: As a result of both. Both are included 21 in work that Wen Tsin has done. And these imbedded structures 22 with inclined waves.
23 MR. TRIFUNAC: So when you do not include the soil 24 structure effects in your calculations, what is it that you 25 take as a contribution from rocking or from torsion?
r~'x
.Y Heritage Reporting Corporation (202) 628 4888
t 1 h r-289 7,
k.) l' MR. KENNEDY: In the final fragility work, the soil 2 Estructure interaction effect.is being included, and the
- 3. statistical incoherence work of Wen Tsin is being included.
4 In the early fragility work, we did not have'the soil 5 structure interaction material available to us. We believe 6 that would generally reduce response, and that it has' generally 7 reduced response; and we believe.that the effects would be 8 fairly small.
9 Therefore,.in the early fragility work, we based'our 10 statistical incoherence estimates, and we did not have~that 11 available for Diablo Canyon, we based it on the work that 12 . Woodward Clyde did using the El Centro differential array, and
~13 some example' nuclear power plant structures placed on-different-(~)'
14- site conditions subjected to ground motion effects that were 15 based upon the El Centro differential array.
16 And so there are to some extent rocking effects and 17 torsion effects included in these estimates. But they were not 18 for Diablo -- these estimates are not for Diablo Canyon-19 specific estimates. And they are approximations.
20 But in the final fragility work, the full soil I- 21 structure interaction information is being taken into account.
22 It is having the effect of increasing most of our fragilities, 23 reducing the risk.
24 MR. SEAVUZZO Bob? I'd like to ask one more 25 question. It's related to the embedment and the rocking Heritage Reporting Corporation (202) 628-4888
. 290
. fg
() 1 motion. I know that one test was conducted from Mathiesen 2 years ago on the CPR where he had looked,at the embedment o-3- question. And what they found experimentally was that after a 4 few' cycles, the analysis of a fixed-base was more accurate for
~
5 the embedments that rocked away from it on-both sides. I 6 -wonder if you might comment on that in the application to 7 what's been done here.
8 1 01. KENNEDY: I think that was a model test, wasn't 9 it?
10 MR. SEAVUZZO: No, the California tube reactor, a 11 small, quiet containment structure that was tested with big 12 shakers.
, 13 MR. KENNEDY: Okay. Well, I certainly agree that 14 when you have smaller scale and strong ground motion, you tend 15 to lose the benefit embedmont effects, after a certain 16 number of cycles. And as a result, you need to reduce the 17 effects of embedment. That question really should be asked of l 18 Wen Tsin as to what he actually did.
19 It's my belief that he did make some reductions in 20 the benefits of embedments in his models, but I don't have the 21 answer to that question.
I 22 MR. SEAVUZZO: That question has been raised for the 23 NRC reviewers and we are in the process of going back and 24 checking on the stress at that interval. Ours is embedded in 25 rock, and we don't expect the same kind of degradation as you O Heritage Boporting Corporation (202) 628-4888
A 291 1 would'at a soil site, butlwe are going back and taking a look 2 at the stress at that interface.
3 MR. KENNEDY: Wasn't that filled with concrete?
4 MR. SEAVUZZO: Yes.
5 MR. KENNEDY - It was, all right.
6 MR. SEAVUZZO: I mean, yes, concrete placed against 7 the rock.
8 MR. KENNEDY: Basically, at the end of phase 2, these 9 are the component's which were identified as being the probable
- 10. dominant contributors to seismic risk at Diablo Canyon. And 11 the diesel generator was considered to be a very important 12 contributor to potential risk at Diablo Canyon, primarily 13 because it had rather low high-tech and a slow probability of
!O
- 14 failure at capacity values. Again, these are average l
L 15 structural acceleration values, in the 3 to 8.5 hertz range.
L 16 And then there was followed by a series of components l
l
' 17 2 through 5 which were important contributors to risk, 18 electrical components, reactor coolant found in containment fan 19 cool-ars, followed by some additional electrical components 20 which were contributors to the risk but to a lesser extent; in 21 a single building, the turbine building.
22 None of the other buildings entered in as 23 contributors to the risk.
24 As a result of this preliminary work, phase 3A we 25 have gone back with an effort to significantly improve capacity l
Heritage Reporting Corporation (202) 628-4888
/
292
!m)-
\/ 1 estimates of all of these components. We have not, for 3A
.2 improved the response estimate, because we wanted to wait until 3 we had all of the response data that you heard about yesterday 4 and we are in the current process of proving the response 5 estimates.-
6 MR. EBERSOLE: Mr. Kennedy, before you take that 7 -down, I have here a piece of paper dated February 5, in which 8 you have sent a letter to the staff providing answers -- to the 9 original list of components identified, approximately 60, to be 10 of interest in this seismic array, you sent that list. It 11 actually has 80 components or pieces of equipment.
12 DR. SEISS: Are you-familiar with what he --
j- 13 MR. KENNEDY: -It doesn't really matter, I'm going to 14 get to the demo --
15 MR. EBERSOLE: I'm going to get to the general point 16 in a minute. We have then tabulated information on 75 of 17 those, but all I'm getting to is this: I look at this list, 75 18 or 80 pieces of equipment and components, and I say that it's 19 got to be a shortfall of the whole show, the whole bit.
20 That it has to be interrelated to the thing called a 21 "Q-list," or Appendix R safe shutdown list, or at least [
t 22 "Shutdown," or somehow expand it to include the obscure details i
23 of equipment not only which are safety classified in their own 24 right, but also such systems as may interface with them in a 25 system directed context. l
(
Heritage Reporting Corporation (202) 628-4888 1
~
r---
r.
293
(^)
N./ 1 For instance, we came in finding the absence of any 2 qualification of fire protection systems which was permitted by 3 the stats, which permits coincident operation of that system, 4 hypothetically, all over the place. -Yet it may'be based on 5 selective operation in certain regions, and cannot tolerate the 6 concept of universal occurrence all over.
7 This would occur in the diesel plants and switchboard 8 rooms where one thinks you are going to uniquely protect the 9 space, but in fact everything is going to go off at once, and i
10 you will disable through your directive effects the whole 11 functional network that gets heat out of the plant, and that's 12 the only thing we're interested in is getting heat out of that ;
_ 13 plant. -
5 -
14 So I would suggest that you've got to review using
[
15 appropriate system interaction or system integration engineers 16 the whole list, the Q-list, the Appendix R list and make a 17 comparative analysis and pick the obscure relays and details of 18 HCLPF operation which tend to remain sort of under the rug 19 until you do that.
20 MR. KENNEDY: .Okay,'the list of components for which 21 we have made fragility estimates was provided to us by 22 Pickford, Lowe and Gehrig in Diablo Canyon PG&E people.
23 MR. EBERSOLE: You've got to do it with your own 24 outfit.
25 MR. KENNEDY: What?
l r
Heritage Reporting Corporation (202) 628-4888 L
e~ -
4 I
~'
294
/ ;
\~ # 11 MR. EBERSOLE: I.think you have to do it with.your 2' own staff.
3 MR. KENNEDY: I don't have systems engineers. .I 4 can't do it.
5 DR. SEISS: -These are'PRA questions.
6 -MR. EBERSOLE: Bob, of PG&D, these are idsntified'as J
7 a list of components identified to be of interest.
8 MR. KENNEDY: Yes?
9 MR. EBERSOLE: Now, does that mean these are the only 10 . components that they have' looked at, or does that mean that ;
11 these are the only components they have looked at but they !
12 think they need the fragilities for?
13 MR. KENNEDY: Let me answer a little bit of that
-U-s 14 first and then I am going to dump it into PG&E or PLG's lap.
15 Basically, there's been a number of other PRAs done i
16 on PWR plants. These components were the components that have 17 been identified as potential contributors to seismic risk and
~
18 other PRAs and that is part of the reason they're on the list.
19 The other part of the reason they're on the list is 20 they were based upon system evaluations at Diablo Canyon done 21 by the systems engineers as components which were most likely 22 contributors to seismic risk. It is a reduced list from the Q 23 list, because to ma;ce these fragility evaluations for i
24 everything on the Q list would be a horrendously costly effort, 25 so it is a significantly reduced down list. But it is the list
} [
Heritage Reporting Corporation (202) 628-4888
b 295 yy 4 s' 1 .that.was identified as being.the most important contributors.to 2 risk.
3 When'we walk-down a plant for these components, we 4 only walk-down a plant for these components. I should make 5 that clear., These are the only components that we're 6 explicitlyflooking at. When we walkL d own a plant for these -
7 components, we are-looking for systems interaction issues that 8 will influence the fragility of these' components and if we find 9 any systems interactions issues that will influence the 10 fragility of these components, we do. include those.
11 So seismic spatial systems interaction is in. The i 12 other systems interaction issues that you talked about really ,
13 are not in.
- () 14 And now, I guess I will turn it over for more 15 explicit --
16 MR. SEISS: No, let's defer it. We're going to hear 17 from PRA people later. Let's go through the fragilities.
18 As of now, the FRA people have given you a list of e
'19 components they want fragilities for. Let's stick to the 20 fragilities and then we'll address this question to how they 21 arrived at that list.
22- MR. KENNEDY: I believe the list that was sent to you 23 includes all of the components which we have estimated i
24 fragilities. I think that is a complete list. I can't swear :
1 25 to it at this time. i Heritage Reporting Corporation (202) 628-4888 i
I i . - _ , , ,
4 l
296 f"Y
(_)
-1 DR. SEISS: I can just look at the number 80 though,.
2 nd say-that_this is foreshortened by a factor of 4. And how T' 3- they-got it'down from 320 to 80 is something we'll have them 4- address.
5 MR. KENNEDY: It is approximate 1y'80 that we have 6- 'made fragility estimates for.- Andlof those 80, these were.the-7 items that were identified to be the potential. dominant t
8 contributors to seismic risk in the phase 2 fragility effort.
9 DR. SEISS: What was your cut off point? Three or 10 something?
11 MR. KENNEDY: What we basically did was we made l
l 12 fragility estimates _for every component that had a median l-13 _ capacity of less than 10 G. For all those components that had
\/ 14 median capacities greater than 10 G, we set their median at 10 15 G; we set their HCLPF at 3 G, the conservative estimate of that 16 HCLPF, We checked that by checking a few of those components.
17 Any component that we believe would have a HCLPF 18 level with respect to acceleration level less than 3 G, we made 19 a detailed estimate.
20 MR. EBERSOLE: Just as a stab, why do you need 21 reactor coolant pumps and containment fan cooling to get a
. 22 plant safe? I don't understand that.
23 DR. SEISS: That's a PRA question. Save that for 24 PLG.
25 MR. KENNEDY: You don't need the offerability of ym
(-) Heritage Reporting Corporation (202) 628-4888
p ; 'V i
. ji
.297- ;
(- 1 . reactor coolant pumps, but'you.do need protection of the
'2 pressure boundary.- Our fragility estimateLis an' estimate of 3 the. protection of the pressure pump.
4 MR.' EBERSOLE: We really don't. call that "the pump.- -
5 1H1.-KENNEDY: In our report, we make it clear.that 6- the fragility estimate of the protection-of the pressure- .
7 boundary basically it was an estimate of.the seal and 8 substantial seal leakage.
9 MR. EBERSOLE: And it's just -- right,.the boundary.
10 Okay.: ,
11 MR. KENNEDY: The ground motion range that dominated l 12 the seismic risk from the early PRA were estimated to be in the-j-c 13 1.75'to 2.75 G spectro acceleration range. Basically, to-put l
u')
14 that in context, the ground motion range that dominated the [
15 seismic risk in the early PRA work and which I believe will 16 continue to as the PRA is refined, is from slightly below the . ,
17 Hosgri evaluation earthquake level, to about 1.5 times the i
18 Hosgri earthquake level.
19 So it is ground motion in general that is somewhat 20 higher than the Hosgri evaluation earthquake level that is 21 dominating the seismic risk. It is high ground motion. U 22 MR. DAVIS: A question on that, Mr. Kennedy, that 23' slide? How did you evaluate the probability of loss of off- !
24 site power to the function of ground motion?
25 MR. KENNEDY: Basically what we used for the f
1 g "
Heritage Reporting Corporation (202) 628-4888
w -
s
- 298
-(): 1 probability ofLoff-site power as a function of ground motion, 2 is we used. damage data that we've collected from utilities in 3 . California, on the relationship between the loss.of offsite 4 power and ground' motions estimated at those locations.
5 Our relationships, we've uued essentially the same 6' relationships in all of the seismic PRAs. .They are not Diablo 7 Canyon specific relationships. They are based upon -- based
-8 purely upon empirical data on the performance of substations in 9 past earthquakes. .The estimates are that there is a 50 percent 10: probability of a loss of off-site power somewhere in the .3 to 11 .35 peak ground acceleration. The numbers I remember in my 12 head are for peak ground acceleration, somewhere in the .3 to 13 .35 G peak ground acceleration; 50 percent probability of loss
() 14 of off-site power.
15 We have found one case of loss of off-site power at 16 very low ground motion levels, about as I recall, .12 G peak 17.. ground acceleration, so our uncertainly bands are such that our 18 HCLPF level for loss of off site power, is about .093. And 19 we've got very large uncertainty bands that range from .09 G, 20 the chance of loss of off-site power, .35 G a 50 percent 21 probability; for the time you get up to .9 G, peak ground 22 acceleration, essentially 100 percent probability of loss of 23 100 percent power.
24 MR. DAVIS: I just would like to point out that this 25 can be a very critical probability distribution, because I
(' Heritage Reporting Corporation (202) 628-4888
_ C
=
, 299 N h 1 suspect plant blackout will dominate seismic risk. This has 2 been the case in previous PRAs. .j 3 MR. KENNEDY: It looks like. ]
4 MR. DAVIS: But at the Coalinga quake, the substation 5 survived and that was a very high ground acceleration.
t I
6 MR. KENNEDY: I think there are some recent i
"_ 7 information out there that would indicate that our estimates of g -
a loss of off-site power are too conservative, because we also !
9 assume they can't recover from this loss of off-site power. We
,- 10 believe that, at least for losses at the lower ground motion N
p- :
11 levels, it is very likely they could recover rather rapidly, so E T 12 I think you're right but there is some potential conservatisms _-
13 in those estimates. _
W Excuse me, I thought I heard him say not 1 14 MR. KERR -u
- a 15 that it was necessarily conservatism, but that there was a 1
, 16 fairly key number that would have a considerable amount of ,
i 17 influence. '
F 18 MR. KENNEDY: Well, okay. I think there are some 19 conservatisms in them. I will not blame --
E. i E $ 20 MR. KERR: No, I didn't mean to say whether there was j m -
21 or was not, I think Mr. Davis was speaking to coservatism
"' 2 22 there.
T 23 MR. KENNEDY: We estimate a 50 percent probability of 24 loss of on-site power at .3 to .35 G. Coalinga saw about .5 G, 25 and did not have a loss of off-site power. We would estimate a
7 0 Heritage Reporting Corporation (202) 628-4880 -
,)
6 . . . . _ . . =
r: ,
9 o
- 300 p(
(/ 1 about a 60 or 70 percent, I don't have the exact number of 2~ probability'of loss of on-site power for around .5 G.
3: There is some-indications that at the_ low ground.
'4' motion levels, you could recover from loss of off-site power-5 that'we have not taken into account.
6 MR. DAVIS: The data seems to suggest that it's a 7- plant-specific situation, and you are not looking at the plant-8I specific design? I think that's what --
9 MR. KENNEDY: 'But I'm not convinced that Lit suggests 10 that it's plant-specific. It really depends -- I mean, this 11 seems to be high amounts of scatter in the data; ceramic 12 insulators are vulnerable; sometimes they have failed;
_13 sometimes they haven't. Sometimes a cortain type of ceramic 14 insulator seems to have behaved very well in an earthquake, and 15 then you get to another earthquake and that same type has not 16 performed well. There is a lot of scatter. We have not done a 17 detailed study for Diablo Canyon of loss of off-site power.
18 MR. EBERSOLE: Did you pay any particular attention 19 to the D.C. supplies for breaker coordination for off-site 20 power and non-safety electrical systems? The reason I ask that 21 is, if you lose the D.C. power supply, you lose all 22 coordinative functions for off-site breakers opening and 23 closing. You might be in an awkward situation where you wanted 24 to open a breaker to stop a process and you couldn't. That 25 would be a curious reversal of what you wanted, such as O Heritage Reporting Corporation (202) 628-4888
l f
y 301 e
/
- 1 Sequoyah -- you might want_to turn it'off.
2 MR. KENNEDY: Well, we certainly have' paid a lot of 3 attention to D.C. power supply within the plant.
4 MR. EBERSOLE: Of course.
5 ~MR. KENNEDY: But maybe you_ guys should ask that --
6 MR. EBERSOLE: This is another set'of batteries out:
7 in the turbine hall somewhere in the yard even.
8 MR. SMITH: We can look at -- Bruce Smith, PRA.
9 MR. EBERSOLE: Yes. Could you inform me about the
- 10. unfortunate reversals being unable to turn things off?
i 11 MR. SMITH: One of the things we looked at in PRA, ;
12 we're especially concerned with a loss of off-sito power and ,
f
,3 13 how we could recover it.
I
\~'J 14 One of the things we did model in the PRA are i
15 operator actions that would be required to close or open 16 breakers under conditions where we did not have control power.
17 MR. EBERSOLE: So you went out in the yard among all !
18 the fallen down equipment and you took a handle on things?
19 MR. SMITH: Exactly. We do have, you know, our 20 rating department does have plans upon doing that -- in fact, 21 we were surprised to find that they did have -- and they do 22 have experience in doing that.
23 MR. EBERSOLE: Working so you could manipulate the ,
24 coordinative functions using muscle?
25 MR. SMITH: Yes, in some cases. !
Heritage Reporting Corporation ,
(202) 628-4888 i
r h
- w. .
f 302
, ~
'( ) 1 MR.-KENNEDY ' With some probability. .
2- MR. SMITH: With some probability of error.or some f
3 probability of success.
4 -But to date, we have_not taken any creditLin that in f
5 the PRA. '
6 MR. KENNEDY: You have not?
7 MR. EBERSOLE: Well, you can't even disconnect short 8 circuits if you don't have D.C. power to do.it.
9 MR. SMITH: You bet.
10 MR. KENNEDY: As I indicated, after completion of 11 phase 2, we revised the capacity estimates of these dominant 12 . contributors; in fact, during the detailed walk-down on the 13 diesel generators that was made between phase 2 and phase 3A,
() 14 we have found additional load paths in the critical areas and 15 were able to significantly increase our estimates of the diesel 16 generator, so that they are no longer the dominant contributor 17 to seismic risk, but they are an important contributor. ,
t 18 These components, 2, 3, and 5, also were able to j 19 raise our capacity estimates the same time a more detailed look !
20 at components 6 and 8, the potential transformers, the diesel i 21 generator control cabinets, led to a reduction of our capacity 22 estimates. The turbine building estimates from a detailed, 23 non-linear time history analysis led to some reduction of our 24 median estimates by keeping of the high-confidence / low-25 probability of failure estimates about the same.
{
("J'
\ Heritage Reportino Corporation (202) 628-4888 I
i 1
303 1- So that some of these dominant contributors from I) 2 phase 2 capacities went up; some went down. The overall 3 conclusions as to the capacity estimates didn't have any 4 substantially different on the average fragilities from those 5 that had been estimated in a preliminary way in phase 2.
6 In general, both the median estimates generally 7 increased, but we also had increases in our estimates in 8 randomness.
9 Th3 overall impact on the phase 2 risk estimates was 10 not small from these -- what was small from these fragility 11 changes -- but the dominant contributors did change some, and 12 I'll leave that to the PRA people to describe the dominant 13 contributors from phase 3.
n I would like to again point out that what you're
(_) 14 15 going to see, and the fragility results that I have presented, 16 are based upon scaling of the Hosgri evaluation response 17 levels. They do not taka into account the soil structure and
?. 8 action work you heard yesterday; they do not take into account 19 the statistical incoherence you heard yesterday; and they took 20 into account the approximate values I gave you.
21 And we are well along in the revising of all of these 22 probabilities to take those factors into account, and what we 23 are finding is there are some significant changes in the 24 fragilities, such that I suspect that the dominant contributor 25 will change some. The most dominant contributors will likely Heritage Reporting Corporation (202) 628-4888
[
304 I ).
I stay _the same.
They are not-going to change enough that they ,
~
2 are_not going to be important contributors.
'~
3 But there will be some changes, that'this more recent i
4 work you've heard about will tend to raise the HCLPF levels on 5 many of the components. With the overall-seismic risk from a 6 fragility standpoint, we will tend to come down some. How 7- much, I don't know. But there will be some raising and so you 8 need to use those dominant contributor lists with some caution.
-9 MR.'EBERSOLE: Mr. Kennedy, this term, "Walk-down,"
10 leaves me a'little bit hung -- I think it was born in the early 11 days of Diablo Canyon and pertained to the structural fc111ng 12 down of one elemen4 onto another. You recall-the vast effort 13 you put in to see if unqualified things could not fall on (G_l' 14 necessary components. But any way, the term, "walk-down" has 15 gotten into the business.
16 But I can never think it's no more than a superficial 17 visual operation in which you notice the special relationship f
18 of one thing to another or others. It doesn't have the depth ,
19 that you apparently connote that it has in determining system f 20 interactive effects in the nense that a three-dimensional study 21 would have of the systems in-depth that you have in one 22 compartmentalized space, do you follow me? I'm just saying you i 23 can't walk around and see anything on the surface of what you ;
i 24 see. You don #t take off cabinet covers; you don't look at the !
25 guts of relays; walk-down has a limited capacity to understand
() Heritage Reporting Corporation i
(202) 628-4888 i
s.
"=
305 I_'l x- 1 spatial interactive effects, and I think it's overbroad in its 2- scope.
3' MR. KENNEDY Well, during seismic walk-downs of 4 plants, including Diablo Canyon, we do take off covers and we 5 do open. cabinets. Walk-down may be the wrong term because 6- there is a fair amount of looking at the equipment.
7 MR. EBERSOLE: It might be.
8 MR. KENNEDY ' I will say that, for seismic _ spatial 9 systems interactions we are looking for components who by their 10 fall, length of fall on these components, and the look is 11 fairly careful. For seismic spatial systems interaction, it is 12 visual and it is judgmental. There has not been any
, _ . 13 recalculations this time.
~
14 MR. EBERSOLE: I made a note on here as a case-in-15 point. You can see hydrogen piping in your-list. Hydrogen 16 piping in the wrong place wher e leakage can be taken out by 17 ventilation would be a wicked thing to have if you lose 18 ventilation.
19 MR. KENNEDY: Yes, I would classify that as something 20 other than seismic spatial systems interaction. Of those 21 interactions, I have not personally been involved in them. I 22 don't know if they have been done or not. Probably should have 23 asked.
24 MR. WHITEt We did include the hazard from hydrogen 25 piping in the spatial interactions.
Heritage Reporting Corporation (202) 628-4888
z 306 l MR. EBERSOLE: It doesn't show in the literature.
- (~)
\._
1-2 MR. KENNEDY: I want to.next go on to the current 3 work. I did not think that so far that the discussion would 4 have taken as long as it has, so I'm going to try to cut'down 5 from this stage on, to keep reasonable time. But-the current' 6 work, we are revising 100 percent of all of the fragilities 7 that we've estimated to include better estimates of structures 8 and equipment responses based upon the re-evaluations that have 9 gone on Diablo Canyon as part of the work you heard yesterday.
- 10 We will also be improving capacity' estimates of a 11 limited number of components. We are including a more ;
12 realistic site-specific spectro shape -- however, it is not ,
13 very different than the one that we used in the fragility
() '14 estimates as shown. We are including median center building 15 response analyses for this site-specific spectro-shape, for all 16 of the buildings. In other words, entirely new floor spectra, 17 This is having a rather significant effect on our 18 fragility estimates. In most cases, it is increasing our i 19 estimates. There are some components that it is reducing. ,
20 Basically our reevaluations have ended up with somewhat i 21 different frequency contents to these buildings, in somewhat 5 22 lower frequency content, may well have been used in the Hosgri 23 evaluation. Heace, for components in about the 8 hertz 24 frequency range, it will result in somewhat reducing of our 25 fragility estimates.
rh Heritage Reporting Corporation (202) 628-4888
307
' s'
' 1 So there will be some shifting. We are incorporating 2 the soil structure interaction effects, including spatial 3 variation of the ground motion.
4 DR. SEISS: When you reduce-the fragility, is that 5- good or bad?
e 6 MR. KENNEDY: That's only I always have trouble with I
7 that. In other words we are reducing our capacities in some 8 cases. In most cases, we are increasing our capacities. I can 9 never remember which way that goes for fragility.
10 MR. SEAVUZZO: Will this information be more 11 available to consultants in the future?
12 MR. KENNEDY: Yes, in fact, tomorrow, we're going to 13 go through a lot of this with some of the consultants in the f
{s) 14 fragility area. And have gone through some of it already.
15 In our fragility work that is currently going on, we 16 use 30 time histories. These 38 time histories have been i
17 composed of 24 components from. empirical records and basically 18 28 components from numerical records. The way we have handled l
19 them, we have doubled the influence of the empirical component, 20 the empirical record relative to the numerical records.
l 21 The empirical record, which said, Item 2 horizontal 12 2 compliments from the empirical record could act in either of 23 the two directions of the Diablo Canyon buildings. We've 24 applied those empirical records twice to each of our building 25 models, once with the north-south empirical record component in O Heritage Reporting Corporation (202) 628-4888
308 I the north-south building direction; secondly with a north-7' ')
2 south empirical component in the east-west building direction.
3 So the empirical components have been applied twice; 4 the numerical components, we did keep the directional effects 5 and only applied them once to our building models, to double 6 the importance of the empirical record.
7 Empirical records that have been chosen, you heard 8 about these yesterday, were chosen specifically to be 9 representative of high ground motions. Ground motions that
'10 might get up into the range that dominated the seismic risk at 11 Diablo Canyon. Ground motions up in this range from 1.75 to 12 2.75 G.
13 All of these records had to be -- most of these
() 14 records, I should oay, have to be scaled up to get up in that 15 range. In other words, the seismic risk from Diablo Canyon is 16 coming from ground motions greater than most of these records 17 recorded, and they've had to be rescaled up to get through to 18 that range. They've been linearally scaled up over the entire 19 frequency range for that purpose.
20 The only exceptions to that statement were the Tabas 21 records, the only record of the empirical records or numerical 22 records -- they all had to be scaled up.
l.
) 23 The Tabas record is up in the range that is 24 contributing heavily to the seismic risk of Diablo Canyon. It 25 is the only record that is up that high. Pacolma Dam record is fl)
'- Heritage Reporting Corporation (202) 628-4888
309
') I close. It takes only a relatively minor scaling up of the 2 Pacolma Dam record. All of the other records had to have very 3 substantial scaling up to get up into this spectrum 4 acceleration range.
5 DR. KERR Mr. Kennedy, the reference is made to 6 "numerically generated Hosgri records," does that refer to the 7 computer program that we heard?
8 MR. KENNEDY: Yes, the semi-empirical as somebody I 9 think said yesterday.
10 DR. KERR As a member for the Society for the 11 Guardianship of the English Language, I object to the use of 12 the term, "records," to refer to a computer generated result.
13 I just want to record my objection.
() 14 MR.. KENNEDY: Computer generated time histories.
15 DR. KERR: Is pseudo-records.
16 MR. KENNEDY: Is pseudo-records, okay. This set of 17 38 records produced response spectra at the site as shown here 18 -- produced a 50 percent -- l'm sorry, one other step was 19 splitt all of these records were scaled to 2 G average spectra 20 acceleration in the 4.8 to 14.7 hertz range. When we found we 21 scaled into that frequency range, we had the most constant 22 relationship between our 84 percent and 50 percent non-accident 23 probability spectra for our frequency range of interest.
24 They're all sceled to 2 G in the 4.8 to 14.7 hertz range; 2G 25 in that range is equivalent to 2.25 G in the 3 to 8.5 hertz
(]
\'
Heritage Reporting Corporation (202) 628-4888
L , ,
o
..,. 310
\/ 1 frequency range. So we could say we scaled all of these 2 essentially to 2.25 G, and the 3 to 8.5 hertz frequency range.
3 They were all then-scaled to'this same common average 4- for the average of their two horizontal components. And from 5 that, we then got spectra that were exceeded half the time by 6 the actual records, plus the pseudo-records,-and the spectra 7' that-were exceeded 16 porcent of the time for our 84 percent a
8 -non-accidents probability spectra. l 9 These distributions do not depend upon the assumption 10 -of the log-Normal distribution. They are the empirical data. l 11 This was exceeded 16 percent of the time and this was exceeded ;
I 12 50 percent of the time.
13 They do tend to fit. It's a.very small number, but 14 they do fit the log-Normal distribution reasonably well.
15 We find that for 4 hertz on up, there is a constant 16 ratio of about 1.25. There is some scatter in that ration, but .
17 it averages constantly about this frequency range of about 18 1.25.
19 The median site-specific spectra that you saw 20 yesterday fits very close to the median of these 30 records.
21 It essentially represents a smoothing of these 38 reccrds, i
22 With these records, and with the median spectrum we 23 saw yesterday, we performed median analyses for each of the l 24 buildings. We did it two ways; one way, which I call I i
25 probablistic, and one way that I will call Deterministic, to !
Heritage Reporting Corporation s (202) 628-4888 i
r 311
/' 'l s' I distinguish-them.
2 The probablistic approach is, we did a whole series 3 - of analyses on the building -- did 38 analyses using each of 4 those records once. Coupled into those 38 analyses we took 5 into account structure, frequency, variability of the building; 6 in other words we made the natural frequency of the building, 7 the stiffness of a random variable, and selected it randomly, 8 made damping of the buildings our random variable selected 9 randomly, and rock modulus properties a random variable
/
10 selected, set, randomly.
11 So they, as our 38 separate analyses of the building, 12 using different building proporties, and different time :
,, 13 histories.
14 Secondly we did that same building model over again, 15 slightly different model but that same region over again, using 16 the median ground spectra, and a single median model of the 17 building.
18 Basically we did this twice because we used two 19 different time histories. One was the Tabas record modified
?.0 to fit the median spectra as Wen Tsin said yesterday; the other 21 was the pacoima Dam record. So there were two analyses, but 22 they both fit the median spectra and used a median building 23 model. And we compared the results.
24 The next thing in your handout package was the 25 properties that we used for the building models in the ,
Heritage Reporting Corporation (202) 628-4880 ,
pu Y . .=
L 0 312
}'8
(_/
1 probability analyses. _ We dJd have rather large variabilities.
2 ForLinstance, we included 90 percent ~ confidence bands on 3 dampening range from 4 percent to 12.5 percent.
4 This'representsan representative median and 84 L
5 percentinon-accedence probability floor spectra for'high up in 6 the auxiliary brilding from the probablistic analysis, from the.
7 38th analysis.
8 And I'El put on top of that the median that came from ,
l 9 the deterministic analysis,.i.e., using a median ground spectra j 10 shape, of a single median building raodel, with these two
.i 11 different records. And you see an excellent fit between the ;
12 two medians.
, 13 We do that.same kind of a fit at all of the other 14 floors in this region of the auxiliary building.
15 Because of that fit, we reached the following 16 decisions for carrying forward the PRA. Basically, we found
- 17. close agreement between the median probablistic and the median 18 deterministic floor spectrum.
19 As a result, the median we believe that the median 20 probablistic floor spectra can be reasonably approximated by 21 doing deterministic spectra using median ground spectra and 22 median structures properties. In other words, that we do not 23 have to take into account multiple analyses and all of these 24 other properties.
25 What we intend to do is to take advantage of that and
.O Herit v e Reporting Corporation (202) 628-4888
,, . . . . . ,- . - -. . = . . ~ ~ - - -
- g -
. d
- g
'v:'
.1; E" ' '
- 313I 1 ,: u'se:'the Emedian',fdeterministih--mediani results Ethat Wen; i Tsin '
~
I' r-2 showed :you' yesterday for thel: final [ fragility 'ef forts.
~
. 31 ' LDR.:KERR IsLthere;reallyfa'reasoneto expect lthat'
~
'4 ;you sh'ould or should"not'have gotten1the1results:thatyyou'are L
h S appealing'.to?
6 MR.JKENNEDY: There's reasons to expect'both that we '
7 :should; and thereLare reasons to expect that we shouldn't have;- ~
'8- And;we didn't know'which would dominate until1 after the
'9 analyses. ,
- 00. If the variability:of'the< ground motion dominates, 11 ~ and the median ground spectrum will then be.a: broad frequency 1 t
- 12 spectra, if th: variability of ground motion dominates,1the
_, 13 variability:of-floor response, then the: median deterministic
(] _ 14J ' floor' response and?the-median probablistic floor response will-15- tend to be similar as they were here.
And I think-that's o, i
/ . 16 because the ground motion variability dominates the variabLlity 17' :of structures. response.
18- If~the variability of frequencies ~in the-structure t 19L were to' dominate structures response, then the-median :
a 20 .probablistic floor spectra should be broader and have love-
, - :21 peaks'than the deterministic spectra because the deterministic 22 spectra are assuming that you know the structure's frequency.
- 23 I think part of the reason why they read very close
- . 24 is that there is also -- they tend to differ some also -- ther.e
'25: cis very high amplifications up through the building. We do not L
O. Heritage Reporting Corporation (202) 628-4888
-~4 e v + + -m y vey,-r-,,,, w w. -,y-,-+,-..e.v- :, y es g-++-y<3 s
J
_ ;_. 314 1 Jfind those in the soil structure interaction analyses of Diablo f
~
- 2 Canyon 1because there is a fair amount of radiation of energy 13 ' packed into the ground in the soil structure interaction study 4 of Diablo. Canyon. So-that-the radiation of energ'; back in to 5 -the ground is rather important and is somewha*..more important 6 than the damping that you use for the structures model.
7 Becauce of that fact, and the wide variability in the 8 timed histories U.at were used, I believe the results are that 9 the two median spectra agree well with eachother is very 10 reasonable. If we had not had the radiation back in the ground 11 such that we would have had much higher amplifications, or if 12 it seemed more important from a variability of structures.
- 13 frequency content, we would not have expected to get the same I ~'l 14 for both ways.
15 DR. KERR In your analysis of structures where the 16 major contributions comes from the stiffness of concrete walls, 17 what assumption do you make about cracking?
18 MR. KENNEDY: In a PRA we have cracked the wall, and 19 we have assumed stiffnesses that are only a ptrtion of the 20 uncracked stiffnesses. That is one of the reascns for the 21 frequency shift between the PRA and earlier work.
22 For fixed base models that is fairly important. For 23 models on soil structure interaction, that is not very 24 imporcant. There is enough frequency shift due to the soil 25- structure interaction, it masks or overcomes the frequency P)
(_
Heritage Reporting Corpo2ation (202) 628-4898
i m ;
"i. s.
315 01D ' shift"due.to; concrete' drafting. <
2 sSoLin some.of these models it's important,fand:in 13 some of them it-is'not important. We have'takenLinto account 4 concreteicracking.
5; DR. KERR : 'Thank you.
6- MR. KENNEDY: And that one of the reasons why we also 7 F. ave such a wide frequency uncertainty, we have very wide 8 frequency.-- we have a coefficient of error -- wp have
.9 logarithmic. standard deviation on stiffness of .5 at Diablo-10L canyon, which~would be'a logarithmic standard deviation on a 11 frequency of . 25. Rather large.
1'2 MR. TRIFUNACt Excuse me, I thought that you 13 indic&ted that the concept- of af tershocks relative .tcf main:
~ ',) ~
14 event has been included somehow through probablistic. analysis, i
! 15 but it'is not-included here.
l 16 MR. KENNEDY: It is included by the fact that some of
~
17- Lthe records that go in to those 38 ' records, have.very high-i j 18 frequency records that go into those 38 records,.have very high E
l 19 frequency content. The Gazli record is a rather high-frequency 20 content record, and the Nahani record is a high-frequency 21 content record.
22 So long as the frequency content of the 38 records, 23 envelopes, is consistent with aftershocks , it is included. If 24 aftershocks or frequency contents different from these 38 1
- r 25. records, then it is not included.
b n Heritage Reporting Corporation (202) 628-4888 M ,- e v. -- 4 , n- -+-m, m,,y- - - . . . , - ,,-,-g- ,,y .-~---.--e-, g w--wn-g ,f~-
j 316 k l" 1~ Because it's only being included.through the 2' variability of these 38 records.
3 ~ MR . TRIFUNAC: Now, which of these 307 -Are you-4- referring to the Tabas or Gazli equipment and so forth?
5 MR. KENNEDY: Yes.
6~ MR. TRIFUNAC: I.see here that --
7 MR. KENNEDY: Gazli is high-frequency. Gazli has a 8 lot of high frequency content, and so does Nahani.
9 MR. TRIFUNAC: Right, but they are large earthquakes.
10 MR. KENNEDY: Yes, and they have high frequency.
11 MR. TRIFUNAC: I agree, but let me finish the point.
12 They are large earthquakes, and therefore, they have large gq 13 spectral amplitudes in the range where you normally spectra, O -14 right? If you took an aftershock, which is a small -- to 15 magnitude 3, magnitude 3.5, aftershock on Hosgri, you have very r
16 small amplitudes in the spectrum range where you actually 17 normalize all of the spectra for your relative-scaling 18 purposes. So the effects will be totally different.
19 MR. KENNEDY: Well,.I think that in spectra, though,
-20 in 14 hertz would not be any higher than Gazli at 14 hertz.
21 Yes, its values at 5 hertz would be very much lower, but it's 22 effects at 14 hertz wouldn't be any higher than Gazli, when 23 Gazli is normalized, this 2.25 G in the 3 to 8.5 hertz range, 24 we get very high spectra acceleration from 14 hertz from Gasli.
25 MR. TRIFUNAC: I appreciate all of v. hat. My point is Of
\J Heritage Reporting Corporation (202) 628-4988
g - -- ,, ,
(
~
317
, f;- 1 'that you make.the conclusion that.whatido you do:probablistic
.2 or-to deterministic calculations, doesn't.make much difference, 3; 'and then~you.went through some justification supporting _that 4' this isireally a consequence of-the fact that spectral shapes
'- 5
- of the excitation are broader, and so'forth.
6 - But that is all fine if.you're looking only at large 7 earthquakes, but I thought that you were promising to_ talk 8 about the aftershocks as well, and I didn't see how that fit 9 the picture.
10 MR.~ KENNEDY: Well, I didn't mean to be promising to 11 be talking'about aftershocks. I thought what I said is that I 12 felt the spectral variability that was included in our 13 fragility evaluations would cover the effect of aftershocks. I
-( ) ~14 still think it covered the effect of aftershocks in that I do' 15 not see how aftershocks are going to produce higher spectral 16 accelerations than those that weLhave included in the fragility 17 ' evaluations.
18 Aftershocks will have-different frequency content, 19 because they will'not have the low frequencies. But there are 20 high frequency spectra accelerations, I don't see how we get 21 higher than either Gazli or Nahani.
! 22 MR. TRIFUNAC: Well, you see, the frequent difference 23 between a main earthquake and an aftershock is that the p 24 earthquake releases a large fault and has a relatively low 25- overall stress drop, average stress drop. Aftershock occurs Heritage Reporting Corporation (202) 628-4888
-en m e ~mv 4 4 4 #
'At"%
- n. ,.
s;
' ~
e I;
< ' te - ,
N , .
h ,
'318-1; where the: stress. concentration isLa-very'small segment of.the.
} .
l2L -fault.1'And sol-usually it'!h'as'.'a(5, l'0, 20 times, more than:20
- 3 ttimes-larger stress drop than
- the. main' shock. :Which means.that;
' .4# -the.high/ frequency end-of the spectrum may be 10 times ther L5; amplitude during the main earthquake, but during theilong
.6 -period of time, the amplitudes are much smaller because it's-a
.7 small dimensional source.
- 8. And so the shape of the : spectrum is .very dif ferent .
9- 'and it'is-small in intermediate.and lengthier motions, but it-10' is higher, andican be substantially higher in the high
~
i 11 frequencies.- I don't see how that comes into your analysis. ' -
12' MR. KENNEDY: I am.not aware, and maybe.I'd better~
- 13 turn this over to'the geoscience people. I~am not aware of an-() 14 ~ aftershock that has produced high' frequency spectra
-15 acceleration higher than about 2 G, which is the kinds of-l 16 ' levels that these earthquakes that we are using in evaluating !
' 17- the fragilities ~ produced. .;
1 18 We're talking here about broad' spectra accelerations 19 in the high frequency range for-Gazli and Nahani, once they're' i 20 ' scaled up, that are over 2 G. And I don't know, what did you I i
- 21. want to --
22 DR. TSAI: Yes, well, my comment to that essentially :
- 23 is the same as Dr. Kennedy's. That is, the records included in ;
i 24 this: suite of time histories has included high frequency 25 records including Gazli, Nahani, and in particular Nahani, that f
. 4 i
f
- l. Heritage Reporting Corporation (202).628-4888 9 y- .-
,,a.-,g we . + -+ , . , .
,..e.-. -rs,,-w , , , , , e ..m.., .__.y ,__,..,m4.- __m,.1%.. - . . . - - , - - - . vr -wrew --w--
i i
,s _ 319 peaki oscillation 2 reaches 2 G, andIin addition to-that, in'the 1
s 21 -proximityfif even scaled up at;1ow frequency range thatutend to;
- 3 ' reinforce'even higher the high frequency component, and the-4 main argument is that, with that' kind of enveloping, and after
~
5 this point, we don't see aftershock'even their high frequency.
6 component are relatively end-reached,but they have not reached 7 that high level.
8 And that with that end-levelling, we really don't see 9 that there is a need to consider aftershock.
10 I think on the other ground, that is, if aftershock 11 has any effect, the building itself will have frequencies shift l
12 to the lower frequency range, and that I would argue if it were 13 the diminished effect of the aftershock.
(~-) 14 DR. LEISS: Thank you. Do I assume that the staff 15 has heard that question? Do we need to follow up on it? Or 16 have-the panel look into it?
l l 17 MR. KENNEDY: For the correction of tnese recent l
l 18 ground motion produced floor spectra, to floor spectra from L
l 19 incoherent ground motion, we are going to use the correction 20 factors for median corrections. We are going to use the l 21 correction factors that Wen Tsin showed yesterday.
22 Basically this is an example of this point high up in 23 the auxiliary building, the relationship between the floor <
'24 spectra from coherent input from median center deterministic 25 analysis, single analysis, versus floor spectra from the O Heritage Reporting Corporation (202) 628-4888
JU6- < .
320
{(} 1-2
.incoherentcinput..
We will use the ratios.as an; estimate;ofithe median 3- effect-due.to incoherence, and, basically,z the-ratios are at-4- Ethis' location'_are shown here.
f
- 5. Essentially about .9 in~this frequency _ range:of' _,
6- around 10 hertz; less at low frequencies; and for' higher 7 frequenclea; and_this reduction of.the higher freg'encies u is 8 due to this rocking and torsional effects that Dr. Trifunac was 9 mentioning-earlier.
But they tend-to range at this location of the OPS
~
11 bu'ilding from about .9 to about .95. Somewhat greater 12 reductions for low frequencies; somewhat less reductions 13 particularly atithe high frequencies than what we had used-in
- 3) 14 up through the phase.3A fragility work.
15 Because these reductions due to incoherence are 16 small, we're not taking very much credit for~ incoherence. In
- 17- fact, looking~over all of the floors and all of the buildings, l
l - 18 these reductions range from about .8.uplto 1.0. There is no
- 19. case that I am aware of that we're taking reduction factors of 20 less than .8 for a median reduction. Because these corrections 21 are small, we have decided not to do variability studies for .
L 22 the PRA on these correction effects.
l 23 Instead we have decided to make judgmental inclusion 24 of variability, a relatively conservative judgmental conclusion 25 of variability. For instance, where we use a media reduction ,
n-U -
Heritage Reporting Corporation -
(202) 628-4888 l ,
L _
I a f 3
~321
- (,.)
\/ JL of .8, we're assuming that there is some possibility that 2 -incoherence could actually increase our: response. Where we're-3 getting.immediate-reduction of: unity, in other words, no 4 reduction, we're admitting to a possibility that we could 5 actually get about a 12 percent: increase in response.
6 We feel that the inclusion of this variability will 7 include -- is more, than what we'd_ find by doing the 8 variability studies, and because the reductions are.small, 9 we've chosen not to do the variability studies on it.
10 DR. SEISS: I don't understand the top line of that, 11 where the median is one -- no, in your table, where you've 12' taken no credit for the incoherence, how does your HCLPF go up
- 13 by 12 percent?
(".s. )
14 MR. KENNEDY: We're just simply assuming that, where ,
1 l- 15 the median reduction is known, is one, there is no credit for L
j 16 incoherence. We are assuming that there could be some l
17 detriment from incoherence from that case.
i l
18 in uther words, we are assuming that there is still l 19 variability associated with that incoherence estimate. And 20 that, in actual fact, that incoherence could possibly increase 1 21 our response.
22 DR. SEISS: Well, what does 1.12 mean?
23 MR. KENNEDY: It means that, on the incoherence, the 24 incoherent response could be 12 percent higher than the 25 coherent response. In other words, this floor spectra could Ik Heritage Reporting Corporation (202) 628-4888
v _
- e. - - -
9
'322
's 1 1
.actually.be' increased --
'2 EDR.lSEISS3 Oh,-okay.
'3 MR.' KENNEDY: --Edue to incoherence.
4 _DR. SEISS: But your HCLPF is'.a G value. Why would 5 that.be?
' ~
6- MR '. KENNEDY: I probably shouldn't have the (f) on 7 -there. I probably_should say high confidence / low probability 8' and leave.the (f) off.
-9 DR SEISS: -Okay.
10 MR: KENNEDY: This is a basically -- this is the 11 value that we would assign 95 percer.t confidence,-five percent 12 probability that the incoherent effects could even be higher 13 than this.
q 14 DR. SEISS: Okaj 15 MR. KENNEDY: In that table it is not a G value.
16~ Similarly for containment uplift, what you saw 17 yesterday, basically what you saw yesterday from Wen ~Tsin, is 18 .the containment _ uplift analyses done for the case where all of
~19 those records were' scaled up to an average spectra acceleration 20 5 percent of spectra acceleration being 2.25 G -- a very high
'21 ' ground motion, but ground motion up centered in the region that 22 has been dominating the contribution to seismic risk.
23 There were scaled up to that ground motion and then 24 Wen Tsin indicated that they did three different records, 25 Tabaz, Pacoima Dam, El Centro No.4, all scaled up to this I)
- Heritage Reporting Corporation (202) 628-4888
-c ,,
323 7S .. . .
-i f 1 . ground motion, and estimated the containment: response including 2 ' base mat uplift, versus;the~ containment response ignoring base 3 mat uplift, and what he showed was that there was very little "
4 effect~due to base mat uplift.
5 In fact,.when you average those three records 6 together, the effect completely disappears throughout most of 7 the building.
8 -What we have decided is that we will make very small 9 corrections for this effect, and we will include variability, 10 but it will be a judgmental -- we've estimated variability as 11 well. We will not be doing variability studies for the PRA.
12 DR. KERR: Mr. Kennedy, again, I have some unease 13 about introducing conservativism in a situation in which k.%.) ' 14, eventually this is going to enter into'a PRA which will involve 15 a very large number of systems and components. It is not clear' 16 to me that one know what is conservative under these 17 circumstances, and I'd feel better if one did a best estimate.
18 MR. KENNEDY: Well, to get a best estimate of these 19 variabilities would require us to do very costly variability 20 studies.
21 DR. KERR: I'm not suggesting a lot of costly 22 studies. I'm suggesting that you have a number that you think 23 is a good number. Now you say, I'm going to make that a I
24 conservative number.
25 MR. KENNEDY: No, our median number, we're using the
, (
Heritage Reporting Corporation
~
(202) 628-4888 '
, , , < . .,, -. ~, , ,-
. p
'324
.<~
- kr ~ 1 .results.--.we're using the results from Wen Tsin's work as'the 2 median ~or best estimate number. We are~ putting in.some
'3 variability.on that number. N o w ,z that variability that we're 4 _
putting in on that number widens the uncertainty band slightly 5 ;in our' fragilities, and thus will lower our HCLPF. capacity 6 slightly.
7 It turns out that with the variability numbers that 8 we've used, the broadening is negligible and the. lowering of 9 the HCLPF is negligible. That's another one of the reasons we 10 chose not to do the variability studies, is that we felt we 11 could conservatively estimate these variabilities and they 12 would have no -- and the conservative estimates of the f-s 13 variabilities would have essentially no effect on our bottom U 14 HCLPF estimates. Because the largest variabilities are in the 15 ground motions themselves.
16 DR. XERR: If you're telling me what you're doing has 17; no influence on the final result, and I suppose I should 18 conclude that I don't hear what you're doing, but I thought it 19 was being introduced as a conservatism, and to me a 20 conservatism means something that you think will produce a L
21 higher risk, than you expect to occur physically.
22 Now, if the conservatism in this case is something 23 that doesn't have any influence in the final result, then I 24 don't.have any final concern.
25 MR. KENNEDY: Well, I share your concern. I do not f'
i' Heritage Reporting Corporation L (202) 628-4888 i
l -
)
5
~
- s 325
.( \
SA 1- like'to see known conservatisms-introduced to the PRA,.and.I
~
2 _did check that point carefully. I believe our variability 3 estimates.are conservative. Our medians we aretusing are the 4 best estimate medians. I think our variabilities are 5 conservative in this case and they have negligible effect on' 6 the end result. They shift the HCLPF levels by about .01.G,' and' 7 we.can't estimate those levels to that kind of level anyway.
8- DR. KERR: If it doesn't have any influence on the 9 final result,-I have no concern.
10 DR. SEISS: Bill, the point is, there's a difference-11 between a conservative estimate of the median'and a 12 conservative estimate of the variability. And he's talking 73 '13 -about conservative on the variability which --
U 14 DR. KERR: Chet,_if it doesn't have any influence on 15 the final result, I'm not concerned.
16 MR. KENNEDY: I don't see either of those-.two factors 17 having any influence on the final result.
18 DR. SEISS: If he were dealing strictly with means-19 and medians without having any HCLPF values, which evolve at 20 uncertainty, there'd obviously be no effect on what he assumed 21 about the uncertainties.
22 But even here, a slight conservatism and the 23 uncertainty doesn't affect the -- variability.
24 MR. KENNEDY: At this stage --
25 O
Heritage Reporting Corporation (202) 628-4888 4
l V o
326.
) 1. 'MR. KENNEDY: At this stage, there are more' subjects 2 to cover, and I don't know whether to cover'them or not. We 3 did do very substantial nonlinear time history. analyses on the 4 turbine building to better estimate its fragilities.-
5 We did these by multiple time history analyses,~using 6 25 different time histories, 21 of which were empirical records 7 and four of which were pseudo time histories from numerically 8 generated pseudo time histories.
9 We wanted to emphasize empirical time histories. We 10 did those studies both varying the structures properties of the 11 turbine building and not varying the structures properties of 12 the turbine building.
13 The results of those studies, and maybe I will pop
. (~') .
- (_/ 14 back to the results -- your handout package has the details, 15 has some details, anyway, of those studies. The bottom line 16 from these nonlinear analyses is that we had a rather 17 significant reduction of our. median estimates.
18 In other words, the simplified approach that we had
.19 used in Phase 2 to estimate the medians, over-estimated the 20 medians. We now know why. We obviously didn't know why before 21 doing the time history analyses.
- 22. We had essentially no effect whatsoever on the HCLPF j 23- estimate, because we also, from doing these nonlinear analyses l~
24 greatly reduced our variabilities.
2$ Basically, the reduction in the median, the turbine
(~)'
Heritage Reporting Corporation
~
(202) 628-4888
x 327[ )' 1 building has a large turbine pedestal in it'which is separate
.2 from the building and there is a gap between the turbine 13 pedestal and the building.
4' The turbine pedestal ~has:a lot of weight, is much 5 more flexible. When you get up to these very high ground 6 motions, ground motions with 6 G, or 4 G, even, average 7 spectral acceleration, much above our risk level, I mean the 8 levels that are dominating the PRA risk. 9 But when you get up to like 4 G ground motion, that 10 turbine pedestal being more flexible closes up easily the gaps, 11 all of the nonlinear analyses run at those real high ground 12 motion levels. The gaps close. 13 The turbine pedestal, being more flexible, and being fm (J 14 heavier, moves where it wants to move _and it forces the shear 15 walls of the turbine building to go'along with it, and the 16 amount that the shear walls of the turbino building move in 17 some cases exceed what we estimated their area drift levels to 18 be. 19- DR. SIESS: What did you assume for that?- 20 MR. KENNEDY: We estimated the median failure drift 21 -levels for the walls in the turbine buildings, we took to be .7 R22 percent of the wall-height for median. We put an uncertainty 23 band on that to where a .5 percent drift represented 16 percent 24 probability of failure, 1 percent drift represented an 84 25 percent probability of failure, .7 percent drift was 50 0 Heritage Reporting Corporation l (202) 628-4888 l
~m .g V '
i 328 L', ,) ' .1 ! percent, a q'iarter. percent drift we-cut off the tails on the 2 fragility: curve such thht a quarter percent drift represented 3 ino probability of failure. 4' So any drift above a quarter percent of the wall 5 height represented some probability of failure.
, 6 DR. SIESS: What would be the drift at the end of the.
7 elastic range? 8 MR. KENNEDY: For the shear walls, it is pretty low. 9 These are very stiff walls. Less than a tenth percent. 10 DR. SIESS: You have two lines on that table labeled 11 ' median. 1'2 MR. KENNEDY: Yes. But this is the median capacity. 13 This is the HCLPF capacity. 5 percent probability, if any, and
' {..}
14 95 percent confidence. This median belongs over here, this 5 l
- 15. percent confidence belongs over here. This line is 5 percent j 16 probability of failure.
l 17 Our best estimate of a 5 percent probability of 19 failure case. This is our 5 percent confidence, 5 percent 19 probability of failure case. 20 All three of these numbers are estimates of 5 percent 21 probability of failure, and a description of our confidence 22 bands on the 5 percent probability of failure case. 23 MR. EBERSOLE: Mr. Kennedy, I wonder if you could 24 help me stretch my memory from some 14 years ago.
~
25 I remember going in the turbine hall of Diablo, and I O Heritage Reporting Corporation (202) 628-4888 i
.i
i
- r. .1 4
329 l
, ,~
1(,) . I think your interest in.it as a structural component or element
-2 is-that it has lots of stuff in it that-many turbine halls 3 don't have. 'It's.got component cooling in.it. .It's got the
-4 diesel plant. s 5 It thus must have critical raw cooling water. What 6 else has'it got in it?
7 .MR. KENNEDY: It has a number of critical electrical 8 components in it. 9 MR. EBERSOLE: So it's really a mixed environment 10 which now includes the turbine itself. 11 Does the turbine itself, spinning at, what is it, 12 3600 or 18? Does it present any kind of a secondary hazard if 13 you shake it.around and the rotor tears into the stator and all e () 14 hell breaks loose? ; 15 MR. KENNEDY: I've never heard or a rotor tearing 16 -into a stator on a turbine from ground shaking. 17 MR.-EBERSOLE: I haven't either. 18 MR. KENNEDY: There is vibration -- 19 MR. E6ERSOLE: Trips. 20 MR. KENNEDY: -- trips on the turbine, such that 21 under this kind of ground shaking you would certainly expect it i 22 to -- 23 MR. EBERSOLE: Well, I remember we even looked into l 24 whether the turbine stop valves would not be frustrated by the i 25 presence of the vibration and concluded that they would work, a L () Heritage Reporting Corporation l (202) 628-4888 !
J f 330
\# 1 long time ago. \
2 That gets the st'am e off, but'it doesn't stop the 3 . turbine. 4 MR. KENNEDY: We have not made a fragility estimate 5 on the turbine itself. I'just don't think that it contributes 6 a seismic -- 7 MR. EBERSOLE: Well, certainly the seals may come out 8 and you're going to have fires. And you don't have any fire
- 9. protection. I'm going _to mess things up for you. I have to.
10 MR. KENNEDY I think that any fire that you would 11 get would be very localized to around the turbine. 12 MR. EBERSOLE: It's a hazard. 13 MR. KENNEDY: And the critical = equipment is not close
.14 to the turbine. It's in the same building but it's not close 15 to the turbine.
16 MR. EBERSOLE: I remember. You put some walls in
~17 fact, did yeu not?
18 MR. KENNEDY: Yes. There's major walls. 19 MR. EBERSOLE: But you do have to preserve the 20 general structural integrity? 21 MR. KENNEDY: Yes. And the turbine and its pedestal 22 are a hazard to the general integrity of the turbine building. 23 And in fact that is what lowered our median risk. 24 But it only becomes a hazard when you get to very 25 -high ground motion levels well above the kinds of ground motion
) ;
I Heritage Reporting Corporation (202) 628-4888 i L
- _ ~,. .-
i. 331 k/- 1- levels that are really contributing to the seismic --
~2 MR. EBERSOLE: Are there any; gyroscopic forces due to 3 the-spinning mass?
4 MR. KENNEDY - Are there any what? 5 MR. EBERSOLE: Gyroscopic forces that compound the-6 problem? 7' MR. KENNEDY: There's probably some, but they would 8 go on to the pedestal itself.
'9 MR. EBERSOLE: Yes. Right.
10 MR. KENNEDY: I don't really have the answer but I 11 don't think they would be very large. 12 MR. EBERSOLE: I don't know. 13 DR. SIESS: Were the turbine stop valves in the list O 14 of components for which you estimated fragilities? 15 MR. KENNEDY: I don't believe so. I don't recall 16- them. No. 17 MR. EBERSOLE: Were the batteries that execute trip? 18 MR. KENNEDY: I need a systems engineer to help out 19- on these questions. 20 DR. SIESS: We'll save that. 21 MR. KENNEDY: I don't know. The main station 22 batteries we estimated fragilities on. 23 MR. EBERSOLE: But they don't do that. 24 MR. KENNEDY: If it's from some other batteries, they 25 are not in. O Heritage Reporting Corporation (202) 628-4888
b 3 332 _ . (j X 1 MR. EBERSOLE: Well, it comes.from other. batteries. 2 MR. KENNEDY: Then they're not in. 3 'The other study.that we've done as a side study is
-4 we've_done a large number of time-history analyses on the 5 auxiliary building, on a very simplified model of the auxiliary 6 building, to benchmark or validate.our standard separation of 7 variables approach to making fragility estimates.
8 We did 200 time histories analyses using 25 different 9 time histories and' varying the priorities of the auxiliary 10 building. 11 We did these in a way that we weighted the properties 12 such that we got mostly high responses because we really wanted 7, . 13 to see whether the separation of variables approach worked ~ - 14 reasonably in predicting high responses. 15 We're really not concerned about accurately 3 16 predicting low responses. , 17 And we found that in fact the results of our 200 time 18 histories which have been weighted using standard statistical 19 techniques to do that, to look more like 2,000 responses, in 20 going out, fairly far out into the high response tails, that 21 the results of these time histories did fit rather well to a 22 lognormal distribution. 23 If they were perfect lognormal, this log of response 24 versus standard normal variant, it would be a straight line. 25 They are not exactly a straight line, but they fit () Heritage Reporting Corporation (202) 628-4888 i
333 I 't g' k / 1 rather well, and they. fit also rather well to estimated 2 responses using the standard. separation of variables approach-3 where these estimated-responses were produced.before running L 4. the time histor'ies. I f_. 5 And so this'is a case of not fitting through the time 6 . histories after the fact, but making the estimated response 7 before. 8 That is one of the cases. I'm not showing the 9 others. I will show a comparison table. 10 MR. KERR: Excuse ma. What conclusions should I draw 11 from the fact that they do almost hit a straight line? Is that 12 good, bad or indifferent?
- 13 MR. KENNEDY: It's good out to -- it basically means
~
14 that our separation of variables approach for predicting the 15 fragility, the responses, the fragility responses, seems to 16 have worked rather well out to I'd~say at least two and a half 17 standard deviations above the median. 18 I'm not so sure that we, there's just not enough 19 points out here. This particular block, it says we are 20 somewhat conservative out here using the separation of , 21 variables. We have other cases where the points out here were 22 higher than the separation of variables. 23 We did this for equipment low in the aux. building, 24 high in the aux. building, four different equipment 25 frequencies. (1) : Heritage Reporting Corporation ; (202) 628-4888
334 I) -1 The conclusions that we drew -- I'm trying to skip 2 through these and I' don't have my things in that order --~but 3 the conclusions that we drew are that we got good agreement in 4 our median estimates within.about 8 percent in all cases. 5 ~We tended to over-estimate our Betas, our 6 variabilities, using the separation of variables approach, 7 particularly for the high frequency equipment, particularly the 8 casos of.14 and 25 hertz equipment, while we got rather good 9 . estimates of our Betas at the low frequency equipment. 10 We did over-estimate at the high frequency equipment. 11 That, even out at 2.4 standard deviations, the separation of 12 variables approach at the low frequencies gave rather good 13 agreement. At the high frequencies, it tended to over- i ( 14 estimate. We have some tendency to over-estimate response of 15 high frequency equipment by this separation of variables 16 approach. 17 The other conclusion out of this that we're making , 18 use of in the PRA is one dealing with response correlation. 19 Basically we looked at the correlation coefficients of response 20 for each of these items of equipment. 21 The symbology here is the 1 means low in the aux. r 22 building; the 2 means high in the aux. building; the 5 is the 23 equipment natural frequency, the 5 hertz; 8 is 8 hertz; 14 is 24 14 hertz, 24 is 24 hertz. 25 What we found is that equipment with the same natural ( lioritage Reporting Corporation ' (202) 628-4888 t i !
??
L .335 (* 3-
. frequency, low in the. building versus high in the building,-
// 1:
2 there was a rather' good correlation ~ coefficient between
- 1
-3 < response.-
4 .For instance,'high in the building at 5 hertz,-the 5 correlation coefficient was .91; at 8 hertz, .88; 14 hertz, 6 .87. It wasn't very good correlation when you got to'24 hertz. 7 In other words, if you knew the response of the
'8 equipment at one location in the building, there was high
~
. 9 correlation with the response at a different location. So 10 that rvsponses of equipment with the same frequency were rather ;
. 11 highly correlated even if they were at significantly different 12 locations in the building.
_; _ 13 On the other hand, equipment at the same location in 14 the building but with different frequencies, there was very low i
~
15 correlation in every case. ; 16 An 8 hertz piece of equipment next to a 5 hertz piece 17 of equipment at the same location, there was very little 18, correlation between their responses, conditional of course on 19 these ground motions all having the same average spectra ; 20 acceleration in this averaging region, that at any specific - 21 response was lowly correlated. . 22 Our conclusions from this are that identical 23 equipment, essentially identical equipment, because identical 24 equipment would have highly correlated capacities. The overall j i 25 capability is associated with the strength capacities and the
- C) l lieritage Reporting Corporation (202) 628-4888 ,
t
~. '
336
.g '
UNM A ~ :1 ' responses.
+
l2. 'If it was. essentially identical:1t would'have highly
- 3. correlated capacities.. If:it.was essentially identical,.it
'4 would have. essentially the same frequency so that essentially 5 ;1dentical-components, we should treat'them.in the PRA as their 6 fragilities being totally correlated with each.other even.if.
7 .they are at different locations within the building. 8 On the other hand, different components which would 9 have different correlations in the capacity and would-tend to
~
10 have different frequencies, different components adjacent to 11 each other should be treated as essentially independent
. 12 . fragility curves.-
13 And these are the assumptions that we are.now
.-'() 14l recommending be included for future versions of the fragility 15 work. I'm sorry. Future versions of the PRA work.
16 That completes the material I wanted to present. 17 DR. SIESS: Thank you. There have been questions 18 during the presentation. Are there any other questions? Mike? 19 MR. TRIFUNAC: I don't have a question, I have a 20 suggestion in the form of a comment, which I mentioned already 21 several times and particularly also during one of the recent 22 workshops on fragilities. 23 And that is that I still have a problem with assuming 24 lognormal distribution for describing structural response. And 25 I think it would be beneficial for you to abandon it, using
.O Heritage Reporting Corporation (202) 628-4888
en , 337
~
1 separation of variables. (s r') 2 I am saying all of this because I did calculations in
~
' 3_ my head. I don't know how significant this would be. But I 4 believe that it would be beneficial for you for the following 5 reason - 'that if you did take a more realistic distribution 6 function describing response of single degree, of multiple 7 degree of freedom systems, the distribution would become 8 skewed. It would become skewed in a way which is similar to 9 the way Rayleigh distribution compares with Gaussian 10 distribution.
11 I speculate that what this would do, that it would 12 deform -- you have a sketch that you showed many times, and you 13 showed it today, on the HCLPF point, where it is. I think that () 14 this would have a tendency to move the HCLPF point to somewhat 15- higher values, which may be beneficial to you. 16 The negative output of that, if it is indeed correct, 17 would be that it would possibly upset a little bit the list of 18 your candidate critical components, like diesel generator, and 19 so forth. It might take some of these off the list, but it 20- might introduce some others on the list. And I have no idea 21 how that would go. 22 But it seems to me that it is something that, first 23 of all, is quite justified on the basis of what we know about 24 response today and it may be very beneficial. I don't see why 25 you don't take advantage of it. Heritage Reporting Corporation (202) 628-4888
338
-1 .That's'really not'a question. It's more a 2 suggestion.
^3 }UR. KENNEDY: Basically, that was one of the primary 4' . purposes of doing this separation of variables study.-And we've 5 done other studies ~1ike that before, including with the 38 runs 6 that Wen Tsing did'through a more complex building model.
7 Every time we have done with these studies, we have 8 found that the responses fit rather well the lognormal 9 distribution. 10 In fact, I don't know of a better distribution to 11 use. If I knew of a better distribution, I would use it. 12 But frankly, everything I've done indicates to me l 13 that they do fit reasonably well the lognormal distribution. O 14 Now, in the past, I've been a strong advocate that.we 15 should cut off the lower tail of the lognormal distribution l ; ! 16 because the tail just keeps going on down to zero, t l 17 Our problem is we don't really know where to cut tbac 18 tail off at. And our separation of variables study showed that 19 we probably should not be cutting off that tail below 2.4 l 20 standard deviations because we tend to fit fairly well out to l l 21 2.4 standard deviations. 22 There are certain frequency ranges r lative to where 23 the floor spectra peaks, where there is a bias to the lognormal ! 24 distribution. 7 25 If the floos spectsA !9 going concave upward, if you ; () Heritage Reporting Corporation * (202) 628-4888 . 9 k W
v 339 () I have estimated your. equipment'at a frequency such that as you 2 shift from that frequency, either lower or to higher 3 frequencies, the floor spectra tends to be concave upward, the
~
^4 lognormal distribution tends to under-estimate the tails and 5 you have to make some corrections for that, which we do make..
6 If you are at the peak, although it didn't show up 7 here in this 8 hertz case, we've seen in some_other cases that 8 if your equipment natural frequency is right at the peak of the 9 floor spectra, such that the floor spectra tend to look concave 10 downward, that the log. normal distribution will tend to 11 overpredict the upper responses. 12 And we sometimes correct for that. But we did not 13 find that effect in this separation of variables study. () 14 I just, I know you've made that comment to me before. 15 I've tried looking into it. I just don't see in the work that
'~
we've done any evidence that we ought to be using something 17 ott.:e than the lognormal distribution. I 'J DR. SIESS: Okay. Let's leave it there for a while. 19 Are there any other questions or comments at this point? 20 (No response) 21 DR. SIESS: Then let's hear from the staff's comments 22 or reactions to the status report. 23 MR. CHOKSH1: Only a very brief comment. We are just 24 getting into the review of fragilities. 25 We received last week a package of a calculation for O Heritage Reporting Corporation (202) 628-480'J
s
-~
340 Kr 1 about 20 componen:w. Tomorrow we are going to meet and discuss 2 all~those calculations and then we are currently planning'a. 3 plant walkdown in this upcoming-outing. 4 So we'really do not have any deta.11od commentsoat 51 this time. 6 D R .~ SIESS: Thank you. 7 MR. SEAVUZZO: Mr. Chairman?- 8 DR. SIESS: Yes. l 9 MR. SEAVUZZO: Is it the function of thio concsittee i 10 to review or eventually review those calculations? 11 DR. SIESS: No, sir. It's the function of the staff. 12 Any advice we could give the staff, they will appreciate. 13 MR. SEAVUZZO: o 14 advice. We have to see the reports to give any 15 DR. SIESS: We can see that you get them. 16 Gentlemen, we could save ten minutes by omitting a < 17 break, but I don't think it's worth ** right ow. But I will 18 suggest that you be back in ten minutes. 19 (Whereupon, a brief yecoss was taken.) 20 MF, CLUFF Bruce Smith is the PG&E engineer that is 21 in charge of managing our probabilistic rick assessment 22 activities on this program, and he will make the presentation. 23- MR. SMITH: I'm Bruce Smith. And I am Project 5 24 Manager for the PRA. I have been working on the PRA since its 25 planning back in 1984. ( Heritage Reporting Corporation (202) 628-4888
, . h ';v +
J\ p . ij 341' k [yqj(~'.) 1 By: training and experience I'm a systems engineer, 2 howe'ver. I'm an electrical engineer. And referring to a 3 question we had earlier about operating substation equipment, I-4 'was a substation operations engineer for about ten years. So I 5' am familiar-with that equipment. 6 My background on Diablo Canyon has been environmental-7' qualification, TMI mods and the IDVP effort. 8 I want to make clear that I am not a PRA expert by' 9 any means. I have Dennis Bley..from Pickard, Lowe &.Garrick 10 sitting at the table over here in case we have any PRA-11 methodology type questions. 12 I'm here mainly to talk about how we're planning to 13 use the PRA and what we have done so far. p) i 14 MR. KERR: Mr. Smith, is there anybody in your 15 company.who is maybe a PRA expert or at least semi-expert? 16 MR. SMITH: We have several engineers who have been 17 trained and worked with PLG during the course of our PRA. 18' MR. KERR They understand the result, the final 19 result, well and its implications? 20 MR. SMITH: That is correct. And these people have 21 actively participated and actually have performed part of the 22 PRA under the supervision of PLG. 23 MR. KERR: Thank you. 24 MR. SMITH: We have chosen Pickard Lowe & Garrick as 25 our consultant and have been working with them for several Heritage Reporting Corporation (202) 628-4888 ;
yyy ; v-y y -
- 7;M"
- m.;;
I f1 &_ r M'_f 342 j("Si - t 1s l - 1 years. I'll skip most'of.this. Except that the point that we
- L ,
12 are. training PG&E-personnel _to work _on the PRA. - : s 1
- 3. .This next-slide I think-is'important-to a:later part im ', '
- , 4 of the' presentation." And-that is the? fact that we are doing a-5 phased.PRA.- r
~ ~
- 6 I would like to-get into a-little analogy;here. I-
- 7. have:a friend who restores automobiles. And this. gentleman 8_ -picked'out.a wreck, or what he would call a classic, at;one a
9 point. 10 He=wasLable to get it running and getLit home,-and 11 wcrk on it. Found out that the brakes were no good, the tires , , i 12 were gone, suspension was corrupt.-
~13 He fixed'it all up, made it at'least drivable. Later '
-C:) -
14 he found out he had some reliability problems. He fixed those. r 15 Ano now he's at the point where he's got a great running car, a : 16 very reliable car. But it looks pretty bad. [ a 17 And I think that is somewhat an analogy to what we're l i 18 doing on the PRA. [ We have, in Phase II,.gone in depti in producing a 19 .- ( 20 model. And based on a lot of judgmental effort, we did a l 21 scoping analysis, as you can see. And in the first iteration, i 122- which we have basically completed, Phase IIIA, we did the j 23 documentation and did some quantification which you will see i 24 later. ! 25 In Phase IIIB, we're now at the point where, like my ;
) !
Heritage Reporting Corporation (202) 628-4888 ; b -
343 l- friend's car, we're working on the things that make this PRA 2 look a lot better, and will provide better documentation and 3 numbers at the end. 4 But I caution you to take these results with some 5 grains of salt because they will change significantly, as my 6 friend's car will, over the course of the PRA.. 7 In 1988, this final phase that we're working on, 8 we're going to be analysing the dominant contributors, thingn 9 that stand out in the PRA, doing the final computer runs and 10 the final report writing. 11 MR. DAVIS: I have a question on your schedulo? 12 MR. SMITH: Sure. 13 MR. DAVIS: According to Mr. Kennedy, he is now doing g () _ 14 the soil / structure interaction modifications to the 15 fragilities. Do you intend to incorporate that new information 16 into the PRA? 17 MR. SMITH: We're the final end of the pipe. And we 18 cannot complete our effort. We're going in parallel with the 19 other parts of the long-term seismi<c program. But we cannot 20 complete our effert and produce results and reports until we 21 get the final fragilities and hazards from t.he rest of the 22 long-term seismic project. 23 So yes, wa are dependent on that. 24 MR. DAVISt But you will still finish in 19887 i 25 MR. SMITH: 1985, according to the present schedule, i o)
\"'
Heritage Reporting Corporation (202) 628-4888
344
\>' 1 MR. EBERSOLE: Mr., Smith, you know, there's'SEPs and 2 'there's IPEs and there's ISEPs,let cetera.. And I had rather 3 hoped that the -ISEP progra:n w7uld collectively _ cause a PRA, a 4 really' comprehensive PRA to be produced for every plant, and be 5 actually a part of the'SFAR, as evidence of really good system 6 -integration.
7 I ree PRAs that crop up to solve a particular 8 problem, localized. They sometimes then are used, you know, 9 the severe accident business. How do you really contemplate 10 you* Diablo PRA -- I'll just put that in quotes -- would will 11 be its characterization when you get done with it? I'd like to 12 hear you say it is an evidence of an integral system
, 13 integration program.
"' MR. SMITH: I plan to talk to that particular issue 14 l
15 at the end of this talk. 16 MR. EBERSOLE: Okay. 17 MR. SMITH: Talk about some of the benefits'that we 18 think we're getting out of the PRA. So if I can, I'd like to , j- 19 delay that toward the end. l-20 MR. EBERSOLE: All right. i l 21 MR. SMITH: This next slide just deals with the type l l 22 of review that we go through as part of the process. 23 We have an independent technical review by Pickard, 24 Lowe & Garrick people who are not working technically on the 25 project. PG&E review, technical review, management licensing
-O) \~
Heritage Reporting Corporation , (202) 628-4888
f kI 345 (j '
-1 review.-
2~ And'then we have some workshops for the staff and ; i
-3 their consultants where we have presented this same type of 4 information.
5 And then'we have also submitted three vclumes of 6 analysis, preliminary analysis, that we have completed.
-7 I would like to go through these slides fairly.
8 quickly on some of the real features of Diablo Canyon. Most of 9 you,'or some of you may be aware of these. These.are things 10 that were highlighted by the PRA. 11 The first two items deal with CCS and aux. salt water 12 systems. 13 These systems are perhaps different than you might g.
\_) 14 find on some other plants, some newer plants. We tend at 15 Diablo Canyon to have more flexibility interconnections between 16 systems in some of the newer plants. There's not this straight 17 train separation that you find in some plants.
18 MR. EBERSOLE: Do you call the plant two individual 19 stalls of one unit each or an integral power plant with cross 20 speeds? 21 MR. SMITH: They are basically integral power plants 22 with cross speeds between the units. In fact, there are 23 essentially two aux. salt water trains per unit, with a 24 crosstie between the units. 25 MR. EBERSOLE: So it's interlaced to help, presumably ( Heritage Reporting Corporation l (202) 628-4888 - I
yc - i , s . ,
, I 346
) .-
'x/ 1 to help, --
;2 MR. SMITH: Exactly.
3 MR. BBERSOLE: -- rather than to economize? 4 )UR. SMITH: Not necessarily. I think it was mainly 5 to reduce the-risk-in the original design. 6 We also have three vital power buses, which is-7 different than'in some plants. We'have a swing diesel ! 8 generator, which means we have five diesel generators with one 9 diesel generator swinging.between a unit 1 bus and a unit 2
- 10. bus.
11 MR. EBERSOLE: How about your DC system? I don't see
-12 it-up-here?
j, . 13 MR. SMITH: The DC system is a three-train system,
~'
14 too.
'15 MR. EBERSOLE: Per unit? Per unit or per plant?
16 MR. SMITH: Three trains;per unit. So there are six 17 batteries, essentially. 18 One point that you will see on the tour is that we 19 have a fairly mild climate. We're right in the coastal area 20 and it really doesn't require a lot of e.ir conditioning. We do 21 have air conditioning for the control room. 22 We don't have a lot of areas that are susceptible to 23 room heatup. And by room heatup we're talking about loss of 24 all ventilation whatsoever. We have redundant ventilation ! 25 systems. But this is PRA space where we're dealing with loss (E) ! Heritage Reporting Corporation (202) 628-4888 : i f r {
; o.
347 () l' of all ventilation. 2 One last item that I added on here that I didn't 3 think about earlier. All of the cables that are installed, 4- safety related cables that are installed in Diablo. Canyon, are , 5 installed-in conduits. We do not use cable trays for Class 1E - 6 cables. 7 Now, there is one exception, in the 4 KV cable 8 spreading rooms. Those rooms do have cable trays, but they 9 only have one vitality. 10 We did-find in the PRA that there were some concerns 11 about 480 volt switch gear room and inverter room heatup. And 12 again, that's loss of all ventilation whatsoever. 13 We looked a relay chatter. We found that it is
- 14 fairly easily recovered, that we have no lockup that isn't 15 readily reset and that we have no restart restrictions. And :
16 that is dealing with the diesel generator loading. What we 17 found is that some plants have restrictions on how many motors 18 that you can 1 cad up at a certain time, i 19 MR. 8BERSOLE: Have you looked at the fire protection 20 system in the context of diesel engine protection which might [ s 21 interrupt this operation? 22 MR. SMITH: We've had this question from the staff, 23 and we are looking at it. But yes, the fire protection system 24 is a seismically-qualified system at Diablo Canyon. 25 MR. EBERSOLE: Oh, it is. Heritage Reporting Corporation (202) 628-4888 i
We
. ,m .
348 1- MR. SMITH . Yes. 2; MR. EBERSOLE: That's unusual.
'3 ;MR. SMITH: Yes, I believe it is.
- 4. MR..EBERSOLE: What about the air systems?'
5' MR. SMITH: The air systems are Class 2. I'm not -- 6 MR.'EBERSOLE: Do they interface with Class 1 air. 7 operated apparatus? 8 MR.-SMITH: They do, yes. 9 MR. EBERSOLE: And so how do yon , Ovide the barrier 10 for intrusion of foreign materials, etc., rather than just , 11 stepwise air failure? 12 MR. SMITH: The air system has 'ilters and so forth
- -13 to prevent the foreign material from getting into the valves.
14 We at least have not experienced a. lot of proelems in that 15 area. 16 MR. EBERSOLE: When you need air but your air system 17 has failed, do you get it from accumulators? 18 MR. SMITH Yes. We have accumulators or bottles. i 19 MR. EBERSOLE: You are aware of the check valve , 20 leakage problems, then, and the possible fact that your users 21 consume more air than you thought, and thus run out of t 22 capacity? 23 MR. SMITH: We have heard of that. I think in the 24 PRA that is handled under things like, you know, equipment 25 failures. O Heritage Reporting Corporation
- (202) 628-4888 ,
- s. .
't1 ,
- , y ,
+ ' ~ -
y _3 q> < o 349
' ~hx-~ '
l' MR.:EBERSOLE ' Okay. , I;only want to hear!that you' 2'- . heard:of it.- - We-have a fa'irly?
. . MR.' SMITH:
'Yes., Let's see.
- 4. reliable, plant,.-as,Mr. Brand'has told you.-
~
s.
- 5:
! M R ..- E B E R S O L E s - Are'you extending the' principle.of s
- 6~ : coincidence _ out into balance of: plant' to ' reduce ' trips?
7 MR.. SMITH ' You'd have to - -
'8 MR. EBERSOLE: Extending the principle-of coincidence
~
9_ to avoid spurious trips from single track apparatus,ulike, you 10 ' know, a-heater level switch.
- ll- -MR. SMITH: I'm not sure that I can answer ~that.
12 ,MR. BRAND: We haven't really examined that in the-R
- 13 PRA.-
14 MR. EBERSOLE: It's one of'the current: efforts, you
- 15 know, to reduce trips.
16 One thing-is to look at the family of single track 17 . trips- out in the b'alance of the plant and do something. about 18 . spurious' occurrence. 19' You know, the janitor can trip your plant if you'
'20 don't watch it.
' :21 MR. SMITH: We're aware of that. That's not analyzed 22 in the PRA at this point in time. And I can't speak to what the plant has done in that direction.
24 MR. EBERSOLE: Only as to the challenged frequency. 25 MR. SMITH: I'm sorry? Heritage Reporting Corporation (202) 628-4888
P ,
. . 350 v
(~'/
~- 1 Pet._EBERSOLE: Only as to the challenged frequency of 2 safety apparatus. -You don't ramp down your main feeds, do you? !
3 -101. SMITH: You mean-as far as the. trip?
~
4 MR.-EBERSOLE: 'I think_your: Westinghouse plant calls
-5 for response of the aux, feeds on every turbine trip. No other
'6 plants do that that I know of.
7 MR.. SMITH: That's true. 8 MR. EBERSOLE: All the rest ramp them down so they 9 keep their coolant source. 10 MR. SMITH: Right. One point we heard that is 11 notable-is that our diesel generators have a self-contained : 12 heat exchanger. We do not use component cooling water to cool 3 13 the diesel generators. 14 'MR. EBERSOLE: Component cooling is a safety heat 15 rejection system. 16 MR. SMITH: Yes, it is. 17 MR. EBERSOLE: Yes. 18 MR. SMITH: We have an isolated site as again you 19 will see on the tour, 20 MR. EBERSOLE: Can you recover the secondary water by 21 use of fire or other water on the secondary site if you lose , ! 22 component cooling? 23 MR. SMITH: We are looking at that right now. That's 24 one of the issues that has come out of the PRA and in fact is 25 what I would call a benefit of the PRA, looking at some of O Heritage Reporting Corporation (202) 620-4888 i 1 L,
1 l i 351 () I those scenarios. ) 2- MR. EBERSOLE: ~ Okay. Great. l 3 MR, SMITH:' We have other means besides fire water to
..4 do.that and we're looking at all those.
5 MR. EBERSOLE: But fire is seismic in this case? 6 MR. SMITH' Right. 7 MR. EBERSOLE: Oh, by the way, is fire dependent on 8 electric power? Fire water? 9 MR. SMITH - We are looking at whether or not there's 10 a gravity feed possibility, especially on the seal loca 11 concern. , 12 MR. EBERSOLE: Are you studying the valves when they 13 have to operate in duress rather than routinely and the e-(~T/ 14 . misapplication of statistics on valves that simply flap back 15 and forth without loads and give a spurious evidence of 16 reliability? 17 MR. SMITH: I'd have to talk to the data analysis 18 people on that. 19 MR. BLEY: We've tried to identify any cases where t 20 the routine test conditions wouldn't match the conditions that 21 we would need under the accident scenarios we're examining. l , 22 MR. EBERSOLE: Good. , 23 MR. BLEY: So we have raised that issue, t 24 MR. SMITH: Okay. Under methodology I'm going to 25 give you just a few graphs and they are just to - [ t ( Heritage Reporting Corporation (202) 628-4888 & i.
352 1 DR. SIESS: Excuse me. 2 MR. SMITH: Sure. 3 DR. SIESS: I think that one of our consultants, Mr. 4 Davis, is the only expert on methodology and he's looked ahead 5 at your graphs here and says he's not particularly interested. 6 It looks like fairly sttadard stuff. 7 MR. SMITH: All right. The only graph that I would 8 ask that we put up is the seismic analysis approach. And it's 9 to deal with a question that came earlier about the list of 10 equipment on fragilities 11 DR. SIESS: Fine. If you can find that one, we'll 12 follow suit. 7- 13 MR. SMITH: And the question was that the seismic
\
14 fragilit.les list looked rather short, there was something like 15 75 teems. 16 One point I want to make clear is that every item 17 that is in the PRA model, and that includes all of the safety 18 related equipment, in some form is included in the fragility 19 list. So that there is e fragility for every item in our 20 system model. 21 Now, the system model may be generic. In some cases, 22 for instance, the conduits, cables, valves, piping, those types 23 of things are generic. 24 It's also generic in that we did one diesel generator 25 to represent several diesel generators. So I feel that that l R) Heritage Reporting Corporation (202) 628-4888
a 353
\> 1 list is not clear that that is covering every' plant item. And 2 in fact-for instance the diesel generator,-I think we've split 3 off some of the components such as control panels, to identify
'4 them as having a special --
~
5 MR. EBERSOLE: I just thought maybe there might be
'6 more nearly four or five hundred specific items you look_at.
7 MR. SMITH: There are four or five hundred items but 8 they are covered more generically. 9 MR. EBERSOLE: Okay. 10 MR. SMITH: Now, we have two approaches. 11 MR. EBERSOLE: Do you integrate those items for the 12 seismic event:1et's say to the same ones you have to have for 13 shutting down for fire? You know, the minimum set?
' 14 MR.. SMITH: Well, maybe you can talk to that better.
15 MR. BLEY: Essentially, that is right. What we did, 16 though is start with the list of everything we built into the 17 model and then simplify that into a short generic list to give 18 to the fragility people so go look at a relay cabinet, rather 19 than listing all of the ones that you would find on the Q list. 20 MR. EBERSOLE: I understand. Well, then in essence I 21 was hoping, you know, you say I look for what I need to shut 22 down. 23 MR. BLEY: Exactly. 24 MR. EBERSOLE: And I don't care about the rest except 25 for the interfacial aspects of things that might disrupt the O Heritage Reporting Corporation (202) 628-4888
V - h&} V -
> 4 y- 354 t h'
' :l' things'that'I shut down.
2 :MR. SMITH:- One point we ought to make clear is we 13 : even included some of the nonsafety. systems as part of this-
- 4 shutdown'-
5 MR.'EBERSOLE:' Right. That gets into.the matter of lyou looking at the influence of control systems on your safety. 7- system so they are in common disarray. - You ' know,; the TAP-4 7-8 . thing. Are you.all-looking at that in.the PRA context? .You 9 know what I mean, the~ TAP-47? .The old classic thing, do I.have
- 10 control systems when they get outside their bounds of amplitude 11 and range.
12 MR.'BLEY: The main place where we're looking at that 13 'is in the relay chatter issue, but also as far as signals that 14 we need to perform the functions-that we're modeling in the 15 plant. 16 MR. EBERSOLE: In all the~ process parameters, they 17 have a driver, they have an amp}itude and rate modulator and 18 they have a receiver. And when the amplitude and rate. 19 modulator gets out of its normal bounds something's got to 20 intercept, consistent with the characteristics and capability 21 of the driving function. I'm just trying to put it in a small 22 package. 23 MR. BLEY: And I guess the simple answer is yes, we 24 are looking at that where it really matters to us and a lot of 25 the places where the greatest detail exists in that area is out O !ieritage Reporting Corporation (202) 628-4888 L
4 .
.i_ ;
r 355-E('s , A>; . 1- :in the1 balance.of the plant. .And what we areireally looking'at
'2: 'on a more actuarial: level isigetting the kinds.ofLinitiating 13- ;eventatthat-we'are modeling throughout the rest of the plant.
MR.'EBERSOLE: One of1the-common ~ problems of the 5 standard analysislof the critical. parameters is:that people- .;
- 6. look'at stepwise changes in its presence or. absence. They -
7 don't look at too much to the right,:to the left,.upcand down,
- 8- backwards.and forwards, and accelerator operation.
9 DoLyou cover the full' range of. parameter movement?- ,
'10- MR. BLEY: We're trying to do that in the cases where ;
I 11- .it directly affects the equipment functions we're modeling. l
'12 But'out in the-balance of the plant, we're not-doing that kind ! . '13 of thing, '
I 14 MR. EBERSOLE: Many times a safety device will 15 respond to a zero function but not to too much, i 16 MR. BLEY: That's right. j 17 MR. EBERSOLE: Okay. ! l 18 MR. SMITH: I would like to caution -- at.this point , l t L 19 we are going to talk about the 3A model that we've put together i 20 and that will produce the results that we are going to talk 21 about later, j l i 22 I would like to caution that we get a lot of 23 comments, particularly from our operations people, that say 24 gee, you know, you guys are doing number crunching but what ,.
?
25 does this really mean? And can you guys just come up with any l C) ! Heritage Reporting Corporation ! (202) 628-4888 i i i
f"' 356 1 number you feel like? 2 And my reply to that is that number one, we're 3 bounded on the fact that we have to be defensible to you and 4 the staff. 5 And on the other end, we, and I caution people to say 6 that we don't always take all of the conservatism out. We 7 don't always come up with the best estimate. In a lot of cases 8 we have taken some conservative cuts. If they don't appear to 9 bother us as far as the results are concerned, we leave them 10 alone. They remain conservative. 11 MR. KERR: I wish you wouldn't do that, for the 12 following reason. 13 I d.o not believe, on a localized basis, you can ever
> 14 be sure that what you are doing locally is conservative 15 generally. Because the assumption you make on a local basis 16 may have an influence in the rest of the plant that is hard to 17 predict.
18 We've seen this in the regulatory arena in which 19 regulations meant to be conservative actually introduce 20 additional risk. 21 And unless you are very certain that you aren't doing 22 that, what you may pick as conservative may turn out not to be. 23 MR. SMITH: I think I should perhaps explain that a 24 little bit more. 25 By conservatism we're talking about assuming the (-
._,s Heritage Reporting Corporation (202) 628-4888 L.
357. k,) -1 failure'o'f a particular component or system. Now, I don't. 2 think that there are very many cases where'we could have a 3 system that, not failing would impact the operation of. thel
~
4 plant or recovery from some scenario. 5' MR. BLEY: Can I add to that? 6- MR. SMITH - Sure. ' 7 MR. BLEY: your-point is well taken. And we have 8 made an effort not to set down a litt of ground rules in the t 19 beginning that operation in this mode is conservative and-10 operation in the other mode is nonconservative, but under every . 11 specific accident scenario we're modeling we look and see for
- 12 _that scenario how would operation of that part of the plant-
,. 13 behave.
14 I think, at least under present capabilities, doing + 15 very realistic modeling of the performanco of all aspects of 16 the integrated plant, is just too big a task. So we have to ; 17 close off some areas through conservative bounding I 18 calculations. 19 But we always look under the next accident scenario , 20 to make sure that what we assumed was a conservative in the i 21 previous case doesn't come back and bite us. You have to be j 22 very careful about that. 23 MR. EBERSOLE : Bill was saying, do you look at the . l i 24 negative aspects of everything you considered an improvement up 25 to that time? O Heritage Reporting Corporation (202) 628-4888
358
/
')
s- 1 MR. BLEY: And we certainly try to do that. 2 MR. EBERSOLE: Well, a safety valve is a case in 3 point. It_might stick open. 4 MR. BLEY: That's right. 5 MR. EBERSOLE: Or the brakes on a car might stick on 6 the right front. 7 MR. BLEY:. That's right. To my language that is a 8 different scenario and we have raised the second scenario as 9 well. In fact, we might look at the same thing performing in 10 the opposite direction in another scenario as a conservative 11 judge on that scenario. 12 MR. EBERSOLE: It's almost an invariant that every
,3 13 good action has a bad component.
(
)
14 MR. SMITH: I just put up here a list of the 15 initiating events that we included in Phase IIIA. These are 16 fairly complete and I imagine they are fairly close to what we 17 will have in IIIB. 18 MR. EBERSOLE: I saw loss of condensor circ. water. 19 A while ago I mentioned the fact that you wanted the 20 loss sometimes and you couldn't get it if you didn't have the 21 batteries to trip the breaks, you know, to avoid flooding the 22 turbine hall as a case in ooint. - 23 MR. SMITH: Although again, being an old substation 24 operator, that is very easily done. The operators do that. 25 MR. EDERSOLE: I recall contemplating a shotgun to [b v Heritage Reporting Corporation (202) 628-4888
359 g (_) 1 shoot the breaker out.
-2 MR. SMITH: Sure. The next slide gives you some idea 3 of some of the systems that we' included, gives you some terms 4 here -- support systems, frontline systems and system 5 unavailability causes. These are terms that we've used to kind 6 of divide the plant up.
7 The next slide is on data ~ analysis. And the only 8 thing there is, the point I'd like to make here is that we're 9 doing plant specific effort and this is somewhat my automobile 10 analogy again in that we're trying to determine if our plant 11 specific equipment ic either a lemon or it's a cherry. And 12 we're trying to come up with some defensible numbers in that 13 area. (s.)' 14 MR. DAVIS: Excuse me. I have a question.s 15 MR. SMITH: Sure. 16 MR. DAVIS: What model are you using for common cause 17 failures? l l 18 MR. BLEY: We're really using a variety of models. In j ! 19 some cases, we actually developed explicit models modeling the 20 cause and the effect. In some cases we're using parametric 21 common cause models related to the extensions on data factor 22 models that have been around. 23 What one has to do if you take that approach, t 24 however, is to go back and very carefully look through all the 25 sources of information that were used to generate -- well, the O Heritage Reporting Corporation (202) 628-4888
g ~ jb _ - H 360 ;
/m. t
. data' base for those kind of events, screen it, so that it is
))- 'l t
~
2- appropriate to the: larger models you'are building and to'the 3- plant in.particular-that you are-looking at.- So we-have done s lF some: extensive work in that' area. 5'
~
.So we do'use a variety of~ approaches.
6 MR..EBERSOLE: Just to take a shot at the first item g 7' 'up there, in a seismic event,-do you contemplate loss of the
- . -8 turbine driven aux. feed water because you have salt water in
. 9 the secondaries, because the condensors failed? ,
10 MR. SMITH: I don't remember that being a scenario. _; i 11 MR. BLEY: Our aux. feed water pumps draw from the: i 12 common stage storage tanks rather than from the condensor. 13' MR. EBERSOLE: The common stage storage tank --
. 14 MR. .BLEY: .That is a pond up the hill. -The safety .
. 15' break backup is a freshwater pond at the top of the hill.
f
. 16 MR. EBERSOLE: So you don't have any saltwater ,
17 sintrusion as a problem? 18 'MR. BLEY: No. . 19 MR. SMITH: One point on this relay chatter slide, i
~
20 it's rather busy, but one point I'd like to make here is that 21 we made the ar.sumption that all of the contacts are going to ! 22 chatter, whether they can or not, and then looked at what the ;
~
23 worst case consequences were. 24 And arguably whether it's a worst case consequence is ! i 25 up to whoever does the analysis. ! t Heritage Reporting Corporation ! (202) 628-4888 ! i i
e- - . 361
's/ 1 _. MR. EBERSOLE: Pardon me. If I can go back a minute.
2 I know you had water from common stage storage. But then the 13 4.Thaust goes to -- no, no. The turbino exhaust goes to the
.4 stack, doesn't it? You. don't condense it. Right. I'd a
'5' forgotten that.
6 MR. SMITH: The point I'd like.to make here about 7 worst case consequences is that in a lot of cases, the people 8 -that did this analysis had to assume that for instance, seven 9' :or eight contacts had to line up to get some worst case
- 10. consequence.
11 And we made that assumption, that all these contacts 12 are going to chatter simultaneously and they're all going to 13 line up and do whatever dire thing they are going to do. O- 14 And then we went through the, looked at what would 15 have to be done to recover, whether or not the system locks up, 16 how it locks up, where it can be reset. And later I'll go into 17 some of the things we found on that. 18 We also did, in Phase IIIA, there were some concerns 19 about room heatup, total loss of ventilation. 20 The rooms that we looked at were those that had PRA 21 components in it, components that we have identified in our 22 model. 23 And we assumed that all ventilation is lost and then 24 did some analyses, standard analyses, on initial temperature, 25 heat loads, and then came up with temperature as a function of fleritage Reporting Corporation (202) 628-4888
,w 362 x , 1 time. 2 And then we looked at the thermal fragilities of all 3 the equipment to see if there were any problems due to this 4 heatup and how long the operators would have to recover. 5 It turned out to be something as simple as opening u 6 doors that rendered this not-a concern. 7 DR. SIESS: The assumption that all the ventilation 8 would be lost, would you call that a conservative assumption or 9 do you have some probabilistic basis for making that 10 assumption? 11 MR. BLEY: That was calculated from the systems 12 model, the ventilation system. 13 DR. SIESS: And it just couldn't take the earthquake? (_) 14 MR. BLEY: Oh, no. This is not primarily earthquake. 15 This is in the non-earthquake scenarios, we are calculating the 16 frequency with which you would lose the ventilation. 17 In earthquake, I don't remember, Bruce. Was that a 18 contributor? 19 MR. SMITH: Not a contributor that I can renaember. 20 What it comes from is either in a lors of power -- well, even 21 that wouldn't be a problem because you depower anyway. 22 But I think it was things like maintenance failure or 23 random equipment failures. 24 MR. EBERSOLE: Do the emergency manuals still fall 25 out on the floor in the control room in an earthquake? That's n)
's Heritage Reporting Corporation (202) 628-4888
363 1 an old-timer. 2 MR. SMITH: Mr. Wogsland, from our operating group 3 back here, may be able to tell you that. I'm not sure. 4 Ward, come on up. 5 And by the way, Ward has been involved in this PRA 6 effort from the start. 7 MR. WOGSLAND: My name is Ward Wogsland. I'm grid 8 engineer. I've been involved in Zion, Zion PRA, and have been 9 doing oversight and support work of Pickard, Lowe & Garrick, 10 and of Bruce, in this effort. But I did work with Pickard, 11 Lowe & Garrick and Dennis on the Zion work and Indian Point 12 with George Plow (ph) a number of years ago. And so I'm fairly, 13 I'm very familiar with the plant integration information into (J 14 PRA. 15 The control room that we have, we have book racks in 16 the control room that are mounted on the floor, anchored down, 17 and our books are mounted in the racks. 18 So I don't believe that's a particular problem. 19 MR. EBERSOLE: This is quite a plant. Do your 20 operators sit in swivel chairs or are they belted in? 21 (Laughter) 22 MR. WOGSLAND: The chairs are free standing. 23 (Laughter) 24 MR. EDERSOLE: Well, apart from the human factor, I 25 see you're coming up with human factors, have you looked at the (m () Heritage Reporting Corporation . (202) 628-4888
i
'364 -,
1 ' operator, response'under1these unusually exciting conditions and i 2' that~ horrible problem _he's got with intercepting, or rather i i 3 interpreting-the'million annunciator panels? ! 4 : MR. SMITH:- You-led right'intoLlt. ! 5 MR. EBERSOLE: 1 think there is en operator overload 6 tat all plants, and notably maybe worse here than.everywhere.
~7- Plus compounded by the earthquake presence. .
8 MR. SMITH: That is correct. I think we've tried to 9 ' quantify that in our human actions analysis. Again, it's a- ! 10 dynamic human actions analysis. . r 11 One of the points I want to make with this graph -- 12 DR. SIESS: You don't want them ducking under the 13 desk. : O 14 MR. EBERSOLE: Or thinking about their. families out a 15 there. ! 16 MR. SMITH: Well, I have a lot of confidence in our 17 operators. I 18 MR. EBERSOLE: They're Marines. , I f 19 DR. MOELLER: You mentioned the calculation of room 20 temperatures as a function of time.
- l 21 Have you actually turned off the air conditioning and ;
t 22 just seen what the temperature rise was? ! I 23 MR. SMITH: No, we have not. We have some difficulty l l 24 doing that under normal operating conditions because we would ; 25 violate our tech. specs. , i L ' Heritage Reporting Corporation (202) 628-4888 l L
.t 365 k _s) 1 DM. h?5'LLER: Are you familiar with the AEOD report 2 on effects of temperature in the control room on various 3 electronic components?
4 MR. SMITH: I'm familiar, not with that report, but i ! 5 am familiar with other instances where things like SSPS have 6 given spurious outputs on loss of ventilation. Yes. 7 8 9 10 11 , r 12 13 14 15 16 ! 17 18 19 20 ! 21 22 23 ~ 24 i l 25 ; t fleritage Reporting Corporation (202) 628-4888 t [
366 : /? J'y . T k )' il MR.' SMITH: One point I want to make on this' human ; 2 factor-analysis is.that we have some very, very heavy' operator 3 input here. In fact, so heavy that it's-delayed our completion i 4 of the model. -j
'5 And. operators really have gotten into this effort.and !
6 have made1guite a bit of input to it, including such things as i . . 7 testing some of the scenarios on the simulator that they had- ! 8 never been trained for, , 9' MR. EBERSOLE: I would hope you might be converging i 10' on something as yet unavailable to us, and that is an answer to ! 11' the question of when do you automate. l 12 MR.. SMITH: I'm not sure that that's going to come ! i 13 out at the RA, but perhaps that's -- O 14 MR. EBERSOLE: Well, you might be getting closer than
. I 15 most anyway. [
16 MR. BLY: The question was when do you autom,wtn' 17 MR. EBERSOLE: We have a big void.in our busL. ass i 18 because there's no answer to that question. I don't know of r r 19 anybody that has a consistent or coherent notion of when they & 20 should automate. , t 21 Maybe it'll come to you all in the course of this [
?
22 business. 23 MR. SMITH: The next slide here -- I put a -- it's j 24 quite a complicated issue. And I don't mean to go too much 25 deep in it. l C) ! Heritage Reporting Corporation ! (202) 628-4888 [ r E
. . _ . . __ _ _ _ _ _ __-__-...___L
f l 367 1 However, the point I want to make here is that we-2 have'tried -- it'has never been done in the PRA before.. We've 3 .tried to do it'from a-qualitative standpoint. ;
-4 We don't' feel yet that we have the methodology or the l
5 knowledge, or tne data to be able to quantify this-to.the point
~
6 where it's meaningful. 7 So we dealt with it from a qualitative standpoint; 8 and this is.what this graph somewhat represents. 9 DR. SIESS: Have you tried to develop the data base?. 10 After all, we've built a hundred plants over a period of 20 L 11 plus years, with varicus degrees of quality control. 12 .And some of the more recent plants have been 13 subjective'to some pretty extensive IDIs, and IDVPs -- just ; l _() -14 reviewed and~so forth, all of which turned up a miscellaneous . i 15 collection of design and construction errors. 16 MR. SMITH: We've attempted to do that; and I can't L 17 say that we've been 100 percent successful. j 18 We don't yet feel like we have somethit:q that's i 19 defensible. And I guess we feel, from a qualitative , 20 standpoint, that what you say is true, that an IDVP should have [ 21 identified most of these design and construction errors prior i I 22 to operation, i 23 DR. SIESS: But I mean those are exampler of the i i 24 kinds of errors that have been made. I'll have t< admit, as I l 25 look at them I can't see very many that would be ery [ t 1 s- , Heritage Reporting Corporation ; (202) 628-4888 ! t i
368 73 LJ 1 significant if I stuck them under a PRA. 2 MR. SMITH: From a qualitative. standpoint that's 3 excetly what we've said. 4 DR. SIESS: But I don't know how to do it 5 quantitat' rely. I've been.the business of constructional 6 engineering for buildings, and here our safety' factors are 7 based.on a probablistic (ph) approach but then we calibrate it-8 against experience, which includes all the design and 9 construction errors. 10 Because 90 percent of the building phase -- we had to 11 do the mistakes. (ph) Now we're beginning to develop a 12 database of errors, not of failures due to errors. f ,s 13 That's what PRA is presumably -- substitutes for that
- 14 kled of failure experience.
15 MR. BLY: One thing --
- 16 DR. SIESS
- I just don't know how you can do it.
17 It's there. 18 MR. BLY: The thing we've tried to do in this 19 tabulation is to look at all of the information and sources -- 20 the studies on design and construction errors that are out 21 there; the data that came out of Diablo's design verification; 22 and bringing people who've been involved in construction 23 projects here. 24 And trying to at least list all of the kinds of 25 design and construction errors that we've seen identified and O Heritage Reporting Corporation (202) 628-4888
1
- 369
's- 1_ that we can hypothesize; and-then categorize them as shown-up-2- here.
3 Either, there.are. things-that in fact we think we 41 have modeled welliinithe PRA, or there are things'that'are 5 modeled through -- they effect the seismic fragilities, or-
'6 they're things that really don't. impact our model very much, or 7 only impact specific pieces of it.
8 So it's a cataloguing and identifying where things 9 have been modeled; where they haven't; and qualitative 10 estimates of what their impact might be on the risk. 11 _DR. SIESS: Is Bob Kennedy still here? 12 (Pause) 13 DR. SIESS: What do you do about design defects or 14 _ construction defects in your fragility analysis? 15 MR. KENNEDY: Bob Kennedy. Basically, we've done
.16 : sensitivity studies for EPRI, the influence of design and 17 construction errors on the seismic PRA results.
18 Those sensitivity studies showed that it would 19 require substantial errors to have a real influence on the PRA . 20 result, but that substantial errors, if they exist, could have 21 a real influence. 22 Now by substantial errors -- if you've got a major 23 item of equipment -- let's say a skirt monitor (ph) reactor
~
24 pressure vessel -- that's not the case here but let's say a 25 skirt monitor reactor pressure vessel; and it's anchored down s Heritage Reporting Corporation (202) 628-4888 h
~~
+
1 ,
.. 370 d
q$): . with'high strength volts.
.2 If the volts have the wrong heat,~and are imbrittled, 3 and have -dif ferent properties than they're supposed to have for 4 a seismic PRA, that would be a sub'stantial error because it 5' would have'a substantial impact on the-fragility estimate for- ,
6 that piece of' equipment. 7 So those kind of errors would influence,'if they 9 exist, leaving out of rebar, out of a major section of laws,-
- 9 would have a substantial influence, if they. exist.
10 We heard a case -- I think it-used to be Dave 11- Okrent's favorite case of some steam generators;that had only 12 three supports rather than four. 13 If that existed in a plant it would have a (b 14 substantial impact. 15 We tried to look for errors that have substantial 16 impacts during the -- - 17 (Someone coughed.) 18 MR. KENNEDY: To be honest, we don't find it.- .Now 19 that means either we.didn't look hard enough or they're not 20 there. 21 We have four.d in past cases unanchored equipment. l-22 And what has happened is equipment's been anchored before the 23 PRA proceeded, or we found battery rods with spacers missing 24 between the batteries; and that's been fixed. 25 We did not find those at Diablo Canyon. () Heritage Reportino Corporation (202) 618-4688
- - , - - -m.,m-nm-.- w - , - . -- . 7 -., - - ,
w- , ,
q:
.~
371
\_f -l We do not put.into our fragility estimates some shift-2~ of the fragilities based upon some' hypothetical-error. We 3 ' don't know how tofdo that..
4 The only way I can see to include them~1n the-
.5 fragilities is to postulate various errors and then to-6 postulate the probability that those errors' exist.
7 And at this stage we're simply not prepared to ( ~ 8 postulate the probabilities of those errors existing. We could 9 generate the fragilities under the assumption they exist. < 10 DR. SIESS: When I was listening to your partial list 11 of substantia 1' errors, it had occurred to me that that might be 12 a pretty good basis for direr- g a quality program.
.. 13 MR. KENNEDY: I thins that could be a very great help
(~
\- ' 14' cause I think there'could be efforts -- there'd be team efforts 15 -- I don't think there'd be one person.
16 But there'd be team efforts from~the technical 17 community to start tabulating down the substantial errors that i 18 could have a significant effect on seismic capability of
'19 equipment; and use that then as a list of the kinds of things 20 that needed to be evaluated in the QA\QC programs.
21 I don't think that's been done yet. And at this 22 time, designer construction errors remains a potentially weak 23 spot in PRAs. 24 Now design errors -- a lot of those we catch; and we l l 25 include those in our fragility evaluation. I'm not sure that
)
l Heritage Reporting Corporation [ (202) 628-4888
c 372 , , ~. (I ' 1 you would call them necessarily errors. 2 :About half the-time they actually increase our , 3 strength rather than reducing. 4 DR. SIESS: Well, there are.certain errors that
~
5 effect function that are called -- or could-be called in start-6 up programs. 7 I'm sure every start-up program _ finds'something 8- that's wrong and fixes it.
- But then you_ don't detect the kind 9 'of things you're talking about -- equipment ~ anchorages, skirts 10 __
11 MR.' KENNEDY: Yes, because they're not loading, l 12- except when the' seismic event comes.
, 13 Now on the other hand when you have an operating
~ _b 14 pump, there's start-up vibrations on an operating pump; they're 15 certainly going to load the anchor bolts. l 16 And if,there are errors on anchor bolts and operating 17 pumps they probably are going to be caught and~ fixed. i~ L 18 But errors on anchorage of a reactor pressure vessel 19- skirt are not geing to be caught and fixed, in all likelihood, , 20 until you have an event like an earth quake. 21 DR. SIESS: Thank you, Bob. 22 hP. SMITH: Okay. At this point, I'd like to get 12 3 into the phase 3A results. And again, I want to caution all 24 those present that,'like my friend's car, we know that it can't 25 get any worse, but we don't know how much better it's going to O Heritage Reporting Corporation (202) 628-4888 o -
p:
. [,
373 q-[ L 1' get when we get finished. 2 Phase.2, aslI mentioned before, was a very generic, 3- -judgement-oriented scoping effort. 'And in that kind of an 4 effort, you do run11nto some things that~we didn't realize. 5- One of-those things is the diesel generator fuel oil 6 transfer system; 'and that's a system that'provides fuel oil to 7 the' diesel generator day tanks. 8 And we'found out-after we did some documentation and 9 analysis that these things weren't as -- the syat'em wasn't as-10 reliable as we had thought it had been in the Phase 2 effort. 11 So let me give you a qualitative look at Phase 3A i 12 first, before we get into numbers. 13 Some of the things that came out - perhaps number
~
14 'one, was the seal LOCA concern.
' 15 MR. EBERSOLE: Isn't that only '.nportant for the case 16 of_ total loss of AC power?
17 MR. SMITH: No, it's also important'for total loss of 18 CCW. 19 MR. EBERSOLE: Okay. 20 MR. SMITH: And the second item deals with that, the 21 fact that the charging punp la dependent on the CCW.
- 22. MR. EBERSOLE: Wsil, yes,'in the long-run you've lost 23 the traction.
24 MR. SMITH: Exactly. , 25 And then the fuel oil transfer system was important O Heritage Reporting Corporation (202) 628-4888
t . P (, e" 374 A_J : .1- because-the diesel generator day; tanks will only allow the ,
- 2. diesel generators.to run at fu11' load forz a matter of hours.
'3: And'for'a_long-term accident we needed more than 4
4 that~. 5" -Relay chatter appeared to be a concern. 6 MR.'EBERSOLE: In this matter of AC power failure --
. 7 we've'-asked this of a good many folks, and I guess.I can ask it 8L .to.you'too -- do you'have any understanding of.the. dynamics of I
~
9- disablement of. functions after you've-totally lost ~AC power? 10 That is, what's the first pinch point where something-11 .is.. lost and you can't get it back because it's. damaged? 12 MR. SMITH: You-mean due to the loss of power?- 13 MR. EBERSOLE: This is a chronologically ordered set of events that sayr after X minutes'or hours somethings gone
~ '
14' 15 and I can't get it back.
- 16 MR. SMITH: I think the seals --
17 MR. EBERSOLE: The seals are a case -- 18 MR. SMITH: -- or -- l . 19 MR. EBERSOLE: -- or one of maybe many items. 20- MR. SMITH: -- one item. Yes. 21 MR. EBERSOT,E: I don't know what else you tried to L 22 run that you can't. 23- I can think of the aux -- you know, the turbine l- !". 24 driven aux pumps. 25 MR. SMITH: The instrumentation is a concern, too. 6 O Heritage Reporting Corporation (202) 620-4888 i
s ;f u? g i U g 375
~
6 1.& h 1 If you've. lost off-site power -- p 2' . MR.'EBERSOLEs. You've' lost most,of the heat-input to
+
- 3 'it'because you've lost AC-power. But:you still have' battery 4, ' input.
-5 MR. SMITH: Right. You have battery input which-6 drives. invertors which gives instrumentation.
7 MR. EBERSOLE: Right. And I-don't know whether it's.
-8 . localized internal heating or not. I don't knoa.
, 9 MR. SMITH:' The-invertors'and the batteries produce 1-0 very little heat.
11 MR. EBERSOLE: Well, the invertors in the long run -- 12 do you have -- well, okay. 13 MR. SMITH: Not enough to cause us problems in.what' 14 we found.
.15 MR. EBERSOLE: Okay.
16 MR. BLY: The specific ordering of.which of those-17 comes first, we don't know.- We've tried to address that; and 18 we're still trying to address that. 19 MR. EBERSOLE: It might be an interesting thing to do 20 to find out, you know, just what comes up first that you can't 2' fix once it's damaged. 22 MR. BLY: For example, reactdr fluent pump seals 23 could go very early, or they could last 15 to 24 hours. 24-- MR. EBERSOLE: And if you could keep injection it 25- wouldn't matter.
~
Heritage Reporting Corporation (202) 628-4888
N.
-376 ^A / 1 MR. BLY: 'And'if you injection it wouldn't matter --
2 .or:get it back, 3 MR.lEBERSOLE: Yes, or get it'back.
~4. MR. BLY: So -- there's been a lot'of effort in some-
.5 of these areas; and still the time period -- because different 6- mechanisms might or might not cause a problem, it's not --
7 we're always. going to have uncertainty in that --
;t - 8 MR. EBERSOLE: I'm sure.
9 MR. BLY: -- time period. 10 .MR. SMITH: Okay. Relay chatter's in issue because 11 'it puts equipment in positions that there weren't in prior;-and 12 may not be in the correct position for whatever scenario you're 13 needing tcf -- h 14 MR. EBERSOLE: Speaking of relays, I sent you a 15 Westinghouse; I guess I can ask you all in a segregated sense, 16 what.are you doing about your breaker problems now common to 17 all of Westinghouse? 18 MR. SMITH: We have General Electric breakers. 19 Oh, you mean the -- 20 (Laughter) 21 MR. SMITH: 4-KV breakers -- if you're talking about 22 the reactor trip breakers, yes they are Westinghouse DP-50s. , 23 MR. EBERSOLE: Well, you know reactor trip breakers 24 have been problems because they had them made in Brazil; the 25 spot welders didn't do such a good job, and the QA on them has
.O Heritage Reporting Corporation (202) 628-4888 t
4 377 3
- . M ~1- folded, an'd so forth.
'2' MR. SMITH: Well, we've made~several' improvements to 3 that system that I'm aware of from-a-systems --
E 4 MR. EBERSOLE: Well,fI've been.at the point of r . 5 interrupting _the extation curve [ph] on'the MG sets for a long 6 time. But on the other hand, you can manually. trip those 7 things. 8 .MR. SMITH: And that's what would happen'in the 9 operating procedures. 10 MR. EBERSOLE: Yes, all-the time. Differential's not 11 all that bad. 12 Well, anyway you're a part of family of investigators
-13 on the car breakers.
O. 14 MR. SMITH: That's correct, l l 15 AC power recovery -- that's recovery following an l' 16 earthquake -- is an important issue that came up; and that 17 issue is twofold. 18 One is the length of time that'the batteries would be-19 available for:the operators to do something. Second item is f l 20 how long will it take to recover off-site power -- how long 21 will it take to get the substation back in operation. 22 And then of course we have revised fragility and 23 hazards curves. 24 MR. EBERSOLE: You could tell me, why do we persist 25 in not having mechanically generated charges for the batteries O Heritage Reporting Corporation (202) 628-4888 ,
9 - . . - ., x, _ _' W 1 c , 1378 h
, - ( )[ 21E just to fixed'in. battery lifefas long?as we live?
2 :MR. SMITH. You mean as far'as'the gasoline powered ~
,3 ' generate'd1 type thing?
4 MR. .EBERSOLE: Sure, whatever. You :know :-- Honan'. 5 .' l['ph)~ 6 ' MR . SMITH:- 'Well, we have those kinds of things on. 7 site ---I mean,.that's something that's on almost any 8 . construction' site and we're still doing construction at' times. 9 MR. EBERSOLE: I suspect you could'use a welding 10 Egene'rator if you just adapted -it. 11 MR. SMITH: I'm sure we could. 12 . An'd you know, we'have a lot of flexibility. We have 13 more than'just one charger per battery. So we do have some
~
14 flexibility there too. 15 This particular graph just gives you a schematic of ,
't 16 what happens in the seal LOCA and why it's a concern.
17 .The charging pump, lube' oil cooling; and it doesn't
^ '
18 say there, but'also seals have to be cooled or else we're going . 19 to be-losing the charging pump. 20 Tha don't know how bad that loss would be -- whether - 21 it would be a total failure or whether it would be a partial 22 failure.
?
23 MR. EBERSOLE: On a boiler, that seal cooling network 24 is not seismic. However, it is here, isn't it?
'25 MR. SMITH: Yes. ,
o Heritage Reporting Corporation (202) 628-4888 - P w ,, u w , ., -
, --n, --+- + ----,p-~,--
, ,rn, ,
m - - I 379 t h. A )- 1- _The diesel fuel _ oil' transfer system 1that we're. 2- -talking about is; cured in' schematic. And there's the'large-3 underground fuel oil' storage tanks with two' redundant pumps.
~
4 One is leading - you :know, if it fails then-the.other one 5 starts. 6 And they feed the five diesel generator day tanks. 7- The day tanks are always at a certain level. And 8 when they reach a set point on the tank level control,_ it 9 starts the pump ~and -- 10 MR. EBERSOLE: You mentioned what I was going to 11 mention. You've got level controls which invariably always 12 interestingly susceptible to earthquake. 13 How do they jump up and down? How do your float 14 valves work? 15 MR. SMITH: Very similar. 16 MR. EBERSOLE: Do they stick at the top end so your B 17 tanks fill up and -- 18 MR. SMITH: I don't know, but we did include the
-19 failure of those valves as part of our --
20 MR. EBERSOLE: Do you have an overflow discharge in 21 case you over-pump the tanks? 22 MR. SMITH: Yes, we do have a discharge but it's not 23 returned to the main tanks. 24 MR. EBERSOLE: You could run it out anywhere. 25 MR. SMITH: You could run it out on the floor. O Heritage Reporting Corporation (202) 628-4888
e
?
9 380 (bslo4 - 1 .That's'where it's. going to be. R -" 2 MR. EBERSOLE: Well,.that produces a-fire hazard,3I 3' guess. 4' . W1. SMITH: Yes, it does. And I-think that's'in'our 5 spacial' interaction. 6 MR .' EBERSOLE: Okay. 7 - MR . SMITH: The point here on the diesel fuel or 8 transfer system that I should make is the reason that it's 9 ,become a contributor.is'the fact that it's in this' case an 10 -over-sized system in some fashion. 11 The reason -- it fills up the datex very quickly'so 12 that --- high capacity pumps. So that what you have is pumps 13 starting and stopping many, many times over a 24 hour period _O 14 which would be the mission time of these diesel generators. l- 15 MR. EBERSOLE: What if it's-too big? [ph] 16 MR. SMITH: So we're looking at that, and I have some .
- 17. -- perhaps later - .
l 18 MR. EBERSOLE: Are those the operators valves up l 19 there that I saw? 20 MR. SMITH: Yes, they are. l 21' MR. EBERSOLE: So there we go about air -- by the 22 way, is your instrument air system dry, non-lubricated pumps? 23 MR. SMITH: These valves are fed-from a separate air 24 system that also provides starting air for the diesel 25 generator. ( l Heritage Reporting Corporation l-l (202) 628-4888 y , - ,. - , - , - - . , --
f
. 381
( ,. . ;, ,
.() 11 'MR.-EBERSOLE:- Okay.
2- MR. . SMITH:' So it has. dryers and --' d3- .MR. EBERSOLE: So,it's -- sized air.
.4 MR. SMITH Yes.
f' 5 Okay, relay chatter -- here's some of the results 6 that we found. And I think these are kind of. obvious results.- 7 One~is-it does effect all systems. We couldn't find-
~
8 any system that it didn't effect. It is easily recovered; and: 9 it will cause many misleading indications and alarms which will 10 effect the operator actions. 1 11 MR. EBERSOLE: Does it lock in some valve functions 12 to the opposite mode -- open to close? 13 MR. SMITH: Yes, we.did find -- 14 MR. EBERSOLE: Isn't that disruptive to the
.15 operators?
16 MR. SMITH: Yes, it is,.very disruptive. And that's-17 one of the issues that we're dealing with in Phase 3D. 18 DR. SIESS: If there is a lock-in due to relay 19 chatter, does the operator know it? 20 MR. SMITH: The indication will tell him, yes. 21 DR. SIESS: He will have an indication? 22 MR. SMITH: Yes. 23 DR. SIESS: In all cases? 24 MR. SMITH: Yes. 25 MR. BLY: Well, what are indications? O Heritage Reporting Corporation i (202) 628-4888
v=, 1(['i f . 1 382
/~T - .
\ ,i .1. MR. SMITH: Well, the position' switch-indications of 2 a~ valve, I thinkfwould, even though they may chatter, they're 3 .still' going to be; good; you're still going to have red'and 4 green lights.
S- MR. EBERSOLE: Speaking of valves, ordinarily they're 6 sort of bi-stable devices -- they open or shut and you don't 7 know with what margin they do it.
-8 =Are you thinking the1new -- what's it called? New 9 valve monitoring systems? What am I thinking about?
10 MR. SMITH: Yes, the MOVATs? 11 MR. EBERSOLE: Yes. 12 MF. SMITH: We're using it for testing. I don't 13 think we're using it on a --
. ,rx
' \l 14 MR. EBERSOLE: Are you getting data on the valves to L
15 ascertain, just like the Salem case, what's the marginal 16- function? 17- You know, how much-was left besides what you needed
- 18. to close an open valve -- to include the Duress case?
! 19- MR. SMITH: I'm not sure that we explicitly have done l 20 that. You mean as part of our maintenance efforts? ! 21 MR. EBERSOLE: Yes, I take the reactor water clean-up 22 line and a hypothetical break of it -- that's a classic one. 23 You must cut the water off or accommodate -- and then you-24 discharge the reactor water out of the primary vessel. 25 Do you know your valves will always shut when they O Heritage Reporting Corporation (202) 628-4888
~
P_ , 383 g7 is 2
- are.-in full flow?
, .q 2 - MR. SMITH: That issue has come up already.. And it's o
;3' :a generic issue,.I,think.
-4 _MR. EBERSOLE: It is. Any other pipes-like that?-
5 Yes, there is -- this beam to aux turbine.is another case'. 6 That's the one, you know, where high temperature 7 closes the valve when you need it to be open because of the 8 loss of electrical. cooling.
-9 You remember. Is.the temperature a trip which is 10 tripped by the loss of ventilation?
11 MR. SMITH: I don't think that our -- I don't think 12 our valve is set up that way.
. -s 13 MR. EBERSOLE: You don't have a temperature --
14 MR. SMITH: No. 15 MR. EBERSOLE: -- trip because of hypothetical 16 leakage of the pipe? 17 MR. SMITH: No. 18 MR. EBERSOLE: Okay. All right. 19 MR. SMITH: No, I'm pretty sure. It'r, SCV-95. 20 MR. EBERSOLE: Okay. 21 MR. SMITH: Let's see -- some important contributors 22 as far as relay chatter concerned are the 4-KV breakers, and 23 that's with protective relay sealing in; and of course the 24 motor operated valve transfer, which we just talked about. 25 And then there's a lock out on the diesel generator O Reporting Corporation Heritage (202) 628-4888
4
'384
(_) 1 in the course. [ph) 2' Next' slide talks about some of the things that we 3 -went into on the relay chatter analysis that were concerns. 4 ~ And this gives you some idea of how we handle those particular 5 issues. 6 One of those is control circuit lock-up. We were 7 wondering about whether or not there would be cases where the 8 circuit would lock up in.such a manner that it couldn't be-9 reset from some normal location such as a control board, 10 We looked for that; did not-find that, although we 11 did find many circuits that actually did latch in, but' fairly 12 easily reset. 13 Load sequencing didn't turn out to be a concern. O 14 Contact arching damage insulation failure and induced voltages 15 were all things that electrical engineers would worry about. 16 We did some analysis on that. And based on those 17 transient analysis those weren't a problem. 18 We also looked at simultaneous closure of reversion i 19 contactor such that you use on a motor operated valve. And we l l 20 do have mechanical and electrical interlocks on those. So they i 21 weren't of concern. 22 We also looked at other things that were non-relays, 23 such as circuit breakers, over-loads, and flow pressure, those 24 types of switches; and those were included in the study. 25 And then we looked at valves at pressure boundaries Heritage Reporting Corporation (202) 628-4888 b > t
385 c 76
', L ) ~~1 .where.the valves might crack.open. And we found that only.if 2 theicontrol circuit latches up and drives it to' the other 3- direction'would we.have some' questions-there. -
4 MR, EBERSOLE: ' Just another shot in the' dark -- some 5 of the plants are holding on to exitation after trip to carry 6 houseloads. 7 But even withoutLthat. Do you look at the 8 possibility of excess frequency and excess voltage from the 9 main turbine generator when it's dropped its main load and ; 10 still carries house load? 11 MR. SMITH: I think we have trips that occur due to
~
12 that. 13 MR. EBERSOLE: Well, I'm looking for the excess again 14 -- too much frequency; too much voltage -- because you've lost 15 your main load; and you may have some overspeed which is --- 16 MR. SMITH: We do have that trip on.the reactor 17 coolant pumps. I'm sure of that. . 18 MR. EBERSOLE: Well, that's because of the flywheel
- 19. effect. I'm talking about damage to the water pumps which you 20 hope ultimately will get back on the diesels.
21 MR. SMITH: Yes. No, I don't think we have 22 considered -- t 23 MR. EBERSOLE: Yes, this is a common effect. It 24 spreads through the whole plant. If you get excess voltage and 25 frequency -- () Heritage Reporting Corporation (202) 628-4888
h 386 xl' 1 MR.' SMITH: Exactly.
- 2. MR. SBERSOLE: -It's just not a turbine problem.
+
-3 MR. SMITH: No, I-don't think we've explicitly-
'4 considered that.
5 MR. DAVIS: Would you put that slide back up again, 6 please? 7 I presume you mean reversing contactors and not 8 reversing contractors. I have heard offreversing contractors. 9 (Laughter) 10 MR. SMITH: Yes, you are_ correct. That is the type 11 -of -- 12 (Laughter) 13 MR. SMITH: Okay. Here is the -- 0 14 MR. EBERSOLE: Do you know any case.of contactors. 15 that precede breakers in opening when they should not? 16 MR. SMITH: As far as the full current type of thing? 17 MR. EBERSOLE: Yes. 18 MR. SMITH: I really doubt it. 19 The short circuit devices are all much higher speed 20 than normal -- 21 MR. EBERSOLE: Yes, if they work right. Sure, I 22 know. 23 MR. SMITH: The results that we've presented here -- 24 we've tried to divide it up into things that would be of 25 interest to this group. O Heritage Reporting Corporation (202) 628-4888 E
I 387
.( i_ . . .
A/ 1. One point I want to make'here is that theselare. based 2- on' point estimates onlyf- ;no uncertainty involved here. 3' Let me give you a:little bit of-a. legend. 4' 'The bar out atithe right-hand side-is a mean bottom 5 end of where we think th'at -- we know that it's no worse than. 6 that'particular value. 7 That's the upper-end as far as core melt risk is 8- concerned. 9 DR. SIESS: What -- 95 percent or 100 percent? I'm 10 not -- l l 11 MR.' SMITH: That's the mean -- mean . ore damage ! '12 frequency. We know that's the upper end of a mean. { 13 The -- O
\ ~14 MR. BLY: Yes, I guess it's an unusual display. And 15 the reason we're showing it this'way is because we're in the j 16 middle of this process.
l 17 And the work we're doing in Phase 3B right now, i 18 looking primarily at those five or six issues Bruce had on the l 19 board up here before kind of generates where we fall. ( 20 If in fact our engineering judgment as this -- in 21 fact it's more than that. It can't even happen now; we've I 22 learned enough. 23 But if in fact on all five or six of those issues 24 things turned up on study to have the worst impacts we could 25 have imagined back when we started Phase 3B we would have been O Heritage Reporting Corporation (202) 628-4888
388 i s
\-) ; 1 'up at the high-hand of this.
~2 If in. fact after the 3B work is done, these issues-3 full out, where-our best judgement tells us now, given the work g.
-4 we've already put'in, we think it is going to fall; we como 5 down about to where that star is at.
6 MR. SMITH: The star is our best estimate of where we 7 think we're going to be at the end of Phase 3B. 8- MR._BLY: At this point in time, given the work we've 9 done. 10 MR. SMITH:- And actually, from some of the work we've 11 done since we did this graph, I personally think we'll do 12 better than that, but that's our best estimate. 13 MR. KENNEDY: This would imply that the seismic [- 14 contributor in your best judgement is a larger contributor to 15 core melt frequency than all of the other non-seismic issues.
.16 About two thirds --
17 MR. BLY: On the best estimate, yes. l 18 MR. KENNEDY: Now there is occasionally among the 19 . staff a discussion that says no identified event -- and I'm not l 20 sure whether a seismic event falls in this category or not -- i' l 21 shall centribute more than 10 percent of the total risk. 22 I don't have any idea whether this part of the staff 23 knows about that other part of the staff; and if they did find 24 out about it whether they would want to invoke that, or whether 25 our seismic event falls into that category; but it's an O Heritage Reporting Corporation (202) 628-4888 L
t 4 389 f l1 interesting: possibility. 21 VOICE: There've been many seismic events.- 3 MR. BLY: .That 's :right. This'is the. sum of.all 4 possible~ seismic events. So you have to think of how you group ; S events. In fact,the other one -- the non-seismic is a group of - 6 a whole' bunch of -- 7 MR. KENNEDY: -I only mentioned it because I had
.8 encountered it previously in:other circumstances.
9 MR.-DAVIS: How many dominant sequences were there'in
- 10 'the seismic initiated events? Can you -- do you remember
- 11. approximately?
12 MR. BLY: Well, it depends on where -- which event 13 you're talking about. The lower level earthquake -- lower
- 14 acceleration -- the main contribution was coming from a very 15 'small chance of failures in the 4160 system.
16 MR. SMITH: We have another graph that deals with 17 that. 18 MR. DAVIS: I got the impression though from 19 Kennedy's. presentation that the major contribution to core melt 20 is up around two to two and a half G -- something like that. l. 21 MR. BLY: The frequency of core damage for 22 earthquakes within a particular frequency range -- 23 DR. SIESS: Go to the next slide, please, while he's 24 talking about that.
- 25 MR. SMITH: This is a break down of that seismic
( Heritage Reporting Corporation (202) 628-4888
\ ;
r 390-k- 1: Lcontribution. And basically what -- we talked'about h'ow this 2 happens, I mean, what drives this particular bar. graph. 3 And number one, it's a convolution;of the-hazards and
~
4 fragilities fact that around two Gs, two and a half Gs it's a 5 hazard. 6 The hazards are going down; the fragilities are going 7 up. And you have a -- 8 DR. SIESS: Now that's a spector. acceleration. And 9 as I remember what Mr. Kennedy said, that would correspond to a 10 zero period acceleration of about one. It's 2.25 or 2.35 is-11 the number. 12 MR. KENNEDY: Bob Kennedy. Yes, you would take these 13 numbers and divide them by about 2.35.
'O 14 DR. SIESS: To get a ground acceleration.
15 MR. KENNEDY: To get a peak ground acceleration, yes. 16 So the two and a quarter G is just slightly below one 17 G peak ground acceleration. 1-8 MR. BLY: And remember, these are the results given 19 in Phase 3A fragilities, discussed earlier. 20 MR. EBERSOLE: Could you put up the previous slide, 21 please? 22 That's the small LOCA -- the big one? 23 MR. SMITH: Yes. Small LOCA but it also includes 24- seal LOCA, which is -- 25 MR. EBERSOLE: Where is loss of aux feedwater? O lieritage Reporting Corporation (202) 628-4888
- ,y _
y 391 [v ') ~ 1; (Pause). 2: MR. BLY: Loss of aux feedwater by itself then slips 3' you back' to bleed ~and feed cooling; and-failure of.both loss of 4- -- both aux feedwater and bleed and' feed is'over in that'over
- ,} box cause it's down --
- 6. MR. EBERSOLE: On my right?
7 MR. BLY: Yes. 8 MR. EBERSOLE: And fire is way down. 9 .MR. SMITH: Let me talk to you about some of.the 10 things that are driving these. The loop, the loss of outside 11 power one is, you know, a contributor to the seal LOCA concern.
, 12- MR. EBERSOLE: That's ultimate loss of.all power, j , 13 isn't it?
O 14 MR. SMITH: Yes. 15 MR. EBERSOLE: Okay. l 16 , MR. SMITH: And then the steam generator tube rupture 17 item is a scenario where we have a steam generator tube l 18 rupture; the sjafety valves stick open; and the operators don't 19 recover. 20 (Pause) 21 MR. SMITH: And the other contains all the other
- 22 scenarios.
23 MR. EBERSOLE: And the fire -- gee, your -- your iron 24 conduit must be pretty good around those wires. 25 MR. DAVIS: Before we leave that slide, I'm sorry Heritage Reporting Corporation (202) 628-4888
-392
(- k' 1
~
in previous PRAs for Westinghouse plants, the dominant risk
~
2 sequences have not-been the same as the high core melt 3 probability sequences. 4 I realize this is just a level one. You're not
~
5 really going'to determine risks. But in both Millstone 3 and , 6 Seabrook the V sequence was the overwhelming contributor to 7 public risk. 8 And I presume that's in your other category here. 9 Do you remember what the probability of that was?
~10 There's been some controversy about that sequence as a result 11 of some work IDCOR did on Seabrook and Zion; and I'm wondering 12 if you have --
13 MR. SMITH: e I don't rememb^r what the frequency was - O \Y 14 - 15 MR.:BLY: The model here -- and I guess I have to 16 take you back to Seabrook. If you want to talk about that one 17 chat isn't what this really involved. 18 But as Seabrook it was the important contributor to 19 the early effects risk, but not the dominant contributor, I 20 think, to the latent effects. 21 MR. EBERSOLE: I believe it was -- 22 MR. BLY: And if you go to the final Seabrook work 23 which is submitted in the site study -- a much more thorough 24 model of the V scenario is in there; and that's the model of V 25 scenario we're using here, which is a detailed model looking at () Heritage Reporting Corporation { (202) 628-4888 1
S. . 393
- 7. .
' ~
's / 1 where you would get. breaks if in_ fact you had the interfacing
~
2 LOCA_-- where.the water would go; and all that sort of detail. 3 'It's over in the other era. It'_s a very low 4 contributor.-- MR. DAVIS: To core melt. 5 - 6 MR. BLY: -- core, core damage frequency, that's
, 7 right.
8 MR. DAVIS: Not necessarily the risk. 9 MR. EBERSOLE: Since you brought that up I have a , 10 letter here dated first of -- 19th of January in which you all 11 are going through some' agonizing about removing the interlocks 12 on the RHR order closing (ph) interlock functions. 13 Have you claimed to that? bgs 14 MR. SMITH: Yes. l 15 MR. EBERSOLE: What are the pros and cons of that 16 because there's certainly some on both sides of the issue. j_ 17 MR. SMITH: Well, the people from Brookhaven have 18 done some studies on shut-down risk. And that's one of the l l 19 things that they're recommending -- is a removal of the closure 20 interlocks. 21 MR. EBERSOLE: I see that Westinghouse specified the 22 bell operator sizing such as they are not capable of opening -- > 23 to give loss of the air's pressure. [ph) i 24 That suggests they, you know, they put a small motor i 25 on it. That also then, as always, suggests that it Won't oper. () Heritage Reporting Corporation (202) 628-4888 [
9
., 394
_>f3 b 'l- when you want-it to.
-i ,
2 .You know, .the negative. aspect'. Omi 3. MR. SMITH: Yes. 4" MR. EBERSOLE: Are.you.througn with that' analysis, s. 5 .and have you sewed that up? 6 MR. SMITH: I'm not inv'Ived with that analysis--- 7' MR. GBERSOLE: It'r a hot operation. 8- MR. SMITH: --~but I'm funiliar with it. . 9 MR. EBERSOLE, 11 o v.. che stove today. This is a 10 current event. 11' MR. D'.(ESLSR s I'm Mike Dresler, (ph) Project Engineer
.12 for Diablo.
q 13- I believe our submittal for removal of those
~
h, 14 interlocks has been submitted to the NRC and is' currently under 15 review in the form of a license -- and they issued it last 16' week. 17- MR. EBERSOLE: So it's still --- 18 Mit . ROOD: Harry-Rood, NRC staff. 19 We last week issued a letter approving their safety 20 evaluation of that issue. It did not require a license 21 amendment. b
- 22. MR. EBERSOLS: So now that you've got a small motor 23 on it, if you had some piping failures, and you still have four
- - 24' othar p mnds of pressure in it, how are you going to close it 25 off? And what happens if you don't?
O Heritage Reporting Corporation (202) 628-4888 6
395 1 MR. BLY: Excuse me, Doug. The small motor, I 2- believe,.unless I'm way off-base, is the motor thatLthey've 3 always had that wouldn't let the valve open'at full system 4- pressure, 2000 pounds, but would allow it to operate.at the 5 pressure ranges you're' describing. 6 MR. EBERSOLE: Yes, yes, that's what it says. It was 7 not capable against full pressure. But it will close (ph) 8 against some out -- force, (ph) consistent with what you have 9 when it's open. 10 Am I correct? 11 MR. BLY: Yes. 12 MR. SMITH: Okay. I picked this graph up and I can't 13 talk too deeply about it because we don't have a lot of 14 information about what Plant A, B, C and D are. 15 In fact, that was something that Gary has not told us 16 specifically. But it does give you an idea of how we -- and 17 again, this is based on best estimate -- where we think we'll 18 be at the end of phase 3B, rather than on 3A results. 19 MR. KENNEDY: What is the significance of that graph 20 as far as PGB (ph) is concerned? 21 MR. SMITH: I don't think it really has any 22 significance other than we were interested in how we compared 23 with other plants as far as uncertainty in the immediate 24 points. 25 MR. KENNEDY: Well, if you were interested in it, now Heritage Reporting Corporation (202) 628-4888
p 5' 396 y . (_) 1- that'you see it it must have some significance. You must have 2 : decide'd that you like the comparison,~or you don't.like it, or 3 -- 4 MR. SMITH: Well, we would like to be as far out to 5 the left as we possibly can, but what we want to know is 6 whether or not we're in a normal range for a plant of our type. 7 DR. SIESS: This includes seismic -- all of them 8 includes seismic? . 9 MR. BLY: I think Dennis would have to answer that 10 one. 11 MR. SMITH: I think that's true, for this group. I 12 think they all include seismic. 13 DR.'SIESS: Because I notice that the Diablo Canyon i) 14 has significantly -- well, noticeably larger range 15 uncertainties with the corresponding difference between mean 16 and median. 17 MR. BLY: Let me address'that real quickly. 18 This -- the mean value -- the circle in this curve is 19 the best estimate. The whole uncertainty bounds reflect the 20 uncertainty as of the beginning of Phase 3B. 21 So that we anticipate at the end of Phase 3B the 22 uncertainty bounds will be smaller. That's what the c*ork in 23 Phase 3B is doing. 24 The thing on here that is best estimate and tied to 25 the other results you've seen is where that circle falls along O Heritage Reporting Corporation (202) 628-4888
. x .
; ~ -:.t-
> ~
-- _ 4_
, 397
,;f't . e
'l M ' ,1, the range.,
DR. SIESS So this is not certainlyicomparable to ; 2 13- Plants:A,fB'and.C. ,
, :o .. .
-4 MR. BLY - Those are completed studies,,that's right.
~5 DR._ SIESS ~ Okay.- And-it's been your~experiencexin
- c -
-; 6 other'plantsithat ittmoved to the left as you got better - _got- ,
~
7- smaller?
'8 '10 0 BLY , It's.been our experience the uncertainty 9 'has gotten smaller. As long as you're very careful _about how ;
~
- 10 .you address the uncertainty early on.in1the-account for.all.
7 x 11: aspects 1of.it,~ the mean can move.either_way. ! 12 MR. KENNEDY: Core melt, as defined'in the PRA,.means. 13 loss'of. capability of cooling the core? s .
~
14~ MR. BLY: That'sLessentially right. Yes. 15- ~MR. EBERSOLE: If I could just step back to the fire . i 16 risk. It's very low here. 17 You said you used conduits on -- all over the plant. 18 f.R . SMITH: That's correct. > 19 MR. EBERSOLE: Eventually, that gets impractical. 20 Don't you have an open-spreading room? !
. 21 MR. SMITH: The cable spreading room below the 22 control roou has all conduits for -- t 23 MR. EBERSOLE: Right on up to the terminal boards to
- 24 the individual --
25 MR. SMITH: Eight on up to the boards. ! Heritage Reporting Corporation ! (202) 628-4888 4 o [
1 'g 398 y\, O :l
-MR. EBERSOLE: Astonishing.
2 MR. SMITH: .Yes, it-is.-
=3 DR. SIESS: .They'll show'it to you later.
14 MR.-SMITH: If you like, I'll'take you down there'and 5- -- 6 MR. EBERSOLE: 7t must be like spaghetti. 7 MR. SMITH: -- and show you the spaghetti that we , 8 have, j 9 MF. EBERSOLE: The last one I saw like that was the 10 mussel shells plant built in 1914. , 11 (Laughter) 12 MR. SMITH - Well, I'll tell you what, I'm an 13 electrical engineer. And we do design in that area. t
'I )
14 And I can't remember what the cost is, but it's an 1.5 astronomical cost to run a conduit in that area. 16 MR. EBERSOLE: I can imagine. 17 MR. KENNEDY: Let me continue. The ten -- minus four 18 number that you have here, I recognize it as being very 19 prt'i.minary. i j; 20 But it cortainly suggests that you are not l 21 outstanding compared to core melt frequency in other plants . 22 that PL & G has analyzed. I
.23 If I were operating a plant, that would suggest to me 24 that I want to know more than core molt frequency. I'd want to 25 have some idea of the associated risk, not'necessarily for this O Heritage Reporting Corporation .
.(202) 628-4888 l t
i
+ - -
c e:
- m. , ><
y , 36 ic . , : _ D'9b , . 0 399' !
,[(1- >7A
- 1
?
.: study,-but if Ijwere a plant operator Ii think I'd be interested:
3 12 -in that.
'3 I'make',that':as a comment not as a. question.
4- DR. SIESS 1 hat occurs-being; consistent' wishing for 5-more than a level one PAA.- - 6 .MR.: SMITH:- I/ have~one. graph-that may address'that-7 tv-some extent. We were talking about -- we heard you question 8~ -yesterday'and were' talking about it. 9- What'we do as part of the' level one is - 'a' level'one-10 -- plus,- if you want-to <:all'it that.: We do include in'the plant 11 damage states the status of the accident mitigation systems 12 which include the fan coolers and.the spray research system.
~
13 :And also we include whether'or not the containment is O~ 14- in tact. 15 DR. SIESS: By intact you mean more than just 16- structurally. MR. SMITH:
- 17 More than just structurally, yes.
18 Whether or not penetrations are all'-- 19 (Microphone problem.) i , . 20 MR. SMITH: And also as part of this effort we're i 21 doing a containment capacity analysis, so we're trying to 22 determine what the actual best estimate capacity is for the 23 containment. , 24 So we are addressing it. We are not doing a level 12 5 - two PRA, but we are addressing some of the issues that you , Heritage Reporting Corporation s (202) 628-4888 1 A e - - - - - - - - n---e - - - , ,-- - +
400
.(m
{, h k.) I
~
would have on a level two PRA. 2 . Well, okay. We're getting around to Phase 3A -- 3B. 3 And=what we're' going to do is to do some-refinement of the 4' analysjs that we did in'3A. 5 We're' going to concentrate on those things that we've 6 identified as dominant contributors; and then we're going to do 7 the report. 8 -The major technical activities that we have in 3B:are 9 of. course the hazards and fragility effort that we're going to 10 be getting from the other pieces of the long term seismic 11 project. 12 Ne.'re going to be doing some work in human action _ - 13 analysis, where it's turned out to be important. We're also
.14 _ going to be doing some human action analysis to address some of 15 the issues that came up in Phase 3A such as relay chatter --
16 .how can the operator recover from relay chatter. 17 Relay chatter analysis, we're looking at, as I said, 18 recovery, and also fragility of the relays. We did a fairly 19 generic fragility for all relays; and we're going to look at 20 that closely to see if we can improve that. 21 We're looking at pressurized thermal shock. And the 22 reason we're looking at it is to make sure that we're , 23 consistent with our Westinghouse analysis that has come out in i 24 the progress of the PRA. 25 And then finally we're going to do the numerical end Heritage Reporting Corporation (202) 628-4888
rl },3 . m. 401 4,3 T_/ -.1 of it - .the quantification and uncertainty analysis. 12 'A big activity is the seal LOCA. And some of the 3 ' questions that we're trying,to~ answer are-how-long before-this 4- -seal is going to leak; and how much willlit leak if it does 5 tauk. 6 And some associated things with that are the. charging 7 . pump life without-CCW, and any recovery efforts that we would
-8 have in-that area, such as fire water, someone mentioned, or 9 cross tie to,another unit.
10 Two,important areas are the diesel fuel oil and off- , 11 site power recovery. And those are mainly the seismic issues. 12 The off. site power recovery hinges -- I know I've got 13 'a slide in here ot-that. I think I'll get to that later. O
- ' This loss of inventory from sea lee.ks 14 MR. EB1R. COLE:
15 is parallel with a loss from PRVs being stuck open. That's 16 another inventory loss. I guess that's the only other one I 17 can think of -- other than something broken. ; 18 It throws you into the possibility of having to feed 19 and bleed because you loose pressure, doesn't it, to translate ; 20 heat to the secondary side. 21 MR. SMITH: The seal LGOA -- I mean, it's.an actual 22 LOCA rather than a -- 23 MR. EBERSOLE: Right. But it also loses -- before it [ 24 loses water it loses pressure which prec1, dea your sending heat 25 to the secondaries. j Heritas) Reporting Corporation (202) 628-4888 i
l(- 402-g 1 4
'-4 1 MR. BLY: It does but we've -- we have been looking 2 at analyses cn1 rebus (ph) cooling with a two phase monitoring 3 system, and that looks pretty good.
4 The real' problem is under.most seal LOCA scenarios, 5 we ' don't have make-up capability. 6 MR. EBERSOLE: Yes. 7 MR. BLY: And that's our problem. 8 MR. EBERSOLE - That would be true with-a PRA, too. ~- 9 MR..BLY: Well, it depends on what sticks with~the 10 PRV -- 11' MR. EBERSOLE: -- In the open -- it's an inventory 12 loss.
,,s 13 _MR. BL6 But if we have power every ihing shuts --
'~] 14 MR. EBERSOLE: Oh, yes. But if you have power you-15 can also take care of the sea leak.
16 MR. BLY: That's right. 17 MR. EBERSOLE: Well, I mean I don't see the 18 difference between a PRV and a sea leak. 19 MR. BLY: Except the loss of power can induce the 20 seal LOCA where the loss of power is not likely to induce the 21 PRV -- 22 MR. EBERSOLE: That's true. Right, true. 23 MR. SMITH: I'll skip over a little bit on seal 24 LOCAs. I think we've talked about that quite a bit, and get to 25 the off-site power recovery task, p (. Heritage Reporting Corporetion (202) 628-4888
y-n, ; - - - f
~
403
.(~
is) 1 MR. DAVIS- Excuse ms.. Before you.do that -- ' s 2 MR. SMITH . Sure. 3L _MR. DAVIS: 'The slide indicates you're going to 4: review Westinghouse NUREG 1150 models -- or models that were 5 used in NUREG 1150. 6 I don't remember exactly what those models_were, but. 7 I think there's some information that's better or more recent 8 -than that. 9 Maybe the NRC can help. I believe.the -- 10_ MR. BLY: That's what we're talking about. There was 11 an elicitation on the seal LOCA issue with Jerry -- che NRC.. guy 12 -- Jerry Jackson, Mick Hitchler from Westinghouse, and a 2: J.ow
- 13 from ACO.
14 MR. DAVIS: Okay. That's good. Now --
~
15 MR. BLY: And that'e what we're waiting for. Thas 1G not ott-yet but it's due out any time now. 17 MR. DAVIS: I was hoping you-could do bette,r than 18 1150. 19 MR. BLY: Next version 1150 is what we're talking i 20 about. 21 MR. DAVIS: Okay. Good. So in other words, all 22 safety issues, I believe, are in their quantitative activity 23 right now. 24 MR. SMITH: In off-site power we're looking at all I 25 the components that are causing this issue. One of them is the O Heritage Reporting Corporation (202) 628-4888
i 404 1 outage durations -- how long would the transmission lines be 2 'out; how long the substations be~out; and also the 125 volt DC
- _3 battery life.
K 4 And wo're talking about things such as shedding load 5 or' completely shutting one battery down -- we're even-talking 16 to the operators about that. 7 We have three batteries; we're saying can you operate c e 8 with two batteries or one battery if you know that you're going 9 to be without off-site power for some significant~1ength of 10 time. 11 These are the kind of questions that we're asking the 12 operators. 13 MR. EBERSOLE: Talk about a class of things that I
\-
14 think you all initiated interest in. It's the modes -- the 15 various modes of operation below full power at which you find 16- astonishingly that Tech Specs simply are going to wonder, and - 17 you don't use them any more. 18 And you begin to put in Taigon (ph) tubing and all 19 sorts of horrible things take place. 20 You're looking at that family of events below the 21 full power level when you're going into shut-down modes and so 22 forth; and the potential for slow but disastrous over-heating - l l 23 - you know the ones I'm talking about, where you had that half-24 pipe case? - 25 MR. SMITH: We have looked at the Brookhaven study. O Heritage Reporting Corporation (202) 628-4888
b. r e t 405 5 1 In fact, the Brookhaven -- we're working with the Brookhaven 2 people; and we're.trying - we haven't included that 3- specifically in:the PRA; and we're trying to. determine right 4 now whether that's something we want to look'at. 5 MR. EBERSOLE: I noticed that you disconnected the 6 ribler (ph) system that you had to put on at the MI2 at the
.7 very time and space that you might have used it.
- 8 (Laughter) 9 MR. EBERSOLE: And you hooked up Taigon tubing 10 instead.
11 -MR. SMITH: That's right. Brookhaven had mentioned 12 that in their study, if you're familiar with that.
, 13 MR. DAVIS: .On this issue, do you assume that manual
14 operation of the-turban driven auxiliary feedwater train is 15 possible with no DC?
16 MR. SMITH: Yes, we do. And we have procedures, I 17 think on doing that. 1 18 MR. DAVIS: Oh, do you? You feel comfortable -- [ s 19 MR. SMITH: Pretty sure -- is that true, Ward? 20 MR. WOGSLAND: Yes, in fact that was one thing they 21 were surprised with, is that we already had that generated and 22 in place even before we looked at it from a PR standpoint. 23 MR. DAVIS: I don't want to go into details now but 24 I'm skeptical about being able to do that. 25 MR. SMITH: Well, let me give you an example, and O Heritage Reporting Corporation (202) 628-4888 t
~
T 1 4'06
~
-k-) -1 ,maybeithis will reduce some of'that. skepticism.
1 2 One of the things that we said that we.could do is to 3 tieftogether buses on,the vital -- we-have.three vital 4-KV 4 buses, and we :said, hey, we can ~ tie those together.
.5 PIU4. people said, well, show us, we don't believe 6 that. And we said, well, we've done it. And prior to
.7 licensing we did have an occasion where we had - .we were-able.
8 'to tie the three. buses together; and I believe we-do have the 9 procedure now. -
-10 ER) it'r, not something that's completely far-fetched, i
11 Our oporators do know how to do those kinds of things. 12 MR. BLY: The whole area of procedures is worth a 13 word hero. We'v0 looked at a Joc of plants and a lot of
.14 procedures; and most all the Westinghouse plants now have 15 procedures that have spun out of the owners group guidellnes. -
16 And'these go much further than any I've seen to 17 filling the gaps that people have raised. Simple example -- in 18 a small LOCA, it wouldn't make sense to go on to open up i 19 recirculatien. ; 20 Nobody's ever done it. Nobody would unless they had l 21 to. This is the only plant I've seen that has all the options 22 for ge+cing on the closed loop RHR [ph) while you're still t 23 los'.ng water. l 24 So they've done a real thorough job in working -- 25 MR. EBERSOLE: Have you got a satisfactory set of - O Heritage Reporting Corporation (202) 628-4888 i, t
t
'l L - . . 407
;( '
i 1 clues for the: operator to tell when it should go into the feed i n 2 bleed mode? 'You don't want him to do it, so -- 3- MR. BLY: But you don't want him to wait too long.: 4- MR. EBERSOLE: That's right. So -- I'm'looking at, 5 you know,ldoes he have a positive point of entry.
, 6 MR. SMITH: Yes, they're much more complete here; and 7 maybe you'll get a chance to look at them.
8- MR.'EBERSOLE: Let me~go back just momentari.1.y to 9 that salt water intrusion problem, realizing the condensors 10 might take a sudden' slug [ph) with salt water in a seismic 11 event. 12 There's no chance that would stick the secondary 13 PRVs, would it? Because you might need the turbine pump, later 14 on -- you might need the whole pressuring section area. (ph) 15 I just heard some stories about salt water sticking 16 recently. And-you will get salt water in it if you lose 17 condensor tubes. 18 MR. BLY: I guess. I'm not familiar with the 19 mechanism that would cause it. I have seen pretty heavily 20 salted up steam systems. 21 MR. EBERSOLE: You've got a high vacuum in the 22 condensor; and it'll be some salt uptake pretty quick. 23 MR. BLY: Oh, sure. 24 MR. EBERSOLE: And whether it's got in and got to a 25 concentration that would bug the equipment, I don't know. Heritage Reporting Corporation * (202) 628-4888 L
- fy 408 Kl 1 KR. BLY: There have been both' commercially and in 2 other service' fairly, extensive-salt. water intrusion events. .I
- i. .. .
+
3 'And I guess:all of those I'm familiar with I've.not seen-4 sticking of equipment. 5 I've seen eventual problems due to corrosion after [ 6 that's occurred. 7 MR. EBERSOLE: Well, it's important to you.to keep 8 pressure in the secondary. 9 MR. SMITH: I should mention, Dennis served on the : 10 enterprise, so he may be familiar with salt water, 11 (Laughter) 12 MR. SMITH: Skip to the relay chatter task. I think ,
-s 13 we've talked about the fuel oil system to some extent, and what
~k )
14 we're going to do about it. 15 We're going to look at relays and try to develop a 16 little bit more specific information rather than trying to use 17 generic information which we did in Phase 3A; and also we're [ 18 going to look at the recovery actions. i 19 And I believe that's all I had to present as far as e 20 graphs. One question that I think we anticipated, and that I , 21 can answer rather succinctly, is what -- basically what have we j 22 gotten from the PRA at this point. E 23 And I think there's a couple of things. One is an 24 identification of vulnerabilities and fuel oil system is an 25 example of that, f f () Heritage Reporting Corporation I (202) 628-4888 i I i
; .g ,
t 409-1 We considered that going 11nto the PRA to be a fairly 2 reliable system. And in fact we still do. But it's_ turned out 3 to'have a failure rate higher than we expected.
~4 And then' number two, I think that we've look at --
5 the PRA has identified that our operators hadn't anticipated;- 6 that need operator. reaction. And one of those_was the loss of-7 CCW to the charging pumps. 8 We had -- wasn't something that the operators were 9 too concerned about.. And during the course of the PRA, .in 10 fact, I think we had an incident at Farley where that exact 11 thing happened. 12 So the PRA has identified those types of incidents
,_q 13 for us.
14 So those two, I think, are the main things that we've 15 gotten so far from the PRA. 16 MR. EBERSOLE: May I ask another little shock 17 question. When you look at the batteries do you look at 18 degradation of the internal plate structure in the context of 19 seismic damage? 20 MR. SMITH: That happens to be one of my favorite 21 topics. 22 MR. EBERSOLE: I would have guessed it, I guess. 23 MR. SMITH: The batteries -- you know, we do testing 24 on batteries periodic.tlly. We also read Ellis Sandia 25 literature about battery life and seismic fragility. O Heritage Reporting Corporation (202) 628-4988
,0 cs -
410 (~~\ - 0 A-J 1 MR.'EBERSOLE: I understand. You can cut it off
-2 now..
3 MR. BAGCHI Mr. Ebersole? 4 MR. EBERSOLE: Yes. 5 MR.' BAGCHI: That we'll come back'to. I know this l 6 was t.ested by -- I believe 535. [ph) It was -- this being I ;7 developed at-that time. 8 So:that consideration has gone into it.. It has -- it 9 hasLa design life incorporated.. 10 MR. SMITH: One of the things that we anticipated, I 11 we've-already changed our batteries out once. 12- MR. EBERSOLE: You know my view of nuclear plants is 13 thent all are like an inverted pyramid sitting on the batteries.- ,
. (s]' 14 MR. SMITH - I agree, since I'm an electrical 15 engineer.
16 (Laughter)
. 17 MR. SMITH: Actually, they're sitting on 4-KV, too.
18 MR. EBERSOLE: Which sits on batteries. 19 MR. SMITH: Exactly. [ 20 DR. SIESS: Let me ask you a question. And don't ; 21 devote more than a minute to the answer. 22 Suppose the Board of Directors of PG & E looked at 23 this and decided that 2 times 10 to the minus 4 probability for 24 core melt didn't make them feel too happy; and they asked you
- 25. what you could do to reduce that by, say, a factor of five? l t
Heritage Reporting Corporation 1 (202) 628-4888 7 _ , ,.._,_ ~ - . - - . .- _
4 p L 411 i.,. A_J- 1 Have you got_some ideas about -- from your PRA as to
~
o ! 2 --
~
- 3 MR. SMITH: Well, almost_specifically_I can tell you 4 we need to work on-electric-power. That would be the1 area that-5 I would be concerned with; I mean, that would give us the most i
6 benefit.. 7 I don't know that it would give us that much benefit, 8 .but it would give us the most benefit. 9 DR. SIESS: There's not any one thing in there I 10 .could see that would count for a factor of five. 11 MR. SMITH: No, not any one thing. 4 12 MR. BLY: The things we're working in Phase 3D'is a 13 . group associated roughly all -- somehow associated with seal
\ LOCA -- under that category. But that's what we're working on. .
14 15 And we've made some progress. 16 DR. SIESS: Even though that small LOCA -- seal LOCA 17 -- looks big it's just not that big'. Reducing that to zero 18 doesn't -- 19 MR. SMITH: No, I don't think that it -- 20 MR. EBERSOLE: -- doesn't give it to you. 21 If you have to go to feed bleed are you going to make 22 up baration all the time? 23 DR. SIESS: No, let me finish. 24 MR. EBERSOLE: Okay. I thought -- 25 DR. SIESS: Mr. Brand's got it. O Heritage Repcrting Corporation (202) 628-4888
Exi 412 / 't 1 MR. BRAND: Let me address Dr. Siess' question a 2 moment because from a policy standpoint that's one of the 3 things I'm looking at, certainly in terms of a major benefit 4 coming out of the PRA in terms of how we, from a policy or 5 management standpoint look at the ultimate safety of our plant. 6 Based on the information I've received thus far from 7 our staff there is no one, or two. single, or small number of 8 items that would allow us to measurably improve the ultimate 9 safety of the plant. , 10 At the same time we view this tool as really an 11 optimal opportunity for us to continue to probe and push the 12 plant -- certainly now while it's still in its early life, but
,3 13 also ultimately throughout its life, assuming that we were to b 14 maintain those tools throughout its plant life.
15 MR. KENNEDY: Let me urge you to urge your PRA people 16 not to be conservative. 17 (Laughter) 18 MR. KENNEDY: But realistic. 19 MR. SMITH: That's high on our agenda. - 20 MR. BRAND: We've heard you make that point before; 21 and I think this meeting, again, underscores your point. And I 22 think we are listening to you a little -- 23 DR. SIESS: It seems to me -- 24 MR. BRAND: -- we'll have to view this -- 25 DR. SIESS: It seems to me possible -- and I ()
/~s Heritage Reporting Corporation (202) 628-4888
3 gt%- - 413 (s' 1 _ recognize you're_not finished yet -- that you have done a very 2 through PRA; you may have picked up things that aren't in the @ 3 PRAs. o
< 4 Although I recognize you have a PL & G representative 5 who is generally familiar with what's being done. So -- you, 6 .in comparing your results with our results, I think you-need to.
7 examine the thoroughness with which these other results have. 8 been achieved. 9 MR. BRAND: Yes, sir. 10 DR. SIESS: Any other questions or comments from 11 other members of the Committee or consultants?
- m 12- MR. EBERSOLE: Well, I think I've been impressed by~
13 this PRA approach in which it's not just a PRA man in charge O- 14 but there's a mix. 15 DR. SIESS: That's what it takes. 16 The staff have any comments to make at this stage of 17 the game? I realize this is the last stage of the project that 18 is pretty early. 19 MR. BAGCHI Well, my request really is that as you 20 make a report to the full committee, perhaps you would 21 incorporate specific advise to the staff. That would be 22 helpful when we review.
^
23 DR. SIESS: Like we gave you yesterday. 24 MR. BAGCHI: Well, more than that. 25 (Laughter) (_/ Heritage Reporting Corporation (202) 628-4888
414 1 MR. BAGCHI: More than that. -- Used to look for 2 3 4 5 6 7 8 9 10 11 12 13 O 14 15 16 17 18 1 19 20 i j 21 22 4 23 ! 24 25 l l O Heritage Reporting Corporation (202) 628-4888 i
U 415
\_> 1. -Mr. Davis had some points --
2 HR. EBERSOLE: Let me ask a question here which will
- 3. reflect my old TVA b'ackground. It's the identification of 4 safety issues which have no-guidelines in the regulatory 5- context.
~
6 If you or your people' identify a safety issun, and 7 you can't find any relationship to a printed regulation or 8 guide, how do you process it and dispose of it? 9 DR. SIESS: You put it under the severe accident 10 policy. 11 (Laughter)
-12 MR. EBERSOLE: No, I'm talking about -- and there are 13 many cases where you can't find a regulatory relation unless
[. ,
' l' 14 it's very general, and a very wide interpretation.
15 Do you have a process of racheting that up to 16 appropriate levels for disposal, or how do you handlo somebody 17 who comes in who says, "I think we've got a problem here but I 18 can't find it in the book"? 19 MR. BRAND: Yes, I think we have that process. I 20 think the process that we employ is the same process that ue 21 have for those issues for which there are stated guidelines. 22 But we do have a process in place that allows our 23 people to make those queries and move them through our 24 organization appropriately. 25 MR. EBERSOLE: -- And all of that happened back east, Heritage Reporting Corporation (202) 628-4888 t=
416 O,. 1 was it -- it wasn't in the book; it wasn't a safety problem. 2 That's not right. 3 DR. SIESS' Gentlemen, 11:57, 58. Meeting's 4 adjourned.
.n 5 (Whereupon, at 11:58 p.m., the meeting was 6 adjourned.)
7 8 9 10 11 12 13 14 15 16 17 18 19 20 l 21 22 23 24 25 1 en
- U Heritage Reporting Corporation (202) 628-4888
i 1 CERTIFICATE () 3 This is to certify that the attached proceedings before the 4 United States Nuclear Regulatory Commission in the matter of: 5 Names DIABLO CANYON LONG TER21 SEIS!!IC PROGRAli 6 7 Docket Number: 8 Place: San Francisco, California I -9 Date: February 24, 1988 10 were held as herein appears, and that this is the original 11 transcript thereof for the file of the United States Nuclear l 12 Regulatory Commission taken stenographically by me and, 13 thereafter reduced to typewriting by me or under the direction 14 of the court reporting company, and that the transcript is a i () 15 16 true and accurate record of the foregoing proceedings, oh M
/S/
17 (Signature typed): 18 Official Reporter i 19 Heritage Reporting Corporation 20 21 22 23 1 24 25 9 () Heritage Reporting (202) 628-4000 Corporation
L' A
,' "'3: ,
f1
^
1 IO I s e i
}
i
]
t DIABLO CANYON FRAGILITY EVALUATION k t i 1
.k l
^
lo R. P. KENNEDY . FEBRUARY 1987 O: g si,mae, u x a ,comiemor
g> , . n
- 4 O
STRU'CTU'RESFRAGILITYDESCRIPTION STRUCTURES' CAPACITIES'ARE DEFINED AS CUMULATIVE-DISTRIBUTION FUNCTION REFERENCED TO RESPONSE OR - INPUT FUNCTION ( FAILURE CAPACITY v A=At au t A = MEDIAN CAPACITY REFERENCED TO. GROUND ACCELERATION l OR RESPONSE QUANTITY (INSTRUCTURE ACCELERATION, FORCE,DI$ PLACEMENT,'ETC.) . t
.. O ARE LOGNORMAL RANDOM VARIABLES, MEDIAN = 1, t
g, t U 8 R, 8 U ARE LOGNORMAL STANDARD DEVIATIONS , 4 e O . g Si,ucio,.i u .cs.mic, com,oia ,n
l;\
,.g ,
jll , [_ s , l - , , e u v , e - t , h , t e .
? / .ic.
wn. e . , sd . e. ei . vf rn ua
,/
cC , yt w t n i e w l r w
. .f,.
i e gf af ri t e n , FD t -
/- a -
- r 7 e l,
e v
/ c c
v a A 0
/ d n
u
-- , . o S g[ 0 r
=
f.- G )
'h O
/ .n I 3
/ i a . (4 ,T A~
h ~- ./ d e . n .T-N t .M o i w ./.
/ .
f-i t a E S E e . r e R v _ oC r u y
/
/
. - /. /,/ f, A
5 l A e c c . P E R E t i
/ ,
./ ~ l V
/ R l a r .U i
g
~a,
/
/ t c
.C
,1 r I , e p Y F ' T
/ S I
/, -
k L l a I e G ll P A R l j, F _ /
/ -
/
l s.- ,
, - - O
- - t
*- 5 F 0 5 r o ,
0_ 7 2 0 t 0 c 1 0 n,
*2 c5. Io}a,_n nO3c_O.3sm
e. 10 . PHASE il FRAGILITIES BASED ON SAME METHODOLOGY WITH'SOMEWHAT GREATER EFFORT THAN PREVIOUS COMMERCIAL PRA'S 9 TPAGILITIESKEYEDTOSPECTRAL-AbCELERATIONS IN THE 3 TO 8,5HZ RANGE SOMEWHAT MORE ACCURATE SOMEWHAT LESS UNCERTAINTY O. t ' NUMBER OF CYCLES APPROPRI ATE FOR LONG DURATION (M >6.2) EARTHQUAKES e 1 i, O' Structural Mechanics Consultine
O - I FRAGILITY DESCRIPTION i A = I Ag
~
f=MEDIANFACTORORSAFETYONCAPACITYABOVETHESAFE SHUTDOWN EARTHOUAKE LEVEL SPECIFIED FOR DESIGN . CONSIDERING STRENGTH. DUCTILITY AND STRUCTURAL RESPONSE t Ag = PEAK GROUND ACCELERATION OF THE HOSGRI EARTHQUAKE O -
"k'u' s RS
- RE is STRENGTH FACTOR i
fy DUCTILITY FACTOR igs STRUCTURE RESPONSE FACTOR i RE = EQUIPMENT RESPONSE FACTOR O Structural Mechanics Consulting
(.
. :0 - .
JPECTRALSHAPEFACTOR f, RATIO OF DESIGN RESPONSE: SPECTRUM AT DESIGN DAMPING TO MEDIAN
- SITE-SPECIFIC RESPONSE SPECTRUM AT MEDIAN DAMPING c TURBINE BLDG.: f = 5.7 HZ DESIGN: 71 DAMPED NEWMARK - HOSGRI ANCHORED TO-O.558 - San = 1.3088 LO MEDIAN: 10% DAMPED SITE-SPECIFIC ANCHORED TO 4
0.758 - SAg 1.360 l 1-
- DESIGN SPECTRUM INCREASED-101 FOR TORSION WHICH ACCOMPANIES TAU FILTERING I-l . 1.30 (1.10) i FSA = = 1.05 e 1.36 4
4 O g a,ono,.mun.*, coa umn,
O . FREQUENCY, CPS t oo 10. 40. 40, 30. 20. 10. 8. ' 6: 4 3. 2. 1. 8 .6 4 .3 .2 .1 3'8' rg .'t , LffAV J 7/Frg&gy',,/sgh&.F l bEiW-g wN y t PP,+ t xrfg'
", 77IW t'e k .i4(h gKh;h\;6 .##gi.%X/. <
N '.NR'fs'12K% ; W W.'yN)@hlMMR>6K'>/Gil.BM.,, .I
' ';, h A x //v' , .
{\x 's; . * , X/ q , s
.y4//
.Ig, g( Q, n . ' %p N >N y .?x.u .e,F /v '..s 9W 03 W,m- . c'.>N .
y& W;<,;<f cf 1ec.
.. . . ;- n. , 7 .s , . , . . . . ,
.. , .1 . . . ,
e s ,..a
., n . s, ,...,.,,,, . .-6i . . . . . , , , ,, s,-
m .s . .. ...2..... ..... ,, ,,s s W . ,,.r..,....,j.,d...,.,,,,,-
.v v , c _ ..; . . , r is . , ...,s, ,,,,,, .. ,s, . . .,.,,fa p('. .
/ / / i /. V N Q,c , t,.' .
,,%t .r, ..
4fi'V// r A
,. , ,2, s ' eW.V.
/} nN'Ly/fy / / yo ' yq/ t/q y.cp g
..C,,).a:.,4 *rn X/ YY l XV'Y'V17.l 'iY1,7 **
%'1, :.Y.//.Y v 7.I'M V X f t 1. t.6.'/XY f ;< QV'Yt l
N.M%r %}1%rz0 b r % X s. 'M2 s ,w n vs rv/.<,DitDiN ,_ i, i y s . v '.:hM%r A0bMR{s
,,- 4 o.
- ,;, s s c.., r ..f .
.s s ..: ,; . r so , v , ... e . , . . x w ,; ..
kTl fN ^ ). /V /M JN K 4 ei ,i .i ;, s /. , .'N ef , 1 N/Aii.aj/se.F.$,V v4 V1 -@? <; ,1,4 s '* g'
'Q. XANXiX ,4M.W/A.KlX.K N W X , g ,
/ e).nlXAbx1sx l ! #1. I ><// /GU
'tH1X/'/N NN kWYVWZ a .W'>NGkN Wh'#N'hl'WhMStN '.N.'\x u )$,# 'Ts'X D.N8}'d>'IMdMs'>'.\Tl@$l'X:OdDf-W 488}'2 M' 7 * ,' o
.' O 10 k,,
g
/, s. ,e.,,v,
(
..,../ ,,
,s , ,.a , - :..[ ....
s,. .,s_ ,
, s 0, 10.
=.
U 5,s. , . r e .. a. fxn_.. D x.s y > . .,: . , v. r. , , , n < . , s. . , r. . P. , . . . i ,so.> M*~/t""'! rs . r,,;. v 6. *a v r/ms.vis.r /. r, ,* 5 g ytw ei.v// /jvra vis ,3 mpo vn .wiv//.,v_y vg .w yw.sugixnrwxnr n.pxaw,i;<g ..g Av_ A O, ,- g ,- v,i.. w i*r v . w+%r . . .. s u
/xerw.as awm.n.r. y . o,xy y w , ,,;. we, 4. g ,.
a v,. >. ,.v.m .,. .y ,v,,n,vu, nv%.i,..,,w..<. & v,,.
, s r.., ,s , , . ~ , s,;
, vn . .
v.<. w ,i 3o a '- w ;<r.m<i.xi e m . . vixon.x
-a, xwn:a.:. .a* . sfN xmi,sn 4,s. _;
MRP VMMr.i,/., ASAN $
'Tx^Wl.W @jx/M2
$~:Wi:M VN AW' d'A% 3X/%W
>Ufx% & MMAM> >fkV/N' $'%X's'MM'/
wa
.3 m . .. 2 x
m w$.N:< e #^q'/
- ,t 7
! DV 3 ,n% $'s$W, ~);'b'.%_ 3 Y $-
.s .. b .T / I*
r' I
' . s 'V.1 i' 0' ' , ' V rvh..V.
l' FV \ './ /.1 _ t . N r . 9 , . . . r1 i' I-A 6 4 . /s /v ., , , . . ("7 , . ,. A N
* ,,3 A r. . ~ _ 'O 8 '
.i'h4 ," I s A w* .s .
'8 ,w%.N /
,e/ /e / .t . t .' i e.4-0
' ' ' " .V ' .# A ' ' "s.' ' '. 'a v' . '* *8
.l~i. . , $. s h . .. U. @ s i // /* 0 e /./Fv $ 'ii fy,.
'a
.y/sv r. ! yqua w .vn.kirv/// A n t 4svsii/1/vevst vs rw T 1 X L'X / .t x X F4'v- X) '> n.:t;t /./ fy IX s/N W'O.,O X 6; ViXvsrt'1,xi'i vli *'
3Gh%( /Xh(W'.3W&Y%$%(/Y *Ca ,c' N '.Hh?H % ( / V Z>( M N LN Y,0 .,
, n
,a , r ., w v , .x m ,, i v. i . i . , . s.v
*0
- p. .'\/Y /s , s., r ,s r, ,s s .u// .
X'V t/M ?N 4;4/ pit' e t sM
*3
,3 X,4,NA 4 GN/ ,/4 J ttgg..pl .t.Vs,y,o, gal It IX .0 s w x.Aix x @ l .t.1 lxrMA.K Nusix^r.7,'xl s%) 4 n/,WWlN'E..Ixw fu;Ix/.o,86\'N WI/N"o oP 'do MW1 P,W}Xf1(hF ,
l ,
~XNTDiN/
sb& b O<$ N?/Nfc Y'Jk.Y/Y H Y'h W/h' $ N'
*g, N s.
es , ,es NM. ,< p ' M'k..h*gy.-M.7
, e < . e "T.f , ../ c pe ,, i s N .W"< i 7 i. ,
C r~ s .--.
.3. i
< x ei-A M#$$s@M*)
7O u . . . . , . ,. . s g y4 ' . .
.,4./
,s .4
/ ' r
*g,
- r. i . e/ rv , .. e y g mN, .
r /T: ,i n .. . . r , , // n_ . vs r. . . c.i e. , 6, yqe v4 r s v
* '4 'I ' ' ' E #* "Y DN ' / y .v., f , .vq , e ...if f ff fy y....';/fv ; xy ry.n xisr y t y f y y y l,'s
- xxi sqvy,f I
l
.o a X'V yx/.rv i xy c 3*P Af'y
.c o s.m o s .o2 .o .e4 .os .os . .s .4 .s .s 1 2 3 4 5 s . t.o .
PERIOD, SEC , DIABLO CANYON MEAN GROUND RESPONSE SPECTRUM (5% DAMP!NG, NORMALIZED TO 1G)
& ~
l l
._ O .
S0TL STRUCTURE-INTERACTTON
' SOIL STRUCTURE INTERACTION EFFECTS'ASSUNED TO RESUL1 ONLY FROM, STATISTICAL INCOHERENCE FROM THE GROUND MTION hWE a
e FOR A'150 TOOT PLAN DIMENSION FREQUENCY'(HZ) - R'EDUCTTON 5 l','O 10 0','9 . t 25 0,8 - 8 LINE'AR REDUCTION PROPORTIONAL TO PLAN DIMENSION . I t 8 O . l v ,- --.-,-,en,ee v, .. ,,.-m_ . , . , , ,,,,n,-,.- , ,. n-. , - ~, e ~ ._.. - - ,,e,-----n , . - - - . . ,,, - , - , , - - . ~, , e,e..m-
~ ' '
~
~-i.
Sy , e. PROBABLE 00MINANT CONTRIBUTORS (FROM PHASE II) . MEDIAN VARIABILITY HCLPF COMPONENT SA (G) BR SU 8A (G)
- 1. DIESEL GENERATOR 4.22 0.19 0.33 1.79' PLUS PERIPHERALS
~
- 2. 4160V SAFEGUARD-RELAY 6.34 0.20 0.50 2.00 DANEL Q 4.49 0.26 2.07
- 3. 120V AC INVERTERS 0.21
.4. REACTOR COOLANT PUMP 5.84 0.30 0.32 2.10
- 5. - C0i?TAINMENT FAN C0CLERS 5.04 0.27 0.28 2.03
- 6. 4160V TRANSFORMERS 5.39 0.25 0.24 2.40-7.- 480V BREAKER CABINET 4.77 0.08 0.23 2.86 l: 8. DIESEL GENERATOR 3.38 0 0.18 2.88 CONTROL CABINET
- 9. TURBINE BUILDING:
l- LOCAL DISTRESS 4.4 0.13 0.25 2.4 OVERALL DISTRESS 6.2 0.30 0.38 2.0 l l 0 c{cpy si,ocio,ous mic. com,emme.
. i GROUND MOTION RANGE THAT-00MINATES SEISMIC RISK (FROM PHASE II PRA)
SEISMIC RISK DOMINATED BY: S: A 1.75G - 2.75G A r,: 0.75G - 1.45G BELOW SA = 1.75G (Ao = 0.750) THE SEISMIC RISK ONLY COMES FROM SEISMIC LOSS OF 0FFSITE POWER COUPLED WITH RANDOM LOSS OF ONSITE POWER ([) . S 0 9 g sau c o, u u n oes cooit
E FRAGILITY 0F PROBABLE DOMINANT: CONTRIBUTORS II)
' Natior at NIS Engineenng Technical
-s Systems
't) . .
- MEDIAN VARIABILITY HCLPF COMPONENT SA (G) 8 8 R 0 SA (0)
- 1. DIESEL GENERATOR 5.2 0.25 0.22 2.4 :
PLUS PERIPHERALS
- 2. 4160V SAFEGUARD RELAY 8.4 0.29 0.29 3.2 PANEL (STRUCTURAL)
- 3. 120V AC INVERTERS 6.0 0.30 0.31 2.2
- 4. REACTOR COOLANT PUMP 10 1 0.32 0.31' 3.6
- 5. CONTAINMENT FAN COOLERS 6.6 0.30 0.39 2.1 O 6. 4160v e0TENTiAt s.0 O.28 0.30 1.9 :
TRANSFORMERS
- 7. 480V BREAXER CABINET >10.0 -- -- --
- 8. DIESEL GENERATOR 4.8 0.28 0.16 2.3 CONTROL CABINET l 9. TURBINE BUILDING:
l LOCAL DISTRESS 4.2 0.24 0.25 1.9 OVERALL DISTRESS 4.8 0.24 0.28 2,0 (1) PNASE IIIA FRAGILITIES FOR COMPONENTS DEEME0 TO BE DOMINANT CONTRIBUTORS BASED ON PNASE II ANALYSIS l O 1611 ;
; ,cq
~
.. m
.. 1
-[QJ
- w
- r CONCLUSION o- '0N THE AVERAGE, PHASELIIIA HCLPF ARE NOT SUBSTANTIALLY
,1
^
DIFFERENT FROM PHASE ~II FRAGILITIES o IN GENERAL, BOTH MEDIANS AND RAND 0MNESS HAVE INCREASED i o HAS ONLY A SMALL IMPACT ON PHASE I-I RISK ESTIMATES o STILL D0.NOT INCLUDE SIGNIFICANT SSI STUDY RESULTS, i WILL-RAISE MANY HCLPF ESTIMATES
) .
, l' i' O structurai sechanics consuiti (CPf l
l L p. l .>- ; l} l Current: Work
. Revising Fragility Estimates of All
. Components to Incorporate:
a More Realistic Site Specific Spectra Shape a Median Centered Responses
. (1 Analyses for All Buildings to Site Specific Spectra a Incorporation of Soil-Structure-Interaction Effects Including Spatial Variation of Ground Motion l-r m
i) ; s ':
~
p. j{)I' 7 REALISTIC SITE SPECIFIC SPECTRA USED o FRAGILITY REEVALUATION.AND= BENCHMARKING USED REALISTIC
- TIME HISTORIES APPROPRIATE FOR AVERAGE 3 TO 8',5 HZ, 5%
~
s DAMPED SPECTRAL ACCELERATIONS FROM 1,75 T0 2,75 G o 12 SCALED EMPIRICAL RECORDS AND 14 SCALED NUMERICAL
. GENERATED.H0SGRI RECORDS ARE BEING'USED FOR PHASE IIIB TABAS, IRAN DAYH00K, IRAN :
GAZLI, USSR PAC 0lMA DAM OTHER SAN FERNANDO RECORDS bx IMPERIAL VALLEY RECORDS PARKFIELD - TEMBLOR c MORGAN HILL - C0Y0TE' LAKE NAHANI l'
- PLEASANT VALLEY PUMPING PLANT-
'o - EACH RECORD'IS SCALED SO SA5%
FOR-AVERAGE.0F TWO HORIZONTAL COMP 0NENTS EQUALS 2,0 G ; o- NS SPECTRA.- 24 EMPIRICAL RECORD COMPONENTS (BOTH COMPONENT ,
+ 14 NS NUMERICAL RECORDS L o. EW SPECTRA - 24 EMPIRICAL RECORD COMP 0NENTS (BOTH COMPONEN
+ 14 EW NUMERICAL RECORDS OL .
Structural Mechanics Ccsnsultir i
u , O. O V 7 1%/-
~
6 i i i i i i i j _i i i i .i i i i 84 PERCENTILE
-- 50 PERCENTILE f
5 cn I e in
- Z 4
. O ! H W i < m W J w 3 . r.'.~.., u '. o '~.
< i i -- s b.
J f N '.' v% 1 4 _j ., g 2 f 3, , I- \
. /~.*
u \ . W N N h ., Q- % % in 1 - - - - - . _ _ _ _ _ _ _ _ i ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' 0 10 e 10 1 10 2 FREQUENCY-CPS
; 50 & 84 PERCENTILE GROUND SPECTRA ( EMP.+ EW-NUM.)
a t PROBABILISTIC AND-DETERMINISTIC MEDIAN FLOOR SPECTRA T o DEVELOPED PROBABILISTIC MEDIAN AND 8tl% NEP FLOOR SPECTRA FOR CORE-WEST PORTION OF AUXILIARY BUILDING USING SIMPLI-FIED BUILDING MODEL AND CLASSI FOR C0HERENT SSI ANALYSES INCLUDING TJME-HISTORY VARIABILITY AND STRUCTURE FREQUENCY AND DAMPING AND ROCK MODULUS VARIABILITY DEVELOPED DETERMINISTIC MEDIAN FLOO.R SPECTRA FOR CORE O o WEST, C0RE EAST, AND CORE CENTRAL PORTIONS OF AUXILIARY BUILDING, TURBINE BUILDING, AND CONTAINMENT USING MORE COMPLEX BUILDING MODELS FOR C0HERENT SSI ANALYSES WITH : 50% NEP GROUND SPECTRUM AND MEDIAN BUILDING MODELS , 1 O : suu a m, i u x s.oic, com,ein c{cpy
'hi O: PROBABlLISTICFLOORSPECTRA -
- o. 38 TIME HISTORY ANALYSES EACH USING DIFFERENT INPUT TIME. HISTORIES (SET OF TWO HORIZONTAL COMP 0NENTS) SCALED
.TO SA5%, 4,8-14',7 '
- SELECTED BUILDING AND ROCK PROPERTIES o MEDIAN PROPERTIES
~
FREQUENCY ~ 8.0 HZ
. STRUCTURE DAMPING = 7%
- O o VARIABILITY MEDIAN a 90% CONFIDENCE BOUNDS-
~
l , STRUCTURE STIFFNESS- l',0 0,50 0.44 - 2,3 I i STRUCTURE DAMPING 7% 0,35 4% - 12,5% ROCK SHEAR MODULUS. 1,0 0.45 0.48 - 2,1 (q>y si,ucturei y eca arc, cons,
~
O O O - 12 i i i i i i ij i i i i i .. , , 84 PERCENTILE
-- 50 PERCENTILE 10 '
on i e in Z 8 o e-o Z E i w i
/
\-
J g / w /
\i u
o / s.
< j
/g \ .
J / / \ i.
~
I4 H /,.-
^
. / j
' \
\
/ \
U
* / / \
/ / \ 's.
m 2
/ ,/ 'Q x__ W
/'
0 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' 10 a 10 1 10 2 FREQUENCY-CPS 50 & 84 PERCENTILE EW SPECTRA AT EL. 140'-0" (SI DAMP.)
+ _ _ _ _ _ _ _ - _ _ _ - ---
uewUFm< O n 3
<oOw3 wm<Faoz mal G 1 1 10 2 4 6 8 0 2 0
a S cS
' aZ,
- l e
dD U A
' pH P
' pS I
~i eNi A r cG t
U na X ' n i E H '
/. *. ' < i C F ' ti I r i O R J RE '
A EQ ' , O U HE 1 ' i EN 0 j SC 1 1. T TY - \. EC P L S ' 1 - 4 ' 0 MM
" o o d d
- i i 0 f i i f
" ' e e, d d
' T P a a b c
' a o, s i m
a 1 ' 0 2 O '
e 7 4 O COMPARIS0N10F"FLOOR' SPECTRA o ' MEDIAN-AND 84% NEP FLOOR SPECTRA - PROBABILISTIC
- DETERMINISTIC o CLOSE AGREEMENT BETWEEN MEDIAN PROBABILISTIC AND DETER-MINISTIC FLOOR SPECTRA FOR AUX BLDG CORE WEST
VARIABILITY DOMINATED BY TIME HISTORY VARIABILITY AND NOT BY STRUCTURE OR ROCK PROPERTY VARIABILITY
MEDIAN PROBABILISTIC FLOOR SPECTRA CAN BE APPR0XI-MATED FOR T!ilS CASE BY DETERMINISTIC SPECTRA USING MEDIAN GROUND SPECTRA AND MEDIAN STRUCTURE PROPERTIES 9 .
EXTRAPOLATE WITH CARE TO OTHER STRUCTURES - SOME SPECTRA BROADENING o' ALL PHASE IIIB FRAGILITIES WILL BE DEVELOPED USING THESE MEDIAN FLOOR SPECTRA o WILL REDUCE RESPONSE VARIABILITY FROM THAT USED IN PHASE II AND IIIA AND THUS INCREASE MANY HCLPF CAPACITIES O
(g>y si,ucio,.i uecs mic, com,uii
~ . .
=
.b CORRECTION FOR GROUND MOTION INC0HERENCE o MEDIAN FACTORS BASED 0N SSI INC0HERENCE: STUDY
SMALL CORRECTION (GENERALLY'BETWEEN 0,8 AND 1.0)
SIMILAR T0-VALUES PREVIOUSLY.USED IN PHASE II AND-IIIA , 1 o VARIABILITY FACTORS WILL BE BASED ON CONSERVATIVE JUDGMENT WITHOUT VARIABILITY STUDIES
-MEDIAN CORRECTION IS SMALL .
VARIABILITY STUDIES ARE COSTLY AND UNWARRANTED FOR-
- d. SUCH A SMALL CORRECTION
WHEN COMBINED WITH OTHER VARIABILITIES, EXPECTED TO HAVE NEGLIGIBLE IMPACT ON COMPONENT HCLPF S 8-MEDIAN U R HCLPF 1.0 0,04 0,03 1,12 0,9 0,06 0,04 1,06 0,8 0,09 0.07 1,04-6 O
g si,uauroi u x semic, com, unio
~
- i e , 0
.4 -
21
' e ,
c n
' e ,
r e
' t h ,
u o p c
' I n n i ,
t r n of r e , e d h e o t C s u U
- r. j o d -
f A ~ mm u u 0 r r 4 t t
' c c ,
1 e e p p S S
- S L P E
- E
- .\.
1 Y
\ CT:
I S, 0 NS m~
' . 1 1 1 EE
' A \ , UH Q
i
!l'- - ,
E E R R [ F 8 - O
!. C i
'j- ,
*- S i ':'
N a r ~
~ X t
G ce l*f U
' N p I S A
P p M U A D ed l i i r ac S S e 2 0 8 6 4 , 0 1 - 1 1 m cI n)U 2Oe-ob4WW3WOU<
- 1_4EyuwDU.) . !ll lll l l ; l ll l1l \l 11 l
O O O - h (o Pf- PH O PPPh 1.2 i i i i i iij i , , , , , ,,
----- S r D AMP I NG ,
1.0
--~~
~~~-~ ~
_~---
- - . ..~~
.~~a. .
g .8 s F-4 0" _a - 4 .6 tr V-U to Q. tn
.4
.2 I
.0 10 e 10 1 10 2 FREQUENCY-CPS AUX NS CORE WEST EL 140' __ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _
E,
'g -
4 w CORRECTION FOR CONTAINMENT UPLIFT o MEDIAN-FACTORS BASED ON SSI UPLIFT ANALYSIS FOR-
~ '
S A3-8,5,5%
~
.O .
CORRECTIONS ARE VERY SMALL l t. l ll 0 VARIABILITY FACTOR WILL BE BASED-0N CONSERVATIVE JUDGMENT WITHOUT' VARIABILITY STUDIES L' . l; i n O struao,.i uun.oic, comiu. c(g>g{
O O ~ O - 12 i i i e i i i i i i i i i i e i
' Sr DAMPING LINEAR (AVERAGE) ,
NONLINEAR (AVERAGE) For 53 (3-8.5 liz) = 2.25 9 , 10 cn 8 i S (n Z 8 ' A, o s-< f F f K
^I
)
d6 o
~
o __1
< 4 m
l- - U w Q. in - 0 ' 10 e gg 1 10 2 FREQUENCY-CPS CONT NS INTERNAL CENTER EL 138'-6"
TURBINE BUILD G NONLINEAR ANALYSIS 5 NTS Enginearng ~ TecL*al Systems PURPOSE: DETERMINE THE INELASTIC LOAD DISTRIBUTION AND SEISMIC RESPONSE OF Tile l'URBINE BUILDING APPROACil: CONDUCT SERIES OF TIME IIISTORY RESPONSE RUNS OF 2D (E-W) TURBINE BUILDING FGHLINEAR MODEL e NONLINEAR MODEL WITil INELASTIC SilEAR WALL, DIAPHRAGM L TURBINE PEDESTAL ELEMENTS 0 25 TIME IllSTORY RUNS AT SPECTRAL ACCELERATIONS OF 3 AND 6 g WITH MEDIAN STRUCTURE PROPERTIES S 50 TIME HISTORY RUNS AT 3, 4, AND 6 g WITH MONTE CARLO VARIATION OF STRUCTURE DAMPING AND STIFFHESS e FAILURE DEFINED AS 0.7% MEDIAN STORY DRIFT WITil 0.16 PROBABILITY OF FAILURE AT O.5% DRIFT 1611
g . TURBINE Bull. DING PROPERTIES
- Nations!
NTS Eng-netong . Technical 4 ^N t
- b. ems-H00AL FREQUENCIES ME0!AN VARIABILITIES WALL fl9 -
8.6 HZ FREQUENCY: sp = 0.25' WALL f31 - 9.5 HZ STIFFNESS: 83 = 0.50 OPERATING. FLOOR - 4.0 HZ PEDESTAL - 3.1 H2 DAMPING: 7% ME0!AN VARIABILITY: 80 = 0.35 ils = 5% to 10% CAPACITY VARIABILITY: sc = 0.25 i ( d O
,$ s 1
0 4 DRIFT O 1611
,,,,-.6.---+- n., - - ~
TifLE PACT OF Jsb Na STRUCTURAL 8Y DATE I I MEOHANICS couygg7, j w"Sual~o g .. .. ..,E , Con /se S-sc<esn e assav,2 0 5 % % % % O : e s b i E w R b O b 2 - ds @n ii d %-kg~ da N N ge
'h n
4,$ -
.N ~~
u A o
.x D g
l , h k' Q
I g TURBINE BUILDING FAILURE PROBABILITY l y,','[ij,'h NTS Engmeenng stem. 0 50 RUNS - VARY EARTHQUAXE MOTION & STRUCTURE PROPERTIES LOGNORMAL j COMPOSITE SA = 4.64, ec = 0.36 S Py A PF - 3.0 11% 11% 4.0 34% 34% 6.0 75% 76% 6 25 RUNS - VARY EARTHQUAXE MOTION ONLY, MEDIAN STRUCTURE PROPERTIES LOGNORMAL RANDOM SA = 4.64, sR = 0.23 O S A E F Py 3.0 3.1% 2.9% 6.0 88% 87% 0 SU" 8 -8R = 0.28 0 INCREASE FOR SSI & EARTHQUAXE COMPONENT COMBINATION S,= 4.8 $U = 0.28 ea = 0.24
- O 1611 l
r-g TURBINE BUILDING OVERALL DISTRESS FRAGILITY -
. Nellinal NTS Eng+,e ir.;
Technical s rsums O - SPECTPAL ACCELERATION PHASE II PHASE III MEDIAN 6.2 G 4.8 G l sg 0.30 0.24 s u 0.38 0.2S HCLPF 2.0 G 2.0 G 5% Pp - 95 % CONF. 2.0 G 2.0 G ME0!AN 3.8 G 3.2 G , 5% CONF. 7.1 G 5.1 G O l l 1 I i l ! O , i l 1611
v INVESTIGATE SEPARATION OF VARIABLES APPROACH TO FRAGILITY ESTIMATION TRADITIONALLY HAVE OBTAINED AN ESTIMATE OF MEDIAN RESPONSE AND ITS VARIABILITY BY LOGNORMAL COMBINATION OF INDIVIDUAL PARAMETER MEDIANS AND VARIABILITY SPECTRAL ACCELERATION STRUCTURE FREQUENCY SHIFT STRUCTURE DAMPING O EQUIPMENT FREQUENCY SHIFT EQUIPMENT DAMPING , PERFORMING BENCHMARKING STUDY TO VALIDATE OR MODIFY THIS APPROACH FOR DIABLO CANYON BENCHMARK CASE: EQUIPMENT RESPONSE IN AUXILLIARY BUILDING 200 LINEAR ELASTIC RESPONSE ANALYSES FOR 8 EQUIPMENT ITEMS BOTH HIGH AND LOW IN AUXILLIARY BUILDING. USES 25 DIFFERENT EARTHOUAKE TIME HISTORIES, ALL SCALED TO SAME AVERAGE SPECTRAL ACCELERATION FROM 5 TO 14 H2 < O ! Structural Mechanics Consulting 4
g O PARAMETER VARIABILITY O ^ Maleenal NIS Engineceing ' Technical s r.o . 9 STRUCTURE FREQUENCY; HEDIAN = 8.1-HZ a = 0.25 e STRUCTURE DAMPING: HEDIAN = 0.07 8 = 0,35 e EQUIPMENT FREQUENCIES; 4 HEDIAN FREQUENCIES CENTERED AT: 24 HZ 14 HZ 8 HZ
- 5 HZ s = 0.2 ,
e EQUIPMENT DAMPING; MEDIAN = 0.05 a = 0.35 e EQUIPMENT LOCATED AT 2 LOCATIONS IN THE STRUCTURE: LOW (ELEV 100') , HIGH (ELEV 140') O
~~ --,.--.,,r- e ,e.er-- - ,-.
O - O - WEIGHTING FUNCTION FOR HONTE CARL 0 TRIALS Nationag NIS Engwieesing - 1ecemnocal Syssesses
, STANDARD NORMAL FUNCTION PROBABILITYDENSITYFUNCTION+{)=he~"
WEIGHTEDPROBABILITYDENSITYFUNCTIONe{)= e" WHERE W{,) IS DETERMINED FROM:
~
3.0 i
- I l 0.275 I
- , I e i e i
_j -1. 2 ,'j _') -0.11f , i p . Wg ) = 0.275 FOR p > -0.11 W{ ) = 2.5p FOR -1.2 1 p 3 -0.11-FOR p <. l.2 W{ ) = 3.0 i
me o ,,
~ Q) e 4
National NTS Engeneering M:~ ' 8 HZ RESPONSE AT ELEV. 140' a 5-
+ SEPARATION OF VARIABL G ESTIMATION
! + o + 9-m
+++
+ .
i 5~_ s k. - r _es g +g _aa-W O i O i. 48 W ,*- I Q-11. o . zo J "._ o h* 9- ++ + _
-4.00 4 .20 4 .40 -t.so -b.so -b.oo o'.so i'.so a'.40 a'.ao
-STANDARD NORMAL. VARIATE (-S) . ..
m
, \, (%
uJ [pr ' COMPARISON OF SEPARATION OF VARI 3BLES AND TlHE HISTORY RESPON < Nageonsi _ Technic al . HIS Engmeer.co Systems . COMPONENT AND LOCATION 5 HZ I 8 HZ 14 HZ 24 HZ 100' 140' 100' 140' 100'. 140' 100' 140' SEPARATION OF MEDIAN 2.? 3.8 3.0 5.3 1.7 2.5 1.4 2.1 VARIABLCS a 0.31 S 53 0.41 0.52 0.32 0.59 0.30 0.26 2.48 6.1 13.6 8.0 18.5 3.7 10.3 2.9 3.9 MONTE CARLD MEDIAN 2.8 3.8 2.9 5.7 1.7 2.9 1.3 2.2 TIME HISTORY a 0.30 0.50 0.40 0.53 0.25 0.44 0.21 0.21 2.4a 5.8 12.2 7.6 20.3 3.0 8.0 2.2 3.8 e' e e- _ _ _ _ _ _ _ M
7 r~ - "
- g 7, j p, m, ..
>[ , (Ql . CONCLUSION -- SEPARATION OF VARIABLES. O 1 SEPARATION-0F VARIABLES-APPROACH SHOWS EXCELLENT' , . CORRELATIONWITH MONTE CARLO TIME HISTORY-METHOD r b 4 10 1 4 0 f I 6 s i t - O sne m , - - a c{cpy ,coe,on 3 4 " etwwr.y w
g _-
- a;.u w .
16' L; O '
~
RESPONSE CORRELATION e. o EXCEPT AT HIGH FREQUENCY (GREATER THAN ABOUT 18 'HZ), RESPONSES OF COMPONENTS WITH.SAME FREQUENCY MOUNTED AT DIFFERENT ELEVATIONS IN STRUCTURE ARE HIGHLY CORRELATED o RESPONSES OF COMPONENTS.WITH DIFFERENT FREQUENCIES ARE-ESSENTIALLY UNCORRELATED EVEN WHEN MOUNTED ON SAME FLOOR o FRAGILITIES OF ESSENTIALLY IDENTICAL COMP 0NENTS WITH SAME FREQUENCIES SHOULD BE TREATED AS DEPENDENT EVEN . WHEN MOUNTED IN STRUCTURE AT DIFFERENT LOCATIONS t o FRAGILITIES OF DiFFERENT COMPONENTS SHOULD'BE TREATED ! AS INDEPENDENT EVEN WHEN MOUNTED ADJACENTLY 4 I I 1 e f o { O : 1 Structural Mechsnics Consulti L
1611 I
> Table 1. Correlation Coefficients Between Component Response RESPONSE R1 ,5 R1 ,8 Rt ,14 R1 ,24 R2 ,5 R2 ,8 R2 ,14 R2 ,24 R1 ,5 1.0 0.25 -0.02 0.04 0.91 0.30 -0.03 0.32 R1 ,8 1.0 0.06 0.07 0.13 0.88 0.16 0.35 R1 ,14 1.0 0.31 ' 0.01 0.02 0.87 0.31 R1 ,24 1.0 -0.08 0.11 0.21 0.57 R2 ,5 1.0 0.26 -0.06 0.38 SYHMETRIC R2 ,8
- 3.0 0.11 ,0.42 R2 ,14 1.0 0.38 R2 '24 1.0 Where gR ,3 = response of "j" hz oscillator at floor "i".
Floor 1 = 100' elevation Floor 2 = 140' elevation 0-7
l' DESIGN AND CONSTRUCTION ERRORS N LOCATION OF ERROR (I)-
...y S
5 . g . . w 8 ERROR et Ame*ICATION E e ALREADY ELSERNME 98 PRA D
. ReAcis ONLY SeSARC FRAaEm g
e NO SIGNIFICANT RFACT ON CORE DANIAGE
. OR RELEASE FREQUENCES e COULD INFLUENCE CORE DARRAGE
" FREQUENCY l
l l e COULD INFLUENCE RELEASE FREQUENCY Pld ard,Lowe ased Osrelsk,Ine, e G #
e . l , c5aYE DCPRA REPORT e EXECUTIVE
SUMMARY
(~30 PAGES) - e TECHNICAL
SUMMARY
(~200-300 PAGES) -
- INTRODUCTION AND
SUMMARY
- PLANT AND SITE PERSPECTIVE -
- TECHNICAL APPROACH
- PLANT MODEL (i.e., SECUENCES, SYSTEMS, RECOVERY)
- DATA
- RESULTS AND CONCLUSIONS
! e RISK MANAGEMENT MODEL AND GUIDE
- INTEGRATED PLANT MODEL
- REPORT MAP
- APPENDICES (METHODOLOGY, EVENT SEQUENCES,
- SYSTEMS, EXTERNAL EVENTS, HUMAN ACTIONS, DATA, '
COMPUTER CODES) Pkkerd. Lesse end Garekk,Inc. e O O
~
rem 6 TS SUBJECT TO CHANGE BEST ESTIMATE RES!JLTS MEAN CORE DAMAGE FREQUENCY NONSEISMIC EVENTS 8'**~' TOTAL - 7.ex19-snEan
< x ,g - 5 _
3 X 1g-5 _ 5 a
> 2 X 10 - 3 -
1.7 X 10 - 5 g 1.2 X 10 - 5 l h 1X19-5 _ e.s x 1e - s 1.9 X 19 - 8 g I I SLOCA LOOP SGTR FIRE OTIER INITIATING EVENT Pkkord. Lesse and Ostridt. One. O O O 1.
MD SU5KCT TO CHANGE l - BEST ESTIMATE RESULTS - PHASE Illa FRAGIUTIES MEAN CORE DAMAGE FREQUENCY ! SEISMIC EVENTS 8 X 10 ~ I -
, ., ,_4 5 X 10 _
4.7 X 10 - 5 .
- 4 x 10 - 5 _
u.
; > - 3.0 X 1g - 5
@ 3x10-5 2.0 X 10 - 5 2.1 X 10 - 5 8 2X10-5 w
E 1.t x 10 - s ! 3 X 10- 8 O I ' ' ' ' l 0.5 1.0 1.5 2.C 2.5 3.0 3.5 4.0 SPECTRAL ACCELERATON (g) i l Pkkord, Lowe and Garthk, fee. i e O O O
KEY PHASE Illa SEISMIC CONTRIBUTORS e OFFSITE POWER e BOP PIPING FRAGILITY TAILS l e MAIN CONTROL BOARDS e 4.16-kV POTENTIAL TRANSFORMERS l e 480V LOAD CENTER l e TURBINE BUILDING SHEAR WALL e 125V DC CHARGERS , o CONTROL ROOM VENTILATION CABINETS i I e CONTAINMENT FAN COOLERS , ! l e 4-kV SWITCHGEAR ra.,e.t- 4 m m.n j i !
! O O O .
l RELAY CHATTER ISSUES 1 CONCERN RESULT e CONTROL CIRCUIT LOCKUP - NO SNEAK CIRCtMTS. MANY LATCHING CIRCtRTS e LOAD SEOUENCNG - NO CONCERN (TEST RESULTS) a CONTACT ARCING DAMA - NOT EXPECTED e INST.tATION FALURE DUE TO ARCING - NOE EXPECTED OVEN VOLTA & e ItOUCED VOLTA 5 - NOE EXPECTED e SIMLA.TAEOUS CLOSURE OF - NOT POSSBLE. (MECHANICAL / REVERSING CONTRACTORS ELECTRICAL INTERLOCKS) e NOtmELAY CHATTER - CIRCUIT BREAMERS APO OMOADS WEL NOT CHATTER (EXCEPT FOR AUXILIARY CONTACTS). TEMPERATURE, PRESSURE, FLOW, AND CONTROL SWITCHES ASSUMED TO CHATTER. e PARflALLY OPEN V** vES AT
- ONLY IF CONTROL CIRCUIT LATCHES PRESSURE DOUNDARIES W W W
n 1 RELAY CHAI It-R t . o RESULTS
~
- AFFECTS ALL SYSTEMS INTERMITTENTLY l - EASILY RECOVERED
- CAUSES MISLEADING INDICATIONS AND ALARMS i
- NO PERMANENT EQUIPMENT DAMAGE
) - IMPORTANT CONTRIBUTORS o 4-kV BREAKER TRIP (PROTECTIVE RELAY SEAL-IN)~ o MOTOR-OPERATED VALVE TRANSFER o DIESEL GENERATOR CONTROL TRIP (LOCKOUT i RELAY)
; o PORV OPENING 1
l l , ) 4 l MAJOR TECHNICAL ACTIVITIES IN PHASE IIIB e UPDATE SEISMIC ANALYSIS e EXPAND AND REFINE HUMAN ACTION ANALYSES e RELAY CHATTER ANALYSIS e REFINE PRESSURIZED THERMAL SHOCK EFFECTS e QUANTIFICATION AND UNCERTAINTY ANALYSIS n.
o i - - - 4 PHASE IIIB OBJECTIVES t e REFINE THE ANALYSIS OF ISSUES HIGHLIGHTED IN PHASE MA i , i e DOCUMENT AND FINALIZE DCPRA REPORT 1 i i i l 1 j ra.,4. i. o u. h.c. O O O .
- , . , , -,,,n.,-, - - , , . . , - - , . , , . _ - . - . . - , - - - , , . . - - - - , . ,
's DIABLO CANYON PRA PHASE lilB ACTIVITIES ~
l l Pldient, Lease and Garridr, One. O O O
l . FULL SCOPE ASSESSMENT (PHASE Ill) , I WESTINGHOUSE PWR PLANTS l ._ PHASE MA 1 T PLANT A [ b "" Q] mc LOWER BOUND UPPER BOUND (STH PERCENTILE) MEDIAN MEAN (95TH PERCENTH.E)
, , ,t y PLANT D 5 I I I I 10-6 10-5 10-4 10-3 10-2 10-1 FREQUENCY OF SEVERE CORE DAMAGE / MELT (EVENTS / REACTOR YEAR) ea w.t - % i.. %
O O O
MAJOR TECHNICAL ACTIVITIES IN PHASE IllB (continued) l e ADDRESS RCP SEAL LOCA
- SEAL LEAK RATES
- CHARGING PUMP DEPENDENCE ON CCW
~
- OFFSITE POWER RECOVERY
- DfESEL FUEL OfL TRANSFER 1
n ,* . 4. t e o . e . n.e. O O O
o OVERVIEW OF PRA STATUS : PHASED PRA , PHASE I (PLANNING) COMPLETE PHASE II (SCOPING) COMPLETE PHASE IIIA (IST ITERATION) 99% COMPLETE PHASE 1118 (FINAL ITERATION) STARTED 1988 ACTIVITIES
- ANALYSIS OF DOMINANT CONTRIBUTORS FINAL COMPUTER RUNS .l FINAL REPORT WRITING .
O O O
DCPRA REVIEW PROCESS.
-TECHNICAL REVIEW BOARD (PICKARD, LOWE, AND GARRICK)
-INDEPENDENT PG&E REVIEW (OPERATIONS / ENGR /MAINT/ TRAINING)
-MANAGEMENT AND LICENSING REVIEW
-3 PRA HORKSHOPS FOR NRC STAFF AND CONSULTANTS
-SUBMITTAL OF 3 VOLUMES OF PRELIMINARY ANALYSIS FOR NRC REVIEW l
i i e G G
O O DIABLO CANYON PROBABILISTIC RISK ASSESSMENT PRA OVERVIEW PRA METIIODOLOGY PHASE IIIA MODEL PHASE IIIA RESULTS PIIASE IIIB ACTIVITIES O O O
. _ . , , _ . _ . _ _a .. .a_ e - m . ._ -.. m.- __ __ _.
O i 4 I i 54 H
% O, i ce O ;
n l (4 N i l 6 I l 6 l I i O! l . : i t 1 i l 1
REL.AY CHAI ItiR ANALYSIS TASK
'e PREPARE LIST OF KEY RELAYS BASED ON PLANT IMPACT' ,
e DEVELOP KEY RELAY DATA BASE ;
- TYPE /MODEL
- MANUFACTURER
- PANEL LOCATION
- ORIENTATION e SELECT OPT!ON FOR KEY RELAYS / PANELS
- OPERATOR RECOVERY / PROCEDURES
- DEVELOP NEW GENERIC FRAGILITY CURVE
- DEVELOP RELAY-SPECIFIC FRAGILITIES n.
, , t 9
t-
1 TASKS TO REFINE PRESSURIZED THERMAL SHOCK ANALYSIS e SUMMARIZE PHASE IIIA FINDINGS e DETERMINE CONSERVATISMS IN APPLYING WCAP-10319 RESULTS l e REVISE VESSEL RUPTURE PROBABILITIES ACCORDINGLY e MODIFY PLANT MODEL TO ACCOMMODATE REVISED APPROACH ru.,4. t ,4 ma. w. e O O
) . QUANTIFICATION AND UNCERTAINTY ANALYSIS e QUANTIFY ALL FOUR STAGES OF THE PLANT MODEL
- NONSEISMIC INITIATORS
- SEISMIC INITIATORS e APPLY RECOVERY FACTORS TO KEY SEQUENCES e IDENTIFY AND DOCUMENT RESULTING KEY SEQUENCES e PROPAGATE UNCERTAINTIES FOR EACH PLANT DAMAGE STATE
- NONSEISMIC INITIATORS
- SEISMIC INITIATORS e DEVELOP PROBABILITY OF FREQUENCY CURVES run.4. t w = m. h=.
g _ .._ . _ . ._. _ . . _ _ _ _ _ _ _ _ . . . , _ _ _ _ . _ . . _ . _ . . . _ _ . , . . . . . _ . - . . _ . _ _ _ . - . . _ - . . _ . _ . . _ . . __ _ _ _._ ____.__ _ ______ __ _ __ _-. _ _ _ _ _ _ _ _ l
I i TASKS TO UPDATE SEISMIC ANALYSIS e REQUIRED INPUT HAZARD CtIRVES FRAGILITY ANALYSIS e COMBINE SEISMIC PROBABILITY DISTRIBLITIONS nenne,t.se W k he.
DIESEL FUEL OIL TRANSFER TASK e IDENTIFY SYSTEM FAILURE CONTRIBUTOR 3
- NONSEISMIC
- SEISMIC
- PERFORM SENSITIVITY ANALYSIS (e.g., TREATMENT OF MULTIPLE STARTS) e INVESTIGATIVE TASKS
- PUMP DESIGN DETAILS
- OPERATING EXPERIENCE
- RECOVERY OPTIONS F' '
i I i . I ! OFFSITE POWER RECOVERY TASK i e COLLECT PLANT-SPECIFIC LINE OUTAGE DURATIONS . i l e INVESTIGATE 125V DC STATION BATTERY CAPACITY l l i e LOAD SHEDDING STRATEGY / PROCEDURES , o EIECOVERY OPTIONS FOLLOWING AN EARTHQUAKE e REVIEW SEISMIC FRAGILITY FOR OFFSITE POWER L l r *.4. t n..km. i e. . I _
i CHARGING PUMP DEPENDENCE ! ON CCW TASK i l l e IDENTIFY CCW AND AUXILIARY SALTWATER SYSTEM l FAILURE CONTRIBUTORS -
- NONSEISMIC l
j - SEISMIC - l e DEFINE CHARGING PUMP OPERATING CONDITIONS i WITHOUT CCW i e INVESTIGATIVE TASKS l - CHARGING PUMP HEATUP RATE ! - ALTERNATE COOLING OPTIONS (INCLUDING PUMP ROTATION) -( - VALIDITY OF SYSTEM FAILURE MODES I
- OPERATOR'S RESPONSE / APPLICABILITY OF PROCEDURES J
Pkkad, t. owe amt G=rkk. Bew.
1 l i
! RCP SEAL LOCA LEAK RATE TASK l
i e IDENTIFY KEY SCENARIOS FROM PHASE lilA i , i e REVIEW WESTINGHOUSE /NUREG-1150 SEAL LEAK RATE l MODELS ? j e COMPUTE CORE UNCOVERY TIMES USING MAAP I I I i ! m t w a u. w. i a . % 1
RCP SEAL LOCA THEISSUE e SEAL COOLING AND CHARGING SYSTEM DEPENDENCE ON CCW e SEAL COOLING AND CHARGING SYSTEM DEPENDENCE ON VITAL AC e SEQUENCE DEPENDENT TIME AVAILABLE FOR RECOVERY ru a.t 4a.,ia. w l
. - ~
! DCPRA - APPROACH l . ! o UTILIZATION OF PICKARD, LOWE AND GARRICK (PLG) ! AS A LEAD PRA CONSULTANT OPERFORMING A PLANT AND SITE SPECIFIC PRA THAT WILL j RELY HEAVILY ON PG&E KNOWLEDGE AND EXPERIENCE ] l 4 oEXTENS1VE PG&E INVOLVEMENT MANAGEMENT OVERSIGHT OF PLG PERFORMING TECHNICAL PRA WORK UNDER PLG DIRECTION REVIEW OF PRA MODELS AND RESULTS i o EXTENSIVE TRAINING OF PG&E PERSONNEL ~ ! IN PRA TECHNIQUES I o TRANSFER OF PRA METHODOLOGY FROM PLG TO PG&E' ,
l HUMAN ACTION ANALYSIS TASK I ! e OPERATIONS REVIEW OF PHASE lilA FORMS { l e REEVALUATE PHASE lilA ACTIONS i e EVALUATE PHASE lilB ACTIONS AFTER IDENTIFYING KEY SCENARIOS )
- - FUEL OIL TRANSFER RECOVERY OPTIONS
! - RELAY CHATTER RECOVERY l - COMPONENT COOLING WATER / CHARGING PUMP RECOVERY I
! e INVESTIGATIVE TASKS
- HEAT LOADS OF EQUIPMENT SERVICED BY 480V
! SWITCHGEAR VENTILATION
- EQUIPMENT TEMPERATURE LIMITS Phhad, Leum med hht, hue.
1' )
-,__m _ _ _ - - . _ -. . . . , . - . . , , , _ . - - - - , - , _
1 l . 1 i l FEATURES OF DIABLO CANYON ) ) e THREE CCW PUMPS AND AUXILIARY SALTWATER l CROSSTIE j e TWO AUXILIARY SALTWATER TRAINS i e THREE VITAL POWER BUSES - l e SWING DIESEL GENERATOR i l i e TWO-TRAIN FUEL OIL SYSTEM I l e MILD CLIMATE (MINIMAL) AIR CONDITIONING i REQUIREMENTS ! i l e NSSS PUMPS AND 4-kV SWITCHGEAR NOT SUSCEPTIBLE l TO ROOM HEATUP ! l e PROTECTIVE RELAYS NOT SUSCEPTIBLE TO ROOM l HEATUP l ru d.t and n=u. h . j s _.
.___-- _ . _ _ - _ _ . . _ _ . . _= _ - -_ .
l . 4 i ! FEATURES OF DIABLO CANYON (continued) 1
'1 i
) ~e 480V SWITCHGEAR/ INVERTER ROOM HEATUP 1 ) e RELAY CHATTER EASILY RECOVERED, NO "LOCKUP" AND l; NO RESTART RESTRICTIONS i j e RELIABLE PLANT WITH FEWER STARTUP PROBLEMS THAN ! MOST i 1 j A sur coarAsaro i e DIESEL GENERATOR HAS A4R40dB HEAT EXCHANGER FOR COOLING , i e ISOLATED SITE ' I e SEISMIC DESIGN l e LARGE BATTERY CAPACITY - i ' i ra .a.t 4o..ia.i e. l l
I 1 FEATURES OF DCPRA i j l e RELAY CHATTER AND RECOVERY i e ENHANCED SEISMIC ANALYSIS
- e DYNAMIC HUMAN ACTIONS ANALYSIS -
e EXPANDED SPATIAL INTERACTIONS ANALYSIS I ) e DESIGN AND CONSTRUCTION ERRORS i I ! e "STAND ALONE" DOCUMENTATION i e QA PROGRAM e DOCUMENT CONTROL (REVISION, REFERENCE) ekt.d. t- w o ke. w.
CD O A O A O > Z EH N Z C4
)
QUANTITATIVE DEFINITION OF RISK. I ANSWERS TO (1) WilAT CAN GO WHONG? (2) WIIATISTilE LIKELillOOD? (3) WIIAT IS Tile DAMAGE? . SCENARIO LIKELlHOOD DAMAGE - l l *1 II 81 j 8 2 f2 82
- s3 I3 x3.
I : sk /N 8 N , 3 R in RISK = <sg. f g. ng > a r
~
- Pkkord, Lowe end Gerrkk,Inc.
e . 4 i, . STRUCTURING SCENARIOS ~ I ru r t.V.i. ;>;gna ., ., ... P u l.T , m . CO.,rA_l m. e
...q .
- - v
- . ... i
. I j A h l SYWTsas Lawst Is b I T T ~
f (LOGIC] I (LOGIC) - z I gggg eoe eoe LEVEL _ _ _
- _Cgg Cg, Cyg C.2in Cl2 C22 j
cAust LEVEL List or *Ig CAUSES LISTOF 5 CAUSES 2 i . j List oP CAUSt1 COMMON _TOIg A;40 I3 $$gAND$t'El e 1 DATA LEVEL EXPERIENCE WITH FREQUENCY OF CAUSEs STEntoft { CENERIC INFO E'
- pggg y
j POPULATION DATA: E 2i D PLANT SPECIFIC DATA EyJI ! 4 i i l Pickeni, Lowe and Garr* _ loc. q _ l .
. /
I I E SEC 47 Y HEAT LONG-TERM RCS II" I
- SsNet 84 f EcosTT AVAILAOLE STAstt go g., SE AL LOCAL-I 80 0 l SECONDA4Y AT SeMet LOCA a <r Y 'I Wel
, CHAfecesco 88E E D RESTOet SECOSIS&WT SAFEff seeKCTeost F04 SUesP SWIP DECestCULAftger DeEAT SSNL . puesp ge8JECTIOsg ECl#CULAY fMP5T RSAREUP STA8EE l
fee SECOMOAAT DE OU A1 - SEEAT Sesset C L I CORE ' ff 1 P y DAasAOE eq.Ese Ases ACS DE PSE SSUtt!A78008, PEES e m " num putCieon.A=O ofeett - l RECIACULATIOce l a 1 . vue 9 r \ . Costfacesasterf C0001Atesestati systfassieSOt.Afeoss ) GoAmActj S 0,TES.it.es A LEs PRl!LIMINARY Ri!SULTS ' SUBJIXT TO CilANGli 1 r l I s COstTAessestest 1 - 4 SYSTEMS F Act E3 I J FIGURE C.4-1. PHASE III GL .RAL TRANSIENT ESD OVERVIEW -
)
q - EVENT TREE WITH BOUNDARY CONDITIONS l l e MODELS EASIER FOR NUCLEAR PLANT OPERATORS AND
- ENGINEERS TO UNDERSTAND .
o DEVELOPMENT FOLLOWS FROM EVENT SEQUENCE ' l DIAGRAMS l e EVENT SEQUENCE DIAGRAM DOCUMENTS EVENT TREE . I e FOCUSES ATTENTION ON KEY DEPENDENCIES l e APPROXIMATIONS TO SIMPLIFY MODEL EXPLICITLY ! CONTROLLED BY ANALYST - l l e QUANTIFICATION REQUIRES CONDITIONAL SPLIT ' FRACTIONS e ENHANCES VALIDITY OF PHASE il STUDY ) 1
- Pkkard, Lowe and Garrio., .nc.
1 M MATRIX FOR MODULAR EVENT TREES - I 1 SUPPORT EARLY PLANT INITIATING DAMAGE EVENTS SYSTEM RESPONSE FRONTLINE STATES FRONTLINE ^ STATES SUPPORT SYSTEMS SYSTETAS SYSTEM 1 EARLY t LONG-TER M y ' 2 RESPONSE MODEL 2 ' RESPONSE 2 2
~
MODEL
- MODEL Y .
A = [sp ] *
- J -
L u (I kpf I L I E=[ejlk] K l s
\
m 11
=I a ji Ie jlk 1 k]Il y k 1
j 1 Pickmd, Lewe and Gwrick, Inc. i w I -
( . i
,;;~ . . . .
='N .
l t _ l N .J. .. ' l lA&Ca Simplified Plant Event Tree Diagram ' ((S)=((1)f(All)f(EllA)f(Ci!Aijf(U!ATC) where ,
, ((S) = the frequency of scenario 5
((1) = the frequency of initiating event 1 1 f(All) = the fraction of times system A succeeds given that ! has happened l f(TilA) = the fraction of times system 8 fails given that I has , happened and A has succeeded ! f(Ci!E) = the fracti6n of times C succeeds given that I has l happened, that A has succeeded, and 8 has failed ! f(U11 ATC) = the fraction of times D fatis given I, A,I, and C 5 ((S)*((l)f u infg *** ;
'%=fu I
3 f a g*** 'nt. *** * '(3) i (
' Pickard, Low and GarHek, Inc. !
i 5
. 1 i
A _ a. m . ma.- s - 4 m _ _ - . .w _ ._ ____s _ _
- _ _ . . _ -
( v$g
-C Dl1, e
O Ell .
}
l O f. l- I Ct - A,f - , l E, 5 "- - V gsm s-E-
~W T / :
5 hm
- n. af!
een y "- w ' g
...g y
g y I O 8,8 \ l ... s e i 1 5 E[E$ a E , 4 o Ig Ii l J... #
~. l -
"ytr...
5 , IiI. M i i [ \ . 4 i
- _ _ _ . - . . - _ _ . . . _ . _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ = _ _ _ _ _ . -_ - . - - - - _ - -
PHASE II i .
, FRONTLINE EVENT TREE MODEL LINKING ~
1 Initiating Events Sy Early hontHu long 4 m Fru tH w Plant Damap
. States Event Trees Event Trees States l 1 ; General Transient : '
i Less of Offsite Power - rSuccess LTIA (w/5r. cooling) : Soccess
' Loss of States : General i
Component Cooling Water LTIIF(Ereed anOced cooling)** LT1 l< 1 - 48 Transient LT2A IllP oTC's success) Tree - Core Daeage
~
(or Austlf ary Salt Water)-* Tree LTfiF(lip and C FrTTIT 5 - A.D.E,II,J ; Loss of 4#0V Switchgear IT/C(iiiFrifls) - Ventilation ~
. 21fTff/Y1Tifli) ~
. : I.T- V//V1 and LIIFor t.5J rail) .
LT2F~{V'8VI anarrTll} 5elsmic fuents +
=
?
1 Small , : Core camage Small LCCA : : LOCA . LT2 1 - A,0,E.II.J Tree
- _0 Tree 2 - A.U E,li J
_ 3 - A,u,E.If,J 4 4 - A,0,E.11.J
, PRIiLIMINARY RESULTS ,
SUBJECT TO CilANGE / I
'5fallar to general transient mapping. - -
Plckard, Towe and f;nrrick, Inc. W
SYSTEMS ANALYSIS PROCEDURE e . QUALITATIVE SYSTEMS ANALYSIS j - SYSTEM FAMILIARIZATION AND DOCUMENTATION 4
- PROBLEM DEFINITION AND COMPONENT /CAUSE SCREENING i l
e LOGIC MODEL DEVELOPMENT ' RELIABILITY BLOCK DIAGRAMS FAULT TREES i: COMPONENT TABLES - j , e ALGEBRAIC MODEL DEVELOPMENT SPLIT FRACTION EQUATIONS , e QUANTIFICATION AND RESULTS EVALUATION SPLIT FRACTION QUANTIFICATION AND CAUSE TABLE
. s RESULTS INTERPRETATION AND COMPARISON 1
IDENTIFICATION OF SYSTEM IMPROVEMENT OPTIONS l ' e SYSTEMS ANALYSIS REPORT l Pickard. Iowe and Garrick, Inc. t . i
. - - _ - . _ _ . . _ _ . _ . , . . , _ . - - - - _ _ _ ______m_,____ _ _ _,__.,,,,,.__.._,_m___,_,-, , _ ,, . m., .. -r__ , , . . . , , - . .-- -, ,, , . ,.m
. -~
l QUALITATIVE SYSTEMS ANALYSIS OUTLINE OF SYSTEM SUtt FILE: SYSTEM FUNCTION FSAR SUCCESS CRITERIA , SUPPORT SYSTEMS SYSTDSS SUPPORTED " ~ SYSTEM OPERATION AND SPECIAL FEATURES * ! e SYSTEM CONFIGURATION e CONTROLS, INDICATORS, AND ALARMS e NORMAL OPERATION - CONTROL ROOM l e TRANSIENT OPERATION - LOCAL PANELS AUTOMATIC ACTIONS e SURVEILLANCElTEST OPERATION MANUAL OPERATOR ACTIONS e MAINTENANCE OPERATION POTENTIAL 'OR EVENT INITIATION TECNNICAL SPECIFICATIONS REQUIREMENTS . i e LCO e SURVEILLANCE REQUIREMENTS
. REFERENCES o FSAR SECTION e TECifMICAL SPECIFICATION SECTION i
e OPERATING PROCEDURES e SURVEILLANCE / TEST PROCEDURES i e OTNER PROCEDURES e OTHER DRAWINGS l e P AND ID e OTHER DOCUMENTS l , a . Plckard, Lowe and Garrick. Inc. ~
*s DCf"# 4 tLIC1pIC Pous e wif AL AC#DC tmlY 2 m
9C SUCCESS CRITERIA e DF SC e Br e SF SC 39 4 MV hws F twnst 29. 400 V bus 2F and I2S V DC remet 21 9
^
SF SC most remata awallable for 24 hovre. S SF SC 9 DF SC DOUNDARY C(DeDITIONS ,' S SF SC 19 Offette Celd failed.
- DF SC
- SC
- SF
- DF SF SPt.IT FRACTION ids S SF SF S
9F SF SFS DC-F , e SF DF 9 SF AS ASSUMPTIONS 9
- BF AS
- 3) There are se tests which will centribute sipalficently 4 I DF AS to the waavailability et watt 2 w8tal 4 RV leus F. 400 V S BF AS 9 SF AS bus 2F er DC b u s 2 8. ,
e _ 8 Dr AS . pr 23 pse credit is etwen for redendent battery chargers ephess 9 AS considering hardware f a6 5 wees. However, tred8% is glwen 9 , SF AS when constdering tests and maintenances during charger 9 SF A9 SF AS performance tests (STP M-82WP and charger matatenente. $t 9 as essweed that the redundant c harger is altened se that S , SF AS there se etwags a charger awa 4 3 abl e. t DF AS 9 SF AS s SF AS 3) adhen tensideriisg meintenance et the 480 V breaters , (S?-2F-42. 52-2C-42. 3?+.H-429 which feed the chargers S WF AS and the 223 V DC evtput breaters (72-2100. 72-2201. 9 Dr AS DF AS 72-23089 feem the c hargers 4* to answeed that the redondento , SF AS chargers (225 and 2389 weald be aligned se that e therger 9e Dr AS se etways avelRable en each DC bus. S
- SF AS e OF AS 49 For DC trains 28 and 22. natatemente of the panel feeder breaters 72-2102 and 72 . W is censidered part of the S Dr AS 9 Dr AS malsetenante of the penet itself. ,
S SF AS s Dr AS S) Malsitenance of the lead center transformer 4 MV feeder S DF AS breeter es essweed to be inc lud ed in the meintenance e I DF AS of the lead tenter transfereer, 4 DF AS e SF AS 69 If en ewel38 erg feeder breatee is in maintenance. 44 is ^ assweed that the asseelated hws would be realigned to 9 I 9F AS S AS the standby startup power swppig. Therefore. st is DF DF AS reqwsred that the startup feeder breater evet open en 9 , Dr As demands thws, fa,Avre to apen en demand remanns a ve3td
- t pr AG fastwee mode even when the aus breater is en maintenance. .
FICURE D 2.1.4-9. EP2AO CRT . (Sheet 1 27 w J _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ . . ~ . - - - -
O _ O Dr As e 9 3r A9 79 Unit 2 AC/DC train H to slightly different free trates F
- and C. Tratn H has one less breater het one addittenal 9 97 AS Dr A9 #wse This dafference as not modeled and does awt atteet e DF As the reswits s i nc e the two fattere f regwenc ies are er the e same order. S Dr AS AS 9 Dr 9
O pr AS 3) There is na malatenance performed om the batteriee ehtsh Br AS esi l l na t e then weawailable e 9 SF A9 p 90 SC SUCCCSS CRIT [RIA e 9 90 SC 90 SC 39 4 MV hws C f un 8 % 2 9. 490 V tus 20 and 823 V DC panel 22 e
- r evet remain awallable For 24 howe s . s .
90 SC DQ SC S 50 BC 3DUNDARY C"JNOI T 3 DNS , 9 DJ SC BC DC 23 Offette Crad failed. AC/DC treln F wait 2 sectessfel. t S 9C BC - S- as i es e,t AC/DC traen r , sited. . O Ba DC 2, 9 90 DC SPt.If FRACTIONS 3Ds 9 30 De - O wa Da e 90 Sr 301 OO-r. Dr-S SC2 D0-F. 90 4 9 SS SF -
'O u Do .
30 AS ASSUMPTIONS
- AS S 99 "
~I Be AJ Saee as t er Dr. e AS S 30 SH SC SUCCCS3 CRITCNIA e ~
r SH sC e SC 19 4 MV tus H twnit 29. 490 V two 2H and $23 V DC pene 23 $ SH nest remata seallable for 24 h owe s . 9
- DH f. C S
'I DH SC S
SH SC NJtY CDMDITIDMS
- DH DC *
.I SC 19 DFrette Crid fatted. AC/DC trains F and 3 wait 2 secteed.
- DH DH SC SC 2B Base as $ estopt AC/DC train O or F Failed. e SH
- s I M 3C o
DH DC 39 Ei,=e as t escept AC/DC tratas r and a passed. DH DC - seuT rDACf roH ses . ,O DH SH e 94 DH 9648 00 4. Br-S. 90-S e 94 Sr . O DH Sr mz oC.r. Br-s. Sc.r er oc-r. DF-r. Sc-S s pH Sr SH3 OC #. 90-F. DC-r s
- SH DH S WH AS ASSUMPTf0NS S
SH AS e DH AS Same as ter Cr. e g DH AS ^ f " FICURE 3.2.2.4 e (Sheet 2 et 2)
0 d 4 t O
%W J r
.$3 sE. _
in G
) LL G
as $ % e4 .n j y w ga6
- r75 r
x _ as Wv 67 h % nt a $b t
@O4 '
1 sEs Ed p d
"s 3 y $ \' ed _
&w
( .) e
,G w ~f n e >s .
ca i g WD ** Od p 3 h 3
>p u<
a do e c tu 1 59 E t. sv E r a E . de< ! di D TE es . . e 1 a, I
- I l O 16 o
- N \E h
, < ,i
=+
*e
*L 1 ABLE D 2. 3. 4-3. EP2AD TABLE %
O LIST Dr COMPON[MTS IN BLOCr DI ACR Art Sheet I er 4 S FleeC? tO%. AND IMITIAL LO.43 OF ACTUATED COMPONCNT #Dr4ER HAJOR CIDEPORENTS E NVIRibdW NTAL STATE SLOCR !,ts' PORT SusitteS STATE STATE MO ENAME AND ID NO 9 FAILURE NODE %
^
Top Event BF . , - Energized Ene*~3 tees De-energises Dietribetten Panel 21 Fall during operetten. 1 ~ ITSSSR Closed ee Is Transfer open. 2TC827 -- Closed C5 72-2102 Energised Energised De-energtred 123 V DC See 28 Feel during operation. -- ITBSIR t Chorged Cheeged As-le 2 123 V Setteeg 21 Dotywt Fallere en , demand. FIDAT9 er , I during operetten. 2TBA1R
- Fwee Fett during operat.en.
- 21FutR Enerstred Energised De-energized Satterg Charger 21 Fall during operation. 400 V Sws 2F ,
3 11DCHR Closed As Is Transfer open. 210937 Closed CS S2-2F-42 Closed Closed As is , CD 72-2800 Transfer opem 2fC32T -- 400 V Bus 2H Energlaed De-energised De-energlsed patteeg Charger 221 Fall dweing operetten. C , 2TBCoet Closed Closed Transfer open. ETCSIT Closed CS 32-2H M Closed Open As is C3 72-2303 Fall to close en , demand. ZTC32C or transfer open during operetten. 210827 - Energised Energised De-energired 4mV see Section F Fall dering operetten. 9 ITSSSR Open Cleoed Ao lo Fall to open en 123 V DC 28 I CB 32-NF-13 demand. ZTCSIO Closed Closed As is Transfer open. ITC5tf 123 V DC 21 , C3 2 -tF RO Energised Emet g i s ed De-energised 4mv Vital Bos F O eDO V L. C. Transporeer 2F Fall dweing operation. , 11ER2R Energised Energtred De-ener g 6 s ee 400 V Sus 2F Feel dwr6ng opevation. l, litStR Open Open , r Fort Damper FD-3 Inadvertent transfer ---- closed. F T DF R S Open Open -- F s r e Damp er F D-9 8 adweetant transfer -- closed. ZTDE RI
-e
__ . - . . _ _ _ _ _ _ _ . _ ..._.__m _ . . . _ . _ _ _ _ . . _ _ . _ _m _ _ . _ - . ___ _ _ . _ . _ _ . _ _ _ _ - ._ _. . . . - _ _ . _ . . _
=
"W HD = FD s S CD = FS e to . Dwn3*NDRCCV e Single train dependent bordware patteres.
DMS = DCPA
- CB72OJ + DCDUS
- DIsch I f at t wees.
D'2 - SAYT + FLwi 9 98ect 2 f a t t eres. DM3 = Ch3 Cit
- C83242 + C372OO
- Esect 3 f e t t wees.
SM4 = 1O e s a ec t 4 re t twees,
- slett 3 tadependent pelleres. S tams = pus 44V + S3C380
- CDlOT DBMS = 2e8'3ClitO + T3CSIO
- Diest 3 dependent f e t t eres (FSeFH)+FSDL Due = LCan~
- SUS 490 e 2eFLFDT
- Block & *stleres.
- 823 V DC distributten penet 21.22.23 Falls. S DC Ptd. - 2 T as s m e T CSF200= 2TC32Tef
- 223 V DC C.5. 72-2002 2202.904 transfers open.
DCSUS = IT3Simei
- 123 V DC two 21.22.23 tetIs dwetag operetten. '
SATT = 275AfD + ITSATRef
- 123 V DC betterg 25.22.23 deeendteperetten Failure. -
F LFC = Ifruimei S 123 V DC twee falle during operetten. ComCR = 2TSCHRet
- Detterg charger 28.22.23 retto dwelag operetten. ,
C03242= 2YC5 81eT
- eDO V AC C.3. 32-2F-42. 2G-42. 34-42 % r en s 7 er o er en. -
C5?200= ITCD2Tei e 323 V DC C.B. 72-2800.2208.2300 transfers open. DU54mv= IT353Ref 9 4 MV wttet ius falls dettag operetten. ,
- C 3. 32 W. HC. te4-3 0 trans f ers op en.
CSIOT = 2TCetter LCan = 2 Tan 2nef
- Lead center transf ormer fello dwe tag operetten.
91'5480= ITBSiner s 400 V wital two fella during operetten. , FLtDT = I T DFR Is f e Femsble link fire deeper imedvertent trans fer .4]esed. ref = MF G NO = PF 4 rF = MDMI + McM2
- MGM3 + MDM4
- M9MS
- M9Ma e Malatenance en single train geopenents.
s Metatenance en blott 3 components. MOME = MDCP94.
- MDCDUS MPN2 = 0. O e Meantenance en blott 2 components. ,
MteM3 =00
- Maentenance en blott 3 soarements.
9 Me 6 n t enanc e en b lec t 4 c omp onent s. t99M4 =00
# Meentenente en 6toch 3 cempenents.
Men 3 - MSS 4MV ,
$ Maintenanc e se blet t L components, Meme = Mt.C a R o M95490 --
etDCPet.= 2M95tre2MCNSD 9 823 V DC distrftetten penet metatenance. NDCDUS= IMSSIFe2McNSD S 323 V DC hws metatweence. , MOS4MVe 2MSSlFerMCNSO 9 4 MV bem maintenance. NLCam = 2Merpre2McNSD 9 Lead Center transformer metatemence. MD54GO. 2Mestre2McNSD 9 400 V hws maintenance. ,
- Prebebsittg et non-recewers free commen eewee te t t ure.
HORCCV= 2HCACS T = 24.0 9 M8ssten time. .L FIOURE D.2.1.4-4 (Steet 2 et 27 O O
- eugp#
esp
l e i W4 Mad *3 Past **ST *Pepid
.8e.8 e8e8 4. o 8 e8=. e4 4 8..es == e.a s *8*=8.es 4. ...-== ese...e a8 name 8 8 8 8 e 8 e 8 8 8 j 8 secas 88C8 8 86c8 8 88c8 8 Scc 8 8 88c8 8 8 88C8 Soc 8 8 e es**a sess 8
, 8 Sec8 8 88c8 8 secle 88c8 8 sccas 88c8 e 888c8 deca 8 y =8 44 e m 8 ! 0 8 8 8 8 8 8 8 0 8 o 8 8 8 8 8 8 8 8 8 8 l 8 e e 8 8 8 8 s 8 esca 8 ! B 0 8 8 8 8 8 8 I Ag gammeg3 goa==.Seeg 8 l I 8 8 I I 8 8 8 8 SeC8 8 8 StC8 I *KS 8K8 8 8 8 8 8 SEC8 8 8 383 geomeg3 geomm.Seet 8 I 8 8 8 0 8 8 8 8 8 eeC8 8 8 84C88 8EC8 GKS 8 8 8 8 8 8 8 38 88emog3 te*==. Seat 8 8 8 6 8 8 8 8 8 8 sees 8 , 8 8 8K8 GKS 8 8 8 8 8 IEC8 e s 3 Beemog3 te*=maleet I t 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 I Sie 8 1 0 0 8 8 8 8 8 8 8 38*d **S84***e 88e88e 8 8 e 8 8 8 8 8 0 8 8 8- 8 8 8 8 8 8 8 8 8 8 5 8 I S., 5 8 I I 8 8 4428 8 E 8 895888 **S84*=*s Ses*Be 8 e I 8 E 8 I 8 3 8E I 8 49E88 8 8958E8 **8*****e 88 ease 8 8 8 8 I I E 8 5 8 1 0 8438 8E 8 8988CB *e48**=ee 88**6e 8 8 8 8 8 8 8 8 0 8 8 8 8 8 8 8 8 8 8 8 8 8 8 5 8 4-I $ 8 8 0 E 8 c8 **e 3e Atr8 8 8 8 I 8 8 8 8s I e 8 I 8 58 **e 38 A628 8 0 8 8 8 E 8 8 5 8 8 8 8 II **e 3s m6al 8 8 8 8 8 8 e 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 5 0 8 E l 8 8 8 8428 e 8 Se8 ese mese I 8 5 8 8 8 8 3 8 86c8 e Segas 8 38 ese mee,8 8 8 8 - I I 8 x 8 86c8 8 44z8 e s 48 *ae mese 8 I I e 8 0 8 8 8 8 8 8 I
~
O 8 8 8 8 8 8 8 8 8 1 I 8 I I E I I 8 8 4 4 5 8 888 **e A8 9 0 '& 8 8 5 8 8 8 8 4 E 8 3 5 8 8 8 es8 **e A8 98 6 8 8 8 8 E 8 8 8 5 8 5 8 8 8 8 m8 **e at 98~e e 8 8 8 8 8 8 8 8 0 0 8 8 8 ~8 8 8 8 4 4 8 8 8 8 8 8 8 8 8 8 438 9846 etteseg 8 8 8 8 8 8 e _4 8 8 8 0~ $ 4 8 8 8 8 8 8 8 8 e w I E8 88 I e w 8 El 88 8 e w 8 C8 28 88 8 gens agg 8 CI g8 33 8 g I sealen 8 eemed 8 sengen 8 semed a sentea 8 semed 8 ea31 e semed I g g e 8 8 8 8 8 e 8 g 8 I'NB e 8 8 8 pattes 8 8 98888 8#*f888g 8 mo88 W eg &geseg g metene Se88 eg3 8 asteapees See880ea , g meteos g i 8 8 8 s B 84eeeme 8 i 8 l 8 8 8 888 ***3884 eeless g eed g ge egnes,3 g ,
' ,n u . ,eog. ' 8 XIH1VW ADN3ON3d30 t
suasu u - annm -
_._, . . _ . _ _ _ _ _ _ _ . . _ _ _ _ . _ _ _ _ _ _____. _ __m. __ . _ _ _ _ . ._. _. .___m . _ . . . _ _ _ _ _ _ . _ _ _ _ _ _ . _ _ _ . . _ _ _ . _ _ _ _ . _ _ _ I i j 4 i SEISMIC ANALYSIS APPROACH l l d 1A '
-W
, .A .
_ . - l e m e.ee : 2:aw SA * '* '
* * * , * .,t l ete I e ie I e 1 e I ese_ ^^_ r_se v
=- ,
- e 2 2 . . .e s j - e -
[ L_
,_ / :/ 3
~
/ ,
u .; ,, __
. - .. , T..!.
3 *
., . ~
a 1 ,. . * -
*s 4
Y Y $ MSW As#ALYS46.F AGGmECAf sOes peesacsepos tem entse pag REY i STIntfCTtfat% l COesPOesteet3 1 j ,$ i. i e
! MAIAS.
Assattseg 1 me t e c *m.nie. I
~
j J ' j I
l l ! SPATIAL INTERACT!ON ANALYSIS l l e ESTABLISH LOCATION OF PRA COMPONENTS IMPORTANT j TO RISK i e ESTABLISH HAZARD SOURCES e IDENTIFY ENVIRONMENTAL HAZARD SCENARIOS l e ESTABLISH PLANT IMPACT FOR ENVIRONMENTAL HAZARD SCENARIOS l 4 l e RANK ENVIRONMENTAL HAZARD SCENARIOS BASED ON L CONSERVATIVE FREQUENCY ESTIMATES I I i l Pickord,l. owe end Gerekk,Inc. s -
DIABLO CANYON PRA PHASE lilA DISCUSSION L Pkkord, lesse and Garrkk,Inc.
~
s
1 i PRA PROCESS i ) i ! e SELECT CONSERVATIVE ASSUMPTIONS
- MODELS
- - DATA e BUILD AND QUANTIFY PRA MODEL
- e BUBBLE UP S!GNiFICANT CONTRIBUTORS I
e EVALUATE ALTERNATIVES l - NO CHANGES l - MODEl.ING IMPROVEMENTS i
- ENGINEERING ANALYSES
- - PROCEDURE CHANGES l
l - HARDWARE CHANGES Pkkord, Lowe and Gervick. Inc. _ _ . _ _ _ _ _ _ . _ . _ _ _ _ _ _ . _ _ _ _ _ _ , .. [
PRELIMINARY RE"'.TS SUQJECT TO CHA..GE SYSTEMS ANALYSES SUPPORT SYSTEMS FRONTLINE SYSTEMS AUXILIARY SALTWATER FAN COOLERS ! ELECTRIC POWER CONTAINMENT ISOLATION i CCW CHARGING / SAFETY INJECTION i VENTILATION AFWS SSPS/ REACTOR TRIP CONTAINMENT BUILDING SPRAY l PRIMARY RELIEF STEAM ISOLATION ) SGTR i RHR/LPl/ RECIRCULATION l V-SEQUENCE TOP EVENTS l SYSTEM UNAVAILABILITY CAUSES HARDWARE . HUMAN ERRORS MAINTENANCE COMMON CAUSE j TESTING l Pidierd, Losse oral Gar Ich. Inc. J l
"s" E E [ T o c $ a 7 1 INITIATING EVENTS INTERNAL EXTERNAL EXCESSIVE LOCA INADVERTENT SAFETY SEISMIC INJECTION FIRE j
LARGE LOCA GENERAL TRANSIENTS (7) g
" A AIRCRAFT ACCIDENTS SMALL LOCA LOSS OF ONE DC BUS TOXIC AND EXPLOSIVES INTERFACING SYSTEMS LOSS OF ASW TRANSPORTATION LOCA (ROAD AND WATER)
SGTR LOSS OF CCW TURBINE MISSILES REACTOR TRIP LOSS OF INSTRUMENT EXTERNAL FLOODING , AIR MILITARY AND SPACE MISSILES i TURBINE TRIP LOSS OF 480V SWITCH- WIND AND TORNADOES GEAR VENTILATION l. EXTERNAL FIRES LOSS OF CONDENSER LOSS OF CONTROL ROOM VACUUM VENTILATION i I CLOSURE OF ALL MISVs ] i STEAM LINE BREAK IN/
! OUT CONTAINMENT , ! rick d. t== and Gaevich. sac.
i J 1
i 1 I i' DATA ANALYSIS i l ! e INITIATING EVENT FREQUENCIES ! e EQUlPMENT FAILURES j . l j e COMMON CAUSE FAILURES e MAINTENANCE
- - FREQUENCY AND DURATION
- SURVEILLANCE TESTING l
- e BAYESIAN UPDATING I
, Pktierd. Lowe end Omelch, lac. a w
l 1 RELAY CHAI It-R .l e APPROACH l ! - IDENTIFY COMPONENTS IMPORTANT TO RISK
- ASSUME ALL CONTACTS CHATTER
- DETERMINE WORST CASE CONSEQUENCES l - DETERMINE RECOVERY MEASURES j eSCOPE -
j THE FOLLOWING SYSTEMS WERE ANALYZED FOR THE EFFECTS OF RELAY CHATTER: l
- AUXILIARY FEEDWATER - INSTRUMENT AC (INVERTERS)
- AUXILIARY SALTWATER - MAKEUP WATER TRANSFER
- COMPONENT COOLING WATER - REACTOR COOLANT i
- CONDENSATE - REACTOR TRIP l - CONTAINMENT ISOLATION - RESIDUAL HEAT REMOVAL .
- CONTAINMENT SPRAY - SAFETY INJECTION / CHARGING
- CONTROL ROOM VENTILATION - TURBINE STEAM
- DIESEL FUEL OIL - VENTILATION i - DIESEL GENERATOR - 4,160V BUS /480V BUS i - DC (125V BATTERIES) - AUTOMATIC BUS TRANSFER SCHEMES
} - FAN COOLERS i i Mdterd Lowe seul Gerekk. Inc.
~
a -
~
DIABLO CANYON PRA ROOM HEATUP ANALYSIS I e OBJECTIVE '
- DETERMINE EFFECT OF VENTILATION SYSTEM FAILUREfs j e APPROACH l - SCREEN VENTILATION SYSTEMS FOR EFFECTS ON PRA
- COMPONENTS i
- IDENTIFY LOCATIONS IN WHICH LOSS OF VENTILATION MAY
! IMPACT PHA COMPONENTS
- - DETERMINE ROOM DIMENSION, INITIAL TEMPERATURE, AND
! HEAT LOADS, ETC. \ - CALCULATE ROOM TEMPERATURE AS A FUNCTION OF TIME FOLLOWING VENTILATION LOSS
- ASSESS EQUIPMENT THERMAL LIMITS FOR KEY TEMPERATURE-SENSITIVE PRA COMPONENTS
[ - IDENTIFY RECOVERY ACTIONS FOR LOCATIONS IN WHICH VENTILATION FAILURE WOULD CAUSE LOSS OF PRA l COMPONENTS l ra me,to :cm a.ii.e. s -
i l, HUMAN ACTIONS ANALYSES i l l e 63 HUMAN ACTIONS ANALYZED l l e PERFORMANCE SHAPING FACTORS l - l e QUESTIONNAIRES FOR QUALITATIVE DESCRIPTION l l e PGandE OPERATIONS INPUT o SIMULATOR INFORMATION l l e MODEL RELATING PSF TO SUCCESS PROBABILITY l l e INCORPORATION INTO PLANT MODEL i l ra.a. t d mm i c. i
,/
l , s > !i 8
- 1 2
1
)
2 0) l-
.,.J D
&)
<W tc CC CL O _< ,
o-W
&)
- 4 I
CL i t
i l
SUMMARY
OF PHASE Il PHASE !!I DIFFERENCES 'l l e ANALYSIS TO TEST PHASE II JUDGMENTS i e DIESEL GENERATOR FUEL OIL TRANSFER SYSTEM i i
- NUMBER OF DIESEL GENERATORS 5 4 3 2 1 l - STARTS IN 6 HOURS 17 15 12 8 5 Pkkerd. Lowe and carekk. Inc.
j . _ v
i I
, RCP SEAL LOCA AND CHARGING l PUMP DEPENDENCE ON CCW 1
RCP SHAFT
. e
- '-SEAL NUMBER 3
- SEAL NUMBER 2 I - SEAL NUMBER 1 i SEAL e E -
f INJECTION N THERMAL _ j BARRIER
- l W CCW a i : : COOLS RCS WATER g
- (TO PROTECT SEALS ANDh LUBE OIL (BEARINGS SEAL INJECTION ON LOSS OF COOLING RM j ra .t o.... .
{ _ _ v
I l i FUEL OIL TRANSFER SYSTEM J G l i V
; FUEL OIL '
) STORAGE l TANK
~
1 i DG22 DG21 DG13 DG12 DG11 DAY TANK DAY TANK DAY TANK DAY TANK DAY TANK 1 i - 0-X OK OK 0-X 0-X ) .\ ' ;O 1 l V i j . FUEL OIL i STORAGE TANK l pu d. t .no o ha. i.e. i b w
PRELIMINAR) RESULTS SUBJECT TO CHANGE PHASE lilA RESULTS I l ,
' g 1 SCOPING l J STUDY l
! 4 * ) TOTAL l 1 [ c n ] NONSEISWRC a i i. 1-l )o 4 * ] SEISWNC i I I i, 1 x 10- 6 1 x 10 - 5 1 x ja-4 1 x 10 - 3 1 x 10 - 2 i j MEAN CORE DAMAGE FREQUENCY , i Pickard, t.asse seul Gm Ich, Inc. 1 r - l_ _ _ _ _ _ _ _ _ _ _ _ .
KEY ENGINEERING ISSUES FOR PHASE IllB INVESTIGATION l
- 1. RCP SEAL LOCA
! 2. CHARGING PUMP DEPENDENCE ON CCW - i
- 3. FUEL OIL TRANSFER SYSTEM i 4. RELAY CHATTER l 5. AC POWER RECOVERY FOLLOWING EARTHQUAKE
! 6. REVISED SEISMIC ANALYSIS i l l 1 l 1 r a . a. t emmw. l J - u i r}}