ML20153F866
| ML20153F866 | |
| Person / Time | |
|---|---|
| Issue date: | 05/05/1988 |
| From: | Advisory Committee on Reactor Safeguards |
| To: | |
| References | |
| ACRS-T-1666, NUDOCS 8805110075 | |
| Download: ML20153F866 (167) | |
Text
e ACPS7L /G46 r
m m
0 v
UNITED STATES O
NUCLEAR REGULATORY COMMISSION UNITED STATES NUCLEAR REGULATORY COMMISSION l
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
)
In the Matter of:
)
)
337th GENERAL MEETING
)
)
i O
t
/
i i
Pages:
1 through 111 l
"7 Place:
Washington, D.C..
. g --, g
- -~ e, -,
]t o
Date:
May 5, 1988
{
.........f.Il O O..#'*e.....
.e,,
f
........................n....3..;s......_
,i v v 1.v 3 n; y7 o.'
HERITAGE REPORTING CORPORATION I
osuarnewws O
1224 L Street, N.W., same det WasMageon, D.C. 20005 (282) 62s. mss l
8805110075 380505 PDR f4C R5 T-1666 DCD I
1 PUBLIC NOTICE BY THE 2
UNITED STATES NUCLEAR REGULATORY COMMISSION'S 3
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS 4
5 6
7 The contents of this stenographic transcript of the 8
proceedings of the Uniteo' States Nuclear Regulatory 9
Commission's Advisory Committee on Reactor Safeguards (ACRS),
10 as reported herein, is an uncorrected record of the discussions 11 recorded at the meeting held on the above date.
12 No' member of the ACRS Staff and no participant at 13 this meeting accepts any responsibility for errors or
{}
inaccuracies of statement or data contained in this transcript.
14 15 16 17 18 19 20 21 22 23 24 25
()
Heritage Reporting Corporation (202) 628-4888
1 i
(:)-
1 UNITED STATES NUCLEAR REGULATORY COMMISSION I
2 ADVISORY COMMITTEE ON REACTOR SAFEGUARDS 3
1 4
In the Matter of:
)
)
5-
)
337th GENERAL MEETING
)
6
)
)
7
- Thursday, 8
May 5, 1988 9
Room 1046 1717 H Street, N.W.
10 Washington, D.C.
20555 i
11 The above-entitled matter came on for hearing, 12 pursuant to notice, at 8:30 a.m.
.()
13 BEFORE:
DR. WILLIAM KERR i
Chairman 14 Professor of Nuclear Engineering l
Director of the Office of Energy Research l
15 University of Michigan Ann Arbor, Mici.Agan
[
16 I
ACRS MEMBERS PRESENT:
17 DR. FORREST J.
REMICK 18 Vice-Chairman Associate Vice-President for Research 19 Professor of Nuclear Engineering The Pennsylvania State University i
20 University Park, Pennsylvania 21 MR. CARLYLE MICHELSON Retired Principal Nuclear Engineer i
i 22 Tennessee Valley Authority Knoxville, Tennessee 23 and Retired Director, Office for Analysis and Evaluation of Operational Data 24 U.S. Nuclear Regulatory Commission I
Washington, D.C.
l
(:)
25 HERITAGE REPORTING CORPORATION -- (202)628-4888
i 2
j R.
i 1
'd 1
MR. DAVID A. WARD Research Manager on Special Assignment 2
E.I. du Pont de Nemours & Company Savannah River Laboratory 3
Aiken, South Carolina 4
DR. HAROLO LEWIS Professor of Physics 5
Department of Physics University of California 6
Santa Barbara, California 7
DR. DADE W. MOELLER Professor of Engineering in Environmental Health 8
Associate Dean for Continuing Education School of Public Health 9
Harvard University-Boston, Massachusetts 10 DR. PAUL G.
SHEWHON 3
11 Professor, Metallurgical Engineering Department Ohio State University 12 Columbus, Ohio t
()
13 DR. CHESTER P.
SIESS Professor Emeritus of Civil Engineering 14 University of Illinois Urbana, Illinois 15 DR. MARTIN STEINDLER 16 Director, Chemical Technology Division Argonne National Laboratory 17 Argonne, Illinois 18 MR. CHARLES J. WYLIE Retired Chief Engineer 19 Electrical Division Duke Power Company 20 Charlotte, North Carolina 21 ACRS COGNIZANT STAFF MEMBER:
22 Raymond E.
Fraley 23 NRC STAFF PRESENTERS:
t 24 John Flack John Lambright F
25 Jerry Wilson
)
L 3
HERITAGE REPORTING CORPORATION -- (202)628-4888
_ _. -. - ~ - _ _ _ _ - - _ _,. - -
3 1
E R. O C E E D _I H G S.
(
2 CHAIRMAN KERR:
The meeting will come to order.
3 This is the first day of the 337th meeting of the ACRS.
4 During today's meeting the Committee is scheduled to discuss 5
fire risk scoping study,-regulatory guides, integrated safety 6
assessment program, individual plant examinations, containment a
7 systems, and to discuss further ACRS subcommittee activities 8
and some staff activities.
9 Items for discussion tomorrow are listed on the 10 bulletin board in the back of the meeting room.
The meeting 11 is being conducted in accordance with the provisions of the 12 Federal Advisory Committee Act and the gov inment and the
()
13 Sunchine Act.
Mr. Raymond Fraley is the designated federal 14 official for the initial portion of the meeting.
15 We have received no written statements or requests 16 to make oral statements.
17 (Items of current interest were discussed off the 18 record.)
19 CHAIRMAN KERR:
We will continue with the business 20 of the Committee which begins with the fire risk scoping study 21 discussion which Mr. Siess is the cognizant subcommittee i
22 chairman.
l 23 MR. MICHELSON:
I don't believe so.
}
()
24 DR. SIESS:
It was typographical error.
25 CHAIRMAN KERR:
CPS Michelson--Mr. Michelson.
HERITAGE REPORTING CORPORATION -- (202)628-4888
]
1 NR. MICHELSON:
Thank you, Mr. Chairman.
This 2
morning I think we will have the opportunity to hear about the 3
fire scoping study which was performed by Sandia, and has now 4
been completed.
We had--part of this work I think was 5
prompted by the ACRS letter of July '86 in which we indicated 6
some concern about the loss of program momentum and 7
information that was resulting from the zeroing out of the 8
fire research at that particular point in time, so this was 9
brought to the attention of the Commission.
They agreed that 10 something should be done, and they authorized the staff or 11 instructed the staff to look into the matter, and to work with I
12 ACRS on seeing what the problems might be.
()
13 That prograa then got well underway by late '86 and 14 we first heard from Sandia in August of
'87, although we had 15 been working informally with them up to that time.
16 The Committee did have a subcommittee meeting at i
17 that time, and we indicated that the progress appeared to be 18 satisfactory, and that we were awaiting the final issuance of 19 the report to provide any views that we would have.
We are 20 now at that point.
We have had a subcommittee meeting, and in 21 August--pardon me.
We had a subcommittee meeting in March, p
22 March
'88, and at which time we had I think very fine det. ailed 23 briefing from the staff and from the Sandia representatives
()
24 who were performing the study.
25 There was I think a very active participation on the i
P HERITAGE REPORTING CORPORATION -- (202)628-4888
m
.\\
5 i
i part of the, of industry representatives in the sense they I
s 2
were in the audience and from time to time we even had some i
3 discussion with.them.
The information for this particular 4
briefing in to be found in Tab 2.
Each member of the 5
Committee has received a copy of the final scoping study, and j
6 we intend today to receive the briefing from the, abbreviated 7
briefing from Sandia.
8 I would not, though, recommend at this time we send 9
9 a letter since the staff is now completing its review and 10 preparing recommendations on where we should go from here.
It 11 is anticipated that these recommendations will be received 12 about July of this year, and at that time, we would come back I
()
13 to the Committee with any further briefing that might be 14 needed, and of course, for a final letter from the Committee 15 concerning whatever views it may have.
i 16 At this time, we should concentrate on just
[
17 understanding what the results of this fire scoping study 18 might be, and keep them in mind as we see whatever the 19 recommendations of the staff will be which will be forthcoming i
20 fairly shortly,'and so 1 do not intend unless the Committee 21 wishes, I would not intend to, to have a letter at this 22 meeting.
Just for now it is primarily for education, so with l
23 that, I would like to turn it over to John Flack, who is the j
()
24 program or project manager for this program, and he has a f
25 rather complete briefing prepared for you.
i HERITAGE REPORTING CORPORATION -- (202)628-4888 I
.. ~
6 1
HR. FLACK:
Good morning.
I am John Flack. :I am 2
with the Advanced Reactors and Generic Issues Branch in the 3
Office of Research.
4 I am the project manager of the scoping study.
The 5
purpose of this meeting is to present to you the results, the 6
conclusions and recommendations of that study.
It was 7
performed at Sandia.
The principal investigators are Mike 8
Bohn, who is the project manager at Sandia; Steve Nowlen, who 9
is-the fire protection engineering expert, was also involved 10 in the NRC fire protection program; John Lambright, who is the 11 fire PRA expert at Sandia; and Vern Nicolette, who is not with 12 us today, but is the code expert, worked on the COMPBRN III
()
13 code, fire code.
14 (Slide) 15 MR. FLACK:
The fire scoping study, the purpose of 16 the firing scoping study was to assess the risk significance 17 and dominant sources of uncertainty associates with fire 18 issues.
The information from this study is to be used to 19 determine the need for additional fire prctection research, 20 and the study uses current state of the art methodology, did 21 not develop any new methodology.
The budget would not allow j
22 that; using the most up-to-date data available, some of which 23 has come out of the NRC fire protection program.
24 It also includes Appendix R backfits, other backfits
()
25 that were made to give a proper risk perspective, and it HERITAGE REPORTING CORPORATION -- (202)628-4888
7 1
considers potential fire risk issues.
()
2 (Slide)-
l 3
HR. FLACK:
The study was set up to focus on 4
specifically the major fire risk contributors, to identify the 5
sources of uncertainty so that any additional research that 6
may be warranted could attempt to narrow these uncertainties, 7
and completeness of the fire risk issues which considers if 8
not all of the potential issues, at least what is believed to 9
be the most aignificant.
10 Okay.
Brief chronology--the fire risk scoping study 11 was initiated in January of
'87.
Tasks 1 and 2 were completed 12 in June of that year which included requantification of four
(
13 PRAs.
In July we presented those results to the subcommittee, 14 ACRS subcommittee, and also got feedback from the subcommittee 15 back into the program.
5 16 In August we presented those results to the Full 17 Committee, and had a letter stating that we were on target and 18 essentially addressing the concerns of the Committee.
19 In January, the first draft was completed.
In 20 February we had a peer review which although over at the end 21 of the presentation, comments were made at that meeting.
22 In March we went again to the subcommittee, second i
I i
23 draft was written, and beetme available in April, which fed 3
l
()
24 back into it the comments up to that time, and here we are in i
l 25 May in the Full Committee.
l l
HERITAGE REPORTING CORPORATION -- (2021628-4888
_ - ~, -
8 1
DR. MOELLER:
Could you remind us how the peer 2
review was done?
Who was involved?
3 MR. FLACK:
Well, the peer review was essentially 4
set up through the lab that identified the experts out in the 5
field that were not initially involved in Task 3, which was a, 6
which is identifying potential issues.
Okay.
Th9y were 7
contacted through the lab.
I don't know if the lab wants to 8
comment on that.
We will wait unti3 later.
9 Under Task 3 where we list all the experts that were 10 involved in that and then the peer review was, was set up to 11 review the scoping study, also to get feedback on the study 12 which could be incorporated into the draft report.
()
13 MR. BOHN:
We identified a number of sources.
It 14 turned out to be thirteen, to gather information on issues 15 that we felt to be outstanding relative to fire, but then we 16 chose another set, an independent set of people to perform the 17 peer review, so there was no overlap, and the peer review was 18 primarily in the utility personnel.
19 DR. MOELLER:
How long did the meeting last and how 20 much ahead of time did they receive the material to read and 21 so forth?
22 MR. LAMBRIGHT:
The group's involvement of peer 23 safety received their information approximately a month before 24 the peer review meeting, and the meeting itself lasted one
()
25 full day.
HERITAGE REPORTING CORPORATION -- (202)628-4888
i 9
1 DR. MOELLER:
Thank you.
.O 2
DR. SHEWMON:
Is there any place I am suppesed to 3
have a schedule who says who is talking or whether this is the 4
4 first speaker or the last speaker we hear?
I am looking at 5
Tab 2 and I find a lot of things, but I don't find that.
6 CHAIRMAN KERR:
Page 1 of Tab 2.
l 7
DR. SHEWMON:
Okay.
That's the schedule for this 3
8 meeting.
9 CHAIRMAN KERR:
I believe so.
That's my 10 interpretation.
11 DR. SHEWMON:
Thank you.
12 HR. FLACK:
If there is no further questions, I L
()
13 would like to turn it over to the researchers.
i 14 CHAIRMAN KERR:
I believe you say one of the things 15 that you hoped to do is to determine whether additional 1
16 rerearch was neided.
17 What criteria do you plan to use to determine
^
18 whether additional research is needed?
19 MR. FLACK:
Well, the study is risk based, so we are 20 looking at--the study has been, the study was risk based, so 21 there are different mechanisms by which to judge the 22 significance of the research that will be done.
23 At the, following the presentation, my section
()
24 leader will comment as to where we will go from here, how it i
25 will be prioritized and so on within the system.
At this i
J 4
HERITAGE REPORTING CORPORATION -- (202)628-4888
d 10 1
point, we don't have a position exactly on what research will
(
2 be done.
3 CHAIRMAN KERR:
I am not asking whether you plan to 4
do more.
I am asking how you are going to decide whether i
5 additional research is needed.
6 MR. FLACK:
Well, I would say at this point in time, 7
we are trying to understand the study.
We are looking through s
8 the study.
I had figures where the sources of uncertainty are 9
coming from, what the risk is, and we will ultimately come 10 down to a form of engineering judgment based on what 11 information is available.
The exact process taking place is 12 now just a reviewing and trying to understand the fire risk
()
13 issues.
14 DR. SHEWHON:
Your answer is you have no criteria at 15 this time, but you think you will know when you get there?
]
16 MR. FLACK:
All--that there are influences on the 17 direction of research based on risk, yes.
18 MR. MICHELSON:
I think he indicated it is strongly 19 risk based because that's the purpose of this study, to i
l 20 ascertain the risk, and if you find some potentially 21 significant outliers as a result, then that would be, direct
[
?
l 22 the research program.
I 23 CHAIRMAN KERR:
He may have meant to say that, but i
(}
24 he didn't.
I was trying to say that.
l l
HERITAGE REPORTING CORPORATION -- (202)628-4888 l
'g 11 1
CHAIRMAM KERR:
He said it was. risk based, but he
.D\\
k/
2 did not mention a particular risk-that was going to or
.3 uncertainty that was going to be used.
4 MR. MICHELSON:
Nor did I.
I didn't indicate which 5
particular one, either, because we don't know yet.
6 DR. SHEWHON:
If there is a large risk, you fix it 7
rather than do more research on it, too, i
8 MR. MICIELSON:
It may require research to determine i
9 a fix.
10 MR. BOHN:
Later on, you will see a slide that ranks 11 the issues relative to their potential order of magnitude, 12 changes in risk, and furthermore, we have attempted to (j.
13 identify which issues are in effect generic which could affect 14 all plants versus those which are highly plant specific and 15 may be important for individual plants.
16 CHAIRMAN KERR:
I was trying to find out what the i
17 NRC staff was going to use as their criteria or the criteria
[
18 for asking for further research.
I recognize that you will be
[
i 19 able to describe the research you have done, j
20 Any further are questions?
I t
a 21 DR. SIESS:
It is so difficult for people to 22 understand.
i 23 MR. FLACK:
I will turn it over to the research t
()
24 engineers that have actually performed the study; Mike Bohn.
25 HR. BOHN:
My name is Mike Bohn from Sandia Labs, I
HERITAGE REPORTING CORPORATION -- (202)628-4888
12 1
and I am on the agenda in two places.
This first brief
(
2 introduction to the program is in the first part of your 3
handout, and the talk on recommendations as to what has come t
4 out of this program is contained in the second half of the 5
presentation, so don't throw it away after I have done 6
basically.
7 I want to just give you a brief overview of what we 8
have done in this project.
9 (Slide) 10 MR. BOHN:
The study was intended to provide a 11 re-examination of past PRAs, which of course, which of course 12 was performed on, commercially perforned in supported PRAs so
()
13 we had four that we looked at.
In modifying these, we took a 14 look at impact of the latest data base that had been generated 15 under NRC sponsorship.
16 DR. MOELLER:
These are not general PRAs, but 17 specifica13y fire PRAs?
18 MR. BOHN:
These came from four general PRAs that 19 had a fire PRA as part of them.
Yes?
20 DR. SHEWMON:
These data bases are how often 21 somebody finds a can of volatile combustible material in an 22 area where it shouldn't be?
I 23 MR. BOHN:
I have a slide on that in just a moment.
24 But yes, that's the nature of the--we have actually three data
()
25 bases.
[
l HERITAGE REPORTING CORPORATION -- (2021628-4888
13 1
Second was to determine the risk significance of a O-2 number of previously unaddressed fire issues, that issues that 3
had not been included in past fire PRAs but issues that had 4
been raised by members of the ACRS, members of the NRC, and 5
the outside fire community, and they were in general supported r
6 by the utility personnel, fire engineers that we talked to.
i 7
Thirdly, an objective was to assess the impact of 8
Appendix R rules, and the modifications that had been made to 1
9 the plants in response to Appendix R and to assess their 10 impact on the dominant accident scenarios that had been I
i 11 identified in the risk assessments, j
12 CHAIRMAN KERR:
What does impact mean?
i
(
13 MR. BOHN:
We were trying to see whether or not the 14 changes that had been incorporated into Appendix R actually 15 made certain dominant sequences cr failure modes go ahead, or i
l 16 whether or not they affected things but still allowed a number 17 of dominant sequences to remain.
18 CHAIRMAN KERR:
So impact in this case means risk?
19 MR. BOHN:
Yes.
20 CHAIRMAN KERR:
Risk of core melt or--
21 HR. BOHN:
That is correct--core damage frequency.
I l
22 However, we could make some, by looking at failure modes, we 23 could, although we didn't quantify it, we could make l
l 24 inferences relative to man-rem at the boundary release measure f
()
l t
25 in effect because in certain cases, we could identify that the l
i l
T HERITAGE REPORTING CORPORATION -- (202)628-4888
14 1
failures that led to the core damage state also failed to O
2 containment, mitigating systems in which case--containment 3
sprays in particular, in which case you would assume a slow 4
build-up and nonetheless a containment overpressure situation.
5 CHAIRMAN KERR:
Thank you.
6 (Slide) 7 MR. BOHN:
The project was broken down into five 8
tasks.
I put this up primarily because it was going to be the c
9 order of presentation.
10 The first task was to reassess four previous PRAs 11 and their sources of uncertainty, and to requantify the fire 12 scenarios, so the first two tasks, John Lambright will be
()
13 talking right after myself.
14 Task 3 was to identify the potential fire issues i
t 15 and, this involved correlating all the questions and 16 information that had come out of the fire protection research 17 program over the past eight year, questions that were raised 18 by the NRC, the ACRS, and to come to a consensus on what were 19 the outstanding issues that remained to be examined.
20 DR. MOELLER:
How do you get a hold of that 21 information?
I 22 MR. BOHN:
Steve Nowlen will have a presentation on 4
23 explaining that.
()
24 DR. MOELLER:
I'll wait.
25 MR. BOHN:
Task 4 and Task 5 were to assess the risk l
1 i
HERITAGE REPORTING CORPORATION -- (202)628-4888
15 1
significance of the potential issues once they had been O
2 identified, and also evaluate the extent to which Appendix R 3
might impact them or mitigate them.
4 The way this will be presented is that we will 5
present, Steve and John will present the conclusions of the 6
research during their talks, and then I will follow up with a i
7 talk, brief talk on the recommendations for research that 8
would be done either by NRC or by the utilities, action by the 9
utilities to help improve the situation.
10 So conclusion and recommendation will be presented 11 first of all for the PRA requantification task, and then 12 secondly for the new issues, and here are the six new issues
)
13 that will be discussed.
They are manual fire fighting 14 effectiveness, control systems interactions, and those two, by 15 the way, we feel are the most important two in terms of 16 priority to come out this study.
i 17 Then there are additionally environmental effects of 18 smoke and inadvertent suppression, which can also be quite l
t 19 important; fire code adequacy, barrier effectiveness, and then
(
l t
20 lastly, seismic fire interactions, so these will all be 21 discussed relative to conclusions and recommendations.
22 (Slide) l 23 HR. BOHM:
For the first two tasks that John
()
24 Lambright is going to discuss, we studied and requantified the 25 four PRAs associated with Limerick, Seabrook, Oconee and I
i i
HERITAGE REPORTING CORPORATION -- (202)628-4888
16 1
Indian Point.
Now those were performed by Piccard, Lowe and 2
Garrick, and one was performed by NUS, but they are 3
essentially the only two companies that have been performing 4
these fires PRAs, so it pretty well covers the field.
L 5
New data was incorporated in four areas, three of
(
6 which are the data base that you asked about--first of all, in 7
fire initiating frequencies, extensive review of all the fire 8
occurrences that could affect safety systems were performed, 9
resulting in about 368 fire occurrences in the data base, and 10 so--
11 DR. SHEWHON:
All nuclear plants?
12 HR. BOHN:
That is correct, all of them, and trash
()
13 can fires and small non-threatening fires were not, were y
i r
14 excluded from the data base.
There is, was a criterion as to 1
15 the severity of the fire and its potential impact before it 16 was included in the data base.
17 Number 3 is that the new fire suppression model.
t 18 Information data base was collected on the times to suppress 19 fires from fire occurrences, and this data base was
,I i
20 significantly greater than exirted before.
In fact, we had, 21 we have 68 fire occurrence suppression times.
It turns out it l
22 is fairly difficult to get good times to suppression from fire v
23 reporting, but previous PRAs were base ( on data base that only
()
24 had eleven occurrences, so this is significantly greater and 25 this could be used to generate a model for fire suppression.
l
_, _ _ _ _ _ _ _ _ _ _ HERITAGE REPORTING CORPORATI0!(-- (202)628-4888___;
n
)
17 l
1 The last data base had to do with cable failure i
3 temperatures.
Past PRAs had often used autoignition 3
temperature or some type of failure.
We used information that j
4 came out of the NRC fire protection research program that had i
5 actual damage temperatures, loss of, loss of resistance if you 6
will to the cables, and it turned out that they are 7
significantly less, of course, than actual cable installation, l
8 flammability temperatures.
9 The last area whera new data was incorporated had to i
t 10 do with the fire code that was used in the PRA.
All the 11 commercial PRAs to date have used a code developed at UCLA 12 called COMPBRN I.
There is an improved version of the
()
13 COMPBRN, COMPBRN III, which we used and evaluated.
It was our 14 attempt to see the impact of the latest, greatest version of l
15 COMPBRN over previous calculations, so these are the four l
i 16 areas that were incorporated in the fire PRA requantification, j
j 17 and there are some limitations which I should point out 13 relative to the requantifications, namely, we did not go back I
19 and visit the plants themselves.
We did not walk them down.
20 We did not develop judgmental factors that go into any fire i
I 21 PRA.
We used the original factors in the PRA.
t i
22 It was only the four areas that I just mentioned l
23 that were indicated.
In particular, plant-specific l
t t
24 partitioning factors, which are used to ratio the building
(}
25 fire frequencies for which we have data, down to the HERITAGE REPORTING CORPORATION -- (202)628-4888
18 1
subcompartment rooms where the fires are hypothesized to 2
occur.
These partitioning factors are very judgmental and 3
they require walk-down, estimate of transient combustible 4
cable content, et cetera.
We used the original partitioning 5
factors in the four PRAs.
6 Finally, two of the four PRAs did not have an 4
7 uncertainty analysis performed with them, so we did not 8
attempt to contruct an uncertainty analysis for the original 9
PRA.
We did perform uncertainty analysis for our 10 requantifications.
I 11 Finally, and most importantly, the six new issues ~
12 that will be discussed here, none of those were incorporated
()
13 in the requantification.
We attempted to address their risk 14 impact separately because again, we did not visit the plants i
15 and we couldn't, we couldn't make judgments as to how much 16 these six issues affected the particular plants being I
17 requantified.
18 DR. MOELLER:
A couple of questions I should have 19 asked on the previous slide--when you said that you collected l
20 the data on so many firee that have occurred, were there 21 lessons to be learned there?
These fires occurred in more, r
22 frequently in certain types of plants versus others and in l
1 23 certain places within plants, or will we hear that later?
]
()
24 MR. BOHN:
I'm sorry.
We didn't bring the slides on 25 that, but Steve I believe can comment on that.
Would you like i
I HERITAGE REPORTING CORPORATION -- (202)628-4888
19 1
to comment on that?
(')N
(
2 HR. LAMBRIGHT:
This is John Lambright from Sandia
)
i 3
National Laboratories.-
)
4 I would like to say that due to the lack of enough 5
-fire occurrences in any specific plant, it is hard to make a 6
judgment as to whether fires are more likely to occur in one 7
type of plant rather-than in other.
8 Typically fires are occurring anywhere from once 9
every six to ten years in a given plant type so any given 4
10 plant may have no more than two or three fire occurrences in 31 its total operating history.
12 DR. HOELLER:
Okay.
Thank you.
The second item, in t
()
13 terms of the PRAs, do they specifically address the control 14 room?
15 HR. BOHN:
They usually do, and they usually assume r
i 16 that a fire has occurred that damages everything in the 17 control room.
It is fairly course resolution.
And we as part t
18 of the control systems interaction study, we attempted a l
19 detailed look at that question, and we had some interesting l
20 conclusions.
21 In regard to your earlier question, from the data it l
t 22 is not clear that the rate of fire occurrences is decreasing i
23 at all.
It seems to be roughly constant.
()
24 HR. LAMBRIGHT:
I would like to make another comment 25 about addressing control room.
l
\\
r HERITAGE REPORTING CORPORATION -- (202)628-4888 f
m 20 1
In the Limerick PRA, NUS methodology, essentially I
2 they would like take a look at one cabinet that allows spread l
3 to adjacent cabinet, where in the Seabrook PRA they took zones 4
within the control room for failure rather than the control 5 -
room as in either one of those cases, but typically, people l
6 look at control room failing as a whole.
7 DR. MOELLER:
Thank you.
8 MR. BOHN:
Well, with this lead-in, I would like to 9
introduce John Lambright, who will discuss the PRA 10 requantification and its implications.
t 11 CHAIRMAN KERR:
Has there ever been a fire in which i
12 all the control room failed?
()
13 UR. LAMBRIGHT:
No.
Good morning.
My name is John 14 Lambright from Sandia National Laboratories, and as Mike 15 mentioned earlier, I will be discussing the results of Tasks 1 l
16 and 2 of the fire scoping study.
17 (Slide) i 18 HR. LAMBRIGHT:
Task 1 and 2 had a two-fold purpose, i
l 19 First, we requantified certain fire scenarios in four previous 20 fire PRAs--the Limerick, oconee, Indian Point, and Seabrook i
21 PRAs--and then we assessed uncertainties also in the four l
22 previous fire PRAs.
23 We discovered four ma]or areas of uncertainty first
()
24 of all, in fire frequencies and partitioning methods, also 1
25 what type of initiator would be probable given a fire HERITAGE REPORTING CORPORATION -- (202)628-4888 J
[
t 21 i
1 occurrence, transients or loss of coolant accidents, certainly 2
in fire propagation modeling, and then in their fire 3
suppression model that they used.
4 CHAIRMAN KERR:
Excuse me.
What was the uncertainty 5
in No. 2?
6 MR. LAMBRIGHT:
For instance, in the Limerick PRA, 7
they said that there wasn't a possibility of having a fire 8
initiated loss of coolant accident occurring, whereas for the l
9 three PL&G PRAs it turned out the dominant contributor was a 10 transient-induced loss of coolant accidents for core damage t
(
11 frequency initiated by fire.
l 12 MR WARD:
What were those?
Opening of, PORV
()
13 opening?
f 14 MR. LAMBRIGHT:
Yes.
That was typical.
Also i
15 reactor coolant pump seal-up as well, but either of those i
i i
l 16 mechanisms contributed.
For the Indian ?oint PRA, it was I
l l
17 about an equal contribution of each, whereas for the Seabrook 1
18 and Oconee PRAs, typically the seal LOCAs being the dominant
{
19 contributor.
20 DR. HOELLER:
On your first one, the partitioning,
(
I I
21 would you elaborate?
I
]
22 MR. LAMBRIGHT:
Okay.
I have another slide that i
23 will elaborate a little bit further on that.
(}
24 (Slide) 25 HR. LAMBRIGHT:
Past PRA fire frequencies were based i
1 i
i HERITAGE REPORTING CORPORATION -- (202)628-4888
22 i
1 on historical dhta either from insurance reports or from h
2 license event reports, and in all cases, judgmental I
3 modifications were needed to derive fire area specific l
f 4
initiating events frequencies.
These partitioning factors 5
were necessary, as I said, to develop for a particular plant 6
area fire initiating frequency.
7 Due to the nature of the data, typically it is 8
reported in the six areas for larger buildings such as aux i
9 building or control room and cable spreading room, and to get i
t 10 a specific fire initiating frequency, either area or severity 4
11 partitioning had to be accomplished.
i 12 DR. SHEWHON:
What were the initiating event and l
O 13 re ta v 1
r viet tio= or a i i tr tive orecea=re -
=r 14 where did they come from?
15 HR. LAMBRIGHT:
Initiating events came from a number 16 of sources, anywhere from transient combustible fires to fires 1
17 due to pumps or diesel generators--whole wide spectrum of i
i 1
18 sources.
r I
19 For the PL&G methodologies for Oconee, Seabrook, and j
20 Indian Point, essentially they smeared all the data and 4
21 modeled it as one transient source for their code l
l calculations, whereas for Limerick methodology--
i I
23 DR. SHEWHON:
Transient means limited amount to burn 24 or somebody is walking through with it?
]
25 HR. LAMBRIGHT:
Typically *, hey said any--they l
1 i
HERITAGE REPORTING CORPORATION -- (202)628-4888
23 l
1 smeared all the data and modeled it as a one-foot pool of oil O-2 for their code calculations, for their COMPBRN calculations,-
i i
3 whereas in the Limerick methodology they took combustible l
4 sources.
They took a look at self-ignited electrical panel f
5 fires, self-ignited cable fires, and transient combustible r
6
. sources.
]'
7 HR. NOWLEN:
The question on transients in this S
8 case, a transient is something that they bring into a plant 9
that is not a part of the installed systems that design the i
j 10 plant.
It is they bring in wood to do some work or they have l
11 trash that is laying around the plant and in situ, or other l
\\
12 types of fuel, the cables and the panels and things.
l 13 HR. LAMBRIGHT:
As I mentioned, a lot uncertainty in 14 the fire initiating sources, major contributor of Limerick, 87
)
15 percent to core melt frequencies was from self-ignited panel 16 fires and self-ignited cabling fires, whereas the PL&G l
17 methodology didn't even address these two particular sources.
3 18 DR. MOELLER:
What is self-ignited?
i i
l 19 MR. LAMBRIGHT:
That would be a fire that would 1
^
'O start within a particular electrical panel, or a fire that 4
l i
)
j 21 was--
[
t i
22 DR. MOELLER:
It was due to some error or something j
23 in the design?
f
~O 24 "a
'inea1our:
ee id1r errer i= the ae ie=.
]
25 possibly a faulty breaker operation--a lot of different l
i
,I I
i HERITAGE REPORTING CORPORATION -- (202)628-4888 l
l 24 I
sources in the data base.
2 DR. HOELLER:
Do you mean it, the source of the i
3 ignition was due to a failure or an error or something within l
t 4
the system itself?
It was not ignited from outside?
l l
5 HR. LAMBRIGHT:
That's right.
Ignited interior to 6
the electrical panel, or say on a splice, for instance, on the i
7 cable itself.
PL&G considered the transient oil fires would I
8 be dominate core damage frequency.
That's all they took a 9
look at.
10 CHAIRMAN KERR:
Did they say how they arrive at that i
11 conclusion?
i 12 HR. LAMBRIGHT:
No, they didn't.
l
()
13 DR. HOELLER:
And is, are pools of oil commonly j
j 14 present throughout nuclear power plants?
i 15 HR. LAMBRIGHT:
Not to my knowledge.
However, using t
16 a pool of oil to model, model the COMPBRN calculations would 17 be representative of a typical heat source for these fires.
1 18 and so that's why they used it.
19 HR. B0HN:
Typically volatiles are things like i
20 acetone used in cleaning, and there are oils, usually small 21 oil pools adjacent to pumps and other things and it is next to 22 things like acetone.
l 23 MR. HICHELSON:
Big snubbers are leaking a fair j
f 24 amount of oil; bearing leaks on oil-cooled systems.
(}
t J
25 HR. LAMBRIGHT:
Most oil fires--
[
I HERITAGE REPORTING CORPORATION -- (202)628-4888 l
-l 25 l'
HR. MICHELSON:
Reactor coolant pumps is a big
(:)
4 2
source of oil.
3 HR. LAMBRIGHT:
You see in the data base it may i
4 occur from-a pump maybe spilling oil on hot lagging or diesel 5
generator exhaust type of fire.
l 6
HR. BOHN:
The use of pool of oil as initiator is 7
just a means of regulating or quantifying the sire of the heat 8
scurce and the equivalent quantities of aceton and other 9
combustibles.
i i
10 DR. HOELLER:
Thank you.
1 11 MR. LAMBRIGHT:
COMPBRN I was used in all the four 12 original PRAs.
Parameter uncertainties were quantified using
(
13 hypercube sampling technique for the input vectors.
Modeling 14' uncertainties were kept with first of all, with PRA by using i
j 15 assumed correction factor multiplier on the time of the damage i
l l
16 of II.
COMPBRN I code was regarded as overly conservative, 17 because the original version neglected heat losses from the I
18 fuel targets.
)
li f
i 19 The fire suppression codels that were constructed in I
i 20 all four PRAs were based on engineering judgment and also a
{
i 1
21 limited data.
The source was HGGR report in 1978 which had I
i 22 eighteen data events, and for instance, for Limerick PRA, and
[
i I
23 Oconee, they only used one suppression model, no matter what l
24 the fire area, no matter there was automatic suppression in f
l
()
L 1
25 detection of the fire, whereas for Indian Point and Seabrook.
O b
I f
HERITAGE REPORTING CORPORATION -- (202)628-4888
26 1
they used two models as a whole for both of the PRAs, so to 2
construe these suppressions as plant-specific to any given 3
area, I don't think you could do that.
~
4 Actual detection and suppression features, as you S
mentioned, weren't really included in the construction of 6
these models.
Also the potential risk for mis-application'of 7
suppression agents was ignored.
j 8
(Slide) 1 9
HR. LAMBRIGHT:
If we take a look at the types of 10 initiators that could be caused given fire occurrence, 11 Limerick did not consider LOCAs credible due to cable 12 separation, whereas if you take a look at the Indian Point 2 0
13
=a oco
>*^ -
11 'oc^
aroauc a dv etr tr i===
14 dominated the fire induced core damage frequency 95 percent.
15 CHAIRMAli KERR:
Does that have anything to do with 16 the fact Limerick is a BWR and Indian Point is a PWR7 1
17 HR. LAMBRIGHT:
That is certainly one effect because l
18 you don't have possibility of having a seal LOCA.
- However, l
19 one of the things I am looking at right now is Peach Bottom 20 which is a BWR, and we haven't been abin to screen out those 21 types of sequences in our screening analysis, having a 22 situation where we have a transient with one or multiple stuck 23 open relief valves leading to LOCA situation.
Sequences 24 aren't going that way in the screening stage at least.
I 25 don't know if it will turn out that they are dominant or not.
i HERITAGE REPORTI!iG CORPORATIOli -- (202)628-4888
i f
I l
27 l
1 but certainly we wouldn't have screened them out initially..
i
(
2 DR. HOELLER:
On that, to be sure I understand it, i
3 are you saying that 95 percent.of all the small LOCAs that are i
4 postulated to occur at Indian Point 2 occur as a result of l
5 fire?
i
-6 HR. LAMBRIGHT:
No.
What I am saying is that 95
'[
t 7
percent of the fire core damage frequency comes from 8
transient-induced small LOCAs.
9 DR. MOELLER:
Okay.
Well then, what percont of all
[
l 10 of the potential induced core melt comes from fire?
i 11 HR. LAMBRIGHT:
For Indian Point plant on initial i
12 study, I think it was roughly around 60 percent, and--60 I
I 13 percent, and most of it, as you see, was coming from
[
()
l 14 transient-induced small LOCAs, j
?
15 DR. SHEWMON:
The small LCCA means here that you in I
I 16 some way mess up the electrical supply to valves so that it i
i 17 doesn't function?
[
i 18 MR. LAMBRIGHT:
What is occurring in this particular i
ti 19 case, you have a fire either in the cable spreading room or 20 electrical tunnel which takes out the support system for the 21 component cooling water system such that it gives you a seal l
i 22 LOCA, and at the same time, fails your high pressure injection 23 function, and also for any end point they assessed the
{
l 24 possibility of having a stuck open relief valve for that
(}
25 particular plant.
It didn't turn out to be a dominant i
l i
HERITAGE REPORTING CORPORATION -- (202)628-4888
28 1
contributor as the seal LOCA contributor was.
2 Then let me make a further point.
For these two 3
particular areas that we looked at, also all the containment 4
spray systems would be failed given a fire too, so not only do 5
you have a potentially high core damage frequency scenario, 6
you also have a potential risk, high risk contributor at the 7
same time.
8-DR. MOELLER:
Could you help me again?
The previous 9
reviews that we have heard, if I understood them, pointed out 10 that the dominant contributor to potential core melt in most 11 nuclear power plants was of, was an event or were events that i2 were seismically related.
()
13 MR. LAMBRIGHT:
From the ten fire PRAs that we have 14 reviewed, the core damage frequency from fire in comparison to 15 seismic and all internal event initiators was roughly 60 16 percent to 70 percent.
17 DR. MOELLER:
And is, then are seismic events the l
18 next biggest category?
19 MR. LAMBRIGHT:
I am not sure of that particular l
20 issue.
l 21 MR. BOHN:
We have looked at a number of past PRAs, 22 and it depends.
Fire tends to be somewhere from a low of 23 maybe 10 percent up to as much as 60 or 70 percent.
- Totally,
()
24 seismic tends to be up around the 50 percent level almost in 25 all of them pretty much, so it is hard to, it is hard to say.
HERITAGE REPORTING CORPORATION -- (202)628-4888
29 1
Some of the past fire PRAs he.ve been excessively Q
2 conservative, including some done by Sandia, so it is probably 3
true that they are roughly comparable, but they always make a 4
_significant contribution to the overall risk, including 5
internal.
6 MR. MICHELSON:
These results are after you have 7
re-evaluated the models and made your data input corrections, 8
and these are presumably more realistic estimates than the 9
previous ones.
10 MR. BOHN:
That is correct.
11 MR. MICHELSON:
Still perhaps conservative.
12 MR. BOHN:
In some areas as we will see, some
(
13 incorporating new data lowered frequencies and raised others, 14 but the net effect was it didn't--still quite significant.
15 DR. MOELLER:
Thank you, i
16 MR. LAMBRIGHT:
I will be addressing that a few 17 slides to come.
18 CHAIRMAN KZRR:
If you had seismic experts review 19 the fire PRAs, the fire experts review the seismic PRAs, what 20 would you guess might happen?
21 MR. BOHL:
Well, being a resident seismic expert at 22 Sandia, I would certainly applaud your goal.
The bottom line 23 is I feel that re-evaluating these PRAE ve have significantly
()
24 improved their reliability and their scrutability and 25 confidence we can handle them and so I have much more HERITAGE REPORTING CORPORATION -- (202)628-4888
30 1
confidence in our requantifications than some of the earlier
(~/\\
\\-
2 work that I was involved in in others, so I think that the 3
two, in terms of their ability to be comparable, especially 4
the methodology that we are currently using, we are performing 5
the 1100 fire PRAs as well as the seismic PRAs, and in our 6
view, the two methodologies are encouraging in terms of 7
comparability.
8 CHAIRMAN KERR:
What I am hearing is that if you 9
eliminated seismic events and fires, you wouldn't have 10 anythius else to worry about hardly?
11 MR. BOHN:
That would in general take care of most 12 of the external events.
That is certainly true.
Only
()
13 rarely--
14 CHAIRMAN KERR:
Wait a minute.
We are only talking 15 about external events.
I thought you were talking about total 16 core melt frequency.
You are only talking about core melt 17 frequency do to external events?
18 MR. BOHN:
I'm sorry.
The question I was asked was 19 what is the relative contribution of say fire to the total 20 core damage frequency?
And there the numbers I quoted 21 somewhere from a low of 10 percent due to fire alone up to 60 22 or 70 percent.
That is relativsly overall total core damage 23 frequency.
()
24 CHAIRMAN KERR; That's what I thought.
25 MR. MICHELSON:
Wouldn't be much less, HERITAGE REPORTING CORPORATION -- (202)628-4888
31 1
DR. MOELLER:
And what portion of the fires are 2
induced by seismic events?
3 MR. LAMBRIGHT:
They haven't been addressed in any 4
of the fire PRAs to date.
That's a new issue that we will be 5
talking about in Steve's presentation.
6 As Mike briefly mentioned earlier, when we 7
requantified the fire scenarios from the four PRAs, we have a 8
new improved data base developed by Sandia's fire program 9
where we developed new initiating event frequencies.
We are 10 using the modified version of the COMPBRN III fire code now.
11 We have developed a suppression model which has incorporated a 12 significantly greater amount of events than in the original
()
13 PRAs and also we have modified the cable damagability 14 temperatures to more realistic values.
15 (Slide) 16 MP.. LAMBRIGHT:
However, some general areas which we 17 couldn't re-evaluate could have a potentially large impact on 18 core damage frequency.
19 First of all, the initiating event frequency 20 calculation method, it was substantially different, as I 21 mentioned earlier, in the Limerick PRA versus the other three 22 PRAs.
I 23 In the Limerick PRA, it took a look at initiating
()
24 events and broke them down into either electrical panels, 25 self-ignited cable fires, or transient combustible sources l
. - HERITAG E REPORTING CORPORATION -. -(202)628-4888
32 1
within a zone, and then tried to look to see if that fire
()
\\~'
2 could take out all safe shutdown methods within a fire area.
3 Whereas for the PL&G methodology, they partitioned fire 4-frequency from a large building to fire area within a building 5
using a number of ratioing factors which had large uncertainty 6
on them, error factors anywhere from four to twenty.
7 Also we didn't change or look for any new initiating 8
events, so in the case of Limerick, for instance, we wouldn't 9
have looked for loss of coolant accidents since it wasn't done 10 in the original study.
As I mentioned, the partitioning 11 method for the PL&G plants used severity ratio.
They tried to 12 assess what percentage of the fires in the data base would be
()
13 large enough to cause requisite damage for having their 14 scenario to occur, and we couldn't re-evaluate that based on 15 out data.
16 Also area ratio within fire zone, since we didn't 17 walk down the plants, we had no way of knowing what the exact 18 area of influence would be for the particular scenarios that 19 were postulated, and also finally, manual versus automatic 20 suppression, they, as I mentioned earlier, developed a 21 suppression distribution where they smeared essentially a 22 distribution over all areas of the plant, and whether or not 23 there is automatic suppression or automatic detection or
()
24 suppression or manual detection, manual suppression, it would 25 have appeared with the same probability of suppressing a fire HERITAGE REPORTING CORPORATION -- (202)628-4883
33 1
within a given timeframe.
g 2
(Slide) 3 MR. LAMBRIGHT:
Our uncertainty analysis. utilized 4
the TEMAC code, and it was performed for all four plants even 5
though Limerick and the Oconee PRAs only listed point 6
estimates.
7 (Slide) 8 MR. LAMBRIGHT:
I brought one slide to show typical 9
uncertainty events on PRA 4, and this is for the Indian Point 10 2.
First of all, the original PRA in our requantified numbers 11 using the TEMAC code, and as you can see, for the PL&G 12 methodology, there seems to be a wide spread typically between
()
13 the median and the fifth percent aisle where we seem to have 14 slightly tighter error bounds for the TEMAC code, and I talked 15 to a person who is an expert in this area, and he felt that we l
16 really couldn't comment on the behav or for the PL&G 17 methodologies because they used discrete probability 18 distribution.
He wasn't sure how many points they sampled, 19 and exactly why the behavior spread out or skewed near the low 20 end of distribution.
However, using TEMAC many times, we feel 21 that our uncertainty bounds are considerably probably more 22 representative than the original PRAs.
23 CHAIRMAN KERR:
What are these numbers that I am
()
24 seeing representing?
25 MR. LAMBRIGHT:
First of all, the original PRA HERITAGE REPORTING CORPORATION -- (202)628-4888
34 1
distributions for fire core damage frequency for various
(
2 areas, the switchgear room electrical tunnel, and the cable 3
spreading room.
4 CHAIRMAN KERR:
No.
When I say on the switchgear, 5
five, one, six, E to the minus five, that is a core melt 6
frequency or what?
7 MR. LAMBRIGHT:
Yes.
That's'the core damage 8
frequency for fires in the switchgear room.
9 CHAIRMAN KERR:
Okay.
If I add all the means up, I 10 ought to get---
11 MR. BOHN:
There will be a total slide to summarize.
12 MR. LAMBRIGHT:
For each one of the plants.
()
13 CHAIRMAN KERR:
I want to be able to check your 14 arithmetic.
15 MR. LAMBRIGHT:
Totally checkable; we have gone 16 through great detail, and Appendix Al where if there is a 17 modification to data or any other source, you could literally
(
18 just go back in and modify the numbers appropriately.
19 MR. BOHN:
That's not the total number of sequences 20 for that particular PRA, so they will not add up, but in the 1
21 report, all sequences are listed with the values which do add 22 up to the numbers you will see on the next slide.
23 CHAIRMAN KERR:
Are these the important ones?
24 MR. BOHN:
Just to show you the types of uncertainty
()
25 bounding that were calculated, and the fact that our HERITAGE REPORTING CORPORATION -- (202)628-4888
35 1
uncertainty bounding are actually less than the commercial O
2 PRAs, and partly it had to do with the method of calculation 3
and perhaps more importantly, they used engineering judgment 4
and a Basian approach to attempt to include plant-to-plant 5
variation in the commercial PRAs, rather than just data based 6
uncertainties.
7 We couldn't evaluate that because it is a judgment 8
based--there was no data on that, so in most cases, we have 9
increased the initial, the fire occurrence frequency, but 10 decreased the uncertainty about it.
11 MR. LAMBRIGHT:
If you take one specific point for 12 the PL&G methodology, for the two PRAs where they expressed 13 uncertainty dis'.ributions, Indian Point and Seabrook, even 14 although Seabrook had more fire events and more operating 15 history which you would think would shrink the uncertainty 16 bounding on initiating event frequency, in fact the initiating 17 event certainly bound increased by about two orders of 18 magnitude on the low end simply because it took a wider i
19 non-informative prior distribution in this calculation method.
20 Yes?
21 DR. SHEWMON:
The requantified is your word to PRA 22 or somebody else's work?
23 MR. LAMBRIGHT:
They are original PRA values.
()
24 Requantified distributions are our numbers before taking into 25 account Appendix R modifications.
e
---HERITAGE REPORTING CORPORATION -
(202)628-4888
36 i
1 CHAIRMAN KERR:
Could you translate requantified as
\\
,~
2 recalculated without losing very much meaning?
3 MR. BOHN:
Yes, but including the three new data 1
i 4
bases, and the COMPBRN III versus COMPBRN I code.
5 DR. MOELLER:
Most of your numbers are higher, 6
although I see I guess the bottom two are lower, but were 7
they--
8 MR. LAMBRIGHT:
Typically what occurred is except 9
for a limited case of areas, the valuen, mean values for core 10 damage frequency all increased, and the reason for that was 11 the value placed on initiating event frequency per fire given, 12 we had a more complete data base.
Those values increased, and
()
13 it seemed that time to damage for COMPBRN typically decreased 14 slightly over the initial PRA estimates, so those two factors 15 predominantly had the effect of, before taking a look at 16 Appendix R modifications, having the mean values for core 17 damage f requency for lil:e I say, 95 percent of the areas 18
- increase, a
19 (Slide) 20 CHAIRMAN KERR:
The differences are well within the 21 uncertainty bank?
22 MR. LAMBRIGHT:
Yes.
That's the case.
23 MR. BOHN:
What is significant is that certain
(')
24 sequences went down and certain sequences went up quite V
25 significantly, although as you point out, the mean value HERITAGE REPORTING CORPORATION -- (202)628-4888
_ ~. _ _ __, ___. _ _._ _ _._
37 1
differences tend to sort of average out, but if one looks at 2
individual sequences, then one finds that there are some 3
substantial differences resulting in conception of risk.
4 MR. LAMBRIGHT:
In the case of Limerick where they 5
used three different fire initiating sources, they only had 6
fire, electrical cabinet fires in our data base whereas we had 7
28, so that fact alone increased that particular initiating 8
event frequency by over a factor of five.
9 Here is a comparison of fire initiating core damage 10 frequency.
This is once again before taking into account 11 Appendix R modifications, and I would like to put point out 12 they are comparing point estimates.
()
13 CHAIRMAN KERR:
Excuse me.
I just, I am just 14 beginning to assimilate something you just said.
You in the 15 28 covered the same time period?
16 MR. LAMBRIGHT:
No.
The time period increased, but 17 the time period might have increased by a factor of 20 or 30 18 percent, whereas the amount of events increased by over a 19 factor of five.
20 CHAIRMAN KERR:
Okay.
So roughly you found about 21 five times as many fires as somebody else did?
22 MR. LAMBRIGHT:
That's right.
23 MR. BOHN:
The time spans were not the same.
24 CHAIRMAN KERR:
There is only 20 percent difference?
()
25 MR. LAMBRIGHT:
That is correct.
l l
HERITAGE REPORTING CORPORATION -- (202)628-4888
38 1
CHAIRMAN KERR:
So this could lead one to believe i
2 that the next investigator could find maybe twice as~many as 3
you did?
4 HR. LAMBRIGHT:
I couldn't draw that conclusion.
It 5
is my feeling that we have a pretty complete data base for 6
significant fires at this point.
However, one of the things 7
we would like to do is since the data base only goes through 8
the end of June 1985 is we would like to be able to 9
continually update this data base.
10 HR. MICHELSON:
That was the computerized data base 11 that the staff has put together, wasn't it?
Set up on a PC?
12 HR. BOHN:
Yes.
It is available on PW.
()
13 MR. MICHELSON:
That was not available to the 14 earlier workers?
15 MR. BOHN:
That is correct.
16 HR. MICHELSON:
They had to go about it a different 17 way.
Staff has accumulated a rather complete data base this 18 time, much more complete.
19 MR. BOHN:
We did find fires in time periods that 20 were not in the original data base developed by Garrick.
21 MR. LAMBRIGHT:
The fire occurrence frequency, as 22 Mike mentioned seems to hold about constant, so while we have 23 about two, three more years of information, there is a lot of I
()
24 events that also occurred during the time period that their 25 data base would have covered at the same time.
HERITAGE REPORTING CORPOP.ATION -- (202)628-4888
1 39 1
I would like to continue, and this is a comparison (y
t l
'/
2 of fire initiating core damage frequency for the four PRAs, 3
and as reported, that's the initial PRA value and 4
requantified, this is our requantified value before Appendix R S
modifications, and for the case for Limerick and Oconee, we 6
are comparing point estimates while for--I'm sorry.
7 For Limerick and Oconee, point estimates, while for 8
Indian Point and Seabrook we are comparing mean values.
And 9
you should notice the overall effect in all these PRAs is, 10 from the factors I mentioned earlier, is we are having an 11 increase in core damage frequency, and in the case of Limerick 12 and Indian Point, it has exceeded the 1E to the minus 4 level
(/
13 now.
14 My next slide will then compare the, these four PRAs 15 again before and after Appendix R modifications.
It is our 16 feeling for Oconee and Seabrook that all Appendix R 17 modifications were completed before the completion of the PRA.
18 Therefore, there is no value listed in this particular column.
19 But if you notice for the two plants where we can 20 make a comparison, for Indian Point, the effect of Appendix R 21 was to roughly reduce the core damage frequency by a little 22 bit over an order of magnitude, and for the case for Limerick, 23 reduced the core damage frequency by about a factor of three.
24 However, it is my feeling if we weren't constrained
()
25 by the original methodology, and took into account automatic HERITAGE REPORTING CORPORATION -- (202)628-4888
40 1
suppression systems at certain pinch points within the 2
Limerick plant, I did an analysis that would say that core 3
damage frequency would probably decrease by about another 4
factor of three, so in both cases roughly the effect of 5
Appendix R seems to be to reduce the core damage frequency due 6
to the fire by about an order of magnitude.
7 (Slide) 8 MR. LAMBRIGHT:
As you notice from the previous 9
slide, fire-induced core damage scenario frequencies are still 10 high and basically because of the followir.g three factors.
11 Fire occurrence frequencies are high, For instance, in the 12 aux building, we have fire occurrences about once every
()
13 thirteen or fourteen years; control room, 4I to the minus 3 14 per year.
In the cable spreading room, plants have cabling 15 pinch points where cables for multiple and redundant safety 16 systems are located.
Therefore, a single fire in every plant 17 that I have visited also has potential to take out a lot of 18 multiple systems at the same time.
Even random failures have 19 to be involved.
20 CHAIRMAN KERR:
The control room fires is one that 21 never occurred, is sort of a business estimate based on 22 something or other, is that right?
23 MR. LAMBRIGHT:
We have in our data base, four 24 control room fires, and I can't remember the exact amount of
{
f 25 operating years.
l l
f HERITAGE REPORTING CORPORATION -- (202)628-4888
41 1
CHAIRMAN'KERR:
The earlier question I asked I guess n('-)
2 is whether control room fires have taken you out?
The answer 3
to that is no?
4 HR. LAMBRIGHT:
What happened in the case of control 5
room fires is even although there have been four control room 6
fires, they have always been contained within the panel and 7
suppressed before they spread to an adjacent panel.
8 CHAIRMAN KERR:
And that data base includes those 9
four that were contained within?
10 HR. LAMBRIGHT:
That's right.
But as in the case of 11 PRAs we requantified, for control room, and control room was 12 only requantified in the case of Seabrook, we did it with-the
()
13 same methodology.
They just said a fire would be contained 14 within a zone of the control room, which was consistent with 15 what we had seen in our data base, so we need to exclude those 16 particular events when we requantified Seabrook.
17 MR. WARD:
John, as I recall, the grandaddy of 18 control room fires was one in Switzerland back in the early 19
'70s, right?
Is that right?
20 MR. LAMBRIGHT:
I am not familiar with that 21 particular event.
22 MR. WYLIE:
I was thinking that was Italy.
23 DR. SHEWMON:
Maybe it was southern Switzerland!
()
24 HR. WARD:
Is that in your data base?
25 MR. LAMBRIGHT:
No.
The only type of fires we have HERITAGE REPORTING CORPORATION -- (202)628-4888
42 1
in our data base are'U.S. light water reactor fires.
We don't
. f}
k/
2 have fires from any foreign plants incorporated in the data 3
base.
4 MR. WARD:
I am kind of vague on this, but I guess I 5
am surprised at that because I thought, I mean before the 6
Browns Ferry, my understanding is before the Browns Ferry 7
fire, there was a serious control room fire.
I thought it 8
was.
Maybe it wasn't.
9 MR. WYLIE:
I was thinking it was Italy.
There was 10 one in Italy.
11 MR. WARD:
Which really got most of the European 12 programs concerned about risk of fire before the NRC was.
()
13 MR. NOWLEN:
There was a large European control room 14 fire.
Ve don't have a lot of information on it, and it is not 15 included in this data base at all.
16 The data base that we are working from is strictly 17 U.S.
commercial reactors.
It doesn't include DOE production 18 facilities and it doesn't include any foreign power 19 plants--all U.S.
commercial reactors.
20 MR. WARD:
Why wouldn't you include foreign?
I mean 21 particularly LWRs which are just like ours.
22 MR. NOWLEN:
Yes.
Well, there is differences in 23 reporting requirements, and availability of information, and
()
24 consistency of reporting.
We can QA the U.S.
reactor data 25 base a little better than we could with that.
HERITAGE REPORTING CORPORATION -- (202)628-4888
43 1
MR. BOHN:
The answer is we probably should.
In Ok,)
2 addition, we are doing fire work for Savannah River, and for 3
the end reactor and especially the Savannah River, the fire 4
occurrence rates are at-least as high as what we have seen in 5
our data base, so the information that is coming out there is 6
quite consistent with what we have already, but due to the 7
scope and limitation of the, putting the fire data base 8
together, we only looked at U.S.
LWRs.
That work went up 9
there, what was it?
10 MR. NOWLEN:
Eighty-five 11 MR. BOHN:
One of our recommendations is this data 12 base ought to be kept updated and improved by NRC on an
()
13 ongoing basis both for their use and for use in individual 14 plant examinations or other utility activities.
15 MR. MICHELSON:
There have been a number of 16 interesting fires since
'85.
The recent Browns Ferry one was 17 probably the biggest.
It was quite extensive.
18 MR. BOHN:
Three panel fires occurred within the 19 past six months.
20 MR. LAMBRIGHT:
To reiterate the point once again, 21 our data base, if you compare it to the data that was used in 22 all four of these original studies, as we roughly have an 23 order of magnitude greater amount of fire occurrences than 24 were used initially.
{}
25 DR. SHEWMON:
Would you explain to me what HERITAGE REPORTING CORPORATION -- (202)628-4888
44 1
non-suppression value is before you leave that slide?
(~2
'i
\\-
2 MR. LAMBRIGHT:
Okay.
A probability of having 3
either an automatic system or the fire team suppress the fire 4
before critical damage occurs to either cabling or equipment.
5 DR. SHEWHON:
This is the non-suppression value, so 6
that means that the suppression system didn't work or--
7 MR. LAMBRIGHT:
That's right.
Either suppression 8
system didn't work, or the fire team wasn't able to get down 9
there in time to suppress the fire before the critical damage 10 for a given scenario would have occurred, and so even if it 11 suppressed fire subsequently, it still would have been 12 conservatively modeled when that particular scenario occurs.
-(
13 HR. BOHN:
Basically in all the evaluations, you 14 predict a time to damage using your fire code.
The code would 15 predict a fire in a corner of a certain pool size would cause 16 an item of cable in another corner to say be damaged in ten 17 minutes.
18 Then that is compared with the suppression model 19 which gives you the probability that it will not be put out 20 prior to ten minutes, and that's what is used in the 21 quantifying these, and as we will see, this becomes a very 22 critical part of the evaluation, and the importance of manual 23 firefighting is highlighted because of the short times that
()
24 are often available for the fire team to reach the equipment.
25 We will see that in the next presentation.
i HERITAGE RdPORTING CORPORATION -- (202)628-4888
i 45 1
CHAIRMAN KERR:
In your Oconee study, did you take 2
-into account the fact that Oconee is using armored cables?
3 MR. LAMBRIGHT:
No.
We didn't, because that wasn't 4
taken into account in the original PRA as well.
We used, 5
modeled the same exact scenarios that they used.
I 6
CHAIRMAN KERR:
Isn't that significant influence on l
7 potential or actual fire damage?
j 8
MR. LAMBRIGHT:
However, I would like to make 9
another point about Oconee.
Oconee modeled it such that the I
10 fire had to spread through like cix or seven trays stacked one 11 on top of each other to have critical damage occur.
They 12 didn't cable trace to know where the critical components were
()
13 within the stacks themselves, so the time of the damage also i
14 could have been significantly less, 15 CHAIRMAN KERR:
You or the original PRA did not take 16 into account that the cables were armored?
17 HR. LAMBRIGHT:
That is correct.
18 MR. MICHELSON:
They were just trying to requantify 19 under these few changes.
20 MR. LAMBRIGHT:
Under the original.
21 CHAIRMAN KERR:
Get some idea of fire risk damage?
22 MR. MICHELSON:
No, no.
They were sticking with the 23 original models except requantifying according to the earlier
()
24 slide.
25 DR. SHEWMON:
You are not nonsense--we don't want to HERITAGE REPORTING CORP 0kATION -- (202)628-4888
46 1
change them, keep getting nonsense out.
\\}'
l 2
MR. MICHELSON:
The amount of the resources 3
available for the work did not permit a complete 4
re-examination of the plant.
It has to be kept in mind when 5
you look at any of these numbers.
6 MR. LAMBRIGHT:
I would like to conclude this 7
portion of the talk by stating that the requantified core 8
damage frequency, as I showed earlier, increased for all four 9
PRAs, and for Indian Point 2 and Limerick, before taking a 10 look at Appendix R modifications, was greater than ten to the 11 minus 4 per year.
And even with Appendix R modifications, the 10 fire-induced core damage frequency still is an important f
()
13 contributor.
Large uncertainties still exist.
14 However, if you could apply a more consistent 15 methodology not constrained by the original analysis 16 assumption, we feel that these uncertainties could be greatly 17 reduced, and typically the uncertainties that seem to dominate 18 were the ones in the initiating event frequencies, and the 19 partitioning methods.
20 CHAIRMAN KERR:
I'm sorry.
I a.a missed your last 21 statement.
You have said if you could apply--
22 MR. LAMBRIGHT:
If could you apply standardized 23 methodology to all four of these PRAs, the large uncertainties
()
24 that still remain could be significantly reduced.
25 MR. BOHN:
With regard to the last question, we have HERITAGE REPORTING CORPORATION -- (202)628-4888
in 47 l'
to plead ignorance a little bit because the expert on fire
/
(
2 code is not here.
I apologize.
However, we did use the 3
original fire codes influent.
Fire damage temperatures were 4
not inconsistent with our current research, and so which 5
actual values were used and whether they were modified to 6
incorporate armored cable where appropriate, I can't answer 7
that question.
You would have to, I would have to get back 8
with the person that did those, but the point I want to make 9
is we used exactly what was in the original fire PRA and 10 that's a fairly recent PRA also.
One would assume they took 11 advantage of that in the original PRA when they could.
12 Now additionally, cable can cook even though it is
()
13 encased or armored if it is actually in flames, so they may 14 have, they may have made that assumption in the original PRA 15 also.
16 HR. WYLIE:
Well, it depends a lot on what you 17 assume is a fire source.
And if it is oil, then you can cook 18 things.
19 MR. BOHN:
Precisely.
20 MR. LAMBRIGHT:
in the case of this particular PRA, 21 oil was the source of fire.
22 MR. WYLIE:
I think that won't happen with the 23 armored cable.
It will not spread.
Now how much effect that
()
24 has on your PRA, I don't know.
25 HR. LAMBRIGHT:
In the case of this particular HERITAGE REPORTING CORPORATION -- (202)628-4888
48 1
scenario, fire had to spread up through seven vertical cable 2
trays to cause the damages.
3 MR. WYLIE:
You would have to have a fire source all 4
the way up because the armored cable won't spread your fire.
5 MR. BOHN:
Or significant source under the cable.
6 MR. LAMBRIGHT:
That's right.
7 MR. BOHN:
Again, we followed the original PRA and 8
their assumptions in that regard.
9 MR. LAMBRIGHT:
That's right.
10 MR. WYLIE:
Other thing, with armored cable, you 11 wouldn't initiate a fire overloads or short.
12 MR. LAMBRIGHT:
Their methodology, they took, they
()
13 caid they took a look at self-ignited cable fires, found them 14 to be insignificant contributors compared to transient sources 15 located approximately directly underneath the cable trays.
16 MR. WYLIE:
What you are saying, self-ignited cable 17 fires are insignificant regardless of whether they are armored i
18 or unarmored?
19 MR. BOHN:
That was the PL&G assumption.
20 MR. WYLIE:
It is not yours?
21 MR. BOHN:
Not necessarily, no.
22 MR. LAMBRIGHT:
Is there any further questions on 23 Tasks 1 and 2?
()
24 I will turn it over to Steve now.
25 CHAIRMAN KERR:
According to my agenda, we are due HERITAGE REPORTING CORPORATION -- (202)628-4888
49 1
for a break at this point.
I would like to follow the agenda f) x-2 and so we will have a--I'm sorry.
3 MR. MICHELSON:
A break was scheduled at 10:30.
4 CHAIRMAN KERR:
It was scheduled at 10:45.
I 5
misread my agenda.
6 MR. MICHELSON:
I think we are right on time.
7 CHAIRMAN KERR:
Please continue.
8 MR. NOWLEN:
We have an extra minute, so this 9
question has come up repeatedly.
This is not a part of the 10 presentation that I have prepared, but since it has come up, I 11 will show it to you.
12 This data, by the way, is presented from the March
()
13 9th subcommittee meeting of the ACRS auxiliary systems, so if 14 you are interested in it, this is the fire data base that wo 15 were using, and it lists the number of events, and I have 16 separated them out a little bit here as to the state of the 17 plant when the fire occurred, and as you see, it is, it bounds 18 up and down a little bit.
Some of these can be correlated 19 with regulatory actions, like implementation of Appendix R 20 Browns Ferry fire, things like that, but basically in the 21 range of 10 to 20 fires a year for the industry, so just as an 22 aside.
23 MR. WARD:
I thought Mike said earlier that there
()
24 was not a trend?
25 MR. NOWLEN:
Well, I'm not willing to bet on this HERITAGE REPORTING CORPORATION -- (202)628-4888
50 1
trend.
This is right at the tail end of the data base.
I 2
think if we updated this data base, you would find these 3
bounce up there as well.
I mean--
4 MR. WARD:
Eighty-three and '84 might be incomplete?
5 MR. NOWLEN:
IT is possible, yes.
Data base goes 6
through June of
'85, and it is hard to say when you get to the 7
tail end how complete your findings are.
8 CHAIRMAN KERR:
I was told the data base probably 9
was accurate.
The investigator wouldn't double the previous 10 number of fires.
11 MR. NOWLEN:
Back here I don't think so.
You are 12 spreading out from '84.
If you identified a couple more, that
()
13 bounce it up here, that is certainly not going to double your 14 frequency overall.
15 DR. SIESS:
I don't understand.
You are using a 16 data base that you know to be incomplete or that you are 17 fairly sure it is incomplete and you don't know?
Just what is 18 this?
19 MR. NOWLEN:
We consider the data base to be the 20 most complete data base available.
Now whether or not there i
21 might be, not be an actual couple fires that aren't included 22 here, that is a possibility, and I'm, and the other thing is I 6
23 am not willing to bet that this trend continues on.
You know,
()
24 it bounces.
25 The question was is there a trend for decreasing (202)628-4938 HERITAGE REPORTING CORPORATIOft
51 1
fire?
And I am saying on the base of this, I would not 2
describe that as a trend.
3 MR. WARD:
I am sure puzzled that there is a lot 4
more plants in those later years.
Is that number of, simple 5
number of events?
6 MR. NOWLEN:
Yes, per year, for calendar year.
7 MR. WARD:
I don't know.
If you divide that by the 8
number of plants, it looks to me like it might be kind of flat 9
up until '82 or something.
Then this is a dramatic decrease.
10 HR. NOWLEN:
That is Appendix R.
Now the 11 statistical significance of this is yet to be demonstrated I 12 think.
That's one of the things we would like to do is update
()
13 this for
'85,
'86,
'87 to get a feel for--
14 DR. SIESS:
What evidence would you have to have to 15 convince you there was a trend?
16 MR. BOHN:
There is historical things going on in 17 there.
The actual task which we were funded to develop, this 18 data base only went on through a certain period of that, and 19 then when we took over, there was some additional work on more 20 limited basis not necessarily funded to look for additional 4
i 21 control room fires and certain areas where we thought we j
i 22 needed additional information.
That work was not complete.
23 But John, what Steve is saying in terms of completeness, what 24 we are really saying, this data base is an order of magnitude
{}
25 greater than anything that has existed before in terms of the HERITAGE REPORTING CORPORATION -- (202)628-4888
52 1
number of events, and the reporting toward the higher end is n
2 not necessarily complete.
3' DR. SIESS:
You said this is ten times as many 4
events as you'had before?
5 MR. MICHELSON:
In the data base.
6 MR. BOHN:
They had roughly 60.
7 DR. SIESS:
The whole volume of this thing is up by 8
a factor of ten over what you had before?
i 9
MR. BOHN:
Went from 60 up to about 368.
It is not 10 ten.
Maybe a factor of five.
11 MR. LAMBRIGHT:
I went through--
12 DR. SIESS:
Excuse me.
Let's get one answer at a
()
13 time.
You go off faster than I can absorb your answer.
[
1 14 The volume under those histograms is measure of 15 total number of events and that's six or eight or ten times 16 higher than you had before.
And they all came early, or was
~
17 the distribution changed as you got the additional 300 data 18 points?
i 19 MR. NOWLEN:
I think the question is what was the 20 timeframe of the original data base?
Ir that we are all 68, 21 the first 68 on the slide here?
22 DR. SIESS:
I was trying to discuss the question
[
23 that had been raised about trends which means I sort of am t
()
I 24 looking at the distribution with time.
25 Now did that change when you got the additional 3007 HERITAGE REPORTING CORPORATION -- (202)628-4888
53 1
MR. NOWLEN:
I am not certain I can answer that 2
question.
3 DR. SIESS:
You certainly would expect as the number 4
of plants grew, if nothing, everything else stayed constant, 5
the number of events per year would go up.
If the number of 6
plants stayed constant, you would expect that to level off.
7 Some people got smarter, we would expect it to go down.
And 8
now there is also questions on, you know, the fact that Browns 9
Ferry, and you suddenly have a jump in the reported fires, so 10 there are other things going on here as well.
Appendix R 11 implementation--
12 MR. BOHN:
I think at the lower end you see the
(
13 reporting requirements were significantly less, and that i
14 impacts it somewhat, too.
15 In addition, this came from both LERs and from 16 insurance claims reporting.
Now we should see something, you 17 know, relative to Appendix R.
However, I think we do not feel 18 that at the upper end that we have exhausted--for example, 19 when John went back to look for particular fires in particular 20 areas, he did not go back to the insurance sources as was done 21 in the original study, so we did not think that the '82,
'84 22 is necessarily complete.
We just saw, for example, the 23 original data base, there were no control room fires, yet we
()
24 heard of some, so John went back to find those and he found 25 certain other fires, too, but it was not complete.
And so--
)
HERITAGE REPORTING CORPORATION -- (202)628-4888 i
e
-l 54 j
1 DR. SIESS:
Did you use reported fires as the base
((../
2 for your analyses, or did you try to establish some 3
relationship between reported fires and real fires?
4 MR. BOHN:
That is correct.
Within the LER data 5
base, it wac, you know, fires that could potentially affect 6
the safety system.
Usually in the insurance fires, they are 7
fairly significant fires or they are not even reported.
8 DR. SIESS:
What about the unreported fires?
What 9
do you do about them?
10 MR. BOHN:
We don't know about those.
They are.
11 there are, there are many small trash can fires that are not, 12 don't have to be reported, will not be reported for insurance
()
13 purposes.
14 DR. SIESS:
Are they important for risk purposes?
15 MR. BOHN:
That's problematic; probably not.
16 CHAIRMAN KERR:
Shall we continue?
17 HR. NOWLEN:
Yes.
Okay.
The bulk of my talk is 18 overview and the conclusions on new issues investigations.
19 (Slide)
F 20 MR. NOWLEN:
The new issues that we were 21 investigating, and I will talk about each one of these in 22 turn, are listed here.
There were some questions as to how we 23 arrived at this list.
24 We started with an initial listing of, by Sandia
(}
t 25 that grew out of the fire protection research program, and HERITAGE REPORTING CORPOXATION -- (?J21628-4888
55 1
went through a procedure where we looked at the experience
~
2 base in terms of information notices, and the data base of 3
reported fires.
4 We also went around the country polling various 5
experts in fire protection research, plant design, and that 6
kind of thing, and we ended up with this as a list of issues 7
that we felt were worthy of investigation, and had basic 8
concurrence by those we polled.
9 (Slide) 10 MR. NOWLEN:
I will give you sort of the bottom line 11 risk impact of each of the issues and then we will go into 12 each one in a little bit more detail.
()
13 What we looked at is for each issue we looked at two 14 factors--the potential increase in the core damage frequency, 15 and the applicability, whether it be very plant-specific or 16 whether it is a problem which would be common to more plants 17 than one or two.
18 With manual firefighting effectiveness, we feel that 19 this is basically an order of magnitude kind of effect, and i
n 20 then it is a relatively generic issue, that the Appendix R 21 guidelines reasonably lacked--well, I shouldn't use that word.
22 They specify a certain level of training for the fire brigade 23 and that level of training is relatively low, and so many i
24 plants may display that kind of an impact on their risk
(}
i 25 estimates if they look at actual responsiveness of their fire HERITAGE REPORTING CORPORATION -- (202)628-4888
7 56 1
brigade.
O 2
CHAIRilAN KERR:
Now let me make sure I understand.
3 You are saying in your view, the existing PRAs estimate core 4
melt frequency of say ten to the minus 4.
If one really knew 5
how ineffective manual firefighting is, it would jump to ten 6
to the minus 3?
Is that the message we should be getting?
7 MR. NOWLEN:
I'm not sure I would pick those 8
particular numbers.
9 CHAIRMAN KERR:
Pick some numbers.
Then tell me how 10 I em supposed to interpret this.
11 MR. LAMBRIGHT:
I can interpret that for you.
I 12 took a look at every scenario from the thirteen of them for
)
13 all four original PRAs, and what the conclusion of my work was 14 that even in a timeframe as little as 20 minutes order of 15 magnitude increase in core damage frequency for any particular 16 fire area, so the timing is what is really critical here.
17 CHAIRMAN KERR:
I am not trying to get the details 18 of how you reached a conclusion.
What I see is a ten and I 19 want to know what it means.
And it says increase in CD.
Now 20 if that means Certificate of Deposit, it is one thing, but if 21 it means core damage frequency--
22 HR. BOHN:
Roughly, yes, that is what that means.
23 We have evaluated certain specific sequences in evaluating
()
24 that, and we have shown that certain important dominant 25 sequences did go up by factors of ten to 30 let's say, and if HERITAGE REPORTING CORPORATION -- (202)628-4888
i 57 i
1 those are significant parts of the PRA, which they typically 2
were, the fire PRA, then it would imply'an increase between 3
ten, 30 and the total core damage frequency.
4 CHAIRMAN KERR:
I am not trying to disagree.
I just 5
want to understand what you meant.
a.
6 HR. BOHN:
Order of magnitude effect sort of r
7 assessment, so you have to look at it in that respect.
If you 8
say order of magnitude, we are talking about somewhere between 9
ter and 30 or something like that, increase.
10 CHAIRMAN KERR:
Order to me means, indeed it says l
11 ten, so all right.
Now if I also take into account control i
12 systems interactions, does that mean I have got another factor
()
13 of ten?
14 HR. BOHN:
If the worst possible case, if everything 15 stacked up in the worst way, then that's what it would imply.
16 However, that's not, of course, the way things work 17 in real life.
I mean real life is--
18 CHAIRMAN KERP:
I want to know, I want to know how 19 to interpret the conclusicus, and I see up there potentially 20 effective judgment, more maaual ineffectiveness, potentially a 21 control system interaction another factor of ten.
Now I could 22 interpret it to mean if I get ten to the minus 4, thereabouts, 23 and take these into account, which is as big as ten to the c
()
24 minus 2.
25 HR. NOWLEN:
No.
That is not--
HERITAGE REPORTING CORPORATION -- (202)628-4888
58 1
HR. BOHN:
If it stacks up enough, then we have two O
2 core melts a year, but that is not--let me interpret this 3
properly for you.
4 This says that we have evaluated the effect in 5
several, one or more or many dominant sequences in the PRA.
6 That is, we have looked at, and we have found that these 7
dominant frequencies, dominant sequence frequencies, did go up 8
by factors greater than ten.
9 Now that is, that does not mean that all sequences 10 will necessarily go up by a factor of ten, but a number of 11 dominant sequences did, so it is potentially possible, yes, 12 that the core damage frequency can go up by a factor of ten or
)
13
- gredter, t
14 However, these are often very accident sequence 15 specific, and plant specific, se it is not likely, and in fact 16 we did not see this--it is not likely that they will all stack 17 up in the worst possible case.
18 DR. LEWIS:
Does the word "potential" mean worst 19 possib'e?
What does the word potential mean?
20 HR. BOHN:
Potential means we have re-evaluated t
21 sequences and found that if this case, say control system I
22 interactions occurred, that situation could occur, we L
23 calculated sequence frequencies that went out by say a factor
()
24 of ten or greater.
25 DR. LEWIS:
You say if it occurs.
HERITAGE REPORTING CORPORATION -- (202)628-4888
59 O-
'l MR. BOHN:
That is correct.
I 2
DR. LEWIS:
You worst case the control system 3
failure, and then calculate the increase so it is not a number 4
we should take as a probabalistic number because you have 5
started with a worst case assumption?
6 MR. BOHN:
Let's look at the second one, control 7
systems interaction.
We examined that by looking at the 8
LaSalle plant for which we have detailed control circuit 9
information.
We postulated a fire in one of the control 10 cabinets initiated by a short.
We evaluated the effect of 11 that fire damaging all the circuitry within that particular 12 cabinet.
()
13 DR. LEWIS:
All the circuitry?
14 MR. BOHN:
All the circuitry, but just in one l
3 15 cabinet, and we found that that, in conjunction with certain I
16 random unavailabilities, raised the core damage frequency by a 17 magnit ce factor of ten or more.
18 DR. LEWIS:
I am just worried about the bottom 19 number being interpreted as a probabalistic number when there 20 has been mixed into the calculation a worst case assumption 21 whose probability is not included because all too many people 22 will not go through this sequence of logic that you have gone 23 through, but they will read it as a genuine probabalistic
()
24 estimate, which apparently it is not.
25 MR. WARD:
I didn't understand.
I thought it was--
HERITAGE REPORTING CORPORATION -- (202)628-4888
I 60 l
1 MR. BOHN:
It is genuine in the sense that we have
,~
2 computed the sequence frequencies that go up by those 3
estimates factors of ten to 30, for these situations that we 4
have postulated there.
5 CHAIRMAN KERR:
I have got four orders of magnitude 6
on that slide.
7 MR. NOWLEN:
I don't believe this is multiplicative.
8 They don't stack up that way.
9 MR. LAMBRIGHT:
I would like to make a comment on 10 why they don't stack up and he then--on control systems 11 interaction.
12 First of all, in the case of the LaSalle study, we sf ~\\
)
13 calculated a particular fire sequence for one subpanel 'f a 14 control panel in the control room that led to a core damage 15 frequency of roughly lE to the minus 5 per year, which was 16 comparable with the core damage frequency from all other 17 internal events at LaSalle.
18 The only conservatism in the entire analysis was the 19 fact that things could fail in the worst case given a fire in 20 the cabinet.
The fire was not allowed to spread, and in the 21 second case for control system interaction, we did a 22 sensitivity analysis on Seabrook, and Indian Point, and 23 Oconee, and we found that if we assumed all six shutdown
()
24 methods failed, the effect could be getting core damage v
25 frequencies on the order of lE to the minus 4 or greater which HERITAGE REPORTING CORPORATION -- (202)628-4888
h 61 1
is an order of magnitude greater than the core-damage 2
frequency from including the Appendix R modifications.
3 Typically, control system interaction issue would 4
not be coupled with manual firefighting, and the PRAs, it took 5
a 1cok at control room.
They did not allow credit for manual 6
firefighting, and would assume a certain area within the 7
control room failed, and the crux of it was whether the remote 8
shutdown panel remained operable or not, or what other 9
additional random failures were required.
So there is a--
10 CHAIRMAN KERR:
I look at it and I see on the 11 right-hand side a generic applicability and on the left-hand 12 side, and I see four orders of magnitude for things that
()
13 aren't directly related and I don't know what other conclusion 14 I can draw _than that there could be a four magnitude 15 difference if one took these into account as compared to the, 16 what the typical PRA would calculate the core damage 17 frequency.
18 HR. LAMBRIGHT:
Let me give you another example.
l 19 HR. BOHN:
The one is--No.
1, they are interrelated, i
20 very much so.
Manual firefighting is closely related to fire l
21 keyed adequacy.
They are intimately related.
Similarly, as 22 John mentioned, the control systeras interaction would not the l
I l
23 sequence that were effective since they don't take into
()
24 account manual suppression, do not, would not be effective, so l
25 these things really don't stack up in our opinion.
_ _ _-.,_ HERITAGE REPORTI!1G CORPORAT10N -- (202)628-4888
~
~
62 1
If you wanted my personal opinion of how they should O
2 be viewed, it looks to us like probably in any given plant, i
3 one of these effects would probably be there on some of its 4
dominant sequences, so the net effect would probably be to 5
push the core damage frequency due to fire up by roughly an
[
6 order of magnitude.
7 DR. LEWIS:
The term "could probably be" there is 8
not a term one should use in doing a PRA because I can go into 9
a PRA at any stage where there is any linguist assigned a 10 probability of ten to the minus 2, say just to invent a 11 number, and say gee, that could happen, and therefore, if it i
12 happened, the bottom line would be increased by a factor of a
()
13 hundred, but that would be an incorrect thing to do in a PRA.
14 Of course, any probability in ; PRA reflects a 15 probability, and the event described could happen, but the way 16 you do a PRA is to assign it the probability of it happening, l
17 and then carry through the calculation, so I think what has 3
i 18 confused Professor Lerr and me at the same time is to take a
j 19 these things which I assume you have done well, and described 20 things that could indeed happen, and then presenting the data 21 as if they change the probability.
22 They don't change a probability, if I understand 1
23 you.
What they do is describe something that could happen j
3 t
()
24 which you have looked into more carefully than people have 25 looked into it before.
1 HERITA GE REPORTING CORPORATION -- (202)628-4888
b 63 1
MR. FLACK:
I think the description.of what they are n()
2 doing is simply a sensitivity study.
3 HR. MICHELSON:
One issue at a time.
4~
MR. BOHN:
Let's talk about what you mentioned.
And 5
in the case of manual firefighting effectiveness, we have now 6
in our suppression model data that shows that for many areas j
7 in a plant, your fire brigade personnel are not going to 8
effectively get down there within 15 or 20 minutes, that is, 9
there is a large probability that those areas will not be
.10 gotten to in that period of time.
11 However, our code calculations show that for those 12 areas, the fire, the fire will be--well, they have to get down
()
13 there between ten and 15 minutes, in other words, a shorter 14 period of time, and the assumption--so given you have that 15 situation, we know there are areas like that in every plant, 16 and the utility personnel that we discussed this with, you 17 know, we said our data tends to show you have got to get there 18 in about 15 minutes, and the fire brigade personnel say och, 19 that's pretty quick, we are not sure we can do that, so we are 20 saying that here, when one identifies, and we did identify 21 sequences for which you had to get down there in five minutes, 22 but our review of manual effectiveness says they probably 4
23 wouldn't get there.
It pushes the frequency up, so I am not 24 so those cases do occur virtually in every plant.
[
()
25 DR. LEWIS:
I understand what you are saying and I HERITAGE REPORTING CORPORATION -- (202)628-4888
64 1
think that if you have discovered that in many cases the 2
probability of putting out the fire has been overestimated, 3
and therefore the probability that field spread has been 4
underestimated because the brigade won't get there in time, 5
then the thing to do is not list it as a potential, but say by 6
golly,-the probability has been underestimated before, it 7
should be increased to whatever it is that you found by 8
looking carefully, and not show it as potential, bv.t show it 9
as an actual increase of the core melt probability.
10 And if you show it--let me finish.
And if you show 11 it as an absolute--I let you finish.
If you show it as a real 12 increase in core melt probability which is what you have just
()
13 told me it is, then by golly, Frofessor Kerr's concern is a 4
14 very real one because these things are going to be added up to 15 each other in which case, you know, I'm worried, and I think 16 he is worried.
17 HR. BCHN:
Well, we did do exactly that 'or a number 18 of sequences, but we did not do all of the effects and all--we 19 did them one at a time, sequentially for a number of different 20 sequences.
We did not attempt to include all these effects in 21 the requantification of the PRA, and that was a point I wanted 22 to make on the earlier slide, so we have put this up just to 23 give you an idea of the effects that we had calculated.
24 DR. LEWIS:
In that case, you should show this as
()
25 not potential.
The word potential is a bad word to use in 1,
1 HERITAGE REPORTING CORPORATION -- (202)628-4888
]
65 r
1 connection with a PRA anyway--show this as an actual increase i
2 in the contribution to core melt probability of those 3
sequences which you have calculated, because from what I 4
4 understand, that's what you have done.
And then people will 5
ask the next question, which is how big is the contribution of 6
those sequences to the overall core melt probability?
Because 7
that's the next step in doing and amending the PRA.
8 MR. WARD:
He already answered that.
He said about 9
a factor of ten it was his judgment.
10 DR. LEWIS:
For these sequences.
4 11 MR. WARD:
No.
On the over--
12 MR. BOHN:
You are correct.
That would be a much O
(/
13 better way to present it, and the report does contain 14 individual results, and also as Dr. rard mentioned, I did give 15 you my impression of what, if you looked at all of these in 16 any particular plant, what its net effect would be.
That was 17 the number that I drew up, probably pushed any particular PRA 5
18 up at least an order of magnitude.
I' 19 DR. LEWIS:
Okay.
So you really do believe that the j
20 core melt probability that we are all carrying in our heads 21 should be increased by a factor of ten?
22 MR. BOHN:
That is correct.
That is my personal 6
23 belief.
()
24 MR. MICHELSON:
Let's move on.
This is impo.can?
j 25 and we have got a number cf equally important items in this i
HERITAGE REFORTING CORPORATION -- (202)628-4888
66 1
particular section.
We have used up nearly, we have used up O
2 our 25 minutes already.
3 MR. WARD:
That's an earthshaking conclusion.
i 4
DR. LEWIS:
It most certainly is.
5 MR. MICHELSON:
We need to discuss a little more the 1
6 composition of each of these.
7 KR. NOWLEN:
Let me move on to the specific 8
conclusions.
9 From manual firefighting, as Mike mentioned, we 10 found scenarios where fires must be suppressed in 5 to 20 11 minutes to prevent damage, and many areas will exist within a 12 plant in which it was not exepeted that all fire brigades will
()
13 reach that area within that timeframe.
14 We also found that fire brigade size and training 15 practices in the industry varied very widely, and that the 16 knowledge of critical plant systems is crucial in that they do 17 the right thing when they arrive at the fire and they don't 18 spray the wrong equipment.
t 19 We also know that dense smoke is going to hamper 20 manual firefighting efforts.
This is shown by both testing r
21 and experience, and that fire suppressed by halon and CO2 are 22 likely to re-ignite if the fire brigade shows up and starts 23 violating the enclosure, opening the barriers and what not, so
()
24 again we feel that this is one issue that is fairly important.
25 (Slide)
I HERITAGE REPORTING CORPORATION -- (202)628-4888
67 t
1 HR. NOWLEN:
With respect to the fire codes and the 2
analytical tools, we found that the accurate calculation of 3
time to damsge is critical for assessing these fire response 4
times.
These two guys are tied real close together.
The time 5
calculated by the code to damage is the key to the timing of 6
sequences, and tying that to the suppression probability i
7 within that timeframe becomes a key to the scenarios'
+
8 significance.
5 9
HR. MICHELSON:
Fire code you are talking about is 10 not the fire code?
This is the calculational code?
11 HR. NOWLEN:
That is analytical model of fire drill; 12 this is not a regulation.
13 None of the existing fire models have been 14 benchmarked against actual enclosure fire tests.
There is 15 data available from the fire protection research program, but 16 it has not been processed due to budgetary constraints, and 17 hence these codes have no data against which to compare 18 themselves to see how accurately they predict fire 19 environments.
[
20 As a part of this study, we took the COMPBRN III l
21 model, made a number of modifications to it in the Task 1 and l
22 2 work on requantifying the PRAs.
We ran into a number of 23 difficulties with application of the code.
There are some
()
24 internal inconsistencies and we found one coding error.
We 1
i 25 went in and to the extent possible, corrected as many of those l
(202)628-4868
-- HERIT. AGE REPORTING CORPORATION --.
I 68 1
problems as we could within the scope of this study, and we O
2 did remove a number of the obvious deficiencies, but in our 3
opinion, the code, even with our modifications, is sometimes 4
conservative and sometimes not conservative.
You need to know 5
some of the tricks of the code in order to get it to give you 6
what would be considered reasonable results based on 7
engineering judgment, so there are.
8 CHAIRMAN KERR:
You talk about lack, when you talk 9
about conservatism or lack of it, you mean in terms of your 10 judgment of what really happens?
11 MR. NOWLEN:
Yes.
Well, I can give you an example.
I I
12 In the original code, one of the things that it doesn't do is
()
13 model heat transfer in the flame, so you can have, for 14 instance, a cable tray immersed in flames for several minutes 15 until your pilot fire goes out, and that cable will never be 16 damaged as predicted by the code.
That's one of the things 17 that we changed in here is that we modified that to include 18 heat transfer mechanisms within the flame.
It is sort of a r
19 gross conservatism.
20 DR. SIESS:
Obvious deficiencies is not--I'm sorry.
21 You have got codes that haven't been benchmarked against real i
22 life?
23 HR. NOWLE';:
That is correct.
()
24 DR. SIESS:
Start with that statement.
Now that 25 doesn't constitute an obvious deficiency?
Not comparing it HERITAGE REPORTING CORPORATION -- (202)628-4888
69 1
with real cenditions, because you don't have any real O
2 conditions to compare it, you are comparing it, what the j
3 deficiencies, a judgment?
4 HR. NOWLEN:
Things like having the cable tray 5
immersed in flames for several minutes and not being damaged.
6 DR. SIESS:
Why was that in there in the first place 7
if that is so obvious?
Is this incompetence of the people 8
that developed the code in the first place?
And if-so, who 9
were they and who paid for it?
10 MR. BOHN:
That particular example is one that in 11 which they did not model vertical, exactly vertical flame heat i
12 transfer, so to use the code, what we found in, the
()
13 practitioners found is you had to locate your fire source some i
14 distance away, a foot away from right underneath the cable.
15 In that case you have got the lateral heat transfer and then t
t 16 it did ignite correctly, but the fuel sources modeled as the i
17 point and so mathematically under the cable was also modeled f
18 as a point, it would not ignite it.
f 19 Now that is not correct, clearly not correct, but 5
20 people that ran this thing learned this thing.
21 DR. SIESS:
Who did the incorrect analysis?
Was I
22 this something that NRC paid for?
4 l
r 23 MR. BOHN:
No, sir.
It was develop d at UCLA in
()
24 order to have a simple fast running fire code to look at to 23 look at risk studies in effect, and they have been, made f
f i
HERITAGE REPORTING CORPORATION -- (202)628-4888
70 l
1 improvements, but our understanding is that it has been, the O
2 improvements have been made by successive people who perhaps 3
made errors in some of the original coding, so in our 4
evaluation, we tried to take as many of the those out as we 5
could.
6
';. SIESS:
I am not talking about coding errors.
I 7
am talking about obvious deficiencies, to use the term that is 8
on the slide.
9 HR. BOHN:
COMPBRN I had a number of obvious 10 deficiencies.
11 DR. SIESS:
You wanted something that was simple l
12 whether it was right or not?
()
13 MR. NOWLEN:
It was never intended to be an answer 14 to all codes.
Unfortunately, it is the only one available, 15 the only fire code available which addresses the unique 16 situation of the nuclear power plant fire, i.e.,
cable fire 17 growth and damage to equipment not directly involved in the 18 fire, so it has been used extensively, but the developers of l
19 the codes themselves will admit it is not perfect, and that it 20 contains a number of shortcomings.
1 21 HR. WARD:
The guy who developed that code thought i
22 he was a hero and you are turning him into a villain.
{
23 DR. SIESS:
I quit.
()
24 MR. BOHN:
The problems with COMPBRN I have been t
25 recognized by the people that have done the PRAs we evaluated, HERITAGE REPORTING CORPORATION -- (202)628-4888
71 1
and it is probably slipped by you on one of the earlier O
2 slides, but what they did was they assumed that COMPBRN had a 3
factor of two conservatism in it, and so they actually J
4 modified the COMPBRN results by this factor of two because.
l S
they recognized its deficiencies.
6 Now that is not, obviously not the right way to go 7
about it.
The right way is to fix it.
Nonetheless, these 8
have been recognized, and you know, the people that did these t
9 studies were, you know, intelligent people and they used the 10 code as intelligently as they could.
11 CHAIRMAN KERR:
Let's consider that sufficient then, i
12 Mr. Lewis?
(
13 DR. LEWIS:
I just want to react with the familiar i
14 speech about the use of the words conservative and not 15 conservative when you are dealing with a PRA because the, I 16 think Dr. Seiss has been using the word deficient and you used 17 the word conservative.
18 The function of a PRA is to do something 19 realistically, not conservatively, and one of the standing i
20 deficiencies in many PRAs is that the people who do it are 21 simply brought up to do things conservatively because they are 22 interested in safety.
l I
23 The purpose of a PRA is not to make things safe, but 1
()
24 to estimate how safe they actually are, not to estimate on the 25 conservative side, so there is nothing wrong with people using i
HERITAGE REPORTING CORPORATION -- (202)628-4888
72 1
a code if they believe that the calculation has been done 2
conservatively, to take out a factor of two conservatively if l
e 3
theit ee*,imate is correct because a PRA calculation should not 4
be done conservatively.
It is wrong to do it conservatively.
5 And that's a familiar speech, but it is a disease that keeps 6
coming up, and it is not directed particularly at you, but I 7
am worried about the, about the picture.
8 MR. MICHELSON:
Let's proceed.
We will never get 9
through all of this today.
10 DR. LEWIS:
I think that is an important point.
11 MR. MICHELSON:
Now we have got to proceed, f
J 12 HR. NOWLEN:
One other aspect that is closely tied
()
13 to the analytical tools is the availability of component and 14 cable fire damage information, and there is very, very little 15 data available in the community on where components and cables
(
16 will be damaged by fire, what sort of mechanisms will cause
[
17 the damage, what damage states they will go into, and what the 18 threshold for that damage is, so all or this information ties f
19 into the code and that code is telling you how long it takes 20 to damage things, and if you don't know about how things l
21 damage, this becomes a large uncertainty.
[
f 22 (Slide) 23 HR. NOWLEN:
With respect to the control systems 24 interaction issue, we looked at some scenarios in which we
()
i 25 assumed that electrical independence was maintained between HERITAGE REPORTING CORPORATION -- (202)628-4888
_ _-____ __- _. _ ___.._ _ _, ~ _ _ __.
)
73 1
the control room and'the remote shutdown panel, and yet we
, O 2
found subtle interactions involving the control room panel 3
fire which can cause a significant core damage contribution.
4 4
These typically involved fire-induced damage coupled with 5
random errors, but they still filtered to the top, and John 6
mentioned the LaSalle scenario in which the one control system 7
interaction scenario examined was on a par with all the 8
internal events core damage initiators.
j 9
We also found it is very difficult to assure that 10 you really do have electrical independence between the remote 11 shutdown panel and the control room.
As I say, we simply l
12 assumed that it did exist.
)
13 Hany remote shutdown panels have limited 14 capabilities and lack of indication that remote shutdown panel l
i 15 may increase risk.
In the case of LaSalle, there was one 16 particular indication on the panel which resulted in an order 17 of magnitude decrease in the significance of the scenario, j
18 Without that one indication, the recovery actions allowed in
[
l 19 the PRA by the operators would be significantly reduced as an 20 operator would not have an indication of the problem, and the 21 chances of his going off and correcting the problem would be j
22 significantly lower, so that the fact is that remote shutdown 23 panels vary widely and that there are certain indications that
()
24 can be very important on the panel.
25 (Slide) i HERITAGE REPORTING CORPORATION -- (202)628-4888
74 1
HR. NOWLEN:
On the total environment survivability, O
2 we looked at a couple of aspects of that.
The first one was 3
the smoke control issue.
4 We have got some tests that demonstrate that within 5
five to eight minute timeframe, you are going to lose 6
visibility in a typical nuclear plant enclosure with typical 7
ventilation conditions.
These are sort of control room-sized 8
enclosures with typical control room type ventilation as high 9
as ten room air changes per hour.
Control room ventilations 10 are not effective in preventing this loss of visibility.
11 We also know, of course, that the smoke will hamper 12 firefighting efforts, and that smoke spread could cause
(
13 inadvertent actuation of fire suppression systems, and this 14 can also hamper brigade access to areas.
You have CO2 systems 15 that you have to ensure they have the breathing fear on before l
16 they go through.
17 The potential impact of this issue is very plant 18 specific.
It is going to be very dependent on the 19 configuration of the plant, the pathways that are available 20 for smoke flow between areas, whether or not the plant has i
21 guidelines on how to handle their ventilation system and the l
22 indication of a fire, whether they isolate the room, whether J
i 23 they try and vent the room, whether they continue to maintain l
()
24 ventilation.
All of these things have a very strong impact on 25 it, and we don't really know much about how components will HERITAGE REPORTING CORPORATIOti -- (202)628 4888
75 i
1 respond to these smokey environments.
- O-2 We have some evidence that there may be some 3
equipment vulnerabilities, especially high voltage switchgear 4
equipment.
There is non-nuclear experience involving tripping 5
of switchgear by smoke, so again, this was a very difficult 6
one to quantify, and we sort of left it at that.
7 The second aspect of the total environment issue 8
that we look at was inadvertent suppression.
The past 9
experience shows that incidents are occurring about ten per I
10 year, or I'm sorry, ten to the minus 2 per year.
11 What we did is we looked at a review of LER data 12 base since the implementation or since the institution of
()
13 Appendix R, and we found a number of incidents reported in the 14 LER data base.
We have looked at these and some of the t
15 significant ones were there was a case of loss of off-site 16 power induced by a sprinkler actuation.
There was one case in i
17 which a plant operating at full power had two control rods 18 repositioned in full withdrawal and they could not be manually 19 repositioned by the control operator and they had to SCRAM the 1
20 plant.
21 There was a case where some redundant systems were l
22 lost, but we found no cases of inadvertent spurious 23 suppression caused by fires, which was one thing that we were
()
24 looking for.
25 One of the questions that had been raised was l
HERITAGE REPORTING CORPORATION -- (202)628-4888
i 76 1
whether or not smoke spreading from room to room in a fire CE) 4 2
might suppress systems in other areas.
The experience base 3
doesn't have any of these cases, so our conclusion was that I
4 without a fire, there was really no significant generic safety
(
5 impact.
The frequency of these incidents was fairly low.
6 Their significance was typically not a real safety issue.
7 There were some cases where some safety systems were affected, 1
3 8
but in general, they veren't that significant.
9 Now with the fire, the case where you have spurious i
10 suppression in other areas due to fire somewhere else, we
]
11 didn't have any incidents, so we feel that it is probably a i
12 small impact based on the experience base, but it is going to
()
13 be very plant specific.
You would have to go in and look at i
]
14 again the smoke spread paths and whether or not you can have 15 inadvertent suppression actuation in an area remote from the 16 fire, expose vulnerable equipment to the inadvertent l
17 suppression, and by that mechanism lose redundant trains.
18 CHAIRMAN KERR:
Mr. Nowlen, could you finish in i
19 about two more minutes?
20 HR. NOWLy.N:
Yes.
I have two more slides, one 1
21 minute a piece.
22 With regard to the barrier effectiveness, there was 23 an issue raised in our solicitation of expert opinions, i
l
()
24 certain individuals had expressed concern that barrier 25 standards do not induce a positive pressure during testing, HERITAGE REPORTING CORPORATION -- (202)628-4888
77 1
and that this,may be an overlooked mechanism of damage.
2 What we found is that if a penetration has a 3
reliability of on the order _of 90 percent or less, you could 4
introduce scenarios that have been screened out that could 5
represent an order of magnitude kind of contribution to the 6
core damage.
7 Now as I say, this is not a multiplier.
This is an 8
adder where we introduced scenarios that were screened out 9
because, for instance, there is a three-hour barrier between 10 two areas housing redundant eq'11pment.
If you now say that i
11 the barrier is only 90 percent effective or so, then you can 12 reintroduce that, that scenario, and it may contribute like an l
r
()
13 order of magnitude increase in the fire risk because again i
14 this is a, a mechanism by which you can violate your passive i
15 barriers which are the preferred method of protection.
16 Now if the reliability of your barriers is on the 17 order of 99 percent or higher, what you find is that the
+
18 random barrier failure frequency will dominate, and so it 19 becomes less of an issue.
This random failure is based on the 20 frequency that they find doors propped open, penetration seals 21 that have been violated due to maintenance activities or 22 penetration seal that hasn't been installed, those kind of 23 things, and--
()
24 HR. MICHELSON:
The barriers you looked at include 25 the one-hour barriers like cable wrap?
HERITAGE REPORTING CORPORATION -- (202)628-4888
9 78 1
1 MR. NOWLEN:
This is a-barrier between areas or the, 2
we were not looking at--
t 3
MR. MICHELSON:
Cable wrap, we are using cable wrap 4
as the one-hour barrier for Appendix R.
t e
t 5
MR. BOHN:
This particular issue, this particular f
6 issue is raised with respect to wall-type penetrations and how i
7 they were tested.
e i
8 MR. MICHELSON:
Dampers-and so forth?
1 9
MR. BOHN:
Yes.
1 10-(Slide) 11 MR. NOWLEN:
And with respect to the final issue, j
12 seismic fire interactions, again there is no data on frequency l
+
j3 g,)
13 of fires during earthquakes at nuclear power plants.
There is 14 a significant base in the non-nuclear industry, but the I
i 15 applicability of that is somewhat questionable.
l 16 There have been no fires at nuclear power plants, 17 and this involved an examination of events at foreign reactors
[
i
[
18 as well, and what we found is that the risk would be primarily 19 due to systems interactions which could easily Le identified 20 during a plant walk-down.
If they knew what to look for, you
+
t 21 could walk down a plant, look for things like fire jockey
[
\\
22 pumps that are on vibration isolations without consideration I
i 23 of seismic stops, so if you had a seismic event, the vibration
[
24 isolation system could be failed.
Something like the simple j
()
i 25 installation of seismic stops would prevent that problem.
[
i l
l HERITAGE REPORTING CORPORATION -- (202)628-4888 f
i
i 79
[
r HR. MICHELSON:
Did you look at the seismic f
1
(:)
2 qualification of the actuation devices, didn't you?
And we j
2 r
3 recently discovered some mercury switches being used to i
4 actuate CO2.
i f
5 HR. NOWLEN:
That is another aspect of the question i
6 is that you want to look at your fire protection systems and 7
ensure that they wouldn't be vulnerable to seismic actuation i
8 mercury switches.
System is tied to simply a smoke detector 9
where you are going to have a lot of dust lofted.
In a 10 seismic event, to a smoke detector, dust looks just like l
11 smoke.
(
12 HR. HICHELSON:
A number of LERs have indicated j
2 i
i
)
13 where such systems existed, that we have got actuations--
14 HR. BOHN:
That is correct.
In the past data base 3
1 15 non-nuclear facilities, they essentially conclude that you J
I 16 could expeat many spurious fire suppression system actuations
[
17 during earthquake event.
I 18 HR. MICHELSON:
Was that in your probabalistic
[
19 determination?
4 20 HR. BOHH:
We did not do a probabalistic 21 determination on the seismic fire because there is no data j
i 22 whatsoever to estimate the ignition of fires during 23 earthquakes.
()
24 HR. MICHELSON:
Or the ignition of fire protection 25 systems which might be just as bad as having a fire, depending
[
HERITAGE REPORTING CORPORATION -- (202)628-4888
t 80
?
1 on the vulnerability of the system.
4 O
i 2
HR. BOHN: 'Yes, except we do know those would be l
3 expected at a nuclear power plant based on past experience.
4 If you have an earthquake, you can expect spurious actuation.
j 5
MR. NOWLEN:
That is correct, so what you would like i
1 6
to do is identify the *Ypos of things that are vulnerable, and j
i 7
have the plants walk down their systems to ensure that they
[
8 don't have any of those--again, the mercury switches, smoke j
?-
}
9 detector, which are the sole actuation method for a system i
f 10 automatically, things of that nature.
l 11 MR. BOHN:
I think an important thing that needs to l
12 be stated here is the fire protection systems in general are l
O 13 not ismicativ aua11fie4.
There is a reautrement for sust t
14 the stand pipes be seismically qualified, and so although i
15 certain manutacturers do make actuation cystems recently that
{
j 16 are in effect seismically qualified, for many years that was f
4 f
17 not the case, and one of our recommendations is that this h
J t
j 18 probably should be reviewed by each plant to understand their l
f l
19 particular vulnerabilities, t
h 20 CHAIRMAN KERR:
Thank you, Mr. Bohn.
Let's take a j
21 break until five of.
i 22 (A brief recess was taken.)
f T
23 CHAIRMAN KERR:
Mr. Bohn, when you are ready.
I i
24 HR. BOHN:
These slides are in the side packet that j
P l
25 I handed out at the second, the second talk of the day
{
I i
L HERITAGE REPORTING CORPORATION -- (202)628-4888 f
i 81 1
entitled the summary and conclusions.
2 (Slido) 3 MR. B0HN:
So John and Steve have summarized the l
6 conclusions that we have drawn from the year-long study that 5
has been performed, and we can draw certain conclusions and
\\
6 recommendations for work that could be performed to improve v
i 7
our understanding of fire risk, to improve the fire safety at' l
8 plants.
4 9
Now what I have tried to it in these slides is i
10 indicate where appropriate whether it is a piece of research l;
6 11 that might be sponsored by NRC or whether it is an action that l
4 12 the utilities could profitably take or whether perhaps it is j
()
13 combination of both.
i 14 And again, I am going to go through the six issues l
15 and from the--first of all, I will start off with what really t
16 needs to be done to improve our perspective of fire risk?
And l
17 I think the general conclusion that we came to in looking at
)
18 past fire PRAs, was based on limited data in the original
{
L 19 PRAs, based on evolving methodology, it is very difficult to I
i f
{
20 compare the results of any particular existing fire PRA.
That 21 is an earlier fire PRA.
22 Nonetheless, you are led to the conclusion that the r
1 23 projected risk is still quite significant relative to say i
24 internal events, so in order to solidify our understanding, j
()
l 25 improve our fire risk perspective, we could do some several i
I I
HERITAGE REPORTING CORPORATION -- (202)628-4888 t
c
i 82 i
i i
things.
[
O-2 First of all, I think we need to develop defensible j
i
-3 estimates of fire occurrence frequencies.
By this I'm i
i 5
4 primarily referring to being able to include plant-to-plant i
I I
5 variations that reflect how keeping the fire watch practices i
I 6
and other differences amongst plants because soae plaats f
l 7
certainly do include, have a high cognizance of fire and do i
8 include these practices, but right now there is no real
[
9 mechanism for giving plants credit for them where they do I
10 implement these practices in risk asses 4ments, i
4 11 Secondly, we need to develop defensible guidelines
[
4 i
12 for partitioning and fire severity factors.
One of the
[
j 13 results of the requantification was the realization that these f
()
14 judgmental factors were driving the uncertainty to a large 1
1 15 extent in the bottom lines.
I i
16 We have actual data on the occurrence frequencies i
i i
1 17 for large buildings, but very it is very judgmental getting j
i 18 down to individual subcompartments, and so their needs to be a 19 reasonable means of doing this.
And it might involve going 20 back and looking at the original fire data base and try to
[
21 develop some guidelines from that.
j i
i l
22 In addition, it would involve developing a
{
l i
23 standardized set of recommendations on which to base your 1
[}
24 partitioning factors, t
l 25 CHAIRMAN KERR:
You think this is a doable task in i
r HERITAGE REPORTING CORPORATION -- (202)628-4888 i
i 83 l
1 some finite time?
Finite resources?
{
O 2
HR. BOHN:
It is certainly doable in the' sense that i
l 3
we can evaluate what has happened in past fire occurrences in I
4 terms of partitioning.
f i
5 In terms of coming up with recommendations, we have 6
a set of recommendations that we have provided to NRC for 7
consideration for use in their individual plant examination
)
8 recomme.tdations, but certainly that could be improved on.
[
9 CHAIRMAN KERR:
The term defensible guidelines is I
10 used, which I assume means--I'm not quite sure what it means.
[
l 11 HR. BOHN:
In the swJ.se it would be consistent with
{
12 past fire occurrences.
()
1?
CHAIRMAN KERR Okay.
14 HR. BOHN:
Third, it is obvious that we do not have 1
15 an adequate fire computer code now for predicting damage of
(
a 16 components.
As we have seen, COMPBRN I was 2enerally 4
17 considered to b9 quite, quite overconservative.
The version j
i 18 COMPBRN III we have now is still conservative and sometimes
[
l I
19 underconservative, but the bottoa line is it is not a tool 1
i 20 that we have confidence in.
i 21 In addition, neithet of them have been benchmarAad l
i 22 against actual fire enclosure data, and so it is clear here 8
23 that we need a tool, not necessarily a huge complex code, but
[
I 24 nonetheless we need some adequate tool for calculating fire f
()
i 25 growth and component damagability duri.ig fires, i
J HERITAGE REPORTING CORPORATION -- (202)628-4688
4 84-1 MR. MICHELSON:
One thing you didn't remark on, of l'h k/
2 course, is the ability of COMPBRN III to predict the migration 3
of heat and smoke from the fire zone to other areas where it
]
4 might potentially actuate fire protection.
You need somehow 5
to know how much smoke gets there if you want to say the yes 6
or no on the, on an inadvertent suppression.
7 MR. BOHN:
We certainly agree.
8 HR. MICHELSON:
COMPBRN doesn't do necessarily a 9
good job of that, either.
10 MR. BOHN:
That is entirely correct.
11 MR. NOWLEN:
Doesn't do it at all.
12 MR. BOHN:
We also need to update and maintain a
()
13 fire data base with regard to occurrence and suppression, and 14 component damage threshold data.
Included in that we think 15 should also be the occurrence of inadvertent fire suppression, 16 so these we would recommend be updated on a current basin by 17 someone.
18 These would be useful both to NRC, they would be 19 useful to the NRR fire protection branch people, and certainly 20 be useful to the industry.
21 It would also form the basis for any individual 22 plant examination, look at fire frequencies which would help 23 standardize those calculations if and when they are performed.
24 Lastly, we feel that it would be useful to take a
()
25 look at fire PRAs for several plants that have not been looked HERITAGE REPORTING CORPORATION -- (202)628-4888
85 1
at already, and to actually perform them from the start as we
~'
2 are doing to Peach Bottom and Surry, and include up-to-date 3
methodologies rather than the evolving methodologies that we 4
observed in the PRAs we requantified to use the latest data-5 bases and also perform the plant walk-downs so that we could 6
make sure that the judgmental factors, partitioning fire 7
severity, et cetera, were included properly.
8 I think an additional fact would be one would be 9
able to get some handle on the impact of Appendix R from a 10 slightly different perspective.
In the four PRAs we looked 11 at, the Appendix R modifications were ongoing at the same time 12 the PRA was done.
In discussing with the original PRA
(
13 practitioners, we found that as the PRA identified certain 14 dominant accident sequences, that tended to influence the 15 extent of the fire PRA--excuse me--the Appendix R 16 modifications, and so it would be interesting to try to assess 17 what were the effect of the Appendix R modifications that were 18 done if they didn't have the insight of the concurrent PRA.
19 MR. MICHELSON:
Before you leave that slide, though, 20 one important issue which I don't believe you really put 21 together in the presentation is the question of having 22 experienced a fire in a given area, while important now, how 23 important is it now to be sure that you do not inadvertently
()
24 suppress in other areas that are adjacent?
You haven't 25 looked at the kind of things that are adjacent.
What is the HERITAGE REPORTING CORPORATION -- (202)628-4888
86 1
probability of those being actuated by the heat and smoke from m
2 the fire and what would be the consequence?
That was not in 3
your study I don't believe.
4 MR. BOHN:
That is correct.
We did look at those 5
issues for the two plants that we are looking at under the 6
NUREG 1150, and we found it was very plant specific and in 7
fact, we did not find for those two plants that the particular 8
combination of rooms for which you could have a fire in one 9
room and spread smoke to an adjacent room, in effect redundant 10 safety systems.
The two plants we looked at seemed to have 11 large gratings in floors, large area.
We could not pin down 12 an area for which that would be significant if true that such
()
13 situations do occur.
14 MR. MICHELSON:
Because random event inadvertent 15 actuation has not been shown to be important, and I think that 16 is correct.
It does knock out safety-related equipment, but 17 never enough of it, but in combination with a fire, it could 18 be quite important, depending on the arrangement.
19 MR. BOHN:
Undoubtedly for any particular plant 20 given those situations, it could have a dramatic effect on 21 that plant's fire r.isk.
That is certainly true.
22 So those were the recommendations that we make 23 relative to the overall perspective of fire risk,
()
24 understanding in a more uniform way what the actual fire 1
25 numbers were.
l HERITAGE REPORTING CORPORATION -- (202)628-4888
87 1
(Slide) p 2
MR. BOHN:
Of course, if one did relook at several 3
plants that had not been examined in the past, one w'ould, of 4
course, try to uniformly incorporate the unaddressed fire 5
issues.
l 6
MR. MICHELSON:
In doing a walk-down, how do you 7
know how hot an area outside of a postulated fire gets and how 8
much smoke to expect in order for the walk-down to make sense?
9 MR. BOHN:
You don't know that, but you examine the 10 conductivity between the room and you look and see if the room 11 with the fire has more than one access, and you look at given 12 that it has maybe only one access like a typical switchgear
()
13 room you often see, then you would look at where the spread of 14 the fire could actually go.
15 MR. MICHELSON:
Write some criteria or instructions 16 to the person doing the walk-down which would kind of 17 highlight this problem for him to look for.
F 18 HR. BOHN:
Absolutely.
Turning to the issue of 19 manual firefighting effectiveness, now I want to say that in 20 these slides, we have not ranked these issues.
However, it is 21 clear that three of then stand out relative to the other.
22 Probably the most important in our view is manual 23 firefighting effectiveness.
The sorts of things that would be
()
24 implied by our conclusions are we need to test some cabinet 25 and cable routing combination to assess the possibility of
- - (202)628 4888 HERITAGE REPORTING CORPORATION --
4 88 1
fire spread cabinet to cabinet.
2 We do have limited data that was performed at Sandia 3
in which a double wall cabinet was considered and fire in one 4
cabinet was shown not to spread in the adjacent cabinet across 5
this double wall cabinet.
However, many times the electrical 6
cabinets have cables that run underneath that trench..This 7
forms an ideal pathway to spread fire.
This has not been 8
looked at in an experimental sense.
9 Similarly, certain cable configurations in which the 10 cable comes out of the top of the cabinet could perhaps form a 11 basis for spreading fire.
This needs to be looked at briefly 12 anyway in a test situation.
)
13 We need to develop guidelines for the required 14 response and extinguishment time for various types of 15 equipment.
For example, we have some data that shows that in 16 a typical cabinet, if you get a fire started, by the time the 17 fire has gone on eight to ten minutes, that fire can no longer 18 be extinguished with a hand-held extinguisher, and this 19 conclusion we drew and was substantiated by utility personnel 20 that reviewed the photographs at various times from this 21 particular experiment, and so different types of equipment 22 have different times before they are sort of totally damaged 23 if you will, and we think we need to understand that better.
()
24 Not all equipment is equally vulnerable.
25 MR. WYLIE:
Would you consider the Browns Ferry fire l
HERITAGE REPORTING CORPORATION -- (202)628-4888
89 1
in that category?
Who went under the control room, cable 2
spread room?
3 MR. BOHN:
I guess I don't know.
4 MR. WYLIE:
Because they put that out with hoses 5
very casily.
6 MR. MICHELSON:
It wasn't a confined area, either, 7
HR. WYLIE:
No, but it went on for three hours, 8
wasn't it?
9 HR. MICHELSON:
More than that.
10 MR. BOHN:
The tests we have were started by a 11 simulated 15 amp short in a typical motor control center, and 12 it was found that within five minutes or so, that the cable,
()
13 the fire had engulfed the entire cabinet.
Everything within 14 it was damaged, and that fire could not be put out with a 15 hand-held CO 2 solution.
16 MR. WYLIE:
What kind of cable was that?
17 MR. NOWLEN:
PPE jacketed.
18 MR. BOHN:
The test was repeated, however, with IEEE 19 cable.
20 MR. WARD:
That was hand-held CO 2 extinguisher and 21 that probably, well, that couldn't have extinguished a cable 22 fire.
It did take--
23 CHAIRMAN KERR:
They didn't use, did not try a 24 garden hose.
One of our consultants said several years ago
(}
25 the best thing to do with an extinguisher is throw it at the HERITAGE REPORTING CORPORATION -- (202)628-4888
90 1
nearest water pipe and hope that it will fracture the water O
2 pipe and spray water on the. fire.
3 MR. BOHN:
That is certainly true, and the recent 4
cabinet fires in which they used CO 2 sprayed in the top of 5
the cabinets which suppressed the fire, smothered it if you 6
will, but then when they reopened the cabinet, the fire 7
re-ignited because the heat was there, as an' example of that 8
sort of situation.
9 Yes?
10 DR. SHEWHON:
I realize that suppression things 11 change with what consultants the NPC has or what staff 12 members, but it seems to me that for a while the NRC didn't
()
13 like water on things and was pushing people toward CO2, and 14 halon, is that correct?
And thus we hcVe insisted that these 15 systems be in plants often?
16 MR. BOHN:
There is no specific requirements as to 17 the particular form of automatic suppression say in the 18 Appendix R guidelines.
19 DR. SHEWMON:
Water is quite acceptable in all 20 situations?
21 MR. MICHELSON:
No.
Depends on the utility as to 22 whether they use it or not.
23 MR. BOHN:
They have to, they have to, they attempt
()
24 to protect the cabinets overhead spray.
I have seen this, 25 that certain instances, the tops of cabinets will be protected
. HERITAGE REPORTING CORPORATION -- (202)628-4888
. _ _ ~ -
91 1
six inches down across the top if they have overhead water C) 2 spray.
They prefer not to do it in general.
3 In electrical cabinetry rooms, they prefer to use t.
4 halon or CO2.
5 MR. MICHELSON:
This is a control room situation, 6
for instance, and it is still unclear, and I have asked it a 7
number of times of utilities, what they are going to do if 8
they get such a cabinet fire, will they bring the hose in from 9
outside or not?
And the answers I get are quite mixed.
10 DR. SHEWMON:
In the cable spreading room?
11 MR. MICHELSON:
Cable spreading room; it is also 12 quite mixed to whether they use it with water..
The NRC likes
()
13 them to use water.
There has been exceptions where they 14 don't.
Beaver Valley I believe was one of them.
They finally 15 settled on CO2.
16 MR. BOHN:
- Okay, In terms of manual firefighting 17 effectiveness, the third bullet is to evaluate fire brigade 18 effectiveness via fire simulations in a number of plants to 19 get an idea of first of all, operator versus fire fighter l
20 systems awareness.
When the fire brigade team goes down and 21 sprays a set of cabinets for a fire, it is necessary that they 22 understand first of all, the power is still on and they aro 1
23 likely to electrocute themselves whether or not there is
()
24 adjacent equipment which could have redundant trains of the 25 same system being danaged by the fire which would also be
{
HERITAGE REPORTING CORPORATION -- (202)628-4888
92 1
damaged by water, in which case they could get two trains if
(")
2 that situation occurs, so there is a requirement for awareness 3
of systems.
4 Now Appendix R currently requires that one member of 5
the fire brigade be knowle;dgeable of systems, but that's the 6
extent to what it says.
7 So in terms of the philosophy of manual firefighting 8
skills, everyone in the plant is defined to be a fire brigade 9
member.
Sometimes they have dedicated fire brigades which 10 come from some related area, but that are not necessarily 11 particularly knowledgeable of the systems aspects, so this 12 question as to what is optimal relative to operator versus 13 firefighters is something that perhaps needs to be explored 14 for effective firefighting.
15 MR. MICHELSON:
Also the question of what happens 16 when you spray the cabinets that is on fire in terms of other 17 unwanted actions as a result of adding water in addition to 18 heat?
We don't know.
19 MR. BOHN:
In addition, I think it would be, we 20 think it is really important to understand how vulnerable 21 plants are to the fact that the fire brigades can't really get 22 there in say a 10 to 15-minute timeframe, which the utility 23 fire brigade members we have talked to agree is a very
()
24 difficult time to get down there.
The typical scenario is 25 they get a fire alarm, first off, they will send someone from HERITAGE REPORTING CORPORATION -- (202)628-4888
93 1
the control room to verify that it is not spurious, because
(~)s
\\-
2 they get many spurious alarms.
Then if there is a real fire, 3
then the fire brigade has to go to its, its assembly point and 4
don the protective equipment, then proceed to the fire area.
5 All of'this, if you have a fire in a sort of remote partHof 6
the plant, it will tend to slow it down.
7 DR. SHEWMON:
Does the guy who goes down to see what 8
set the alarm off have access to an extinguisher or is he just 9
there with his radio saying yes and no?
10 MR. LAMBRIGHT:
The person that typically, the 11 person typically responding to the fire would be whoever the 12 senior leader would be for a given plant, and his function is
()
13 not to go down and extinguish the fire per se, but respond to 14 whether there actually is a fire.
15 DR. SHEWMON:
The answer to my question is no, he 16 would not have an extinguisher?
17 MR. BOHN:
No, that is not the answer.
The answer 18 is his primary job is to go down and respond back to the 19 control room.
20 DR. SHEWMON:
I understand that.
21 MR. BOHN:
However, one would certainly assume and I 22 am sure as the primary response he will grab a fire 23 extinguisher if it is small and start spraying.
They are not
()
24 precluded from doing that.
25 DR. SHEWMON:
Thank you.
HERITAGE REPORTING CORPORATION -- (202)628-4888
~
94 1
MR. BOHN:
So trying to respond then is something YJ 2
that we think could be looked at just by drills in an actual 3
plant.
4 Room accessibility again is something to be 5
considered.
Many of these rooms are locked and you have to 6
find the appropriate keys.
When we walked down these plants 7
and we want to look in a particular room, we have to ask the 8
operators which the key is, and sometimes we spend five or ter.
9 minutes digging out what the correct key is.
So these are the 10 sort of things you could evaluate by fire simulations at a 11 particular plant that would I think shed some very useful 12 light on how well manual firefighting is effective, actually
()
13 effective.
14 You have to understand that Appendix R minimum 15 requirements for training is four sessions a year, only one of 16 which requires that you hold a hose, and the hose does not 17 have to be spraying.
Now that doesn't mean that many plants 18 don't do a much better job because I think they do, but there 19 is the minimum requirements nonetheless--three movies and a 20 simulation if you will.
21 Okay.
Lastly, we think that it would benefit a 22 particular utility to map the fire areas versus the critical 23 response time for equipment in there so they would really
()
24 understand--if they assume that a manual firefighter can get 25 down to a certain critical motor control center area, they HERITAGE REPORTING CORPORATION -- (202)628-4888
95 1
would really understand whether he could actually do that in a
,n 2
10 to 15-minute time span, but it might result in them putting 3
in more automatic systems or more adequate alarm systems, 4
something like that, but we think that their perspective en 5
how fast they can get there is not, is not consonant with the 6
response tince that are really required.
7 (Slide) 8 HR. BOHN:
Continuing on for the recommendatione 9
for, to approve manual firefighting effectiveness, we think 10 that a plant should consider plant-specific training in areas 11 where equipment is susceptible to sprays so that the fire 12 brigade members really do understand what is critical and
()
13 what, what other damage could be done during the manual 14 firefighting and make sure that they understand that for 15 electrical fires, the proper use of fog nozzles and the fog 16 nozzles are indeed available for those areas to prevent 17 electrical shock coming back through the hose.
18 HR. MICHELSON:
Are you also recommending that they 19 figure out whether they can even see in those areas when they 20 get a postulated fire?
21 HR. BOHN:
That is absolutely true.
Our tests show, 22 and certainly the Browns Ferry experience indicates that 23 within five to six minutes after a cabinet fire, the level of
()
24 smoke has gone down nearly to the floor and you cannot see the 25 fire.
We have several tests that show that, and that happened HERITAGE REPORTING CORPORATION -- (202)628-4888
96 1
in Browns Ferry.
You cannot actually see where to spray the
(~T 2
hose.
3 CHAIRMAN KERR:
I know we got you started later than 4
scheduled, but if you can help us make up a little time, it 5
would be appreciated.
6 MR. BOHN:
You bet.
I'm, I have about four or five 7
more slides, so I will try to speed it up.
8 Lastly, and maybe most importantly, to evaluate the 9
fire brigade effectiveness under actual fire conditions, 10 including dense and toxic smoke, that is instead of the il minimal training which does not require actually fire 12 simulation, to actually make sure that the brigades get
()
13 training in fire touers so that they experience dense smoke, 14 lack of visibility and the possibility of toxic smoke, because 15 certainly toxic smoke will arise due to the cables burning 16 itself.
t 17 It is also true that halon after a fire has been 18 going- -well, even though halon is supposed to be breathable, 19 it is at the start, but after halon has become very, very hot, 20 it does give you products that are noxious basically.
A 21 number of fire brigade people we talked to indicated that 22 although in theory halon is breathable, in responding to a hot i
23 fire, they had no confidence that it was breathable.
That's,
()
24 by the way, why they go from CO2 to halon systems, because 25 halon in principle is only 6 percent effectiveness, and it is HERITAGE REPORTING CORPORATION -- (202)628-4883
97 1-breathable, but that does not necessarily seem to be true.
O 2
(Slide) 3 MR. BOHN:
Turning to fire code adequacy, we have 4
already said we don't have a reliable tool for calculating 5
fire growth and spread.
We think that the correlation that 6
COMPBRN is based on needs to be benchmarked against data.
7 They have not been.
We think COMPSRN III, perhaps the CHAM t
8 code that has been partially under development by NRC, or some 9
code really needs to be benchmarked against actual fire 10 enclosure data.
We need to validate these tools.
11 We think that there needs to be a fairly simple 12 model, somewhat more complex that COMPBRN, but still
)
13 relatively simple, maybe a control volume type code or flow 14 code capable of accurately predicting component damage times,
[
15 needs to be developed and benchmarked, 16 And the uncertainties in that code 17 characterize--that really doesn't exist.
Now we need 18 something that can be operated for a PRA which requires maybe 19 10 to 15 runs per fire area, in the critical fire area where 20 spread of fire is, is significant, so the code has to be 21 something that can be run multiple times in a workable 22 timeframe and an hour run is not acceptable.
t 25 Lastly, and this is coupled with some of the others,
()
24 we need to understand, we have to have consistent and i
25 validated criteria for predicting when cables are damaged.
We HERITAGE REPORTING CORPORATION -- (202)628-4888
98 1
havr. limited data now on some cable types.
That should really O
2 be extended to encompass most of the cable types that you see 3
in plants today.
4 (Slide) 5 MR. BOHN:
In terms of relative priority, fire code 6-adequacy is closely second let's say behind manual 7
firefighting suppression effectiveness, and the third 8
which we also consider one of the top three is control systems 9
interactions.
10 I think there needs to be a method and guidelines 11 for detailed review of the independence between control room 12 panels and remote shutdown panels that is currently required
()
13 in Appendix R.
However implemented, that verifying that is 14 quite difficult.
It probably, if such a method were 15 developed, it should be applied to several older plants to try 16 to identify the most likely sources for significant control 17 room interactions, and then this should be applied on a 18 plant-specific basis by the utility.
19 Lastly, we think it would be very beneficial to 20 perform a review of the capabilities of remote shutdown 21 panels.
Currently remote shutdown panels primarily serve the 22 function of isolating the containment, that is, de-enable 23 valves.
They have limited capability for actual control, as
()
24 per the guidelines of Appendix R.
25 So we think a review should be done to determine the HERITAGE REPORTING CORPORATION - _(202)628-4888
99 1
robustness of their capabilities for control over a wide range
. (~2 J
\\-
2 of accident scenarios, and perhaps some recommendation would 3
be prepared such that you could enhance the capabilities of 4
existing remote shutdown centers, and it is our view that it 5
is possible to enhance the remote shutdown panel, to perform 6
feed and bleed operations such that you would get a fair 7
amount of benefits of say a postulated remote, remote shutdown 8
facility, as considered under TAP A-45, 9
The last three are relatively less important in our 10 view, although not to be disregarded.
11 In terms of spurious suppression, we have said it is 12 very plant specific, but a plant could identify areas where
()
13 combination of smoke spread could hamper firefighting or cause 14 inadvertent suppression.
The utility could review its layout q
15 for that if there were appropriate guidelines.
16 Plants should review fire / water drainage paths.
17 There have been cases of fire system suppression systems have
.i 18 leaked into important circuitry and caused problems.
19 In terms of fire environments, we have already 20 mentioned expanding the cable fire data base to develop 21 criteria for damage.
We think that some investigation of the 22 impact of in-cabinet gaseous fire suppression systems should 23 be done.
It is possible that static charging or low 24 temperatures induced by spraying CO2 directly in the cabinetry
()
l could cause adverse effects to those circuitry, and of course, l
25 I
I HERITAGE REPORTING CORPORATION -- (202)628-4888
r 100 1
in some cases, it is proposed or has been used to install 2
suppressions right in cabinets within the control room, so 3
there are possible adverse effects that should be examined.
4 And lastly, Steve mentlened earlier that some 5
investigation of the effect of smoke on high voltage systems 6
should be done because there have been cases where the high
- /
voltage equipment has been triggered by just smoke 8
particulates alone.
9 In terms of barrier effectiveness, this was the 10 issue of whether or not the current method in which 11 penetration barriers are tested--we think that one ought to go 12 back and look at the reliability of barriers from their past
()
13 experience in actual fires, and we understand from the 14 utilities that in a qualitatiss sense, they actually have 15 performed quite well.
16 We think some limited test data should be done to 17 verify this.
In particular, we think it is more of an age 18 question than anything else, so the testing should be 19 performed on aged penetration seals.
Yes?
20 DR. SHEWMON:
The barriers I think of doors which 21 are specified which I would guess wouldn't change too much 22 where tests are done.
You are thinking of other kinds?
23 MR. BOHN:
We are particularly thinking of sort of
()
24 fiberglass or polymeric seals around pipes and electrical 25 penetration and in walls.
HERITAGE REPORTIllG CORPORATION -- (202)628-4888
101 1
DR. SHEWMON:
Fine.
Thank you, i
-s
(
\\
'2 (Slide) 3 MR. BOHN:
And those are the only kind to which the 4
issue applies really.
5 Lastly, seismic fire interaction, we think it would 6
be useful to formalize guidelines for identification of 7
potential seismic fire interactions because most of them can 8
be picked up during a walk-down.
We have a good knowledge 9
either from past fire experience in non-nuclear facilities, or 10 from things we have seen during our many seismic and fire 11 walk-downs to identify these things.
12 And then the utility could use these guidelines to
()
1?
Verform walk-down of their plant to eliminate them perhaps 14 under, as part of A-46, and during the walk-downs that are 15 proposed for that, they could also incorporate those.
We have 16 a preliminary set of recommendations as to what should be 17 looked at.
18 In particular, very importantly is the CO2 system.
19 There are large volume tanks, don't tend to be anchored, and 20 so during an earthquake if you had a fire, but your large tank 21 six foot long has slid and broken the small pipes that go into 22 it, you would lose your CO2 system for other areas and you 23 would dump the entire CO2 inventory in the one room and give
(}
24 you problems like occurred at Surry.
25 MR. MICHELSON:
Or you can push the wall out.
HERITAGE REPORTING CORPORATION -- (202)628-4888 j
102 1
1 HR. BOHN:
There are a number of things.
We did s
\\-
2 have a set of preliminary guidelines that could be used, and 3
our feeling is these things could be eliminated just by an 4
intelligent walk-down and a set of guidelines.
5 So that summarizes the implications and 6
recommendations that come from the look we have taken.
Three 7
of them we think are say top drawer items, and the last three 8
we think are important, but of less overall generic 9
significance.
10 If there are any questions before I go on?
L 11 CHAIRMAN KARR:
Questions?
I guess not.
Thank you, 12 Mr. Bohn.
()
13 HR. FLACK:
Okay.
We are running a little bit late, 14 so I will briefly go over the comments that we received 15 stemming from a peer review meeting we had on February 23rd.
16 Fifteen organizations were contacted.
It was 17 strictly on a voluntary basis that they came.
The following 18 people did come--Brian Dolan from Duke Power; M.ke Kominski, 19 Wisconsin Electric Power; Jerry Philabaum from Philadelphia 20 Electric Company; and Jean Sursock from EPRI.
21 We also received additional comments from Edison 22 Electric Institute, and Arizona Public Service Company.
I 23 have summarized those comments here.
Basically I have broken 24 them down into general comments and comments that, addressing
()
25 each of the specific issues which we have just gone over.
t
, ____.H_ERITAGE F EPORTING CORPORATION - _( 202) 628-4 8 8 8 __
J 103 1
General comments were research should find the h
2 pitfalls so utilities know what to do in order to lower risk.
3 Management is generally concerned with what they are getting 4
for their money, if they have spent a lot on Appendix R cost 5
effectiveness.
6 Will we see a real reduction in risk or just 7
finetuning the uncertainties?
It's difficult, some members 8
would way it was difficult to extrapolate test data to real 9
world situation.
Actual experience is a better indicator.
10 However, it was questionable, other members pointed out, 11 whether limited nuclear exporience is sufficient to draw 12 conclusions.
()
13 (Slide) 14 MR. FLACK:
There was concern expressed over the 15 large uncertainties in the bottom line, especially when 16 considering the bounding analysis.
17 It was questionable whether future research will be 18 able to narrow uncertainties much further, and study had a 19 pessimistic viewpoint which may be mis interpreted and result i
20 in unnecessary backfits, and another comment, potential issues 21 are unaddressed by the selected method of risk evaluation 22 only, sure to point out that they do not represent unaddressed 23 issues for nuclear power plants.
24 MR. WARD:
I don't understand that one, John.
Could
(}
25 you give me an example?
i HERITAGE REPORTING CORPORATION -- (202)628-4888
104 1
MR. FLACK:
What they are saying basically here, 2
unaddressed as far as PRAs, fire PRAs, are concerned, but they 3
are not, they are addressed out in the field.
They didn't 4
want it to sound they were just not addressed, unaddressed.
5 Is that plain?
6 MR. WARD:
There are some issues that have been 7
identified that the PRA doesn't capture?
8 MR. FLACK:
Yes.
The potential unaddressed issues, 9
yes.
10 The six areas which we have covered before; they 11 just wanted to let it be known that they are addressing these I
12 issues.
()
13 MR. MICHELSON:
To varying degrees.
14 MR. FLACK:
Unaddressed as far as the PRAs are 15 concerned; i
16 okay.
Under manual fire ighting, they pointed out 17 since the Browns Ferry fire there has been significant t
18 improvements in training and equipment and therefore, Browns
[
4 19 Ferry should nut be a case in point.
(
20 They have disagreed with that dedicated versus 4
21 non-dedicated fire brigade conclusion.
They were conclusions 22 that had been drawn in the earlier draft and have subsequently 23 been updated to reflect that comment.
()
24 Cannot determine effectiveness by simply knowing the 25 number of brigade members.
This was response to surveys.
HERITAGE REPORTING CORPORATION -- (202)628-4888
105 1
Regarding training deficiencies, need to expand on O
2 what was actually performed in the field, and that proper 3
smoke tower training can address the types of problems which 4
the brigade members may encounter.
5 (Slide) 6 MR. FLACK:
Under control systems interactions, they 7
pointed out that operators should be given more credit for 8
recovery, that systems interaction analyses are detailed and 9
thorough when they were performed.
Following Appendix R, they 10 were performed in great detail.
11 It is cheaper sometimes they pointed out to go in 12 and physically separate systems rather than trace down all the
()
13 interactions which could take place.
14 They disagreed on the time to detect control room 15 fires.
They said that ventilation fans blow smoke out.
16 Cabinet smoke detectors would trip, and corridors are 3
i 17 frequently traveled, all leading to fast response to a fire in 18 the control room.
I i
19 DR. SHEWMON:
If ventilation fans blow out smoke to 20 a detector, how does that aid detection?
21 MR. FLACK:
They felt the operator would smell it or 22 or maybe detect--they weren't that specific on that.
i 23 MR. HOWLEN:
That varied quite extensively between
()
24 plants.
Scme plants actually use the control panels as return 25 air plenums so that they would not blow smoke out, blow smoke 4
HERITAGE REPORTING CORPORATION -- (202)628-4888
106 1
out in the air handling system which may or may not have a
~
[\\')
~
2 detection system in it.
3' MR. SHEWMON:
Fine.
Thank you.
4 MR. FLACK:
Components survivability, we should not 5
conclude that some poorly designed systems represent 6
industry-wide practice.
It is important to differentiate 7
between types of systems when modeling suppression, and EPRI 8
has just issued a request for proposal to study the effects of 9
fire suppression, and that adequate design followed by good 10 maintenance program will assure high reliability of the fire 11 suppression system.
12 On the barrier survivability, based on previous
()
13 non-nuclear experience, barriers have demonstrated excellent 14 performance.
They felt that actual fire performance of fire 15 barriers tested with much ASTM methods provide a statistical 16 base for continued use of ASTM testing methods.
17 Differential pressure across the barrier is believed 4
18 to have little impact on barrier integrity, and analysis, the 19 analysis that is done is conservative because the amount of 4
20 combustible material that is available is limited, and that 21 actually hose tests and time-temperature curves are more F
22 conservative than most European tests.
23 Under seismic fire, the results were helpful for 24 utilities since they gave them something to look for at their l
(
()
25 plant.
That was a list that was developed in the record.
l (202)628-4888 HERITAGE REPORTING CORPORATION --
107 1
The statement that seismic / fire interactions has O
2 largely been dismissed is not accurate, and that the data is, 3
they agreed is insufficient to identify this issue as 4
significant.
5 And the last comments on the adequacy of analytic 6
tools, they agreed with the limitations that exist with regard 7
to the modeling of fire conditions, and they felt that, the 8
one writing comment, the entire report is based on failure of 9
COMPBRN III to agree with available data, or there was 10 insufficient data to support the conclusion.that summarize the 11 comments.
12 There is no, if there is no questions, any
(
13 questions?
There is no questions, I would like to turn the 14 rest of the presentation ovnr to my section leader, Jerry 15 Wilson.
16 CHAIRMAN KERR:
Thank you.
i 17 MR. WILSON:
I am Jerry Wilson with the Office of 18 Research.
19 I would like to wrap it up for today and say that 20 now that we have the consultant's report, we are going to be 21 reviewing their conclusions and recommendations, and I engaged I
22 in three different actions which the staff will take.
23 For those issues that are covered by current
()
24 regulations, we would recommend to NRR that they pursue 25 further implementation that will address the implementation of a
HERITAGE REPORTING CORPORATION -- (202)628-4888
108 l
1 the regulations.
(~h 1
2 For issues that aren't currently covered by 3
regulations, put that into our generic issue prioritization 4
scheme, and for issues that require further R and D such as 5
improvement of analytical methods, we would put that in our 6
research prioritization scheme.
7 I anticipate that we will have that all documented 8
and copies sent to the ACRS by the end of the fiscal year.
9 CHAIRMAN KERR:
Question?
I 10 MR. MICHELSON:
Does that mean by October 1st you 11 will tell us which went into category 1, 2 and 3?
12 MR. WILSON:
Yes.
(
13 MR. MICHELSON:
Not before October 1st?
14 MR. WILSON:
I anticipate sooner than that.
15 MR. MICHELSON:
I was under the impression it was 16 going to be a lot sooner than that.
In other words, you don't i
k 17 feel that there is an urgency such that it needs to be done r
18 any sooner?
19 MR. WILSON:
Right.
20 CHAIRMAN KERR:
In attempting to deal with some of 21 the recommendations, is it your feeling that one could deal 22 with mest of these appropriately by changing the let's say l
23 Appendix R considerably, or does one have to do a j
()
24 plant-by-plant PRA, or none of the above?
Have you reached 25 any conclusion on this yet?
s HERITAGE REPORTING CORPORATION (2021628-4888
109.
l 1
MR. WILSON:
We have not reached a conclusion.
./^)
\\#
2 CHAIRMAN KERR:
Any further questions?
3 Thank you very much, gentlemen, for I think a very 4
complete and comprehensive report.
5 MR. MICHELSON:
There is a question to the 6
Committee, and that is whether or not the Committee is willing 7
to wait for October 1st to write a letter on the subject which 8
apparently is when we will know what the actions are, or do 9
you want a letter prepared soon?
10 CHAIRMAN KERR:
What would your recommendation be?
11 MR. MICHELSON:
My recommendation was originally 12 going to be to wait because I thought it was going to be July
()
13 that we would see the plan.
It is going to be October 1st, 14 and I believe I would recommend that we proceed with a letter.
15 CHAIRMAN KERR:
You mean write before we see--
16 MR. MICHELSON:
Because it is going to be another 17 six months, five months, seeing what they are going to do.
I 18 think we have waited long--this program has dragged on and on.
19 CHAIRMAN KERR:
Would the letter primarily say we 20 think things should be done faster, or would it comment on 21 incomplete--
22 MR. MICHELSON:
Bring to the attention of the 23 Connission the work was done and that we have the following
()
24 important comments on the work if there are.
If there aren't 25 any important comments, we shouldn't write a letter, and your HERITAGE REPORTING CORPORATION -- (202)628-4888
110 1
comments then, the Commission should decide whether they want A
2 to wait until October to hear the staff's side of it or 3
whether they will push the staff to get it done a little 4
quickar.
5 Absent any pressure on our part, the staff will 6
indeed take until the 1st of October to tell us, so it is the 7
wish of the Committee as to whether they would like to 8
proceed.
9 CHAIRMAN KERR:
Is it your view that the principal 10 purpose of the letter should to be try to speed up the 11 process?
r 12 HR. MICHELSON:
That's right.
()
13 CHAIRMAN KERR:
It does not take a very long letter 14 to do that.
15 MR. MICHELSON:
That's right.
16 DR. SHEWMON:
It may take some reasons, that risk is 17 so high that it is an imminent concern or something.
18 MR. MICHELSON:
Here they have told us the risks are 19 anywhere from 10 to 60 percent of total core melt.
That's a 20 larger contributor than most of the stuff that we spend 21 aimless hours on.
j 22 CHAIRMAN KERR:
Seismic risk is about 50 percent, 23 which says to me that you know, really the, what we ought to
()
24 be looking at is external initiators an( forget about internal i
25 initiators.
NEEITAGE REPORTING CORPORATION -- (202)628-4888
r-
~
111 1
MR. MICHELSON:
Precisely; I wasn't going to mix up 2
the seismic with this one.
That's another problem.
3 CHAIRMAN KERR:
You can't.
I mean it seems to me 4
that unless one concentrates on the total picture, looking at 5
little pieces may give one--
6 HR. MICHELSGN:
It is a big enough little piece so 7
that I think it stands on its own merits, if we believe that, 8
the work that was done.
9 CHAIRMAN KERR:
Carl, I would say if you are i
10 convinced we should try to speed up the process, I would urge l
11 tnat you draft a letter and wo will talk about it on Saturday, 12 whatever we get some time during the week.
()
13 Is there further comment or questions?
Again, we 14 thank you, and we will go to the next item on the agenda, i
15 which is regulatory guides.
Mr. Wylie?
16 (The issue of regulatory guides was discussed off 17 the record, and at Noon, the meeting was recessed, to 18 reconvene at 1:00 p.m.
the same day.)
19 20 21 22 23
(
25 HERITAGE REPORTING CORPORATIO!! -- (202)628-4888
1 CERTIFICATE
()
2 3
This is to certify that the attached proceedings before the 4
United States Nuclear Regulatory Commission in the matter of:
5 Name:
Advisory Committee on Reactor Safety, 337th General Meeting 7
Docket Number:
8 Place Washington, D.C, 9
Date:
May 5, 1988 10 were held as herej-appears, and that this is the original 11 transcript thet "r for the file of the United States Nuclear 12 Regulatory Commission taken stenographically by me and, thereafter reduced to typewriting by me or under the direction 13 14 of the court reporting company, and that the transcript is a 15 true and accurate record of the foregoing proceedings.
16
/S/[Sn W
- ct, 17 (Signature typed):
Catherine S.
Boyd 18 Official Reporter 19 Heritage Reporting Corporation 20 21 22 23 24 1
25 O
Heritage Reporting Corporation (202) 628-4888
O O
O FIRE RISK RESEARCH BRIEFING TO THE 337th ACRS FU LL COMMITTEE MAY 5,
1988 PRESENTED BY JOHN H. FLACK RISK ASSESSMENT ENGINEER ADVANCED REACTORS AND GEN ERIC ISSUES BRANCH i
DIVISION OF REGULATORY APPLICATIONS OFFICE OF NUCLEAR REGULATORY RESEARCH j
i
O O
O PURPOSE:
To Present the Results of the Fire Risk Scoping Study PERFORMED BY:
Sa n dia National La borato ry Principal Investigators:
Mike Bohn Steve Nowien l
John Lambrig ht Vern Nicolette
FIRE RISK SCOPING STUDY PURPOSE:
To assess the risk significance and dominant sources of uncertainty associated with fire ris k issues.
CURRENT STATE-OF-THE-ART METHODOLOGY MOST U P-TO-DATE DATA APPENDIX R
BACKFITS CONSIDERATION OF POTENTI AL FIRE RISK ISSUES
v O
O O
l RESU LTS FOCUS ON:
i MAJOR FIRE RISK CONTRIBUTORS.
ROOT SOURCES OF UNCERTAINTY.
f I
4 I
COM PLETEN ESS i
r 1
i
O O
O CHRONOLOGY
- Jan, 1987 Fire Risk Scoping Study initiated.
- June, 1987 Tasks 1
& 2 Completed.
J uly, 1987 ACRS Subcommittee on Auxiliary Systems.
Aug u st, 1987 328th ACRS Full Committee.
- Jon, 1988 First D ra ft of Fire Risk Scoping Study.
)
- Feb, 1988 Peer Review Meeting.
I M a rc h, 1988 ACRS Subcommittee on Au xiliary Systems.
l Ap ril, 1988 Secon~d Dra f t Fire Risk Scoping S tu dy.
i
- May, 1988 337th ACRS Full Committee
O O
O I
l l
]
FIRE RISK SCOPING STUDY OVERVIEW AND RECOMMENDATIONS i
l BY i'
MICHAEL P, B0HN 4
PRESENTED TO I
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS i
I i
MAY 5, 1988 1
i
1 O
o o
1 l
OBJECTIVES 1
l THIS STUDY IS INTENDED TO l
- PROVIDE A UNIFIED RE-EXAMINATION OF PAST FIRE PRAS INCLUDING
)'
NEW NRC-SPONSORED DATA BASES 1
- DETERMINE THE RISK SIGNIFICANCE OF A NUMBER OF PREVIOUSLY-UNADDRESSED FIRE ISSUES
- ASSESS THE IMPACT OF APPENDIX R RULES AND MODIFICATIONS ON I
l THE DOMINANT ACCIDENT SCENARIOS IDENTIFIED FROM A PRA PERSPECTIVE.
i l
t
1-i O
{
D W
M M
V M>
m b
M 4
a.
~
s M
>=
D M
A E
E t
- f. =
E&
w m
4 m
W e
[
it et 3
4 m
r 1
E m
W g g
o wt E ~w W
@i
= Em a
w
_,i
=
W 1
C E
eC h
hw 9
E
=
-w 1
K
=
W y
et e
O M
E W
E m
l W
W W
W 4
U J
E l
g eC m
W gm b
~
=
kw W
4 m
a
=
W g4 N
W M
(
o W
+
p M
E g
iE w
l
=
m W-m gl m
b=WD3 w
m o
W e
m m
W D
m d.
N 4
w i
.)
M N
M T
It) e e
m e
m
.i et 4
et et et 1
I I
O i
1 1
q
j.
O O
O 1
l CONCLUSIONS AND RECOM4ENDATIONS WILL BE PRESENTED I
i FOR EACH AREA STUDIED IN THE SCOPING STUDY I
J t
j PRA REQUANTIFICATION l
NEW ISSUES l
MANUAL FIRE FIGHTING EFFECTIVENESS CONTROL SYSTEMS INTERACTIONS i
TOTAL ENVIRONMENT EFFECTS (SMOKE & INADVERTENT SUPPRESSION) l FIRE CODE ADEQUACY B RRIER EFFECTIVENESS I
SEISMIC-FIRE INTERACTIONS j
I i
I i
,,.,----.,r-r---
,--.,----m-.,-
-..e
,,--.,-.---r
-- ~
~w.-.
j.
l
'O o
o j
TASK 1 AND 2 FIRE PRA REQUANTIFICATION SC PE 0
)
- WE HAVE REVIEWED LIMERICK, SEABROOK, OCONEE & INDIAN l
POINT PRAS l
j
- MEW DATA WAS INCORPORATED IN
- FIRE INITIATOR FREQUENCIES
- COMPBRN III FIRE GROWTH CALCULATIONS l
i
- NEW FIRE SUPPRESSION MODEL l
- BEST ESTIMATE CABLE FAILURE TEMPERATURES l
l
O O
O i
J _MPORTANT LIMITATIONS OF TASK 1 AND_2_ WORKS _COPI
- PLANT-SPECIFIC "PARTITIONING FACTORS" BASED ON JUDGEMENT AND THE R UNCERTAINTY WERE NOT UPDATED J
- IF NO. UNCERTAINTY ANALYSIS IN ORIGINAL PRA, NO ATTEMPT WAS MADE TO CONSTRUCT ONE FOR COMPARISON WITN REQUANTIFIED RESULTS l
l l
- NO "NEW ISSUES" WERE INCORPORATED l
l I
~
1 l
1 l
l-
4 O
O O
RECOMMENDATIONS TO IMPROVE FIRE RISK PERSPECTIVE
- DEVELOP DEFENSIBLE ESTIMATES OF FIRE OCCURRENCE FREQUENCY UNCERTAINTIES REFLECTING PLANT-TO-PLANT VARIATIONS (HOUSEKEEPING, l
l FIRE WATCH PRACTICES, ETC.)
l l
- DEVELOP DEFENSIBLE GUIDELINES FOR PARTITIONING AND FIRE SEVERITY l
FACTORS
- VALIDATE C'OMPBRN AND MORE ADVANCED FIRE CODE AGAINST ACTUAL FIRE l
ENCLOSURE TEST DATA i
i I
- UPDATE AND MAINTAIN FIRE DATA BASES (OCCURRENCE, SUPPRESSION AND 1
l COMPONENT DAMAGE THRESH 0LD DATA) l
- PERFORM FIRE PRAS AT SEVERAL PLANTS - INCLUDING PLANT WALKDOWN AND THE "UNADDRESSED" ISSUES - TO GET A TRUE PERSPECTIVE OF THE FIRE RISK i
l
l O
O O
i i
J REC 00MENDATIONS TO IMPROVE MANUAL FIRE FIGHTING EFFECTIVENESS L
- TEST CABINET / CABLE ROUTING COMBINATIONS TO ASSESS FIRE SPREAD TIMES (RESEARCH) 1
- DEVELOP GUIDELINES FOR REQUIRED RESPONSE AND EXTINGUISHMENT TIME FOR VARIOUS EQUIPMENT TYPES (RESEARCH) f EVALUATE FIRE BRIGADE EFFECTIVENESS VIA FIRE SIMULATIONS AT SEVERAL a
i j
PLANTS TO. ASSESS i
i
- OPERATOR VS FIRE FIGHTER SYSTEMS AWARENESS i
l
- TIME TO RESPOND l
- ROOM ACCESSIBILITY (KEYS, ETC.)
(RESEARCH) i l
l l
PLANT SPECIFIC MAPPING OF FIRE AREAS VERSUS CRITICAL RESPONSE TIME (UTILITY) f i
L d.___.,.__,.-,_____-__.
O O
O RECOMENDATIONS TO IMPROVE MANUAL FIRE FIGHTING EFFECTIVENESS (CONT'D)
- PLANT-SPECIFIC TRAINING ON AREAS WITH EQUIPMENT SUSCEPTIBLE TO SPRAY - USE OF PROPER FOG N0ZZLES - AND AREA COMBINATIONS WHERE SMOKE SPREAD AND AUTO SUPPRESSION ACTIVATION COULD HAMPER ACCESS.
(UTILITY)
- EVALUATE FIRE BRIGADE EFFECTIVENESS UNDER ACTUAL FIRE CONDITIONS INCLUDING DENSE AND T0XIC SM0KE.
(RESEARCH) l
2 O
O O
4 j
RECOMENDATIONS TO IMPROVE FIRE CODE ADEQUACY l
i i
VALIDATE CORRELATIONS IN COMPBRN III AGAINST DATA (RESEARCH)
L k
BENCHMARK COMPBRN III AND CHAM CODES AGAINST FIRE ENCLOSURE TEST DATA l
4 TO DETERMINE THEIR LIMITS OF APPLICABILITY (RESEARCH)
I DEVELOP CONTROL VOLUME OR FLOW CODE CAPABLE OF ACCURATELY PREDICTING COMPONENT DAMAGE TIMES.
BENCHMARK AGAINST FIRE ENCLOSURE DATA.
ASSESS UNCERTAINTIES IN CODE PREDICTED RESULTS.
(RESEARCH) l DEVELOP CONSISTENT AND VALIDATED CRITERIA FOR PREDICTING CABLE / COMPONENT DAS*GE IN FIRE CODES.
i f
l
[
l
~
\\.
l O
O O
RECOMtENDATIONS TO REDUCE CONTROL SYSTEMS INTERACTIONS DEVELOP A METHOD & GUIDELINES FOR DETAILED REVIEW OF ELECTRICAL INDEPENDENCE BETWEEN CONTROL ROOM PANELS AND REMOTE SHUTDOWN PANEL (RESEARCH).
APPLY TO SEVERAL PLANTS (ESPECIALLY OLDER) TO EVALUATE LIKELY SOURCES OF CONTROL SYSTEMS INTERACTIONS (RESEARCH)
APPLY ON A PLANT-SPECIFIC BASIS (UTILITY)
PERFORM REVIEW 0F REMOTE SHUTDOWN PANEL INDICATIONS AND CAPABILITIES TO DETERMINE THEIR "ROBUSTNESS" OVER A WIDE RANGE OF ACCIDENT SCENARIOS.
PREPARE RECOM4ENDATIONS FOR ENHANCEMENT.
(RESEARCH)
O REC 0 MEN 0ATIONsFORTOTALENvANNENTSURVIVABILITYISSUESO SPURIOUS SUPPRESSION j
l
- PLANT SPECIFIC REVIEW 0F AREA COMBINATIONS WHERE SM0KE SPREAD COULD i
HAMPER FIRE FIGHTING, OR CAUSE INADVERTENT SUPPRESSION ACTUATION.
l (UTILITY).
- PLANT REVIEWS OF FIRE WATER DRAINAGE PATHS (UTILITY) l
- FURTHER REVIEW OF EXPERIENCE BASE TO IDENTIFY VULNERABILITIES l
(RESEARCH)
FIRE ENVIRONMENTS
- EXPAND CABLE FAILURE DATA BASE, DEVELOP CONSISTENT FAILURE CRITERIA FOR USE WITH CODES (RESEARCH) t
- INVESTIGATE IMPACT OF IN-CABINET GASEOUS FIRE SUPPRESSION SYSTEMS ON i
j CONTROL CIRCUITRY (STATIC CHARGES, LOW TEMPERATURES) (RESEARCH)
I i
l t
j
- INVESTIGATE IMPACT OF SMOKE ON HIGH VOLTAGE EQUIPMENT (RESEARCH) l
O O
O l
l RECOMENDATION BARRIER EFFECTIVENESS ISSUE i
EVALUATE RELIABILITY OF BARRIERS FROM EXPERIENCE AND TEST DATA i
i i
(RESEARCH) i l
i
- TESTING ON AGED PENETRATION SEALS UNDER FIRE CONDITIONS TO i
EVALUATE BARRIER RELIABILITY.
(RESEARCH) i i
I I
j 1
i
s O
O i
i RECOMENDATION SEISMIC FIRE INTERACTION ISSUE
- FORMALIZE GUIDELINES FOR IDENTIFICATION OF POTENTIAL SEISMIC-FIRE VULNERABILITIES (RESEARCH)
J
- PERFORM WALKDOWN OF PLANT TO ELIMINATE POTENTIAL SEISMIC-FIRE j
l INTERACTION MECHANISMS.
(UTILITY) l 1
I l
l I
l l
l l
1
-r a
__-a,.
__m m
-s._w
-a
- w em h---a a%
m.
- a.2 m
---A
-n-A--&-A m-.L m-E M
-^*- -+
0 O
O FIRE RISK SCOPING STUDY RESULTS FROM TASKS 1 AND 2 BY JOHN LAMBRIGHT PRESENTED TO ADVISORY COMITTEE ON REACTOR SAFEGUARDS MAY 5, 1988 l
I l
O O
O FIRE RISK SCOPING STUDY. RESULTS FROM TASKS 1 AND 2 i
- REQUANTIFICATION OF FIRE SCENARIOS IN FOUR PREVIOUS FIRE PRAS LIMERICK OCONEE SEABROOK f
INDIAN POINT 2 i
i
- ASSESS UNCERTAINTIES IN FOUR PREVIOUS FIRE PRAS I
t l
i -
i O
O O
i l
l FOUR MAJOR AREAS OF UNCERTAINTIES IN PREVIOUS FIRE PRAS i
- FIRE FREQUENCIES & PARTITIONING i
i i
1
- TRANSIENTS VS. LOCAS l'
l I
- FIRE PROPAGATION MODELING
- FIRE SUPPRESSION MODEL e
l l
i
.I
)
i 2
~
i O
o O
I j
PAST PRA FIRE FREQUENCIES
- BASED ON HISTORICAL DATA FROM US LWR EXPERIENCE F
- JUDGEMENTAL MODIFICATIONS USED TO FREQUENCIES DERIVED FROM SUCH l
DATA l
- PARTITIONING NECESSARY TO OBTAIN FIRE ZONE SPECIFIC INITIATING EVENT FREQUENCIES
- UNCERTAINTY IN FIRE INITIATING SOURCES
- MAJOR CONTRIBUTION AT LIMERICK TO CORE MELT (87%) FROM f
l SELF-IGNITED PANEL FIRES AND SELF-IGNITED CABLE FIRES I
- PL&G CONSIDERED TRANSIENT OIL FIRES WOULD DOMINATE CORE DAMAGE FREQUENCY l.
4 l
i O
O O
i l
PAST PRA FIRE _ PROPAGATION MODELING UNCERTAINTIES
'l
- COMP 8RN I USED IN ALL FOUR ORIGINAL PRAS
[
L
- PARAMETER UNCERTAINTIES I
- QUANTIFIED USING LATIN HYPERCUBE SAMPLING FOR INPUT j
VECTORS E
b
- MODELING UNCERTAINTIES i
y
- PL&G PRAS USED AN ASSUMED CORRECTION FACTOR MULTIPLIER OF 2 ON DAMAGE TIME (ASSUMED LOGNORMAL) 1 1
- REGARDED AS OVERLY CONSERVATIVE BECAUSE THE ORIGINAL j
VERSION NEGLECTED HEAT LOSSES FROM FUEL ELEMENTS 1
+
4 p
}
o o
O l
!l l
EIRE SUPPRESSION UNCERTAINTIES I
O 1
t
- ENGINEERING JUDGEMENT IS USED TO CONSTRUCT SUPPRESSION MODEL 0
t BASED ON LIMITED DATA l'
- ACTUAL PLANT DETECTION AND SUPPRESSION FEATURES ARE CURSORILY INCLUDED IN CONSTRUCTION OF SUPPRESSION MODEL l
4
- TNE POTENTIAL RISK FROM MISAPPLICATION OF SUPPRESSION AGENTS IS i
l IGNORED I
I I
I t
^
^~^
^ ~ ^ ~ '
^:
~
^ ^ ~ '
I O
g Mb<
bd aC M
4 M
A E
W i
M W
b a
a.
Seg E
M o
J W
W E
E u
W aa
-l W
C i
J V
=J C.
M Q
A W
M W
W SC i
W ME M
ew she O-x M
in N
W W
eC Ch La M
C OC U
i i
to J
M c
E 6
J SC at W
W J
4 O
O J
M l
M E
E O
N i
O M
U C
>==
j D
H E
b-en M
i
=
E W
M M
4 W
O h
5 k$
e M
SC E
to
.1 W
M W
e e
E to M
M
=J 4
a I
r 1
l O
i l
I i
l
i O
O O
j REQUANTIFICATION OF FIRE SCENARIOS IN FOUR PREVIOUS FIRE PRAS BASED ON CHANGES IM I
L i
- INITIATING EVENT FREQUENCIES l
l
- COMPBRN CALCULATIONS
- SUPPRESSION MODEL
- CABLE DAMAGEABILITY TEMPERATURE I
i i
i r
j (3
GENERALAREASIITRE-EVALUATED
()
i' WHICH COULD HAVE A MAJOR IMPACT ON CORE DAMAGE FREQUENCY
- INITIATING EVENT FREQUENCY CALCULATION METHOD
- PARTITI0FING OF FIRE FREQUENCY FROM A LARGE BUILDING TO A i
i FIRE AREA WITHIN THE BUILDING l
i
- INITIATING EVENT DEFINITION (TRANSIENT VS. LOCAS)
- SEVERITY RATIO i
i
- AREA RATIO WITHIN FIRE ZONE i
i
- MANUAL VS. AUTOMATIC SUPPRESSION
l-o o
o UNCERTAINTY ANALYSIS
)
l
- UTILIZED THE TOP EVENT MATRIX ANALYSIS CODE (TEMAC) i l
- PERFORMED FOR ALL FOUR PLANTS EVEN THOUGH LIMERICK AND OCONEE i
l PRAS ONLY LISTED POINT ESTIMATES i
i i
i i
l l
i
Indian Point 2 Sequence Frequency Distributions (yr*1)
Location /Dasage PRA Requantified State Distribution Distribution Switchgear Room 5th Percentile 1.8E 8 2.8E-7 Median 5.4E 6 1.3E-5 Mean 5.6E-5 1.4E-4 95th Percentile 1.3E 4 7.8E-4 Electrical Tunnel SLF 5th Percentile 2.1E-8 2.3E 6 Median 3.7E 6 2.0E 5 Mean 2.4E 5 1.1E 4 95th Percentile 6.7E-5 4.3E 4 SET Sth Percentile 2.4E 10 2.7E-8 Median 3.7E 5 1.7E 7 Mean 1.0E 7 5.9E 7 95th Percentile 3.2E 7 1.6E 6 SE 5th Percentile 1.9E 8 1.2E 7 hh Median 6.1I 6 5.9E-6 Mean 5.6E 5 6.9E 5 95th Percentile 1.2E 4 4.1E 4 Cable Spreading Roo=
TEFC 5th Percentile 1.9E 10 2.5E 10 Median 6.0E 8 5.CE 8 Mean 1.6E 6 8.5E 7 95th Percentile 4.0E 6 3.6E 6 TE Sch Percentile
<1.0E 10 3.2E 11 Median 7.2E 9 7.3E 9 Mean 3.0E 7 1.1E 7 l
95th Percentile 5.9E 7 4.7E 7 O
i
~'
t O
O O
FIRE INITIATED CORE DAMAGE FREQUENCY (YR-1) i AS REPORTED REQUANTIFIED i
LIMERICKa 2.3 E-5 1.6 E-4 i
INDIAN POINT # 2a 6.5 E-5 2.0 E-4 OCONEEa 1.3 E-5 2.0 E-5 SEABROOKs 2.1 E-5 4.6 E-5 A POINT ESTIMATE a MEAN
o o
O REQUANTIFIED FIRE INITIATED CORE DAMAGE FREQUENCY (YR-1)
WITHOUT APPENDIX R WITH APPENDIX R l
MODIFICATION MODIFICATION 1
l LIMERICK 1.6 E-4 5.9 E-5 j
INDIAN POINT # 2 2.0 E-4 8.8 E-6 i
l OCONEEa 2.0 E-5 l
SEABROOKA 4.6 E-5 a APPENDIX R MODIFICATIONS COMPLETED BEFORE COMPLETION OF PRA l
l l
l 4
I
i O
O O
FIRE INDUCED CORE DAMAGE SCENARIO FREQUENCIES ARE STILL HIGH BECAUSE l
1
- FIRE OCCURRENCE FREQUENCIES ARE HIGH l
(AUX BLDG 6.4E-2/YR, CONTROL ROOM 4.4E-3/YR, CABLE SPREAD 2.7E-3/YR) l l
-l l
- PLANTS HAVE CABLING "PINCH POINTS" WHERE CABLES FOR MULTIPLE i
l AND REDUNDANT SYSTEMS CAN BE AFFECTED BY A SINGLE FIRE ALTHOUGH, DUE TO APPENDIX R, RANDOM FAILURES ARE INVOLVED
- EXPERIENCED BASED PROBABILITY OF NON-SUPPRESSION VALUES HAVE INCREASED SINCE ORIGINAL PRAS HAVE BEEN PERFORMED PRAS PERFORMED.
i l
l o
o O
TASKS 1 AND 2
SUMMARY
- REQUANTIFIED CORE MELT FREQUENCY INCREASED FOR ALL FOUR PRAS I
- INDIAN POINT #2 AND LIMERICK > 10-4 YR-1 l
l l
- EVEN WITH APPENDIX R MODIFICATIONS, FIRE INDUCED CORE DAMAGE FREQUENCY STILL AN IMPORTANT CONTRIBUTOR
- LARGE UNCERTAINTIES STILL EXIST j
- COULD BE REDUCED BY APPLYING A CONSISTENT METHODOLOGY
);i TO ALL FOUR PRAS WHICH IS NOT CONSTRAINED BY THE l
)
ORIGINAL ANALYSIS ASSUMPTIONS l
l
O O
O FIRE RISK SCOPING STUDY I
PEER R EVI EW 4
l (Fe bru a ry 23, 1988) j Review Members in Attendance:
j B ria n Dola n,
Duke Power 4
Mike Ka mins ki, Wisco n so n Electric Power J e rry P hila b a u m, Philadelphia Electric Co.
l l
J ea n - Pie rre S u rsoc k,
EPRI j
i Additional Comments:
Eddison Electric Institute Fire Protection Committee 4
l Arizo n a P u blic Se rvic e Company i
O O
o GENERAL COMMENTS i
l l
t j
- Research should find the pitf alls so utilities i
j know what to do in order to lower risk.
4 i
)
Management is concerned with what they are getting for their money (cost effective sa f ety).
I
- Will we see o real reduction in risk or lust a fine tuning i
of the uncertaintles?
i Difficult to extrapolate test data to real world l
l situations; actual experience is a better Indicator.
Questionable whether limited nuclear experience i
j 1s sufficient to draw conclusions.
I
O O
e
GENERAL COMMENT
S (Contin u ed) l Concern over large uncertainties in botto m li n e,
especially when co n side rin g th e bounding analysis.
Questionable whether future research will be able to n a rrow u n certainties much f u rt he r.
pessimistic viewpoint which m ay be Study has a misinterpreted and result in unnecessary backfits.
Potential issues are unaddressed by the selected method of risk evaluation o n ly, and do not re present u na d d ressed risk issues for nuclear plants.
0 D
v l
MANUAL FIRE FIGHTING COMMENTS Since the Browns Ferry fire there has been significant improvemenis in training and equipment.
Disagreed with dedicainJ vs. non-dedicated fire brigad e conclusions.
l Cannot determine ef f ectiveness by simply knowing the number of brigade me m be rs.
Regarding training deficiencies, need to expand on what l
l was actually being performed in the field.
- Proper smoke tower training con address the types of problems which the brigad e may encounter, i
i 1
(
m=m m
l O
O O
I i
CONTROL SYSTEM INTERACTION COM MENTS 1
- Operators should be given more c redit f o r recovery.
- Systems Interaction analyses are detailed and very thorough.
Cheaper to go In and physically separate systems rather than trace down Interactions.
l Disagreed on the time to detect control room panel fires,
- ventilation fans would blow smoke out
- ca binei sm o ke d etecto rs would trip t
- corridors are frequently travelled.
l 1
i
g COMPONENT SURVIVABILITY COMMENTS
- Should n oi conclude t h at some poo rly desig ned sys tem s represent industry wide p ro ciice.
- Important to dif f e re ntia te between types of sytems when modeling su pp ression.
- EPRI has issued a "R e q u e st For Proposol" to study the effects of fire suppression.
1
- Adequate design followed by a good maintenance p rog ra m will in su re high reliability of the fire i
su ppression systems.
l l
9 BARRIER SURVIVABILITY COMMENTS Based on previous (non-nuclear) experience, ba rrie rs have demonstrated excellent pe rf o rm a nce.
Actual fire perf orma n ce of fire ba rriers tested with ASTM
- methods, provide a staiistical basis for continued use of ASTM testing methods.
- Dif fe re ntial p ress u re across the barrier is believed to have lit t le impact on ba rrie r integ rity.
Analysis is conservative because the amount of combustible material available is limited.
i i
Hose stream tests and ti m e -te m p e ra tu re cu rves a re l
m ore conservative than some European tests.
y O
O l
i SEISMIC FIRE CO M M ENTS 4
l l
1 Results were h el pf ul to utilities, gave them j
something to look for at their plant.
(
The statement that "seismic / fire inte ra ction s has i
l largely been dismissed" is not accu rate-l 1
~
)
Data is insufficient to identify this issue as 1
l sig nifico nt.
1 i
1 i
g O
n.
i I
I i
l COM M ENTS ON ADEQUA.CY OF AN ALYTICAL TOOLS l
l 1
FOR THE TREATMENT OF FIRE i
- Agreed with the limitations that exist with regard to modeling fire condition.
1 i
- The entire repo rt is based on the failu re of COMPBRN lit to agree with available data.
I
1 4
The Fire Risk Scoping Study l
Overview and Conclusions 1,
of the NewIssues Investigations i
I Presented to the i
Advisory Committee on Reactor Safeguards i
May 5, 't988 By Steven P. Nowlen Sandia National Laboratories l
1 Conclusions Will be Presented for Each New Issue Studied in the Scoping Study 1
l New Issues l
Manual Fire Fighting Effectiveness Control Systems Interactions Total Environment Effects j
(Smoke & Inadvertent Suppression) l Fire Code Adequacy Barrier Effectiveness Seismic-Fire Interactions i
e i
l l
Potential Core Damage Frequency Impact of the i
"Unaddressed" Fire Issues Was Assessed Potential Issue Increase in CD Anolicability l
Manual Fire Fighting Effectiveness 0(10)
Generic Control Systems Interaction 0(10)
Generic Total Environment Survivability Smoke Control
?
Plant Specific
- l Inadvertent Suppression
?
Plant Specific
- Collateral Damage Small Generic Fire Code Adequacy 0(10)
Generic
- i l
Barrier Effectiveness 0(10)
Generic l
Seismic Fire Interaction Small Generic-Could Eliminate By i
Walkdown g
- Related to Fire Fighting Effectiveness i
I i
Conclusions on Manual Fire Fighting Effectiveness Must Suppress Fires in 5-20 Minutes to be Effective 1
1 Many Areas Exist Where Critical Damage is Likely Before Fire Brigade l
can Effectively Reach Them Fire Brigade Size and Training Vary Widely i
i Knowledge of Critical Plant Systems Crucial i
l l
Dense Smoke Will Hamper Efforts l
1 Fires Suppressed by Halon or CO2 Likely to Re-ignite
Conclusions on Fire Code Adequacy Accurate Calculation of Time to Component Damage is Critical to Assessing Fire Suppression Response Time.
No Existing Fire Code Benchmarked Against Actual Enclosure Fire Tests Even with SNL Fixes to Remove Obvious Deficiencies, COMPBRN III Gives Results Sometimes Conservative - Sometimes Non-Conservative
)
Cable / Component Fire Damage Threshhold Data Very Limited.
l l
l
Conclusions on Control Systems Interactions Even with Electrical Independence Between Control Room and Remote Shutdown Panel Functions, Subtle Interactions Involving Control Room Panel Fires and Random Failures can Cause Significant Core Damage Scenarios.
It is Very Difficult to Assure Electrical Independence.
Many Remote Shutdown Panels have Limited Capability. Lack of Good Indication at RSP may Increase Risk.
t
a_
C Conclusions on Total Environment Survivability
- Smoke Control Tests Show Total Loss of Visibility in 5-8 Minutes Given Typical Cabinet Fire Normal Control Room Ventilation Rates not Effective in Preventing Loss.of Visibility.
Experience Shows Smoke Will Hamper Manual Fire Fighting.
Smoke Spread Could Cause Inadvertent Fire Suppression System Actuation and Hamper Fire Brigade Access.
Potential Impacts Very Plant Specific.
w Conclusions on Total Environment Survivability
-Inadvertent Suppression Past Experience Shows Inadvertent Suppression Actuations Occur with Frequency About 10-2 Per Year One Case of LOSP One Case of Control Rod Withdrawl One Case With Redundant Systems Lost No Cases Associated with Fires Conclusion
- Without Fire, No Significant, Generic Safety Impact
- With Fire, Small Impact, Very Plant Specific 4
~
i
~
m q
i i
l Conclusions on Barrier Effectiveness Issue Penetration Barrier Failures are Main Issue. Aging Cracks or Barrier i
Test Methods May Result is Less Barrier Reliability than Anticipated.
If Penetration Barrier Reliability Less Than 90%, Could Increase Core
~
j Damage Frequency by 0(10).
i If Reliability Above 99%, Random Barrier Failure Frequency I
Dominates.
l j
i I
i i
i s
i
a l
i Conclusions on Seismic Fire Interaction Issue No Data on Frequency of Fires During Earthquakes at Nuclear Power Plants Risk Due Primarily to Systems Interactions Which Could be Identified During Plant Walkdown.