ML19270H429
| ML19270H429 | |
| Person / Time | |
|---|---|
| Issue date: | 09/07/1978 |
| From: | Bradford P, Hendrie J, Kennedy R NRC COMMISSION (OCM) |
| To: | |
| Shared Package | |
| ML19270H427 | List: |
| References | |
| NUDOCS 7907270220 | |
| Download: ML19270H429 (66) | |
Text
I
~
J 1
E m.
T'#- l
- - - N U C1.E A R REG U L AT O RY CO MMIS SI O N i
IN THE MATiER OF:
3EETING WITH RISK ASSESSMENT REVIEN GROUP o
Place - Washinct n, D, C.
Octe -
'"h=sday, Septe=e: 7, 1978 P=ges 1 - 64 l
790727e220 7.enen.:
',$T10A?.550 AC5 -:: uE?J.L E.E?0?;.
"S, DiC.
l OTc.nlRepor:ers t
i
.: Lt Ncrih C==irci Streer l
Weshing =n, C.~ 20001 NAT:C NWIC E COVY.uGE - O A!!.Y
.il'f..
1 t
t.
s DISCl. AIMER This is. an unofficial transcript of a meeting of the Unf ted Sta'as Nuclear Regulatory Commission held on Septe=ber 7, 1978 in the Ccmmission's offices at 1717 H Street, n.
W., Washingr.ca, D. C.
The meeting was'open to public' attendance and observation.
Th'is transcript' has'not been reviewed, corrected, or edited, and it may contain inaccuracles
( '.
The transcript is intended solely for eeneral informa'tfonal purposes.
As provided by 10 CFR 9.103, it is not part of the femal or informal record of decision of the matters discussed.
Expressions of opinion in this transcript do not necessarily reflect final de~ terminations or beliefs.
No pleading or other paper may be filed with the Commission in any proceeding as the result of or addressed to any state =ent or arg: ment contained herein, except as the Commission may authorize.
g.
d e
t
t.
i' la jeti-r, 1.
UNITED STATES OF AMERICA cr 9244 2
NUCLEAR REGULATORY COMMISSION elaine 3
~
4 IIEETING WITH RISK ASSESSMENT REVIEW GROUP S'
6 Room.1130 1717 H Street, N.W.
7 Washington, D. C.
8 Thursday, September 7,.1978 9
The Commission. met, pursuant to notice, at 10:00 a.m.
10 BEFORE:
11 DR. JOSEPH M. HENDRIE, Chairman PETER A. BPADFOIID, Commissioner-12 RICHARD T. KENNEDY, Commissioner VICTOR GILIN3KY, Commissioner s
I
/
13 JOHN F. AHEARNE, Commissioner 14 ALSO PRESENT:
15 HERBERT J, C. KOUTS WILLIAM D. ROWE 16 HAROLD W.
LEWIS.
ROBERT J.
BUDNITZ 17 FFANK VON HIPPEL WALTER B. LOVENSTEIN 18 J, KELLEY 19 20 21 22
~
1 24 wreenen nonm. anc.
23
(-
6 2
- 9244, j,
PROCEEDINGS EJ 2
CEAIRMAN EENDRIE:'
If we could come to orde$, please 3
Our colleagues will join us soon.
The Commission meets this 4
morning to hear a report from its Risk Assessment Review Group.
5 This group was constituted over a year ago, and I believe, met 6
for the first time in late August, just a shade over a year,
7 ago.
8
.The group was constituted to review the reactor.
9 safety study, WASE-1400, and the various comments that had been 10 made upon it, and to. report to the Commission upon these matter 11 and p,erhaps more importantly, looking to the future, to develop 12 advice and recommendations for the Commission on the future f) 13 development and use of risk assessment methodology.
14 The group is headed by Dr. IIal Lewis of the Universi 15 of California, and i.ncludes Herber.t Kouts from Brookhaven 16 Laboratory, Walt Lowenstein from the Electric Power'Research 17 Institute, Bill Rowe from the Environmental Protection Agency, 18 Frank Von Hippel from Princeton, Frederick Zachareisen from
~
19 Cal Tech, who, I understand is in Geneva this morning, and is 20 perhaps enjoying that more than he might enjoy Washington still 21 in the sm mer heat.
22 DR. LEWIS:
Especially since he is on vacation today 23 CEAIRMAN EENDRIE:
And finally, Bob Budnitz, from 24 Lawrence Berkeley Laboratory, whom the Commission has lately co-Federed Rego,sers, Inc.
25 hired away'from tne university.
He will be put to work in
c' 3
8 1
research here, on the Commission staff.
2 We welcome you all to this meeting, and your report 3
has been long coming, and I know you have labored mightily 4
and not without difficulty to come to it, and the Commission 5
will be interested to hear you out.
6 Now, I will turn the floor over to you, and you 1,ead 7
us where you eould like to take us.
8 DR. LEWIS :
Well, thank you Joe.
I will say it
~9 is a pleasure to be here, and, because this really is the light 10 at the end of the tunnel.
Whether it is light remains for 11 you to decide.
I have here with me, copies of the summary to 12 the report, and as a m<wsure of how mightily we have labored
. )'
13 up the last monent.
Our' findings and recommendations are now i
14 being typed, and will be available to everybody in the ne:<t 15 few minutes, but certainly by the time we are finished.
16 On our full repert, we have some editorial changes 17 still to be made, but it will be available, I am confident, 18 wi+ hin the next few days, assuming that, the facilities of NRC 19 are up to the job, it's about 100 pages.
20 CEAIRMAN ENDRIE:
I assure you that printing papers 21 is one of the things that w: do best.
22 DR. L N :
We have learned during our study to che 23 I every'hing like that that anyone tell's us.
24 (Laughter. )
Am-Federal Repo,wrs. Irte.
25 DR. LEWIS:
As you said, we were constituted
e 8
1, as a result of an interchange between the former Chairman 2
of the Commission and Congressman Udall, and as you recall, 3
our objectives were negotiated during.a period a year ago-4 Spring,.and started out with a request from Congressman Udall,
5-.to rewrite the executive summary of WASH-14 00, the Rasmussen 6
report; the Commission after some consultation responded.that y
that would not be a useful~ expenditure 'of time, and instead 8
constituted this group with a charter to do several things, 9
one of which was to clarify'the achievements and limitations 10 of WASE-14 00, 'th:< Rasnussen report, and another was to assess 11 the peer reviews thereon, and another, which was to look at 12 the current state of that' kind of risk assessment methodology, l
,j 13 and make recommendations to you' on how and whether it can be il 1 integrated'into the licensing and regulatory processs.
15 I note with some amusement that when we took on 16 the job in July of last year, we said that we would be finished 17 by-December 31 of last year; I'm' not clear on whether we have 18 had two or three extensions, but the system has been very kind 19 in granting us more time.. We now don't want any more time, 20 because we are-finished.
21 We do have a spectrum of views on the group, as you 22 know, and therefore, the report, although our report will be 23 n9eimous, there will be a couple of things which were 24 negotiated, and a couple of caveats that one or more members cm.Feaeral Report,n. Inc.
25 may want to make on some of the findings.
,l 5
2 1
I will try to report to you on our unanimous views, 2
I'll try to report to you as' fairly as I can the views of the 3
committee, and not my own, which differ, of course, in some a
respects, and then perhaps at the end, to the extent that you 5
want to elicit whatever views others may hold, 'I think you sho'u 6
do =o.
7 Furthermore, if you should ask. questions to whi.ch 8
I don't know the answer, I will turn to my fri'. ends who _do.
9 That is more effective than making up the answers.
10 There. are two ways in which I could go about doing 11 this.,
I was given,a total of eight minutes to prepare this 12 presentation, and that's quite long enough.
I can do it.in.
)!
13 two ways:
one is to go through our findings and recommenda-14 tions when they come, but I think that would probably not 15 be wise, because each of them rests on a sol.id base in the '
16 report itself which we are not going to give you dor a few days 17 I think it would be more useful if I.were to go 18 through the repcrt the way it is laid out; tell you what I can 19 about what is going to be in the report when we release it to 20 you, and have it lead to the finding or conclusion or recom-21 mention which you will have in writing before you at this ' time.
22 The report will have only minnr changes from the 23 current version to what you will finally get.
Now, I have to 24 begin by saying that,we have had outstanding cooperation from u Federal Reportws, Inc.
25 - everybody.- We have taken on a job which everyone knows is
6 I
controversial, it's one on which a great deal of heat has 2
been expended, a lot of nasty words have been said, and th_ere 3
are contesting parties.
We jtst have to say that everyone
~
4 has been cooperative with us.
The nuclear community, the 5
staff of the Nuclear Regulatory Commission, the people who hate 6
nuclear energy, people who are indifferent, everyone has helped 7
us, and we are very grateful for that.
8 We have found that in the assessment of peer comment 9
on WASE-1400, the Rasmussen report, it has been vs y difficult 10 to get at the facts, because in a certain sense," the peer revie 11 process has gone downhill.
It was never very good, but it has 12 gone downhill on WASH-1400 in the sense that the kind of -
13 quality s+-=A=rds we in the technical community usually place 14 on peer review, these standards have not been upheld.
15 It has come to a point where anybody can say anythin 16 about WASE-1400, and it is counted by NRC as a peer review 17 co= ment.
So that the admixture of sound and meaningful peer 18 cocnents with those which are not sound.and meaningful has 19 produced a kind of hodge-podge through which we have had to -
20 burrow in.doing our job, 21 On the other side of the coin, we have found that 22 the staff of NRC has become extremely defensive about peer 23 co=ments.
Perhaps it is not surprising when you get a lot 24 of nonsense in addition to reasonable criticism, but it is a
.e. Federal Reporters, Inc.
25 fact that tihe normal response to peer co= ment - and I don't
7 j
mean that in a pejorative sense, it just seems to be that way --
2 has been a defensive response before trying to analyze the 3
comment, so there are sins on both sides here, which has made 4
it very difficult to get at the substance.
5 COMMISSIONER GILINSKY:
Let.me ask, are you talking 6
about comments before the publication of the report?
7 DR. LEWISI Both, both.
The period before the publi 8
cation of the report, the period between draft e..d final was 9
very short, and I will have a great deal to say about the 10 ! peer review process as we go along, but there was essentially 1
11 no tine between the draft and the final to make peer comments, 12 altho Igh a number of people responded, and we will say that 13 as we go along.
i.
14 Most of-the peer comments have come after the publi-15 cation of the final report, and in fact, many of them were 16 elicited during the testimony to our committee.
We dug a lot 17 out of the woodwork that was latent in the community,. so -
18 COMMISSIONER KENNEDY:
Will you~ explain why it was 19 latent, and not presented earlier?
20 DR. LEWIS:
I can't explain that, and there is no 21 point in my guessing on.your time.
It is simply a fact that 22 there is a lot out there, and the communication on this par-i 23 ticular subject has not been good.
It has been poisoned by 24 a lot of extraneous issues other than technical review of a
- ..r.e.r.: a.corwei arc.
25 technical " report.
I
w 8
1 On top of that, WASH-1400 is a very complex and'long 2 i report.
It stands a foot high, which is half the thickness 3
of the draft version, but that's because it was printed on 4
thinner paper.
It is very hard to read, and all of us who 5
have tried to read and understand the drift of 'an arcfmemt 6
have had trouble, sometimes it takes years.
7 So-it is very difficult to accomplish,a serious 8
technical peer review, and the comments are therefore same-9 what garbled.
I will speak more about that as we go along.
10 Let ne,, then, if it is agraeable, sort of go. through 11 our table of contents, and tell you what the sort form of -
12 th,e report is, and then, as we go along, you can interrupt.
-)
13 freely, or my friends will interrupt.
14 By golly, it did work.
Here is a set of findings 15 and recommendations which have, I warn you have not been proof-16 read.
It may be occasionally, where a ".not" appears in a sen-17 tence, it has been left out.
18 (Laughter. )~
19 GAIRMAN HENDRT?,:
We will regard these then, as 20 being withirt say, oh, minus one, plus or minus one,,or a mere 21 reflection.
22 DR. LEWIS:
In all fairness,'they were agreed to i
.iby all of us, but as I say, have not been proofed, so if the 23 24 job has been done well, this is it.
I doubt if any " nots" have ce.Federst Reporters, Inc.
25
' been left out, but only time will tell.
In any case, those of
e e a ge 1
us who-have a stake in the findings and recommendations will e
2 notice if " nots" have been left out before we are through 3
this morning, so you can take these as real, although we will 4
proof them.
5 Let me then go through our report.
We have divided 6
the report into two major sections, in that we have a set 7
of sections which we think are of interest to the Commission 8
in its rolicy making role.
That is, there are things which 9
are meaningful in terms of our broad charter to you, which, 10 as I understand it, is to help you understand what has_been 11 done in risk assessment methodolony, specifically, WASH-1400, 12 and how you can.use it in the '.ature.
And the first.few
,j 13 roman numeral sections are gered to that purpose.
14 Then we recognized early in the game, that there 15 were a lot of things we needed to say in the report, but didn't 16 fit coherently into the main thrust of the argument, but still 17 couldn't be overlooked, for example, the Brown's Ferry fire.
la Issues of risk perception.
Things of that kind, so we'have 19 a collection of what we call disjoint special items at the 20 end, which are part of the text which are agreed to by all 21 of us, except where noted, but which you should read if you 22 want to know the background, but you.lon't need to do your job.
23 We have fiEst, a section on risk assesment methodolo 24 There has been, as you know, a lot of flak about the method-w r.o m i n.co n m.inc.
25 ology used'in WASE-1400, which is the so-called fault-tree / eve.
.i 10 I
tree methodology.
To remind you, an event-tree starts with sometl$ing that happens,and follows all possible futures from 2
3 it, whereas a fault-tree asks for the. misbehavior of a componen m
4 and asks for those pasts which could. have led to the misbehavio 5-WASH-1400 used a technique which was innovative, which consisted of putting the two together to find witat the
~
6 pcssible sequences were.
The methodology has been criticized.
7 8
We find that the methodology is-sound.
There are other com-9 peting methodologies which are not in logical structure, 10 I different, but which have different empha_ses.
11 The problem is that a methodology of this kind which 12 tries to look through accident sequences fo'r complex systems I
i sj 13 like a nuclear power plant, must deal with zillions of accident 14 sequences.
I don't know what zillions are, but they are certair 15 larger than billions.. And, it.is just physically impossible 16 to do that.
So that the alternative method logies that have 17 been discussed are really other ways of picking out of this intractable mass of accident sequences some, a small subset 18 19 that you can deal with, and for which you ~can make an argument 20 that it is enough, and I will come back to the question of lcompletenesslater.
21 22 There.is one class of things which are not in princi 23 but are in fact, not tractable through the fault-tree / event-tre 24 methodology, and there are things which involve continuum varia m.Feoers! Reportres. If%
25 for example, the fault-tree / event-tree methodology proceeds
yy j.
through, as you know, end gates and ore gates in the logic' 2
structure.
That is to say, one deals wit'- a valve by asking 3
whether it is open or closed.
Now in fact, a valve has many
.s
~
states, all of us who have dealt with valves knows it has many 4
5 states between open and closed,and the am;unt of leakage, 6
the character of leakage and such things is an important subjec 7
This is not, thi's is an example of.a continuum varinale which' 8
is not naturally present in the fault-tree / event-tree methodolo 9
But other than than the methodology we found is 10 sound, and should be used more.
The question of the specific 11 implementation of the WASH-1400 will be a good part of our sub-12 ject today.
We do think that NRC should use a sound applica w
O 13 tion of _he methodology,'which rests on the provision of an 14 adequate data base, and all those good things that are required 15 to make the methodology work, should use it more effectively 16 than it does in allocating resources.
17 For example, if it were to be learned in a responsib 18 way by a certain type of accident, is really not a major con-19 tributor to a risk in a reactor, you ought not to spend much 20 ; regulatory and licensing time on it, or spend much of your 3
21 research budget on it.
That is, if you can believe that.
22 There are many things one can learn usir.g this s
23 methodology which ought to help you match your rescurces agains 24 the risk of accident,in nuclear reactors.
We will find as we
' cm Federst Reporters, Inc.
25 go along, that although some steps have been taken in that
12 I
direction, you can do more.
2 WASE-1400 itself, and I'm going to give you the 3
imnediate one-liners before going on, WASH-1400 itself, we find 4
has lots of defects.
It was a very complex job, very hard 5
job.
It was done on a time scale, it was new.
There really 6
was an inadequate data base.
We think it was not done as well 7
as we would have desired.
I will go into great detail for a
the reasons for this statement.
9 We asked ourselves whether that means that we feel -
10 I that the risk of an accident portrayed in WSH-1400 was greater 11 or less than the truth, and we were unable to determine whether 12 it is too high or too low.
I doubt that we could do so even m
v 13 if we were given the three years and the megabucks that were 14 involved.
What we are certain of is the error bounds 'shown 15 in WASH-1400 for this risk estimate, are too small..
16 You may recall that the error bounds given were 17 a factor of five, high or low.
We are sure that that is 18 overconfident.
But, if you ask me, as I am sure you would 19 like to, how much overconfident, I can't answer that either.
20 We have found too many things either done badly, or done with 21 an inadequate data base to believe that the answer is within 22 a factor of five of the truth.
We don't know which way, but 23 we don' t believe that it is within a factor of 5, or that it 24 can be stated to be within -- but of course, fortuitously, it co Fooerst Reconen. Inc.
25 could be exact.
But, we don't believe that one can be confide.
.r 13 1
that it is within a factor of 5 of the truth.
4 2
Okay, we have then a section on statistics.
You know that there has been a great deal of commentary on th'e s
I I
~
4 statistics and techniques used in WASH-1400, and in particular, 3~
on the extensive use of what is called' a log normal distribu-6 tion, and the conclusion from it which is a technique for 7. handling in particular common cause failures, about which I wil 8
have more to say, which is the so-called square root bounding 9' technique, we find that the square root bounding technique 10 1 is an abomination.-
It just is rotten.
11 There is a problem, and the problem is that in some 12 cases, and the most egregious example is the case of the BWR i
m
\\_)
13 control rods. - In some cases there really are not enough data 14 to support a solid statistical support of the problem, yet,.
15 in the context 'of WASH-1400, one was required to provide a 16 numerical answer, and an error bound.
And just to use the 17 case of, the BWR control rods, as an example, I assume you are 18 - familiar with that case, and I don't need to spell it out, but 19 just to use it as an example, it was a case in which it was 20 judged that the failure of three control rods would pose 21 a problem for scram in a reactor, and the problem was therefore 22.i to compute the probability that any three control rods might 1
It 23 ' fail at the same time, any three adjacent control rods.
The definition of adjacent is ambiguous in WASE-14 00 24 i n.=mn. inc. I 25 (but that's really a small point.
The technique used was.to
14 i
ask for the probability that a single control rod would fail, 2 I and to say that the r.ost optimistic element would be that three
! adjacent control rods would fail independently, and then take 3l 4
the initial probability and cube it to find the probability 5
that the three adjacent ones would fail independently.
6 I might just comment that parenthetically, even 7
here there is.an example of the conservatism which permeates 8
the whole report, which is thati the definition of failure for 9
single control rod was failure to insert completely.
An exampi 10 of a continuum variable; if it inserted half way it might
~
11 very well do a good share of the job, but failure to insert 12 completely was used as the. definition of' failure.
]
13 The lower limit of the estimate that the three 14 adjacent was just the cube of 1. The upper. limit would be an 15 estimate based on the single failure probability, with some' 16 estimate for what fraction of the time cr-nn cause failures 17 would have caused that single failure probability.
That 18 number was given as 1 percent in the ' report.
19 Then, Le final answer was taken by taking the pro-20 duct of these two co=pletely different estimates, based on 21 completely different models, taking the product, and taking 22 the square root, that's the square root bounding model.
We 23 think it is just very arbitrary unless one has a sufficient 24 data base to establish a log normal distribution, it makes no
=e Federal Reporters, Inc.
25 sense.
And yet, one has to ask what one can do better if
15 EJ 1
one lacks a data base on scram failure.
2 Well, there are many things one can do, and as.you 3
note, the staff has recently issued a NUREG-0460, which is 4
based on the observation that a particular German reactor has 5
had a potential scram failure in the sense that the scram 6
system was disabled, and had it been called on, which it was 7
not, it would have failed, and they have used that one statisti 8
as a data base to assess the probability of scram failure.
9 It turns out to be much larger than WASH-1400.
10 That is; probably better, but it is not clear. on what 11 one can do, and now I speak personally for myself., and not 12 for the group, I think one would do much better in the case 13 in which there isn't a data base adequate to support a numerica 14 answer, not give a numerical answer, or give a bound on a 15 numerical answer rather than force oneself through questionable 16 procedures into an answer.
In any case, we deal with many 17 of these questions in our section on statistics.
18 Question of completeness is our next section.
Again 19 one that has been leveled at WASH-1400, the issue is that amon 20 the billions or =illions of sequences, were they all accounted 21 for.
The answer is obviously, no.
No one can account for them 22 So, you have to ask yourself what do you mean by 23' completeness.
Do you Imean that these people stopped too soon; 24 do you mean they left out significant sequences; do you mean eeJederal Reporters, Inc.
25 ;
that they didn't add up the sequences properly.
What do you
16 mean?
2 I guess our-concensus view is that we observed that 3
no new significant sequences have come up since WASH-1400.
s i
4 That's not ent$irely true.
Some sequences have been brought 5-up not contained in WASH-1400, and many of those have not 6
been quantified, so one doesn't know whether they are sig-7 nificant.
8 At least one sequence has occurred in practice. -- th 9
Brown's Ferry fire.
Which is not an important part of WASH-140 lo ! It is a common cau'se failure, which we will come back to under.
11 its specific title a bit later.
But still, in the sense that 12 somebody has said that valve A could lead to.a failure type J3 B, which could lead to a loss of material C.
That hasn't 14 really happened.
15 Yet, we'all have, I think we are all of mixed minds-16 on this.
We haven't seen something new come up; we can't 17 prove completeness in any reasonable sense, and we have a 18 nagging concern.
It is not a quantifiable concern,. but it's 19 a nagging concern.
Time may help, but the issue of completenes 20! remains open.
21 I think it is fair to say we have no evidence that 22 major + hings were left out of WASH-1400, but the. question is 23 almost in principle not a closable question.
There's the 24 section on co=plet'eness.
- M*r=' Reporters, Ire.
25 The section on data base in many cases is weak.
17 1
We encourage NRC to fill in the crucial points, but that yo'u 2
would have known anyvaf.
3 The section on health effects, or disease effects.
4 We have a controversy about whether to call it health effects, 5
which is the traditional terminology, and it has been pointed 6
out to us that there.are no health effects from. radiation.
7 On the other hand, probably that hasn't been established 8
beyond doubt.
9 CHAIRMAN EENDRIE:
That seems a stride forward..
10 DR. LEWIS:
Well, we decided not to quarrn about 11 that.
We had enough thing to quarrel about.
We understand 12 that there are effects on health, and that they are generally.
13 bad effects.
14 Here the issues are very complex.
It is a curious' 15 paradox that probably more is known about the effects of 16 radiation on organisns, than is known about most other disease 17 effects.
Not all, but most other disease effects.
But, becans 18 it has been an important subject, there is almost luxury of 19 data, and competent people working in the field; yet, it',is 20 controversial and the answers are not in, and the specific 21 issues we spent time on two issnes, One is the acute LD 50-60 22 issue, which has to do with fatalities from an accident, and 23 I guess in my mind, th'e more important question of latent fa-24 talities.
" ::n.Fedevel Reporters, ite.
25 I might, parenthetically say here, that une of,the
18 I
great achievements of WASH-1400 was to highlight the latent 2
fatalities-of a reactor accident really for the first time.
3, It did highlight them.
There was controversy about how 4
they were calculated, but they were certainly highlighted.
5 The latent cancers in particular are the subject of a great deal of controversy in the field.
The issues go 6
7 on the names of the linearity hypothesis, the plateau question, 8
the relative versus absolute risk codels.
We have a great 9
deal to say about those; we simply found that the iss as are 10 controversial.
We.' had parades of experts come before us, and 11 generally speak well of each other, but disagree politely.
12 We are not able to _ tell which one is right.
13 There are real uncertainties in the field.
The s
14 uncertainties are not a big factor.
We have decided for our-15 selves that, at worst, the uncertainties in the lat: 4t, effects 16 could lead to a factor perhaps of 5, just bridging the entire 17 spectrum of uncertainty, in the nu=ber of latent cancers from 18 the accident, but probably not more than that.
19 But even that is a judgment call-based on our having 20i read the experts, and I think we a e simply going tc live with 21 that uncertainty for a long time.
How one makes public policy I
22 in the fact of uncertainty is your business, to which we will 23 try to contribute.
24 Then we have, after the discussion of health, we ee Federal Recorters, Inc.
25 have a set'of disjoint items.
yg 1
First and foremost, Brown's Ferry.
The question 2
of Brown's Ferry, apart from the fact that it'is a very large 3
accident from which one ought to be able to. learn a great. deal, 4
the popular question is was it a near miss.
And that is the 5
terminology often used.
We have decided that that is not a 6
useful question.
And it is not a useful question for'a lot y
of reasons.
8 I'd like to go into a bit of detail on this, because 9
it is an important disjoint item.
It is related, I might say 10 to WASE-1400 because, again, to remind you, there is contained 11 in Appendix 11 of WASE-1400 a discussion of the Brown's Ferry 12 accident, which, of course, occurred'between the draft And
]
13 the final report, and the discussion does the following.
It 14 takes the existance of the fire as given, and asks what the 15 probability is, using WASH-14 00 methods, that given the fire,,
16 there could have been a core melt.
17 The discussion concludes with the result that the 18 probability of a core melt was a part in 300, approximately, 19 given the sequence of event", the alternate sequences of events 20 that could have occurred.
21 The final conclusion' there is that one in 300 is 22 okay, because there has only been one such fire in roughly a i
23 i few hundred years of reactor experience, and therefore if you 24, multiply by those probabilities, one in 300, you get something ca.e.rw neco<wn. sac.
25 which is below the WASE-1400 estimate for a core melt, and
20 EJ that's the bottom line of the WASE-1400 Appendix 11 discussion.
j.
2]Wefindthediscussionuseless.
I don' t mean that, that's an overstatement, strike it.
3l We find the discussion not entirely to the point 4
5 for the following reason.
If you want to discuss alternate 6
futures, you must be.very careful to do your statistics reas,onal 7
well, and for. example, if a meteor were to come through the roof of this room and land here in the middle of the table, 8
9 I think we would not say the the Lord was mad at us for doing 10 what we are doing today, we~ would say that it was a random 11 impact of the neteor, and it could just as well have' landed on 12 the Department of Energy, as on the Nuclear Regulatory Com-13 mission.
14 (Laughter.)
15 DR. LEWIS:
But, in any, case, the, point is, that" 16 the point at which the meteor has its impact is truly a random j7 effect.
There is a certain level of -incidence of meteors on 18 the top of the atmosphere, which, incidentally, we know much 19 better 4-han we did at the time WASH-1400 was written, and it 20 makes a big to do about the relative probabilt,3y of being hit 1
21 on the head by a meteor, and being damaged by a nuclear acciden 22 The point is that it is a random event.
So, in orde to take a thing which 'has truly happened in a certain way, and 23 24 ask yourself whether it could have happened in some other way,
- z. =.F.o.r.: n.corien, inc.
25 you must look at the thing, and ask yourself where there were
21
'EJ 1
random events.
The real issue, putting it in the most extreme 2,
1.,rm, if Brown's Ferry were to happen again, it wc,uld end.the 3
same way Brown's Ferry happened, because Brown's Ferry happened 4
So, you must look very carefully for random things before you 5
ask what is the probability that something else, in this case, 6
a core melt, could have occurred.
-7 It turns out there are some random things,'but there 8
are not all that many.
And the analysis in Appendix 11 to 9
WASH-1400, assigns random variables in car view uncritically 10 to things which ar.e deterministic, and I will give you an Il example.
12 In the discussion of Brown's Ferry, you may recall, 13 in the event,. seven of the relief valves failed very early 14 in the fire, and the remaining four failed 5-1/2 hours into 15 the fire.
The period of greatest danger was when all the 16 relief valves were failed, and if that had happened.much 17 earlier when the reactor was losing water more rapidly, it 18 would have been a more serious situation.
19 So, one is concerned about whether these last four 20 relief valves could have failed earlier.
What was done in 21 WASH-1400, was to take the event, regard it as a random event, 22 fit it to an exponential decay law for the probability of the 23 event, and therei;y compute +'>at there was a probability of 63 24 percent that it could have failed a.arlier aan the 5-1/2 hours.
=-F.omi n.conm. inc.
25 Now, if you look at what caused it, it turns out
22 1
that the last four failed becasue they were pneumatic valves; 2 i they run off an air reservoir; the air reservoir ran out of 3
air, and so the last four valves failed.
That's the reason.
4 And that's a causal event, not a probabalistic one.
I can use 5
the same argu.'ent to prove that my car, which gets 250 miles
~
6 on a tank full of gas has a 15 percent probability of getting 7
500 next time I drive it.
And, we know that's wrong.
8 So, it's a simple example of the place where one 9
has used probability, where one shouldn't' have used probability 10 So, we regard the question of whether Brown's Ferry was a near 11 miss as not a viable question, unless one goes back and does 12 the deterministic things deterministically, and the random T
13 things, randomly.
J 14 On the other hand, one can learn a great deal from 15 Brown's Ferry.
It is the worst accident there has been, 16 and we think one should learn a great deal from it.
For 17 example, one does learn, as a fact, by reading WASE-1400, that 18 fires are not regarded as a major cause, or do not come 19 through as a aajor cause of reactor accidents in WAS'H-1400.
20 Yet, it is so, that the biggest one we have had was started by 21 a fire.
22 Even here, one has to be a little careful, because 23 I there is a statistical trap into which peop1'e fall.
We haven't 24 ; cleared this with the group, so let me say this on my own.
..--,q 25 There is a' statistical trap into which people fall which comes
23 i
from looking at a selected population of things which have 2
happened, and ask what is the probability that they could have happened, and concluding that the statistics are wrong.
'For 3
4 example, suppose there are a million different events that 5
could have happened, each of which has a probability of one 6
in a million:
the Redskins going to the Super Bowl.
You know, 7
your list of things which have a probability of one in a 8
million. Suppose there are a million of them.
This year, one 9
of them is going to happen.
And then if I look at that one, 10 I and say, my God, the probability of that happening was one in 11 a million, and yet. it happened, the statistical treatment 12 was wrong, I would be making a terrible mistake.
I would have
)
13 done the denominator wrong.
14 So it is important not to make that nistake.
- And, 15 where it is true that WASH-1400 does not highlight _ fires as 16 a cause of accidents, and WASH-1400 was caused by a fire -
s 17 VOICE:
Brown's Ferry.
18 DR. LEWIS:
Brown's Ferry.
What did I say?
19 VOICE:
WASH-1400.
20 DR. LEWIS:
I hope there are no professional 21 psychiatrists in the audience.
It doesn't necessarily follow 22 that the estimate of the probability of the fire given in WASH-23 is wrong, because one is there selecting something which has 24. happened, and one doesn't want to fall into the trap.
a.p.e,w a.corwes. iac. I 25 Nonetheless, it is an important thing, it is not
24 I
highlighted in WASH-1400, and it happened.
We have a dis-2 cussion of Brown's Ferry.-
3 We have a discussion of common cause failures.
It 4
is a very complext subject.
WASH-1400_has a treatment of 3-. common cause failures which is in our, view, not entirely 6
satisfactory.
It is not clear that one, can make a satisfactory 7
treatment of common cause failures.
8 What is true about common cause failures, and,this 9
not a statement about WASH-1400, but simply about reactor 10 safety, is that th'e causes that go into common cause failures 11 are denumerable.
There are not all that many of them.
When 12 it is talking there about fires, earthquakes, floods, things
~%
i 13 like that, and we have a list of nine of them in our report.
14 It is prob' ably better to assure safety by looking at the cause 15 of common cause failures and making your stand there, it is 16 better to do that than to try to assess safety through trying 17 to quantify common cause failures.
18 That is as far as we have gotten wif.h that.
There 19 is a discussion of>that.
~
20 We have a discussion of htman factors, which are 21 very difficult.
Humans can cut both ways humans do. dumb thin 22 At Brown's Ferry, a himn started the fire by doing a dumb i
23 thing.
On the other hand, also at Brown's Ferry, humans saved 24 the plant by adapting to the existing circumstances in a way w.ome seconm. anc.
25 that could not be accounted for in the WASH-1400 methodology.
25 1
So humans can start things, and mess things up.-
2 Some of my friends use strenger words, but it is equally true 3,
that humans, especially on the tine scale associated with_ a 4
4l reactor accident, have a major possibility of actively coping.
5 As we all know, to our dismay, I fly airplanes,- and I know, 6
that if we did a WASH-1400 analysis of airplanes, and -left 7
out t' t possibility that the pilot looks at his situation,. and 8
does w at is normally right, if we left that out, we'd have 9
a much higher prediction for accidents ; caused by flying than 10 ! is in fa:::t the case..
11 You can' t quantify that, but it's there.
That-12 wasn't wuantified in WASH-1400, but by the same token, the.
13 possibility for people to cause. accidents was well quantified.
14 We find that in the handling of hnman factors, 15 WASH-14 00 did as well as one can do in this _very difficult fiel 16 They interviewed, generally speaking, the right people, took 17 views from the right people; there's a rotten data base; nobody 18 really knows how to factor human into these equations.
They 19 did as well as one can do, but it is a tough subject, and there 20 is a lot of room there for maneuver.
21 There may well be an enormous amount of conservatism 22 there is certainly some nonconservatism in WASH-14 00.
We 23' can't quantify it.
We' have a section on human factors.
24 We have a section called format and scrutability.
co-Federna Rooorters, Ire 25 WASH-14 00 is truly inscrutable.
Anyone who has tried to work
26 i~
with the thing and learned how any given calculation was 2
really~ done, comes away drinking heavily.
l (Laughter. )
4 F
5 6
~
-7 8
9 10 11 12 13 e
14
~
15 e
16 17 18 19 20 21 23 end EJ 24 cm-Federet Recursers, Inc.
end EJ l
i
~
i jeri 1 1-It's just very hard and I don't know whether even er 9244 EJ B 2
having an index would have helped, so you can look thingsfup.
3 But it is a fact that you find yourself, when you seriously 4
try to look at the report, you find -- and try to study how 5
something is done, you find yourself shunted around from 6
appendix to appendix' in a way that does become frustrating..
7 It's just not a very scrutable report.
We are sorry about tha 8
but it is a fact.
And everyone has noticed it.
9 There is a separate issue on formht and sciutability 10 which is the famous executive summary.
The executive summary -
11 I think we have consensus on this -- was, in the first place, 12 not a summary of the report.
That is to say, when one writes
./
13 a report normally -- and I am not sure whether we will do it 14 with our report - but normally one writes' a whole report which 15 is meant for people to read, really read, who need to know wha 16 was done.
17 One writes often.either s'ubsidiary sections or 18 technical appendices. for people who are working so that th'ey 19 can really know and if they want to duplicate something and 20 then ene writes an executive su==ary for people who are busy 21 sitting at hearings like this and don't have time to read all 22 the paper that is churned outside their organization, which is
~
23 a concise st m of what there is at the level that a decisior 24 maker needs it.
?. decisionmaker may not have the background
..r.aerei Reporters. anc.
25 to read the technical part.
So it is a summary.-
I
, 28 jeri~2-1 The executive summary to WASH-1400 was not that.
2 And that has generated a lot of heat.
It.was not a summary of 3l the report.
It, for example, left out the latent fatalities, i
4 which were in the main body of the report, which were one of,
5-the mair. achi'evements of the report.
They were simply left ou 6
It was written as a public mtement that resctors 7
are safe compared to the other natural risks to which we are 8
exposed.
We think it is a pity that it was portrayed as a 9
s"mm' y of WASH-1400; it is. not that.
We find it unfortunate 10 I that it was attached to WASH-1400 because then it mascueraded 11 as a,si m ary of the report.
We think that if AEC or NRC had wished to interpret WASH-1400 for thle public as saying that 12
)
13 reactors are sufficiently safe, it would have been entirely 14 proper to do that in a separate document using the results of 15 WASH-1400.
But it should certainly not have been represented 16 as a su= mary of the report and attached. to the report.
It's 17 our view on that separate item of the executive si-7 18 COEiISSIONER BRADFORD:
What do you mean by entirel 19 proper?
Do you mean that the report in fact would support tha 20 document if it were separate from it, or do you mean that that 21 would be a tactic --
22 DR. I2WIS:
I mean it would have been a tactically 23 proper thing to do.
It would have generated a great deal of 24 heat.
But the issues would have been fought on different ca.Feceral Reportm. Inc.
l 25 grounds. 'They would have been fought on just the grounds you
29
'eri 3' I
mentioned, whether the report supported such a conclusion.
2 They would not have been fought on the question of whether 3
this is a proper representation of the report as part of'the 4
report itself.
5-You.know, of course, the exacutive summary is the 6
most widely. read part of the report.
And we differ among our-7 selves how important it is.
There are some who. feel -- on our 8
committee who feel that we ought to go gungho in kicking. the 9
executive summary. I feel the executive summary was a real 10 mistake, but it is history and it should be regarded. as II history.
We do differ on that point.
When I say " proper.," I 12 mean only that if after the issuance of WASE-1400 properly w
i 13 summwized for the decisionmaker, if after that appearance, 14 NRC had decided to make a public document available saying that
~
15 even in a strident' tone that reactors are safe, because lord.
16 knows there's plenty of stuff in a strident tone that says 17 reactors are unsafe, if it wanted to do that using or even mis-18 using the r.aterial in WASH-1400, that would have been proper.
19 The fight would have occurred anyway but it would have occurred 20 on that document as a docunent and not on WASH-1400 as such.
21 I think the executive sn7mf has thrown a lot of 22 bad stuff at WASH-1400 itself which it could have avoided, 23 comment about executive sur Amt.
24 The peer review process, I have touched on that
=eJaseral Reportm, Inc.
25 earlier.
We have a section on that.
It was too short; it was
30
' eri 4 1
not well organized; it was not selective.
And yet it is j
~
2 terribly important.
3 CHAIRMAN HENDRIE:
This in particular, I take it
~
4 you refer to the publication of the draf t report and the 5-
. period of time.that was allowed for the distribution of the
~6 same.
7 DR. LEWIS:
That's correct.
In between.
There 8
really wa.sn't tine to do an adequate peer review in between.
9 But even subsequent to that, there was a problem in communi-10 1 cation between the peer reviewers and the WASH-1400 staff.
II There really were,' legitimate peer review comments which were 12 not acted upon by the WASH-1400 staff which really, I think by
~
./
13 most reasonable standards, should have been.
The inclusion of 14 latent fatalities in the executive summary, for example, be-15
-cause it was a part of the report.
That was mentioned very 16 early and it never happened.
17 Now mind you, I am not saying and we are not saying 18 that latent fatalities should be added to acute fatalities in 19 talking about reactor _ risk but they also shouldn't be excluded 20 Somewhere in between there is the truth.
There's no question 21 that telling a person that he has a tenth of a percent chance 22 of contracting cancer in a.few decades is different from tell-t 23 ing he is going to be killed tomorrow.
That has a different 24 i= pact and one has to account for that difference.
And it is
. Lee Faceret Reporters, Inc.
25 neither correct to add or omit.
And that was a peer comment 4
i
31 jeri 5 1
that was made that was not acted on.
2,-
There are a fair number.of others.
There are also 3
a lot of nonsensi' cal peer comments that were quite properly s
4 not acted upon.
There just wa:n't time to filter out the.
5 kind of interaction between the report writers and the.techni-6 cal community that it takes to put a thing like this in shape.
7 This is a very complicated report, hard to read,.and it just 8
takes time ~in the technical community, the only.way we know.
9 Once you write something to make sure that it meets t51e 10 !
standards of the cominunity.
You can never make sure it is Il correct; but you can make sure it meets the standards of the
~
12 community by expr> sing it to the community for a given length '
13 of time.
14 We do it in physics by publication.
And we let 15 people write nasty letters to the'same journals saying that 16 Harold Lewis is an idiot, doesn' t he know ~that 2 - and 2 are 5; 17 you know, things like that.
But you have to have a peer revich 18 process.
Otherwise you can't gat a thing like this to be 19 technically credible within the community.
And r.he peer review 20 process was flawed in many ways, which are in our report and I 21 don't particularly want to spell it out.
It 's j ust didn ' t war 22 very well.
23' The section on earthquakes, here there are alternat 24 i ways of calculating the probability of an earthquake and the
- r.Faceest Reporters. Irc. l 25 j effect of an earthquake on a plant.
There have been some repo.
32 jer3. 6 I
hit the light recently which assess the probability of an e[rthquakeaslargerthaniscontainedinWASH-1400.
You will 2
3 recall that it is a function not only of the probability of 4
large ground motion near a quake but also the probability of 5
damage within the plant.
6 WASH-1400, in assessing the probability of damage 7
within the plant assumes that two systems must fail because of 8
the earthquake, takes the probability of one failing, squares 9
it according to the probability that they fail independently, 10 takes the probability of one, saying that they are a common 11 cause, and takes the square root of the product.
We don't 12 like that procedure in > oiler water, control rods; we don-'t lik 13 that procedure in earthquake analysis.
14 We feel that the earthquake subject is very 15 important.
I personally regard it so important that I arrange 16 for an earthquake to occur in Ganta Barbara just last month so 17 I could at first hand assess the effects of a magnitude 5 18 earthquake.
And I made some measurements which 'are contained 19 within our report.
It's really quite important.
20 CHAIRMAN HENDRIE:
I hope you are not going to send 21 us a bill for that earthquake.
22 DR. LEWIS:
The damage in my house was limited but 23 serious and the vorst thing to me was a large gallen jug of 24 Hawaiian punch. concentrate which fell among a31 the broken ca.Federsi Reporwn, Inc.
25 glass.
It made a bit of a mess.
But our university library,
t 33
'eri 7 1-which has a million and a quarter books in it found 400,000 of 2
them on the floor after the earthquake.
And if you can imagine 3l sending students to reshelve 400,000 books in order, you knew l
4 we will be living with that earthquake for a long time.
S In any case, we feel that the analysis necessary 6
to really quantify the earthquake risk has not been well 7
enough done.
It is a high priority and we would like to see
~
8 it done better.
Do more work on it.
9 Risk perception, we have a philosophical section 10 on risk perception.
You struggle with that issue all the time.
11 The perception of. risk depends on many things other than the 12 risk itself.
It depends on the perception of how personal the 13 risk is viewed by the perceiver.
It rests to some extent on 14 the credibility of the Nuclear Regulatory Cnmmission, which 15 tells people what the risk is.
It rests on many things.
They 16 are.all contained in our section.
17 We dealt, with some difficulty, with a specific 18 disjoint item which came up, because it has come up and it cam 19 up in our hearings.
And that is an allegation made by the 20 Union of Ccncerned Scientists that the WASH-1400 staff was 21 dishonest from the start and was geared to produce an accept-22 able estimate of the risk of nuclear reactors.
23 This is formalized in part on the so-called quality 24 assurance / quality control issue associated with a partiet.lar
- r Feders! Reporters, Inc.
25 employee of NRC who wrote a couple of memoranda which have bee.
34 jeti 8' I
released which outlined different ways of approaching the 2
quality assurance question in the report, one of which was 3
dishonest and one of which was honest.
And many of us.r'egard 4
it as remarkable that even a pair of options, one of whi.:h was 5
dishonest, ever hit the table, although the final choice of 3 6
staff was not to select the ons that was dishonest.
7 Nonetheless a representativo 6f the Union of 8
Concerned Scientists appeared at one of our meetings in a 9
public meeting and specifically made the accusation of dis-10 !
honesty explicitl:ir.
- Therefore, we felt we had to deal with it 11 We spoke to the NRC employee specifically, one of us did.
12 We.have received a communication from' him on the subject.-
We 13 havc looked into it and we have asked the UCS accuser.whether.
14 he has further evidence, other than what we had on the record.
15 He said no.
16 We feel that when you accuse people of dishonesty 17 that's a pretty rough accusation.
We often accuse each other 18 of error and that's not bad.
We have all survived lo.ts-of 19 errors.
But we feel that an accusation of dishonesty' 20 deserves a standard of proof matched to the seriousness of the 21 allegation.
And that sta' dard was not only not met but didn't n
22 even come close.
So we have to say that we do not agree with 23 the assertion that the WASH-1400 study was " deliberately 24 dishonest," which were the words used to us.
Am-Feoeraf Renorters, Inc.
25 I will speak for myself now and not for the group
35 jeri 9 '
1 in saying that I regard such an accusation without proof as 2
despicable.
It is not the way we do business.
But for the 3
group, we have a finding in your hands which says that we 4
find WASH-1400 to be a conscientious and honest effort to apply 5-this methodology.
And that's all I want to say about this.
6 Now we have had some disagreement within our:. group 7
about the wording of this finding and I feel I should say that 8
to you.
9 okay.
I am running out of time so let me just 10 say we have a couple of -other disjoint things which are perhap 11 at a different, level of importance.
We have one on the curren 12 role of probabilistic methods in your racket.
We have one on 13 issues of violence, sabotage and wars.
Sabotage, you know, wa 14 left out of WASH-1400.
That fact was openly stated, so it is 15 not an issue in that sense.
It is very hard to quantify 16 sabotage.
It is important to be prepared-against it.
17 There is an issue which is perhaps less important 18 to the United States than ta others, which is that acts of 19 violence do include war.
And if one is dealing with questions 20 of war involving the United States, those are -- the other 21 issues associated with that transcend the safety of-nuclear 22 reac tors.
But there are parts of the world which are beginnin 23 to be filled with nuclear reactors in which conventional wars 24 I are not -- well, are foreseeable.
And in those cases the
. m Federal R eporters, Inc.
25 questions of hardening against accidental or deliberate impact 1
36 jeri LO' 1
of conventional war, we think that is an issue which ought to 2I be considered in the siting and construction of reactors else-3 where.
Not your ' business, but we wanted to mention it.
4 We have a section on ATWS, on transient without 5-scram.
We lay. into the control rod calculation in WASE-1400, 6
again, beca",e it was bad.
The other calculation used, which
'7 is the NUREG-04 60 -- I hope I have the nummar right --
8 calculation which is to observe that in the German reactor, the 9
Kohl reactor, there was a case in which the plastic on a set 10 I of relays was inadecuately cured and that applie d to all the-Il relays, so had the scram system been called upon,.it would 12 have failed, even though it was not called upon.
.13 That number is used as a data base for a single 14 event.
Whether this is a useful thing to do, I just don't kno 15 It is a data base but it.is a unique data base and probably 16 people would be looking at that particular failure mode now.
17 on the. other hand, using the form of statistics that says, 18 well, this is one of a large number of potential design and 19 production defect failures, then it is legitimate to use that 20 as a single case and take the number from it.
It's a very 21 difficult problem.
It's a case in which there is simply no 22 data base and one has to somehow upper bound the probability 23 of scran failure.
And.we discuss this in our section on A'IWS,
24 We speak of design defe. cts indejendently as a sourc w.o re seconm. nac.
25 of failure, unquantifiable but real.
You know, it is possible i
b l-
't 37 eri.11.
I and there is no way of quantifying it; that there is some 2
terrible Achilles heel in the design of reactors.
It cannot 3
be excluded on pure reason.
We don ' t know what it is.
No rea 4
to change our way of running our lives because of it, but it 5
is important to recognize that it is possible.
6 The example I like is the Takoma Narrows Bridge.
7 The Takoma Narrows Bridge fell down; bridges don't fall down, 8
you know, but the Takoma Narrows Bridge fell down in a 9
moderate wind because there wa s an error in design of the 10 bridge which led the bridge to not be invulnerable to uncommon 11 vortices and thc r never been a bridge before that at 12 moderate wind speeds was vulnerable.
There are lots of them, 13 as subsequent studies showed, which were vulnerable at higher
/
14 wind speeds.
It was a case in which there was a design error 15 because one didn't know enough.
And we don't know eve _mfthing, 16 so it is a nagging possibility.
17 In fact, there is an amusing fact about 'takoma 18 Narrows which you may not -- well, there are two amusing facts 19 which you may not know.
One is, that when the state decided 20 to collect insurance on the bridge, it turned out that the 21 state enp.'.oyee whose job it was to deposit the premiums on the 22 insurance had put them in his own pocket because he knew 23' perfectly well that bridges don't fall down and he regarded 24 the insurance as a vaste of money, which he pocketed.
w e.re a.mnm. inc.
25 (Laughter.)
'i 38 jeri 12' 1
He's got to be :the unluckiest investor in creation.
2 But more to the point, af ter the bridge collapsed, 3
the governor announced th'at h_e was going to save money on 4
redesign of the ' bridge and use the same blueprints because the 5
were attested by the bridge building community as a sound 6
design.
'7 (Laughter.)
8 And that's perhaps a more poignant reaction.
9 So its only that, you know, there's always a 10 possibility that there is something that you don't know.
Ther 11 is. <
s 12 COMMISSIONER AHEARNE:
At Takoma didn't they also 13 use a new alloy steel in some of the supports?
14 DR. LEWIS:
I don't think that was the issue.
The 15 issue was that a tortional oscillation was set up which was -
16 COMMISSIONER AHEARNE:
Yeah, I remember that.
I 17 thought also there was --
18
'DR. T:2WIS:
I don't know that as a fact.
It may.
19 be, but I don't know it.
Certainly -- I don't know if you 20 have seen the films of the bridge collapse.
The tortional 21 oscillations were really pretty impressive.
They were the 22 new thing.
And in fact, if you look at the films, you are 23 impressed at how long the bridge stayed there.
So I-think the 24 steel was pretty good.
2 5,e,r.: nemorms. Inc.
25 That's pretty much the summary of our report..
1 I forgot to say an important thing, my friend tells eri 13 2
me.
Use in the regulatory process.
I mentioned that, of' 3
course, we have findings there.
I did mention very briefly 4
at the beginning, there is quite naturally conservatism in 5
NRC and reluctance to do new things.
We feel,'as I said, that 6
although steps have 'been taken to use the metho'dology in th_e 7
regulatory process, we feel more can be done in helping you 8
to allocate your resources.
We specifically do not mean that' 9
you should use the results of WASH-1400 uncritically beca' se u
10 I hope we are leaving you with an impression that we feel that iI anything in it that is used, one should review and reassess 12 and make sure it's righ before before using it.
There. are 13 uncertainties in it.
14 But..L.
da process itself, one'should be more The whole point of risk assessment' m'ethodology.k.s 15 rational.
16 to be rational and not to operate out of the' hip pocket in 17 allocating resources and research to the alleviation of risk ir 18 nuclear reactors.
We think you can do more and we urge yciu to 19 do more.
20 I am open to questions; my friends are open to 21 questions, and my friends may dissent from everything that I 22 have said, in which case they should speak up.
23 CHAIRMAN HENDRIE:
I thank you ve.nr much for an 24 excellent senny and a very interesting one, and why don't I a.Feeni Raponm, Inc.
25 ask of all if my colleagues have questions they might direct I
40 eri'14 1-to you which presumably deal with collegial, group and then let 2
us see if members of your group would like to make some 3
individual comments and the Commissioners, in turn, can have-4 at.them on an individual basis.
5 I would note that, sort of for his remark on the 6
general configuration here, that questions that the Commission 7
direct their comments they may make are by way of better 8
understanding of the things that you are reporting to us 9
this morning.
Your report is well along.
I want there to be 10 absolutely no doubt that the Commission is not in any way 11 suggesting any sort of changes by anything that it may say 12 this morning.
The group's report and its integrity is very 13 important to.us.
You understand that and we understand that
'l4 and I would like to have the room as well understand that 15 very clearly and explicitly.
And the transcript here shows 16 that explicitly here.
And that is the reason I am making the-17 comment.
18 DR. LEWIS:
Well, may I just add two things to l9 that comment.
As you know we have been at this for.a,.i.-
f 20 year and during that period I have in fact spoken to a number 21 of the Commissioners and I have say, again for the record, that 22 in every case the co= ment you made has been made, call it as 23 you see it regardless, and we have tried to do so.
24 I will add one amplification to what you say and
...s.o
.i senorren, inc.
25 that is that having experienced the temperament of this group
jeri 16-1 methodology -- the principal problem. Found in the PWR, which 2
was a check valve problem -- was new to people and they went 3
and addrested it in every reactor to which it was applicable. -
4 So, in fact when the report appeared, that problem wasn' t 5
there, even though it was the principal one discovered.
6 That's a good use of the methodology.
But I think
-7 that any multiplication or extension to other reactors is 8
inapplicable.
You have to study them in detail.
And that, of 9
course, leads to all other sorts of questions; about standard-10 ization which are not our business here today.
11 COMMISSIONER GILINKSY:
What can you say then about d
12 overall risk as a result of the studies and your review of 13 the report.
I mean, you say you are not sure whether the numb 14 are high or low and you thinP the error bounds are probably 15 wrong, but you don't know by how much, and in fact the results 16 may only apply to these two reactors.
Where does that leave 17 us?
18 DR. LEWIS:
Where it leaves.you is, of courese, 19 your problem.
20 (Laughter. )
21 DR. LEWIS:
But I understand what you are asking 22 and I can't answer it,
- really can't.
That is to say, I cann 23 say anything that would -- speaking personally for myself now 24 that would be in the nature that based on this report, this a.r.o.r.: neporters, are.
25 implementation of risk assessment methodology -- which we real
41 jeri 15 1
over the last year, if you were to make a suggestion for 2
change you might achieve the opposite of what you intend.
3 CHAIRMAN HENDRIE:
Sounds like a good group. -
4 COMMISSIONER AHEARNE:
At some stage before this 5
morning is over what I had wanted to do is to ask Hal and the 6
other members for any individual comments they had.
Hal, you 7
mentioned at several points that these were your own personal 8
views on it and I am interested in exploring any other 9
personal views you have and any that the individual members 10 have; whether that's appropriate now or Inter, it doesn't make 11 any difference to.me.
12 CHAIRMAN HENDRIE:
Well, let's see if there are 13 other comments or questions that are sort of directed at the 14 collegial group, and then let's turn to that.
15 COMMISSIONER GILINSKY:
I want to ask -- the report 16 of course, dealt with two reactors, the 3WR and PWR.
To 17 what exten.t do you feel the results are applicable to other 18 reactors?
19 DR. LEWIS:
.We feel that's a problem that they. are 20 not -- it is clearly not.
In a number of cases there are 21 things which were good about those reactors which are not 22 reproduced elsewhere.
In other cases there were bad ones.
23 In fact the report itself, when it appeared, didn't even 24 refer to the deal with the two reactors it studied because -
2.s.o.i nenorrers, inc.
25 and this is a fine example of the use of risk assessment
43 jeri 17 1
to want to distinguish from the methodologies -- based on this 2
implementation that I can say anything about the population of 3
reactors that we have now.
That would be useful to you'.
I
.s 4
can't say anything.
5 We have an actuarial base of course which is not 6
all that far from the formal probability predicted in WASH-140 7
but that is not what we are discussing here.
8 COMMISSIONER GILINSKY:
Are you saying, in effect, 9
that one can' t really make use of the integrated results, 10 that the only thing you can really with are the relative 11 probabilities?
d 12 DR. LEWIS:
I have to be careful about this because 13 I have lost the view which I held when we did the APS study an 14 which appears in the APS study -- you may remember that there 15 is a sentence in the APS studies, to whic5L I subscribed at the 16 time, which said we didn't have confidence that the absolute 17 probabilities predicted in WASH-1400 but we believed that it 18 had nerit in assessing the relative probabilities of different 19 accident chains.
I am personally moving away fro:n that posi-20 tion because if one finds different accident chains and one t
21 feels that -- let me just invent a ter2n -- that the error band 22 is a factor of 100 on one of them, I don't see how one can 23 assess its relative importance compared to the other.
24 Now clearly there are extreme cases.
There are ce Feders! Reporters, Irt.
25 things which one can look at and say are'by many orders of
44 jeri 18.
1 of magnitude not very important.
And that is clearly a useful 2
application of the methodology.
Or you may discover something 3
you hadn't thought of before.
It's a little less likely.
4 But on the specific issue of whether I can learn 5
anything from WASH-1400 that will enable me t'o say whether 6
reactors are acceptably safe or not, I can't.
I don't learn 7
it from-WASH-1400.
Speaking for myself.
8 COMMISSIONER GILINSKY:
Well, the " acceptably" 9
comes by comparison.
But I think one was looking for somo 10 judgment about what the risks were.
Are you saying you really 11 can't say anything about that?
12 DR. LEWIS:
We have said collegially that we can't 13 say whether the probability of core melt in WASH-1400 is high 14 or low, but we know that the error band is bigger.
Now the 15 error band doesn't have to be very much bigger to begin to but 16 into reactor experience.
And reactor experience is there at
~
17 a certain level.
18
'So I guess -- I am speaking for myself here - I 19 personally have learned a great deal about reactor safety, 20 from studying WASH-1400.
I. have become much smarter.
But the 21 way in which I have become much smarter does not include 22 substituting a number out of 1400 for the reactor experience 23 number in my head as my estimate of the probability of core 24 melt.
- ce.Feoerst Reportees, Inc.
25 For example, we tend when we say there are wide
45 jeri 19 1
error bounds, people tend to concentrate on the upper bound, 2
as I just did, by saying that it abuts the accident experienc 3
I would personally now, speaking for myself,;regar, 4
it as not at all unthinkable that the results given in 5
WASH-1400 is three orders of magnitude too high.
And if that 6
were the case, one would learn very little relevar....
So I
'7 can't answer your questions.
8 COMMISSIONER GILINSKY:
Well, let me ask you a 9
little more about the relative risks of an accident.
You wer 10 urging us to make greater use of this kind of understanding.
11 DR. LEWIS :
Yes.
12 COMMISSIONER GILINSKY:
At the same time you are 13 saying that the results apply really to these reactors and 14 I've got to be very careful about --
15 DR. LEWIS:
Okay.
Let me smell out then -- I'. : <.
16 understand the confusion here.
There are things in WASH-1400 17 there are many things.
One is -- and that is what you were 18 alluding to when you spoke of relative risks -- there is a 19 bottom line number, which ir 5 X 10-5, for the probability of 20 a core melt.
That is kind of bottom line number which comes 21 adding probabilities of a very, very large number of accident 22 sequences.
23 I am saying that I don't learn much from that.
Wh 24 I urge you to use the methodology, it is in the subsystems fo ace.Faceret Reporters, Inc.
25 which there is a data base and where the universe of discours
46
'eri 20' 1
can be sensibly and logically bounded to be within the 2
capabilities of people and things to compute.
3 Just as a minor point, you know, in assessing how 4
many diesels one uses, or the differen.t kinds of pumps, steam-S'. driven, electrical-driven pumps, things like that, subsyst' ems 6
which can be carved out which are contributors to the safety 7
, of a reactor, you need to use this kind of methodology.
Now 8
there is more than that.
There are some things which re. ally 9
one can presumably say are by eight orders of magnitude too 10 small.
Clearly you don'-t waste time on them.
11 CCMMISSIONER GILINSKY:
Well, but in some common 12 sense way, the staff applies this sort of understanding.
i 13 Perhaps not in a systematic way.
But people obviously devote
~
1 their effort to things they think are important and less to 15 things they think~are unimportant.
I guess w1iat would be
~
16 interesting is from your experience do you see major 17 disparities.
18 DR. LEWIS:
Yes.
Well, for example, suppose.for 19 the noment we adopt the position that believe everything that 20 is in WASH,-1400.
Working hypothesis.
Fair enough.
21 It says for example that large LOCA is not a major 22 ! contributor to risk.
That is not reflected in the regulatory 23 and research programs.
Transients are a major contributor to l
24 l risk not reflected.
There are really realig;.ments to the a.Faderal Reportus, Inc.
mee fois 25 '
extenu that one believes the results.
I ti
=te 1l -
47 fis jeri 20
^
Now, by the same token, looking into the future, 'as j.
2 one carves out specific issues and does them better than was 3
done in WASH-1400 or validates the Wash-1400 results where we 1
4 have concerns about them, one could use this information to 5
assign your priorities, really, just for that purpose.
But not 6
the bottom line.
7 DR..KOUTS:
Can I do some modifying there?
I think -
8 CHAIRMAN HENDRIE:
If we've run out of collegial 9
comments, why don't we start down members of the group?
10 DR. KOUTS:
Well, I didn 't want to make this an 11 overall cor=nent.
I just wanted to comment on that particular 12 point, where I would go somewhat farther than Hal and say that-13 the particular estimates of overall rir.,k associat id with nuclea 14 power as they were given in WASH-1400 may have son.e doubt 15 atthched to them, and do have some doubt attached to them, '
16 because of reason.s that we have uncovered uid perhaps other 17 reasons that other people know about.
And so those particular 18 numbers may not be the best guides to the question of risk-19 associated with nuclear power.
20 But I think if we could fix those things up, if we 21 could repair the damage, the defects, that we've seen in the 22 report, and in.u oduce better ways of -- introduce the better :
23 ways of calculating that we think we know about, we would arri 24 at central estimates of risk, which we still might find to be necorwri,ine.l the most dependable values we can generate in any way at this
..F.o.r.:
25 I
=te2*
48 time.
And for lack of anything better, we would use those, I w uld recoI= lend we use those in policy determinations.
2 DR. LEWIS:
I don't disagree with that.
3 DR. KOUTS:
But this goes beyond simple views on the methodology -
5 DR. LEWIS:
It goes beyond the methodology.
COMMISSIONER GILINSKY:
But you_ are saying that the.
ones that are available now are not --
8 DR. KOUTS :
We 've learned a great deal since issuance of that report, some things just becacse science has pro-g gressed and some things because of the detailed introspection g
whicli this has generated, this process we're ending, I hope, g
And we found a number of things th.at really ought to be now.
g fixed up, yes.
~
g COMMISSIO'NER GILINKSY:
But you're saying, in effect, 15 that it is possible to arrive at these I:rumbers?
16 DR. KOUTS:
Yes, I think so.
7 DR. ROWE:
There is a qualific:ation, and we make it.
/
We say that in the case where there is an adequate data base 39 available, you can arrive at probably reasonable numbers.
In g
the cases where they're not and you have to use some subjective value judgments, at least we can do a bounding analysis, which can be useful.
And th'ere are some cases when you don't even have that; you may use it because there "s nothing else you have g
2.F.e, car nenorms. sac.
available,.and as long as you properly qualify it, it can be 3
"mte-3
, 49 us eful.
So this is -- the point is, you've got 'to put the variotis parts of it together and use what is 'useful at the z
highest level of confidence and use what is useful at the least 3,
l level of confidence, as long as you properly put it in perspec-5 DR. VON HIPPEL:
I think Hal likes the orig'inal way of presentation.
That's fairly representative of the lowest 7
common denominator of the group, that I think the emphasis has been misplaced on the bottom line number in WASH-1400.
I mean, 9
I Personally, I represent an end of the spectrum in the opinion 10 of this group, am doubtful that the methodology for this botton 33 line' number sums over the sequences that are identified, 12 in luding their very large uncertainties, and then estimates 13 of what is left out, that that bottom line number really can y
15 give us, with high confidence, a number for the total probabi-lity, for example, of a particular end point to which all these 16 things contribute -- core melt; 'that one can get a confident 37 estimate that that probability is well below the actuarial limi 18 that we have today, the experience limit.
19 ONSSIOER GILINSKY:
Well, why is that, because of 20 the sheer complexity of. the problem?
g DR. VON HIPPEL:
Yeah, just having -- this is, again, g
my personal opinion - having looked in all the ranges, compar-23 ing the WASH-1400 calculations with alternative calculations 24 DFoccW Reponm, Inc.
3 on major contributors, well, contributors that may be considere I
mte 4 -
j minor in WASH-1400 that other people 's calculations would 2
indicate are major; and seeing the complexity of real things 3,
that happen in real reactors, such as loss of DC power supplies 4
and the ways in which those things happen,' has given me a 5
subjective error bar on that number, which goes all the way up 6
to the actual experience limit.
7 And. in fact, I think if we didn.'t know that there has 8
been no meltdown to date,. I don ' t think we - you know, if one.
9 did WASH-1400 in a box without that knowledge, I think that a 10 realistic error ba_r would go above the experience to date.
11 DR..ROWE:.
I think -- I was going to say,
I. think in 12 separating the framework and the methodology from the numbers -
13 and Hal has pointed this. out a number of times - we do think 14 the framework provides a way to go.
There are a lot of holes 15 in the framework tliat have to be filled in' before one believes -
16 either absolute or relative risk numbers.
And we ' recommend 17 that these holes continue to be looked at and filled, that this 18 be a way of identifying where the holes are, and that you go 19 about a procedure to fill these.
20 DR. LEWIS:
I guess there is really a general consen-21 sus that we want people to be smarter.
22 (Laughter.)
23 DR. LEWIS:
'It takes time.
But we want to encourage 24 people to be analytical in the approach to reactor safety.
And
,r.5ec,ral Reporters, Inc.
25 as Frank has emphasized, there is a difference, in which most
mts' 5 51 1
of us, I believe, feel that there is light at the end of th'e 2
tunnel, as Herb and Bill has said, and Frank does ' not.
That's 3
a subject of judgment.
But we certainly don't want to di,scoura 4
people from using a rational approach tot tlie assessment of 5
these risks.
6 COMMISSIONER GILINSKY:
Do you have anything to say-7 about a further study or another comprehensive study?
8 DR. LEWIS:
Well, I suppose our study has to be 9
subjected to peer review.
10 COMMISSIONER GILINSKY:
Well, I don't mean an audit 11 of your activities.
But I mean like a review, in effect, of -
12 DR. LEWIS:
I have no comment on that.
I know it's -
13 been batted about.
I ha.ve no comment on that.
14 DR. ROWE:
I think we talked about it, to the extent 15 that we're not so sure that that by itself %;ould be useful,'
16 moreso than would be the continued staff r ttentio$1 to the 17 various soft places anc'. problem areas, to put it together, and 18 have the staff in the normal way bring these together, than to 19 have continued studies of this type.
20 COMMISSIONER GILINSKY:
I don't mean necessarily 21 going over the same ground to redo the calculation, but I'mean 22 a new comprehensive study of reactor safety.
23 DR. ROWE:
I suspect it might be more important, as 24 Hal has pointed out, to look at those soft spots and concentrat a.pasmi neoorms. anc.
25 on the soft spots, which are maybe the high levels of uncertain
mte ;6-52 1
that really haven't been quantified.
2 COMMISSIONER GILINSKY:
Are you saying in effect that 3
the existing study provides a framework that one can build on?
4 CHAIRMAN HrNDRIE:
I think the methodology and most,o 5
the analysis does that.
6 DR. ROWE:
By methodology, you exclude statist'ical 7
methodology?
8 CHAIRMAN HENDRIE:
Yes.
I'm talking about the fault 9
tree, event tree, framework.
10 You tve been trying to get a' comment on?
11 COMMISSIONER BRADFORD:
I'm not wise to insist on it, 12 but along the lines of what you were just talking about with
,13 regard to the further study effort, you referred, Hal, several 14 times to an inadequate data base, and I wasn't sure it was 15 always the same inadequacy.
What's the difference between what 16 we have now and an adequate data base?
Is. it in terms of 17 additional years of operation, additional reactors to study?
18 DR. LEWIS:
Okay.
There are several levels to the 19 question.
Certainly, additional years of operation will provid 20 a better data base, both by lowering the actuarial limit, which 21 I regard as not so important as some of. rty friends do, but also 22 in allowing time for more of the precursors of accidents to 23 appear.
And the more abnormal things that happen in the reacto 24 population and the more you take advantage of them to get weeeme nmmn, sne.
25 smarter, the safer the reactors w'ill'be.
So there's certainly
m,te 7 53 that.
I g
But additionally, there are places in the data base, 3
in the component data base that goes into the estimates in WASH-1400 which are weak.
For example, if one reads the sectio 4
n pipe failures, the data base that goes into pipe failures, 5-which are sort of the beginning of many of the accident 6
sequences, these data are weak.
Now,. we have had more experi-7 en e.
We can perhaps organize them in a better way.
We can 8
fill them up to some extent.
9 10 There are places like that where component failure 11 probabilities can be filled up more effectively.
j2 COMiISSIONER BRADFORD:
Even with the existing material to work with?
13 ja DR. LEWIS:
Well, we have more than we had at the tim of WASH-14 00.
15 16 DR. ROWE:
There are a number of things going on now 17 which are contributing to it, human factors as an example.
18 There are some studies on human factors going on, which may contribute to the data base.
So it's not just the component 39 e-C failures, but there are specific areas where we just have an 20 21 absence of knowledge that you can focus upon by looking at 22 similar kinds of systems.
COMMISSIONER'BRADFORD:
If I understood you, it's a 23 24 mixture.
There are some areas you can cure by doing a better m-Fet!eral Reporters, Inc.
25 job with the data that's really there, and other areas that
. m te
,8-54 1
time itself will provide more.
I 2
DR. LEWIS:
And.there's a third area,.where irou-can 3
go out and seek better data.
4 CO!O!ISSIONER BRADFORD:
But in terms of the areas 5
that you want to wait for perhaps more time and perhaps more 6
abnormal occurrences, is that a matter of doubling or. tripling, 7
or 20 or 30 times away from what you feel. is adequate?
8 DRO. LEWIS:
Adequate, it will probably never be 9
adequate.
You can get smarter, but you can't get smart.
.I 10 don ' t think I' d.want.to set a goal.
11 DR. VON HIPPEL:
We really haven't developed this 12 area'very much.
13 DR. LOWENSTEIN:
You know, there is always this 14 problem.
You're generating data, you get smarter, and you try 15 to fit into the sys'tems the benefit of that smartness.
So you'
~
16
.ilways going to have inadequate data.
You allow for some 17 education along the way.
18 COMMISSIONER BRADFORD:
I guess that depends a' lot 19 on your purpose.
If we were talking about automobiles, would 20 you say that we had an inadequate data base for assessing the 21 probabilities?
Or airplanes, for that matter?
22 DR. LOWENSTEIN:
It depends how precise you need to 23 know.
In a sense, we have a very good data base.
But judging 24 from the nu=ber of recalls on detail, it may be inadequate for en Federna Re.mrters. Irc.
25 this purpose.
And perhaps it should be --
- mte,* 9 55 1.
DR. ROKE:
Or it may be a changing perception of' the 2
use you want to put the data base for.
3 DR. BUDNITZ:
I have two comments, and one of them 4
is along this line.
To me, a great finding of WASH-1400, which 5
surprises nobody when you -- it was like Miss America.
After 6
you've discovered her, everybody said, oh, gee -- was the rqle 7
of human failures in risk.
Whatever risk there is and however 8
the future quantification comes out, we have learned that much, 9
if not most, of the risk is due to human failures of several 10 kinds.
11 There's,the human who begins an accident. ' You know, 12 the candle at Brown's Ferry.
There 's the human who doesn.'t return valve setting to its proper position after prepping or 13 14 maintenance or callibration, and then.it's 'not available when 15 called upon.
And if you ask how much t aat contributes, thehr 16 don't really know, but they'd say half, pernaps 75 percent, of 17 the event trees have important places where it's the human tha 18 cade that mistake, rather than, to: cxample, something that 19 just, in the normal course of events, fails, you know, without 20 a specific human thing that happened, and so it's not availabl 21 Now, we don'r know enough about those himn failures, 22 about ha to correct them or about how to go after remedying 23 that situation, and we won't know enough until we gather the 24 data and then do the, risk assessment over.
It's a very import a:r-Federal Recotwrs, Inc.
25, area of application which this study has guided us to, but whe.
I
- mte d.0
. 56 much more work is required.
3 The other side of it is -- and' Hal "said this very 2
well -- that there is little or no way for the methodology to 3,
I incorporate the ingenuity of a human during an accident, runnin 4
soMedng g some ju dgged de in oder to 7
a m
a an 5-make a contro,1 mechanism available that wasn't there.
And that 6
is an intrinsic conservatism in the calculations.
And there's 7
eeP inskde ourselves n
waY p
a n.
u we a nN 8
that, as Hal says, to him it's the most comforting thing of 9
all in terms of how the overall system will'be safer than we 10 think.
I think I shouldn't quote from him, but he has said jj that.'
~
12 Now, that's --
13 DR. LEWIS:
If I've said it, I have to own up to it.
jg DR. BUDNITZ:
Well, you can say sometliing different.
15 Now, that's an important area whe e we've learned; 16 and where the whole regulatory process must, in my view -- in j7 fact, we say it here in the recommendations - concentrate more 18 attention than has been concentrated to date.
39 Now,.there's another point -- so that's point number 20 one.
I have a second comment.
21 CHAIRMAN HENDRIE:
Before yor get to the second 3
comment --
3 s
DR. BUDNITZ:
Su:ge.
24
- r.Feoeral Reporters, Inc.,
CHAIRMAN HENDRIE:
Isn't it then consistent with 3
11 I
i
mte'll
~
57 that to put greater emphasis upon the quality of the personnel j
involved in the process?
2 DR. BUDNITZ:
That could be, or it could be the 3
procedures that they follow aren't good.
I heard an example 4
just recently of a 12-step thing, when you take something apart 5
to fix it.
And it said, do this, do this, do this, do this.
6 And then when they got to the end, it said, undo it all.
You 7
know, turn it back.
Instead of having a check reference to go 8
back.
And of course, sometimes those things are left not right 9
10 That's an example of where a procedural, administrative 11 approach can remedy a failing which enters into the risk.
d 12 Now, you know, it's just something that has to ha done in detail.
And the methodology and its lessons can make' 13 34 that approach much more rational than had been.
15 I want to make a completely 'different point.
And 16 here I think it's most useful just to read a couple of the reemendations and-findings.
It's on the very last page.
It 17 18 says:
" Fault tree-event tree analyses should be among 'the 19 principal means used to deal with generic safety issues, formu-20 late new regulatory requirements, assess and revalidate exist-21 ing regulatory requirements, and evaluate new designs."
- Now, 22 that's a very strong recommendation, which I think we not only 23 all concur in, but we mean it to be very strong.
24 And then the next point is important, too.
It says:
ce Federsi Reportres, Inc.
25 "NRC should encourage closer coordination among the research
mte?I2 58 and analysis staff and the licensing and regulatory staff,- in rder to promote the effective use of these techniques."
2 Now, let me turn back a page, in the second one from the top, recommendations, page 1:
"Use RSS. probabilistic methodology" -- this is a recommendation -
"more effective,1y to guide the reactor safety research program."
.And then some other stuff.. There's another e.cample.
7 E * * '" E " 9"* '
8 read you -- I guess it 's the one on the top of page 5 -of the 9
10 !
findings:
" WASH-1.400 was directed to make a realistic estimate f risk.
When using this information in the regulatory 11 Proedss, the usual conservatisms must be incorporated."
And 12 we all concurred in that..
g g
These together can synthesize, I.think, our collegial ew n
e a
a mm e med, u
ga e usual 15 nservatisms must also be used, nat only to mask'our uncer-16 tainty, but because even 3.n the face. of what we think is g
certainty, conservatisms are an appropriate regulatory approach 18 The thing that we had much discussion about was the 39 fact that in the implementation of WASE-1400, in the study 20 itself, regulatory conservatism entered into the realistic 21 calculations in a way that we cannot quantify, but we know is g
' there.
For example, in the calculation itself, the BWR scram 3
. f ailures is an example.
Three-rod failure was said to be a 24 I
- .. Federal Reportm, Inc.
scram failure, and in fact, the data base for that was tnat it
,3, d
rte 113-59 j,
went all tha way down and 90 percent of the way in, and didn 't 2
get all the way in, and that was a rod failure which entered 3
into the scram failure.
But we know that isn 't so.
But who 4
knows?
5 This conservatism is not only in the calculation, and 6
inevitably so.
But then it has to be aoplied on top of that 7
as it's used for regulatory purposes.
It's a very difficult 8
thing to try to put in.
Still, I think the words are as strong 9
as they can be:
Use the methodology to guide this, this, this 10 and this.
And I think that we mean it.
Is that a fair consen-11 sus?
12 DR. VON HIPPEL:
The methodology?
~
13 DR. BUDNITZ:
Right, the methodology.
14 CHAIRMAN HENDRIE:
I'm still interested in whether an 15 of the others have any other comments they ~ haven't made.
16 Walter?
17 DR. LOWENSTEIN:
No, I really think that Hal simmn-18 rized the collegial view as well as, I guess, my own personal 19 view on most instances.
If I differed with some of expositions 20 it's on detail.
If-I differ with Hal's view on these things, 21, I think they're on matters of detail rather than on what you 22 might call the larger issues that he's focusing on.
23 DR. ROWE:
I' was going to say that the general thing 24 that happened to us -- we came from diverse backgrounds and m.Fecefal Racerters, Inc.
25 some diverse upinions on this when we started.
And as we went i
ate '14 60 threrugh this and talked to an awful lot of people, many of us ame away with much different perspectives than we had when we 2
came into it.
And while there still are some differences. among 3
us, it is a surprising consensus that we've.all come to, and I 4
think that's reflected on the report.
There are one or two 5
places where we have surprising differences.
6 One thing that I have emphasis on -- emphasis, not a 7
ma r
e en e --
s dat we have sepeated de proWhes 8
and the consequences and where they're looked at.
And we 've 9
talked an awful lot about the recommendations and the probabi-10 lities.
We also said in our recor::=endations and the consequen-
);
es, that the way they were formed, for a number of reasons, 12
~
not just the health effects, but also the meteorology and a 13 number of other atmospheric dispersion models, that we shouldn' jg 15 rely up n those, that those ought to be redone, especially due 16 to the fact that there 's new information now becorning available And I just wanted to emphasize that we have to talk g
about that aspect as well as the probability aspect.
Further-18 39 more, that you can't use the average values that we use when y n're talking about application to a specific plant.
20 COMMISSIONER KENNEDY:
This is a point you're making g
on the second page.
22 I
9 23 to make sure that that wasn't overshadowed.
It's not a point 24 3 5 oere neconers. Inc.
of disagreement; it's just a point of emphasis.
3 I
mtel15
, 61 DR. VON HIPPEL:
I agree with most, although I 2
disagrea with some details, of what Hal has said.
I'd like to emphasize that this is a highly negotiated document.
We.really 3
4 did come from very, very different positions.
We include stro 5
people who are on record as strongly defending and strongly 6
criticizing WASH-1400.
And it is remarkable, the level of,---
7 I think the nonscientific community will be surprised by the ~
8 amount of agreement we've been able to achieve.
9 I'd like -
10 COMMISSIONER KENNEDY:
The nonscientific community 11 or.the scientific community?
12 DR. VON HIPPEL:
I mean, the fact that -- no, the 13 nonscientific community will.be-surprised,. you know, that we 14 basically all have a respect for truth.
~
15 (Laughter.)
16 DR.' VON HIPPEL:
And that's not universal.
17 I'd like to stress the document itself, though, 18 because it is a highly negotiated document, and I don't th' ink 19 any verbal summary, even one as skillful as Hal's was in most 20 respects, can do it justice.
We do each put different weights 21 on different findings, and if we started to explain in detail 22 the different weights that we put on different findings, I thin]
23 then the consensus, to a degree, the consensus that we have bee 24 able to achieve would start coming apart.
And so I'm being cm.Focers! Recorners. Inc.
25 very restrained.
mee.10
- 62 j.
But I would -- you know, I would stress myself the 2
importance of the findings. and the recommendations section, 3
which we worked over for two multi-day sessions.
We formally-1 4
only had about an hour to work over the summary statement this 5
morning, and so I myself put greater weight on the findings and 6
recommendations.
7 CHAIRMAH HENDRIE:
Any comments,?
8 (No response.)
9 I would..like to ask Hal, at the risk of having some 10 of this magnificent consensus come apart -- and I recognize the 11 difficulty a chairman always has --
12 DR. LEWIS:
You can ' t begin to.,
}
13 (Laughter.)
14 CEAIRMAN HENDRIE:
Are there any.other points that '
15 you would like to m'ake as an individual member, as opposed 'to 16 speaking in the collegial sense?
17 DR. LEWIS:
No, no, I don't think so.
Thank you for 18 asking.
I don 't think so.
I took the liberty during my.
19 presentation to make the ones that I felt were important, and 20 I'll let it. go at that.
You'll all hear from all of us anyway.
21 You know that.
22 CEAIRMAN HENDRIE:
Gentlemen,' you 've come pretty much 23 to the end of a very l'ong and very difficult job, for which the 24 Commission owes you an immense debt.
I don't want to praise a Feceret Reporters, Inc.
25 you too fulsomely until the report is actually printed, lest
\\\\
.mte 17 63 you go ir:nediately into collapse at the very last minute.
But 2
I trust that it will promptly issue, and it's clear that the 3
Commission owes you a great debt of gratitude on this work, and 4
we're very appreciative that you've all been able and willing 5
to stick to it.
6 I've observed myself, on a body noted.for its sometim 7
abrasive internal deliberations, and I'm.very glad that you've, 8
in effect, ground on each other for a whole year.
If there Is.
9 a particular value in which the whole of your effort is in some 10 sense greater than the individual contributions, it i~s precisel
~
11 in that interaction, the testing and retesting of.the individua 12 views and analyses that you had held and made when you came in,
[
13 and the subsequent adjustment of those as you've interacted 14 with each other.
And I think that sometimes we create commit-15 tees to do jobs in ' order to sort of spread' the load. out and' 16 diffuse the responsibility.
I think in this case' the value of 17 the group and the reason of the group was precisely to produce 18 this interaction between different points of view, and to let 19 the final product reflect that.
And without that long inter-20 action between you, I don't think the report would have nearly 21 the value that I believe the repor:. you are producing will in 22 fact have.
23 I look upon 'it as an extremely important document fo.
24 the Cor:=ission, the Commission staff, and I assure you we will
- Feceral Reporters, tric. g 23 pay very, ve.m.f close attention to the recommendations.
l
. :: frei 18 64 "g'"
s g
May I simply close finally --
2 DR. ROWE:
I. think we'd be remis.s as a group for not wanting to point out that we've had associated with us an:
3 official member of the Commission acting as. secretary.
1 4
I think j
I speak for all of us in saying that his action and the work 5-that he's put onto this thing has made this thing quite a f
7 success.
And I think' it's important to recognize his contribu-3 tion as well here.
9 DR. LEWIS:
I should have said that.
John Austin has been a pillar of strength through this whole thing.
10 He's been.
11 superb, and you ought to know that.
12 CHAIRFAN HENDRIE:
We're glad that he was able to 5
serve well and effectively, anti ve'll take due note of your 13 14 accolades.
Let me just close, then, by once more giving to each 15 for the considerable personal sacrifices which you 've made the 16 Commission's very profound thanks, and our assurance that we 17 will indeed.take to heart the product of your labors, and you 18 will in due time see them in an assortment of ways.
19 Thank you 20 very much.
(Whereupon, at 12:00 o' clock noon, the meeting was 21 C
22 adj ourned.)
I 23 24 ocetat Reportees. 88%
$5 r