ML20010C435
| ML20010C435 | |
| Person / Time | |
|---|---|
| Issue date: | 07/24/1981 |
| From: | NRC |
| To: | |
| Shared Package | |
| ML20010C401 | List: |
| References | |
| NUDOCS 8108200076 | |
| Download: ML20010C435 (50) | |
Text
--
i s
n 1
j UNITED STATES OF AMERICA 2
NUCLEAR REGULATORY COMMISSION 3
4 PUBLIC MEETING 5
SAFETY GOAL WORKSHOP" h
8 6
SESSION 4 e
7 SOCIAL EVALUATION
[
8 J
d 9
Cliffside Inn 10 Harper's F,'rry, West Virginia i5j 11 Friday, July 24, 1981 3
g 12 The meeting was convened, pursuant to notice, at E
y 13 8:35 a.m.,
with P.
Slovic, Decision Research, presiding.
=
l 14 PRESENT:
N E
2 15 P. SLOVIC N.
BRADBURN s
j 16 D. RATHBUN R. T.
PAGE A
p 17 D. OKRENT E.
ZEBROSKI 18 C. WHIPPLE L.
LAVE E
{
19 T. COCHRAN A. MAZUR 5
20 D. MAC LEAN C.
PERROW 21 E. O'DONNELL D.
BRIDENBAUGH 22 J. SILBERG G. COREY i
23 E. WEISS 24 H.
INGRAM 25 S. DERBY l
8108200076 810724 ALDERSON REPORTING COMPANY. INC.
2 i
l PgQQQEQ[QQQ 2
MR. RATHBUN:
Good morning again.
This morning we are 3
going to hear from Paul Slovic, who was chairman of the Social 4
Evaluation Panel, and Letween now and 10:00 o' clock he will lead 5
our plenary session on this subject.
Paul?
e A9 3
6 MR. SLOVIC:
I don't have any special remarks to 2;?rt e
R R
7 this session other than to provide an orientation to what I think N
3 is the topic for this morning, social evaluation.
O m
9 Perhaps one way to think about the discussion is to i
O g
10 recall the comments that our panel had yesterday about the fact El 11 that there was a very rich and complex discussion of social, a
d 12 political, and ethical issues that somehow did not seem adequately E=
y 13 reflected in the discussion statement, and I would like to put
=
5 14 forth as a general question how can this be addressed, and how w
2 15 could a document such as this be made more adequate in that E
J 16 regard?
I d
17 i One other general set of remarks that I would like to E
18 emphasize, again elaborating on something I said yesterday, is
=
19 something that seems very curious to me, and really a puzzle, and A
20 that is that the approach to setting safety goals is so much 21 directed from the standpoint of the probability side, the 22 probabilistic assessment, and so little directed from the side 23 l of social value questions, when they seem at least equally l
24 important, or perhaps one could argue that the value issues are 25 ;
even more central to the statement of what the goals should be.
ALDERSON REPORTING COMPANY. INC.
3 1
Let me also just emphasize that in any statement of 2
goals there are all these implicit assumptions that are value 3
laden, and just to go through a half a dozen that appear to be 4
inherent in the present document, but not always spelled out as e
5 such. I just want to list a few of these.
A N
s 6,
one is'that there is some notion of risk aversion here e
7 in the document, that is, the fact that large accidents should be s
5 8
given special treatment, and I wonder whether this is-- this is N
d
=
9 an assumption.
I wonder whether there is a corollary here that Y
10 small accidents need not be given any special attention. I sort 3j 11 of question whether that really is the case.
5 d
12 Also, there is an assumption here that we can deal with E=
13 issues of equity by making the stated goals small enough so t hat E
14 we don't have to worry about it.
If it is the case that du!
15 verifiability of the low probability levels is in doubt, then one E
16 wonders whether or not one can simply pass off the equity problem 3
A d
17 ;
that easily, because who knows whether the probabilities are that M
18 l small, and there are some serious questions there.
=
H E
19 There is an assumption that current levels of risk from A
20 competing energy technologies are meaningful benchmarks against 21 which
-o develop a standard for nuclear power, and I think this 22 is open for debate.
There is implicit in this document, I think, l
23 {
a rather narrow focus on what the relevant costs are.
I think 24 this is sort of oriented towards health effects as the basic cost 25,
to consider.
ALDERSON REPORTING COMPANY. INC.
I 4
f i
i l
One cculd consider a much broader spectrum of costs j
i 2l f accidents that take into account a wide range of economic and 3
ther social consequences that would stem from nuclear accidents.
t 4l That would imply very different goals.
5l There is the statement that some risks are unacceptable e
A I
N 6!
regardless of benefit, that there are no benefits that can e
m j
7-offset certain risks, and this seems to be a very strong value
~
8l laden or ethical statement which I wonder if -- whether or not l
a g
9j that is justifiable.
Y S
these are just a few of the value issues that are E
10 i n
i j
jj inherent in this document, and I would just like to open the
.i 12 floor to discussion about the general problem here.
David?
z I
I 3
13 -
MR. OKRENT:
I would like to take your comments and
=
0.
E 14, use them to. pose three questions to you.
The first one is, what d
I 15 ! criterion, qualitative or quantitative, would you propose with d_
I T
16 regard to risk aversion, since there is none?
The second is,if j7f you don't like the proposed benchmark, what benchmark do you 5
E 18 propose and why?
E t
j9,
And the third one is, with regard to equity,what y
n 20 ; number is small enough that if you have reasonable confidence 21 l y u are there or below, the equity problem is in fact handled a
22 ; in that fashion?
1" 23 MR. SLOVIC:
With regard to risk aversion, I see --
24 a first of all, I see all these questions as questions that need
- i 25, addressing by a broad group of people.
I only have my own l
i I
0 j
ALDERSON REPORTING COMPANY, INC.
I
l 5
l 1 ; opinions on this, and I have a more specified opinion with regard I
2 to risk aversion than some of the other questions that you asked.
3 I think that we don't know exactly how to treat risk 4l aversion.
The problem is one -- is a confounding between aversion i
e 5
to large accidents and the problem of uncertainty,and it is not at 9
M 6i all clear to me that ir there was no uncertainty in this area of e
Rg 7
risk, that if we knew what the probabilities were, and we knew s
3 8
what the magnitudes of the accidents were, we could say for sure n
d I
d 9l that there is a very low probability of an accident that takes, Y
E 10 you know, 800 lives, but we knew what that probability was E
E 11 Precisely, then I don't think we would find the same sort of 5
d 12 aversion to the accidents that we find in a situation where we E=
d 13 don't really know what the probability is within several orders S
i S
14 of magnitude, and where the consequences perceived as possibly db 2
15 being many orders of magnitude greater than 800, so it sort of x=
.]
16 an unbounded character to the perception of nuclear risks which A
d 17 :
leads people to act in a risk averse way but which shouldn't x
5 18 l enter in necessarily to the modeling of risk aversion.
U 19,l
[
I don't know that I'm being very clear here, but I think
=5 i
20 !
that dealing with risk aversion is a modeling problem much like 1
21 '
the modeling that goes on in probabilistic risk assessment, and I
22 that we need to unconfound questions of uncertainty from 23 '
questions of attitudes towards large accidents, okay, because I 24 J think you can see in society a very inconsistent response towards ll 25 large accidents.
?
ALDERSON REPORTING COMPANY. INC.
I 6
I l
1 !
There are many potentially large accidents such as i
I 2i those with dams that, you know, where there isn't much social 3
concern at all, and I think that is a function of the fact that 4,
people see the dam accident as sort of isolated and bounded and g
5 certain other ways.
N N
6l MR. OKRENT:
You are being clear but not very e
i R
R 7
specific, and you are not telling me what Slovic thinks.
3 8
MR. SLOVIC:
Well, I think that there should not be any l
0 i
9j special -- that there shouldn't be a risk aversion coefficient d
5 10 built into this, that one should treat, in terms of societal d
I 11 risk, you should weight the number of lives, expected number of B
J 12 lives lost by -- you know, you shouldn't weight it one way or s
5 13 another.
The alpha would equal one.
That probably if one E
i E
14 l studied this issue carefully and tried to unconfound issues of w
15 uncertainty from it, that an alpha of one is probably the most
]r=
16 i defensible.
s j
4 l
d 17 MR. RATHBUN:
Let me just -- I think -- in other words I
18 '
'ather that what you do, Paul, is have a qualitative treatment, E
19,
and Juld not do as David suggested in his proposal, an 0
explict and a quantitative treatment, and I am wondering.
20 l l
21 l
.R. SLOVIC:
Well, one could argue quantitatively 22 i that if you want to minimize the number of expected fatalities, 23 '
that an alpha of one is recommended from that perspective as l
l 24 !I well.
I d.
l l
25 MR. WHIPPLE:
I want to jump in on this question.
I 1
t i
i ALDERSON REPORTING COMPANY. INC.
.. -. =
)
7 4
f I
i I
1 agree with your answer.
I think it is half the answer.
But the 1
2I second half, or the question now is what sort of attitude do you 3
have towards uncertainty, and do you have to develop a procedure i
j 4
for being averse to highly uncertain risks as opposed to better
)
e 5
known risks, and how do you do that?
l 9
l I
E 6i MR. SLOVIC:
Yes.
Some people want to respond.
Let me o
l R
i 8
7 just, since you addressed it to me -- I think that that is the I
~
j 8
way to address the risk aversion problem, is by deciding what I
dc 9
your attitude towards uncertainty is, and how much conservatism Y
10 you want to build in due to the uncertainty, but not necessarily 3
3 11 due to an aversion to large accidents.
Tom, you had a --
i i
<5 i,
d 12 MR. COCHRAN:
I want to disagree with your statement z
4
=
j i
d 13 l about alpha should be one.
I think you unnecessarily complicate j
E i
i 14 the problem by using these Greek symbols, and that risk aversion, d
j 15 if you believe that there is such a thing out there, I think d
j j
16 dictates that the goal or the rule be that --to ensure that you
^
l 1
y 17 ;
don't have any big accidents.
I d
l 18 '
You don't have to talk about alpha being one or.2 or
?
E 19,
whatever.
People don't want accidents.
They don't want these i,
5 i
]
20 big accidents, and that has got to be part of the -- part of t'
21 your regulatory requirement, and if you can't show it, if you 22 can't demonstrate that you are there, there is going to be an I
i 23 equity problem, and the fact is you can't demonstrate that you j
24 I are there.
l i
9 25j And the more you use the probabilistic risk assessment,
[
ALDERSON REPORTING COMPANV. INC.
i
-,____.,m-_,
-,_,,-,,-y-
f 8
I i
1 l I think, the less confidence people are going to have that you 2'
are close to it, and that is why I would go back to putting the 3
goal, one of the goals on just no catastrophic accidents, no 4
core melts with significant fatalities or -- and contamination, 3
l c
5l and that I think, if you could demonstrate that you could meet 8
i j
6I that, David, I think you would have met your equity limit problem.
K I
s 7'
MR. SLOVIC:
Neal?
Go ahead.
A j
8.
MR. OKRENT :
I must say, I think the equity is an G
9 individual risk thing to me, and so I visualize it somewhat i
3 10 differently, but I would rather not depart from the risk z=
i j
11 aversion discussion that is still going on.
u j
12 MR. SLOVIC:
Tom, with regard to your statement, I E
i j
13 ;
think risk aversion is used in different ways by different
=j 14l people, and maybe we should try to clarify a little bit the use.
s i
=
I
{
15 '
One was of, you know -- we are averse to large I
16 -
accidents, that they are bad, they shouldn't be-- we should do j
i p
17 everything we can -- we shou?l try to take proper steps to E
18 prevent them.
E I
19 '
A somewhat different use is, we should take specially 5
f n
1 20 l
-- you know, we should put special emphasis on accidents that l
21 l might take a large number of lives, even if they are, you know, I
u il 22 ] very unlikely, just because losing many lives in a single ai 23] incident is particularly aversive.
It is even worre than -- just 24j because of the sheer number of lives, there are other sorts of l
25 things going on.
That we should devote special emphasic to that.
l.
d ALDERSON REPORTING COMPANY. INC.
l
9 l
l 1
j 1 h And you know, I think that is -- the way I interpret your l
{
2l statement is that catastrophes are bad, you know, because they 1
I 3
take so man, lives, and they should not be allowed, but that 1
1 j
4l doesn't necessarily imply a risk aversion in this other sense.
i e
5 Doug?
3 P.
I N
6l MR. MAC LEAN:
It's not -- I don't think the point 4
-Tg 7
is so much just that catastrophes are bad and they shouldn't be 4
s E
8 allowed.
I agree entirely with you that some catastrophes seem i.
e
=
9l to bother the public much more than others do, and although this i
i I
j
@ ' 10 may really muck up any attempt to treat risk aversion in a I
z 4
=
i 2
11 quantitative model, I think that we have to be prepared to face i
?
)
d 12 the fact that people might be averse, more averse to some risks z
i 5
l 3
d 13 j than they are to others, even though quantitatively they turn out 5
E 14 to be equivalent.
I mean, people just -- I mean, there are d
i
=
i 2
15 l qualitative differences in different risks, and I think these are d
i
.)
16l terribly important to a lot of people, that people just don't l
j p
17 ! want to take certain risks and are not so bothered about others.
5 l
1 7
18 l That is a problem with any single number that is
)
5
}
19 supposed to measure risk aversion.
Uncertainty is one thing, 1
a l
20!
but I mean, hell, I took my motorcycle down to the hardware store j
l 21 l to buy a smoke detector device for my house the other day, and I i
l 22 '
don't think this is irrational.
23 (General laughter.)
4
>R. O 'lX}NELL: You know, risk aversion seems to be --
l 24,i.
i, 25j it seems to support the view that it is all right to kill 50,000 I
i i
i i
d a
ALDERSON REPORTING COMPANY. INC.
I i
.~
r 10 I
3 1. or 100,000 people one at a time over some period of time, so long i
l 2j as you do it so it doesn't create a fuss, but if one kills 100 3
people, such as in the Kansas City hotel accident, or 50,000 4
people in a large accident that would occur once in a million I
e years, that somehow is less acceptable.
E n
6 On a rational basis I just can't accept that.
It e
'l R
a 7
doesn't make any sense to me.
Risk aversion, introducing alphas I
s 3
8 and factors like that, seems to support that kind of logic, which n
d 9l is not logical.
=
l E
10 The other thing that a risk aversion factor does, I
,i s_
j 5
11 think, in practice is that it does in fact -- it introduces a
's d
12,
bias in the regulatory arena such that resources and attention z
i
=
j s
13 are driven to those very low probability,large consequence E
j i
E 14 i events, with the result that necessarily less resources are 4
w b!
15 given to those accidents that may be more probable, of less a
1 16 consequence, that really make up the majority of the risk.
s A
i d
17 l So I think -- I agrees.with you that we ought to s
i 5
18 i be neutral on this, and that we should not introduce factors that I
- s j
[
19 l introduce this bias.
3 n
20 With respect to the questions of equity, I don't think 6
I j
21 i we are focusing very;well.
The equity question is, as Dave said, 22 an individual risk question, and it has to be addressed in that i
23 manner.
4 i
24]
MR. SLOVIC:
Chip?
I 25 MR. SILBERG:
I guess this gets back to the basic i
a 1
ll ALDERSON REPORTING COMPANY. INC.
11 1
i i
j j j question of what is the purpose in a safety goal, and I think what Tom is saying is, the safety goal ought to reflect 2
pe pie s perceptions, people out there, and I.think what Ed is.l.,
3
)
4l saying is, no, the safety goal ought to be a more normati e i
e 5
requirement that ought to reflect what people ought to think, g
I n
8 6j what rational people sitting around the table agree on.
4 7
My own feeling is that a safety goal ought not to be 8
what Laople consider as less rational views of life than other 1
N views of life, and if we agree that taking 50,000 lives one at a 9
1 2
S time is really about ther, same as taking 50,000 lives all at 10 c
3 i
g jj once, then we oughtynot to prejudice the regulatory outcome by
<>}
1 j2 adding some additional weight impact.
3 13 l MR. SLOVIC:
Norman?
i e
t l
S 14 l MR. BRADBURN:
I would like to make a slight protest 4
w H
f 15 against this excessive individualism that we have been assuming.
d l
~.
There are other units ada
..C ues in society besides the 16 l, 3
w
]
g j7 l individual.
There L
... Lies, and there are kin groups, and 4
S i
there are communities, and I think there is a sense in which j
E 18 1
(
J j9 !
two deaths in one family, or a whole family being wiped out, are 5
i n
j 20 valued in a different way than distribueed deaths whica are not i
l
]
2j !
-- which are different, families or communities or whatever the J
l units are.
22 1
23,i I think this is the sense in which -- at least one j
24 ] sense in which it is rational, at least from some value 4
25j perspective, that collective deaths are valued in a diff' rent way i
4 ALDERSON REPORTING COMPANY, INC.
_..__._y..
.v., -
2-twr
w
_=
12 e
k 1
than the same number of individual deaths.
i 2
I think also the point -- we shouldn't leave the point i
3 about the relationship between uncertainty and the feeling about i
4 the certainty -- the probabilities that one wants.
It is a point i
5l l
e
. that came up very' late in our discussion.
We mentioned ic i
0 3
6l yesterday, and I think it shouldn't be lost, the point you were R
i 7
making.
sj 8l MR. SLOVIC:
Helen?
o 9
MS. INGRAM:
I just would like to second that remark d
b 10,
about probability.
I think any time when experts' cerceptions i
3 i,
11 are so far from the public, it is arrogant to assume that the j
B
(
12 experts are right.
It may be that they have defined the problem i
E
\\
l s
13 i incorrectly, and that they need to look harder at how the public
=
]
n l
5 14 is looking at things, and it seems to me pr, sibly rational that w
i
=
_l 2
15 given large amounts of uncertainty or perceived uncertainty, a
i
=
1 y
16 which may be justified, that large risks are harder to take in,
^
l 6
17 l say, a great uncertainty, especially when the results are very w
i
=
M 18 serious.
1
=
w
}
19 Prceumably, if we are talking about 50,000 deaths that 4
M r
j 20,
may occur one at a time, the real probabilities in the course of l
21 those events may become much more clear, so people can
)
{
22,
determine their positions on things.
A catastrophic accident t
j 23 '
is complete.
All 50,000 are gone at once.
That seems to me a i
]
24 l perfectly rational point of view, and I think we need to look l
i 25 harder at how the public feels about things, not immediately t
i ALDERSON REPORTING COMPANY, INC.
I i
T
,5 n
--w
---,----,,-n,n
l 13 i
1 assume that because you are a Ph.D. in economics or a Ph.D. in i
2i psychology, that you have a particular corner on the question of l
3 f values, which is really an individual and a human matter.
We need 4,
to understand, not presume we know.
5l MR. SLOVIC:
I agree with tha.
c E
l n
G 6;
MS. INGRAM:
Thank you.
o 7l E
5 MR. SLOVIC:
Ellen?
j E
l j
8f MS. WEISS:
I think that to suggest that 50,000 deaths d
I 9l strung out over some period of time would be an equivalent event
=
Y l
10 to an event which causes 50,000 deaths all at once is z=
t i
manifestly untrue.
I mean, one has to look at what would g
11 a
d 12 ;
accompany an incident that would cause 50,000 deaths.
You would z
=
i 13 have -- it would be an incident where you would wipe out some j
14 ;
community or numbers of communities permanently.
It would b
i 15 contaminate hundreds of square miles permanently.
Thousands of a
y 16 j square miles would require decontamination.
So you are talking s
17 about an event that in human terms has far greater consequences
=
5 18 :
than 50,000 people dying over 50 years.
i
?
[
19.
MR. SLOVIC:
Toby?
5 20 ;
MR. PAGE: I agree with a lot of what has just been said, and I think it is clear that there is no -- and it would be hard 21 1
22 'l for experts sitting around a room to come upuwith a number 23 that -- but I also think it would be useful if we distincuished 1
i 24j two sources for what is ccamonly called risk aversion, one being l
r 25.; the classical one that is treated in economics -~ which has to 0
i ALDERSON REPORTING COMPANY, INC.
i 14 i
k i
i i
}
1 ! do with the way individuals value low consequence, high 2 a probability events versus low probability, high consequence 3
' events.
There is a whole literature and a lot of experimental 1
i 1
-4 literature on this.
It is a clearly defined technical concept.
i 1
l e
5 It is useful to think about it in these terms.
1 0
j 6!
The second way of approaching it I think gets mixed i
i R
R 7
in and confused, and that is that when we are dealing with low l-s j
j 8
probability events, that there is a great deal of concern that j
d 4
c 9
the estimates, the probability numbers themselves are more -- and f
5
_4 l'
10 if we have an estimate of an accident of, say, 10 and by the z=
-3 g
11 way, we don't know whether it might be 10-5 or 10
, and we put 3
4
-5 g
12 a -- on all of that, putting half the weight on 10 and half i
=
{'
j 13 the weight on 10 leads you to a different expected value
-3 I
_4 l
l 14l entirely than putting the entire weight on 10
, so if you begin i
{
2 15l to spread out your probability distribution at the low range you I
16 j
get something in a value that looks like risk aversion, in the A
I d
17 ;
i a
' sense that you put more weight on the higher end.
1
=
j M
18 So, if we -- I think if we realize that a lot of the i
=
4 H
}
}
19, concern about risk aversion is that people, including ourselves, l
5 i
i 20 l do not take nearly as seriously these low probability estimates 21 l as firm numbers as we do the high numbers, 10~
10
-1 s
j j
22 '
frequency information; we believe in them a lot more than the I
I i
23 10-6 That sort of thing.
I j
24 So my feeling is that a lot of the concern for risk I
e, I
i 25 l aversion has to do with the greater concern for the sof tness
)
0 i
V I
h ALDERSON REPORTING COMPANY. INC.
4
o o
15
)
)
. of the analysis when we get down to these numbers.
2l MR. SLOVIC:
Ed?
3 MR. O'DONNELL:
I agree with that, and I think that 4
does in fact introduce a form of risk aversion in itself.
That
.i i
e 5
is, if you are dealing with uncertainty, and if you are not going 3
N 8
6, to make decisions in areas where you have great uncertainties 4
e E
l 8
7 without something else, such as defense in depth, well, that i
I s
]
8 8
introduces risk aversion.
To put another factor in, such as t
i d
d 9
an alpha, is just another bias.
i 5
j 10 with regard to the process, I agree that the experts 3
3 5
11 cannot sit around in a room such as this and say, this is the 4
3
}
d 12 basic goal.
I view this as kind of a two-phase process, that z
I j
3 l
j j
13 i some attempt should be made to come up with a rational or j
j 14 normative approach to setting goals, but then -- and that should 1
t f
15 be presented to the public and to the Congress to see what i
i 1
=
l j
16,
factors might be introduced to account for public perceptions and l
17,
public desires, that really they want -- they may want nuclear w
l 5
18 power to be safer than some other technology, but that is not a 19 normative consideration tha t should be introduced in the
.I b
l 20 beginning.
J l
j 21 i MR. SLOVIC:
Ed Zebroski.
1 22,
MR. ZEBROSKI:
I think one of the criteria that we i
23 '
really haven't discussed enough is the coherency and consistency i
24 ] of -- data, and I think one of the problems one has with risk 4
d
!i I
l 25j aversion coherency and consistency objectives is the ease with d
j i
ALDERSON REPORTING COMPANY, INC.
i
16 1' which risk aversion perceptions are manipulative, and we have a 2
wonderful body of literature by Conaman, Perceby, and Slovic and i
3 Associates on how -- what are the factors that manipulate your 4
perception.
If people would see the movie Jaws, suddenly Miami
}
g 5
beaches look very ri' sky, and you would have a high risk aversion l
0 i
j 6
coefficient, but a rather volatile one.
R j
g 7
So, I think it would fail the test of coherency and i
3 8
consistency, because it is likely to be subject to manipulation i
d
+
d 9
with time.
i I
10 Secondly, why would you put in a risk aversion I
5 l
5 11 coefficient?
I think partly to show responsibility from the
<3 3
d 12 standpoint of a legislator or a regulator, but if the measurement i
E=
4
]
d 13 is a volatile one, I think it has the weakness and public a
o
=
l 14 !
perception of whether it is a reasonable thing to do, and I
i I
2 15 i secondly, I think the whole concept of, I am permitting so many l
w i
16 l deaths as a sign of responsible regulation is basically a loser.
l S
I 1
m
]
p 17 !
So, to coin a phrase, I think thanatology is a dead end.
I E
i 18l!
MR. SLOVIC:
Lester?
=
4 g
i E
19 I MR. LAVE:
First of all, I have a procedural x
I 1
A s
f 20 l suggestion.
In many years of conferences at Brookings, it turns j
21 out that it is easier to manage discussion if when people have i
j 22 something to say, they turn their name tag up like so, and that 23 makes it easier for the chairman to see who has got something j'
24 ] to say instead of waving your hand dl the time, and I suggest we 1
1 1
25 do that.
I.
I j
. ALDERSON REPORTING COMPANY. INC.
I e
I 17 I
1 Secondly, on substantive matters, there are two things 1
2l running around.
One is this question of the alpha value.
- Sorry, i
I 3l Tom.
I think it is extremely difficult to manage the notion that 4l it is worse to have several people die at one time than one at a r
5!
time, in terms of technical analysis, trying to figure out what e
i H
i G
6 ',
you do about that.
o E
h
{
7 I certainly agree with Norman Bradburn's comment that s
j-8!
we feel differently when a family or a community is wiped out, I
O
\\
d 9l but it still is very difficult to get in any mechanistic way of i
i b
10 l having an alpha of 1.3 and do an analysis.
Maybe that is E
E 11I simply an objection to the way that David Okrent formulated it
<a d
12 before.
z E
13,
The second one is that there is a second mattur that is E
i A
14 l running around, and since the Kansas City hotel was brought up, i
E 15 or the MGM fire, I would guess that there is a second reason why 5
j 16 ;
we want to pay particular attention to large consequence A
p 17 '
accidents, and that is, often when accidents occur there was a l
5 18 l technological reason which is reparable.
That is, we can figure
?
i
[
19 -
out why it happened and lower the probability of it happening in l'
5n 20 l the future.
a i
21 ?
And one of the things that the MGM hotel showed us was i
d 6
!i l
22 that # ire control systems and fire alarm systems in hotels 23 I weren 't all that we might have assumed that they were, and that I
i i
24 1 for relatively small amounts of money, that you could have loweredi u
I 25 l those risks, and even on an expected value basis, that the i
i, j
i ALDERSON REPORTING COMPANY, INC.
I
o 18 1
fire protection systems in commercial buildings don't look all I
2 that great, and that has led, slowly, to some changes in fire 3
protection codes around the country for commercial buildings.
l 4i Part o f what the Kansas City hotel seems to have shown i
5!
us so far is that perhaps a little bit more atten". ion during s
N j
6; the inspection process during the building could be used to avert R
\\
I H
7 some disasters.
Again, I think that in the Kansas City case l
j 8
you probably don 't have an implication for design so much as you d
o 9
have an implication for carrying out what is happening.
5 5
10 When you apply those to nuclear reactors, there was an 3
2 11 awful lot of attention devoted to TMI to try to find out why
<?
d 12 that occurred, why it went so far, and I think that people z
E I
d 13 ;
reasonably feel that not only that event but events of that ilk E
14 are less probable in the future, and so I would have guessed d=
E 15 that when you are talking about high consequence accidents, you 16 are talking about failures of some sort that probably are not
]
A
-7
-9 t'
17 described by these estimates of 10 or 10 and a reason for d
i E
18 !
devoting special attention to mitigation of injury in such
=
s r
}
19 cases is that these things might well occur mo.
often than you n
t 20 thought they would because there is something that you left out 21 !
in your analysis.
I l
i t
i 22 ]
And so you might want to pay special attention to l
3 I
23j mitigation of effects of these high consequence accidents, j
24 i because they are teaching you something technologically, tha t i
25 i you don't have a static system.
They are not simply that one in ti I
ALDERSON REPORTING COMPANY. INC.
j 19 j
i
_j l a trillion observation that you are pulling out of the hat, but 1
l I
2 they are saying that technologically something was going on, or 11 at least there is a failure in human beings to try and i
3 i
4I consistently interpret our past mistakes as things that should e
5 have been reparable.
I mean, we always look at something and 9
i 8
6 say, gee, if only I had been smart, I would have seen that e
i G
j g
7 wouldn't occur, and I will change my way in the future.
We tend 8i to do that too much.
But in this case I think there really is a i
n d
j
=;
9 very strong element of that that should be done.
i 3
{
10 MR. SLOVIC:
Yes. We have referred to this at times, z
i 5
11 '
the accidents as signals, the notion that t' 2 impact of an 4
3 l
d 12 accident is a function of the signal potential, what it tell.c us z
5 d
13 about the degree to which this hazard is being well managed, or a
i
=
f 14 is under control, and again it relates to the issues of 15 uncertainty and the confounding of uncertainty and reaction to i
I w=
16 an accident.
i 3
i i
=
l j
g 17 j Helen?
I i
5 18 ;
MS. INGRAM:
I simply wanted to respond to the question i
~
r I
l
[
19 i about the difference in what is really a technical decision and x
i n
20 !
what is a value position.
This document seems to have reacted i
q 21 l to the question of uncertainty about values by deciding not to I
i l
22faddress them at all, by pretending the value questions don't i
23 '
exist.
i l
24,
For example, on Page 7, it takes a number of -- it i
f 25j is followed by a number of value positions begun by the statement, j
t i
i ALDERSON REPORTING COMPANY. INC.
l 20 i
r j
1 we believe that the premises stated below are generally self-I 1
2 evident.
I think that is misleading, because what happens after 3
this are a number of value positions where there is great 4
uncertainty and lots of debate, and the document itself would be l
5 so much improved if those questions were elucidated, if the 9
6l question of what is a benchmark, that it is a reasonable thing j
g 7l i
R
}
to say that this much risk in society, and that ought to be our 1
e
{
y, 8
goal.
That is a value position, and it needs to be stated, so d
y 9
that if people whose job it is to represent the public values Z
t g
10 address that, they know this is something on which nuclear z
a I
j j
11 engineers have no more special information than anybody else.
a
{
j 12 They simply thought hard about the problem, and this is what they 4
1 j
j 13 think.
1
=
l z
i l
5 14 I think it is really terribly important in this s
2j 15 document to have more discussion, and the fact that our group i
y 16 may not come up with a consensus isn't a reason not to do that.
d I7 f In fact, it is a greater reason to do that.
It is terribly I
l j
}
18 '
important, I think.
i 5
I9 ;
MR. SLOVIC:
Allan?
E l
20 MR. MAZUR:
I would just like to -- if you will forgive 21,
me, but as a Ph.D. sociologist I insist that people really don't l
22 i
want to have a catastrophic accident.
I don't think that
~
l 23 '
violates --
f 24 f MS. INGRAM:
I don't disagree with you.
I 4
j 25 l MR. MAZUR:
Okay.
And I also wanted to agree with the i
\\
l t
i ALDERSON REPORTING COMPANY. INC.
I I
j
. _ _ ~ _
21 j.
statements that these contorted risk aversion formulas are just l
2 awful, and I don't see why we can't go to what is essentially 3
Tom's position, and slightly modify it, and consider as a goal 4
that a large accident, however you want to define it, is just not c
5 acceptable, or I guess we would equivalently say, has some very a
f nj 6
low probability of occurring, but we can't quite say zero.
Why 7
indeed can!t that; be an explicitly stated goal without worrying a
3 8
about alphas, which are virtually arbitrary and meaningless kinac n
d d
9 of things?
i 10 Why can 't a straightforward goal simpl'y be that a large E
5 11 accident defined some way, that more than a very tiny probability
<3 d
12 of a large accident is simply not acceptable?
z
~
l d
MR. SLOVIC:
Chip Perrow.
13 l 2
i 14 MR. PERROW:
I would like to recycle through some of_ the i
w 2
15 things that..have been going,on here.
I think they are very 5
l 16 important, and I agree with a number of them.
Okrent started 5
m f
17 i out, when I came in anyway, with a series of three questions about E
5 18
-- that he addressed to Paul, and I won't repeat the questions out 5
4 10 :
they were all tougn questions, very hard to answer, and I think 8
i n
20 the attitude there was that since these are tough questions, we 21 should ignore them, and the evidence that this might be the 22 attitude is because in the report of Panel C for the last meeting 23 we raised a number of tough questions and said that they we e 24 ! being ignored in the draft reports we had before us,and that they I.
I I
25 should be seriously considered even if they can't be answered,
,i ALDERSON REPORTING COMPANY. INC.
._.4 r---y._-_..---
r-.-.
-p
-._.,w
w
-,-,---- - - - - ~ ~,
-r,,
-r<-.
l
- 22 I
i as Helen has just pointel out, and I was appalled when I got the 2,
latest draft to find out that yes, these are tough questions, and I
l 3 l we shall ignore them.
They had not been built into the i
4 deliberations.
e 5
Then it turns out, well, they are not so tough, and E
I n
s 6: there are two lines of thought that are going on here that say e
7 they are not so tough.
One is that one death is equal to another s
3 8,, deauh, and as Ed says, I just can't accept that, that there ir a
d q
9 any difference between deaths.
That, I am afraid, appals me, and
?
E 10 I think we -- I have heard -- I don't have to speak on that, f_
11 because people have answered that kind of thing.
<b
'd 12 !
I much more concerned about one that Lester Lave just E=
0 p
13 l brought up.
I don't know his position on this, so I may be 9.
14 1 misreading the implication, but just in case somebody would get w
l E
15 this implication, the implication is that in our accident is our
/
y 16 salvation.
That is, we are going to learn trom accidents, and if x
d 17 we only have the Triangle Shirt Factory accident, then we will 5
i E
18 l improve the exits and so forth.
Incidentally,..we are probably E
i t
19 I going to have another one, because the problem there was sweat y
i n
+
20 l shops and not the fire safety so much, because fire safety --
l 21 are rarely inspected, and now they are worried in New York City, 22 some people I know, about the possibility of some more of those 23 accidents.
l 24 j But the point was that we learn from accidents, you say,I i
25 ( from failures.
I would say we don't, and this is precisely one 0j ALDERSON REPORTING COMPANY, INC.
l 23 I
l difference between the unrational public and the rational expert.
I 2i The ru51ic response to a large catastrophy is the statement, 3
they.asured us all along that this wouldn't happen, and now it 4l has happened, arid therefore I no longer trust them.
What the 5l experts are saying is, well, if we have a good core melt and we i
e 0
I 3
6 find that it doesn't break containment, th 'o e can be 4
i G
s 7
reassured, or more explicitly, if we cat the generation of
- j 8 l hydrogen which was unpredicted in this accident, then we don't a
i d
9' have to change the licensing rules for the next plant or TMI 1 i
i y
10 l and restart should be able to cope with more than 5 percent z
I
=
l g
11 hydrogen being generated because now we have learned about the u
g 12 generation of hydrogen and we can stick something in there and E
i d
13 l fix it.
So in the accident is our salvation.
E j
14 f Now, the experts are saying, we will get better and E
2 15 better.
Sometimes we will.
We do learn.
They put lifeboats, s
y 16,
more lifebow on boats after the Titanic.
That wasn't really i
d 17 '
the problem.
The problem was on the failure they were dealing s
5 18 j with.
But I think the public is saying, our experience is, you E
19 !
don't learn.
Our experience tells us that things are not 5
1 20 getting safer when you keep having catastrophic accidents, and 21 l I think in this case the public is right, because these systems 22 '
are so' complex, so interdependent, so tightly coupled, that 23 ] there will L; another way to generate hydrogen, not literally, l
24j but t:ere will be another way to generate the kina er problem as j
I I
25j Three Mile Island that can't be solved by-learni g from that ALDERSON REPORTING COMPANY. INC.
l
)
24 I
1 accident.
i 2'
One final point.
We have had at least two accidents i
3 that have been made more serious as a result of learning from I
j 4I T'.aree Mile Island.
The instructions were sent out, don't cut e
5 back on high pressure injection, and in those two accidents the 2
I l
6l operators said that is. silly imt we've got to do it, because
}
R l
7 presumably -- because the order nave come down on the basis of j
8; Three Mile Island.
So even that experience is counter-J 9
productive because the systems are so different, and the i0 1
y 10 l accidencs are so unique and so different that in any one case z
1
=
t i
j 11 you cannot say,do this or do that.
E f
- j 12 MR. SLOVIC:
Chip, how would you apply this to the i
E y
13 I risk aversion concept specifically?
=
l 14 l MR. PERROW:
I am saying the public has a better sense
}
E IS '
of risk aversion, of the e.tcaning of catastrophic accidents than x
t
=
1 J
16 the experts do because the experts say that they are our 1
l j
d 17 1
salvation, we can learn from them.
There fo re, the fact that 4
w t
=
i 5
18 j the public has a high dread of low probability, high consequence 5
19 )
{
}
accidents is not some mistake that they have and which we 20l i
should get them in here and run them through some Baysean i
4 21 statistics, and teach them that they should be worried more about 22 l smokers than they should about nuclear plants, that is not going j
23{
to work.
They have very good grounds for this belief that some 1
24 l4 pecole here would like to say is a not very sensible belief.
t' 25 :i MR. SLOVIC:
How would you incorporate that into a l
J' 1
f ALDERSON REPORTING COMPANY. INC.
i i
a n
,, ~,, -, - -,
.-n-,--
25 I
specification of a safety goal?
]
+
MR. PERROW:
I think the specification for spiety goals 1
3 should go much, much higher than 10~
for the probability of a 4
core meltdown and breach of containment.
{
e 5
MR. SLOVIC:
It seems to me that yoi addressing D
a 6l the problem of uncertainty as opposed to the impact of a large
~
i 1
R j
R 7
accident per se.
i i
j 8
MR. PERRON:
They are both bound together. If we had i
d
.i d
9 uncertainty about this, if we were going to have -- if we had i
5 10 1,000 planta, we could be sure that there was going to be a l
i I
I 11 Three Mile Island but worse every three years or four years, 2
1
)
d 12 l maybe we would learn, but I don't think we will ever get to that z
5 i
s 13 point, so the uncertainty and the catastrophic thing, those i
E 14 issues are intertwined.
I w
l i
2 15 MR. SILBERG:
You are saying there is no way to i
w i
j y
16 reduce the uncertainty.
1 s
i d
17 i MR. PERROW: Eight, and I think this has to be addressed i
w i
=
l j
18 j in the safety goal report, and I think it just has not been i
5 j
E 19 lr addressed.
]
A 20 MR. SLOVIC:
I want to call on David. He started all 21 this.
j 22 i Did you get your answer?
Maybe more than you want.
i 1
23 !
MR. OKRENT:
I am glad you did what you just did, 24 ] trying to press people making comments for, what would you do.
In l
i 25 j fact, that's what I tried initially to say, Paul Slovic, what I
3 i
I I
ALDERSON REPORTING COMPANY, INC.
I I
e_,,-
- _. - - -.. ~. -., _. _ _.. _..., _, - _ _ _. _ _ _ _. _.,.. _.. _ -. _, _
l 26 I
would 3,2 do, instead of giving general comments about how what 2-Sege and Rathbun did wasn't good enough, and how they didn't 3
elucidate it enough, and so forth, or they left out a criteria.
4 I think it is more useful to try to propose what should be done, I
g 5j and of course Tom has proposed a point of view, but I think not j
6 enough other people have come in with this sort of proposal, and R
7 I was trying to do that.
A j
8 If I could make one small comment on what O'Donnell d
y 9
said, I guess I have a problem with the industry position that 3
5 10 if we look at the larger accidents in some kind of risk averse
_3 11 way, we will be not looking at small accidents for some reason u
N 12 or we won't have enough resources.
I think that is somewhat of E
i j
13 l a fiction, and it is time that the industry stopped saying it i
z 5
14 that way.
E 15 In fact, the concern with the small accidents that j
16 they are talking about is that they might become big accidents, I
N 17 either big economically or big for the public health or both, 5
i E
18 l and really, let me just make that plea to the industry.
The NRC C
i k
19 '
g l
in fact is risk averse.
The AEC was risk averse in certain ways.
=
20 l You know, they did not agree to put reactors in the middle of 21 New York.
That was a kind of risk aversion.
But they don't have i
22 a stated policy per se or safety goals.
I think you find it 4
23 there, and I guess I would still advance the point of view that l
l 24 ] even if you have some kind of qualitative criteria in risk l
l i
25 ) aversion that might be, I am not going to site in populated l
0 l
l l
ALDERSON REPORTING COMPANY. INC.
t l
l 27 I
1 areas or whatever it is, you can if you wish use a mathematical 2l form, and I am not going to defend the simple one that was shown 3l in NUREG-0739, to provide incentives.
In other words, when you 4!
put, for example, an alpha of 1.2 or 1.3, and if there is an 1
I g
5 ALARA criterion, this then provides an incentive.
It costs you j
N 6f more money to take a site that is closer to more people, or R
s 7'
closer to some more expensive, more valuable economic resource, s
f, 8
than a more removed site, given those two choices.
U 9
?-.
You have more money that you are forced to think about 10 for improvements, as it were.
/
i II l 4
So, this was really the perspective from which I myself N
12 l and Mike Grisemyer said, we think it is reasonable to put in some E
i a
13,
kind of risk aversion into what I would call the mathematics,
=
g
=
i w
i 5
I4 l okay, as an incentive in the ALARA area.
5 I
E 15 '
MR. SLOVIC:
Dale?
s I
I6 g
AR. BRIDENBAUGH:
Yes. I just wanted to support this zj 17 '
idea that there should be -- we should be aole to define, not
=
l
}
18 this group, but it should be possible to define a large accident
=
r g
19 l that is not acceptable.
There are different ways that you car i
5 20 i do that, and of course if you make the alpha high enough,it 21{
happens that way, but I think it would be easier and more l
22 !
understandable to the put'lic to just make a statement similar to 1
23 '
what Tom Cochran is proposing.
l l
24]
One of the reasons that I think that is desirable is, l
25
! we talked about it quite a bit yesterday afternoon, and that is, l
i s
)
ALDERSON REPORTING COMPANY, INC.
28 l
- i a
1 '
what is the meeting of a safety goal?
Is it a requirement?
Is
]
2 it something that you shoot for?
Is it binding on the i
3 regulatory agency and the industry?
And we really didn't answer i
4 that question.
But in observing what is going on at this meeting a
g 5,
and the past meeting, I can't help but have the' feeling that the i
0 j
j 6!
intent or the way this whole thing is heading is to develop a i
R j
7 numerical goal that pretty well represents what the~ industry l
s j
j 8
feels has already been achieved.
Therefore, don't set the goal i
0o 9
higher than you think you can defend as having already b een 3
i i
e I
]
g 10 j obtained, and I don't think that is a good idea.
I think if you z
i
=
i j
11 are going to set a safety goal, you ought to set a safety goal 2
i g
12 that gives you something to shoot fo r, that demonstrates very 5
I i
j 13 l clearly when you haven't met it, and therefore you have to take
{
=
l m
i j
g 14 l specific action in order to bring about improvement as time goes i
2 15 !
by and as you get more plants.
5 i
f 16 f MR. SLOVIC:
Steve?
l
=
f d
17,
MR. DERBY:
I have two comments.
One of them is, there u
=
]
18 l seems to be a rational-non-rational notion of one or two deaths 1
=
l 1
19 '
added together are a large number of deaths all at once.
It is 5
I 1
20 linear to think that 50,000 deaths one at a time is the same a
s 21 I thing as one 50,000-death accident.
It is linear thought.
It is 22 ;
not sensible,.it is not rational, it is merely a very narrow l
23!
definition.
4 l
24j So I would like to take out this notion of sort of l
a i
1 25 ] righteous words of rationality and sensibility compared to i
1 i
I I
4 i
ALDERSON REPORTING COMPANY, INC.
J
~.
29' l
j 1 ;
non-rationality and insensibility in trying to discuss the i
4 2
effects and the value judgments of large accidents.
I think they 4
4 I
3 are words that don't really lead you in the right direction.
4 The other point is also on the notion of formal models, 4
1 j
e 5
and that is the use of alphas, the Von Neuman Morganstern risk E
{
j 6l aversion concepts and things like that, I would hate to see g
7 brought into a standard.
I think you can use that as a guide to r
a j
8 try to come up with what the goal is, but I think the problem d
d 9
is richer and harder than the problems that these people have i
i O
t g
10 solved, and so to formalize this equation as a framework for E
11 talking abo:t the problem I think is narrowing the. focus to the j
j' 3
12 l j
j point where you can't put a lot o.f things into it.
E l
j j
13 If you are going to make something formal, make it
=
)
14 '
nice, clean, and simple.
Getting to that formality is probably 4
w i
k i
2 15 pretty hard and sticky.
Certainly not linear.
That is a hope.
]
5 l
y 16 {
If you are asking me what I would do, I would take it out.
A L
)
d 17 MR. PERROW:
What would you propose that is nice, x=
1
)3 5
18 clean, and simple, if I can follow --
P i
19 j MR. DERBY:
As a goal, I would have a complementary 1
5 t
i 1
20l.
cumulative probability distribution on fatalities, and I would 21 draw that as a limit curve.
I would not --
1 I
l 22 l MS. WEISS :
Steve, could you explain that?
3 23 '
(General laughter. )
24]
MR.OERBY:
A complementary cumulative probability 25j distribution.
It is a risk curve.
It is not frequency, it is i
H ii ALDERSON REPORTING COMPANY, INC.
1 i
,n.~v
,.n.,--
-.-.,.-e.
,e n
30 I
l 1
probability.
It starts at one and falls to zero, and the bottom 4
I
{
2 is -- fatalities, and we call that a nice, clean representation.
-3 Lester Lave had a longer word for it, but it is a nice 4
probabilistic concept.
The equation is calculated.
There'is
]
g 5
no dispute as to what the devil it is.
And that would be a goal.
t 0
j 6l I think that could be forged out of incredible. debate on where I
1 g
{
7 to draw that curve, and how far does it go, and where does it s
]
j 8
-start, and you can meet that, if that is your goal.
It is a very
}
d
[
9 simple calculation to meet.
But how to get there by solving 2
10 another set of equations, I believe, sets a very narrow agenda z
i
=
j j
11 for getting to that goal.
I think the use of formal models to j
12 get there is closing out a lot of opportunity for debate and I
5 13 l
{
g a lot of information that doesn 't come in this nice formula 's i
m i
j 14 parameters.
I e
i 2
15 MR. S LOVIC:
Chris?
{
I 16;I j
g MR. WHIPPLE:
Yesterday we talked about how we needed l
i I
d 17 :
to think of the goals in an operational sense, and that is,
$c
+
w 18 consistent for a plan for how they would be implemented, and
{
}
19 l while I can conceptually see some merit in the idea that Allan l
5 i
j 20 l and Tom and Dale have mentioned of having a consequence above a l
]
21 certain amount cannot be permitted unless the probability;is i
l 22 less than some extremely small number. The difficulty is the 1
4 23!
unverifiability of anything you might choose there.
i i
{
24 ;i As we talked yesterday, you've really got to pick your l
l 4
l 25 j shots on where probabilistic assessment can help you and where s
l j
ALDERSON REPORTING COMPANY. INC.
I
+
mr-me
.e.,,,,--n-,
-m-----n-n
-e---
,,~m-mm
-n,w,--m w e e
---w
r-,
---v-w~--w--r-,-
,-----mv n----r-w e--
mm
-w-
--ev--,<
[
31 4
l you can use probabilistic safety goals, and where you have to 2i depend on traditional judgment, deterministic type criteria, and l
3 p I am afraid this area is one where PRA is at its weakest, where i
4{
the uncertainties are the greatest, and the likelihood that you I
5l could ever do anything even if you had a goal I think is very e
M i
n 8
6 small.
i e
7 Now, Lester's point that the probability of 8
catastrophe is often much greater than the assessed probability, n
i 0
i e
9l that there are outliers that you find and presumably clean up i
l 10 '
each time they happen, but you still sit and wait for the next E!
11 outlier to show up, I think is very accurate, but I think that B
d 12 brings us back to the point you made initially about being z
5 5
13 averse to uncertainty, and given the impossibility of i
e i
14 !
verification at these low levels, and the large uncertainty that a
'c!
15,
is bound to be associated with them, I think we are ahead of the s
l J
16 l game addressing the uncertainty directly rather than the s
A i
i 17 consequence.
5 E
ja l MR. SLOVIC:
Ed?
i
?
[
19,l MR. O'DONNELL:
I think that one thing that is missing
=
5 i
20 !
from the goal proposal set forth in this document is some way i
I 21 I of dealing with, as Dale said, the need to do better than we 22 have, and the goals I don't think should merely state a number 23 below which if you can demonstrate you have met that, saying i
I 24 j you are home and free.
j 4
l 25 I think there should always be this concept of i
i l
l 4
i 1
ALDERSON REPORTING COMPANY, INC.
I
I 32 1 ! ALARA or cost benefit balancing.
That is, if you can improve I
2i -- or you can make things safer at a reasonable cost, one should i
4 3
do that, and one should have to do that.
The ALARA principle is 1
l 4
stated in very vague terms in this document, and I think it is i
g 5
very important that this is to be an operational tool, that that l
0 i
]
j 6!
be very explicitly stated in quantitative values for decision g
j 8
7 making in some sort of cost benefit terms.
2 s
i 8
MR. SLOVIC:
Allan?
1 m
d d
9 MR. MAZUR:
I would like to make a bit of a counter-
.j h[
?
10 argument from my own position earlier, and that was that I 3
l l
11 didn't think there was any way of showing that goals were i
3 j
j 12 implemented with PRA, which;I would still maintain, but i
E i
y 13 interestingly, there is a way o f showing that goals aren 't t
j j
14l achieved by PRA, which is, of course, also very uncertain, but i
b l
15li nonetheless does wave a flag for caution.
w=
j 16 Now, in this particular case, judging from Chris's A
)
p 17 l recount to me of some of the discussion of last night, there w
f x
i j
18 was a notion that the probability of a core melt is what, 10-4 3
P
{
19 ;
was the figure, and that given a core melt, that there is a M
l 20 l flat probability distribution for a very large number of t
I 21 deaths, and that that presumably, the condition of probability
?
l 22 is fairly high of a third of a death or more, which suggests 23 ' without too much problematic assumption that the probability;of I
24 some moderately high number of deaths is not too many orders of l
8 I
l 25 l magnitude different from 10-4 t
ALDERSON REPORTING COMPANY, INC.
1 33 I
l 1
l 1 !
Now, it strikes me that there is an argument that says, j
l l
2l 1
I
. gee, there really is a probability of having a fairly large
{
i 3
number of deaths at a level that is somewhat higher than many 1
t i
4 of us are thinking that there ought to be, so in that sense it I
I 5l seems to me quite realistic to put these small probabilities on e
O 3
6 catastrophe, not so much because it is difficult to verify that J'
n l
8 7
you have achieved it, but it is a point to -- you can attack it b
A j
8l the other way, make the point that you clearly haven ' t d
]
d 9
achieved it.
1 i
l h
10 It seems to me that seems to be well within the
?
1 j
11 calculation of probability, and in fact may even be the case.
a d
12 MR. SLOVIC:
Again, I am not sure that that is pertinent 3
I j
g= 13,
to the risk aversion issue.
I mean, you could say that the m
{
j 14 probability of a serious accident is high enough that we should 1
+=
{
2 15 lower that probability, but it doesn't speak specifically to risk M=
y 16 aversion in the sense of giving extra weight to the large j
17 accident above and beyond the fact that it is a large accident.
=
j 18 MR. MAZUR:
I am sorry.
I was premising the notion l
E 19 again on the thought that one reflects risk aversion very, very n
j 20 directly by having a goal that says that, as Chris was j
21 responding, that the probability of a large accident must be very 1
1 22 : small.
That simple statement.
That itself is a goal.
i.
j 23 l MR. SLOVIC:
There is a terminological problem here.
t l
24 l I think we have been using and are continuing to use risk t
i i
I j
25, aversion in several different senses, and probably we shouldn ' t e
i 1
f ALDERSON REPORTING COMPANY,INC.
i i
l 34 l
i j.
try to untangle that any further at this point, but I think it j
2 illustrates a problem in this whole area that often words are i
i 3
used in very different ways, technical ways, sort of just l
4l common sense ways, and whatever in a document like this I
e 5
terminology should probably be made as clear as possible.
R N
s 6!
I am wondering if we should pursue the risk aversion e
7 question which we, I think, have demonstrated is as complex as 4
I 8
we thought when we came into the discussion,or instead turn to n
0
=
9 the question of levels of safety and what would be an Y
E 10 appropriate benchmark.
This is the second of David's questions, E
E 11 and it was in response to my comment that I saw some very definite
<3 d
12 assumptions in the quantitative goals that were put forth in this 3=
5 13 !
document with regard to the logic of looking around at what J
E E
14 society is accepting from other risks and other energy risks as du!
15 l a guide to determining what would be acceptable in the nuclear E
J 16 industry, and you kl.vw, I don't have an alternative benchmark to Ej 17 propose.
I am not sure that that is a bad way to go about it.
x i
I 5
18 I am just saying that that does make certain assumptions.
It I
E I
19 l assumes, you know, that there is something meaningful about the 5
20 current level of risk that we tolerate from coal fired electrical 21 l plants.
I 22,
I think that you can take this argument, and if you are i
23 :
really serious about it, you should say, well, we really don't 24 ;
need any safety goals for nuclear power, because what is now is t
l a
25 i acceptable, you know, that nuclear power is now acceptable, l
i l
i ALDERSON REPORTING COMPANY. INC.
I
)
i 35 I
1 !
because it is, and why go through this exercise?
It seems that l
2' the logic of determining the appropriate value is really 3
depending very heavily on these comparisons, and there are a lot l
4 of assumptions there, and I think they need to be. debated.
That e
5 was my point.
N I
I 3
6 MR. ZEBROSKI:
Point of order, Paul.
i e
I R
2 7
MR. SLOVIC:
Yes.
N j
8 MR. ZEBROSKI:
I think -- I am troubled that all of the r
d d
9 discussion on this has fuzzed up the distinction between the 10 qualitative and the quantitative goals.
I think many of the E
G 11 statements are much more acceptable in the qualitative goal sense
<B d
12 than in the quantitative goal sense.
Perhaps further speakers 3
5 13 ;
will come back to this.
I thought we had near unanimity among a
I 14 i the panel that in the qualitative goal universe, the distinction d
I u
a 5
15 i of, shall we say, greater intensity of attention on large 5
J 16 accidents was universally ag"eed to, and -- well, nearly so.
E l
g' 17 '
That is what I got from the panel reporter's summary, that there 5
l M
18 '
was considerable --
U
}
19 ;
MR. COCHRAN:
I have a point of order. He has gone i
5 1
20 !
beyond his point of order.
I 21 l MR. ZEBROSKI:
I have to leave in two minutes,so if I 22 may finish the thought, then you can rebut, Tom.
23 '
MR. COCH RAN :
All right.
24 MR. ZEBROSKI :
I think in the quantitative goal, I 25 l just wonder if alpha effects, unless very large, aren't really ALDERSON REPORTING COMPANY,INC.
36 l
ja i
I covered over by the inherent uncertainty of the-quantitative 1
2l issue of verification, and so I -- to respond to the very good k
3s comments by Helen and Ellen, it seems to me we ought to offer i
4!
one other distinction, that one could have a very rational and I
e 5,
acceptable qualitative goal, and one could have as a tool within N
I j
6' the regulatory process to think about these things with whatever R
7i mathematical tools, arcane, esoteric, and incomprehensible though I
,u
\\
g 8l they may be, if those tools are useful to get consistency in how d
E, 9!
you approach the regulatory question, they should be used on 0
10 !
their own merits without regard to whether or not they are z
l 11 understandable to the lay public.
They should rise or fall by
~
p B
l j
12 !
the acceptance of a peer group of people who are critical in the E
i 13 I technology, but they should not rise or fall on whether or not w
i 3
14 -
you can make them unders tandable to people who don ' t like E
15,
algebra.
In other words, they are tools as distinct from goals.
d 1
y 16 '
MR. SLOVIC:
Well, I think that is touching upon z
d 17 a very important issue, the degree to which, you know, a E
G 18 l technical group can override the strong feelings of the public 5
l 19 ;
on whose behalf they are --
5 l
20 l MR. ZEBROSKI: No, I am not saying that at all.
I am i
21 l saying that the qualitative goal must-- that is why I am l
j i
22 j agreeing with Helen's comment especially, that the qualitative J
23)l rule must indeed be in some kind of synchronism with the 24j realistic public perception, but implementation of that with I
i 25 some degree of consistency may require some quantitative tools l
l
-ii ALDERSON REPORTING COMPANY. INC.
37 1 l which are not everyday language.
I mean, that is just inherent I
2 l in the way ti.e technology works.
There are many examples of 1
3 this where, you know, the dentist will do things which he will i
4l explain to you in very simple terms, but which he can only -- may n
5 take years to learn and use very elaborate language to describe.
O i
6l So the distinction between a tool which supports the E
8 7
qualitative goal and the statement of the goal itself, it s
j 8
seems to me, is what we can separate.
d d
9 MR. SLOVIC:
Well, obviously, every member of the Y
10 l public doesn't have to understand the mathematics of risk-Z l
=
1 1
g 11 aversion, but I would think that the process should be open to a
f 12 l critique by representatives of the public who do understand these 5
I
(
g 13 I methods and who are, you know, concerned about them, and that
=
y 14 { there would be the flexibility there to debate the technical l
i l
2 15 I application of the qualitative goal.
S i
l y
16 MR. ZEBROSKI: Put another way, we are not really l
A g
17 j debating whether or not you should use PRA for goals.
The issue w
18 l F
E is that PRA is a tool which has evident limitations, even to the 5
l
{
19lmost enthusiastic exponents, and the fact that it has those A
f 20 l limitations, however, doesn't mean you shouldn't use it for the 21 areas where it has some applicability, but that tends to eliminate 22 l it from practical use in a goal-: sense, because those limitations 23 are fairly fundamental.
24 j MR. SLOVIC: Dennis, did you have a comment?
25]
MR. RATHBUN:
Yes, I did.
I think, as 1 hear it with e
f 3
ALDERSON REPORTING COMPANY, INC.
38 4
l l
1 i regard to the question of risk aversion, one proposal --a 2
couple of things emerged to me.
Highlighting the difficulties of I
3 an explicit or quantitative treatment such as Dave has proposed.
4 Tom has proposed, and others also, Dale, an alternative safety e
5 goal, which emphasizes an aversion to a major catastrophic l
0 6
nuclear accident, and Chris has expressed some misgivings about R
R 7
the verifiability of that.
We will have to go back and think sj 8
about that.
d d
9 With regard to the question that we are taking up now, 5
E 10 I would like to try and get some sentiment from the workshop on l
Z=
l g
11 the relevance of comparisons to acciden.9 in other electrical u
j 12 energy generating technologies, o,r techi. iogies, or whatever, E
13 what members and participants in the workshop would propose that 14 we look at if we attempt to back away from the type of l
E 15 comparison that we have now.
5 l
J 16 The third thing that I would like to get back to E
g 17 i eventually is this notion that we had in the paper now which I I
l M
18 think we will have to re-examine, and that is the self-evidency E
l
{
19 of the premises, and how we should treat -- your advice on how n
i 20 we should treat implicit, buried, let's say, for now, in the 21 paper, value statements.
But I think what Paul wanted to go 22 into at this point was the comparison to other technologies.
23 '
MR. SLOVIC:
Norman?
24{
MR. BRADBURN:
Thank you.
I wanted to ask a question i
l 25 j in response to your thing, but first I will just answer your r
a l
l ALDERSON REPORTING COMPANY. INC.
39 l
I j question directly, at least as to what my opinion is, and that 5
2 7 is to what the proper comparisons might be.
I think you have I
3 to look at comparisons along several different dimensions, and i
4l certainly other energy supply, or particularly electricity, 5l ways of producing electricity, is one relevant comparison.
a 3
I n
i 3
6l However, in light of this discussion we have been having about e
G s
7 !
the big accident kind of problem, t.he collective things, I think a
8 8!
you have to look at some other things which may be outside the i
o 9l energy area, which have the commonality of the dimension of i
=
10 probabilities of big accidents, so you can't just look at one d
5 11,
kind of comparison.
It seems to me you have to look at several, i
l d
12 l and find out what the relevance is.
There may be others, but z
5 I
s 13 '
those are the two that strike me straight off.
E S
14 l Now, hydroelectric dams are perhaps an example oJ du 2
15l where the two come together, and the particular quest. ion that s
i i
j 16 '
I wanted to ask whether anybody sort of knew offhand as to what x
p 17 the probabilities of failure are figured in dam construction on a:
i E
18 l the one side, and secondly, another one that occured to me is r
l
}
19 ;
certain long-span constructions liAe roofs of the Hartford n
20 l Arena or places like that.
21 i Now, those are presumably places where there are l
i l
22 ll standards, and the uncertainty is probably considerably 23 different, because there is longer experience, and I guess 4
24 j in an area of that sort there is probably less uncertai.nty about i
25 what failure rates are, but does anyone know --
i ALDERSON REPORTING COMPANY. INC.
1 i
c 5
40 l
l j j MR. SLOVIC:
Dave Okrent has looked at the dam risks.
I 2
He could probably give-you some statistics on it or probabilities l 3tj in that regard, but do you want to comment on that, Dave?
i.
4f MR. OKRENT:
Well, I note that there have been some 5
dam iV'. lures which led to the deaths of a thousand or more e
M i
n e
N 6 !
people in the last 20 years, but not in the U.
S.
In the U.
S.
o 1
R 2
7I we have had lesser numbers failing.
You do not have PRA's on I
n dams.
They are equally expensive to do, and they may be 3
8l n
d d
9l equally uncertain when you are all done, and of course there are 5
6 10 i many kinds of failures, just as there are different kinds of 2
1 i
11 r accidents for reactors.
We have a spectrum of consequences.
5 I
I d
12 I There are other events you can also calculate like the chance of 6
=
s 13,
a plane crashing into a football field and so forth, and they E
14 '
come out in a similar ball parh.
d e
i 2
15 I agree with you that that is an additional benchmark, d
1 i
.. ~
16,
however.
If in fact nuclear seems to be much larger in this s
r A
d 17 regard than other things, people would have to. I think, think 5
i 18 '
about this.
E b
I.
19 If I can make one brief comment while I have the floor, !
n 20 it relates to this.
In my own opinion, in NUREG-0729, the ICRS l
t 21 i proposal, there was a fairly sianificant sten alonc the lines of l
!i 22 'l -- what you would call large accidents, and the a:proach taken 23 there was the following.
In the first, place, it stated tnat 24 j there should be both prevention and mitigation, and there were i
i 25 requirements proposed on both prevention and mitigation, and if 1
1 1
'l ALDERSON REPORTING COMPANY. INC.
l 41 l
3 i you Jcok at the goals that were asked fcr mitigation, it was 1
2" asked that you meet a one in 100 chance of a large release given l
3 3 but the core had melted.
4:
Now, if you remember Matt Taylor's presentation last l
5.i night, one of his reactors in fact seemed to have met that e
s n
j 6i roughly.
The other in fact had higher probability. So that R
7 would say if you took the numbers in NUREG-0739 for future f
!1 "j
8 plants, it would affect the design of plants so that they were d
9i at least as good as his estimate of one of those.
5 E
10 !
So I would say that in other words there was an effort b
l 2
11 j to reduce the chance of large accidents.
I don't think we tried s
i
=
12 to s ay it was -- should be zero, because I don't thin:c it can be z
E i
5 13,
zero unless you say 50 miles or 30 miles, some very large Ej 14 distance away from people.
2 15 MR. S LOVIC :
Another comment with regard to the dams.
i a=
j 16 While at least as of 1976 or 1977 the whole approach to managing z
d 17 dam safety was to deny that there was any risk, that dams would a=
E 18 be built so that there was no risk of failure.
It was treated l
5
{
19 deterministically and probabilistic calculations were not done.
A 4
20J I think the nuclear industry is far in advance of l
21 that profession in terms of its sensitivity to the need for a
22)recognizinguncertaintyandanalyzingitwithprobabilist.ic l
1 23 I methods.
i 1
i 24 1 Tom?
l 25 MR. COCHRAN:
I have a couple of comments.
One, on ALDERSON REPORTING COMRANY. INC.
l
l
~
l 4,
a i
!l 4
j benchmark, I don't think the NRC should set goals for this I
i 2 ] technology that either propagate or endorse as acceptable i
l 3
l f ailures of other technologies, like unsafe dams or excess 4
i 4
pollution from coal-fired plants, and so forth, and I.think there l
e 5
has been a great deal of that going on in setting numbers, risk E
1 n
j 6
numbers for nuclear power in some of the papers we have seen.
R t
g 7:
I would like to get either Ed or David to help unravel I
s I
i 8
a problem I have.
I proposed, as you have requesed, an alter-I n
l d
c 9, native goal., We make these very low likelihood of any serious 8
1 10 g core melt event with fatalities or major accidents. Now, one d
5 ji. might have supposed that you and the industry, you -- I won't
^
l
<s g
12 say you, but the ACRS members and the industry would jump on that Z-=
d 13 and say, we can do it, and this problem would largely go away,
~=
s ja because in fact that was the dogma of the AEC for many years, d
i 15 that these things were incredible, and it was being done, and s'
J 16 only recently, with -- have we gone through this Rasmussen and j
h' 17 j -- probability risk assessment analysis and come up with numbers y
l 5
18, that suggest that -- and had some accidents like TMI that suggest l
19 that maybe you are not doing it.
1 n
I'
~
20 Then we see these numbers like a goal for probability of j\\
~4 l
21 a core malt of 10 and I am somewhat amused that this number t
22 ; happens to fall right there in the range of what these Rasmussen l
23 l analyses predict is the probability of a core melt, and my first l
24 question is to ask, why can't you do much better?
I mean, why 25 can't you say that core melt accidents should be closer to ALDERSON REPORTING COMPANY. INC.
43 n
l3x10
-6
?
I mean, we had some numbers from this last night that j
4 2i said of the seven reactors we have already built, we can -- one l
-5th 3
of them was -- predicted a core melt of 3 x 10 per reactor W
4f year, and there are lots of things we could still do. We could e
5 put stainless steel rods in instead of zircalloy so we don't get M
i nj 6-hydrogen generation.
We can build the containments better, and Rg 7l co forth.
\\
8; And so mv first question is, why don't you accept a l
0 9 i lower number, and why can't you get there?
And then I want to 5
6 10 '
try to unravel that from the question of, do you pick a higher l
2 E
11 number because of this verification problem, because you can't l
<t d
12 prove you can get to the lower number, or do you really believe i
z 1
i
=
t 13 you can't get to the lower number, o r you shouldn ' t try to get E
14 '
to the lower number except through some sort of ALARA mumbo-dw E
15,
j umbo that I don't see really works very well, but the final a=
1 16 statement I wish to make is --
r j
17 MR. SLOVIC:
Do you want David to answer that?
5 h
18 MR. COCHRAN:
Okay.
I'd like to add also, why y K 5
10-4, Ed, why --
i t
19
?
il 20 '
MR. O ' DO NNELL :
I think the more basic question is, j
21 why pick any number on that particular type of issue, because 1
22 1 core melt is a subordinate -- core melt does not necessarily i
- i l
result in the public f atalities or injuries.
23 1
24 MR. COCH RAN :
It's a defense in depth.
25 MR. O'DONNELL:
Yes, we have defense in depth. So the ALDERSON REPORTING COMPANY. INC.
I
I 44 3
(l j
fact that you may have damage or melt the core does not l
2 necessarily mean that dua public is going to be harmed in any h significant degree.
l 3
t 4
MR. COCHRAN:
If you believe in defense in depth, you i
5!
would want a conditional probability that is low for core melt e
R
\\
n I
6-that is quite independent of the conditional probability that R
R 7,
the containment will hold if you have a core melt.
That is why n
I was --
3 8'
n J
=
9 MR. O'DONNELL:
Well, you want defense in depth, you i
(
.5 10 ;
have defense in depth.
You legislate it deterministically. That 2
t
=
l 2
11 is all we have.
The idea -- the primary safety goalstare l
a d
12,
basically those directly related to public health and safety, L-=
l N
13 that is, protecting the individual, ensuring that the societal E
14 risk is commensurate with the benefits derived.
The only reason Ce 1
2 15 you would have a goal of any sort on the core melt is that you i
5_
}
~
16 want to make sure that you are doing everything, or you are doing s
1:
p 17 a reasonable amount to prevent accidents in the first place, I
s E
18, and this relates somewhat to the uncertainties involved in 1
=_
C 19 l doing the calculation.
Ea 1
_4 i
20 l Now, your question of why not get 10 down to
,I
-5 21 3 x 10 Well, that :aay not be a very cost effective way or --
I 1
22 ; it may not be a very wide group of resources that is putting l
l all of -- that may cost 23 a great deal of money, and may not be i
i 24 cost effective.
You may be better of f putting money into 25 mitigation features, and those decisions ought to be made on the ALDERSON REPORTING COMPANY. INC.
I
l 45 l
l 1 1 basis of the cost benefit balancing.
That is, if you have met h
2 these primary goals, getting below the goals should be a decision n
-i 34 based on looking at the costs and the benefits derived from any I
4' design change.
1 g
5 There is no reason from first principles why j
A 1
-5
-4 3
6-3 x 10 is a better goal than 10 A_
6 7i MR. COCHRAN:
Well, I think the reason I would say, t
8 the staff has the 10~ -- well, no, the staff doesn't have the d
_4
-4 0
9i 10 David has the 10 right?
z.
O i
g 10 L
MR. OKRENT:
They do also, but let me comment on the z
e
=
i j
ll questions before I do have to run to the airport.
I have already w
j 12 said elsewhere in writing that I think it would be difficult to 3
E y
13
- argue with high confidence for a core melt probability smaller
=
r
-4 5
14 '
than 10 In fact, I think it is hard to argue with high b:
-4 r
15 confidence for 10 if you allow for all sources of the E
y 16 initiating event which were not included, by the way, in your i
^
-5 y.
17 3 x 10 number you saw yesterday for Indian Point, because there E
y 18 were a series of initiators not -- or defects and so forth not
?
i 19,
included in that evaluation, even if you accept those numcers, 5
i 20 ] which I reserve the right not to.
1 2I So I have my own technical evaluation which says that J
22 1 this is a problem, to get to that lower number, and it is one of f
1 l
23 ] the reasons, by no means the only one, that I myself think for 24 ! future reactors we should strike for like one in 100 containment 25 ' capability.
In other words, given the care melt, the averace i
c i
ALDERSON REPORTING COMPANY. INC.
s 46 4"
4 3
i core melt, the chance of a large release is small, quite small.
I 2'
I w uld not be opposed, in fact, I have urged a variety 0
3 {l f design changes that would move you in the direction that you
.I 4
would like, a lower probability of core melt, and I think in fact 5!
many of these, if done, would have a salutory effect.
I am not c
R I
n i
3 6i sure because of design errors or whatever and things of this o
N I
g 7'
sort that -- you know, human error doesn't occur only at the I
j 8'
operator level.
You would have a very high ccnfidence in a 0
-5 g
9l number like 10 even if my pet improvements were made of a b
l E
10 ;
Preventive nature.
A i
3 11 So I come down in the end with what I will call
<u d
12 l prevention and mitigation to get in the direction of your z
3 N
13 principal goal of a fairly -- you know, you are fa^.rly confident E
S 14 that the probability of a large accident is really small, and d
15 I don ' t want to depend only on the prevention approach.
=
J 16 MR. SLOVIC:
Ellen, do you have a comment?
z p
17 MS. ING RAM :
My point was back when we were talking s
l 7
18 about comparisons, so maybe if anybody else wants to speak to
=
I t
19 '
this point.
3n 20,'
MR. SLOVIC:
Mr. Corey?
21 I, MR. COREY:
I was going -- as far as I am concerned, li d
22 4 I was going to just say the same thing you were, Ellen.
t i
I 23 MS. WEISS:
I really wanted to second wnat I heard l
1 24 ; to be some of the implications of what you were saying, Paul, and 25 what Tom said, the fact that there have been failures from dams ALDERSC. J REPORTING COMP ANY. INC.
)
l 47 4
1 ; and deaths from coal mining.
It doesn't seem to me to provide 2,
any justification for saying, for nuclear power we will set a 3
goal that accepts an equivalent numbcr of deaths.
I mean, that 1
4 is not a goal, that is just an excuse.
And I assume when the dam i
5:
builders built thos e d ams, that they didn't -- they built them e
R L
n 3
6' the best way that they could with the goal that they would not e
R l
E 7'
fail.
If they made mistakes a nd failed to foresee events which 3
8 led to failure, that is no excuse for people who design n
- J c
9' nuclear plants to tolerate -- to build that failure in in the 5
10 l construction of nuclear plants.
z=
2 11 MR. SLOVIC:
Mr. Corey?
5 l
d 12 l MR. COREY :
In responding to Mr. Rathbun 's question,
z 5
i 13 t it seems to me that the logical comparison to make when you are
+
A 14,
comparing nuclear risks is to compare nuclear risk with the 2:
E 15 '
risk involved in providing alternative energy supply.
I said 5
i j
16 that in the workshop yesterday.
It seems to me that that is ez d
17 a reasonable comparison to make.
Presumably we need energy in a
=
1 E
18,
some form, and we can get it with nuclear power or we can get
=
i-t H
l 19 '
it some other way, and therefore a straight comparison of the A
I 20 ] risks involved in nuclear versus alternative ways seems to me to lbea logical matter.
l 21 lI 22 1 If I may, Paul, since we are moving along, I would like l
23 to address one other question, and I think it isthequestionthat{
i 24 J David 0Arent raised before I came in, although I am not absolutely!
l i
i 25 sure, but at any rate, how should we deal with the problem of i
1 ALDERSON REPORTING COMPANY. INC.
i 48 s
I 1 'i assessing risk and establishing goals.
It seems to me, Number l
I 2 i One, for the time being we should not try to establish fixed, 4
4 a
3 r quantitative safety goals.
Number Two, the NRC staff should in l
I 4
my opinion continue to use probabilistic risk assessments, tools, l t
p.
5l to improve upon them, I think, and to develop whatever other S
i j
6l techniques and tools can be developed to help the URC do its R_
l Q
7l jcb, and Number Three, the results of using those tools should s
i j
8' be implemented as stated on the bottom of Page 25 in the report, G
9i as I said yesterday.
5 E
10 MR. SLOVIC:
Thank you.
I think we have time for one E
=
j 11 ;
more comment, and then Tony Romano has some remarks before the l
3 i
d 12 break.
Lester?
d:
y 13
- MR. LAVE:
Yes, I will be very brief.
Our workshop i
=
l r
g 14 in Palo Alto emphasized that the way of trying to set goals was l
b i
=
I g
15 a to compare comparable risk.
The question is, what risks are l
E i
i y
16 comparable? I think Gordon Corey just said it very well.
It is i
s d
17 clear that when you are talking about generating electricity, that s
I 5
18 you have to talk about what the alternative is. Either the risks
=
19 due to conservation or the risks due to other energy sources.
l 5
4 20 j But you should never look backward at what is going on 21 in coal mines or what went on 100 years ago in coal mines and i
22 i say, aha, see, that is what we have to strive for. When you are 1
ltalkingabout comparable risks, what is ccmparable is a new 23 24 i plant of a new technology that would be built today, as best one 25 can estimate it.
ALDERSON REPORTING COMPANY, INC.
l
49 o
i k
1 ;
Now, nobody is going to permit the kinds of fatalities
$ in coal mines that we had in the forties and fifties or thirties.
2ls 3 l Nobody is going to permit a new coal plant to be built that l
4 puts out the amount of air pollution that plants,in the forties e
5 and fifties did, and so it is very important that one looks at a
j 6'
new plants now to get that comparison.
I hope that that kind of 7l E
thing will draw people together, I mean that kind of statement s
I j
8, will draw people together.
U
=
9 That is, we shouldn't be looking backwards and saying,
'I 10 h gee, just because there is one coal plant in the United States 2
t
=
-3 2
11 that has accident hazards that are 10 per year, that that is 5
8 d
12 !
all we need to meet for nuclear.
z 5
j 13 l MR. SLOVIC:
I would like to thank you for the
=
14 discussion this morning.
d i
=
E 15 i (Whereupon, at 10:00 a.m. a recess was taken.)
s I
i 16 '
s A
d 17 m
4 5
18 !
l 5
19 <
3 9j 20;j t
21 i
l o
22 !
23j l
24 25 1
ALDERSON REPORTING COMPANY, INC.
O NUCLEAR REGULATORY COMMISSION This is Oc certify that the attached proceecings before the in :he matter o fi SAFETY GOAL MORKSHOP - SESSION 4 - SOCIAL EVALUATION Date of ?roceeding:
July 24, 1981 i
Docket !!umb er:
l Place of Proceeding: Harpers Ferrv, West Vircinia I
'4ere held as herein appears, and. :ha: this is the Original :ranscrip:
thereof for the-file of :he ccccission.
I Rossie Sutton Official Reporter (Typed)
.3 J
i A
Cfficial Reporter (Signature)
_