Reliability and Probabilistic Risk Assessment - July 11, 2000
UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
MEETING: RELIABILITY AND PROBABILISTIC
11545 Rockville Pike, Room T2-B3
Tuesday, July 11, 2000
The committee met, pursuant to notice, at 1:00
GEORGE APOSTOLAKIS, Chairman, ACRS
MARIO BONACA, Member, ACRS
THOMAS KRESS, Member, ACRS
DANA POWERS, Member, ACRS
WILLIAM SHACK, Member, ACRS
JACK SIEBER, Member, ACRS
ROBERT UHRIG, Member, ACRS
ROBERT SEALE, Member, ACRS. PARTICIPANTS:
MICHAEL T. MARKLEY, ACRS Staff Engineer
ERIC HASKIN, RES
T. KING, RES
MARY DROUIN, RES
BOB CHRISTIE, Performance Technology, Inc.. C O N T E N T S
Proposed Schedule 4
Introductory Statement 4
Risk Informed Part 50 Framework 6
SECY 98-300 Option 3 Framework 128 . [1:00 p.m.]
DR. APOSTOLAKIS: The meeting will now come to
order. This is a meeting of the ACRS Subcommittee on the
Reliability and Probablistic Risk Assessment. I am George
Apostolakis, Chairman of the Subcommittee. ACRS members in
attendance are: Mario Bonaca, Tom Kress, Dana Powers,
William Shack, Jack Sieber, Robert Uhrig, and Robert Seale.
The purpose of this meeting is to discuss the NRC
framework for risk informing 10 C.F.R. Part 50, described in
SECY 0086 and related matters. The Subcommittee will gather
information, analyze the relevant issues and facts, and
formulate proposed positions and actions as appropriate for
deliberation by the full committee.
Michael T. Markley is the cognizant ACRS staff
engineer for this meeting. The notice of this meeting was
not published in the Federal Register in sufficient time to
inform the public properly. Therefore, we will keep the
transcript record open for ten additional days subsequent to
the availability if this transcript to the public to enable
persons desiring to have written comments and oral
statements entered into the official record to do so. A
transcript of the meeting is being kept, and will be made
available as stated in the Federal Register notice -- as it
is usually stated in the Federal Register notice.
It is requested that speakers first identify
themselves and speak with sufficient clarity and volume so
that they can be readily heard. We have received no written
comments from members of the public regarding today's
meeting. However, Mr. Bob Christie of Performance
Technology, Incorporated has requested time to make a
presentation concerning the staff's proposed framework for
risk informing 10 C.F.R. 50, Option 3, and the proposed
revision to 10 C.F.R. 50.44.
We will now proceed with the meeting, and I call
upon Mr. Tom King of the Office of Research to begin.
MR. POWERS: Bob, before we begin, I need to point
out that I'll have an organizational conflict of interest
over the discussions of 50.44.
DR. APOSTOLAKIS: So noted.
MR. KING: Okay. Thank you, Mr. Chairman. For
the record, I'm Tom King from NRC's Office of Research.
With me at the table are Mary Drouin, section leader in the
PRA branch in research and Eric Haskins, who's I believe a
subcontractor to Sandia, who's one of our prime contractors
on the whole Option 3 risk informing Part 50.
We prepared a presentation to walk you through the
framework, but I know when we set this meeting up, you said
you had a number of questions on it, and I didn't know what
the best way to proceed is. If you want us to just start
through our high level presentation and you bring your
questions up at the right time, or do you have another way
you'd like to proceed?
DR. APOSTOLAKIS: Unless the members have a
different opinion, I think that would help us focus our
attention. So, we can start with the prepared slides.
MR. KING: Okay. What we prepared was sort of at
a high level to walk through what is the framework, what are
the key elements of it, particularly things like defense in
depth and the risk guidelines, and when we get to the end,
some of the key issues, particular the policy issues. So,
Mary will take the lead and walk you through the
presentation. Thank you, Mary.
MS. DROUIN: Thank you, Tom. I'm not sure if it's
Starting with the framework at high level, in
putting together the framework, we put it together such
that, you know, it was going to apply when we looked at the
regulations to risk inform them, but it would go beyond just
the actual regulations, but also to help us when we looked
at the regulatory guides and look at DBS's, et cetera. It
was to help us both screen and formulate the technical
When I talk about screening, to decide whether or
not a particular technical requirement should be in the
regulation, whether it really had any risk significance, and
then looking at the concern that was on the table and in
formulating the regulation, how should that particular
requirement be formulated from a risk informed perspective.
So, given that kind of where we were heading, we
put together a framework. We feel it's what we call a risk
informed, defense in depth approach. Then you ask well,
what do we mean by that, and that kind of leads into the
next bullet. That is based upon the strategies of
prevention and mitigation. In looking at those particular
strategies, we were mapping those directly from the
cornerstones from the plant oversight, the cornerstones for
safe nuclear power plant operation. I don't know if I've
quoted that directly, but we wanted to be consistent with
the cornerstones, and so we're looking at prevention and
mitigation. We'll come a little bit later because there's
strategies for prevention and mitigation which we will also
tie back to the cornerstones.
Then looking at prevention and mitigation, what do
you do with these? We also came up with what we call
tactics, and that would go about telling you how you would
implement those two strategies. The analysis under which
you were going to formulate your requirements would also
take into consideration both your design bases and your
severe accidents so you were able to focus on, you know,
your risk significant accidents.
DR. APOSTOLAKIS: Maybe this is a good time to
raise some comments.
MS. DROUIN: Sure.
DR. APOSTOLAKIS: I'll try to also give you
comments on the document.
MS. DROUIN: Right. We have copies of the
document with us.
DR. APOSTOLAKIS: Which I understand you have
already updated, but I'm sure the comments apply.
MR. KING: Yes, this has been a living document
from the March version to an updated version, and ultimately
in August, we hope to send the Commission sort of a final
MS. DROUIN: You have the update. We gave you the
updated last time, so I don't know which one you're going to
be talking to.
DR. APOSTOLAKIS: I'm going to be talking to the
one before, but I will give you the section numbers.
MS. DROUIN: Okay.
DR. APOSTOLAKIS: As I started saying at the
previous Subcommittee meeting -- that was a Subcommittee
meeting, right? It seems to me this is really what we call
in the paper from the ACRS the preliminary proposal. It's
not just defense in depth. It's defense in depth at the
high level, but then it's rationalist at lower levels.
I believe the document itself says this because in
chapter two, you do have discussion of the structuralist
model, but then in chapter four, under the question number
three, how will uncertains be accounted for in risk
informing existing regulatory requirements. I think that
survived. You have, in the paragraph there, decisions
regarding lower level defense in depth in the form of
redundancy or diversity are generally well suited to this
type of analysis, and that refers to PRA, which is
The other point, when I read chapter two, I mean,
there was this discussion of the structuralist approach and
that this is a high level defense in depth. There is sort
of a vacuum there. It doesn't tell you what to do after
that, okay? However, if you go to chapter four, realize
that the rationalist approach is at the lower level.
So, I was wondering whether the discussion in
chapter two can be edited to say look, this is structuralist
at the high level, and rationalist to the extent possible at
lower levels, which is a preliminary proposal, okay?
MR. KING: Yeah, I think --
DR. APOSTOLAKIS: So, that's all my -- I mean, it
sort of leaves it hanging there, and then you realize, you
know, two chapters later that it's really rationalist at
lower levels. I don't know if anybody -- Dr. Kress will say
that it's rationalist all the way. The point is that the
structuralist at the top, because of the uncertainties, are
very large, but we don't want to push it that far.
The other thing about the structuralist, the
discussion in chapter two, there is a lot of discussion
that's borrowed again from the ACRS letter, but it seems to
me that the primary point which should be made is that we
have always tried to handle uncertainty, but before 1975, we
really didn't have quantitative measures of uncertainty.
So, defense in depth and safety margins was a way to handle
them. After that, we've quantified uncertainty, so now we
can use some quantitative measures to evaluate how good
these defense in depth measures are.
I think we need a little paragraph here, somewhere
in Section 2.2, the defense in depth approach, explaining
that. It would fit very nicely because it says here defense
in depth has evolved since the first research reactors were
designed in the 1940's, you know, and go from there. Then
it jumps too quickly into the rationalist instruction.
Before that, we can say that it was a way of handling the
uncertainties that people knew existed in these analyses,
but they could not quantify it.
MS. DROUIN: George, have you marked up a version
with these comments that we could --
DR. APOSTOLAKIS: Yes.
MS. DROUIN: And you will give those to us?
DR. APOSTOLAKIS: I usually don't, but I will turn
to my cognizant federal official here.
MR. MARKLEY: George, you can share your comments
with the staff.
DR. APOSTOLAKIS: I am sharing my comments with
the staff now.
MR. MARKLEY: But did he share his mark-up with
the staff. Well, as much as you can share in a public
DR. APOSTOLAKIS: Can I share it with my friend,
MR. KRESS: We have done this in the past.
MR. KING: We have done it.
DR. APOSTOLAKIS: I have no problem sharing.
MS. DROUIN: That would help us.
DR. APOSTOLAKIS: You would probably need to call
me to decipher some of these things here. These are what
intended to be personal notes.
MR. KRESS: Well, you have to delete all the swear
DR. APOSTOLAKIS: No, some of it is in Greek. No,
MS. DROUIN: I have a Greek dictionary.
MR. POWERS: The ones of high praise for the
MS. DROUIN: No, I mean, I'm taking notes, and of
course, there's a transcript.
DR. APOSTOLAKIS: I understand.
MS. DROUIN: But it does help to have your
DR. APOSTOLAKIS: We can easily run a copy, in
fact, right now so you have it in front of you as we talk.
MS. DROUIN: That would be even --
MR. MARKLEY: We can do that.
DR. APOSTOLAKIS: Yeah.
MR. BONACA: Can I get --
DR. APOSTOLAKIS: Now, there is --
MR. BONACA: I would like to make a comment to
DR. APOSTOLAKIS: Yes.
MR. BONACA: I have no objections. I as a member
don't particularly care about this discussion of
structuralistic position, at least in this document. I
mean, I understand where we're going. We're discussing at
length. It sounds as if the world is divided into two type
of people that think differently. I believe there are
historical reasons why regionally people have structuralist
thought, because they didn't have risk analysis.
DR. APOSTOLAKIS: That's what I just told you.
MR. BONACA: Yeah. I'm saying that the way this
has developed --
DR. APOSTOLAKIS: It was a matter of handling
MR. BONACA: -- you know, somewhat makes you
believe that, you know, like two different religions.
DR. APOSTOLAKIS: Yeah, that's why we need this
MR. BONACA: Yeah, and I believe I'm in both
DR. APOSTOLAKIS: Yeah. The reason was, you know,
essentially handling uncertainty, and in the first 20 years
or so, we didn't have the benefit of quantification. So,
people resulted to these two principles.
MR. BONACA: I don't think that we should
encourage almost a conflict today, just recognize the
reality of where we can from. We didn't have risk analysis.
We didn't have the ability of evaluating that. So, that's
all I wanted to say.
DR. APOSTOLAKIS: Well, there's no conflict. The
point is that even after risk analysis came along, there are
still big holes. So, you have to go back and revert to the
original approach of defense and depth in margins. It's very
simple idea. We're not talking about anything which is
deeply philosophical here. I think it's important,
especially for people who haven't seen these things before.
You have one paragraph explaining.
MS. DROUIN: I agree. Let me translate what I
think I'm hearing, and if what I heard is incorrect. I
mean, what I'm hearing is that at a basic level, in the way
we have developed this framework with this structuralist and
this rationalist, that the concept is not in question.
DR. APOSTOLAKIS: No.
MS. DROUIN: It's a good explanation of how we've
DR. APOSTOLAKIS: Right.
MS. DROUIN: So, I welcome any comments that helps
clarify and explain that better.
DR. APOSTOLAKIS: Yeah.
MS. DROUIN: But I'm not hearing that there's a
basic disagreement with the concept of how the framework has
DR. APOSTOLAKIS: There will be some more comments
as we go along, but fundamentally you're right.
MS. DROUIN: Okay.
DR. APOSTOLAKIS: I mean, you're not going to hear
no, throw this out of the window, at least not from me.
Subcommittee member, of course, may have different views,
but I thought, and I believe the committee itself had seen
the overall framework months ago.
MS. DROUIN: Yes.
DR. APOSTOLAKIS: And there was praise. So, you
can certainly change your minds today, but it's not my
MR. BONACA: My only comment was, again, that on
page two. Maybe there is, would be a full recognition of
the fact that there are historical reasons why, I mean,
there were more structuralists before and there are more
DR. APOSTOLAKIS: They agreed to do this. That's
what Mary said.
MR. BONACA: That's all I wanted to point out.
DR. APOSTOLAKIS: The historical reasons is the
lack of quantification of uncertainty.
MR. KING: There are a number of places in here
where I think some additional explanation would be useful,
some better packaging.
MS. DROUIN: Yes.
MR. MARKLEY: I'd just like to say one thing. The
version that George has is the same one that was issued for
public comment. So, anyone that's referring to it from the
transcript discussion as best we can reference the sections,
it would help people better understand that because that
particular document is available in the PDR.
DR. APOSTOLAKIS: Okay, that's what I'm using.
MS. DROUIN: But the only difference between the
version that was in the SECY and the May version is chapter
DR. APOSTOLAKIS: Okay.
MS. DROUIN: Any differences in chapters one, two,
three and four would have been editorial. Maybe we caught
typos, but it didn't change.
DR. APOSTOLAKIS: If I bring up a point that you
have already responded to, just tell me and we'll go on.
The only reason I haven't read the new version is I was out
of the country the last week, so I didn't have time to do
it. So, basically that was my comment, that chapter two,
especially 2.2, where we talk about the defense in depth
approach. Actually -- yeah, it should be edited to
accommodate, you know, my comments, Mario's comments, and
create a better package. The other question --
THE COURT: Can I ask one more question?
DR. APOSTOLAKIS: Yes.
MS. DROUIN: Because I got a sense at our last
meeting, and maybe that was incorrect, that even the
organizations from chapters two and chapter three to chapter
four could really be completely repackaged. It wasn't just
adding paragraphs here and there. I got the sense that
certain things and certain chapters, whole parts of it
should come out and be moved elsewhere. I just wanted to
revisit if that was --
DR. APOSTOLAKIS: Ask me which ones. I wouldn't
be able to tell you right now, but I didn't find that
problem when I read it.
MS. DROUIN: Okay.
MR. KING: Well, you mentioned we overemphasized
that, you know, we're not to rely too much on the numbers.
We kept saying it over and over again.
DR. APOSTOLAKIS: Yeah, there were editorial
MR. KING: That kind of thing.
DR. APOSTOLAKIS: Yeah. I mean, starting right
away in Section 2.1, however, quantitative objectives would
not generally appear in specific regulations, and then
that's repeated every chance you get. So, immediately I
thought of a common cause there. I mean, there must be a
reason. Somebody was scare that numbers might appear
someplace and commit a cardinal sin.
MR. KING: Like NEI.
DR. APOSTOLAKIS: Huh?
MR. KING: Like NEI.
MS. DROUIN: Well, NEI had the exact opposite
interpretation. They read it saying numbers were going to
appear in the regulation, so it may be in the eye of the
DR. APOSTOLAKIS: Before we leave the issue of
defense in depth, I realize some of these things come from
our paper, and I suspect that paper is not perfect. The
structuralist says here, or traditionalist model asserts
that defense in depth is embodied in the structure of the
regulations and in the design of the facilities built to
comply with those regulations.
I've heard many people tell me that this is a
lousy definition. If you want to define the structuralist
approach, you are not doing it.
MR. KRESS: That's a description of it, not a
DR. APOSTOLAKIS: It's a description, really of
why it was. So, I think in the context of the historical
context of lack of quantification and handling uncertainty
by adding successive compensatory measures, I think that
would go a long way towards explaining the structuralist
approach. And then it was embedded into the regulation. We
did that everywhere. Okay? You see that paragraph?
MR. KING: Yeah.
DR. APOSTOLAKIS: The other thing that's missing
here, and I think that's a more important point. Do you
want to say anything about safety margins, because defense
in depth and safety margins were really the cornerstones
handling uncertainty the traditional system, were they not?
Not just defense in depth. So, the question might come up,
why is this new approach an overemphasizing defense in depth
and not safety margins?
MR. KING: To me, safety margins are an element of
defense in depth, and maybe we'll talk about that.
DR. APOSTOLAKIS: You can view them as such.
There's no question.
MR. KING: At the high level, it's defense in
depth. Safety margins are maybe a tactic. We call them
tactics, to implement defense in depth and to account for
uncertainties in some things. I mean, to me, safety margins
really are more applicable when you're talking about I want
to do a deterministic calculation and meet some acceptance
criteria, but I've got some uncertainty in how good my code
is, and I've got some uncertainty in how good my acceptance
criteria are, so I throw some margins in there. I back off
on the acceptance criteria. I do a conservative
calculation. I do 95 percent confidence. When you're
getting into things like reliability and availability, the
term safety margins to me doesn't fit that well.
DR. APOSTOLAKIS: No, it doesn't.
MR. KING: And then you're getting into
redundancy, diversity, that kind of stuff.
DR. APOSTOLAKIS: Well, the whole point is that in
regulatory guide 1174, as you know, safety margins have
their own box.
MR. KING: Right.
DR. APOSTOLAKIS: Alongside with defense in depth.
So now, this new document takes margins and puts them under
defense in depth, which is fine with me, but maybe we should
say that. I don't know, maybe we shouldn't.
MR. KRESS: The safety margins I'm familiar with
mostly apply to success criteria, which deal with, in an
uncertainty sense, very well in PRA's.
MR. KING: Right.
MR. KRESS: So, you know, that's why we have
safety margins, because you -- but maybe you could tie it in
that sense together.
MR. KING: But George is right in that in 1174,
they're sort of two equal separate things.
MR. KRESS: Yeah.
MR. KING: But in the framework, they're not
MR. KRESS: I don't think they should have been
equal in the 1.174. I like them better being part of
defense in depth, actually.
DR. APOSTOLAKIS: I have no problem with that.
MR. KING: I mean, to some extent, the framework
represents an evolution of thinking since 1.174 was put
DR. APOSTOLAKIS: And I think it would be
perfectly all right to say that because, you know, not
everyone is up to speed, you know, with these documents as
you guys are. Just explaining, I mean, when you say 2.2
defense in depth, maybe somewhere there you say that by the
way, defense -- I mean, safety margins are considered an
element of defense in depth.
MR. KING: Yeah.
DR. APOSTOLAKIS: So the last comment on 2.2 is
this, you know, bring back the rationalist at lower levels
with the uncertainties, which is as I say, the preliminary
framework we purport. In section two on -- I just saw my
notes -- there is a problem that I think we have to resolve
once and for all, though.
The first paragraph on the right-hand column --oh,
boy. This is not the new version, right?
MR. KRESS: I've got the old version. What page
are you on?
DR. APOSTOLAKIS: The NRC has established safety
MS. DROUIN: What page are you on?
MR. KING: Yeah, what page are you on?
DR. APOSTOLAKIS: Section 2.1.
MR. KING: Okay.
DR. APOSTOLAKIS: Yeah, the revision dated May,
2000. Actually, the paragraphs starts on the left-hand
MR. KING: We found it, yeah.
DR. APOSTOLAKIS: Although licensees are not
required to demonstrate that they meet the quantitative
goals, comparisons of PRA to IPE results to the goals are
common. Okay, my problem is that okay, then it goes on and
says it is proposed in this document that quantitative
objectives be used to provide guidelines for risk informing
The first paragraph of this section talks about
the Commission referring to this standard as either the
adequate protection or the no undue risk standard. What are
we talking about, goals or undue risk, and what should the
numbers be, the ones that are being used in the framework?
Should they be goals or numbers representing in some sense
MR. KING: They're not numbers representing
adequate protection. They're guidelines for the staff to
use when they take a look at the existing regulations and
have to decide are there holes in the regulations? Are
there things in the regulations that are overkill and really
aren't contributing much to safety? There are sort of
guidelines for what's the baseline of safety that we're
shooting for where we can quantify that. The fact that
they're based upon the Commission's safety goals means that
they're really written at the level of how safe is safe
They're not intended to define adequate
protection. They're not something that we expect licensees
to come in and have to demonstrate compliance with either.
That was an issue NEI raised. They looked at the framework
and said my God, am I going to have to as a licensee do some
analysis to show that I meet this framework? No, the
framework is for the staff.
MR. KRESS: I think it's entirely appropriate that
you write your regulations in a sense that you're trying to
meet the safety goals and stay away from adequate
MR. KING: Yeah, you don't have to deal with the
adequate protection question.
MR. KRESS: You don't have to deal with it that
MR. KING: And besides --
MR. KRESS: I don't think we have to resolve
anything. I think that's what safety goals were intended to
MS. DROUIN: Well, the other thing, by having a
guideline, you certainly don't want to write a requirement
that's going to be more stringent than the safety goals.
MR. KRESS: Yeah, you certainly don't want to do
MS. DROUIN: And so, by having the quantitative
guidelines, that kind of gives us the measure so that we
know that we aren't going past the safety goals.
MR. KRESS: The problem I may have with it is that
the current regulations have objectives that are more than
just safety goals. They're objectives of limiting fission
product releases, and most of your quantitative guidance has
to do with core damage frequency and conditional containment
failure probability, and you don't deal with fission
products at all. I have a little bit of a problem handling
that because I don't think you can get away from fission
MR. KING: And there are things --
MS. DROUIN: I don't think we did.
MR. KRESS: Well, I didn't see anything that
would, say, deal with something like 10 C.F.R. 100 because
it's not a risk dominant concept. It's just a way to limit
the releases of certain levels of fission products to
certain frequencies, in my mind, that's what it is. But
those aren't risk dominant, and they don't involve
containment failure. They do involve some level of core
MR. KING: It's the same issue as anticipated
operational occurrences. They're not risk dominant, but we
don't want to throw away the concept. We still want plants
to be able to ride through those without any damage. So,
there are things like that that we have to deal with.
MR. KRESS: And I guess you still deal with those
by maintaining the design basis accident concept somehow.
MR. KING: We can deal with those at the
structuralist level. See, there are some things, and we'll
get to a viewgraph on that. There are some things --
MR. BONACA: I mean, although you're not looking
really at those goals numerically as criteria that you have
to meet, you're still using this as a way of looking at
possible gaps in the fabric of regulation.
MR. KING: Right.
MR. BONACA: Okay.
MR. KING: And particularly severe accident gaps.
MR. BONACA: Okay, so you're looking at it, yeah.
DR. APOSTOLAKIS: In the past, thought, when the
staff wrote regulations, you had in mind adequate
protection, didn't you?
MR. KING: No, number regulations were based upon
safety enhancement, station black-out. You had to do cost
DR. APOSTOLAKIS: Yeah.
MR. KING: That's not adequate protection.
DR. APOSTOLAKIS: Right, but you had to do a cost
MR. KING: Yeah.
DR. APOSTOLAKIS: And now we're talking about the
framework, the regulatory framework which will actually not
be based on adequate protection, and there will be no cost
MR. KRESS: Voluntary.
MR. KING: Well --
DR. APOSTOLAKIS: It is voluntary.
DR. SHACK: Well, there is a question of how you
apply cost benefits. Is it regulation by regulation, which
is the way we do it now, or is there some overall kind of
cost benefit? That seems to be the --
MR. KING: The whole issue f backfit cost benefit
is a policy issue the Commission's going to have to deal
with, and it could range anywhere from maybe since this
whole thing is voluntary and we're just providing an
alternative to an existing regulation, maybe we don't need
to do any backfit analysis. Or, maybe we do need to do some
backfit analysis, and if we do, is it piece by piece within
a regulation, or do you bundle all the changes within a
regulation together and do it?
MR. BONACA: Again, it's interesting, but still,
for example, say that you have a generic safety issue that
you raise. Okay, you could bring it through this process,
and then through that determine that we have adopted the
regulation for the specific issue. Then you can use a
criteria like ten to the minus six cut off point to
determine whether or not you should do something about that.
MR. KING: Yes.
MR. BONACA: So, you're still not looking at the
absolute, you know, CBF nerve, but you're possibly using
criteria of that type to determine whether or not you should
go ahead with that enhancement.
DR. APOSTOLAKIS: I think I'm a little
uncomfortable with this, and I wonder if there can be -- I
mean, is the industry aware of what you're doing, of the
distinction between adequate protection and goals and they
have agreed that what you're doing is okay? I've heard them
complain that this is becoming too risk based, but that's
not the same complaint. That's their usual whining. The
question is do they fully understand that you are using
numbers that were meant to be indicators of safe enough
rather than adequate protection?
MR. KING: We met with the industry, discussed
this issue specifically with them, among other issues, and
they did not object. They seemed to understand.
MR. KRESS: As long as you're not trying to force
each individual plant to meet those numbers, I don't think
they'd have any problem with it.
MR. KING: Again, we're making generic changes to
the generic set of regulations. We're not asking each
licensee to come in and demonstrate that he meets these
MR. KRESS: They shouldn't have any problem with
it because I suspect they feel that the bulk of the plants
already meet them anyway.
When you go to the Commission with these policy
issues like funneling cost benefit and selective
implementation, are you going to make your recommendations
to them as to which way you'd like for them to go, or are
you just going to leave it to them and say tell us what you
MR. KING: I think we ought to make a
recommendation. That would be my view.
MR. KRESS: Do you want some input from us on
those particular kind of issues then, or is that something
MR. KING: These are going to be issues we're
going to have to address in our August paper.
MR. KRESS: August paper.
MR. KING: Yeah, I would like the Committee's
views, and we have a timing -- since you don't have an
August meeting, there's a timing issue about how we get
that, but yes, I would like to get those views.
DR. APOSTOLAKIS: I think the Commission also
should be presented with a clear statement that this is
different from adequate protection. This is what we're
basing it on for a certain reason. The center for strategic
-- CSIS, asked that the Commission define adequate
protection. My concerns is that if we start doing things
like this, pretty soon before you know it, simply because
these are the only numbers that are available, they will
become measures of adequate protection.
MR. KRESS: And I would wave the flag and say good
for them. They won't because nobody's going to define
adequate protection, and nobody's going to put numbers into
the regulations that each individual plant has to meet, not
while I'm still on the Committee they probably won't.
MR. BONACA: These plants were not really licensed
on the basis of this goal. They were licensed on the basis
of meeting the regulation, whatever regulation was in place
at the time. I don't think that there is an ambiguity
MR. KRESS: There's not much of an ambiguity
because you can write the regulations, but they have to be
read in such a way that they can be met in a variety of
ways, so that some plants will exceed the goals and some
DR. APOSTOLAKIS: Let me put it in a different
way. If Option 3 is based on these numbers, and by the way,
there was a lot of discussion later of some other subsidiary
numbers that are even more stringent for certain reasons.
Are we de facto ratcheting up the regulation?
MR. KRESS: Well, we've always had this in the
regulations. It's in the cost benefit rule and the
regulatory analysis rule.
MR. KING: The reg analysis guidelines are in
effect based upon the safety goal numbers.
MR. KRESS: And you have a safety goal policy
statement that says policy is to write the regulations so
that they try on average to achieve those kind of things, so
it's been there. Actually, they show up in 1.174. It's the
same numbers there, basically.
DR. APOSTOLAKIS: No one has the same concerns.
It's not an objective. It's just a --
MR. KRESS: I've always been concerned that
adequate protection has never had a quantified value to it
because I think it's a little bit strange that we have such
a concept and don't have a quantified number associated with
it, but I think that's a different problem than this one. I
think it's a different issue.
DR. APOSTOLAKIS: So the least we can do then is
in Section 2.1, maybe at the paragraph explain the number we
are about to use are not protection numbers.
MR. KING: I think the document ought to be clear
as to what these numbers represent in the relation to
DR. APOSTOLAKIS: Before we go on, I've been
advised that it would be best at the end of this meeting you
return the copy that we just gave you. Feel free to copy
anything you want right now.
MS. DROUIN: We can't read it anyway.
DR. APOSTOLAKIS: I suspected that.
MS. DROUIN: I mean, the Xerox wasn't good enough.
DR. APOSTOLAKIS: But you can ask me to explain.
That's during the meeting. Mario? I thought you had a
So, I think we beat this to death. Yes?
MR. POWERS: A question. When you focus your
examination of the regulations on risk analyses, you usually
have available to you things like core damage frequency and
maybe even things like large early release frequency. I'm
wondering what about late small release frequencies, which
also has to be a part of public protection. How do you get
measures on that to look at?
MR. KRESS: That was one of the things I had in
mind when I said they didn't focus on some of the
objectives, and that was one of them. Late small releases
would be something I would worry about.
MR. POWERS: I just made that up because it's --
DR. APOSTOLAKIS: Late small?
MR. KRESS: Yeah.
MR. POWERS: It's the opposite of large early
MR. KRESS: Yeah, and normally you can get it out
of a PRA or some measure of it, but not very good. The
regulations deal with those things, some of the ones that we
have now. So, yeah, that was one of my concerns. How do
you deal with those kinds of things?
DR. APOSTOLAKIS: I think they have something on
that in the document somewhere. It specifically deals with
MR. POWERS: Containment failures.
DR. APOSTOLAKIS: A quantitative objective of less
than .1 containment failure probabilities proposed, yeah,
for a late large release in a core damage accident.
MR. KRESS: Yeah, but that's a late large. We're
talking about a late small.
DR. APOSTOLAKIS: Late small?
MR. KRESS: And I would worry about early smalls
also due to some of the bypass accidents.
MR. POWERS: I guess what I'm asking about is here
are some numbers that I don't see usually coming out of the
PRA's. I don't know that it's not possible to get them, but
for instance, PRA's usually, in the past at least, ones I'm
familiar with usually said all accidents are over at 24
hours, whether they are or not.
When I look at the regulations and I see things
that seem to suggest that you've got to maintain control
room capability with outside intervention for seven days and
30 days and things like that, these are numbers that are not
addressed by the PRA, and I'm wondering how you handle that
kind of issue.
MS. DROUIN: When you go back and you look at the
IPE's, there were certainly a select group. I don't recall
how big the group was that terminated the accident at 24
hours, regardless of where they were. You did have analyses
that went through the whole progression. We certainly have
numbers from the IPE's on late releases.
What you don't see out of the IPE's in terms of
numbers are numbers associated with small releases. We
don't have those kinds of quantitative numbers from the
PRA's. That's not to say we don't have them. We just
don't have them. Data that we have all tend to center
around your large releases both early and late. Then some
of the late stuff, you know, is not consistent because of
how they define late and how far they took the analysis.
So, on the late you kind of have a mixture.
MR. KING: Where we don't have risk information,
seven days or something like that, we're going to have to
make judgment using something else, or we leave that piece
of the regulation alone.
MR. KRESS: A conditional containment failure
probability could address if it's the full conditional over
a whole time period. Do you limit the conditional
containment failure probability to conditional early
containment failure, or is it the whole conditional?
DR. APOSTOLAKIS: I think that's what I read,
recover late, but also large.
MR. KING: It covers late, but what does late
mean? Is late 24 hours? Is late -- I don't think late is
DR. APOSTOLAKIS: I think it's late, the opposite
of early. Early is I think, the way I understand it, about
three hours, is it not?
MR. KING: The neighborhood of four to six hours
kind of thing.
MS. DROUIN: Yes.
DR. APOSTOLAKIS: Then late must be after that.
MR. KING: But there's a cut-off. I mean, you
don't carry the analysis for 30 days.
DR. APOSTOLAKIS: No, no.
MR. HASKIN: By the same token, small is a
complement of large. I mean, if you have the large and you
have your PRA, you could extract the small release, but we
haven't established it.
DR. APOSTOLAKIS: I'm going to be confused now by
my colleagues because the two colleagues that have been
speaking for the last ten minutes were the ones that
objected to using frequent -- well, not so much the one to
my right -- frequency consequence curves and the CDF and
LERF point values were good enough.
MR. KRESS: I like frequency consequence curves.
I just think --
THE COURT: But what happened? I mean --
MR. KRESS: Well, CDF and LERF happen to be a very
convenient, in my mind, is a convenient representation of an
FC curve, if you draw lines correctly between them.
DR. APOSTOLAKIS: Well, and the argument that --
MR. KRESS: And if you use the whole LC part of
that and not just the end parts.
DR. APOSTOLAKIS: But in terms of early
fatalities, I thought that was the argument that, you know,
which is what really counts. CDF and LERF point values were
MR. KRESS: Yeah, and that was based on the
concept that early fatalities generally dominated.
DR. APOSTOLAKIS: Right.
MR. KRESS: That if you covered those, you've
covered the goals you would have for late, and there are no
goals for that.
DR. APOSTOLAKIS: So, what are we talking about
MR. KRESS: Well, basically the same issue there,
and that's what -- in my mind, when I talk about goals that
involve small releases or late releases or late small
releases, I'm talking about frequency consequence curves
because that --
DR. APOSTOLAKIS: Well, Dave probably doesn't.
MR. KRESS: Well, that's what I had in mind when I
talk about them. You have a frequency limit on fission
product releases of any magnitude, and it deals with the
whole range, the small, early. So, I'm certainly not
opposed to FC curves. I think they're a good concept, and
they can come out of the PRA if you do it correctly.
THE COURT: Well, and the other question is
whether you need 84, as Dana was arguing, or you can pick
one representative, like cesium.
MR. KRESS: Oh, yeah, I think that's an issue, but
if I were going to pick one representative, I'd probably
DR. APOSTOLAKIS: You would or wouldn't?
MR. KRESS: I would, probably, yeah, because in my
mind, I can generally correlate the iodine release with
MR. POWERS: You can if you persist in believing
the releases are driven by the high temperature component.
You can't if they're driven by the aquus chemistry
MR. KRESS: Yeah, then you have a problem.
MR. POWERS: In late, you're very likely to lose
MR. KRESS: Yeah, you'd have to do something else
with the late part, and if you had an error ingression
problem, cesium may not be the -- you'd have to have a
different cesium curve. You could still use cesium,
MR. POWERS: And of course, you have at least one
member of the committee that believes you always have an
error ingression problem.
MR. KRESS: Yeah, or maybe two. Maybe two members
in this Committee. Huh?
DR. APOSTOLAKIS: What advise are we giving this
MR. KRESS: Well, I don't know. I think the
advice is that we think CDF, CCFP or LERF --
DR. APOSTOLAKIS: They like CCFP.
MR. KRESS: Yeah, are fairly good things to use
because they're practical and they do incorporate things,
but we believe they're limiting. There are other objectives
in the regulations that they want us to think with and deal
with, and we just don't want them to forget those when they
DR. APOSTOLAKIS: I'm under the impression the
Commission decided recently on the safety goal issue that
plant contamination ought to be part of the revision?
MR. KING: That's right.
DR. APOSTOLAKIS: So, if that's the case, then why
are we discussing this? Did they say anything about late
releases and consequences? My impression was, as I recall,
that they like things as they are now.
MR. KING: It's basically approved updating the
policy to be current with the practices that are in place
DR. APOSTOLAKIS: So, there isn't an issue
anymore. The policy issue has been resolve. I mean, we
could choose to write a letter to the Commission, but I
don't think they're staffing writing Option 3.
MR. KRESS: The staff is concerned in building a
policy, I think.
DR. APOSTOLAKIS: But this is not the right forum.
Before we leave 2.2, which is really an important section in
my view, from page 2.3, there is a discussion of the
paragraph that starts in applying the strategies, good
engineering practices will be maintained. The general
design criteria of providing many concise statements of good
engineering practice. For example, negative prompt
feedback, emergency AC power. I never considered those as
good engineering practice. This is not what I understand by
good engineering practice. I mean, these are deliberate,
rationalist measures that have been imposed on nuclear
engineering. In the west and in the United States, you
would not design a reactor with a positive feedback option.
Not you will have emergency power. I mean, to call these
good engineering practices seems to me you're stretching the
meaning of the term.
MR. KRESS: It's not my usual determination of
that term either.
DR. APOSTOLAKIS: Yeah. Then I read on in this
paragraph, requirements that the fuel design limits not be
exceeded in anticipated operation occurrences, and that the
extent of fuel damage be limited in design basis, accidents
will be maintained. Now, why do you want to make this, to
commit yourself to the fact that design basis accidents will
be maintained? What if later on we become wiser and we come
up with something better?
MR. KING: No, this is not intended to say we're
going to stick with today's definition of design basis
accidents. I mean, I think clearly one of the things we may
do is come up with a new set of design basis accidents that
reflect risk consideration. What we were trying to say here
is we're not going to go in and make the acceptance
criteria, core melt, for example.
DR. APOSTOLAKIS: I think you've made it very
clear earlier that you're not going to put the numbers up
there. In my view, this paragraph more hurts than helps.
Would you lose anything if you just deleted it?
MR. KING: In reading it now, I think clearly it
needs to be modified, but I would like to come back to this
issue of good engineering practices because I don't agree
with everything you've said.
DR. APOSTOLAKIS: Okay. Well, good engineering
practice, we've heard those words a lot from South Texas in
the context of quality assurance.
MR. KING: To me, there's an element of good
engineering practices that fits under the structuralist
label, and I would put prompt negative feedback coefficient
as one of those. I mean, that's not something you're going
to decide based upon risk information.
DR. APOSTOLAKIS: No, no.
MR. KING: I mean, that's something you want in
the plant. I don't care what the PRA says. That's what you
DR. APOSTOLAKIS: I wouldn't call it a good
engineering practice either.
MR. KING: What would you call it?
DR. APOSTOLAKIS: I don't know.
MR. KRESS: A general design criteria.
DR. APOSTOLAKIS: Yeah, I mean, it's a design.
Good engineering practice usually means, you know, I will
pick a better quality pencil, given the option, and you
know, that kind of low level stuff.
MR. KING: I think use of codes and standards to
me would be a good engineering practice. You're not going
to decide that based upon your PRA. You're going to say
hey, I'm going to design stuff to the ASME code or whatever.
MR. BONACA: But I think it's more than good
engineering practice in a certain way. I mean, for example,
look at the word containment. I mean, containment. Maybe
it's a good example to talk about. Containment, would we
eliminate containment if we could demonstrate that you don't
DR. APOSTOLAKIS: If you convince me that you
don't need it, yes.
MR. KRESS: Yes, thank you.
DR. APOSTOLAKIS: I want to be on record that I
MR. KRESS: Thank you, George, thank you.
DR. APOSTOLAKIS: This seems to be a sacred cow,
like this -- we don't want it to be risk based. It has to
be risk informed. I mean, come on. Yes, if you can
convince me, why should I have it.
MR. BONACA: Remember that the Titanic didn't have
enough life boats because it was unsinkable.
MS. DROUIN: Maybe one way to look at these things
is that they're the outcomes of your good engineering
practice. If you want to go in and say maybe your good
engineering practice is to have diversity redundancy, and
how do you go about achieving that. Well, you have a
containment, you have multiple barriers.
DR. APOSTOLAKIS: I know, but even you in here,
you know, you say the general design criteria provide many
concise statements of good engineering practice. For
example, negative prompt feedback, emergency core cooling
containment, are all called for in the GDC's and deemed
essential to the defense and depth approach. So, you are
going now from good engineering practice to defense in
depth. I don't know, I think most people don't pay
attention to these things, but I don't --
MR. KRESS: I would be tempted, George, just to
call that a good defense in depth approach.
DR. APOSTOLAKIS: Right, that's what it is.
MR. KING: I'd say these defense and depth
measures will be maintained no matter what.
DR. APOSTOLAKIS: But I wouldn't call them good
MR. POWERS: I agree with that.
DR. APOSTOLAKIS: It's the structuralist approach.
MR. KRESS: They're structural defense in depth.
DR. APOSTOLAKIS: It the structuralist defense and
MR. BONACA: And I believe the uncertainty issues
to the issue.
DR. APOSTOLAKIS: But the negative prompt feedback
is not defense in depth, is it?
MR. KRESS: Not at --
DR. APOSTOLAKIS: I mean, it's a matter of
controlling the thing.
MR. KRESS: Not traditionally, but --
MR. BONACA: No, but there's a preventative
defence in depth concept.
MR. KING: I tell you, that's an important
fundamental aspect you want your design to have.
DR. APOSTOLAKIS: But you want to control it.
MR. KING: It's because you don't want to rely on
a control system to do it.
DR. APOSTOLAKIS: Yeah, inherently.
MR. KING: So, I mean, it's an inherent defense --
DR. APOSTOLAKIS: That's not what I'm saying, that
I believe that if you delete this paragraph and maybe other
few lines that make clear the stuff we just discussed, like
yes, there are essential elements of defense and depth that
will be maintained. Then I don't know what you want to say
about the GDC's, but --
MR. KING: I agree. This paragraph needs some
work, but I think basically what we're trying to say is
reflected on our slide four in that there are some elements
of defense and depth that are going to be implemented
regardless of what the risk analysis says, and there's going
to be others that are going to be dependent upon the risk
DR. APOSTOLAKIS: You don't have to put it that
way. You don't have to put it that way. It will be
MR. KING: All right, and I think we ought to try
and list those, or list examples of those.
DR. APOSTOLAKIS: And then there is this
discussion here that goes into detail that the reader, you
know, for example, it's been suggested that the number of
regulatory requirements have led to the fuel design limits
during normal operation could be eliminated, because the
intent of GDC 10 is being met for commercial reasons so the
requirements are not risk significant. I mean, I probably
have to read the whole report to understand what this means.
MR. KING: Well, it means we found some things
that don't have any risk implications and you really don't
need them as requirements to keep the fuel intact. So,
those are candidates for removal.
DR. APOSTOLAKIS: Okay. If you put it that way,
it's a general statement that I understand, but referring to
GDC 10 and so on.
MR. KING: Like I said --
DR. APOSTOLAKIS: There are too many thoughts in
this paragraph. I mean, it's really loaded with thoughts
that somebody obviously is a veteran of regulatory wars.
Now -- oh, you're now in your tactics.
MR. KING: Well, we put that up because you
brought up the issue of negative power coefficient.
MS. DROUIN: Our viewgraphs, George, do not
necessarily follow the outline of the report.
DR. APOSTOLAKIS: Okay.
MS. DROUIN: I like the way we're going through
DR. APOSTOLAKIS: Okay, great.
MS. DROUIN: I mean, to me, this is very helpful.
DR. APOSTOLAKIS: On 2.5, off the old and the new
one -- maybe that's what I can do, huh?
MS. DROUIN: So you like 2.4?
MR. KRESS: You can't assume that.
DR. APOSTOLAKIS: Silence means -- I think the
strategy was all right.
MS. DROUIN: There's not a 2.5.
DR. APOSTOLAKIS: Page 2.5.
MS. DROUIN: Oh, page 2.5.
DR. APOSTOLAKIS: It's still section 2.3. The
four high level defense in depth strategies on the left,
paragraph on the left.
MS. DROUIN: Yes.
DR. APOSTOLAKIS: The four high level defense in
depth -- you don't have that figured. Do you have a
transparency of the figure?
MR. KRESS: It's figure 2.1.
MS. DROUIN: Yes, we do.
DR. APOSTOLAKIS: Yeah, but I mean to show it up
MR. KING: It's not in your hand-out.
DR. APOSTOLAKIS: Right, right, right. So the
four high level defense in depth strategies are
intentionally more focused than the reactor safety
cornerstones. Now, the reactor safety cornerstones are
initiating events, right?
MR. KING: Yeah, the four that are up there.
DR. APOSTOLAKIS: Why are these more focused?
MR. HASKIN: Basically, we're not dealing with
fuel handling, fuel storage, rad waste storage tank rupture,
those types of requirements in this pass-through. We're
concentrating on the core damage frequency. You don't get
core damage from rad waste tank rupture, for example.
DR. APOSTOLAKIS: So these, don't worry about
these things. You don't have those here, do you? These are
MR. HASKIN: As we proceed to the quantitative
goals, yes. I mean, it doesn't show on this viewgraph. If
you read the cornerstones, they're concerned with accident
prevention no matter what the accident is, whether it can
lead to core damage or not.
DR. APOSTOLAKIS: But isn't the original statement
from the Commission on QHO's referring to operations at the
nuclear power plant? They don't really distinguish between
reactor accidents and other kinds of releases.
MR. HASKIN: Right. I think this is an insight
that we obtained from risk assessment, is those types of
accidents really aren't the risk dominant ones. Hence, our
focus on core damage.
DR. APOSTOLAKIS: But you could have public --
MR. HASKIN: Conceivably, there could be
regulations dealing with those types of accidents that could
impose excess burden. If they did, we'd take a look at
them. You know, our framework strategies and quantitative
objectives, we didn't feel that that was the appropriate
place to focus.
DR. APOSTOLAKIS: Okay. The paragraph in
describing the cornerstones, the strategies to break this
down? I suggest you delete everything up until containment
failure for all conceivable accidents. Then you say, the
use of all four strategies constitutes a high level defense
in depth approach, which compensates for the limitations of
the individual strategies, and assures that the risks to
public health and safety will be low and consistent with the
top level objectives. Everything else is not needed.
I'm a little bit uncomfortable the way this is
going, though. I mean, do the other members of the
subcommittee feel this is a good Subcommittee meeting?
MR. POWERS: I think you're belaboring individual
DR. APOSTOLAKIS: Well, we are hitting big and
small here, but we have had no other opportunity to discuss
this report, which is one of the most important reports the
staff is going to release, and I'm not allowed to give them
my comments in private. So, I don't know how to do it.
MR. MARKLEY: George, what we could do is if you
have some higher level comments, we could provide them to
them in like a memo or something of that nature, which we
could also, you know, make available to the public. I think
that would be the better way of doing it, the editorial type
things, unless they have a real major impact.
DR. APOSTOLAKIS: The editorial I agree.
MR. MARKLEY: yeah.
DR. APOSTOLAKIS: At the high level, no. We
should discuss them now.
MR. MARKLEY: Yes.
DR. APOSTOLAKIS: So, there is a table 2.1 on page
2-6. Tactics and examples of related regulatory documents.
You say that the safety goals under PRA's IPE's is a tactic?
I thought that was the top objectives that was driving
everything. See safety goals under PRA's, IPE's?
MR. KING: Yeah, I see it.
MR. KRESS: I think it's out of place in this
DR. APOSTOLAKIS: It's completely out of place,
MR. KING: Yeah, I tend to agree. I'm not sure
why that's there.
DR. APOSTOLAKIS: And then next to special
treatment, there's a parenthesis, non-scope. I don't
understand what that means, non-scope? You see special
treatment in the same column?
MR. KING: Uh-huh.
MS. DROUIN: I can't answer why the word non-scope
is there. It doesn't make sense to me.
DR. APOSTOLAKIS: Fine. You will look at it later
and take action.
Now there is something that's a little bit more of
substance. At the beginning of 3.0, quantitative objectives
for the framework, the second sentence of the first
paragraph. The intent is to develop requirements which
retain deterministic characteristics in such a way that
compliance will provide reasonable assurance. Now, why do
you want to insist that they will be deterministic? I mean,
in the maintenance rule, not all requirements are
deterministic, and it's already out in the street. The
revised oversight process certainly uses frequencies for
initiating events, unavailabilities for protective systems
and other things. Not all requirements will necessarily be
MS. DROUIN: I think all we're trying to say
there, and it was probably a misuse of the word, is that a
defense in depth characteristics.
DR. APOSTOLAKIS: Now, this comes back to my
earlier comment that you're protesting too much various
places. You know, we will never use probability numbers as
part of the regulation. I saw this as being consistent with
that approach, which there is no reason. I mean, some
numbers come naturally in some places, like the frequency of
initiating events, or unavailabilities, the maintenance
MR. KING: Well, we sort of went into this with
the ground rule that we wouldn't put the risk numbers in the
DR. APOSTOLAKIS: I think you mean higher level.
You mean core damage frequencies.
MR. KING: Or damage frequencies.
DR. APOSTOLAKIS: Yeah, and I agree with that
because the whole framework, as you say correctly here, will
make sure that if you meet those lower level requirements,
then you meet the top. I don't think you should be so
absolute as to say that even at the lower levels, the
probability risk numbers are out of the question. There's
no reason for that. You already have regulations that use
them. Maybe there should be a distinction of some sort.
MR. KING: I don't think we have regulations that
use them. I think we have reg guides and SRP's that use
them, and I think in our -- in this effort --
DR. APOSTOLAKIS: Well, maintenance rule is a
regulation, is it not?
MR. KING: Well, reg guide is a one way, one
acceptable way to meet the regulation.
DR. APOSTOLAKIS: No, but the maintenance rules.
MR. KING: Oh, is a regulation.
DR. APOSTOLAKIS: Regulation.
MR. KING: Yes, that's a regulation.
DR. APOSTOLAKIS: It doesn't have numbers itself,
but it asks the licensee to specify such a number.
MR. KING: Yeah, and that's reasonable.
DR. APOSTOLAKIS: And that's reasonable.
MR. KING: And in this effort, it might be
reasonable to have a reg guide or something that has numbers
DR. APOSTOLAKIS: That's right, but also in this
effort, you should say that, you know, you will try to have
deterministic requirements as much as you can, but in some
instances, at low levels, you know, some unavailabilities or
frequencies would be natural to use, e.g., see maintenance
rule. So, people would not be shocked, that it's something
that they have already done.
MS. DROUIN: No, it is very conceivable that in
the implementing documents of a regulation, you could
certainly see numbers, but we're just talking about the
actual requirement in the regulation.
DR. APOSTOLAKIS: But I mean, the way I see it,
Mary, is that you are taking an extreme position. You are
saying we will not produce anything that will be
probablistic. They will all be deterministic
characteristics, and that's not what you intend.
MS. DROUIN: No, that's not what we intend.
That's not what we're saying.
DR. APOSTOLAKIS: Yeah. Now, in the next column,
you go into the details of how the average individual risk
is determined by dividing the number of fatalities to one
mile, weighted by the frequency of accident and so on and so
on. I didn't think that that was part of the regulation.
This is just what happened.
MR. KRESS: That's a part of the definition in the
safety goal policy.
DR. APOSTOLAKIS: Is it in the safety goal policy
statement, that you should do it that way? I thought you
were postulate an individual being there 24 hours.
MS. DROUIN: I'm sorry, George. What was your
comment? I missed it.
DR. APOSTOLAKIS: The bullet on the right. It
starts the individual risk.
MS. DROUIN: Correct.
DR. APOSTOLAKIS: Towards the end, it describes
how the average individual risk is determined.
MR. KRESS: My understand is this is exactly how
you do it.
MS. DROUIN: This is --
DR. APOSTOLAKIS: No, this is how you do it, but
is it how it is stated in the QHO statement?
MR. KRESS: Yeah.
DR. APOSTOLAKIS: I'm not sure.
MR. KRESS: Somewhere it's stated that way. I'm
not sure it's in the QHO.
DR. APOSTOLAKIS: I'm not sure it's in the QHO's.
MR. KRESS: But that's been my complaint for a
long time. What this does is treat all sides the same.
That means an accident that kills 100,000 on one site is
treated the same as an accident that kills 10,000 people on
DR. APOSTOLAKIS: Because you are exacting the
MR. KRESS: I'm saying they don't have a real
DR. APOSTOLAKIS: They don't have societal, but
I'm objecting to even describing how the average individual
risk is calculated. It seems to me --
MR. KRESS: Well, I think this is my understanding
of how it's done.
DR. APOSTOLAKIS: Does anyone know?
MR. KRESS: I'm sure this is how it's done.
DR. APOSTOLAKIS: Yeah.
MR. KRESS: I'm not sure where it tells them it's
DR. APOSTOLAKIS: That's where my question is.
MR. KRESS: And I thought it was in the safety
goal policy statement.
MR. HASKIN: It's a direct quote, but I'm not sure
DR. APOSTOLAKIS: I'm sorry?
MR. HASKIN: It's a direct quote.
DR. APOSTOLAKIS: From the goal?
MR. KRESS: I've read it somewhere.
MR. HASKIN: Because I tried to write it a
DR. APOSTOLAKIS: From the safety goals --
MR. KRESS: I've read it somewhere, and I think
it's in the safety goal policy statement.
MR. HASKIN: I don't confirm that.
DR. APOSTOLAKIS: That statement needs to be
defined. Now again, here is a defensive thing. The
right-hand column near the bottom, the QHO's and related
subsidiary objective set forth in the section applied to
mean risk measures. Unfortunately, the QHO's are difficult
to apply. Simply replacing existing regulations with the
QHO's would be an entirely risk based approach, heaven
forbid, which you would not assure defense in depth. I
mean, gee whiz, we are grown people, guys. Come on. I
mean, it's again, this fear of risk based and trying to
soothe somebody because they're bound to explode.
I don't think you need statements like that. Take
them out. This is what we're doing because it makes sense.
It's a rational way to do things. To try to apologize --
and then on the next page you say that the subsidiary
quantitative objectives are developed from the QHO's and are
generally consistent with subsidiary goals. Well, we have a
report from a former fellow that says that's not true for
the CDF. A CDF of ten to the minus four, in fact, is not
consistent with the QHO's. It's much more stringent.
MR. KRESS: That's why they put the word generally
MR. UHRIG: That's with a factor of ten? Well,
but the subsidiary goals, the Commission gave us the ten to
the minus four back in 1990.
DR. APOSTOLAKIS: But then --
MR. UHRIG: I mean, that's a subsidiary goal.
DR. APOSTOLAKIS: Do you have your figure 3.1 in a
transparency? Can you put it up?
MS. DROUIN: Yeah.
DR. APOSTOLAKIS: Okay. So, one question that was
raised earlier is under mitigate, condition of probability
of early containment failure, right? Why don't we have
late. Is that a question that you guys raised?
MR. KRESS: That was one of them, yeah.
MR. POWERS: Another question Dr. Kress frequently
raises and might be raised in connection with these figures,
is suppose that I have done an analysis and persuaded myself
that my plant has a core damage frequency of less than ten
to the minus four per year. I think Dr. Kress might well
ask gee, at what confidence level do you think that you have
met that criteria on ten to the minus four per year? I
wonder if you have a confidence level in mind for these
MR. KING: We have a mean value in mind. Maybe we
ought to go to our viewgraph.
MR. KRESS: You can also have a confidence level
associated with the mean. Even just saying the mean, to me,
implies a 50 percent confidence level in the mean, but I'm
not sure that's what you mean or not.
MR. KING: No.
DR. APOSTOLAKIS: They mean that mean value is
calculated in a PRA.
MR. KING: Right.
DR. APOSTOLAKIS: So the curve that gives you the
core damage frequency until you find the mean value.
MR. KRESS: Yeah, and if you put uncertainties --
DR. APOSTOLAKIS: On the curve?
MR. KRESS: On the curve, you get a confidence
level in the mean.
DR. APOSTOLAKIS: Yeah, but nobody --
MR. KING: No, but no PRA does that.
MR. KRESS: You won't get it out of the PRA, but
that's the way you do --
DR. APOSTOLAKIS: You don't get it out of
MR. KING: The reason we selected the mean was
simply because that's what the safety goal would -- that's
what would be consistent with the safety goal. The safety
goal is essentially based on the means.
DR. APOSTOLAKIS: Perhaps what you need here,
Eric, is a similar statement like the one in 1.174, that if
the mean values are close to these limits, there ought to be
increased management potential as to the issue of
MS. DROUIN: But remember, we're not applying
these numbers on a plant specific basis, and I think that's
a very important --
MR. KRESS: There's no way of knowing whether the
mean value approach is there or not.
MS. DROUIN: -- difference here that comes into
DR. APOSTOLAKIS: So, what are we applying them
MR. KRESS: How they write the regulation.
MS. DROUIN: How we write the regulation
generically. We're not writing a regulation to a specific
DR. APOSTOLAKIS: That's true.
MR. KRESS: And in fact, I'm not sure how you
translate these numbers into the regulation unless you do it
like the intent, and that's looking at limits on the
anticipated initiators and frequent initiators. Unless you
get down with that level, you are actually -- you know, you
actually have to write the regulation in such a way that you
meet these frequencies. I don't know exactly how you do
that, and that's kind of an interesting --
DR. APOSTOLAKIS: Generic business. I mean, let's
say the 19 PWR's have a greater damage frequency, so none of
these regulations would apply to them.
MS. DROUIN: I'm sorry?
DR. APOSTOLAKIS: We have right now, as you point
out right now, as you point out in the report under ID's, 19
PWR units where CDF is greater than ten to the minus four.
Some would say there are many more because they have not
included a number of other contributors, but you look at
numbers, I mean, the reports right now, it says 19.
MR. KING: Then we ought to look at what's causing
MS. DROUIN: What's causing that.
MR. KING: Look at the 19 and say what's causing
that. If there's some generic item that's causing that to
happen, maybe we ought to think about a change to the
regulations to take care of that generic item.
DR. APOSTOLAKIS: Say it's not generic. With the
regulations you produce here, then it should not be
available to these people because they're already above the
MR. KRESS: This won't appear in your regulations.
They'll have something else in the regulations. This very
well could meet those regulations, even as they are now and
not meet these numbers.
DR. APOSTOLAKIS: It's not clear to me what the
use of these would mean.
MR. KRESS: It's not to me either. I'm not sure
how you translate this into a regulation.
MR. BONACA: Well, the interesting thing about
this table is that the column on the left, anticipating the
issue, et cetera. That is exactly what you have in the core
licensing basis. Even those frequencies in the aggregate
are consistent with what is using the standards that goes
into the current accident analysis.
MR. KRESS: The DBA. These are the DBA's.
MR. BONACA: Exactly, the frequency. Now, the
only thing that didn't exist then was then the condition or
core damage probability, et cetera. Now, what you're
looking at then is specific requirements implemented that
would address conditional core damage probability for an
anticipated initiator and how in the aggregate they would
contribute ten to the minus four. That's what you would be
looking at, right? You would be looking at the effect of
requirements imposed to deal with conditional core damage
probability for a set of anticipated initiators.
MR. KRESS: Yeah, I was guessing they would use
these numbers, just like they do now, in designing a set of
design basis accidents, and in describing the
characteristics, you have to go into the medium.
MR. BONACA: Before you can link it.
DR. APOSTOLAKIS: But would these design base
accident apply to the current generation of reactors or the
MR. KRESS: It would apply to any reactor that
volunteers to use the risk base.
DR. APOSTOLAKIS: I mean, my plant has already
been designed against it.
MR. KRESS: So, that's the whole question of the
MR. BONACA: But no, taking the example of the
loca. You know, you go to the frequent initiator. Why
don't we just talk about an example to make it clear to all
of us. If I would put the loca inside the frequent
MR. HASKIN: Yes.
MR. BONACA: Less than ten to the minus two, the
reactor in the aggregate of all infrequent initiators. Then
I would look at the effectiveness of these is yes. I would
have to demonstrate that the requirements should impose, you
know, that the conditional core damage probability resulting
would be less than ten to the minus two. That would be a
criterion to impose the expectation on this is yes,
MR. HASKIN: Right, but you can take that example
a little further, and you can say well, there's a certain
class of large locas, large pipe break locas, which may
actually have frequencies of ten to the minus six per year.
Then we ought to be looking at eliminating those as design
MR. BONACA: Exactly, so what you would be looking
at is, you know, whatever you have in excess. Then there
would be a way of trimming whatever you do not have in
excess, so you're lacking. You may have new considerations.
MR. HASKIN: Right, and I think in retrospect in
practice, we'll wind up applying this more to accident
classes than to the three classes as we've, you know,
enumerated from here.
MR. KING: I think there's two aspects. There's
the aggregate aspect. There's the overall core damage
frequency LERF number. Are they being met, or is there some
class of plants that isn't meeting it? Then there's the
looking at individual accident classes. There we've chosen,
you know, one-tenth of the numbers in this table as sort of
a guideline to look at individual acts and classes. You
don't want one accident class to chew up the whole --
DR. APOSTOLAKIS: Which brings me to another
thing. I mean, this seems to be a very severe application
of the concept of defense in depth. You start out with
defense in depth. Because of defense in depth, you have
prevention mitigation. Because of defense in depth, you
have the four columns limiting the figures of initiators and
so on. So, you have a multiple application of defense in
MR. KRESS: Even down to the ten percent.
DR. APOSTOLAKIS: And then in the footnote, you're
throwing a bomb.
MR. KRESS: Yeah, that's a real one there.
DR. APOSTOLAKIS: That's really a big one. You're
saying even for each one of these, no individual accident
sequence should contribute more than ten percent of the
goal, which is -- the first time I saw that was in Sizewell
MR. KRESS: That's a very interesting concept.
DR. APOSTOLAKIS: The question is why. I don't
think it's even discussed in the text. Even if it is
discussed, I mean, the question is why did you feel that
there was a need for such a multiple application of defense
MR. KRESS: And then my comment in the
subcommittee meeting on that was why just ten percent on the
contribution to the main? Why not a percentage applied to
the -- related to the uncertainty also?
DR. APOSTOLAKIS: That would make it even worse,
but you see the point, Tom?
MR. KING: I understand your point.
DR. APOSTOLAKIS: There's this little thing that
says when applying the quantitative guidelines. In smaller
font in the note, it really is a major constraint.
MR. KING: And I'm probably largely responsible
for putting that in there.
DR. APOSTOLAKIS: Well --
MR. KING: I mean, it seems to me what this table
is doing is trying to say what's the balance we'd like to
see when we employ defense in depth? I mean, we always talk
about prevention and mitigation, but this is an attempt to
quantify what's the balance between prevention and
mitigation, and even cut it a little finer, looking across
initiating events, prevent core damage and so forth. But it
doesn't seem unreasonable to say okay, even if you achieve
this balance, I don't want one accident to chew up the
whole, you know, the whole segment of that balance I'm
trying to achieve, and therefore, I took it a step further
and said let's, you know, let's look and see if we do have
one accident like that. Let's have some criteria in here
that will force us to take a hard look and maybe make some
judgment that even that's too much.
MR. KRESS: This is our whole argument we had in
the joint subcommittee on risk allocation.
DR. APOSTOLAKIS: Yeah.
MR. KRESS: And I was advocating this risk
allocation, but I was advocating that you also have to
factor into it the uncertainties, contribution due to each
of these sequences. You can't just use the ten percent.
DR. APOSTOLAKIS: In essence, that's what Tom is
MR. KRESS: Yeah, but --
DR. APOSTOLAKIS: He says based on what I know
about the uncertainties. I think a ten percent value is
reasonable, I mean, without becoming explicit.
MR. KING: If you're concerned that this might be
too rigidly applied --
DR. APOSTOLAKIS: Let me give you what my concern
is. First of all, as I say, the first time I saw this was
in the context of the Sizewell B debate, and I was very
impressed by it. At that time, I happened to be advised in
the director of the new production reactor, DOE. Everybody
thought it was great until we actually did some
calculations, and the seismic risk was way up there.
MR. KRESS: And there wasn't anything you could do
DR. APOSTOLAKIS: And there is nothing you can do
about it. I mean, unless you are willing to double your
budget to bring down the seismic stuff, there's nothing you
can do about it. So, it's not always feasible. I think you
can find -- that's one argument, and I think you can find
different words that express the same thought without
sticking the ten percent there. Maybe you want to say that
it would be desirable to have a balanced design in the sense
that no accident sequence unduly dominates something like
that. So now, if you have this thing with the seismic where
you can't do much about it, at least you can say well, it
says unduly, it says this, you know, that kind of thing.
The second point here is that again, because
you're using -- see, I come to my earlier comment. You are
using goals, not adequate protection. I mean, how far down
are you going to go? You have defense in depth already
twice. Now you are adding another defense in depth measure
on a goal, which is good enough. I understand doing that on
adequate protection numbers, but not on the goal. But
that's more of a philosophical objection. I think the other
one is more practical.
MR. KING: Yeah. I understand your first line.
It's a good point. Maybe we ought to -- it comes across too
DR. APOSTOLAKIS: Yeah. That was, in fact, the
compromise and then --
MR. KRESS: Yeah, we had to compromise.
DR. APOSTOLAKIS: -- the letter that Dr. Kress
referred to, that we kept referring to a balanced design,
balanced -- not allocation, balanced defense in depth
MS. DROUIN: But I do think that you need to at
least have a starting point. I agree that I don't think you
should have -- you shouldn't apply it rigidly, and I think
you're going to need to take it on a case by case.
Personally, I don't like words like unduly. I
don't know what that means. Does that mean ten, 50, 60, 70?
You know, you ask ten people, ten people will give you a
different answer. I do think that you ought to have
something that you start from. You don't apply it strictly.
You know, you go in there, but at least, you know, that's
the ballpark of where you're at.
DR. APOSTOLAKIS: I think, Mary, you're going to
have that problem with a lot of risk informed regulatory
documents, that you have to be deliberately fuzzy, because
the moment you put numbers down, you know, why 60 percent?
Why not five percent or ten percent. I think it means you
would like to see it sort of an equal distribution, but
you're willing to accept something else, and you're willing
to listen to arguments. That's what it means.
MR. KING: Yeah, I mean, this is a level of detail
one step below what's in 1.174 in terms of having another
DR. APOSTOLAKIS: Yeah, that's right. This is the
first time we're using this.
MR. KING: Right.
DR. APOSTOLAKIS: And I also would recommend that
you take it out of the note, or leave it there but also
discuss it somewhere else, because it's really an important
thing. In fact, raise Mary's concern in that we recognize
that, you know, you can't have --
MS. DROUIN: We took the concepts from 1.174 that,
using your words, are fuzzy, and tried to come up with some
working definitions, recognizing they were just working
definitions, that it was a place to start. They weren't
absolutes, but at least it was something that, you know, we
could put our hands on. Then, as you look at it on an
individual basis, does it make sense to use a two percent.
Maybe for this particular case, a 50 percent is better, but
absent anything else, what's your fallback position? I
think, you know, you owe it to people to tell people what
DR. APOSTOLAKIS: You'll get into trouble with ten
MS. DROUIN: That's my personal opinion.
DR. APOSTOLAKIS: I bet you. Go back and look at
the IPE's. It's a routine. It's a routine that one or two
initiators dominate everything else.
MR. SEIBER: Cold pump seal failures.
MR. KING: Yeah.
MS. DROUIN: Yeah.
MR. SEALE: I can just see you now going through
and de-rating the quality of your design so that you've got
DR. APOSTOLAKIS: This was not the intent.
MS. DROUIN: But I think that if you go to the
point where you don't want something to unduly dominate,
then you owe it to explain what you mean by that.
MR. POWERS: My understanding, Tom, of your idea
behind your phrase there was it might really be that if I
have a ten to the minus seven plant, that it's nine times
ten to the minus eight of that is one accident sequence, and
that's fine. You don't have any troubles with that. That
it's only if I have a ten to the minus -- I have a nine
times ten to the minus fifth plant. If you don't want to
see eight times ten to the minus fifth of that coming from
one sequence. Is that my reading of it?
MR. KING: Yeah.
MR. POWERS: So, no, I don't think that you're
going to run into a problem. I don't think this phraseology
leads you to de-rating things to get competition because I
think we do have plants in the IPE's where because of the
way they did things and the way they've been things, there
are dominant accident sequences that make up big fractions
of the total, but it's a small total.
MR. KRESS: Yeah, but that concept doesn't come
across here, I don't think, that you know, that was one of
my argument I tried to make with those, that if the overall
risk number of a plant is very small, I don't care if it's
completely dominating the one accident if it's very small,
and any uncertainties are dealt with, but if you're up there
near the margins, then I worry about one accident dominating
it, but that makes it a sliding scale, in essence. You
know, a flat ten percent doesn't capture that sliding scale
concept to me very well.
MR. POWERS: Well, okay, maybe there's some
language in there, but I'm saying how I read it.
DR. APOSTOLAKIS: I think there is also an issue
here of cost benefit because I think even if it is, you
know, one accident that says, that has about 70 percent of
the core damage frequency, and you are close to ten to the
minus four, if that accident is, as I say, as was the case
with NPR, and it would nearly double your budget, just to
bring that down by a factor of three or four or five, would
you do it, since you are already meeting the goal?
MR. KRESS: Yeah, that's the concept.
DR. APOSTOLAKIS: So, I think the phraseology
should be vague enough to allow for flexibility but also
send a message that really a balanced design would be highly
desirable, balanced in the sense of nothing really
dominates. The reality is that the one or two things will
MR. POWERS: I would encourage the language to be
specific enough to allow some flexibility rather than vague
enough to be.
I wonder if I could come back to something Eric
said. You said gee, we might find some LOCA's that have ten
to the minus six probabilities and less that we should think
about getting rid of from the regulations. My mind turns to
the issue of reactivity control accidents. I don't think
they -- I'm not familiar with any PRA where reactivity
control accidents dominate a risk. The reason they don't is
the plants are designed by all these other regulations and
by Appendix A largely, so that that doesn't happen.
MR. HASKIN: Yeah, as a matter of fact, I agree
with that. Then our preliminary screening of DBA's that was
an accident that appeared high on our list of things that
would deserve risk informing. What we found from the
industry feedback was not too many complaints in terms of
excess burden of no overwhelming feedback from industry in
support of that being one of our early regulations to look
at risk informing, or DBA's to look at risk informing.
There was much more concern about the Part 46, but I think
we did a ranking of DBA's, and that was one of the top three
to look at from a risk informed perspective, specifically
the rod ejection accident.
MR. POWERS: I guess what I'm saying, the question
I'm asking is you have a probablist analysis tool or an
intuition on probablistics that's based on plans to comply
with a bunch of regulations that are addressing systems, but
you don't address those accident anymore because they so
completely eliminate the concern.
If you use that tool to say should I complete have
eliminated that concern or not, you come up with the answer
no, I shouldn't have deleted it because it's so low in risk,
I should not be regulating this. You don't have a mechanism
to say now, if I don't regulate it and they do start
creating, doing things in the plant that cause reactivity
control problems that yes, it will show up high on my risk
analysis, it's not built into the tool. It's not built into
your tuition right now.
MR. HASKIN: Right, and that was another reason
for not pursuing that one from the outside because as you're
aware I'm sure, as you go to high burn-up fuel, there's a
set of questions in terms of fuel performance that comes up
that I don't think anybody has fully analyzed yet. We just
felt we were in a better position to start with 46 rather
than -- and it was more important in the overall scheme of
MR. POWERS: Might have a little visibility on.
MR. HASKIN: Yeah.
MR. POWERS: Let me ask another question.
MR. KING: I don't totally agree with your
premise on the previous question, by the way.
MR. POWERS: Oh, okay. Maybe you could tell me
more about that.
MR. KING: If it turns out the design basis
reactivity insertion actions is and the PWR's are rod
ejection, you're right. On a risk assessment, it doesn't
MR. POWERS: It barely shows up.
MR. KING: Barely shows up, but if for some reason
that is causing operational pen at least and would make
sense in terms of a burden reduction, would really remove
some unnecessary burden. I mean, the question is one, what
would you replace it with and two, what's your basis for
saying that the probability is low enough that yeah, it
falls below, you know, the things we worry about on this
table of frequency.
MR. POWERS: If you replaced it, do you still fall
below the place where you'd worry about it?
MR. KING: I mean, you'd replace it with something
that would fit within this framework of frequencies here. I
think the real question is how confident are you that the
designs and the other regulations really insure that the one
that's in there today is going to stay low. It's not going
to pop up again if you change the regulation.
MR. POWERS: Yeah, that is the question.
MR. KING: That's the real question.
MR. POWERS: Uh-huh.
MR. KING: And I think, and even though Eric said
yeah, this is not number one on the industry's list of
things to change, we may eventually get to that one.
MR. POWERS: Well, the question is now, suppose you
-- I don't know what you change, but suppose you change
something, and an all knowing PRA says that became a ten to
the minus third event. How do you find that out?
MR. KING: You mean the rod ejection all of a
sudden becomes ten to the minus three because you changed
MR. POWERS: By hypothesis it does, and I'm asking
the question how now -- you've come through and you've said
gee, the regulations I've got now that keep the rod ejection
accident down low are just an unnecessary overwhelming
burden, that it's silly to do this. It's preventing people
from using fuel to high burn-ups, and there are lots of
society benefits I can find for using fuel to high burn-ups.
Tangible benefits to the public, regardless of what the
economics is to a plant. Now, what do you do? Do you
have a mechanism that allows you to go back and say okay, I
now incorporate this change into my PRA tool? I don't know
what you're using. Maybe you're using some unpeer reviewed
thing or maybe it's a wonderful thing. I have no idea, and
see whether the probability came up?
MR. KING: I think what you do is you say I'm not
going to keep the rod ejection as my design basis accident
anymore. I think you'd make sure the regulation has enough
teeth in it that the design permits rod ejection accidents,
whether it's, you know, you put some design requirement on
the control rod housing or you know, maybe you already have
something in place that does the job, but I think you'd have
to look at the regulation and see, you know, the frequency
of that is very low today. What in my requirements keeps it
very low, and if it's already in there, fine. If it isn't,
you might want to put something in that keeps it very low.
I mean, that's my approach to that.
MR. POWERS: I guess I'm still puzzled by that,
but I'd like to ask another question that puzzles me. One
of the questions that's come up recent years has been gee,
our steam generator tubes and PWR's are suffering a certain
amount of degradation. Do we have the possibility that as
accident progress, we fail those tubes, and that's a problem
for us because it creates a bypass of the containment and
risk numbers usually result.
People have done a variety of analyses and
whatnot, and they come back and say gee, there's these
wonderful natural convection phenomena that show up that
cause the system to rupture someplace else first. So, we
don't have any problem with that. When you ask them, how
much of that phenomena that they're hypothesizing, were they
able to see at TMI? They respond by saying well, no, there
was no heating in the peripheral regions of the core at TMI.
You're a little bit surprised, and you say gee, are we
analyzing accidents in our PRA tools or the associated
analyses that are stylized to the point that they're not
giving us good information on details of regulations.
So, I guess what I'm asking you is do we know that
the analyses that we're using are, in fact, sufficiently
detailed to address some of these questions of details of
MR. KING: That's a generic question that applies
to more than steam generator?
MR. POWERS: Yeah, I just used that as an example
to motivate my question.
MR. KING: I'm not sure there's a -- I'm not sure
the framework addresses that, and I'm not sure that we can
answer it at this meeting. I think one of the things I know
Mary and I have talked about is in the PRA standards effort,
what are they saying about making sure we have verified,
validated analytical tools that we're using to calculate the
phenomena. You may have the best vent trees, fault trees in
the world, but if your, you know, thermohydraulic tools
aren't any good, your PRA isn't any good. I'm not sure we
have a complete answer to that question.
MR. HASKIN: I think as we get into risk informing
the individual regulations, things like that are going to
appear. In fact, we've already seen some of those in Part
46. We're just going to have to deal with them as we go.
Whether that means that people are going to have to revise
PRA's or we're going to have to do some internal work, we
haven't progressed far enough to determine, but there are
biases both in the deterministic analyses and in risk
assessments that overlook certain phenomenologically
plausible scenarios. So, you just -- we simply have to be
aware of those to the extent that we can as we're proceeding
with the risk informing.
Now, certainly if it's a matter of, you know,
group A says an accident's going to progress one way and
group B says an accident's going to progress another way,
that's an uncertainty that has to be dealt with. Maybe that
precludes us from doing something in risk informed space
because it's an uncertainty that's either unresolved or not
resolvable within the time frame that we're looking at. I
think the only way to address that is on a regulation by
regulation basis as we go through these things.
MR. POWERS: I guess I would feel more comfortable
if those kinds of issues were spelled out a little more
clearly. Just say that that phenomenological activity needs
to become part of the overall strategy here.
MR. KING: I agree. I think that's a good point.
In fact, one of the things we've talked about doing, since
most of the industry PRA's are based upon the map code, how
well do we understand the map code, its strengths, its
weaknesses, so we can ask these kinds of questions, and
we've kicked around the idea of doing a review of the map
code so we could get some idea of the strengths and
limitations. For a number of reasons, that hasn't started
yet, but it's right to the heart of your question.
DR. APOSTOLAKIS: Can we take a break now until
DR. APOSTOLAKIS: Okay, we're back in session.
Anything else that the Subcommittee members have on this
figure 3-1, which would you please put back up? It's very
You know, a big question in my mind is how does
one use a figure like this to actually write individual
regulations, but I guess we're going to see examples at some
MR. KING: Yeah, I mean, we use this in 5044,
which we went through with the Subcommittee a week or so
DR. APOSTOLAKIS: Yeah.
MR. KING: And since it was really talking about
containment performance, we really only use the numbers in
the one column, but we basically took risk information for
the cross section of plants that we had. We looked at how
well the conditional containment performance was under
hydrogen combustion type events. We found some cases where
those numbers were met and some cases where they weren't,
and then the question is okay, for those that weren't, what
would it take to eliminate the problem that's causing the
performance issue, and that's where we talked about, for
Mark III's and ice condensers during station black-out,
maybe the igniters ought to have alternate power supply
because that was really the thing that was driving that.
So, I mean, that's the way it worked.
DR. APOSTOLAKIS: Okay, anyone has a comment on --
MR. KRESS: So, the regulation would say Mark
III's and ice condensers need alternate power supplies to
the igniters, something like that?
MR. KING: Something like that.
MS. DROUIN: No, no. The regulation would say
that we want them to have a hydrogen control system that is
capable of meeting their risk significant accidents. Now,
they can come in and either show that station black-out as
an accident is not contributing, or if station black-out is
contributing, they can deal with it. So, we are not
prescribing that they have to have D.C. power back-up, but
we have analyses, you know, that show for some ice
condensers and for some Mark III's that they're having
relatively high conditional containment failure
probabilities against our guideline, and we're saying okay,
that's a concern.
MR. KRESS: That's not necessarily a risk
MR. KING: If station black-out is very low
frequency, as Mary said, and somebody could come in and make
that case, then that would be good enough.
MS. DROUIN: That would be good enough.
MR. KING: But if it isn't, if it's a fairly
dominant contributor, then they're going to -- how do I deal
with hydrogen combustion and hydrogen control under station
black-out conditions? Mary's right. The way we're -- the
alternative we're developing is allow some flexibility on a
licensee to come in and say this is how I'm going to do it.
It doesn't say well, you'll hook it up to, you know, the 1-E
batteries or something like that. It's not that specific.
MR. BONACA: I have just one more question on the
figure which we raised during the Subcommittee meeting when
Mr. Christie made a presentation. He had some concepts, you
know, similar. The issue is here you're looking at, for
example, containment failure probability, and there you're
really introducing a criteria that is used typically in
probablistic risk assessment. I mean, you're looking at the
ultimate capability of containment, which is not really what
the units are committed to in core design. It would be to a
different kind of volume.
So, you're really introducing an interesting
mixing of criteria there, isn't it?
MR. KING: You're saying, for example, you're
worried about aging concerns and maybe that containment is
strength is not the same?
MR. BONACA: Now, clearly you're talking about
LERF here, so LERF is a large early release, but I'm just
wrestling with that a little bit because you're taking now,
for example, 5044, and you're running it through the process
of this type. In doing so, you're now -- 5044 included a
number of criteria, okay, within 5044 that included, for
example, a certain containment capability, which had to do
purely with the design pressure of the containment, which
was somewhat a very conservative estimate of the capability
of the containment. It allowed for aging of the
containment. It allowed for uncertainties of all kinds, and
to credit only for one-third of the pressure capability.
Now, you're running the same regulation through,
but you are using different criteria to make a judgment on
the performance of the containment. Now, you may do the
same without the criteria that pertained to the condition of
the core damage probability, for example, best capability or
piping rupture that you'll assume and so on and so forth.
MR. KING: You're right.
MR. BONACA: Maybe you're right. I just --
MR. KING: Using ultimate strength in containment,
for example, not design pressure. You know, how much has
that ultimate strength eroded by things like corrosion over
time. That's not factored into the analysis.
MR. BONACA: Because I mean, again, in current
evaluations, you are taking credit for what the utilities
are committed to, which is a conservative assumption of
that, and here you're now -- all right.
MR. KRESS: Let me ask another question. These
regulations that you're redoing, if DOE or some
representative of DOE comes in with the Gen IV reactor,
would these regulations still apply, or were they flexible
enough that they can fit that kind of a different concept
into them? Let's assume that Gen IV turns out to be an
HTGR, and they come in and say well, we don't really need a
containment, so we don't need to meet this ten to the minus
one because we don't have any such thing as a LERF anyway.
This is so low that forget about it. Will they accommodate
that, or will you just have to do a special view of it?
MR. KING: Well, my own view is the framework
would accommodate that.
MR. KRESS: The framework?
MR. KING: The framework would accommodate that.
DR. APOSTOLAKIS: If you replace a conditional
early containment failure probability by conditional early
radioactivity release probability, then I don't need to go
MR. KRESS: That's right.
DR. APOSTOLAKIS: If they can demonstrate that
it's less than this.
MR. KRESS: At a certain probability --
DR. APOSTOLAKIS: This assumes you have a
containment, but you really --
MR. KING: Or confinement. I mean, it doesn't
have to be the traditional containment building. It could
be a confinement building like Fort St. Vrain had. They
didn't have a containment building.
DR. APOSTOLAKIS: It's really the release you're
MR. KING: Yeah.
DR. APOSTOLAKIS: Not so much a structure.
MR. KING: Large early release.
DR. APOSTOLAKIS: Now, another point I want to
raise here is the issue of external events. I remember that
somewhere in the document we're saying that some initiators
cannot really be handled in a neat way like, you know,
frequency, condition or core damage and so on. Then you
move up, right? Like an earthquake is shaking the whole
plant, and it's not really very meaningful to talk about the
condition of core damage probability separately from the
containment failure probability.
In fact, I'm not even sure. You can talk about
prevention versus mitigation. You have to go even higher
than that. The goals themselves perhaps, or the release.
MR. KING: I think you can still talk prevention.
DR. APOSTOLAKIS: As one package, the release of
radioactivity. So, you have ten to the minus five or six,
whatever you want, because the same event affects both
prevention and mitigation, doesn't it?
MR. KING: But I think you can still talk the two
separately for an earthquake because --
DR. APOSTOLAKIS: The question is whether you want
to have a goal for separate, I mean, if you have such a
major event. Now, you classify them as rare events a little
later, as rare initiators.
MS. DROUIN: That we did.
DR. APOSTOLAKIS: Right?
MR. KING: Right.
MS. DROUIN: Yes.
MR. HASKIN: There were some things that are going
to be low on frequency or not going to regulate. I mean,
we're not going to require people to design for rare events,
but as Tom mentioned earlier, if there's something that
needs to be done to assure that a low frequency event stays
a low frequency event, that may be within the purview of the
MR. KING: Yeah, that's a good point.
DR. APOSTOLAKIS: What does it mean that you do
not require them to design for rare events?
MR. KING: It depends on what you're talking
about. If you're talking about, you know, vessel rupture,
the reason we're confident the vessel is going to hold
together is because of the design requirements.
MR. KRESS: Well, let me throw in my example for
that one. I hate to bring it up now, but the spent fuel
pool, the problem with the fire and the loss of cooling or
the loss of inventory. It's a rare event. It falls in that
category in terms of frequency. The problem I may have with
your table is that you probably assume it's already
conditional of early containment failure probability of one.
You forget about the containment. So, the question is is
this ten to the minus -- the value of the product of the
conditional core damage and the initiator frequency, is it a
sufficient criteria to guide a regulation on a spent fuel
I would say it's not, and that's because this goes
back to my original concept that you're not really dealing
directly with fission products here. You're dealing with a
source term related to the reactor. You get a different
source term for the spent fuel pool. You might get a
different LERF value for the prompt fatality goal. So, I
wouldn't apply this table directly to the spent fuel pool,
because I would have a different set of criteria. I'm
hoping that that doesn't happen just because this table
MR. KING: Your point's a good point in that the
LERF value we're using here, and we're using 1174, was
derived from a reactor --
MR. KRESS: Arrived from a reactor, yeah.
MR. KING: -- accident, a single core. A spent
fuel pool could have several core loads of fuel. The timing
is different. You know, the air oxidation or whatever is
different. LERF could be different for a spent fuel pool to
meet the same safety goals.
MR. KRESS: Right.
MR. KING: But the concept of a table like this
could be developed.
MR. KRESS: Could be developed specifically for
MR. KING: Yes. I mean, we didn't have the spent
fuel pool in mind when this was put together.
MR. KRESS: But when you come to a regulation that
involves the spent fuel pool, you would keep the concept of
the actual quantitative goal in mind close to these numbers
MR. KING: You could develop quantitative goals,
back them out consistent with the overall safety goal
DR. APOSTOLAKIS: So if I were to do a seismic
analysis -- of course, I can have a spectrum of earthquakes,
so some of them would be infrequent initiators and some of
them will be rare initiators, and I would have to look at
the analysis and judge whether I met the various criteria
there? Is that how it's going to work?
MR. KING: I think in principle that's how it
would work. I don't think we've gotten into a seismic event
yet. I don't know. Mary, have you thought anymore about
MS. DROUIN: Not beyond what we have here.
MR. KING: But in principle, that's the way it
DR. APOSTOLAKIS: No, these things tend to
MR. HASKIN: Other than the fact that we are
getting into that as we proceed with Part 46, we'll
obviously have to consider that aspect of risk.
DR. APOSTOLAKIS: Anything else on this figure?
MR. BONACA: Just I would like to add one thing.
I raised an issue before containment. Clearly what was
making me somewhat uneasy was the fact of not linking it to
the uncertainty section, to the section that later on talks
about this, you know, regarding uncertainty. That should
take care of it. Simply, I'm still thinking about the fact
that most PRA take credit for, you know, that capability of
containment. This is an example of what to take credit for.
I'm not sure that the uncertainty analysis recognizes, for
example, aging the way we have seen it in license renewal,
where the capability of the containment is significantly
reduced. It seems to be, at least.
MR. KRESS: Well, once again, it's a sort of a
MR. BONACA: Yeah.
MR. KRESS: And PRA's don't deal well with
uncertainties and success criteria. I think that's
DR. APOSTOLAKIS: Well, there are model
uncertainties, and I think you see more and more concern
about model uncertainties.
MR. KRESS: Yeah, but they rarely show up in the
success criteria, though, even the model.
DR. APOSTOLAKIS: Yeah, but we will start pushing
MR. KRESS: Yeah, well we need to deal with it.
DR. APOSTOLAKIS: And one of the reasons why I
don't like the word best estimate code. The thermohydraulic
arena is that I think they should make a statement of the
uncertainties due to modeling approximations to the extent
they can. It's about time we started doing that and forget
about the best estimate.
MR. BONACA: Yeah. I guess we got to see some
examples of how they run through this table.
DR. APOSTOLAKIS: This is not cast in stone.
MR. BONACA: Oh, no, no, no.
MR. KING: You raised a good point. Maybe the
uncertainty chapter ought to touch on that.
MR. BONACA: No, and I think certainly that
actually attaches on the issue in general. Maybe some
specific discussion of how you resolve some of these issues
will be helpful, if you know already how to get there.
MR. KING: At least pose the question. When
you're going through this, you'd have to think about those
kinds of things.
DR. APOSTOLAKIS: Okay, I have a few more comments
after this figure. Anyone wants to continue debating the
figure? Okay, page 3-3. There is a series of bullets on
the left-hand column. The last one says no quantitative
objective is proposed for conditional individual fatality
probability because existing PRA's demonstrate that the
QHO's can generally be met through the preceding three
quantitative objectives. For individual -- conditional
individual fatality probability.
MR. KRESS: That means they just didn't use the
DR. APOSTOLAKIS: QHO's don't talk about
conditional probabilities. They talk about absolute, one
tenth of one percent.
MR. HASKIN: What we're saying is we didn't
establish a goal for column four. We were just using the --
DR. APOSTOLAKIS: Then it says however, off site
protective actions are essential to protect the public.
MR. KRESS: This is conditional on the other three
things happening, the conditional only. It's conditional on
the other three things, yeah.
DR. APOSTOLAKIS: On the other three, yeah. But
do you agree with the last sentence? Off site protective
measures are essential to protect the public?
MR. KRESS: You can't meet the safety goals
MR. KING: The risk calculations --
DR. APOSTOLAKIS: See, that's what bothers me
about this business of individual risk. It's not individual
risk, because if we're individual risk, whether we evacuate
or not should be relevant. The way it's calculated, by
calculating societal and dividing by the number, why that's
not individual. That's a misnomer then.
MR. KRESS: It's a misnomer, but it's --
DR. APOSTOLAKIS: It's a misnomer.
MR. KRESS: It's too late to do anything about
DR. APOSTOLAKIS: Individual means you take an
individual like in the Yucca Mountain thing, and you put
them there, and you say 24 hours, they're there.
MR. KRESS: That's why it's always given the
DR. APOSTOLAKIS: So, it's a representative
MR. KRESS: No, it's not that.
DR. APOSTOLAKIS: Societal normalized by the
number of people.
MR. KRESS: No, it's societal risk at all.
DR. APOSTOLAKIS: It's normalized.
MR. KRESS: It's normalized. That's why it's not
a societal risk.
DR. APOSTOLAKIS: But it's a measure because if
you evacuate, you effect really the societal risk, not
individual. Can we do something about it, though? Is there
any place where we can -- this is not individual risk
because individual risk should be independent of whether you
evacuate or not. That's the whole point, that you assume
the guy's there. Probability of death of an individual
who's 24 hours right there. Unless, again, like the
unavailability business, we are using terms that are not
used by the wider community.
MR. KRESS: I think we are.
DR. APOSTOLAKIS: I think we are.
MR. KING: Early versions of this framework table
did have numbers in that fourth column, the emergency
planning column. We decided that didn't make sense for a
couple of reasons. You know, one is if we're talking
regulations that deal with design and operation, the
emergency planning column doesn't deal with that.
MR. POWERS: Why doesn't emergency deal with
MR. KING: Operating the plant?
MR. POWERS: Yeah, plant, the plant operations.
Activities of the operator. The operator sees conditions in
the plant. He's called upon to make judgments that activate
various aspects of the emergency procedures, or the
emergency planning activities.
MR. KING: Yeah, but I think in effect what you're
saying is independent of the design, we want that emergency
planning capability. You want the ability to have off site
evacuation. That's independent of the design.
MR. KING: ...minus fifth CDF doesn't matter
MR. APOSTOLAKIS: That's a defense-in-depth
MR. KING: Defense-in-depth measure --
MR. APOSTOLAKIS: But this is different from
saying that it essential to protect the public.
MR. KRESS: It is.
MR. APOSTOLAKIS: It may happen in some cases to be
but I would be much more comfortable saying it is a
defense-in-depth measure than saying it is essential.
MR. KING: Except remember the LERF value that we
the 10(-5th) LERF value that was backed out of safety goal
calculations. Those safety goal calculations made
assumptions on evacuation. So that LERF does have a
built-in assumption that evacuation takes place.
MR. KRESS: Absolutely.
MR. KING: And if you didn't make that assumption
you would have a different LERF value. So in that respect
it is --
MR. KRESS: Now you have a (inaudible) on the book
associating the siding -- (inaudible) siding.
MR. KING: Yes.
MR. KRESS: And those, some aspects deal with
population distributions around the plant or --
MR. KING: Population centers.
MR. KRESS: Population centers, basically. And I
presume, those sort of regulations would be considered
defense-in-depth and you won't try to touch those with risk
informing them. I mean, I can see how you could do some
risk informing of that but --
MR. KING: Part 100 is not high on our list of
look at in risk informing. I won't say it will never be
looked at, but it is not in the top ten.
MR. KRESS: What is your point? But you know what
it does it limits -- limits the number of sides (inaudible)
MR. KING: (inaudible) top 10.
MR. APOSTOLAKIS: Would you say that for a plant
that is relatively modern, like South Texas Project, has a
very low core damage frequency. That that side protected
action are essential to protect the public?
MR. KRESS: No.
MR. APOSTOLAKIS: So I suggest you change the
MR. KRESS: It is very plant specific.
MR. APOSTOLAKIS: It is very plant specific and the
word "essential" -- I think essential --
MR. KRESS: Well, if you are talking about on the
average for the whole population of plants, generically it
MR. KING: But I think what is fair to say is that
our LERF value is based upon calculations that make
assumptions on evacuations. Therefore, --
MR. APOSTOLAKIS: Then on the right-hand side
column, we have again the same political statement that
would not be prudent to simple replace existing regulations
with a quantitative objectives. Compliance with, you know,
it is the same like in page 21, so.
Then on the next page, unless somebody has a
question, 3-4: "Anticipated initiators are either expected
to occur or may well occur." Seems to me that is a
(inaudible). Something that is expected, turning out I'm an
English expert. Something that is expected to occur --
MR. SEALE: May well occur.
MR. APOSTOLAKIS: May well occur, right.
MR. KRESS: Where are you reading at?
DR. SHACK: Second column, second paragraph.
MR. APOSTOLAKIS: Anticipated initiators is way
down bottom, that's right. I mean, it is not essential, it
just struck me.
MR. KRESS: Yeah, it is a little strange.
MR. APOSTOLAKIS: It is the same thing.
DR. SHACK: Give it to your tech editor.
MR. APOSTOLAKIS: Now, in the text, of course, the
initiators are placed in -- according to their frequency in
equalities in the table, you just have the upper bond it is
understood, I suppose. The lower bond is the upper bond of
the next category.
"Even when common cause failures and human errors are
accounted for most plants can meet the proposed plant state
I guess he means no power, huh? There is a
sentence I don't understand on page 3-5. So, "The
defense-in-depth approach does not ignore rare events,"
okay. Righr-hand side column.
"Tactics such as research, and so on, are applied
to validate the low frequency of rare initiators.
Generally, however, a risk informed regulation will not
require plant structure systems and components be
specifically designed to cope with rare initiators."
I don't know. I mean, do you really know
MR. KRESS: Unless it changes the initiators
(inaudible) frequency itself. I mean, by (inaudible)
MR. APOSTOLAKIS: Sometimes it is rare because of
the design, right?
MR. KRESS: Maybe.
MR. KING: Well, I mean, we don't design for
meteorites. We don't design for huge earthquakes.
MR. APOSTOLAKIS: Meteorites, we don't. But LOCAs
we have a lot to do with, even big earthquakes.
MR. KING: Well, this initiator there are the ones
you just do not worry about like meteorites because they are
so right. But then there are the others that Eric was
talking about the reason they are rare is because the design
is so rare.
MR. APOSTOLAKIS: The classic example in California
is that an earthquake of a certain magnitude occurring in
Los Angeles, and occurring in Guatemala has very different
consequences. So our designs must have something to do with
the rare initiator and its consequences, right?
MR. KING: But if you are building a plant in
Florida you are not going to design for the big earthquake
in California. You would design for whatever you think
makes sense for Florida.
MR. APOSTOLAKIS: Right. And that is what you are
MR. KING: Yes.
MR. APOSTOLAKIS: That components be specifically
designed to cope with a rare initiator?
MR. BONOCA: Because you conceive that the
earthquake in California to be (inaudible) in Florida?
MR. KING: Yeah. There are some things that are
sight specific: earthquakes, hurricanes.
MR. APOSTOLAKIS: But that is not the fear. I mean,
that is something that comes naturally for risk inform the
regulation the regulation. A risk informed regulation will
not require plant (inaudible) to be specifically designed to
cope with rare initiators.
MR. BONOCA: It is general (inaudible)
MR. APOSTOLAKIS: I think a risk informed approach
would look at all spectrum of all possible accidents and
Now if what --
MR. SHACK: If all he is saying is that the
initiator has a low enough frequency you don't design for
asteroids, because the --
DR. POWERS: But that question has come up before.
I am trying to have things that are risk significant, but if
I only compute frequencies, how do I know what is
significant? I mean, clearly to my mind there must surely
be some frequency below which I am not concerned. I am not
concerned about the meteorite.
MR. KING: Yes.
DR. POWERS: I mean, I refuse to let myself be
concerned about that --
MR. KRESS: Yeah, because then (inaudible) the
whole inventory in the core and then -- as a release and
still meet the goals, yeah. So small.
DR. POWERS: Is that the reason that some how I
know that if I smack this thing with an asteroid and I
explode out of it the entire core inventory plus the spin
fuel, that even then multiplied times the probability that
the asteroid -- I know that is some intuitional --
MR. KRESS: I would guess so. That would be the
only logical way, yes.
DR. POWER: I would guess that that is the only way
I would do it. Now I ask myself what is that number that
allows me to quit calculating frequencies?
MR. KRESS: It ought to be calculable.
DR. POWERS: It ought to be estimatible. They
ought to be able to put that down on a piece of paper --
MR. KRESS: I never seen it actually.
DR. POWERS: -- and look at it. I don't know what
MR. KING: Wouldn't that be the safety goal
numbers. The early fatality numbers?
MR. KRESS: Oh, yeah. But he just wants to know
what the frequency is.
DR. POWERS: Yeah, I want to know -- I got a
safety going risk but it is a big chore to go from
frequencies to that risk number, because Dr. Kress would get
all over my case because I used the wrong force term, or
something like that. And Dr. Powers would climb on because
I used the wrong dispersion code. And there are a lot of
funny people in on this committee, they are very difficult
to get along with. And so, it would be a lot easier if you
can tell me if your frequency is below this thou shalt
forget about it.
MR. HASKINS: The number we are currently using is
10(-6th). That is the number that has been used for
aircraft impact, for example.
DR. POWERS: So you actually have a number that you
would like to use as 10(-6th)?
MR. KING: Yes.
MR. HASKINS: That is the number we are currently
using, and that number appears in the standard review plans.
MR. KRESS: Now the question you have is, why?
What is the basis of it?
DR. POWERS: I don't believe that number.
MR. KRESS: It seems a little high, doesn't it?
DR. POWERS: Yes
MR. HASKINS: The number is based on ten percent of
10(-5th). The ten percent came from the earlier discussion
and the 10(-5th) comes from the figure.
MR. APOSTOLAKIS: But not in the standard review
plan, because there were not thinking that way.
MR. HASKINS: No, but also it was consistent
MR. KRESS: See that number has to be different for
aircraft impacting the thing versus my asteroid impacting
MR. APOSTOLAKIS: My question here is, isn't this
an issue handled naturally by the existence of safety goals.
Why do I have to worry about it? I mean, that is the
MR. KRESS: Well, from the point --
MR. APOSTOLAKIS: And later on in fact it says, "To
focus on reducing risks associated with rare initiators
would draw attention away from, and potentially increase
risk associated with more likely initiators."
Why do you have to worry about it? I mean,
you have their goals; you met them. There are always
initiators that are much less frequent, you know.
MR. KRESS: That's because the goals are not
revealed into the -- aren't explicit in the regulation.
They are only implicit.
MR. APOSTOLAKIS: But not with this document. They
would be very implicityly, right.
MR. KRESS: They will be implicit; they won't be
explicit. That is the problem. In they are implicit which
means you don't have to really meet them. You might meet --
the probability of meeting is very good.
MR. APOSTOLAKIS: There must be a better way of
handling it though. Because, I mean, that is the whole idea
of a goal. I am recognizing explicitly -- acknowledging
explicitly that there is such a thing as a residual risk.
MR. KRESS: But the only way to deal with that is
to have the numbers actually built in explicitly in your
MR. APOSTOLAKIS: That is worst.
MR. KRESS: Which is a no-no, right now.
MR. APOSTOLAKIS: But then it is risk based. We
will all go to hell.
MR. KRESS: Oh, yes. Then it's a no-no.
MR. APOSTOLAKIS: Oh, no, no, no. No risk base
MR. KING: No risk base.
MR. APOSTOLAKIS: So you guys agree that this crazy
ideology makes sense?
MR. KING: We can look at it again, but the idea is
there are some things that are below concern and frequency
is one way to look at them.
MR. APOSTOLAKIS: All right, anything else here
from anyone else. Next page; Additional Thoughts on
Quantitative Objectives." Again, I don't believe that some
of your subsidiaries want (inaudible) objectives.
I mean it says, that there are no risk
arguments for setting subsidiary quantitative objective most
stringent. Second paragraph from the left. And I think we
made it clear that CDF is more stringent, 10(-4th), is it
not. The CDF 10 (-4th) goal is more stringent then one
would derive from the QHOs. independently of what the
MR. KING: Actually, that word is --
MR. APOSTOLAKIS: And that is (inaudible)
MR. KRESS: Yeah, that is definitely true.
MR. APOSTOLAKIS: But in several places this
document says, "That the subsidiary goal should not be more
stringent the QHOs." And I think that language should be
softened because some things are more stringent.
DR. SHACK: How about more stringent than the
commissiion safety goals?
MR. KING: Then is should be the subsidiary
MR. APOSTOLAKIS: The 10(-4th) is still there?
MR. KING: The subsidiary objective 10(-4th) was
endorsed by the commission.
MR. KRESS: It is still there.
MR. APOSTOLAKIS: Later, though. Not in the
original statement, as I recall.
MR. KING: Not in the original statement. In the
MR. APOSTOLAKIS: Oh, well. Okay. So we have
elevated it, then.
MR. KRESS: Well, in the defactor.
MR. APOSTOLAKIS: Now there is another statement in
the next paragraph why there is no basis for being more
stringent than the QHO's both defense-in-depth and
uncertainties, which tend to grow as postulated accidents
proceeding time, influence, blah, blah, blah. "The uncertain
extent to grow as postulated accidents proceed in time."
Is this something you all agree on.
Okay. Since we do, it is just that it is a new thought that
is thrown in there as a secondary clause that says --
MR. HASKINS: You can look at 1150 in that
MR. APOSTOLAKIS: Yeah, yeah, yeah.
DR. POWERS: I mean, it is the whole basis for
being reluctant to go beyond Level 1 kinds of PRAs, because
we can't get technical consensus on the Level 2. I think if
we could get it on Level 2's it would not be hard to get it
on Level 3. Level 2 is a challenge for us right now.
MR. HASKINS: You will still have a problem
MR. KING: Level 3 is a challenge.
MR. APOSTOLAKIS: Page 3-7 guidelines for each
suggested to leave the table, it's motherhood and apple pie.
It really doesn't say anything. How do limit the frequency
of accident initiators. Provide assurance of the combine
frequency less than one per year; provide assurance that the
combine frequency, I mean, sure. It is almost a
(inaudible). I think the table does not provide any
(inaudible)size should be limited. Limit the probability of
core damage. How do you do that? By providing assurance of
the probability of core damage is less than 10(-4th). Is
that the same thing?
MR. KRESS: Well, it just expands on
MR APOSTOLAKIS: It is assuming.
MR. KRESS: So it is an expansion of -- what do you
mean by that. It is an expansion of what you mean by the
MR. APOSTOLAKIS: It restates the figure. It
really does not add anything.
MR. KRESS: Yes, it does.
MR. APOSTOLAKIS: It does.
MR. KRESS: I mean it restates the figure, yes.
MR. APOSTOLAKIS: Does the staff members feel
otherwise, strongly, to keep it? You think about it. Take
it into advisement and you make a decision in the proper
MR. KRESS: In the statement under strategy four
does add some stuff.
MR. APOSTOLAKIS: I'm sorry, four?
MR. KRESS: Yes. It talks about this (inaudible).
MR. APOSTOLAKIS: It is probably the only one that
goes a little beyond.
MR. KRESS: Yes. It goes a little beyond what you
are talking about.
MR. APOSTOLAKIS: But if you look at the first --
at the top of the page on the right, Table 3-1 provides a
list of guidelines. That is not what that is.
MR. KRESS: It is not guidelines.
MR. APOSTOLAKIS: It is not guidelines.
MR. KING: We will look at it.
MS. DROVIN: All Table 3-1 was meant to do was to
take the information and summarize it from the text. So if
you wanted to go to a Table and see it all without having to
go through the text, that is all it was meant to do. Just
and aide to the reader.
MR. KRESS: I did have a question on page 3.6,
George. 3-6, in right-hand column, second paragraph and
this is the old version.
MR. APOSTOLAKIS: Yes.
MR. KRESS: And the underline to specifically deal
with latent cancers a quantitative object of Point 1 is
proposed for the probability of a late large release in a
core damage accident.
Now the Point 1, and I am not sure where it
comes from, and I am not sure why the latent cancers is
relegated strictly to late large releases of -- I guess you,
the rationale is that your guideline is already on large
early release. Deal with that part of it. Now you are
talking about the residual, the late part, and if you set
conditional containment probability on that, covered the
whole spectrum, I am not sure where the Point 1, how it had
just derive from thinking along those lines and then going
to the quantitative help objective on latent cancers.
MR. KING: Well, the Point 1 is just a parallel to
the Point 1 for large early -- or conditional containment
failure probability for large early releases. We call it
large early release which derived from --
MR. KRESS: I just -- what my question was more
specifically, because I am not sure if I started from the
latent death quantitative health objective and worked
backwards to get these conditional containment (inaudible).
That I would actually get this number. You know, I have not
seen that exercise done as of now.
DR. POWERS: I don't think we went through that
MR. KRESS: So my question is, is the Point 1 the
right value to use?
DR. POWERS: It is a good question. It is one we
pulled out to be parallel to the early --
MR. KRESS: And I am not sure that parallel is as
DR. POWERS: I don't know, Mary or Eric may want to
say more about that.
MR. KRESS: Yes.
MS. DROVIN: You have a different recommendation.
MR. KRESS: No, I haven't done the exercise either.
But I was starting from the latent death quantitative health
objective, and do something like you did to get a LERF and
show that the --
MR. HASKINS: Let me just say one thing. There are
source terms that can be, and have been, postulated and
purees, for late containment failures where you have core
concrete interactions going on in the absence and overlying
pool that can be some very bad source terms that could
actually result in fatalities beyond the ten miles zone.
Certainly, we would want to have a consistent goal for those
types of things. Now if the driving force turns out to be
lightened cancers then there may be a more appropriate
numerical goal. We simply put that down as a strawman at
MR. KING: I think it is a fair question. We ought
to think about it some more.
MR. KRESS: If it is a strawman, it is all right.
But I would want to see a technical basis for it based on
the thinking if you went to it alert.
MR. KING: Okay.
MR. APOSTOLAKIS: I think on page 3-8, you are
given some thoughts on the core damage of containment
failure that do not make sense there, but they make sense
after I read page 4-1, where you make a distinction between
the risk assessment perspective and the design bases
For example, on 3-8, there is a risk -- the
second paragraph on the left, "A risk significant level of
core damage found that is specified in the ACC as acceptance
criteria. ACC has accepted criteria to permit only 1
I am sitting here, you know, everything
I have read so far has something to do with risk. Now you
are throwing in these new ideas. But then when I read later
on the Safety Margins definition, and so on, that makes more
sense. So, editorially, I think some how you should make it
clear. Maybe shift this discussion to the next chapter,
statement of uncertainties because I think we are mixing the
And my second question is, isn't it
inconceivable that at some point that the whole frame work
that you have presented can naturally can be used to define
safety modules that was to move now from risk to design
bases? Seems to me that would make sense that you may need
additional guidelines for better ability and so on. I mean,
this thing of a one percent of a (inaudible) being allowed
to be oxidized, I don't know if that could change but that
is just an example.
And working backwards, one should be able to
do that; shouldn't one.
MR. KING: Like working backwards to what?
MR. APOSTOLAKIS: Backwards from the top ladders
that you have. Is that written down.
MR. KING: You mean to the actual QHOs themselves
and the --
MR. APOSTOLAKIS: The QHOs I go down to prevention
mitigation; go down to (inaudible) initiators; condition of
core damage probability.
MR. KING: And then see what kind of QHO you get
and compare that to the safety goal QHOs; is that what you
are talking about?
MR. APOSTOLAKIS: No, no, no. What I mean, as I
move down now, and I am going to the conditional containment
failure probability, and I can derive a number, given that
have core damage frequency of a initiator, cannot I say from
there, "Gee, the safety margin that I want in terms of
probabilities from my containment, is this." Rather than
come from the outside and give the safety margin and do just
I mean safety margins defined on the next
pages are probability that the designer process will
perform an intended function. Can I do that?
MR. KRESS: I don't think so. Because you are
dealing with means already and that is --
MR. APOSTOLAKIS: I can make contributions of
distributions. I mean civil engineers do that all the time,
don't they? I can play with the distribution of the
strength, distribution of the strengths. Some nice
mathematics and say, "This is the number." They are too
strong right now, aren't they?
MR. KRESS: Maybe. I have a different question on
that same page short. Similar to yours.
MR. APOSTOLAKIS: Which page, 3-8?
MR. KRESS: 3-8. Into my view the (inaudible) for
50.46, dealing with the ECCS acceptance criteria is the only
risk component I see to that is that the acceptance criteria
gives you a level of assurance and when you turn on the
ECCS, and when I say terminate the accident, and not lead to
an uncoolable geometry later on. So you have these safety
margins in there to deal with the fact of how well the ECCS
has to work. It is a statement of how good the ECCS has to
be in order to assure a coolable geometry that doesn't get
you in trouble later on. That to me is -- I don't see how
you risk inform that because you have to ask yourself, well,
what probability am I willing to live with that this doesn't
actually terminate the accident like I thought it was going
to. It doesn't lead to bowing and the swelling of the clad
and uncoolable geometries later on. It leads me back into
And I don't know how you deal with that in a
risk informed regulations, because what you have to have
there is some probability that you would live -- or
frequency or willing to live with on this thing and I don't
think we have defined any criteria like that in here.
MR. APOSTOLAKIS: You are absolutely right. I mean
it is always some conflict conceived.
MR. KRESS: So, it is a defense-in-depth concept to
me that I don't see how you can go in to say Appendix K and
K 50.46 and change anything, because you don't have a risk
informed way to change that.
MR. SIEBER: Well, I think you have to go beyond
that. 50.46, to my memory can avow as a settlement of a
lawsuit and it concerns scientist. The final acceptance
criteria came out and so, being that there is an element of
the legal basis in there, it is not clear that you now can
lateral say that I am going to risk inform 50.46, because
then the basis of the law suit settlement --
MR. KING: Jim, I am not sure about the legal
MR. KRESS: I think the voluntary aspect, you have
to get around the legal, but --
MR. BONOCA: It is very significant for them to
(inaudible) standard. Let me give you an example what is
significant. We have seen a number of power plants, PWRs,
recently, where they have found that the MPHS under certain
conditions was inadequate. The condition was literally and
open containment. Apathetic conditions, which means in the
limiting design conditions in which you design a
containment, you may have entrainment and you may have
essentially a problem with PSH. But for all other ranges, I
will see that coming through and risk inform regulation that
particular condition may be almost eliminated because how
did it get there.
MR. KING: You get there with failing containers.
MR. BONACA: So if you design for a more realistic
range of conditions, you will find that all containments had
in fact recirculation capability. See that is a fundamental
change and maybe is the right change, but I think without
implication, that absolutely, because now we have perform
PRAs or power plants which were designed with (inaudible)
material to meet certain limiting conditions that may not be
realistic. In the future, we will be using PRAs to evaluate
PRA designed plants, which therefore, have different success
criteria, maybe. Maybe that is what is going to be
reflected and what (inauible)but isn't this an important
MR. KING: Let me get back to your question on
50.46. I mean, it seems to me you can risk inform. You can
risk inform it in two ways. One is the large break LOCA
still makes sense and all the assumptions that go with it,
given probability and risk arguments. You can take a look at
it from that respect. You can also look at -- I mean, I
think clearly you want to maintain coolable geometry that is
the ultimate goal. If you have a LOCA you don't want to
loose coolable geometry. But is the 2200 degrees and the 17
percent oxidation are they the right numbers to do that.
Maybe they are very conservative.
Have you new information that says, no, there
are better numbers that will help you achieve that. You
know, maybe they'll be more restrictive, less restrictive.
To me that is also risk informing in the sense that you are
trying to take the best available. Maybe do some best
estimate calculations and then you got to figure out what
margin you want to account for uncertainties, but to me that
is part of misinforming the regulation. So I think you can
get into all aspects of 50.46. Either from probability risk
or, you know, best technology.
MR. KRESS: Yeah, I would have to agree with you. I
think those are risk informed concepts.
MR. KING: It won't be easy. Eric has been heavily
involved and working with Westinghouse.
MR. KRESS: You know, I kind of view the concept of
the large break LOCA as also adding margin, and I don't know
how you deal with that versus the margin you have in the
temperature, the peak clad temperature thing. Because I
view just requiring this thing to deal with the large break
LOCA as it sits there a margin concept.
MR. KING: It adds margin for LOCA but what does it
do for PTS?
MR. KRESS: Not very much.
MR. KING: It may make it worst. You have high
capacity pumps that shove a lot of cold water in,
particularly under a small break LOCA when you are still
pressurized, you know?
MR. KRESS: Yeah.
MR. KING: Maybe that is not a good idea.
MR. SIEBER: Small break LOCA --
DR. POWERS: I think this discussion on the
oxidation and the 17 percent raises a question, I am not
sure how you deal with it. And I am not sure how general
the situation is but the specifics of the 17 percent
oxidation and the use of baker adjust kenitics for
calculating that oxidation
And you say, gee, I want to do a more realistic
analysis here. Baker Just, I know how he got his numbers,
he threw balls in the water and things like that to get
those numbers, they are not applicable. I have got better
things, Cathcart, Paul, and a couple of guys at the PNL did
some better work. A couple of guys from Germany have done
better work on those kinetics. And I can go apply those and
if I do, I will get much lower oxidation levels.
But that presumes that the temperature history of
a LOCA is this very benign history which is a rampup to a
specified temperature and a hold for a certain period of
If, in fact, the temperatures, as they probably
do, go through a rampup to a peak, they drop back down and
then rise up to a hold point, then you put thermal stress,
create thermal stresses in that oxide so it spalls, you get
breakaway oxidation and it goes much, much faster than
Cathcart, Paul or Hobson, or any of those other people,
because they were looking at planchettes that were flat.
They didn't have the curvature problems of clads and things
All of these things that people understood when
they set up, when they were looking at 50.46 in setting it
up, and they said, gee, we will try to bound all those
effects by using Baker Just. And so, if you try to do
something more realistic there, then you have to take these
more realistic scenarios on the temperatures, and it gets
into a complicated nightmare that is very troublesome.
When you look for sections to risk-inform, do you
go back into that history far enough to know what kind of
cans of worms that you are opening up in looking at these
MR. HASKIN: There has already been some work done
on those sorts of replacements in the Appendix K
calculations, and the answer is, yeah, you have kind of got
to go back. Norm Lauben is the one that is doing that work,
and he has got enough historical perspective that he has an
appreciation for most of those things.
But you are right, the point is as you start
trying to make Appendix K calculations less conservative in
some respects, you have got to look at what the original
intent was, because you can't have your Appendix K
calculations becoming less conservative than what you would
get with the best estimate in 95 percent, for example. And
you can easily get yourself into that situation if you just
start relaxing Appendix K assumptions right and left.
So, he has got a procedure that he is going
through to make sure that that doesn't happen.
MS. DROUIN: If you take 50.44, for example, and
we had just said, let's just look at the rule and bring risk
insights, it would have been a much easier process to deal
with. But I don't think, in all fairness, you can do it
that way. I think you have to go back and look at the
technical basis. Why did it come around the way it did?
You know, what were all the assumptions and everything? And
that was the difficulty. I mean, so that when you finally
do get to here is your alternative for risk-informing this,
you know, you haven't dropped something, for example,
because of other reasons that were in there, that you just
had no idea about.
So, on all of these, be it 50.44, 50.46, part of
our process is going back and understanding the technical
DR. KRESS: But is that spelled out in the
framework document somehow, that that would be part of the
MR. KING: It is spelled out in the plan we sent
to the Commission before we even wrote the framework
document, that we have to go back in and look at the
technical basis. That was back in November last year.
MS. DROUIN: It was in the plan.
DR. KRESS: Now, I remember, you had to actually
add those words.
MS. DROUIN: And I don't know in the version you
have how well that was explained. In the May version of
Chapter 5 that is talked to very explicitly, that you have
to go back and look at that. You have to see how these are
tied to the other regulations. And maybe you did something
in this regulation, you know, you backed off on it because
it is covered elsewhere.
So that is very explicitly in the newer version of
the framework. I don't remember, in the version you have,
how well that was explained.
MR. KING: It is not an easy task. I mean in
50.44, it took so long precisely because of going back and
looking at all the places that hydrogen issues show up, and
looking at things like, well, was the analysis just for
in-vessel hydrogen generation or did it consider ex-vessel
hydrogen generation? And, you know, how long did the
generation take place, and all that, and it took a long
I mean I expected 50.44 to proceed much faster
than it did, but then Mary clued me in on what was going on.
And, you know, it is not an easy task.
DR. POWERS: You always bring back news.
MR. KING: Yeah.
DR. KRESS: Yeah, I see in Section 5, you actually
have a whole section on assessing the technical basis of the
regulations in relationship to other regulations. I guess
that is where it is dealt with.
CHAIRMAN APOSTOLAKIS: Okay. First of all, let me
understand how this works. Mr. Christie is supposed to
present his views at 4:00. Now, this is not going to be
done by 4:00. Should we interrupt, tell Mr. Christie make a
presentation and then we will come back to the document? Or
DR. KRESS: If they can accommodate that
arrangement, I think that would be the way to do it.
CHAIRMAN APOSTOLAKIS: Because a schedule is a
DR. KRESS: Yes.
CHAIRMAN APOSTOLAKIS: Can you accommodate that?
MS. DROUIN: I have to leave.
CHAIRMAN APOSTOLAKIS: What time? What time do
you have to leave, Mary?
MS. DROUIN: I have to leave by quarter of 5:00
CHAIRMAN APOSTOLAKIS: Okay. And Eric?
SPEAKER: I have got an 8:00 flight.
CHAIRMAN APOSTOLAKIS: Oh, 8:00 is fine.
SPEAKER: I am taking Metro.
CHAIRMAN APOSTOLAKIS: Do you think you guys can
handle it without Mary?
MR. KING: It will be tough. We will try.
MS. DROUIN: They can handle it. I am scared what
they are going to agree to without me that I am going to
have to live with.
DR. SEALE: Well, you can handle it.
MS. DROUIN: I can handle it.
DR. KRESS: This is our chance.
DR. POWERS: Now is your chance, Tom.
CHAIRMAN APOSTOLAKIS: Yeah, I don't expect it
will be much longer.
Mr. Christie, are you going to stick to your time
schedule, 15 minutes, or you will need more? It depends on
DR. KRESS: It depends on what we ask.
CHAIRMAN APOSTOLAKIS: Anyway, we will try to be
reasonable. Okay. We will try to be reasonable, but I
think we should start at 4:00 with Mr. Christie, because
that is the scheduled time.
So maybe we can go on for another five minutes and
then break for five, and then come back with Mr. Christie.
Anything on 4-1? Yeah, I do. The first paragraph
on the left says NUREG-1489 provides a more tutorial
discussion in terms of methods in uncertainty analysis.
This reference is from 1994. I would rather have you cite
Regulatory Guide 1.174, which reflects the more recent
thinking. There is a lot of discussion on uncertainties
there and how to handle them in the context of that
Then on the next page, 4-2, it says, for example,
compliance with the ECCS acceptance criteria of 10 CFR 50.46
can be demonstrated using best estimate codes provided that
uncertainty is accounted for. I don't think a best estimate
code means anything. How do you account for uncertainties
in a best estimate code? It seems to me you have to have a
quantitative statement of how accurate the model is.
DR. KRESS: I don't know if we have a good
definition of what a best estimate code is.
CHAIRMAN APOSTOLAKIS: There is no definition.
Yeah, I think the whole thing has to be revisited. Now, I
can see how, by having a distribution of the output of the
code, one defines acceptance criteria in such a way that in
the future, all you have to do is run the code with point
estimates, because you don't want to do uncertainties all
the time. Right? And then you compare with the acceptance
criteria which have now in them, built into them, the
uncertainties in the code.
MR. KING: But I think you are right, a best
estimate code, for it to work, you have got to have an idea
of what the uncertainty band about the estimate is.
CHAIRMAN APOSTOLAKIS: So, would you please
rephrase this to make sure that it is not offensive.
DR. BONACA: I have a question on this, however.
Doesn't the NRC right now allow for what they call a best
MR. KING: Yes. Yes.
CHAIRMAN APOSTOLAKIS: Yeah.
DR. BONACA: So there is a definition of it.
MR. KING: But you have to know the uncertainty
MR. HASKIN: There is a Reg. Guide that governs
how you do that.
CHAIRMAN APOSTOLAKIS: That does what?
MR. HASKIN: This is an option under Part 46.
DR. BONACA: There is a definition right now in
place for a best estimate LOCA now. Yes, there is.
MR. HASKIN: Yeah, there is a Reg. Guide that
governs what it is.
CHAIRMAN APOSTOLAKIS: What does it say? What
does it say? It says use best estimate codes.
MR. KING: But you have to quantify the
DR. BONACA: You have to quantify the uncertainty,
compare it to the Appendix K. You have -- it is a very
elaborate process. Now, how best estimate it is, I cannot
tell you, but --
MR. HASKIN: And including looking at the biases
in the code modeling. So, I mean there is a whole Reg.
Guide that deals with this.
CHAIRMAN APOSTOLAKIS: Okay.
DR. KRESS: On that same page, George, they have
this intriguing paragraph at the end of the second column,
"As in considering a change to the existing regulatory
requirements, it is important to estimate the overall impact
on the actual plant changes that would ensue."
I think that is not only important, I think it is
mandatory to do a good risk-informing job. But I don't know
how you are going to do that. And my question is, how do
you implement that requirement? Is this an iterative
process where you will make a rule and then go to the plants
and say, now what will you change based on this rule? Or
will you guys try to guesstimate what they will change?
MR. KING: We would have to make some estimate of
what the changes would be. I mean we have to have something
DR. KRESS: And then estimate what the effect is
MR. KING: Estimate what the effect is on risk.
DR. KRESS: But isn't that plant-specific?
Wouldn't you have to do it for each plant and see if, on the
aggregate, you still meet your goals? Yeah, it is just the
implementation of this, I am not sure how you are going to
go about doing it.
MR. KING: I mean it is no different than when we
modify a regulation today. We have to basically do the same
thing. Pick an estimate of what the change is risk is.
DR. KRESS: When you make a regulatory analysis.
MR. KING: Yeah, regulatory analysis based on what
we think the licensees are going to do to comply with that
DR. KRESS: So this would be similar to a normal
MR. KING: Similar to a regulatory analysis. We
don't look at 103 plants, we pick a few representative
DR. KRESS: Pick out, you will out enough of them
that you have got it covered.
CHAIRMAN APOSTOLAKIS: The righthand side, top of
the page, there is a sentence that I think is great. "To
the extent possible, revised and new deterministic
parameters will be based on probabilistic considerations."
This is really a very good statement. I think it should
have been stated much earlier as well.
And I think the statements that I have been
complaining about, you know, about not using probabilities
and risk measures in the thing, can be combined with this to
give a nice little paragraph that explains where you are
coming from, because this really is the essence of it.
MR. KING: Yeah, this is the main them of this
CHAIRMAN APOSTOLAKIS: Yeah, this is the essence.
But, again, revised or new deterministic
parameters, that does not exclude the possibility of having
some of these requirements use probabilistic language, like
unavailabilities, the maintenance rule and so on.
MR. KING: So if we are not risk-based, George?
CHAIRMAN APOSTOLAKIS: Huh?
MR. KING: As long as we are not risk-based.
CHAIRMAN APOSTOLAKIS: This is awful. Can you
MS. DROUIN: We now know what button to push.
CHAIRMAN APOSTOLAKIS: To be vigilant not to
become risk-based, because then -- I suggest that we stop
now for a few minutes. I don't think there are too many
comments after that, but I think we should take a short
break and then allow Mr. Christie to present his views.
I take it you don't have anything. It says here,
status of proposed revision to 10 CFR 50.44. You can do
MR. KING: We covered that at last subcommittee
meeting. We will cover it at the full committee tomorrow.
MS. DROUIN: Tomorrow.
MR. KING: Tomorrow.
CHAIRMAN APOSTOLAKIS: We will cover in the full
committee, yes, that is true.
Okay. So we are recess for seven minutes.
CHAIRMAN APOSTOLAKIS: We are back in session.
Mr. Robert Christie has the floor.
DR. KRESS: Would you introduce yourself so we
know who we are --
CHAIRMAN APOSTOLAKIS: I think Mr. Christie is
well known to the committee, and it is late in the day.
MR. CHRISTIE: My name is Bob Christie. I am the
owner of a firm, Knoxville, Tennessee consulting firm called
Performance Technology. I have been in the commercial
electric power business, nuclear and some other things, risk
and reliability related, for about 26 years. And I think
that today what I would like to start out with is a couple
of corrections administrative in nature.
One is, George, I am not talking about hydrogen
50.44 today, that is tomorrow.
CHAIRMAN APOSTOLAKIS: Okay.
MR. CHRISTIE: Okay. The second is there is,
again, brought up in the discussion today, this statement
that I am supposedly to have made that when we proposed
something in 50.44, we were talking about ultimate
containment capability. And I just want to go back again on
the record, during the meeting last week on the 29th, you
know, I will just read what I said.
I don't see anywhere in that proposed rulemaking
that says we are using the ultimate capability. We are just
saying that the large drives are going to check their
containment capability. We don't use ultimate capacity. We
just say for high probability events, check your containment
capability. Whether you use design, whether you use
ultimate, whatever you do, that is for the people at the
plant to decide how to do that, and they have done that in
their previous works.
The last thing I would like to check, the staff
has stated today that no one has objected to the framework
document, and that is definitely untrue. In the February
meeting that we had to talk about the framework document,
which isn't exactly the same as what we have had today, and
we had probably one of the most, as I stated in our March
1st meeting, the most contentious meetings I have ever been
to with respect to the interactions of the industry and the
So, there were violent objections to the framework
document in that meeting. We came to you and we went on the
record on the March 1st. We provided you documentation
that, you know, put down our objections to what the
framework document is. And then last, I guess, last week,
you know, you called me up and asked me, did you hear what
they said to me, and do you agree to that, you know, et
cetera? I said, no, I don't agree to that framework
So, it is not true that the industry has not
objected to the framework document. That is not true at
Okay. What I would like to talk to -- those are
just to clarify some other things. I would like to talk to
you today. I would like to, you know, just quick, five
subjects, adequate protection, the policy statement on
safety goals, the June 15th, 1990 staff requirements, goal
allocation and summary. And I don't think we have to spend
a lot of time on it. This is, again, the primary basis -- I
mean primary responsibility-wise with the people running it,
and the Nuclear Regulatory Commission is the adequate
protection, public health and safety.
The definition that is generally used, and we will
talk about that, if a nuclear power plant is in compliance
with the regulations, it is presumed that nuclear unit
provides adequate protection of public health and safety. I
think that is pretty well straightforward.
DR. KRESS: Except what do you mean by presumed?
MR. CHRISTIE: I have no idea. I am just quoting
from what the NRC puts in their paperwork.
DR. KRESS: Okay.
MR. CHRISTIE: I am just saying that is the
DR. KRESS: Okay.
MR. CHRISTIE: Okay. We have the 1986 Nuclear
Regulatory Commission policy statement on safety goals for
the operation of nuclear power plants. It consists of two
parts. Basically, the qualitative part having to do with
individual members of the public, and then the societal risk
to life and so. We have seen those before.
From the qualitative goals, we go down to what is
known as the quantitative health effects objectives. And
here we see the .1 percent rules. .1 percent rule means .1
percent of the background. For accident fatalities, that is
called the top fatality goal, and then we have got the
latent cancer fatalities, which, again, is .1 percent.
Where do the numbers come from after that? It is
pointed out in the staff presentation, you figure out the
numbers and so you get approximately 100,000 accidental
deaths per year in, you know, 200 million. So it comes to
one in 2,000. For latent cancers, it is 400,000 per 200
million, and it is one in 500. Okay.
I think, you know, that is pretty straightforward.
I want to talk to you now about this document that
has been referred to a couple of times, and I don't know
whether you have read it or not. I have read it millions of
times. But this is the -- you know, the document defines
how safe is safe enough in adequate protection. And the
first statement that is in here is that the Commission
agrees that it must not depart from or be seen as obscuring
the arguments made in court defending the backfit rule.
Okay. These arguments clearly establish that there is a
level of safety that is referred to as adequate protection.
This is a level that must be assured without regard to cost
and, thus, without invoking the procedures required by the
Beyond adequate protection, if the NRC decides to
consider enhancements to safety, costs must be considered
and a cost benefit analysis required by the backfit rule
must be performed.
The safety goals, on the other hand, are silent on
the issue of cost, but do provide a definition of how safe
is safe enough that should be seen as guidance on how far to
go when proposing safety enhancements, including those
considered under the backfit rule.
And on your point, Tom, where did we get the bit
about -- read the footnote on a related point. The
presumption is that compliance with our regulations provides
adequate protection. I don't know what presumption means.
I don't like the definition, but that is something else.
Okay. What the Commissioners, to me, clearly said
in that, that there is a level of adequate protection
somewhere, and there is also a thing called -- how safe is
safe enough? And the quantitative health objective is to
find, how safe is safe enough?
So, let's take the individual, which, you know,
everybody generally agrees is the most restrictive. And you
look at it and you take the background as one in 2,000, and
you go to .1 percent, and it comes out to be one in 2
million. And below that point, the NRC is not to impose
requirements even if cost beneficial. That is pretty clear
to me in the safety goal policy statement.
Okay. Now, they said somewhere above that line
lines something called adequate protection. Okay. And this
is my curve from my Delphi Process, where I asked everybody
that I knew, including the members of the ACRS, to
contribute to the Delphi Process to see if we could get this
curve defined with adequate protection. And as I told you
before, we didn't get a heck of a lot of response. No one
from the NRC, including no member of the ACRS, responded.
We got a limited from the industry, and it turns out that
most of the people are coming in between 1 and 10 percent,
with kind of a mean, you know, somewhere 3 to 5, as I have
told you before.
So, now, if that were true, this would be the
curve that would be used to define adequate protection.
Above that line, the NRC imposes requirements without regard
to cost, between the adequate protection line and this "how
safe is safe enough?" that we would be using the 10 CFR
50.109. Why do I have 10 CFR 10.109 on there? Sorry about
that. I just caught that.
And in there, we would be using the 200 person-rem
conversion factor based on the $3 million value for health
DR. KRESS: So your Delphi Process would say
currently, today, adequate protection is like a factor of 10
above the safety goals?
MR. CHRISTIE: I would say 30 to 50, but a factor
of 10 for sure.
DR. KRESS: Just go out to the --
MR. CHRISTIE: Three to 5 percent versus .1.
Adequate protection over here is about 3 to 5 percent.
CHAIRMAN APOSTOLAKIS: Of what?
MR. CHRISTIE: Of background. Which is an order
of 30 to 50 more than .1 percent.
DR. KRESS: But this was just a Delphi Process.
MR. CHRISTIE: Right.
DR. KRESS: So it is --
MR. CHRISTIE: You know, again, I have had a
conversation with Dr. Joe Murphy about this, and Joe says
people don't think in terms of 3 to 5 percent, they think --
DR. KRESS: In terms of 10.
MR. CHRISTIE: Ten.
DR. KRESS: Ten.
MR. CHRISTIE: So, you know, what he recommends is
we go -- well, he doesn't recommend, but he said it would be
more logical if they used just 1 percent.
DR. KRESS: Yeah.
CHAIRMAN APOSTOLAKIS: For adequate protection.
MR. CHRISTIE: For adequate protection.
DR. KRESS: That would be the factor of 10.
CHAIRMAN APOSTOLAKIS: Of what, 1 percent of the
MR. CHRISTIE: 1 percent of background.
DR. KRESS: It is a factor of 10.
MR. CHRISTIE: Everything is --
CHAIRMAN APOSTOLAKIS: 10 to the minus 6 then.
DR. KRESS: Yeah.
CHAIRMAN APOSTOLAKIS: No.
MR. CHRISTIE: It would be if you used 1 percent.
For me, it would be 5 times to the minus per year.
CHAIRMAN APOSTOLAKIS: Right. So, an order of
magnitude above the safe enough.
MR. CHRISTIE: Right.
CHAIRMAN APOSTOLAKIS: Yeah.
MR. CHRISTIE: And in that region between 1
percent and .1 percent, we would be using the backfit rule.
CHAIRMAN APOSTOLAKIS: Well, this is a three
region approach we have asked for several times, right?
DR. KRESS: That is exactly what it is.
CHAIRMAN APOSTOLAKIS: And the staff and the
industry are against it.
MR. CHRISTIE: The staff and what industry? I am
for it. I know that the next head of the ANS is for it.
CHAIRMAN APOSTOLAKIS: NEI was against it.
MR. CHRISTIE: Well, I am just telling you that
there are -- you know, industry is not represented entirely
DR. KRESS: The staff actually uses this approach.
CHAIRMAN APOSTOLAKIS: This is like risk-based.
MR. KING: This is, in effect, what the staff
uses. We are not against it.
DR. KRESS: This is, in effect, what the staff
CHAIRMAN APOSTOLAKIS: I'm sorry?
MR. KING: This is, in effect, what the staff
uses, a three region approach.
DR. KRESS: I mean whether it is spelled out or
not, that is what they use.
MR. KING: I mean we just haven't --
CHAIRMAN APOSTOLAKIS: That is what we said in our
letter, Tom, that, in effect, people are using it.
MR. KING: Exactly. We just haven't put a number
on adequate protection.
DR. KRESS: Right.
CHAIRMAN APOSTOLAKIS: And, of course, we did it
in the context of CDF and you might object to doing that for
the CDF. I know you --
MR. CHRISTIE: Depending on which CDF is chosen.
CHAIRMAN APOSTOLAKIS: You like the risk?
MR. CHRISTIE: We will get to that point in a
CHAIRMAN APOSTOLAKIS: Oh, I never doubted that,
MR. CHRISTIE: Okay. And you have heard me
before. The best requirements are those that define a
criteria to be met, do not specify how to meet. The
criteria should lead to a comprehensive approach to the
whole plant, defining the overall criteria is better than
defining a set of lower criteria.
All right. And then, again, we get to the
problem. The public health risk is different for each unit,
and it changes with time. All right.
DR. KRESS: What happens if you cure all the
MR. CHRISTIE: We have been through that last --
two years ago, Tom. If you cure all the cancers, the latent
doesn't even appear anymore and we are down to the
DR. KRESS: No, no, the latent then becomes
MR. CHRISTIE: No, the latents is zero. You can't
have a latent cancer fatality anymore.
DR. KRESS: Yeah, but you have got --
MR. CHRISTIE: Zero is zero, Tom.
DR. KRESS: -- .1 percent of zero that you meet
MR. CHRISTIE: No, Tom, zero is zero. 1 percent
of zero is zero. 100 percent of zero is zero.
DR. KRESS: It is hard to meet zero is what I am
MR. CHRISTIE: But if they cured cancer, then we
wouldn't have latents.
Okay. Now, the staff is in here.
CHAIRMAN APOSTOLAKIS: Yes.
MR. CHRISTIE: And this is what we, you know,
again, we object to this pretty vehemently. I mean it is
such a complicated affair. Just think about writing your
PRA three times now. You are going to run your PRA for --
what do they call them? -- anticipated initiation. You are
group all the ones that are, you know, somewhere between 10
to the minus 2 and 1, and you are going to run it again
between all the ones that 10 to the minus 5. You can do all
you can all your conditions, containment capabilities.
DR. POWERS: I guess I am lost. Why do you have
to run them multiple times?
MR. CHRISTIE: How are you going to separate out
the transients that are between 10 to the minus 2 and 1? If
you just run it once, all of them are lumped together and
all your event trees, et cetera, et cetera.
DR. POWERS: No, they are not. No, not my PRAs,
they aren't. All spit out nice and separately.
DR. KRESS: And when you say --
MR. CHRISTIE: No.
DR. KRESS: When you say you are going to have to
do this, do you think this is going to be the licensees
having to do this?
MR. CHRISTIE: Absolutely.
DR. KRESS: Oh, I thought NRC was going to do
MR. CHRISTIE: Well, what code is it is going to
do, code is the NRC going to use to do this for a
plant-specific PRA that varies individually from plant to
DR. KRESS: The same ones that they use now for
regulatory analyses is what I was told.
MR. CHRISTIE: Right. Right. And do you think
they are going to get accurate results?
DR. KRESS: I don't know. Accurate enough for
making regulations maybe.
MR. CHRISTIE: Anyway, this is a very --
CHAIRMAN APOSTOLAKIS: Your argument, Bob, is that
what really matters is the bottom box, quantitative health
objectives, is that your argument?
MR. CHRISTIE: No. What matters to me is adequate
protection of public health and safety. That is my limit.
CHAIRMAN APOSTOLAKIS: I thought you are objecting
then -- this is a clarification. You are objecting to two
things that the staff is doing. First, they are going
deeper than just the quantitative health objectives. You
have always argued that that is the only thing that matters.
MR. CHRISTIE: No. I have never argued it is the
only thing that matters. I have always argued --
CHAIRMAN APOSTOLAKIS: You have argued this for
MR. CHRISTIE: I have always argued that adequate
protection of public health and safety is the standard to
which plants are licensed and the thing that counted. And
if you want to go below that, then you can go down to the
At the time I was saying it might be easier if all
we did was demonstrate we are below the quantitative health
objectives, because then we wouldn't even worry about the
backfit rule. But the standard that I am held to today, at
every license in the in the United States, is adequate
protection of public health and safety.
CHAIRMAN APOSTOLAKIS: So you are -- I mean that
is why I raised the issue three hours ago. Would the fact
that we are using the goals create any problems?
DR. SHACK: You have to meet the regulations.
Now, the regulations give you adequate protection, but
nobody pretends that they are based solely on adequate
protection. They clearly include safety enhancements that
go beyond adequate protection.
MR. CHRISTIE: If you use the backfit rule.
DR. SHACK: Yeah.
MR. CHRISTIE: Right. That is the scheme that all
of us are licensed to, and the scheme that is embedded in
the law. And if you want to change that scheme, you have
got to go back and change the law and go through the
Congress and go through the courts.
CHAIRMAN APOSTOLAKIS: I'm confused.
MR. CHRISTIE: That's the way life is, George, in
licensing a nuclear power plant.
CHAIRMAN APOSTOLAKIS: I understand that. I am
trying to understand your objection to what the Staff is
MR. CHRISTIE: I am objecting that the standard
that they are now setting to us is how safe is safe enough,
which to me is -- not only they are not setting it as how
safe is safe enough, they are even setting it below.
They are not even going to write rules for
adequate protection in a risk-informed rule.
CHAIRMAN APOSTOLAKIS: So they are -- you are
MR. CHRISTIE: They are changing the Atomic Energy
CHAIRMAN APOSTOLAKIS: You are objecting to their
use of the goals to write regulations without looking at the
MR. CHRISTIE: Right. Absolutely.
CHAIRMAN APOSTOLAKIS: Which is something we
The second one, which you surprise me today, I
thought you have always argued that what really matters is
ultimate public health and safety -- fine, and not CDF and
conditional core damage and conditional --
MR. CHRISTIE: Absolutely.
CHAIRMAN APOSTOLAKIS: So you are also objecting
in terms of that figure to all the boxes they have above
the -- that talk about prevention versus mitigation?
MR. CHRISTIE: Absolutely.
CHAIRMAN APOSTOLAKIS: Well, now I understand.
MR. CHRISTIE: You have got your letter from Dr.
Seale to Chairman Jackson, you know -- we all know that. If
you choose a core damage frequency of 10 to the minus 4, you
are more conservative than the quantitative health
CHAIRMAN APOSTOLAKIS: But the Staff told us today
that the Commission has blessed this.
MR. CHRISTIE: Well, I will talk about that in a
minute. As a matter of fact, here's the next slide.
The Staff says in the June 15th memorandum from
the Commissioners to Staff that they gave them permission to
use the 10 to the minus 4, and let's start looking at what
they say in that memorandum.
It says implementation and safety goals may
require development and use of, quote, "partition
objectives. In general the additional objectives should not
introduce additional conservatism. The Staff should bring
its recommendation on use of each subset of the area
objectives to the Commission in the context of the specific
issue for which it will be useful and appropriate and
explain its compatibility with the safety goals."
Then they go on to say, "For the purposes of
implementation the Staff may establish subsidiary
quantitative core damage frequency and containment
performance objectives through partitioning of a large
release guideline. These subsidiary objectives should anchor
or provide guidance on minimum acceptance criteria for
prevention and mitigation and thus assure an appropriate
multibarrier defense-in-depth balance in design. Such
subsidiary objectives should be consistent with the large
release guideline and not introduce additional conservatisms
so as to create de facto new large release guideline."
And as we know, in the safety goals they couldn't
even come up with the large release guideline.
They said a core damage probability of less than
one in 10,000 per year reactor appears to be a very useful
subsidiary benchmark in making judgment about that portion
of the regulations which are directed toward accident
Okay? Now how are you going to get that and still
be, quote, "not introduce some additional conservatisms" is
the dilemma that I have with a problem with.
CHAIRMAN APOSTOLAKIS: Would you say these SRMS is
MR. CHRISTIE: I would say that statement is not
DR. KRESS: Do you know what core damage frequency
the current regulations were designed on an average to
MR. CHRISTIE: I have no idea -- it would depend
on each individual plant would have its own specific core
damage frequency that would meet the regulation.
DR. KRESS: Well, and that could vary. On the
average, this might be the number they were shooting for
with the current regulations and still use it --
MR. CHRISTIE: Could be.
DR. KRESS: -- and might not add any more
MR. CHRISTIE: Could be, but if I again look at
the fact that I am working off the QHOs and a 10 to the
minus 4 QHO -- I mean 10 to the minus 4 core damage
frequency drives me down below the quantitative health
objectives, then obviously this is inadequate protection
because adequate protection according to this document is
above the quantitative health objective.
DR. KRESS: If it drives every plant down below,
then maybe on the average.
MR. CHRISTIE: So in the average it is all right
to have half of the plants get more restrictive, to pick the
10 to the minus 4 as the core damage frequency --
DR. KRESS: Yes, that's what you mean by that.
MR. CHRISTIE: It's great that half of the plants
don't meet that -- couldn't meet the QHOs, don't meet the
DR. KRESS: Right. I think that's what you mean
MR. CHRISTIE: I'm sorry. I'm not going to have
half of the plants in the United States not meeting a
subsidiary objective --
DR. KRESS: So you would rather have a fixed
number that they all have to meet?
MR. CHRISTIE: I would like for them to meet
adequate protection, which I hope some day to be able to
define in terms of background. I would like for them, below
that, to go with the backfit rule, and I would like for them
not to have to meet any NRC regulations --
CHAIRMAN APOSTOLAKIS: What you are asking, Bob,
is really for the Commission to define adequate protection
in terms of some frequency number, some risk number.
MR. CHRISTIE: It would solve a lot of problems.
But, you know, again, that's their -- now go look at the
next statement however. The Commission has no objection to
the use of a 0.1 containment conditional failure probability
for the evolutionary design -- for the evolutionary design.
Now there's some thought that when they put this
in the same paragraph and partitioning that they also meant
the core damage probability of less than one in 10,000 per
year to be for the evolutionary designs also, because the
Commissioners in 19 --
CHAIRMAN APOSTOLAKIS: 1990 --
MR. CHRISTIE: -90, when they were writing this
knew that the existing plants didn't meet the 10 to the
CHAIRMAN APOSTOLAKIS: And they were not aware of
Rick Sherry's work I don't think at the time.
DR. SEALE: But they were aware of the criteria
that were in the utility requirements document for the
evolutionary designs that talked about 10 to the minus 5th.
CHAIRMAN APOSTOLAKIS: That's correct.
MR. CHRISTIE: Again, I am just going off what is
stated here in the document.
CHAIRMAN APOSTOLAKIS: So what is your point from
MR. CHRISTIE: I just want one more slide on the
large early release frequency because I want to --
CHAIRMAN APOSTOLAKIS: By the way, just as a
clarification, Tom, when you told us earlier that an SRM
would confirm the 10 to the minus 4, you are referring to
MR. KING: Yes.
CHAIRMAN APOSTOLAKIS: Okay, thank you.
MR. CHRISTIE: Okay -- and Tom Kress and I will go
over it again.
On the large early release frequency of 10 to the
minus 5, you got the statement in the SECY that says, hey,
we took the 1150 plants, we raised them all up to a certain
level, we added a whole bunch of stuff up, and we picked the
10 to the minus 5 on the basis of those that are closest to
the QHOs, okay? -- and Tom said, no, that's not exactly what
we did is we drew a line, a kind of median, and so basically
half of the plants if we choose 10 to the minus 5, choosing
10 to the minus 5 is more restrictive for half of the plants
and maybe not that restrictive for the other half to meet
DR. KRESS: I don't think either of those is what
has actually happened.
MR. CHRISTIE: Do you believe that there are
plants in the United States who have a large early release
frequency in your definition let's say of a 10 to the minus
4 and still meet the QHOs?
DR. KRESS: 10 to the minus 4 and still meet the
QHOs? There may be some limited sites --
MR. CHRISTIE: So some plants in the United States
could have a large early release in 10 to the minus 4 and
still meet the quantitative health objectives?
DR. KRESS: It's possible, depending on the sites.
MR. CHRISTIE: Okay, so again picking a 10 to the
minus 5 means that there's some plants in the United States
that's more restrictive than the quantitative health
DR. KRESS: Probably yes.
MR. CHRISTIE: Okay. That's what I think too --
CHAIRMAN APOSTOLAKIS: Let's look at summaries.
Let's look at the summaries.
MR. CHRISTIE: All right, summary. And here's --
so we know where we are at.
Staff is proposing requirements for risk-informed
regulations for the existing nuclear units far beyond the
standard of adequate protection, okay? They're saying this
is voluntary, so if we stick in the existing deterministic
rule the standard is adequate protection. If we go over
into the probabilistic world, the standard is how safe is
safe enough. Oh, boy.
Number two, the Staff is ignoring the requirement
of the backfit rule when the Staff wishes to go beyond
adequate protection. The Staff is ignoring the express wish
of the NRC Commissioners in the thing where they said not
only said use it, but don't give any appearance of not using
Three -- the Staff is proposing subsidiary
quantitative objectives for existing plants that the NRC
Commissioners said were applicable to the evolutionary
plants -- to .1 for containment conditional probability is
for evolutionary plants, new plants.
The framework they are using today uses .1.
CHAIRMAN APOSTOLAKIS: But not the 10 to the minus
4, as Dr. Seale pointed out. They were aware of the 10 to
the minus 5 --
MR. CHRISTIE: I don't know what that 10 to the
minus 4 meant.
CHAIRMAN APOSTOLAKIS: I think it was an allowance
for the existing plants.
MR. CHRISTIE: But at least to .1 is for the
evolutionary plants and now the Staff is applying it to
CHAIRMAN APOSTOLAKIS: I must say that I am really
troubled by the first bullet -- I mean the first comment
there, as I stated earlier today.
I think we are using quantitative health
objectives that were meant to indicate safe enough without
using them in option three, which is intended to be for
I think Bob is right.
DR. KRESS: I don't think so, George. I think
we're mixing up the concept of requiring individual plants
to meet their goal versus writing the regulations so that on
the average they can meet the goal.
CHAIRMAN APOSTOLAKIS: But the individual
regulations will be imposed upon the individual plants --
DR. KRESS: As they are now, and I think they had
the same goals in mind when they wrote the --
CHAIRMAN APOSTOLAKIS: Well, that's where we don't
know. We don't what the goal was --
DR. KRESS: Now we don't know.
CHAIRMAN APOSTOLAKIS: -- when they were writing
something in 1973.
DR. KRESS: I know we don't know that.
DR. SHACK: Any new regulation will have to meet
the backfit rule.
DR. KRESS: That's right.
CHAIRMAN APOSTOLAKIS: I don't understand that.
If you use option three, it won't.
DR. KRESS: They will probably do the backfit
anyway, even if it's volunteer.
CHAIRMAN APOSTOLAKIS: Wait a minute, wait a
minute. If I use option three, and it is finished, okay? --
and the NRC writes a regulation --
DR. KRESS: Right.
CHAIRMAN APOSTOLAKIS: -- then they will have to
use the backfit rule to justify it?
DR. KRESS: They won't have to, but they will
because it's sort of standard procedure nowadays.
CHAIRMAN APOSTOLAKIS: What do you mean, they
don't have to but they will?
MR. KING: That is one of the issues that we are
going to raise to the Commission.
There is a line of thought that when we propose an
alternative, a risk-informed alternative, if that
risk-informed alternative adds additional requirements there
ought to be a backfit analysis to show that those would pass
the backfit test.
MR. CHRISTIE: But you don't have to ask for the
policy statement. The policy statement already exists --
June 15, 1999 the Commissioners told you that in the
implementation of the safety goals you use the backfit rule.
Don't even be perceived as not using the backfit rule.
CHAIRMAN APOSTOLAKIS: I am afraid we are
ratcheting up the regulations if we use the goals.
DR. KRESS: I don't think we know that.
DR. BONACA: I don't think we do.
CHAIRMAN APOSTOLAKIS: You don't think we do what?
DR. KRESS: Look at the plants now. Where do they
fall as a spectrum of plants on a safety goal plot? I don't
know where they fall, but I'll bet you most of them meet the
safety goals and that is because --
CHAIRMAN APOSTOLAKIS: Not according to the
gentleman to my left.
DR. KRESS: Well, you know, we can debate this
issue, but I would say --
CHAIRMAN APOSTOLAKIS: Well, let's take the
subsidiary goals. You know they don't.
DR. KRESS: Most of them probably do.
CHAIRMAN APOSTOLAKIS: Most of them do but --
DR. KRESS: That is why you write the regulations
so that the average meets them and if most of them meet it,
well, the average meets it.
CHAIRMAN APOSTOLAKIS: I am very uncomfortable
with that concept, that on the average if I look at the
population they meet it, because all they need is one
DR. KRESS: I've always had --
CHAIRMAN APOSTOLAKIS: An accident on the
DR. KRESS: I have always had a problem with that.
CHAIRMAN APOSTOLAKIS: That bothers me.
DR. KRESS: I have always had a problem with that
CHAIRMAN APOSTOLAKIS: Yes.
DR. KRESS: But that is the way the system is said
to be designed. I have had a problem --
CHAIRMAN APOSTOLAKIS: We're supposed to fix the
DR. KRESS: Okay, but that is a different issue.
CHAIRMAN APOSTOLAKIS: I don't understand that.
When we issue a particular rule --
DR. KRESS: To fix that issue you have to go to
saying each plant has to be required to meet a particular
CDF or a particular LERF, and the Commission has just over
and over said they are not going to do that, so, you know, I
would like to change that, and I have fought to change that,
but we are not getting very far with it.
CHAIRMAN APOSTOLAKIS: But that doesn't mean that
I can develop a new approach to Part 50 that has some
fundamental if not flaws, fundamental questions embedded in
DR. KRESS: I think there are questions but I
think we disagree on whether or not this is actually a
ratchet or not.
CHAIRMAN APOSTOLAKIS: I am willing to be
convinced otherwise, guys, but right now I think we are
DR. KRESS: Well, the question of the ratchet is
what does the current set of plants meet in terms of the
How many of them meet it? How many of them don't?
I think you would say then with the new
regulations are we requiring more of them to meet it or are
you still going to get the same distribution?
I don't know the answer to these questions but I
have no --
DR. BONACA: I agree with Tom, but the other
thing, adequate protection right now means that you meet the
regulation, the regulation that we have in the books.
CHAIRMAN APOSTOLAKIS: That is not what it means.
The Commissioner makes it very clear.
If you meet the regulations, there is a
presumption of adequate protection. If there is adequate
protection, that does not necessarily mean you meet all the
regulations, and it is not necessary and sufficient. It is
necessary, not sufficient -- which leaves open a huge hole
for us to justify, you know, a lot of things.
MR. CHRISTIE: No, I do not believe it leaves a
huge hole to justify other things except by use of the
backfit rule. It is not a huge hole. It is a very
DR. BONACA: I agree with that. I agree.
MR. CHRISTIE: So all we have to do from now on is
everything that we want to add to the thing say it meets the
backfit rule, go through the 51.09 analysis and make sure it
meets the backfit rule.
DR. BONACA: The question I have is if we did not
use the goals but we used some definition of adequate
protection, how would we go about risk-informing Part 50? I
mean has anybody got any idea how to do that?
DR. KRESS: You would have to explicitly spell
DR. BONACA: Exactly.
DR. KRESS: So since we can't, we have to use
something different, because we can't explicitly spell those
CHAIRMAN APOSTOLAKIS: But you see --
DR. BONACA: We wrote a letter, right?
CHAIRMAN APOSTOLAKIS: But is it inconceivable
that as an agency we'll want to have it both ways?
We don't want to define adequate protection
quantitatively and yet we want to risk-inform the
DR. BONACA: But the point though that I would
like to make is that, first of all, we had a regulation in
place before there was any definition or goals,
Then we began to talk about goals and, you know,
these goals these members here, they were the AIF, American
International Forum put them forth in the early '80s, late
'70s, and now there has been always an intent of filling the
gap with regulation, with the use of PRA.
How have we done that? By looking at CDF and LERF
and in cases we found there was a gap in the regulation,
that there was no adequate protection on some plants, there
were some changes made in the research of vulnerabilities.
There was a process that ended up with the IPEs and so on
and so forth.
So I don't think this concept here presented by
NRC is inconsistent with all these developments.
CHAIRMAN APOSTOLAKIS: It is inconsistent.
MR. CHRISTIE: George, let me --
CHAIRMAN APOSTOLAKIS: 1.174 -- it refers to
individual plants and says if you're changing CDF from what
you have now, which presumably satisfies adequate protection
criteria, is this and that, we approve or we don't approve.
This is a very different approach. This starts
with the QHOs.
DR. KRESS: It ought to be. It ought to be --
CHAIRMAN APOSTOLAKIS: It doesn't start with the
existing CDF at this plant.
DR. KRESS: George, it ought to be a different
approach -- 1.174 can be viewed as a formal way to do
CHAIRMAN APOSTOLAKIS: Changes.
DR. BONACA: It has to be plant-specific.
DR. KRESS: And this is not exemptions. This is
rewriting the rules and they should not be -- the two of
them shouldn't be thought of except in general principles,
you know, the principles are applicable in both cases but
they are not the same thing.
CHAIRMAN APOSTOLAKIS: My point is that in 1.174
precisely because you are talking about changes the issue
did not arise, because you are not relying fundamentally on
the QHOs. You are saying we are allowing this guy to
operate, therefore there is a presumption of adequate
protection. Now he wants to change it a little bit and here
are some rules.
Here we are taking a very different approach.
MR. KING: No.
CHAIRMAN APOSTOLAKIS: We are not looking at what
the guy is doing. We are starting with the QHOs.
MR. KING: 1.174 started with the QHOs and backed
CHAIRMAN APOSTOLAKIS: No.
DR. KRESS: They did.
MR. KING: LERF, 10 to the minus 5th LERF number
shows up in 1.174 as well as the 10 to the minus 4 CDF as
cutoffs as to where --
CHAIRMAN APOSTOLAKIS: No, but there you are
allowing to go higher. Come on.
MR. KING: But the changes have to be, you know,
so small that they are almost negligible but there is a
cut-off in there and it is based upon the safety goals. It
is not based upon adequate protection.
CHAIRMAN APOSTOLAKIS: You look at the figures you
have. They go to the right and then there is a zig-zag line
that says, you know, we really don't know where this is, but
that is exactly the adequate protection, that you are
approaching now regions where even a small delta is not
DR. KRESS: That is because you are dealing with
individual plants, not the body of regulations.
DR. BONACA: Exactly. That is the point -- 1.174
applies to the individual plants.
DR. KRESS: If you are going to risk-inform the
regulations, in my opinion you are going to have to have in
mind some risk objectives that you are trying to achieve.
That is the only way you can risk-inform it.
Now if I were going to use as my risk objectives
this factor of 10 above the safety goals that we saw on Bob
Christie's plants, I would be in real trouble, I think. I
think that would be real mistake to try to risk-inform the
DR. SHACK: And who would be -- the meaning of the
safety goal at that point?
DR. KRESS: Nobody.
CHAIRMAN APOSTOLAKIS: No, the goal I think was
stated very well. It is if you are below the goal, we leave
you alone. We don't even raise the issue of backfit.
DR. KRESS: That is when you deal with specific
MR. CHRISTIE: Agreed.
CHAIRMAN APOSTOLAKIS: That is exactly what it was
always supposed to be.
DR. KRESS: But that's when you deal with specific
MR. CHRISTIE: And that was the intent of the
CHAIRMAN APOSTOLAKIS: I don't think anybody is
MR. CHRISTIE: Okay. Let me explain. We do not
need this framework to risk-inform the regulations.
CHAIRMAN APOSTOLAKIS: Why not?
MR. CHRISTIE: Well, because we have another
framework which we will talk about again tomorrow, which I
have already explained to you on the hydrogen, where the
framework is. You go through the existing regulations. You
retain what is effective and efficient in addressing public
health risk. You add what is necessary that comes out of
your risk assessments and you delete what is not effective
CHAIRMAN APOSTOLAKIS: Well, then the Staff might
tell you that we are developing all this to help us decide
what is necessary.
MR. CHRISTIE: Well, they can have a framework
where half the plants in the United States don't meet the
regulations, which to me is just a debacle of the nth order.
It's almost ludicrous when you think about it.
Do I want to be an in the half of the plants that
don't meet it and go out there and say, hey, I don't meet
the regulations but it's all right?
DR. KRESS: Well, there's going to be two sets of
regulations and you'll have to meet one or the other.
I think that is a problem we have had but I don't
see any way around that.
MR. CHRISTIE: Okay. We are going to have a set
of regulations where the plants that can meet all the
quantitative health objectives and treat quantitative health
objectives and even below the quantitative health objectives
as adequate protection and meet all those are going to be
over here in one set of rules and then we are going to have
another set which can't meet all the quantitative health
objectives as adequate protection and the subsidiaries, et
cetera, and they are going to be up at another level?
DR. KRESS: Probably. That is probably what is
going to happen.
CHAIRMAN APOSTOLAKIS: That is another thing that
makes me very uncomfortable, Tom.
DR. KRESS: I know, but --
CHAIRMAN APOSTOLAKIS: And you know very well -- I
think all of us know that a lot of licensees will start to
pick and choose.
DR. KRESS: Of course they will.
CHAIRMAN APOSTOLAKIS: Then what happens?
DR. KRESS: That is the issue of selective
implementation, that we have got a policy to deal with, and
we even wrote a letter once that we think you are going to
end up with a dual set of regulations and don't see any way
MR. CHRISTIE: Well, I see a lot of ways around
There's no doubt in my mind I see a lot of ways
around it. I can risk-inform the regulations for everyone
using just exactly the framework that we just explained.
I don't think it is the most optimum but it
certainly can be done.
DR. KRESS: Well, you would make a new set of
regulations that would be mandatory for everybody --
MR. CHRISTIE: Absolutely.
DR. KRESS: -- a different framework.
MR. CHRISTIE: The proposed petition for 50.44 is
a mandatory application for petition for everyone. It
doesn't set -- and that is another problem we will talk
about tomorrow. It's got nothing to do with Option 3.
DR. KRESS: That is certainly another option. We
debated among ourselves some about that option.
CHAIRMAN APOSTOLAKIS: Let me understand something
in the context of the framework that was presented.
Let's say that in one of the entries for
infrequent initiators one of the facilities happens to be
higher than the numbers we have here and we say it's a goal,
it doesn't matter.
Then Quad Cities happens. Right or wrong,
somebody comes up with a sequence that is really violating
the goals, the numbers that the Staff has shown.
Where does the Staff start flying people over
there to find out what is happening and in fact order a
shutdown or the licensee itself shuts down the plant because
they feel they have entered the region now where we are
talking about adequate protection and all this remains with
this framework obscure.
DR. KRESS: I think they make a judgment call,
just like they do now.
CHAIRMAN APOSTOLAKIS: That's right, so this is
not helping that way.
DR. KRESS: It is not helping their part. They
will have to make a judgment call just like they do now.
CHAIRMAN APOSTOLAKIS: I think what makes it worse
is that you are allowing a plant to be higher than some of
these numbers but we don't know how high they are allowed to
be before we reach adequate protection.
DR. KRESS: Well, you and I know.
CHAIRMAN APOSTOLAKIS: I know.
DR. KRESS: I think the Staff knows.
CHAIRMAN APOSTOLAKIS: I think everybody agrees
that if you are in the 10 to the minus 3 for the reactor
core damage frequency, I mean the licensee itself proved
that they believe that's too high. They shut down.
I don't know why we have this great reluctance
MR. CHRISTIE: Are they shutting down because of
public health risk --
CHAIRMAN APOSTOLAKIS: No.
MR. CHRISTIE: -- that they can't survive it, or
are they shutting down because the investment risk is too
high to lose $4 billion?
CHAIRMAN APOSTOLAKIS: I think it is really a
combination, and you can add to that one probably the
reaction from the NRC and so on, but the truth of the matter
is that it seems to be consensus that a 10 to the minus 3 or
higher core damage frequency is something that we have to
act immediately upon.
DR. KRESS: Yes.
CHAIRMAN APOSTOLAKIS: I don't think you can
dispute that. Now you can argue about the reasons but this
is the truth.
DR. KRESS: So why don't we codify --
CHAIRMAN APOSTOLAKIS: Why don't we recognize
DR. KRESS: Why don't we codify that --
CHAIRMAN APOSTOLAKIS: And yet the Commission
says, the Commission disapproved the proposed change to
elevate the qualitative statement of the prevention --
qualitative even -- of severe core damage accidents to a
qualitative safety goal. The Commission just flat out
disapproved it and now we can't do anything about it.
MR. CHRISTIE: I don't think -- you are just
focusing on the core damage frequency, which again is not
the reason for the Nuclear Regulatory Commission's
The reason for the Nuclear Regulatory Commission's
existence is the public health effects. The radiation that
is contained in the fission products in the core have a
significant impact during accidents, if we have accidents,
on the public health risk, and it is the role of the Nuclear
CHAIRMAN APOSTOLAKIS: I wonder, Bob, if there was
a core damage incident tomorrow and the New York Times and
Washington Post were after the Commission whether you would
stand up and say, hey, nobody was killed. Why are you
putting them on the hot seat?
MR. CHRISTIE: Absolutely. I would stand up there
without even blinking.
CHAIRMAN APOSTOLAKIS: I don't think the reporters
would listen to you.
MR. CHRISTIE: Again, you are predicating
everything on the front page of the New York Times and the
Washington Post and I know that the Staff does it also,
because I sat in on the meetings and they told me so, and
that is not an appropriate measure for the regulations of
the Nuclear Regulatory Commission nor the behavior of the
CHAIRMAN APOSTOLAKIS: What is appropriate it
seems to me is what the American people say is their role.
Anyway, I think we are running out of time here.
Thank you very much.
I don't know, now do you want to continue with --
MR. KING: It's up to you. Mary mentioned to me
that Section 5 has changed quite a bit, so if you have
comments on the old Section 5 it's probably not worth taking
time to go through them.
MS. DROUIN: I mean we have --
CHAIRMAN APOSTOLAKIS: I have comments on Section
4 and then I would be very happy to skip 5.
MS. DROUIN: I mean we substantially reworked 5.
The new version doesn't look anything like the old version.
CHAIRMAN APOSTOLAKIS: Will we have another chance
to look at 5, or the new 5?
MS. DROUIN: I mean you have it.
DR. KRESS: We have it here.
MS. DROUIN: I mean a chance to talk to you about
it. Tomorrow or the next day?
MR. KING: It won't be tomorrow. We are not --
the subject isn't on the agenda for tomorrow but certainly
we would ultimately like a letter on the framework in
support of our August paper. Now the timing of how to get
that we have got to talk about.
DR. KRESS: That's probably, that's something we
ought to talk about now, I guess.
MR. MARKLEY: The next meeting is August 30
through September 1st.
CHAIRMAN APOSTOLAKIS: Maybe we should postpone
then the discussion until then. I have a few comments but
they are --
DR. SHACK: Let me just ask what is intended in
Table 5-1 where you have this regulatory coverage of some
accidents important to risk.
I mean is the implication here that you need a
regulation addressing each of the accidents important to
risk, or can you take the argument that if you have achieved
a sufficiently low level of risk by some other means you
don't need to explicitly cover them?
DR. KRESS: Which Figure 5-1 are you looking at?
DR. SHACK: Table 5-1.
CHAIRMAN APOSTOLAKIS: Table 5-1.
MS. DROUIN: It's been awhile since I have looked
The only thing we were trying to say, and you
know, this is an early version, I can't tell you what the
new version looks like, is that we wanted to step back and
look at Part 50 and look at what are the risk insights
coming out of your PRAs -- what are your dominant accident
sequences, what are the contributors, and just kind of match
them up to the regulations to see at a quick, high level,
back-of-the envelope approach is there some glaring hole in
Is there some dominant accident class or some
contributor, some large contributor that is not being
addressed by a regulation.
This was just showing how -- you know, we went
through the accident types that are important to core damage
and to LERF and tried to map up where these are being
covered by the regulations, not to read any more into that
table than that.
DR. SHACK: You did seem to be determined to
preserve the concept of a design basis accident, and that
would somehow seem to be inconsistent with this notion that
all these accident classes -- or these would become new
design basis accidents?
MS. DROUIN: Not that these should become new.
This is just looking at here are the accidents coming out of
the PRAs, that PRAs have told us are important.
And we actually have regulations that are matching
up. And if there is something that is not a match, then we
would then therefore want to go look at to say this is a
hole in the regulation.
DR. SHACK: Okay, so we want to think about this?
MS. DROUIN: Yes, it's just a high level screening
tool for us.
CHAIRMAN APOSTOLAKIS: Have you really thought
hard about this position that the concept of design basis
accident should be preserved?
MS. DROUIN: Thought hard about it? I'm not sure
what you mean by hard. We've thought about it.
CHAIRMAN APOSTOLAKIS: You just took it for
granted that this is a good idea? We are doing a lot of
things that really go away beyond the design basis
DR. SEALE: Like steam generator tubes.
CHAIRMAN APOSTOLAKIS: Yes. This is a beautiful
paragraph here, right below that table. Some
risk-significant accident types and related events do not
find any mention in the current regulations.
So I don't know that the DBAs are something worth
MS. DROUIN: It may not be those DBAs, but I do
think the concept of having accidents against which you want
to design, personally I think that's a good concept.
CHAIRMAN APOSTOLAKIS: It is, but it could take a
different form than the current form of design basis
MS. DROUIN: That's all we're saying. We want to
keep the concept.
CHAIRMAN APOSTOLAKIS: Yes.
MS. DROUIN: We aren't saying it's necessarily the
DBAs that are the books now will still be the same ones and
in the same form.
DR. POWERS: Your affection for the concept is
because of its design facility that it provides?
MS. DROUIN: I'm sorry, I didn't hear you.
DR. POWERS: The reason you have an affection for
this concept is because of the facility it provides to
MS. DROUIN: Yes.
MR. KING: That's part of it, and part of it is
unless you go risk-based, I think you're pretty much forced
to come into some sort of design basis accident concept. So
as long as those design basis accidents really reflect the
risk considerations, and --
CHAIRMAN APOSTOLAKIS: But there is always a
MR. KING: Sure. There's a residual risk, even if
you want risk-based.
DR. BONACA: And under a probabilistic regime, I
mean, you would have to postulate some event to determine
the sizing of your pumps or the way you're injecting.
DR. POWERS: Well, what I'm asking -- you might
well do that, but the question I'm asking is, if it's just
the design and not this question of risk-based, and just a
question of design, why does the NRC want to get involved?
Why ought not that be up to the designer hypothesizing
anything you want to do?
Now, this other issue that Tom brings up, which is
that, in essence, we don't want to become risk-based because
we don't think our risk analysis tools are yet comprehensive
enough and robust enough that we can rely strictly on risk,
then gets into this question of there are certain kinds of
features of plants that we want to proscribe.
We want to proscribe plants that have an
end-stable reactivity profile. Just don't even bring them
to me, they're forbidden.
I think I'm putting words in your mouth, but
DR. KRESS: No, that's an accurate statement of
what I propose.
DR. POWERS: Once again, we come to the
possibility that because we can't set up a completely
self-consistent regime here, we have to legislate against
certain kinds of alternatives.
DR. KRESS: That makes sense to me.
DR. POWERS: So, we can't get rid of the concept
of design basis accidents.
DR. KRESS: Yes. I think even in the extreme of a
risk-based system, your design basis accident becomes the
whole spectrum of PRA.
DR. POWERS: Well, it does, but it might be
different for each plant.
DR. KRESS: It would be different for each plant.
DR. POWERS: Yes, or certainly each plant type.
DR. KRESS: It certainly would be plant-specific
design basis accident.
DR. POWERS: We discussed that before, because of
different locations, the risk profile must necessarily be
DR. KRESS: Yes, and because of the different
DR. POWERS: All you've succeeded in doing is
persuading them, boy, I'm glad I'm not doing your job.
DR. KRESS: We don't think this is an easy job.
CHAIRMAN APOSTOLAKIS: I think the most
fundamental issue here is the adequate protection versus the
use of goals.
DR. KRESS: Yes. I think we have a fundamental
difference of opinion among some of the members on that
DR. POWERS: I think that's been adequately
reflected in the e-mail traffic.
DR. BONACA: To which we did not contribute
CHAIRMAN APOSTOLAKIS: Say again?
DR. BONACA: For which -- I follow that traffic
very, very carefully. I was going to help in a couple of
times, and I said, well, it's already said.
CHAIRMAN APOSTOLAKIS: I would say it's reflected
in today's discussion as well.
DR. KRESS: We probably should have added Tom and
Mary to this, and put our e-mail address on that. It might
have been interesting.
DR. BONACA: I would like to say at some point --
CHAIRMAN APOSTOLAKIS: How are we going to solve
DR. BONACA: Wait. If we --
DR. SHACK: We're going to out-vote you, George.
[Discussion off the record.]
DR. BONACA: If we decided to go with adequate
protection, then we would have to quantify it in terms of
some risk measure. And I don't see why the one that we
propose would not be appropriate on an average basis.
CHAIRMAN APOSTOLAKIS: That's what bothers me,
that because it's the only risk measure that we have, we
resort to using that without thinking that its intent was
DR. SHACK: I still come back to my argument that
if you have goals, that your regulatory system ought to be
out to achieve those goals.
DR. KRESS: That's exactly what the Commission
said they were intended for.
MR. KING: I remind you that this issue is going
to go to the Commission in our August paper. I mean, we're
not going to decide this unilaterally. We're going to lay
this out for the Commission that we're using their safety
goals, subsidiary objectives for this purpose, and, in
effect what it means is, we're risk-informing the
regulations to achieve the level of safety that they -- that
was stated as their expectation in this Agency in the safety
DR. KRESS: I think that's --
CHAIRMAN APOSTOLAKIS: So now someone meets all
the regulations under Option 3, and then you promise to
leave them alone and you will never impose additional
regulations on them because they are safe enough?
MR. KING: Yes, unless some new information comes
DR. KRESS: If you could write perfect
CHAIRMAN APOSTOLAKIS: And the staff should
incorporate in the reactor safety goal policy statement that
it is the Commission policy that safety goals are goals and
Now, what difference would that make? That some
plant may violate them and when you issue the rule or
whatever rule --
DR. KRESS: That's what it means; that's exactly
what it means.
CHAIRMAN APOSTOLAKIS: I don't understand that.
DR. KRESS: That's why they're goals.
CHAIRMAN APOSTOLAKIS: They issue a rule that
flows from this framework, what does that mean?
DR. KRESS: That means some plants will --
CHAIRMAN APOSTOLAKIS: Will not follow the rule?
DR. KRESS: No. They will follow the rules. It
says that the rules cannot be written so perfectly that
every plant in how it operates and how it designs and how it
runs its things, will not meet those numbers. It will meet
the rules, but it will not meet those specific numbers.
And that's the way the rules are written now, and
that's the situation we have now.
MR. SIEBER: Some plants may choose not to
risk-inform that rule.
DR. SHACK: Even if you assume --
DR. KRESS: Even if you choose all of them --
DR. SHACK: -- choose to do it, that's going to be
a fact of life, that when they do it, because they've done
it on a generic basis, some will make and some won't.
DR. KRESS: More than likely, the ones that choose
to do it will be the ones that know they can make it easily
But that's not something we ought to debate.
CHAIRMAN APOSTOLAKIS: I don't think you're right
with your argument that if I have goals, I should strive to
achieve them, and otherwise, why do I have them? No.
The goals were specifically interpreted as meaning
that if you are below those, there's no reason to even
consider additional safety measures.
That doesn't mean that I will establish -- it's
safe enough; that's what it says.
DR. KRESS: I don't think that's what they said.
DR. SHACK: I think there was an expectation that
they would do that on the average.
DR. KRESS: On the average. I think that's
explicitly stated in there.
CHAIRMAN APOSTOLAKIS: But it's a safe-enough
DR. POWERS: There was an expectation that --
DR. KRESS: They made it on the average.
CHAIRMAN APOSTOLAKIS: That's different from the
expectation. The protection of the goal is that it's safe
enough in the sense that I will not ask you to do anything
DR. KRESS: That's an additional interpretation.
CHAIRMAN APOSTOLAKIS: This is a --
DR. KRESS: Related to the backfit rule.
DR. POWERS: My perception, based on what I have
read about the history of things was they were interested in
the question of, is the body of regulation such that things
are safe enough now, or is there some tremendous omission
that we've made out there in this body of regulation?
And one needs only think about the history prior
to Appendix R and 50.48. Here's a tremendous omission.
I think they were really interested in, is there
evidence of some tremendous omission? And we derive that
evidence by looking at a lot of individuals, and seeing if,
on the average, things are well away from where our goals
DR. KRESS: That's a good way to look at it.
DR. POWERS: But it doesn't obviate our problem.
If we want to regulate, if we want to go approve a
particular installation, what it wants to do, you need some
number to compare against, and they're just not giving it to
DR. KRESS: Yes. They're not giving us that
number at all.
DR. POWERS: And I don't know why. I mean --
DR. KRESS: I don't know why they're reluctant to
DR. POWERS: They kind of have, in that they said
go ahead and use 1.174, okay, for changes. And I should
say, for the formal -- what method of granting exemptions,
go ahead and use this number, which seems to me a lot more
tenuous of a thing than saying prove the operation of this
I seems to me that granting exemptions, I'd be
more cautious with than just a general running of the plant.
DR. KRESS: It's curious.
DR. POWERS: Makes me glad that I don't have their
DR. KRESS: We've argued that they ought to do
CHAIRMAN APOSTOLAKIS: Are there any other
MR. HASKIN: I just want to make one comment.
CHAIRMAN APOSTOLAKIS: Yes.
MR. HASKIN: While I was at Sandia, one of the
jobs I had was to supervise the group that developed the
methods to do the uncertainty analysis in 1150. If you guys
recall, the uncertainty distributions on core damage
frequency from 1150, typically from the fifth to the 95th
percentile is two orders of magnitude.
If I establish a goal for core damage frequency at
adequate protection at 10 to the minus three, and I base
that on a mean, that means that there is a five percent
chance that my actual core damage frequency could be as high
as 10 to the minus two.
One of the things you haven't considered in this
discussion of safety goals versus adequate protection is
that if you want to set an objective based on adequate
protection for things like core damage frequency, you want
to base that on a mean, or do you want to put some
DR. KRESS: That's right.
DR. POWERS: It may have been neglected in this
discussion, but it has been neglected before, and I will
hasten to point out that the suggestion has been the 95
percentile for that particular goal.
DR. KRESS: Puts you right back to the safety
goals again. That's what I was kind of arguing all along.
CHAIRMAN APOSTOLAKIS: Well, this will come after
we decide what the limit -- what the goal is. And in 1.174,
we just -- because, you know, it was too soon for these
things, so we said there will be increased management
attention, which means a qualitative treatment.
But you're right, that this uncertainty has to be
taken into account. But that does not eliminate the
fundamental concern of goal versus adequate protection.
DR. POWERS: If you're striving for the safety
goal, I feel relatively comfortable making comparisons based
CHAIRMAN APOSTOLAKIS: So now you're saying that
on the basis of defense-in-depth, I will use goals for
MR. HASKIN: No, it's just something that has to
be considered in the deliberation.
DR. POWERS: What do we know about the history of
the nuclear inspection -- anyone in their goal limit --
CHAIRMAN APOSTOLAKIS: We have this in -- see,
the thing about documents that come out of Europe, as far as
I can tell, is that this concern about uncertainty that we
have here is not there.
So they tell you, individual risk, if it's less
than 10 to the minus six, it's broadly acceptable by
society, what we would call safe enough.
If it's 10 to the minus four or higher, it is
unacceptable, what we would call adequate protection is
In between, it's tolerable, in other words,
DR. POWERS: I'm aware of the words.
CHAIRMAN APOSTOLAKIS: But the fact that 10 to the
minus six may have two orders of magnitude uncertainty up
and down does not appear, at least explicitly, does not
DR. POWERS: So they must be prescribing methods
CHAIRMAN APOSTOLAKIS: Not that I have seen. The
documents that I have seen do not.
DR. POWERS: Well, maybe they're not prescribing a
particular method, but they're prescribing that the method
be acceptable to the regulator.
CHAIRMAN APOSTOLAKIS: Oh, I'm sure it has to be.
DR. KRESS: They can deal with it in that
DR. POWERS: It's much more collegial and a less
confrontational system. So if regulatory and regulatee
agree on a method of analysis, yes, you can omit the
question of uncertainties.
DR. KRESS: Because it's implicit in that.
DR. POWERS: It's been implicit in the way you've
developed it, yes.
CHAIRMAN APOSTOLAKIS: Yes.
DR. POWERS: If you guys would just get along with
your licensees, instead of being so damn confrontational.
CHAIRMAN APOSTOLAKIS: Any other comments from the
members on issues that have not been addressed?
MR. KING: Do you want to talk timing of the next
CHAIRMAN APOSTOLAKIS: This is a good idea. You
are supposed to send something to the Commission in August?
MR. KING: End of August, we owe a paper to the
Commission that will have the framework, the policy issue in
DR. POWERS: Why don't you come present that
package to us at our September meeting?
DR. KRESS: And our letter would follow.
CHAIRMAN APOSTOLAKIS: Is that too late? That's
what my question was; when is the Commission expected to do
something about it, so that September will not be late?
MR. KING: I think if we could meet at your
September meeting, it would probably work out okay.
CHAIRMAN APOSTOLAKIS: Can we write the letter at
the September meeting?
DR. KRESS: Yes, I think we could do that.
George, that would be a nice letter for you to write.
DR. SHACK: I can work on my added comments.
DR. KRESS: You and I can get together and work on
our added comments.
MR. KING: Which means you probably want another
Subcommittee so that we can delve more into the policy
CHAIRMAN APOSTOLAKIS: I don't think another
Subcommittee will help, but I think we can have a three-hour
DR. POWERS: I think we would have great
difficulty trying to schedule another Subcommittee meeting,
but on the other hand, I think this is -- we've examined and
debated this enough that if you could just appear before us
and say here are the things that we call your attention to
that have been changed dramatically from what we've talked
to you about before --
I mean, just do that in a very terse fashion,
because, quite frankly, as has been apparent, we've been
going over this thing literally line-by-line.
CHAIRMAN APOSTOLAKIS: The other thing is, you
know, it depends on what they say in the letter to the --
the report they send to the Commission.
If they pose an issue and they say, look, we can
interpret it this way or that way, tell us what to do, then
I don't think there is a question of added comments, right?
DR. KRESS: Probably.
CHAIRMAN APOSTOLAKIS: If the staff comes back and
is de facto using certain things in a certain way, then some
members may disagree, so it depends very much on what you
guys write in the final.
MR. KING: You'd be interested in how we frame the
issue and what our recommendation is.
DR. KRESS: Yes, definitely that.
CHAIRMAN APOSTOLAKIS: If you offer options to the
Commission, arguing, you know, both sides of the coin, I
don't see why we would disagree.
DR. KRESS: They may have a preferred option.
DR. POWERS: I would presume the Commission would
like to get our thoughts on which option to take.
DR. KRESS: Or if there is another option.
DR. POWERS: This sage and insightful one that Tom
comes up with or the thing that we come up with.
CHAIRMAN APOSTOLAKIS: So we will have then a
presentation during the September meeting with the
expectation that we will write a letter at the time.
MR. KING: Okay, sounds good.
CHAIRMAN APOSTOLAKIS: Okay. Anything else?
DR. KRESS: That's a two-hour session.
CHAIRMAN APOSTOLAKIS: Members of the public?
DR. POWERS: I would try to make it short, the
CHAIRMAN APOSTOLAKIS: No, I think we should have
enough time for discussion. The presentation itself can be
DR. POWERS: I think I'm very familiar with the
agenda for September, and I know there's no room to make it
a lengthy period, and I'm going to talk to you guys. I
think September is going to be a four-day meeting.
CHAIRMAN APOSTOLAKIS: The fourth day is Labor
DR. KRESS: I don't care; I'm retired.
DR. POWERS: Second of all, I don't think that's
the issue that we're confronting.
They will have a fait de complis; they will have a
list of options that are so well explained that there's no
clarification. All they need to do is call our attention to
DR. KRESS: All we need is to have the documents
ahead of time.
DR. POWERS: Dr. Shack will be able to formulate
his clear and insightful points of view that will
undoubtedly be endorsed by the rest of the Committee at the
expense of others.
CHAIRMAN APOSTOLAKIS: I am not that these
comments deserve to be in the transcript. Does anyone have
anything of substance to say?
CHAIRMAN APOSTOLAKIS: Thank you very much, both
the staff and Mr. Christie. This was a very informative
meeting today. I hope the debate will not be repeated in
DR. SHACK: We tried to head it off with the
e-mails, George, but --
DR. KRESS: It didn't work.
CHAIRMAN APOSTOLAKIS: So we are adjourned. Thank
[Whereupon, at 5:07 p.m., the meeting was
Page Last Reviewed/Updated Tuesday, July 12, 2016