489th Meeting - February 7, 2002
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
489th Meeting
Docket Number: (not applicable)
Location: Rockville, Maryland
Date: Thursday, February 7, 2002
Work Order No.: NRC-214 Pages 1-368
NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers
1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005
(202) 234-4433. UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
+ + + + +
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
(ACRS)
489TH MEETING
+ + + + +
THURSDAY
FEBRUARY 7, 2002
+ + + + +
ROCKVILLE, MARYLAND
The ACRS met at the Nuclear Regulatory
Commission, Two White Flint North, Room T2B3, 11545
Rockville Pike, at 8:30 a.m., George E. Apostolakis,
Chairman, presiding.
COMMITTEE MEMBERS:
GEORGE E. APOSTOLAKIS, Chairman
MARIO V. BONACA, Vice Chairman
F. PETER FORD
THOMAS S. KRESS
DANA A. POWERS
STEPHEN L. ROSEN
WILLIAM J. SHACK
JOHN D. SIEBER
GRAHAM B. WALLIS
ACNW MEMBERS PRESENT:
B. JOHN GARRICK, ACNW, Acting Chairman
MILTON N. LEVENSON
STAFF PRESENT:
MICHAEL T. MARKLEY
SAM DURAISWAMY
HOWARD J. LARSON
SHER BAHADUR
CAROL A. HARRIS
NOEL DUDLEY
JOHN T. LARKINS
ALSO PRESENT:
VICTOR H. RANSOM, Invited Expert
MARK CUNNINGHAM
FRANK GILLESPIE
STEWART MAGRUDER
RICHARD BARRET
LAWRENCE KOKAJKO
SAMUEL J. COLLINS
ASHOK THADANI
WILLIAM D. TRAVERS
MARTIN VIRGILIO
GARY HOLAHAN. C-O-N-T-E-N-T-S
Opening Remarks by the ACRS Chairman . . . . . . . 4
Risk-Informed Regulation Implementation Plan . . . 7
Meeting with EDO and Office Directors of NRR,. . .72
NMSS, and RES
Proposed Final Revision to Regualtory Guide. . . 181
1.174 and SRP Chapeter 19
PTS Technical Bases Reevaluation Project . . . . 249
Adjourn. . . . . . . . . . . . . . . . . . . . . 368
. P-R-O-C-E-E-D-I-N-G-S
(8:34 a.m.)
CHAIRMAN APOSTOLAKIS: The meeting will
now come to order. This is the first day of the 489th
Meeting of the Advisory Committee on Reactor
Safeguards. During today's meeting the Committee will
consider the following, this Conformed Regulation
Implementation Plan, meeting with the ADO and the
Office Directors of NRR, NMSS and RES, status report
on the proposed final revision to Regulatory Guide
1.174, and Standard Review Plan, Chapter 19, PTS
Technical Basis Re-evaluation Project, Proposed ACRS
reports. ACNW Members John Garrick and Milt Levenson
will participate in the meeting with ADO and the NRC
Office Directors.
This meeting is being conducted in
accordance with the provisions of the Federal Advisory
Committee Act. Dr. John T. Larkins is a designated
federal official for the initial portion of the
meeting.
We have received no written comments or
requests for time to make oral statements from members
of the public regarding today's sessions. A
transcript of portions of the meeting is being kept,
and it is requested that the speakers use one of the
microphones, identify themselves, and speak with
sufficient clarity and volume so that they can be
readily heard.
I will begin with some items of current
interest. I'm pleased to announce that the Commission
has approved the appointment of Dr. Victor Ransom to
the ACLS, subject to final clearance review. Dr.
Ransom is attending this meeting as an observer.
Welcome.
Amarjit Singh is going on rotation for
three months as Senior Project Manager in the
Probabalistic Risk Analysis Branch of the Office of
Research, effective February 10th. And I am happy and
sad to announce that Mr. Noel Dudley is leaving the
ACLS - will you stand up, Noel - to join the Office of
Nuclear Reactor Regulation as a Senior Project Manager
with the License Renewal Section, effective February
10th, as well.
Noel, as we all know, has been one of the
most valuable members of the staff. He has been with
ACLS for eight years, and he has made significant
contributions in several areas, including License
Renewal, Steam Generator Cube Integrity, Materials and
Metallurgy, the Licensing of AP-600, Safeguards and
Human Factors.
I'm happy to congratulate Noel, but I'm
also sad, as I said, that he's leaving us, and I think
he deserves a round of applause.
MEMBER ROSEN: I think we'll also miss Jit
over the three months he -
CHAIRMAN APOSTOLAKIS: Jit is coming back
in three months. I don't have to say too much about
him.
MEMBER ROSEN: Can I get a - can I be
sure of that?
CHAIRMAN APOSTOLAKIS: Now one other item
of interest, in the hand-out, those items of interest,
please go to pages 34 and on, and find information on
the Annual - the 14th Annual Regulatory Information
Conference, which will be held in Washington next
March, March 5th, 6th, and 7th. So if any members
decide that they would like to go, please let us know.
MEMBER POWERS: When is our meeting in
March?
CHAIRMAN APOSTOLAKIS: When is our
meeting? The last date.
MEMBER POWERS: Yeah.
MR. DUDLEY: I'd like to add to that that
the registration form for the meeting is on - is the
last page of your package.
CHAIRMAN APOSTOLAKIS: Yeah.
MR. DUDLEY: You can also register on-
line.
CHAIRMAN APOSTOLAKIS: Okay. Are there
any comments or announcements from the members?
Hearing none, we'll proceed with the agenda. The
first item is Overview of Risk Informed Regulation
Implementation Plan. Dr. Shack, you will lead the
Committee through this.
MEMBER SHACK: Okay. We've, of course,
spent a lot -
CHAIRMAN APOSTOLAKIS: Bill, your
microphone.
MEMBER SHACK: We've spent a lot of time
discussing individual issues under Risk Informed
Regulation. I think this is the first time that I can
recall reviewing the overall Risk Informed
Implementation Plan, and I guess Mark Cunningham is
going to lead the presentation for it.
MR. CUNNINGHAM: Thank you, sir. You're
right. I think - if the Committee has been briefed
on the Implementation Plan, it's been a long time ago,
so I think the intent of our briefing today is to both
give you some ideas of the general structure of the
plan, and the rationale for the plan as it's laid out.
And then give you some - describe some of the more
important initiatives that are embedded in the plan.
I'm Mark Cunningham in the Office of
Research. With me is Stu Magruder and Frank Gillespie
from NRR. Joining us shortly will be Lawrence Kokajko
from Office of Nuclear Materials Safety and
Safeguards.
We've got four parts to the presentation
today. I'm going to give you an overview of the new
Implementation Plan format and content, and some of
the rationale of why it looks the way it does.
The Implementation Plan, the substance of
the Implementation Plan is organized by a strategic
arena. Frank and Stu will talk about the reactor
arena work, including both major initiatives and
challenges facing the staff right now, and at least
one approach for identifying how we - for how we
would identify new areas, or new regulatory activities
to be risk informed in the reactor arena.
Lawrence will then talk about the
materials and waste arenas, again some important
issues that are facing them, some upcoming milestones,
and how they are now using - what process they're now
using to identify what else in NMSS they'd like to
risk inform. And then I'll come back at the end with
some next steps that we're taking.
By way of background, those of you who
have been on the Committee for a while may recall that
there used to be a PRA Implementation Plan, and in the
late 1990s, in 1999, we received some rather
considerable criticism of that from the General
Accounting Office, and the GAO advocated that to - in
order to really effectively implement NRC's PRA Policy
Statement, we needed to develop a comprehensive
strategy for risk informing NRC's Regulatory
activities. In 1999, the Chairman made a commitment
that we would modify the plan, and try to accomplish
what GAO was interested in. Next slide.
We've gone through two iterations of the
plan in the terms of the format of the plan since
then. We had a March 2000 version that went to the
Commission, and the Commission gave us some guidance
at that point on three specific areas. They said they
wanted to hear more about internal communications, of
how we talk internally and bring the staff together on
how the benefits of risk informing are requirements.
They also wanted -
MEMBER WALLIS: Can I ask you something?
MR. CUNNINGHAM: Yes, sir.
MEMBER WALLIS: You have a plan. Is it
clear what the goals are? Is it clear what the
objectives are before you have a plan to get there?
MR. CUNNINGHAM: I believe so. Yes, sir.
We could give you some examples when we get done here,
if you'd like.
MEMBER WALLIS: Okay.
MR. CUNNINGHAM: As I say, the - at the
high level, the purpose - the - what we're trying to
do in Risk Informed Regulation is oriented to the
strategic plan goals of the agency. They're much
broader than this, and everything that we talk about
in the plan is linked to the accomplishment of a
specific, what they call strategy, in the strategic
plan.
MEMBER WALLIS: It just seems to be -
Risk Informing seems to be sort of a method rather
than an objective. Is more Risk Informing better, or
is it itself a means to some other end, which is
greater than itself?
MR. CUNNINGHAM: It's a means to another
end, which is - and one way to think about it is to
improve the focus of our Regulatory activities on the
most safety important issues and topics.
MEMBER WALLIS: Okay. Then you to have
some measures of those successes.
MR. CUNNINGHAM: Yes.
MEMBER WALLIS: So you say we have Risk
Informed this regulation, and in so doing, we have
achieved some objectives which are measurable on some
scale.
MR. CUNNINGHAM: That's correct.
MEMBER WALLIS: It would be very useful to
have that.
MR. CUNNINGHAM: That's correct. That's
the -
MEMBER WALLIS: We have reduced the number
of pages in 10 CFR by 50 percent, or whatever.
MR. CUNNINGHAM: Yes, that's right.
Probably the most obvious success so far has been in
the new Oversight process used by inspection, where we
are focusing our inspection activities on the most
safety important issues, and that's led to a lot of
challenges as well, but that's probably the biggest
success.
CHAIRMAN APOSTOLAKIS: I thought the most
successful one was the Risk Informed In-Service
Inspection. They use a number of metrics, and it
looks really good.
MR. CUNNINGHAM: That's another one. I
think of it in terms of the - how the - there's a
large number of inspectors in this agency out in the
regions, and how they spend the time on a day-to-day
basis is dramatically different now than it was five
or ten years ago. So in terms of NRC resources,
certainly they're being allocated much differently
today, and I think, to a much better focus on safety.
Okay. So at any rate, in 2000, the
Commission asked for better communications internally,
a plan for better communications internally,
explicitly talk about staff training requirements, and
to come back to the Commission, and tell them of
impediments that we see in progress for achieving the
goals of the PRA Policy Statement, so that led to an
October 2000 version.
In January of 2001, the Commission came
back with more specific instructions to us. They're
shown on slide five. They wanted to have a better
idea of the priorities of individual activities within
the Implementation Plan. They wanted to see more
detailed communications plans. That really means
activity specific communications plans. They wanted
to know what resources were being applied to what
activities. They wanted to bring in performance-based
regulation for us to identify where performance-based
- the performance-based policy aspects of what we're
doing is brought into the Risk Informed areas.
Also, to identify critical path items, and
important, what they call cross-cutting activities,
activities that have implications for a number of
different Regulatory activities. And so on slide six,
you see the current version of the Implementation Plan
that was sent to the Commission in December. SECY
010218, does provide the priorities determined by the
Implementing Offices.
MEMBER WALLIS: I'm sorry. This goes back
to my question. It's all about activities, isn't it?
MR. CUNNINGHAM: Yes.
MEMBER WALLIS: And you've got keep sight
of where you're going.
MR. CUNNINGHAM: Okay. If you would hold
off on that one slide, we'll try and come back to
that, or a couple of slides.
Again, we have added general and specific
discussions of communication - and project specific
discussions of communications activities, resources
for FY01 and 02, where it was appropriate to bring in
the performance-based discussion, and identify cross-
cutting activities.
So in terms of what you have in front of
you now, there's two basic sections to the
Implementation Plan. One is the background and basis
for why we're Risk Informing our activities, and this
is a combination of the essence of the Commission's
1995 Policy Statement, and the strategic plan that was
issued a couple of years ago.
It also talks in general terms about how
we decide what activities to Risk Inform. There are
factors - there are a set of factors that are in
there that are oriented towards making the decision
whether or not to Risk Inform a particular Regulatory
activity. They're at a more general level. Each of
the arenas implements them in a somewhat different
way. Again, it provides communications plans,
training programs.
Part Two is, again, much more specifically
oriented with the strategic plan, and the details of
the strategic plan. If you recall, in the strategic
plan, there are a set of agency strategic goals, and
there are performance goals to accomplish the
strategic goals. And there are strategies to
accomplish the performance goals. What you'll -
MEMBER KRESS: Where is it you discuss the
impediments?
MR. CUNNINGHAM: I'm sorry?
MEMBER KRESS: Where is it you discuss the
impediments that the Commission has asked for?
MR. CUNNINGHAM: They come in in the -
probably the - typically in the body of the
Commission paper. If we feel it's an impediment that
the Commission can do something about, if it's
necessary for the Commission to do something.
MEMBER KRESS: It's in the Commission
paper, not in the plan.
MR. CUNNINGHAM: Typically, yes.
MEMBER KRESS: Okay.
MR. CUNNINGHAM: That's right. The
Commission - in the impediments here, the Commission
was - I think - I believe it was Commissioner
McGaffigan, I'm going to say, if there's something
that we can do to help move this along, tell us and
we'll see what we can do.
MEMBER KRESS: I see.
MR. CUNNINGHAM: And Part Two is described
and is organized into two chapters, and focused by
arenas. The reactor safety arena is one chapter, and
we have the materials and waste arenas combined in a
second chapter.
Going to slide eight, this is where, Dr.
Wallis, we tried to get into how do we relate the what
to the why, if you will. This is kind of an example
of the activity descriptions that you'll provide -
you'll see in the plan.
Up in the upper left corner are the agency
performance goals and strategies that are relevant to
a particular activity, so we've oriented the work that
we're doing to say we need this body of work to
accomplish this strategy that's laid out in the
agency's strategic.
MEMBER WALLIS: Well, what does it say
there under the performance goals?
MR. CUNNINGHAM: Well, let me pick - I'll
pick one as an example, see what -
MEMBER WALLIS: I can't read it. I'm sure
you can't.
MR. CUNNINGHAM: Well, I'll pick it -
I've got the book in front of me, so I'll try to -
okay. Let's - 8-8, there we go. Okay. So an agency
primary performance goal, picking one example is,
"maintain safety, protection of the environment and
the common defense and security." Very high level
goal. Okay.
There are a number of strategies that the
agency has defined to accomplish that goal. Strategy
eight is, "We will continue to develop and
incrementally use Risk Informed, and where
appropriate, less prescriptive performance-based
Regulatory approaches to maintain safety." And so
what you'll get out of that is a set of projects that
are intended to accomplish that strategy.
MEMBER WALLIS: It just seemed to me
that's such a high level statement. I'm not sure it's
very useful for planning a particular activity.
MR. CUNNINGHAM: The challenge - I think
the biggest challenge we faced in this is what is the
necessary and sufficient set of projects needed to
accomplish that strategy. And this is - and Frank
will get into some of this later. We have a set of
activities -
MEMBER WALLIS: Well, you said you're
already doing that with the present regulations.
You've already met that strategy with the present
regulations, so you've got to have something else
which tells you what the payoff is for Risk Informing.
MR. CUNNINGHAM: Okay. Well, let's - I'm
not sure - the strategic plan was something that -
CHAIRMAN APOSTOLAKIS: Well, could you
read the goals again? It's safety?
MR. CUNNINGHAM: The perform - the higher
level goal -
CHAIRMAN APOSTOLAKIS: Higher level, yeah.
MR. CUNNINGHAM: Maintain safety,
protection in the environment, and the common defense
and security.
MEMBER WALLIS: We do that already.
CHAIRMAN APOSTOLAKIS: That's not why
we're Risk Informing the regulation.
MR. CUNNINGHAM: I agree.
CHAIRMAN APOSTOLAKIS: These are boundary
conditions actually.
MR. CUNNINGHAM: This is a very high
performance goal, and there are many things that the
staff does to accomplish that goal. A subset of those
are Risk Informed activities.
CHAIRMAN APOSTOLAKIS: But shouldn't we
say somewhere in there that the whole idea is to
remove unnecessary burden and -
MR. CUNNINGHAM: For example, on 1-H you
have a secondary performance goal, which is to reduce
unnecessary burden.
CHAIRMAN APOSTOLAKIS: Yeah. There has to
be a goal somewhere.
MR. CUNNINGHAM: That's correct.
CHAIRMAN APOSTOLAKIS: Because that's
what's driving all this. We're not just conforming
the regulations to maintain safety.
MEMBER KRESS: Increase sufficiency of the
regulation -
CHAIRMAN APOSTOLAKIS: Yeah.
MEMBER KRESS: - and to reduce burden
where it's appropriate.
CHAIRMAN APOSTOLAKIS: That's right.
Exactly.
MEMBER KRESS: I think we've stated those
somewhere.
MR. CUNNINGHAM: That's right. And I
picked one activity out of a bunch of activities.
There are other - that was - the one I talked about,
maintain safety, is a performance goal. Another
performance goal is to reduce unnecessary Regulatory
burden. And there's -
MEMBER WALLIS: Well, I think that you
ought to give the safety prong much more weight. And
my view is that if you really did Risk Informed
Regulations, you'd have a far better idea, and the
public would have a far better idea of how safety is
really being maintained. That's a much better goal
than this rather diffuse thing of reducing burden and
being efficient.
CHAIRMAN APOSTOLAKIS: But these goals are
not part of the plan. Right?
MR. CUNNINGHAM: Those goals are part of
the strategic plan of the agencies.
CHAIRMAN APOSTOLAKIS: The strategic plan.
Yeah.
MR. CUNNINGHAM: Not the Risk Informed
Regulation.
CHAIRMAN APOSTOLAKIS: Yeah, so we can't
really debate them.
MR. CUNNINGHAM: If you'd like - there's
another set of people you can debate those with, if
you'd like, but that's not us.
MEMBER WALLIS: I'm trying to be helpful.
I think if you could get these objectives in a better
- in a more specific form, it might be easier to plan
the activities.
MR. CUNNINGHAM: Agreed.
MEMBER WALLIS: If you have something
that's too general and too vague, then any activity
will do.
MR. CUNNINGHAM: And that's where you get
down to the strategy, and then what you need to
accomplish the strategy.
CHAIRMAN APOSTOLAKIS: Well, I understand
all this, but it seems to me that all this rests on
the assumption that you already have the activities.
In other words, given an activity, I have these
things, you know, how does it - which goal does it
serve, through which strategy, and so on. Isn't the
most important part though, how to come up with the
activities? What to do? I mean, given an activity,
yeah, you can always give some justification. Is that
part of the plan?
MR. CUNNINGHAM: Yes, it is.
CHAIRMAN APOSTOLAKIS: Okay.
MR. CUNNINGHAM: In the sense of the plan
as you see it in the Commission paper is the -
actually, the August 2001 version of the plan, there's
a commitment in there, and there's a statement in Part
One that says this is, in effect, what we're doing
today. And we're going to describe a process in Part
One that says we're going to go out and continue to
identify, seek out and identify whether or not there
are other things, other Regulatory activities that
need to be Risk Informed. That is a very key piece of
the Implementation Plan.
One of the big criticisms from GAO was
you've given me a catalogue of what you're doing. You
haven't told me where you want to be and how you're
going to get there, and that is a key piece.
CHAIRMAN APOSTOLAKIS: So the selection of
the activities.
MR. CUNNINGHAM: The selection of the
activities is a key piece.
CHAIRMAN APOSTOLAKIS: Is a key piece
which means what now? There is a methodology for
doing this, or -
MR. CUNNINGHAM: There is a methodology in
place in the waste and reactor arenas that you can -
you'll hear about later. The methodology for that in
the reactor arena is still evolving. We have an IOU
to the Commission, basically, the next version of the
Implementation Plan, which is due in June, will
describe in much more detail the process that we're
going to use, and whatever results we have to date.
But you're right, to get at the real goal, to
accomplish the policy statement, to accomplish the
intent of the strategic plan, you have to have that
piece of it. And that has been a legitimate criticism
of previous versions of this. It didn't show that
path forward.
I'm leading into what Frank wants to talk
about this morning, because he has some ideas on how
we could do that. But the staff is working several
different ways, in several different activities to lay
out that future looking part of the plan.
MEMBER KRESS: Do you plan to talk about
the waste arena today too?
MR. CUNNINGHAM: Yes. Yes.
MEMBER KRESS: I understand, they don't
plan to use PRA. How can you risk inform any activity
without PRA, is the question I might have.
MR. CUNNINGHAM: One of the challenges in
trying to bring together all three arenas is a history
of different terminology and things. And they say -
they may say we don't use PRA, but we use performance
assessment. And then how is - then you get into how
is that different, and first blush, I'm not sure it's
very different. They've evolved separately.
MEMBER KRESS: They may, in fact, have
some sort of a risk analysis, you're saying.
MR. CUNNINGHAM: Yes. That's correct.
They just may not call it that, if you will. And then
when we get into the materials arena, again they are
thinking of how risk assessment is to be used there.
And you'll hear about how they're going to develop
safety goals and that sort of thing, so that we're
trying to bring things that have been done separately,
under a common set of - a common footing, if you
will.
So at any rate, the intent of the plan as
we show on slide eight is to link activities to the
strategic plan, performance goals, and to the more
detailed strategies. Basically then, we provide a
consistent set of information on individual
activities, and so that becomes, if you will, the
description of - for the Commission of what we've
said are the priorities, and what the resources are
associated with individual activities.
Perhaps I'll just go on from there, and
just say what we can do now is maybe turn to page -
slide nine, which is the reactor safety arena. And
we'll get into some more of the substance, some
initiatives, and their ideas on how they're going to
proceed to identify other activities. Frank or Stu.
MR. GILLESPIE: Yeah. I'm going to start
off, because I was told I have to absolutely put what
you're about to hear in context. And the context is
it has no management approval, other than having been
shown to people. And what we're groping with is not
in any way in conflict with the plan, the four
strategic goals of the agency, the four strategic
goals. And you might say when you pick a task,
whether you want to do that task or not, is how it
contributes to efficiency, and effectiveness, and
burden reduction. So those goals help prioritize, but
not necessarily select.
The other thing we were grappling with,
and what I'm suggesting is this is the rule making
group in NRR grappling with risk as integral. When
you change one rule, have you made the other rules
more important? There's some fundamental questions
you have to address when you start picking things one
at a time. And I think you know that I was kind of
involved in the early part of - the first nine months
anyway, of the oversight process. And what we've done
is looked back on that and said what did we learn from
that? How did we structure it? And I actually went
back and looked at the old NEI white paper, and what
we ended up with didn't look like anything what they
had suggested in the first place. The pieces were
there, but it came out different as it evolved.
So what you're about to hear is our best
thinking as reflected on some view graphs, to try to
structure or develop a structure that is a whole, that
is going to have to have some lower level objectives
that would allow us to explain how things fit
together, because we found ourselves doing 50.46,
50.44, petitions on heat curves, Zircaloid, Zirca, and
how does all this fit together? And how does it fit
together with things like the safety goal, the
subsidiary safety goals we use. Where do all these
things fit? How do we structure what we're doing so
we know what we're either creating or destroying, and
we can explain it, so this is very preliminary
thinking.
We're happy to have feedback. This has
not been displayed virtually to anybody. We sent
these view graphs around yesterday, but it will give
you an idea where Stu, who is kind of project
managing, how do we organize rule making together in
a Risk Informed manner. Tried to understand some
discussions we had, so this is very preliminary. No
one owns it but the rule making group right now, so
I'm going to let Stu go through it, and let's see if
we can answer some questions. I hope it does provoke
some questions, because we need an active dialogue on
how this all fits together.
MR. MAGRUDER: Thanks, Frank. Let me
first go through some of the significant items from
the reactor safety arena, if I could have slide number
nine. There it is.
These are just a sample of the many
activities from the plan, but we just wanted to
highlight these, and see if the Committee had any
questions about the status of any of these real
quickly.
We like to think that Risk Informed
Technical Specifications are one of the successes that
we've had in this area, that would fall under what we
term Option One of the 98300 Plan. No rule changes
are required for this. However, within the current
regulations we're making a lot of progress, and
getting a better safety focus on tech specs.
We've talked briefly about the oversight
process, the significance determination process is
evolving in that area. And Mark talked a lot about
how the inspection have been more focused on safety
important equipment.
Option Three, in general, is changes to
the Technical Requirements, and Part 50. The 50.44
rule making, which you've heard about before, is close
to proposed rule stage that should be coming out
within a month or two for comment. 50.46, there's
been a lot of discussion recently about how to
approach that, how to break that down into more
manageable pieces, maybe, but that's progressing as
well.
50.69, which is the proposed name for Option Two
Rule Making. I'm sure you've had a lot of discussion
on that. That is progressing, although there's some
challenges in there we'll talk about. And obviously,
the significant work on the PRA Standards with ASME,
ANS, and NEI on guidance. Slide ten, please.
MEMBER SHACK: One of the things that, you
know, we always come back to is, you know, how much
risk information do you have to Risk Inform with.
MR. MAGRUDER: Uh-huh.
MEMBER SHACK: And as I went through the
plan, I tried to sort of sum up everything that was
involved in getting risk information, you know. And
I came up with a tenth of an FTE on the standards
work, you know. Now I see that's one of the major
initiatives, and it gets a whole tenth of an FTE. You
know, 1.5 for SAPHIRE, two for SPAR which is, you
know, sort of fundamental for my level three analysis.
It just doesn't seem like, you know - we make the
words that the significance determination process is,
you know, the key ingredient we have to be working on
in the ROP, and yet I can come up with maybe three
FTEs out of the whole effort that seem focused on
improving the SDP process. You know, I hear words,
and then somehow the resources connected with them
don't seem to be commensurate.
MR. CUNNINGHAM: Okay. Just to be clear
on the standards, the large fraction, the vast
majority of the work going on in PRA standards is
being done outside of NRC, by ASME and by ANS, and so
it's not that it's not important, but our role in that
is rather limited.
MEMBER SHACK: No, but the notion of how
the standards and the whole PRA Review Process are
incorporated - you know, one of the difficulties we
have every time there's a Risk Informed application,
somebody trots in a PRA, you know. What is it good
for? Is it good enough? And somehow, I don't see any
emphasis in here on how we're going to use the
standards, we're going to use the review process as
part of a tool for the NRC to make that judgment as to
whether the PRA is applicable.
MR. CUNNINGHAM: Okay. Now one of the
challenges of the Implementation Plan is how do you
capture inter-relationships among activities. The PRA
Standards work is our support to the standards setting
organizations. How we use that is in another activity
in the Implementation Plan, which is Risk Inform,
develop of guidance for Risk Informed applications, or
something like that. So it goes - it's another area
of the plan.
One of the real frustrations we've had is
how do you show those types of relationships, so that
somebody picking up the document would understand that
there's more than just that tenth of an FTE associated
with making the decisions on how we use PRA in
licensing. That being said, that's one of the
continuing challenges with the plan.
That being said, after lunch, I guess
you're going to hear about where we're going on Reg
Guide 1.174. And that - you'll hear a lot this
afternoon on how we intend to endorse the ASME
Standards, and ANS work, and how we're going to bring
the NEI 0002 into this, and all that sort of thing,
and that's going to be discussed after lunch. And
it's different than what shows up in the plan, because
the work has evolved considerably in the last - our
ideas on how to do that have evolved considerably in
the last six months or so. That's another challenge
for the plan, how do you maintain a current, if you
will.
MR. BARRET: If I could interrupt, my name
is Richard Barret. I'm with NRR. I'd like to just
take a second to address your statement about the
Reactor Oversight process, Significance Determination
process.
There is a fair bit of effort, and a lot
of thought going on right now on the subject of where
are we going in the future with the Significance
Determination process. We have a Phase Two - a Phase
One methodology that we're happy with. We have a
Phase Two methodology that is a work in progress, very
much a work in progress. And we have a Phase Three
methodology, which also is a work in progress.
We are currently having discussions within
the staff as to how much emphasis we would give to
Phase Two versus Phase Three. And also, where our
priorities and our resources will go in developing and
finalizing particularly the Phase Two tool, but also
the SPAR models. And questions are being raised as
to, for instance, do we want to accelerate the bench
marking of those methodologies, so I just want you to
know that the question of the quality of the
Significance Determination Process Tools is very much
on our radar screen right now, and we -
MEMBER SHACK: It doesn't seem to be
highlighted very well in the plan though. You know,
I look at - you know, they've got like a priority of
six or seven for the SPAR stuff -
MR. CUNNINGHAM: Yes.
MEMBER SHACK: - which sort of says yeah,
okay. You know, we'll think about it when we get to
it.
MR. CUNNINGHAM: Yeah. You're right. In
August of 2001, that was a reflection of where the
agency was and the importance of the SPAR work.
Again, one of the challenges is that perspective has
changed considerably over the last three or four - as
we've developed more experience with SDP, so the plan
- and so the challenge is how do you continue to
reflect that, and somebody can look at this plan and
see that it's up to date. SPAR, I think it's
recognized, as Rich was alluding to, there's a much
better recognition today of the importance of the SPAR
models in the context of the three phases of the SDP.
MR. MAGRUDER: Okay. Real quickly, on
page 10, I just wanted to highlight that there are
obviously some challenges ahead of us in Risk Informed
Regulation, and these are just some of the areas that
I wanted to highlight. You can look through those.
The last one, I think, Dr. Wallis, we'll
address in the next couple of slides here, which is
how we see this all fitting together. This is - as
Frank said, these are preliminary views of this.
MEMBER ROSEN: Did you skip the third one
for a reason?
MR. MAGRUDER: Just for time, but I'd be
happy - if you have a question about it.
MEMBER ROSEN: I don't know what it means.
MR. MAGRUDER: The Risk Informed
Environment is an effort within NRR to - well, it's
got several phases. Basically, the goal is to try to
make risk information more available to the staff and
have the staff more open to using Risk Informed
Regulation - Risk Informed methods in their day-to-
day work.
MEMBER POWERS: If I'm a Project Manager,
let's say in NRR, and say I'm - I have a job, maybe
power upright maybe, and I say gee, I want to know
whether this is a risky operation. How do I get -
how do I find out? Say it's - let me just be very -
to be specific -
MR. MAGRUDER: Uh-huh.
MEMBER POWERS: Say I've got Indian Point
II and they want to power upright.
MR. MAGRUDER: Right.
MEMBER POWERS: How do I go about getting
risk information on Indian Point II?
MR. MAGRUDER: Well, I think Rich will
probably talk - if I can impose on Rich to talk some
more about that, but I think generally, the Project
Manager goes to Rich's staff in NRR, and ask them to
look at the submittal from the licensee.
MR. BARRET: Dana, we have - I guess we
could say there are two classes of license amendments
like that, those that come in that are flagged by the
licensee as being Risk Informed license amendments.
And then, of course, it's obvious that they should be
reviewed by the risk staff. But then there's another
class, a much larger class of license amendments that
- oh, and by the way, that first class - no, let me
not say that. That's not true.
MEMBER POWERS: Well, I'm - regardless of
what the licensee has submitted, I just want to know.
MR. BARRET: You mean, just woke up one
morning and you just want to know about -
MEMBER POWERS: I got a thing in front of
me from this licensee, wants to do something.
MR. BARRET: Okay.
MEMBER POWERS: He's got his case laid
out, perhaps using risk information, perhaps not. I
just want to know -
MR. BARRET: Right.
MEMBER POWERS: Because I've got - I
mean, you've already made a decision that if I want
this information, I can get it.
MR. BARRET: Yes.
MEMBER POWERS: Okay. And now I want to
know whether I should be asking for more or less,
because I want to know if it's risky, because it's
different.
MR. BARRET: Yeah.
MEMBER POWERS: It's changed the plan.
MR. BARRET: Yeah. And that's - one of
the things we did in SECY 98300 was we raised this
very issue, and that is, what if a licensee submits
something, and it meets all of our regulations, and it
looks like it's consistent with our current design
basis, and it looks like something we should just
approve based on our deterministic regulations. And
yet, you have - you wonder, is this risky anyway.
And we put in place a process for questioning that,
and it's - without going into a lot of detail, it's
sort of a three step process where we first ask
ourselves does this represent a special circumstance?
MEMBER POWERS: I mean, I understand what
you have to do if you want to go to the licensee, or
you want to factor it in. I just want to know. I
haven't decided yet whether this is risky or not. I
don't know.
MR. BARRET: They would come to our staff.
They should come to our staff, the Probabalistic
Safety Assessment Branch, and get a read on it from
us.
MEMBER POWERS: And you do that by running
a SPAR model, or -
MR. BARRET: Well, you know, it depends on
the question. You know, there are - as you all know,
there are many questions where -
MEMBER POWERS: I'm in the last bastion of
the risk ignorant. All I know is there's a thing
called risk, and I want to know what it is for my
particular plant and things like this. I haven't got
a clue what question to ask.
MR. BARRET: Right. And you have the
license amendment in front of you, and I - you know,
as you know, a risk analyst can kind of look at an
issue and pretty quickly get a sense of whether it
tends to be risk significant, or it tends not to be
risk significant. And it may not go any farther than
that. It may be that we could say right off the bat
that this doesn't - this is not in the range of the
risk significant -
MEMBER POWERS: Okay. Suppose the
licensee says gee, the NRC is making me inspect the
upper heads of my reactor vessels all the time. And
boy, that's a super pain to do because I've got all
this insulation on there. What I'm - the licensee is
proposing to do, is he's going to change out this
insulation that he's got for this new micro porous
insulation that's kind of elevated above a head, and
you can get it off real easy, and do this inspection
all the time.
How do - and so the Project Manager wants
to know whether that's - that has any risk
significance or not, because I mean, it's change but
it doesn't seem like a very big change. I mean, one
insulation for another. How does he find out?
MR. BARRET: He would come to us, I would
say, if he had that curiosity. And we would look at
the issue, and you know, ask ourselves some key
questions about what affect would this have on
initiating events, what would have on the availability
and reliability of systems? Would it have any affect
on operator actions, containment performance, you
know. And if there was some plausible impact on risk,
then we might look deeper. We might look at the SPAR
model -
MEMBER POWERS: You think your risk models
that you have model something as detailed as the
insulation on the upper head?
MR. BARRET: Off the top of my head, I
don't think that example would be in a PRA. In fact,
a great de al of what you might see in the way of
license amendments you would not find explicitly
modeled in a PRA.
MEMBER POWERS: And suppose I told you
that this micro porous insulation that's proposed to
use is extremely friable, and in a blow down system
will produce a lot of particulum?
MR. BARRET: You know, I think that would
be one of the questions that we would ask ourselves,
for instance. You know, we - we're cognizant of the
work that's being going on with strainer blockage and
the - now that we're working on the PWR sumps issue,
so that would be one of the questions. Sure, we would
ask ourselves that.
CHAIRMAN APOSTOLAKIS: Well, when we say
risk information, do we all understand the same thing?
I - the reason why I'm saying this is I was surprised
recently in talking to some industry people that they
don't think that uncertainty analysis is necessary,
and the staff is not using it.
MR. CUNNINGHAM: That would be a surprise
to me too.
CHAIRMAN APOSTOLAKIS: Well, I mean, if we
look at the petitions from the industry, the risk
informed things, are they doing explicitly risk - I
mean, uncertainty assessments, or is it a qualitative
discussion as it is in 1174, which says if you come
close to the boundary, management will pay attention,
will do something and thing about it.
MR. CUNNINGHAM: I suspect 1174 is -
today is as precise a characterization of how to deal
with uncertainties as we've got.
CHAIRMAN APOSTOLAKIS: So if I go with
1174, then I don't need to do it explicitly.
MR. CUNNINGHAM: It depends on what the
issue is, and what the decision you're trying to
achieve is.
CHAIRMAN APOSTOLAKIS: But does NRR, for
example, when you review a request, you look for
explicit statements of uncertainty, or is it a
qualitative discussion is good enough.
MR. BARRET: It's generally qualitative
discussion. I mean -
CHAIRMAN APOSTOLAKIS: Ahh, see.
MR. BARRET: Well, again I want to make
sure we're talking about the same thing. I mean, when
we say qualitative you want to know if you're close to
an edge, if you're close to a catastrophic change in
the picture, or are you looking at something where if
the temperature is five degrees higher everything
changes, or you know -
CHAIRMAN APOSTOLAKIS: Yeah, these are
sensitivities really, but I mean -
MR. BARRET: Some of them have to do with
margin, for instance.
CHAIRMAN APOSTOLAKIS: Yeah. Doing
standard uncertainty analysis, standard means that
you're right on something, is that kind of stuff.
MR. BARRET: Right.
CHAIRMAN APOSTOLAKIS: It is a trivial
matter these days.
MR. BARRET: Yeah.
CHAIRMAN APOSTOLAKIS: I mean, with the
computer programs that are available and so on. And
yet, you know, from talking to people I get the
impression that they don't think that that's something
that's necessary. And why is it not necessary,
because the staff does not request it. And that came
as a surprise to me. Now I know when we're developing
rules here, we're thinking about uncertainty all the
time.
MR. BARRET: Yes.
CHAIRMAN APOSTOLAKIS: But when it comes
to interacting with the licensees, evidently there's
a different philosophy.
MR. CUNNINGHAM: Again, there's an element
of -
MEMBER ROSEN: But later today we are
going to talk about the PTS.
CHAIRMAN APOSTOLAKIS: Yes.
MR. CUNNINGHAM: And then there's a
tremendous amount of thinking about uncertainty.
CHAIRMAN APOSTOLAKIS: Because that's us.
The industry is not submitting anything there. We are
doing that. And as I said, we are very sensitive to
that issue when we are developing studies ourselves.
MR. CUNNINGHAM: In a sense, that gets at
the point of what's the decision being made. The
change for an individual license, associated with an
individual license amendment may not necessitate that
sophisticated of analysis. The change of a rule that
could affect whether or not we have a PTS rule for a
dozen or 15 plants looking for life extension, license
extension is -
CHAIRMAN APOSTOLAKIS: I hope you could
resolve that in the ASME Standard. That was a major
issue of disagreement.
MR. CUNNINGHAM: I hope it is too. I'm
not - I suspect it is not.
CHAIRMAN APOSTOLAKIS: Is not. I mean,
it's true that we can do a lot of things without a
rigorous uncertainty analysis. It's very true.
MR. CUNNINGHAM: Yes.
CHAIRMAN APOSTOLAKIS: I don't doubt that.
The thing that's missing, like in many other places,
is under what conditions can you do that, under what
conditions can you do something else. And we don't
seem to be paying attention to these things.
MR. CUNNINGHAM: I guess conceptually the
three column approach in the ASME Standard is a step
towards trying to lay out when you could do very
simple, and when you need to do more sophisticated, or
very sophisticated analyses. Whether it accomplishes
it for - in this particular area, I'm -
CHAIRMAN APOSTOLAKIS: Well, it's been a
while since I saw that, but the I mean, the second
column said use mean values.
MR. CUNNINGHAM: Yes.
CHAIRMAN APOSTOLAKIS: And I don't know
how you can use mean values if you haven't done an
uncertainty analysis.
MEMBER KRESS: Yeah, and that -
CHAIRMAN APOSTOLAKIS: If you declare them
that they are mean values, then it's okay. All right.
MEMBER KRESS: That brings another
question to mind, and that is what we normally see
from the submittals are "best estimates", which they
declare to be a mean. They don't declare it, but it's
understood that this is a mean.
I've never seen a study by this
organization, or any other, that actually took what
would be a best estimate, which to my mind, you go in
with all the parameters that you can - as input and
part of the code that does it, and you try to pick
your mean values for those, and end up with the final
product, is my view of what that best estimate is.
I've never seen a study that really compared that
number to the real mean that you would get by
quantifying the full uncertainty. Is that anywhere in
your plan, because it seems to me like a key issue
these days. You don't really know what you're getting
from these things unless you have that.
CHAIRMAN APOSTOLAKIS: Exactly.
MEMBER KRESS: And I just don't see that
task in the plan anywhere.
MR. CUNNINGHAM: Okay. No, that -
CHAIRMAN APOSTOLAKIS: Sensitivity
analysis is used a lot.
MR. CUNNINGHAM: Yes.
CHAIRMAN APOSTOLAKIS: Which is really not
used, it's abused, so -
MR. CUNNINGHAM: I guess there's two
things to Dr. Kress' point. There were studies done
ages ago to look at that issue, and I don't know - I
don't even know that it was really documented very
extensively but, you know, this was perhaps 15 or 20
years ago. I can remember somebody looking at that
issue and saying the difference was a factor of two or
three in the value, if you will. Whether that has any
substance today, I don't know.
MEMBER KRESS: It would be useful to dig
that out. I didn't realize that existed.
MR. CUNNINGHAM: There had been work on
that ages ago, and I'm not trying to defend it or
anything.
CHAIRMAN APOSTOLAKIS: Yeah. And this is
part of our confusion. I mean, you don't know.
MR. CUNNINGHAM: Yeah.
CHAIRMAN APOSTOLAKIS: It depends a lot on
how complex a problem is.
MR. CUNNINGHAM: Exactly. Exactly. And
the characteristics of the underlying distributions,
and all that sort of thing.
MEMBER KRESS: But if I knew I was no
further off from the mean than a factor of two, for
example, I wouldn't worry much about it.
MR. CUNNINGHAM: Again, that was an
example done years ago, and probably with the WASH
1400 PRA Models, and all of the baggage that goes with
those, if you will. But the issue -
MEMBER KRESS: But is there a general
conclusion, like if you did the central estimate, you
end with the best estimate, can you make a statement
like you always end up with a number that's higher
than the mean?
CHAIRMAN APOSTOLAKIS: No, I don't think
so.
MR. CUNNINGHAM: No.
CHAIRMAN APOSTOLAKIS: I don't think so.
MR. CUNNINGHAM: I suspect if you did the
mean, you would -
CHAIRMAN APOSTOLAKIS: The kind of study
you want has not been done. I agree with Mark that
there have been pieces here and there.
MR. CUNNINGHAM: Yes.
CHAIRMAN APOSTOLAKIS: But the
comprehensive study that looks at that has not been
done, and I'm not sure it can come up with general
conclusions, because it will depend a lot on what
functions you're dealing with.
MR. CUNNINGHAM: Yeah. I think typically
you tend to see mean values being higher than - in
the few examples I can think of, the mean values to be
higher -
CHAIRMAN APOSTOLAKIS: The rigorous mean.
MR. CUNNINGHAM: The rigorous mean -
MEMBER KRESS: Has to be higher than this.
MR. CUNNINGHAM: Higher than the best
estimate.
MEMBER KRESS: Okay.
CHAIRMAN APOSTOLAKIS: Also, it depends on
how you handle the correlations and all of that.
MR. CUNNINGHAM: Yes. Yes.
CHAIRMAN APOSTOLAKIS: It is true though,
it seems to me, that the industry does not feel that
they have to do uncertainty analysis when they come to
you, or not to you, to NRR. And again, we see that
right now in the NEI document on Option Two. You will
find the word sensitivity many, many times, but not
uncertainty. And I don't know why they feel that by
putting everything at the 95th percentile and carrying
out the calculation is more meaningful than doing an
uncertainty analysis. I just don't understand that,
but we will discuss that with them when the time
comes.
MR. MAGRUDER: Okay. Let me move on -
CHAIRMAN APOSTOLAKIS: But does the risk
information, in your mind, include the uncertain?
MR. CUNNINGHAM: In my mind?
CHAIRMAN APOSTOLAKIS: Yeah.
MR. CUNNINGHAM: Every time. Yes, sir.
CHAIRMAN APOSTOLAKIS: So all these
documents will do that. Rich?
MR. BARRET: Well, I think there's a lot
of value in sensitivity analysis, and it tells you -
if you combine it with some sense of how wide, you
know, the variances might be. I mean, if I believe
that reliabilities might go down by a factor - or
unreliability might go up by a factor of ten, then I
do a sensitivity analysis around that estimate, I
think I've learned something from that.
CHAIRMAN APOSTOLAKIS: But it's not a
substitute for uncertainty analysis. I mean, you
learn something from sensitivity. In fact, if you do
a rigorous uncertainty analysis, and you can structure
your sensitivity analysis around that by not just
changing point values, but maybe changing
distributions and so on, which would be a much more
meaningful thing.
MR. BARRET: Right.
CHAIRMAN APOSTOLAKIS: But as a rule, I
don't think you guys require rigorous uncertainty
estimation.
MEMBER KRESS: And along that same thing,
Rich, I'm not sure I know or have seen any written
material of what constitutes a rigorous uncertainty
analysis. I mean, rigorous sensitivity analysis,
because what's usually done, you take various
parameters you think you might have a sensitivity to,
and you change them one at a time. Sometimes you
change all of them together.
MR. BARRET: Yes.
MEMBER KRESS: But I don't know what
constitutes a rigorous sensitivity analysis. Those
two don't do it for me at all. And, you know,
sensitivity is a whole output space.
CHAIRMAN APOSTOLAKIS: Actually, the place
will have done a lot of sophisticated sensitivity
analysis and performance assessment. There are all
chapters there where they do all sorts of things. In
fact, they are so sophisticated that simple minds like
ours have difficulty following what they are doing,
because they had, you know, statisticians develop, you
know, using the latest methods. So there is a gap, I
think, you know, between doing very trivial stuff and
very sophisticated stuff, doing something in between.
And for reactors I haven't see that, that kind of
analysis.
MR. CUNNINGHAM: Yeah, that's true. And
as you alluded to earlier, the PTS work that we're
doing is closer -
CHAIRMAN APOSTOLAKIS: It's closer -
MR. CUNNINGHAM: - on the scale to the
performance assessment work -
CHAIRMAN APOSTOLAKIS: Yeah, among us boys
again. I mean, I will -
MR. CUNNINGHAM: That's right. You know,
and as part of that project we're trying to sort out,
now how do you capture the importance of the
uncertainties, the relative importance of different
uncertainties in the process, and all of -
CHAIRMAN APOSTOLAKIS: For example, in
Option Two, we are categorizing systems and components
using the expert panel with a very input, being the
importance measures. Now the importance measures are
uncertain themselves.
MR. CUNNINGHAM: Yes.
CHAIRMAN APOSTOLAKIS: And that's nowhere
to be found.
MEMBER POWERS: George, one of the things
that you've raised is the relative weakness of
performance measures as a tool for understanding what
the risk assessment is telling you. Is anyone trying
to develop better tools for telling you what the risk
assessment is commenting on these things, since we use
- I mean, we have a lot of work going on now that
involves the categorization of things in both events
and hardware. And Professor Apostolakis has written
magnificently on why one should not attach great
significance to things like fusel vessely or risk
achievement worth, and risk reduction worth.
MR. CUNNINGHAM: Yes, there actually is.
There's some work going on in my group. Using as an
example, the categorization process used for South
Texas to say, you know, that as you've said before and
others have said, the fusel vessely and the other
importance measurements were designed for a particular
purpose, and now we're kind of using them for a
different purpose. And we're asking the question in
this project of given how things were done in the
South Texas example, if you will, is there a better -
an alternative formulation of an importance measure,
an importance calculation that might make more sense
given that application. We're doing some work in that
area right now, but - partially at Brookhaven and
partially at the University of Maryland. At some
point, it may be appropriate to come back and talk to
the Committee about that.
MEMBER ROSEN: I think South Texas would
have an interest in it, as well.
MR. CUNNINGHAM: That's probably true.
Yes.
MEMBER POWERS: Minor.
CHAIRMAN APOSTOLAKIS: Shall we go on?
MR. MAGRUDER: Sounds good. Let's go to
slide eleven, please. As Frank mentioned, the NRR
rule making group has been looking at how our rule
making, or what the next steps should be for risk
informed rule making, along with discussions with
Office of Research, obviously. And this is
preliminary information here, but one of the goals
that we think should have is to proceed with risk
informed regulations such that our rules start to
converge with the processes that we have in place.
We perceive that there may be a gap
between some of the activities in place, and the rules
to support the activities, and I'll talk a little bit
about that in a couple of slides here. And of course,
with the - we want to follow the principles that we
have laid out, dimension depth, safety margins, and
consistency with the safety goals that the Commission
laid out.
MEMBER WALLIS: Now I read this slide
ahead of time and tried to think about what it meant,
and I don't really understand it. And it seems to me,
you're always going to have some measure defense in
depth, you're going to have some round of safety
margins, but the regulations never tell you what
defense in depth is, how you measure it, or what
safety margins should be, or how you measure them, so
it seems to me that they need to be risk informed.
You need to - when you ask the question how much
defense in depth is necessary, that should be a risk
informed decision. And when you ask the question how
big should the safety margins be, that should be a
risk informed decision, so they're not on some other
plain or some other measure. And you should be risk
informing those ideas themselves.
MR. MAGRUDER: If we could go to the next
slide -
MEMBER WALLIS: Otherwise, you'll always
be arguing, or someone will always say well we need
more defense in depth. You'll never reach a
conclusion.
MR. GILLESPIE: I think -
MR. MAGRUDER: Graham, we're in violent
agreement.
MEMBER POWERS: Well, maybe I'm not. You
know, I worry. We create this intellectual construct
called risk information, which in examining it
closely, you find all these deficiencies. I mean,
there are uncertainties here, there are uncertainties
there. We don't know whether this is included. You
know, you go through and you do an uncertainty
analysis. George will just excoriate you because all
you've done is do parameter uncertainty, and you
haven't worried about model uncertainty, things like
that. And at some point, you have to ask what if I'm
just completely wrong about all this stuff? And I
think that that's where you start asking for defense
in depth. And if you try to justify defense in depth
based on the construct you're trying to protect
yourself from, you're going to get into a paradox
that's going to leave you vulnerable, I think. And so
I'd be very careful about using risk information to
guide my selection of defense in depth.
Now you will find on this august buddy
certain people called rationalists, and I encourage
them to think carefully about self-referencing sets
before they try to advocate the use of risk
information to guide themselves on defense in depth.
MEMBER WALLIS: Well, what you're simply
saying is that you've got to be more sophisticated
about how you interpret and use risk information. But
essentially, when you make decisions about defense in
depth, it's made to change the risk. And you're
making it - because - if you make it on some other
basis than using PRAs, it's because you don't believe
the PRAs, and that's information, that's risk
information too. You're still making decisions based
on -
CHAIRMAN APOSTOLAKIS: It seems to me that
Option Three provides an example of - I mean,
actually set certain criteria, you know, attribution,
prevention, mitigation quantitatively, so there is
some degree of quantitative judgment already
established in regulation, so are you referring to
that here, or -
MR. GILLESPIE: Yeah. If - let me ask
Stu to jump in. Let me jump to the problem we had.
The problem we had was -
MEMBER WALLIS: Well, Frank, you were in
violent agreement with me, so -
MR. GILLESPIE: Yeah, I am. No, actually
I'm kind of - we're groping with those same
questions. And, Dana, I think we're also in agreement
with you, which is why we started in the last
several -
MEMBER POWERS: You didn't frown at all.
MR. GILLESPIE: Which is why we started
this kind of construct. If you noticed, the first two
view graphs were a list of rules, and they were being
treated as independent rules. And I'll give you an
example. Top of a report in the Pass said you didn't
need a sampling system. Well, that was because we had
a sampling system required by the Hydrogen Rule. And
for a year and a half, the Hydrogen Rule people wrote
a rule that said you didn't need a sampling system
because the Pass people had it. And after about two
years of work, got these two people to talk to each
other, and we realized what we really needed was a
single sampling system, not two.
What's that evidence of? That's evidence
of when you start getting so many individual efforts
going, and you haven't fit them into any kind of
construct. And it's not that the one we've got up
here is the right one, it's kind of the first one that
Stu and I kicked around, and some smart people gave us
some input on, to try to start to pull things
together. And the next step in this is to fill in
some information. You'll see on there that we've got
the - under "Accident Prevention", we've got the
surrogate safety goal, if you would, of 10 to the
minus 4. What we did was just to put some plugs in of
where some things fit that currently exist. You could
see the oversight cornerstones in there. Those blocks
are in there.
We went to a slightly lower level just in
this picture, and the concept would be to try to
balance things like defense in depth, is to write an
objective for each of the blocks you would see, or
however this might evolve. It would be kind of a
regulatory objective. What's our objective? And we
took initiating events, and just chatting about it
with some brainstorming we said well, you know what,
there's transients and there's accident events, like
true events. What rules do we have that deal with
transients, which might be like station blackout,
things that happen that we don't necessary have all
the regulations on, but we have some. And what
regulations fit under each of these categories.
And then if you - and I don't mean by
whole regulation, by 50.46, or 50 - I mean by
subparagraph, so that you get the truly like things
that deal with phenomena together, recognizing that
one regulation like single failure criteria could
actually go across the board. And then when you
change something, if we change 50.44 you can ask
what's its affect on all the other objectives.
What we're trying to do is kind of link -
in DOT process is link our current body of
regulations, which people would agree are not
necessarily written in a risk informed context,
although safety was in the context of the minds of the
writers, with some set of risk informed objectives.
And I say risk informed objectives because, just
because an initiator is so low, doesn't mean you
shouldn't have a mitigation function for it. So what
we're trying to do is provide a structure in which to
consider defense in depth. It's not totally driven by
the PRA, it's driven by some of our deterministic
thinking which we might not have totally let go of.
But the important piece of where we're going with
this, or where we think we might be going if it
matures further, would be to write objectives for each
of these blocks, sort the rules, the pieces of rules
under this, and say now let's look at it. Is that
rule needed to meet that objective?
The important piece, as we found in the
ROP, was coming up with a consensus objective of what
is containment there for. And coming up with that one
or two sentence objective was, once we did that in the
ROP program, the rest was easy. The rest was actually
quite easy. It was implementation.
So in trying to organize our thought
processes, this was our first cut. You can see under
the LERF box, we've got less than E minus 5. Well,
that's not a conditional number, we recognize. That's
kind of an absolute number, but there's also some
conditional numbers, and we didn't - Stu didn't have
a chance to find it, but there used to be like a
conditional number on containment that was thrown
around in some literature in the agency also. And
so - I think it was .1, so what we're doing is right
now is kind of searching around -
MEMBER KRESS: That was for the full
conditional failure.
MR. GILLESPIE: Full conditional failure,
yeah. So what we're doing now is searching around for
all of this policy guidance and saying how can we
create a structure and fit it in? How can we then
take the next step and create an objective for each of
these blocks which is risk informed, and it may be
qualitative, it may be a number, it may be both. And
we picture kind of a database - a spreadsheet with
little Xs as a starting point to index our thinking
relative to grouping the regulations, and then say
does this regulation - is it needed to meet that
objective? Can these two be combined with a more
performance oriented wording, like the example I used
on sampling systems, to meet the objective?
If I take this rule out, am I more
dependent on the other rules for meeting the
objective? And it's those kind of questions that
we're running into as we're dealing with things in
isolation one at a time.
This doesn't compete, by the way. You
prioritize what you work on by the four agency goals,
which is what's in the plan.
MR. MAGRUDER: Right.
MR. GILLESPIE: This is a thought process
to help us make sure we structure things. And if we
destroy a rule, we know its impact is a synergistic
affect in the whole. Let's skip to the next slide.
MEMBER SHACK: Didn't the research try to
do that with their framework document?
MR. GILLESPIE: Well, we did. We took -
we scavenged a lot, you'll see in here, from the
research document which was used to prioritize Option
Three. But this thought process is the body of rules,
and it's kind of independent of Option Two/Option
Three. And we're not using it to prioritize. What
we're doing is using it kind of like an analytic
approach, or an analytic tool to say do we really know
what we're doing when we change this rule, or this
paragraph, and know what its impact is overall, in
kind of a risk informed structure. I'm not saying
it's perfect, but it's a risk informed indexing, and
now let's go in and change things, but let's
understand how the impact is. Let's have something
that helps us display and understand the impact.
MR. MAGRUDER: Yeah, this is - I guess
this is kind of an expansion of the framework
document.
MEMBER WALLIS: Have you used this for
something like 50.46? And I can see looking at - I
can see doing it with one regulation like 50.46. The
difficulty though is, how does that relate to all the
other regulations because I don't - bringing all the
regulations into this kind of a box you just showed us
is going to be very difficult.
MR. GILLESPIE: Well, I think the first
cut is not that difficult. It's a matter of just
sitting down and taking the various subparagraphs, and
separating them in some logic manner, grouping the
phenomenological things together.
MEMBER WALLIS: But they're also dependent
on all the other regulations, so the interaction is
not going to be -
MR. GILLESPIE: And that's what I'm
concerned with, is we're trying to get a handle as
those interactions. When I destroy a rule, have I
become more dependent on the rest? And so this is our
thought process. I picture the next thing being the
bottom row of blocks here along the top of a database,
and all of our regulations in some logical subset,
because you can't deal with them in big pieces.
Otherwise, you've kind of got to get them down, and
also recognize underneath each one of those little
paragraphs in the regulations is a big body of
guidance.
I'm not saying it's simple, but what we're
trying to do is get our thought processes away from
thinking in isolation, and thinking in context, and
take advantage of some of the thinking that went into
the ROP and its development. And going to the next
step, I'll tell you what I think the importance of
being able to articulate what the objective of
containment is, what the objective of protecting the
fuel is, or protecting the primary circuit. What are
our expectations? Is in the next diagram -
MEMBER KRESS: Before you go to that
Polish firing squad diagram, I have a comment about
this one I'd like to -
MR. GILLESPIE: Remember, this is only me
and Stu, and Cindy.
MEMBER KRESS: If you look at the row of
initiating events, mitigation, containment, emergency
planning, that's sort of a shorthand for PRA.
MR. GILLESPIE: Uh-huh.
MEMBER KRESS: That's what goes into a PRA
when you calculate the risk. And what you - I
understand your thinking is that you're going to look
at the body of regulations you now have to see where
you deal with these things, and see how you might risk
inform that part of it. The problem - the partial
problem I have with that is, the assumption - there's
an implied assumption there that what you now have is
the right balance among those things. Let's talk
about balance as a defense in depth concept.
MR. GILLESPIE: Okay.
MEMBER KRESS: Now I'm going to reveal my
rationalist viewpoint here, but what - when I look at
this, and if I did this in a PRA for each individual
plant, I would get different contributions along those
from each plant to establishing its risk status at -
with risk, some thing at the end of the PRA. And I
would get different numbers, different contributions
for those depending on the reactor, and the type, and
so forth. And so, I have to ask myself, what is the
contribution among those that I find acceptable from
the defense in depth standpoint? That's a question I
ask myself over and over. And I've never seen a
rationalization of any kind of criteria, other than
what we already have, which is kind of 10 to the minus
4, with nothing - with no sequences that really -
outstanding versus - and 10 to the minus 5. Those
are the only two. Why are those appropriate in my
mind, and what do they have to do with the
uncertainties in the determination of each point along
the line?
Now as a rationalist, I would say when I
ask myself what if I'm wrong as a defense in depth
concept? What if I'm wrong, and how do I accommodate
that? I rephrase the questions and say what
confidence do I have in my answer in the PRA? And
that gives me a measure of how much I think I'm wrong,
if you do the uncertainty wrong. So I think the
rationalist approach can accommodate a structurist
thinking, because I think you can answer that
question, what if I'm wrong, to some extent. And how
wrong am I, and then I can have a handle with which to
tie how much defense in depth I need, or where does it
need to be put. And I don't see that in here at all,
that handle.
MR. GILLESPIE: I agree with you, because
that's - kind of our next step would be - and let me
go back to what I said. No matter how low you can
force initiating events down in the mathematical
modeling, mitigating the core damage frequency
accident is still a requirement, which means you need
to set up a set of objectives. This is what we're
grappling - this is why you don't see an objective
written down there. It's just a title right now, is
independent of how good you can make your plant, it
doesn't matter. WE still expect this, this is being
able to mitigate the accident, and this in being able
to contain it. And I think some of that thinking went
into writing the regulations over the years, but no
one wrote it down. It wasn't in a structured,
necessarily, way. And what I'm suggesting is this how
we and the staff are trying to at least get some
structure to our thinking, to start putting those
questions on the table, because that would get to,
well what is the objective of mitigation systems?
CHAIRMAN APOSTOLAKIS: Now I have a
problem here. At 10:00 we have the Officer Directors
coming, and you, gentlemen, have to wrap it up in two
minutes.
MR. GILLESPIE: If I could just jump to
the next slide -
CHAIRMAN APOSTOLAKIS: Tell us what the
most important thing is from your presentation.
MR. GILLESPIE: Okay. The next slide -
MEMBER ROSEN: You're going to wrap up the
entire discussion, or just the reactor arena
discussion, because we had two subjects, and we've
talked about -
CHAIRMAN APOSTOLAKIS: This section is
being wrapped up in two minutes.
MEMBER ROSEN: We talked about half of
what we came to listen to.
CHAIRMAN APOSTOLAKIS: Well -
MR. GILLESPIE: Let me jump just to the
next slide so I can wrap up our piece, and why I think
the first slide is important. The first slide, which
sets up a set of objectives, basically starts setting
up a standard. And in this slide you see some of the
things going around the outside that we've been doing
kind of in isolation. And one would ask, why does the
ASME Code have a low categorization in ISI and IST,
which could be different from Option Two's Risk Three,
and why are they treated different? And what's the
relationship between that and configuration management
under A-4, which also has a scheme for having
something that's called just leave it to the skill of
the trades? And why are all those thresholds
articulated in a different way, all in different
places? And this is an endeavor to say as we're
becoming more risk informed, we need to bring all of
these things we've put out in the last three or four
years together, and start using the same thresholds if
they're, in fact, supposed to represent the same
safety level of action or inaction. And we would see
- the first picture I showed you is actually fitting
in, is kind of being the common risk informed
objectives that would fit into this kind of central
wheel.
It does mean going back and looking at how
we articulated some other things in the past. For
example, on A-4, because the way we did it might not
have been the right way. We might be more informed by
some of the things we're doing in Option Two right
now, is we do need to bring these things together, and
a central set of objectives is one way applying it
across the board to do that. Creating those
objectives, I recognize, is going to be difficult
because defense in depth is a balance. How do you
consider uncertainty? But I need a structure to
answer those question - to even ask those questions,
and I'd like to have a structure I can put the
questions in, so that when I get the answers, I know
how they fit together. And that's what we're trying.
I just thought I'd put this on the table as - it's a
little different.
MR. MAGRUDER: I'm sure we'll talk a lot
about this, but we need to turn over to Lawrence.
CHAIRMAN APOSTOLAKIS: I don't know for
how long.
MR. KOKAJKO: I can do it very quickly. My
name is Lawrence Kokajko. I'm the Section Chief of
the Risk Task Group in the Office of Nuclear Material
Safety and Safeguards. I have not appeared before the
ACRS before, but I have appeared before the ACRS/ACNW
Risk Sub-Committee, and so some of what I'm going to
talk about, they have already heard.
Just very briefly, NMSS through the RIRIP
has embarked on a way to modify the regulatory
framework across a spectrum of regulated activities,
all the way from small seal sources devices, all the
way through spent fuel, storage, transportation, fixed
gauges, the gaseous diffusion plants, fuel fabrication
facilities, and others. It's not the homogenous group
that NRR is. And consequently, we have to be a little
more creative in how we risk inform the regulatory
framework.
MEMBER ROSEN: I'd say it's even less
homogenous than the reactor safety arena.
MR. KOKAJKO: Even less.
MEMBER ROSEN: Yes.
MR. KOKAJKO: We like to think we're the
more interesting program in all this. We have gone
through conducting eight case studies recently. We've
developed some final screening considerations, and
implementing guidance to help us to determine what is
amenable for risk informing within the office. We've
initiated staff training like NRR has, and we're
looking forward to training, perhaps - having
advanced training later on for the staff. And we've
completed a number of our activities in December.
We are now looking at implementing a Phase
Two approach, where we're looking at what is amenable,
and looking at cross-cutting measure. I believe
Margaret Federline has appeared before you some months
ago, and she indicated that.
We are going to categorize any
improvements through the PBPM process and prioritize
them to see which ones are most effective.
CHAIRMAN APOSTOLAKIS: Through what
process?
MR. KOKAJKO: PBPM, Planning, Budgeting,
and Program Management Process. It's the way to
prioritize the - what we should do next. Although
this work is ongoing, we will implement it in a Phase
Three, there are things that are going on within NMSS
right now, which we're taking advantage of.
A couple of things that we will - that
are being worked on is changing the inspection manual
chapter for the fuel cycle facilities to be risk
informed. Also, for Uranium recovery we're doing the
same thing. Part 72 Geological and Seismological
Siting Criteria in SFPO is - will be risk informed.
It will also match what is going on in the reactor
arena.
CHAIRMAN APOSTOLAKIS: In June you're
going to issue another plan, version of the plan.
Right?
MR. KOKAJKO: Yes, sir.
CHAIRMAN APOSTOLAKIS: And we will meet
again?
MR. KOKAJKO: I hope, yes.
CHAIRMAN APOSTOLAKIS: So then we'll start
with you.
MR. KOKAJKO: Okay. That would be fine.
Hopefully by June, we will have some other products
available.
CHAIRMAN APOSTOLAKIS: Good.
MR. KOKAJKO: The other thing I'd like to
say is we are working on developing draft safety goals
with the Office of Nuclear Regulatory Research. The
Sub-Committee, we've presented the three-tiered
approach, and I think it was received pretty well.
We'll also be going to PSAM in June, and
we're looking forward to that.
CHAIRMAN APOSTOLAKIS: You'll do what in
PSAM?
MR. KOKAJKO: We're going to make about -
I think we're going to have about 12 presentations at
PSAM in June.
CHAIRMAN APOSTOLAKIS: Okay.
MR. KOKAJKO: That's it in a nutshell.
CHAIRMAN APOSTOLAKIS: You choose good
conferences to do it.
MEMBER ROSEN: Well, it certainly was
quick. Right?
CHAIRMAN APOSTOLAKIS: But no, we will
have another occasion to discuss these things -
MEMBER ROSEN: I think this is a very
important area.
CHAIRMAN APOSTOLAKIS: - in the next few
months.
MEMBER ROSEN: And we gave the full
Committee, really a very short shrift of it, and I
think it deserves a lot more discussion.
CHAIRMAN APOSTOLAKIS: If we can only
settle the issue of defense in depth, then everything
will be going very quickly at these meetings.
MEMBER POWERS: It'S very well settled.
We know what it is.
CHAIRMAN APOSTOLAKIS: Dr. Shack.
MEMBER SHACK: It's back to you. I guess
we're done.
CHAIRMAN APOSTOLAKIS: Thank you very
much, gentlemen. We'll recess until 10:00.
(Off the record at 9:52 a.m.)
(On the record at 10:04 a.m.)
CHAIRMAN APOSTOLAKIS: We're back in
session. Well, on behalf of the ACRS, I would like to
welcome the Executive Director for Operation, Dr.
Travers. Director of the Office of Research, Ashok
Thadani. Director of the Office of -
MEMBER SHACK: It says NMSS in there,
doesn't it?
MR. COLLINS: I got a promotion on way
over.
CHAIRMAN APOSTOLAKIS: Nuclear Reactor
Regulation, Mr. Sam Collins. And the Director of the
Office of Nuclear Material Safety and Safeguards, Mr.
Marty Virgilio. We're looking forward to discussing
items of mutual interest with EDO and the Office
Directors, and I understand this is the first time
that either committee has had such an opportunity to
exchange ideas and thoughts on both ongoing and future
high priority activities with the Directors. I
understand Dr. Garrick, Acting Chairman of the ACNW
also has a few opening comments.
MR. GARRICK: Thanks, George. Good
morning. My name is John Garrick, the Acting Chairman
of the ACNW. The reason that I'm Acting is that
neither our Chairman, George Hornberger, nor our Vice
Chairman, Ray Wymer, could join us today. They do
send their regrets. I am joined by the very able
Committee Member, Milt Levenson, and he will - he and
I will have to do the best we can to represent the
Committee.
I would like to add my welcome to that of
Dr. Apostolakis. This happens to be an extremely
timely event, because the ACNW will be holding its
retreat, its planning activity later this month, and
one of our goals is to select a limited number of high
priority issues for the Committee to concentrate on
the next year, and beyond. And these presentations,
I'm sure, are going to be very helpful in that whole
selection process, to ensure that the Committee and
the NRC Staff share the same high priority issues, so
we look forward to that very much. Thank you.
CHAIRMAN APOSTOLAKIS: Bill.
MR. TRAVERS: Thank you very much. Mr.
Chairman, a limited number of high priority issues is
a goal we have. We haven't managed to get to that
point yet, but we do appreciate the opportunity to be
here before both committees. There was a time in
NRC/AEC history when we used to report to a joint
committee of Congress, and some look forward, or look
fondly on those days, but we are certainly happy to be
here today with you to talk about some of our current
issues, some of the issues that you've been
addressing, and have a discussion, a dialogue on some
of the activities that we see, and perhaps you see
going forward.
Each of the Office Directors with me
today, and I should mention that Bill Kane and Carl
Paperiello, my Deputies, are here as well, are looking
forward to a dialogue. We have a presentation. I'd
like to make just a few brief comments at the outset,
and leave plenty of time for the dialogue to following
after the presentations.
I have to tell you that we are in the
midst of an awfully dynamic time. I think you
recognize that. It's, in my estimation, probably one
of the most dynamic periods in NRC history.
Certainly, we've had some of those in the past, after
Three Mile Island, and at other times. But if you
look across the spectrum of activities that we are
involved in, and certainly you are involved in in your
role, it is a daunting scope, and we are anxious to
continue to make progress in many areas.
Let me begin by just touching on some of
the areas where I think we have been particularly
benefitted by activities of both the ACRS and the
ACNW, and just touch on a few of those. I think we'll
probably touch on some of these in the presentations
to follow, as well.
Certainly, in the area of advanced or
future reactors, ACRS has been significantly involved,
and we've been the beneficiary of some of your
thinking in that regard. The ACRS workshop in June of
2001 on the regulatory challenges for future reactor
design was helpful. I've seen Dr. Powers' report
recently on the PBMR, which raised a number of issues
that I'm sure will be of interest as we likely get
into further reviews on that reactor design.
There have been - there has been
participation by at least a member or two on the
research sponsored workshop on the high temperature
gas cooled reactor safety and research issues in
October of last year. In the area of core power
uprates, the ACRS has been a contributor in our review
of license amendment requests. And we'd like to note
that your positive support for the reactor fuel
confirmatory research program is helpful, as well.
As I said, I'm going to mention just a few topics.
There are many more.
Turning to ACNW -
MEMBER POWERS: Well, before you advance,
could you touch upon those where you think the advice
from the Committee was of limited help?
MR. TRAVERS: That's an interesting
challenge. We like to take all -
MEMBER POWERS: I know you want to put a
positive spin on it -
MR. TRAVERS: That's right.
MEMBER POWERS: - but, I mean, telling us
what we've done good, it's nice and we appreciate it.
But we're not going to get better if you don't bring
to our attention the things that haven't been helpful.
MR. TRAVERS: Well, let me propose that I
start with where we've gotten good things.
MEMBER POWERS: Okay.
MR. TRAVERS: And perhaps the more
appropriate place to go further than that would be in
the discussion that follows, if that's acceptable.
CHAIRMAN APOSTOLAKIS: If, of course,
there are any.
MR. TRAVERS: In the area of - let me
continue positively, if I may. In the area of license
renewal, the ACRS has been involved in the review of
the generic documents that have been produced,
including the Sandia Review Plan, in our efforts to
institutionalize some of the lessons learned in the
generic age and lessons learned document.
I was going to mention in the ACNW's case,
a number of positive interactions have occurred as
well, including the input on the Draft Policy
Statement on Decommissioning Criteria for the West
Valley Demonstration Project. This has really helped
us in our sense to identify areas where clarifications
that should have been made, were made.
We've had interactions, as well, with ACNW
on the development of the Decommissioning Standard
Review Plan. We think those have been particularly
productive. And I'd have to comment on our very
positive view of the recommendations that ACRS made,
and ACNW made separately, on the safety and waste
research plans in the Office of Research.
As you know, we largely agreed with the
key recommendations in those reports. And, in fact,
we've incorporated many of the ACRS' recommendations
into our planning and budget process for the outlying
years.
One thing that I personally would like to
thank you for, and in concluding my opening statement
is, your efforts at my request to look at a differing
professional opinion, a rather complex one that had
been under review for quite some time, we took the
rather unusual step of asking the Committee if they
would act as a technical review of this issue.
We take these issues that are raised by
our staff very, very seriously. And we think that the
effort that you put into the review of those technical
issues was very well done, and has helped not only put
some closure to that issue, but identify a path in
some activities moving forward to help further assure
ourselves with confidence that we have, in fact, those
steam generator issues well covered.
MEMBER POWERS: Let me interject. You're
- just to remind everyone that the only reason it was
possible for the ACRS to arrive at any conclusion in
that area was the fact that your staff and the
differing professional opinion authors could provide
such effective support, and effective presentations,
forthright, frank, complete description of their
various technical opinions, in a very clear fashion.
When we get that kind of support from your staff, I
think the Committee is far more effective, than if we
have to delve into things, and try to find them for
ourselves. We're not very good at that.
MR. TRAVERS: I should ask you for areas
where we haven't done well, but perhaps we could take
that up -
MEMBER POWERS: Well, we'll go into a
couple of them by day's over.
MR. TRAVERS: Well, I think the last thing
I'll mention in terms of our interactions with ACNW,
and certainly things we see moving forward, are the
interactions that relate to our identification of the
key technical issues in the Yucca Mountain project.
I know the Commission has asked for some insight from
ACNW on these issues. WE're happy to work with you on
a roll-up of those things, a discussion of the
significance of the various items that are on that
list, and talk about a path moving forward.
So with that sort of brief and positive
beginning, why don't I turn to Marty Virgilio, who is
going to begin a presentation that we have planned in
each of the program offices. Marty is going to talk
about, obviously, NMSS programs. Ashok is going to
talk about research, and Sam will, aside from what it
says on his placque, he will talk about the Office of
Nuclear Reactor Regulation. Marty.
MR. VIRGILIO: Good. Thank you, Bill.
Good morning, and thank you all for this
opportunity to meet with you today to discuss some of
NMSS' current and ongoing activities that we consider
of high priority, and the ones that we value your
input on, continuous dialogue and advice.
There are a number of high priority issues
for NMSS that we'll be discussing with you today.
Many of these issues represent what we consider
technical resolutions to first of a kind applications
for radioactive waste transportation and storage, and
disposal.
New standards are being developed around
these issues, with what I consider worldwide interest.
Compliance with these new standards are being
demonstrated for the most part through modeling, with
assumptions where we lack empirical data. That makes
it rather challenging for all of us, I think. And
there's a high degree of public interest in all of
these activities.
MEMBER POWERS: Actually, that easy to do
it. If you've got no data, and you just have to live
on assumptions, those kinds of models are great to
develop.
MR. VIRGILIO: Difficult to defend.
MEMBER POWERS: Difficult to defend.
MR. VIRGILIO: Right. We acknowledge that
you've already provided us good advice in a number of
areas involving high level waste, as Bill alluded to
in his opening remarks, particularly on the KTIs. And
we appreciate your continued feedback to us on the
total system performance assessments, and some of the
issues that we're dealing with there. And we just
recently got a memo from you dated January 17th on
this topic.
In the interest of time, I want to focus
on just a few of the current and future issues that I
see as the most significant, so if we could have slide
two, please.
Today, these represent the four issues
that I want to spend my time on, your time on. These
are activities in the nuclear waste safety arena, and
related programs that we believe will require
continued ongoing discussion and consultation with,
between the NMSS Staff and the ACRS and ACNW.
The first pertains to high level waste and
repository issues, particularly the resolution of the
KTIs, and subsequent performance confirmation.
There's ongoing and increased interest in this area
involving waste package, transportation safety issues,
and our reviews in this area, and some of the other
studies that are planned and ongoing, and I'll touch
on them.
The second activity involves
decommissioning and site cleanup issues, as well as
technical issues related to demonstration of
compliance with the license termination rule.
The third activity pertains to
enhancements of NMSS' risk informed approaches. And
the fourth area includes anticipatory and confirmatory
research areas in the waste arena.
There's this over arching need that I just
want to make sure that you're sensitive to within
NMSS, that we continue to make continuous improvements
to our program, that we continue to challenge
ourselves to seek most efficient and effective
solutions to the problems that face us. And I just
want to make sure that you're as sensitive to that as
we are. If I can have slide three, please.
DOE now has officially announced its
intent to recommend Yucca Mountain to the President,
and so now we continue to prepare for the license
application in light of that announcement. There are
a number of important activities underway today with
the staff, and you have interacted with us on several
of these. First, the key technical issues.
The focus of the key technical issues and
related agreements, when I think of those together, is
to make sure that when we finish that, we have
provided DOE guidance on the information they need to
submit a sufficient license application. The scope
and level of detail is based on what would be needed
to provide the requisite confidence regarding
demonstration of compliance with 10CFR Part 63.
The performance assessment approach which
is embodied in the Yucca Mountain Review Plan has been
used to derive the risk insights for prioritizing and
integrating these key technical issues. The key
technical issues themselves add a varying complexity.
While DOE must satisfactorily address the
KTIs to prepare a sufficient license application, it
must take into account the risk significance in
defining the scope of its response on each of these
agreements.
The relative importance of these key
technical issues and related agreements may be
qualitatively assessed using a combination of factors,
such as the risk significance of the associated
structures, systems and components, and the processes,
the number and complexity of the agreements associated
with each of the KTIs, and of course, stakeholder
issues and concerns, as well.
The agreements that we've reached with -
between NRC and DOE in the pre-licensing application
process are based on extensive staff review of DOE's
technical case, and subsequent identification of gaps
in DOE's supporting information. So our future
technical meetings with Doe will continue to focus on
these gaps and provide strategies and answers that
will help close each of these gaps.
MEMBER POWERS: You guys prepared and
developed a program of, I hesitate to call it, so much
research as technical investigations let's call it,
because I think it was a mixture of research and
technical support, around a set of what you call KTIs,
Key Technical Issues. And I think the Committee wrote
a report some time back, saying gee, what a great idea
this is, and how useful this is. They very much liked
it. Is that kind of an approach still going on? I
mean, do those KTIs evolve, or they - they're the
ones that you set up a long time ago, and you see no
reason to change them?
MR. VIRGILIO: The KTIs have not evolved.
What I think are some of the sub-issues, and how we
are focusing on the sub-issues that I think have
evolved is we've gained more information and more
insights about the site. So if you look at the ten
KTIs, or nine technical plus the performance
assessment, they have held constant for the - you
know, for as long as I've been associated with the
program. But what has evolved is our thinking around
some of the sub-issues.
MEMBER POWERS: Uh-huh.
MR. VIRGILIO: While the titles haven't
changed, I think the way we've looked at them, the
scope and depth, the amount of energy we've put in
them has changed based on our assessment of their
significance, and how that's changed as we've gotten
more information, as we've worked through these
issues.
MEMBER POWERS: Have you thought about
applying that kind of an approach to things like the
transportation issues? I mean, many of our
transportation regulations and approaches for safety
in transportation are fairly geriatric, and people are
interested in perhaps upgrading those. Have you
thought about applying that in that same area?
MR. VIRGILIO: No, I think we're going to
approach that from a different perspective. What I'd
like to do there is take a more risk informed
approach, instead of setting out at the front end
saying that there are this set of a dozen particular
issues that I want to focus on.
What we're working cooperatively with
research today, is to get more insights around
storage, dry cast storage, and working on
transportation issues as well, in a more risk informed
approach right from the beginning, and focus our
attention based on what the risk information is
telling us. Where we have large uncertainties, for
example. Well, overall the system might not be posing
a large risk, the value of doing these kinds of
assessments is to tell you where you've got large
uncertainties, where you've got margins that you might
be able to focus on, and where you've got maybe small
margins -
MEMBER POWERS: So instead of KTIs, you'll
have LUIs, Large Uncertainty Issues and things like
that.
MR. VIRGILIO: I don't know -
MR. THADANI: Then I will touch on this
issue -
MEMBER POWERS: Good.
MR. THADANI: - when I brief you about
what we're doing, and it's really basically along the
lines of what Marty is saying. And one of my issues
is going to be, this is an area where we're going to
interact with you, and make sure that if there are any
issues we're not considering, that we have the benefit
of your thoughts on that.
MEMBER POWERS: Well, the - I just
comment that the KTI approach, since I have limited
overlap with the issues of Yucca Mountain, was an
extremely effective way, I think, of persuading me
where the research needed to be done, because it had
been - it was systematic and whatnot. Doing it with
risk I think is no less systematic, and maybe more
justifiable, but the articulation of these things was
just a very effective -
MR. THADANI: Since you're on this issue,
the approach we are using, of course, is phenomena
identification ranking table type of an approach, so
I think it's fairly systematic.
MEMBER POWERS: Yeah. I mean, a similar
sort of thing.
MR. THADANI: Yeah.
MEMBER POWERS: Yeah. That - I have to
say that that - your staff has been effective in
taking a concept focused largely in thermal
hydraulics, and seeing how they can apply it in other
areas.
MR. THADANI: Yes.
MEMBER POWERS: And it's very impressive.
I mean, that - I think when people ask is there
creativity in the NRC Research Program, that's one of
the areas I'd point to.
MR. TRAVERS: I just have to comment on
one thing mentioned, and one part of your statement
indicated that there are those who would like to see
changes in those requirements, and some would like to
see risk informing. There are others, however, who
really aren't looking for change in the regulatory
scheme, so sometimes there's a balancing of what we're
doing in the context of stakeholder interest on the
part of the industry.
MEMBER POWERS: Making that judgment is
why you get the big bucks, sir.
MR. TRAVERS: It's a balancing act at
times.
MR. VIRGILIO: Just to close on the KTIs,
I wanted to recognize that there's international data
and experience that we're trying to draw on as we work
forward on the high level waste repository issues. I
believe that the NRC Staff and the Committee, as well,
needs to utilize the experience gained by the Finns
and the Swedes to the maximum extent possible, so not
only leveraging our dollars, but also leveraging our
decision making, and I think it has an opportunity to
enhance public confidence, as well.
That's really all I wanted to say about
that issue. If we could move on to slide five,
please. There's been increased national attention on
spent fuel transportation and storage issues. These
have been stimulated by the Baltimore Tunnel fire.
CHAIRMAN APOSTOLAKIS: Why did you skip
four?
MR. VIRGILIO: Oh, I'm sorry. I think I
covered it. It was all covered in my notes. I'm
sorry if -
CHAIRMAN APOSTOLAKIS: Well, I've got a
question on the fourth.
MR. VIRGILIO: Sure.
CHAIRMAN APOSTOLAKIS: Can we go back to
four?
MR. VIRGILIO: Yes, please.
CHAIRMAN APOSTOLAKIS: This issue of
uncertainty and realistic assessment and all that
stuff, I'm pretty sure you have significant model
uncertainties in the performance assessment, and we do
also in reactors. Now the way we are handling them to
the extent that they can be handled is using defense
in depth. How do you guys do it?
MR. VIRGILIO: We also use the defense in
depth -
CHAIRMAN APOSTOLAKIS: Is that defense in
depth, really?
MR. VIRGILIO: - approach for the
repository. That is part of the process.
CHAIRMAN APOSTOLAKIS: Multiple barriers,
and that's it?
MR. VIRGILIO: Multiple - yeah, an
approach that looks at multiple barriers, that looks
at the fuel, the waste package, and it also then takes
into consideration the repository, the transport.
Each of those provide a certain measure of defense in
the process.
CHAIRMAN APOSTOLAKIS: And people compound
this -
MEMBER POWERS: It's a good approach,
George.
CHAIRMAN APOSTOLAKIS: Huh?
MEMBER POWERS: It's a good approach,
defense in depth.
CHAIRMAN APOSTOLAKIS: Well, I'm not so
sure. I think it's a very different application of
defense in depth in barriers.
MR. VIRGILIO: Yes, it is. We've written
several papers on it that we'd be happy to share with
you, if you -
MEMBER KRESS: The only variable you have
access to there in terms of what you can do is the
cask. I mean, you've got the fuel. It's already put
into some sort of form. You've got the repository
external, so the defense in depth you have access to
is what you - how you design the cask. Now my
question is, how do you know how good of a cask you
define to get the appropriate defense in depth you
need?
MR. VIRGILIO: It is not limited alone to
the cask. I think you have to look at the entire
system. I think there are things that you can do with
regard to how you store the waste, how you back fill
behind the waste. There are a number of variables
that you have in addition to the natural barriers that
are provided. Am I sorry. Your question then was?
MEMBER KRESS: My question was how do you
know when you've got enough of that, including those
other things?
MR. VIRGILIO: Through tests and through
modeling.
MEMBER KRESS: You have a criteria for how
much change in some risk measure that you want this
cask to give you?
MR. VIRGILIO: You can do sensitivity
analysis, and you can do modeling in terms of - you
know, what would be the affect of you decrease the
performance of the cask, for example. Take the
package, and so you make some assumptions about how
well it's going to perform. You can model and
decrease, you know, the performance around a
particular element.
MEMBER KRESS: The performance is measured
by some release somewhere, or some contamination -
MR. VIRGILIO: By the standards that have
been established and incorporated in our rules. The
standards are established by EPA. DOE brings forward
the license application. NRC assesses that
application, so there are a number of federal agencies
involved in this. But EPA has set the standards that
really are looking at what the affect might be on some
hypothetical resident in the vicinity of the facility.
MEMBER KRESS: So you do sensitivity
analysis to -
MR. VIRGILIO: Yes, in part to understand
the contribution and affect.
MEMBER KRESS: Contribution, and do you
know - when you do a sensitivity analysis do you -
what? Put some parameters at their 95 percentile
values or something like that?
MR. VIRGILIO: And vary them? Yeah.
MEMBER KRESS: Vary them.
MR. VIRGILIO: As to what degree of
performance you're getting, for example, out of the
cask, or out of the package.
MEMBER KRESS: Okay.
MEMBER ROSEN: On your third bullet, how
relevant is that experience to the plans at Yucca
Mountain. In the reactor safety area we update PRAs
with relevant experience from many thousand years of
reactor experience. Is there analogous value to that
data that you're getting out of others experience in
repository performance?
MR. VIRGILIO: I think that it's coming in
now in terms of what we're getting from the Finns and
the Swedes, and I think that we'll see more from WIPP
as we get more involved in what DOE has done, so there
is some experience. But we rely heavily on the
modeling.
CHAIRMAN APOSTOLAKIS: What is consistency
in treatment of uncertainties?
MR. VIRGILIO: One of the things that we
want to make sure that we're doing is in approaching
the KTIs in a somewhat consistent manner, in terms of
not having more conservatism in one, and less in
another, for some of the same factors. It's a
reconciliation of how we're approaching this.
Try to put everybody on the same playing
field, where we can. That's - I think it's a very
important factor that we don't unnecessarily treat or
use conservatisms around certain assumptions, which in
your last letter to us, I think the over arching issue
there was, because it can bias the results. And I
think that's very appropriate guidance.
MR. GARRICK: I think one thing that might
be important to Tom Kress' question is the very
different situation that exists with respect to the
activity called site characterization, very different
from the reactor problem.
The site characterization program is
designed to really deal with the question of how much
protection are we getting from the natural setting,
and so that's a component of the defense in depth.
MEMBER KRESS: It seems like it's a highly
uncertain -
MR. GARRICK: It is a highly uncertain,
but on the other hand, if you look at the work that's
going on, that's where most of the work is taking
place, is in better understanding the performance of
the natural setting. There's a lot of emphasis on the
waste package, and that is very much an engineering
effort, but the activity associated with the site
characterization is very involved, and very extensive,
and has been going on for many years, and there have
been many lessons learned. And one of the lessons
that probably is the most important from that whole
process is that it is very difficult to quantify the
uncertainties associated with the performance of the
natural setting. But I think that in general, they
have learned along the way what is more important, and
are focusing on those barriers, and those parts of the
natural setting that are going to have the greatest
impact on performance, such as how water moves through
the mountain. And so I think that, to pick up on
Marty's response, that it comes from both places, it's
clearly correct, that there needs to be a component of
defense in depth with respect to the engineer portion,
which is primary what is called the near field, and
primarily what is called the waste package. And there
needs to be an expose of the ability of the natural
setting to provide backup when and how a source term
is actually developed. When you see the billions of
dollars that are spent, most of it is spent in just
trying to go as far in the direction as reasonable, to
quantify the geological setting.
MR. VIRGILIO: Okay. If we can then move
on to slide five. Okay. There's been increased
national attention on spent fuel transportation. And
as I said earlier, this has been stimulated by the
Baltimore Tunnel fire, the terrorist acts of September
11th, and other things, including DOE's announcement
of Yucca Mountain. Staff is, and will continue to
seek the ACNW's views and guidance on critical safety
issues pertaining to spent fuel transportation issues.
The staff itself has been engaged in
review of spent fuel transportation packages, and the
performance of spent fuel transportation packages in
severe design basis accidents, beyond design basis
accidents, what we call the package performance study.
And we've been working very cooperatively with
research in this area.
We've developed a test plan that will be
issued in the near future for comment. In addition,
the Staff is also going to be involved in activities
to validate the structural computer models that we use
in risk assessment around transportation of spent
fuel.
We've also recognized at the National
Academy, The National Research Council Board of
Radioactive Waste Management, will begin a broad
transportation based study focused on spent fuel
issues. This study is going to start in April of this
year, and Staff will solicit the Committee's comments
on the study.
NRC may also conduct additional
transportation vulnerability studies in response to
the terrorist attack of September 11th, and our
proposals are currently before the Commission in that
area.
MR. LEVENSON: I have a question on
transportation. You've listed it under high level
waste. What about other - at the moment, the big
transportation process going on is stuff going to
WIPP. It turns out that in many cases, it isn't the
NRC licensed cask that's controlling much of anything,
but DOT regulations. What's - do we have the same
fuzzy area with high level waste?
MR. VIRGILIO: No, that - I don't -
well, first of all, I don't think it's very fuzzy at
all. I think that the roles and responsibilities
around transportation of waste in this country are
fairly clear with regard to NRC's responsibilities,
Department of Transportation's responsibilities.
With regard to high level waste, this is,
you know, NRC and DOE responsibilities. This is not
going to involve the Department of Transportation, to
the same extent that you see for low level waste, and
for other materials that are being transported around
the country today. They still have some
responsibilities associated with the conveyance, be it
the truck or the rail conveyance, but there's a lot
more NRC requirements, if you will, around high level
waste transportation.
That's changing though. I will recognize
that there's a lot of change going on today around
transportation of low level radioactive waste as well.
We're working cooperatively with the Department of
Transportation, Customs, and other organizations to
make sure that we understand and refine all the
different levels of protection, if you will, around
transportation packages. That's, in part, being
stimulated by our response to the terrorist attack.
MR. LEVENSON: Is there some regulation,
or rule, or law that decreases the role of DOT for
high level waste, compared to the waste going to WIPP?
MR. VIRGILIO: I'd have to get back to you
on that. I'm -
MR. LEVENSON: Because the issues - I
mean, the shipping container for WIPP is licensed by
NRC. It's an NRC DOE and that's fine, but DOT has all
kinds of miscellaneous requirements arising from the
conveyance, which is - I just spent the last two days
in an Academy meeting on WIPP. And the controlling
thing resulting in maybe hundreds of millions of
dollars per year additional expenditure rises from
requirements of DOT, not NRC.
MR. TRAVERS: DOT will still have those
conveyance requirements, as applicable to high level
waste as they are to the WIPP shipments. I think what
we were trying to convey was that in the context of
the detailed reviews that are going to be conducted
associated with high level waste transportation,
they're more rigorous.
Certainly, we went through a certification
process for the cask or the - is that what it's
called at WIPP? And I guess I'm not familiar with the
specific DOT issues that may be limiting, or causing
greater expense in connection with WIPP, but we think,
when it comes to high level waste, it's much more
likely that the reviews and the technical requirements
of NRC are likely to dominate, if you will, versus
some of the safety conveyance requirements of DOT.
They're going to be applicable. DOT still has a
principal, primary role in transportation throughout
the country of anything that involves hazardous waste
shipments.
MR. LEVENSON: Who is responsible, or is
anybody, for looking at that interface, because I know
I've seen what the problem has arisen at WIPP, that
there's some DOT requirements - right now, the big
issue is because remote handled waste going to WIPP
can't necessarily be handled and looked at. You can't
necessarily meet the DOT requirements, and with spent
fuel, you're going to have that in spades.
MR. TRAVERS: Yeah. If you look at what
historically has been the case, we've actually shipped
high level waste around in this country. Not that
much, but it's happened over the years. And am I
saying there won't be issues associated with a much
larger project? I'm not, but I think we have fair
experience in the interaction of NRC requirements and
DOT requirements as they apply to high level waste
shipments in the country.
I don't know if any are actively going on
right now, but they have over the years.
MR. VIRGILIO: There are a handful per
year, moving fuel from one facility to another right
now, so we do have some limited experience.
MR. TRAVERS: But it's a good issue, and
I think it's one when you envision a much larger
expansive project, that is worthy of consideration,
and we'll take that as a challenge.
MR. VIRGILIO: Any other questions? If we
can move on to slide six. In the area of
decommissioning, NMSS Staff is currently evaluating
activities and looking for ways to further risk inform
our technical reviews. Here's an area where we
believe that the ACNW input and guidance has been and
will continue to be helpful.
We're currently working on a consolidated
decommissioning guidance project, where we'll update,
and risk inform, and improve our technical reviews.
The consolidated guidance project is a three volume
set covering decommissioning process, characterization
surveys, and radiological criteria, financial
assurance, recordkeeping, and timeliness of our
reviews.
The ACNW review of the consolidated
guidance will help ensure that we achieve a clear,
complete, and comprehensive set of guidance. The
first volume of this three volume set has just been
published.
Staff is also actively engaged in
evaluating options for long term stewardship for
decommissioned sites, and financial issues pertaining
to cleanup activities, and will be continuing to
interact with the Committee on those issues.
Dose modeling for complex sites, and
consistencies, and conceptual models, as well as the
selection of parameters, and probabalistic dose
analysis are being addressed today in collaboration
with other federal agencies.
Important issues here are being addressed
as well involving partial site releases and
radionuclide transport and pathways. We'll continue
to interact with the Committee around some of these
issues. If I could move to slide seven.
CHAIRMAN APOSTOLAKIS: Now you mentioned
long term stewardship of decommission site. How long
is long term?
MR. VIRGILIO: For hundreds of years, as
a matter of fact -
CHAIRMAN APOSTOLAKIS: Wow.
MR. VIRGILIO: - for some of the sites
that we have, some of the complex decommissioning
sites that we're looking at. We're - right now we're
looking at Department of Energy for taking some of
these sites. However, there are other options being
considered for some of the sites. We're looking at
states for some of the sites. We're looking at Tribal
Governments. We're looking at a number of options in
terms of ensuring enduring institutional controls, and
forcible institutional controls, responsibility for
the long term care and protection of these sites.
CHAIRMAN APOSTOLAKIS: Interesting.
MR. VIRGILIO: It's a very significant
issue for us today.
CHAIRMAN APOSTOLAKIS: I know it's a
significant issue for DOE, for their side. I didn't
realize it was significant for you, as well.
MR. VIRGILIO: Yes. And we're working
with DOE and the states, and the Tribal Governments to
try to find a success path around some of these
issues.
CHAIRMAN APOSTOLAKIS: So we're going to
have again long time periods. Right?
MR. VIRGILIO: Possibly. With, you know
- yes, with institutional controls on some of these
sites. On slide seven, I just wanted to - and I know
we've met several times with the Joint Committee
around risk informing the NMSS programs. We are
continuing to work to further risk inform our decision
making. We've briefed you in the past on this issue.
Today we're engaged in integration of case
studies that were done, eight case studies, and I
believe the staff has briefed you on this. We're
addressing lessons learned and exploring applying a
risk informed approach to other activities within
NMSS.
One of the most significant tasks that
we're working on today is to develop risk metrics and
safety goals appropriate for the materials and waste
arena activities. We see this as very challenging,
but we've been collaborating with the Joint Committee
around these issues, and will continue to work with
you as we move forward.
MEMBER POWERS: When you speak of risk
metrics, could you clarify what you mean by that?
MR. VIRGILIO: Some of that is somewhat
preliminary at this time, but -
MEMBER POWERS: Well, just give me the
analogy to reactor world that you're -
MR. VIRGILIO: If you look at the safety
goals at the fairly high level for reactors, those are
some of the same things we're looking at in terms of
societal risk, individual risk, those type of measures
in metrics.
MEMBER POWERS: Yeah, I know what you mean
now. I'm looking at things like risk achievement
worth, or something like that.
MR. VIRGILIO: Not to that level of detail
or sophistication.
MEMBER POWERS: I understand.
CHAIRMAN APOSTOLAKIS: Let's say deep
diag.
MR. VIRGILIO: Yes.
CHAIRMAN APOSTOLAKIS: Period.
MEMBER ROSEN: Useful detail.
MR. VIRGILIO: We're also continuing with
our training program to make sure that we're uniformly
and consistently applying some of these risk informed
decision making across NMSS.
The last slide I wanted to just touch on,
confirmatory and anticipatory research. And I met
with the Committee before on this, and so has
Margaret. If you think about high level waste, Sub-
Part F of Part 63 requires DOE to submit a performance
confirmation plan as part of their application.
This plan should, in fact, lay out a
program that identifies some of the key assumptions
for the overall site performance assessment. The plan
will also take into account some risk insights, and
develop new and continuous analysis, tests and
experiments that probe and challenge the assumptions
and technical basis for the licensing case.
Anticipatory - and we'll continue to work
on that issue. As far as the anticipatory research,
I think we need to consider what could be needed in
the future, and most importantly, prioritize what
should be done, prioritizing the funding for
anticipatory research needs to integrate external
stakeholder input, the ACNW/ACRS recommendations, and
NMSS views, and of course, researches, technical
insights and planning.
Margaret Federline and I have both
discussed this issue with the ACNW in the past, and we
look forward to further discussion around these
issues.
IN closing, I just welcome your continued
contributions and guidance to the Staff. Your
critical review contributes to ensuring public health
and safety, and enhancing public confidence in the
NRC, so we appreciate your continued interactions.
And I thank you for the opportunity to meet with you
today.
MR. TRAVERS: Sam, we're going to continue
with your presentation, if that's all right.
MR. COLLINS: Very good. Good morning.
I'm Sam Collins. I'm the Director of the Office of
Nuclear Reactor Regulation. I believe I've had the
opportunity in various capacities to speak to you as
individuals. I'd like to welcome Steve Rosen. Steve
and I have worked at a couple of forums together, the
most recently being South Texas. And I think it's an
indication of your stage in career when you have been
at least two places working with one person through
your various stages, but it does lend perspective.
And I think on Steve's part, I never really envisioned
you to be a government employee, but I welcome you.
MEMBER ROSEN: Thank you very much, Sam.
I'm a special government employee.
(Laughter)
MR. LEVENSON: Special means no benefits.
MEMBER ROSEN: That's exactly one of the
things it means.
MR. COLLINS: What I hope to accomplish in
the next period of presentation and questions, is to
provide a broad overview of some of the areas within
the purview of the Office of Nuclear Reactor
Regulation that are of interest to the Committees.
We have a very large span of control in
the Office of Nuclear Reactor Regulation, most of them
being programmatic, so our processes are areas that we
focus on, as well as the technical decision making
within those processes.
In the past, the ACRS has been involved in
both of those forums, and we do have continuing
challenges to achieve success, as we define it. And
I will clarify some of those as I go through.
We appreciate the opportunity to continue
what I would call constructive exchange. And clearly,
that means an exchange of views, in some cases
differing views with perspectives taken into
consideration, and we recognize that both our staffs
play a key role in defining issues.
Here with me today I have members of the
Executive and Leadership Team from the Office of
Nuclear Reactor Regulation. They're here not only to
gain perspective from the presentations and the
questions, but if necessary, to help support responses
to your questions.
I will address a number of cross-cutting
issues today within the areas of presentation, and I'd
like to acknowledge that the three offices here really
work in partnership to address many of the agency's
challenges. We rely heavily on the technical
expertise within the Office of Research to support the
decision making with our programs, and Marty and I
have cross-cutting arena areas in the area of
decommissioning, and we're working on those program
structures also to become more efficient and
effective, and to align those processes between the
offices.
Let me go into the areas of interest, and
begin by talking with technical specifications.
Technical specifications are probably the most
important guidance that's provided to the facilities
within the bounds of operation of the plants. Those
of you who have experience with testing research
reactors in the power reactor community, understand
that this is where the regulations are applied in the
control room 24 hours a day, under varying conditions.
We do have guidance from the Commission.
It's one of the areas of challenge, to risk inform
programs. I think Ashok will be talking about that
also. Marty has mentioned it in his arena, and along
with our efforts to standardize the technical
specifications, we are embarking on a program to risk
inform various areas of the technical specifications.
This is one of our fundamental focuses now in
improving that technical specification tool.
We do have risk informed decision making
guidance, and Marty responded to a question of Dr.
Apostolakis having to do with risk informed decision
making. And we have Reg Guide 1174, was developed in
partnership primarily by the Office of Research, which
does include a fairly prescriptive decision making
process, although some of the inputs to those process
might not be prescriptive, of how to consider the
deterministic aspect, as well as the risk and the
consequences in those variables to come to a good risk
informed decision. We've had some revisions to that
process in the past, and I think it will continue to
be refined as we apply the program itself.
In the area of technical specifications in
that risk informed decision making process, we're
really going to focus today on two areas that we
believe the ACRS involvement will be important.
Although we have one proposal on missed surveillances
that's been approved by the staff, and two others
which are modified in state and mod change flexibility
should be complete by the end of the summer. There
are two areas, one being configuration risk management
for completion times, which would permit managed
temporary extension of existing completion times
within a limiting condition of operation, and that
concept is currently under development.
The second being the risk significant
scope for technical specifications, which would review
tech specs to remove systems, are included solely
because they were judged as risk significant at one
time, and have now been shown by analysis not to be.
Those two initiatives under the eight
total initiatives under risk informed tech specs will
be provided to the Commission, as a part of our
process, as major policy areas. And we will -
CHAIRMAN APOSTOLAKIS: What exactly is an
initiative?
MR. COLLINS: - be giving those to the
ACRS.
CHAIRMAN APOSTOLAKIS: Sam, when you say
initiative, what do you mean? Do you mean the
agencies doing this, or the industry has requested
that something like this happen, and they are
proposing something?
MR. COLLINS: Right. That's a good
question. We have been working with the industry and
our stakeholders, primarily through NEI as leveraging
the industry to provide the input to the areas that
they believe the risk informed decision making could
be applied. We're focusing on eight of those areas.
This is being done in a stakeholder environment.
We're continuing to involve not only the industry
through NEI, but the other stakeholders with public
meetings on the progress of these initiatives. And,
of course, as changes to the tech specs, they would be
subject to public comment.
In the process sense, we're using a new
implementation, or a revised implementation process
which provides for what we would call a pre-screened
amendment review, in that we put out a template, and
as the licensee meets that template, then they are
automatically allowed to change their technical
specifications, but that's on the process end.
CHAIRMAN APOSTOLAKIS: So configuration of
risk management for completion times, that means that
- what, that something has failed, and I'm looking at
the new configuration during my risk assessment. And
I decide that, you know, the risk would be acceptable
if I completed repair or whatever is required by such
and such time?
MR. COLLINS: Yes. I think you have
captured it. Right.
CHAIRMAN APOSTOLAKIS: Now the - you
know, this creates an interesting problem. It seems
to me that PRA was done, as was done in the early 70s
and the last 25, 30 years, was not really developed
for this kind of thing. It was developed for a study
state long term kind of assessment, developing
frequencies of core damage or accident sequences and
so on. And now we are rushing into applications for
which the baseline risk information was not intended.
That doesn't mean that it's inappropriate, but that's
not why it was developed in the first place. And in
particular, talking about time dependent situations -
I mean, I don't think PRA was really very good at
that. We are averaging too many things. We are
averaging the input of periodic tests, this and that.
Now when it comes to real time applications, I don't
know that we really have thought about it very well.
And this is just one example where we're - I think we
are rushing into applications for which the original
tool was not designed.
Now some - that doesn't mean that, you
know, it's useless. Most of the information is there,
but I think we need to really be careful, and pay
attention to the fact that now it's a different
application. Now you've been trying to say something.
MR. THADANI: I'll touch on -
CHAIRMAN APOSTOLAKIS: Sam, yes.
MR. COLLINS: I think your caution is
appropriate. However, I'm not sure rushing into it is
the right connotation. I think if we were to look at
the history of the maintenance rule, which is really
configuration management giving risk insights, that
was a fairly deliberate process. I would call it
deliberate. The industry would call it excruciating
probably, and it's not that different, other than the
conditions set by operational conditions, which would
result in corrective maintenance, rather than the
prescriptive - the conditions that are pre-planned
for corrective maintenance. So the configuration is
a little more tenuous, perhaps, and less able to be
pre-planned, so the process is twisted that way. But
I do think that we have a history with the maintenance
rule that would indicate that these tools are
appropriate, but the input has to be correct.
MR. THADANI: George, I understand the
point you're making, and I recognize that time
dependence is not built into today's PRAs, and so
there are some limitations in what we have in front of
us. On the other hand, we have much better
understanding of where the significant risks might be.
We have these tools, in spite of the imperfections and
uncertainties in these analyses. I think these tools
are very valuable in bringing better discipline to the
set of requirements that are embodied in the technical
specifications. Particularly, as you know, the tech
specs were developed with the concept of one change at
a time. Reality is different than that, so
configuration management, to the extent one can take
advantage of the plant models, so to speak, I think is
a step forward, is the right thing to do, as long as
we're sensitive that there are some limitations in the
tools that we're utilizing. So it's progress, and we
need to be looking at the issues of dynamic aspects as
we go forward. I think this is the right direction to
go in, and not to wait until we finish everything.
CHAIRMAN APOSTOLAKIS: I agree that it is
the right direction. It's just the availability of
the appropriate tools that worries me.
MEMBER POWERS: Well, I guess the question
I'd ask is, do we even have the appropriate tools
right now? You've gotten -- the IPEEE insights
document comes out. It says gee, all these
operational things that you guys are talking about,
that's half the risk. I mean, is it appropriate to do
analyses of configuration management taking into
account half the risk?
MR. THADANI: I think the answer to that,
in my view, is yes. External initiators are going to
look at, and the same way as you would internal
initiators. I think I can agree with you that half
the risk may be from external initiators. That does
not mean that the components, and systems, and
structures within the plant, you shouldn't look at
from the best risk perspective you can get. And the
industry, I think to me that's progress. Industry is
making better use of IPEs and IPEEEs, and that mode.
I recognize the limitations. You recognize the
limitations. I mean, the spectrum of these studies
and analyses, but nevertheless, I mean is there
something better we can use to risk inform various
activities?
MEMBER POWERS: Well, let's make sure -
MR. THADANI: I don't think there is.
MEMBER POWERS: - we understand, that one
of the external initiators you're talking about, it's
actually an internal fire.
MR. THADANI: Yes, I understand that.
MEMBER POWERS: And it seems to me that we
have - I mean, the IPEEE insights document is a wake-
up call that says you've got a problem when you're
using the conventional PRA tool, that you're ignoring
half the risk. And I think there's no question if
your current PRA tool comes back and says this
component is important, or this configuration is
important. The answer is yeah, it is. It's when it
comes back and says well, you can live with this, that
you have the question because of the incompleteness of
your tools.
CHAIRMAN APOSTOLAKIS: John, you wanted to
say something.
MR. GARRICK: Well, I just wanted to
comment that I think that as far as this time
dependent question is concerned, it's correct that
most of the logic diagrams are static models. But it
is also important to note that a lot of the
applications have been with respect to dynamic
situations. And the way that is often addressed is in
the context of discretizing that dynamic situation in
such a manner that you can assemble a set of so-called
steady state or static models, in a manner that will
represent a dynamic representation.
Examples of that, and where it's done more
than in the reactor field, is in the chemical field
when you're trying to do a risk assessment of a
process, where you fundamentally divide that process
up into a series of unit operations. And you connect
the individual unit operations with pinchpoints that
logically and reasonably defensible.
The same thing is true with respect to
developing a risk assessment of something like the
space shuttle, where you map an entire mission. And
the way that's often been done is with some clever
discretizing of the model, and defining of the input
and output states, that do a reasonable job of
representing what's going on. So my only comment is
that there's a great deal of ingenuity being applied
to some of these models that goes beyond what we're
having described here.
CHAIRMAN APOSTOLAKIS: Yeah. My concern
was not so much how to handle time, because I agree
with John, that discretizing has worked very well. My
concern with configuration management is when I change
the configuration, and we saw that in the calculations
of the risk achievement work, which actually does do
some of that. When I change the configuration, and I
use now the new PRA, which is usually a variation of
what I already have, am I doing it correctly? Because
if one component is down, or more than one component
are down, several terms in the PRA are affected. And
some of them in a subtle way, and I'm not sure that we
have - I'm not saying we cannot do it. I'm just
saying we have not really thought about it very
carefully, and established rules how to do it and so
on. The time dependent part I agree with John with.
In other words, don't misunderstand me.
I'm not saying we should go to the dynamic PRA that
some groups are proposing. So far we don't seem to
have that compelling reason to do that, but the
configurations worry me a little bit.
VICE CHAIRMAN BONACA: Well, one thing we
have to also reflect is what's in tech specs right
now, and what is the basis for them. And typically,
there isn't a basis. I mean, it's just - there were
numbers often times coming from - so I totally agree
on the need of cautiousness.
Mr. APOSTOLAKIS: But this is true for -
there is no logical basis for anything that's not PRA
based.
CHAIRMAN APOSTOLAKIS: Okay.
MR. COLLINS: Okay. Again, the goal here,
as Dr. Powers indicated, I guess I would debate the
words that the goal is to live with this. The goal is
really to acknowledge that there is a trade-off
between the risk of taking an action with a dynamic
power plant that's called for by the license, which
may be ramping down in power, as opposed to a steady
-state operation with a calculated length of time that
provides for recovery of the equipment, so that's the
balance we're trying to achieve.
The next area I'd like to focus on - and
I thank you for your questions. It's good debate - is
rule making. This is clearly an area where the
agency, again, is interdependent. I'm going to talk
about a few areas, specifically 50.44 and 50.46 that
are part of the Option Three Rule Making Area where
research has leave to define the criteria. I'm in
rule making space now, which is really when we're
talking to the ACRS concerning the application of
these tools themselves. And clearly, many of these
are of interest to the ACRS, and will likely be
activities that we will bring to the ACRS.
50.44 is combustion gas control systems.
We're looking at a final rule in early fiscal year
`03. 50.69 is a risk informed treatment of structure,
systems and components. It's called the Option Two.
South Texas, which you're familiar with, we understand
the issues from the ACRS having to do with South
Texas. We could call that a proof of concept, but
there are some differences between the approaches to
50.69. We are wrestling, as you are, with the
categorization and the treatment balance.
CHAIRMAN APOSTOLAKIS: We have a Sub-
Committee meeting scheduled for later this month to
discuss the NEI document.
MR. COLLINS: Okay. We have proposed rule
in spring of 2002, final rule in fiscal year `03. We
are getting feedback from South Texas on the
application of the South Texas proofer concept which
was a license exemption, and we're continuing to learn
from that process also.
The ECCS acceptance criteria 50.46, the
technical reviews are ongoing. Ashok has lead for
that. We're looking at what we call an "unbundling"
of the technical issues which is the separation of the
interdependencies in that area. We do have a paper
that was provided to the Commission in July of 2001.
It's SECY 0133, and as a result of the recent arena
update to the Commission which occurred last week, we
anticipate that we'll be requested to update that
paper, and that will provide for clarification of the
challenges that the staff has in looking at 50.46.
50.61, pressurized thermal shock. The
technical reviews are ongoing. And again, we'll rely
on the input from research. There's a possible
proposed rule in fiscal year `03 in that area.
50.48, fire protection rule making.
Again, the regulatory framework in the adoption for
NFPA 805 and the proposed rule in fiscal year `03.
MEMBER POWERS: Sam, do you think anybody
is going to take advantage of NFPA 805? Is there any
advantage to the 805 to take?
MR. COLLINS: I think it's going to have
to be looked at on a case by case basis, where
licensees would look at the ability to bring realism
to the code, if you will. Where fire protection
systems already exist, there will be a balance between
maintaining those systems, and/or modifying those
systems, and there has to be a value associated with
that, so the answer to your question, I think, right
now is indeterminate. And I think it relies fairly
heavily on how the staff use the implementation
guidelines with the fire code. How much leeway will
we allow for a mix and much, if you will, of the
regulations.
MEMBER POWERS: Pick and choose among the
things?
MR. COLLINS: Right.
CHAIRMAN APOSTOLAKIS: I'm a little bit -
I'm confused actually. I don't understand how a major
technical society can come up with something that is
not very useful to a major technical agency. In fact,
it's almost useless. I don't understand that. How
can that happen? Do you have any thoughts on that?
I mean, the ASME Standard on PRA, we have
disagreements with them, I mean the staff does, but
there is some substance there. You can see what
they're doing. The other guys are - I mean, in one
of the earlier versions you're supposed to do the
deterministic stuff, and then you, you know, do a risk
assessment on top of it. I mean, a complete
misunderstanding of why one does a risk assessment.
It's a mystery to me how that can happen. I mean, are
we approaching fire protection in such a different way
from the National Fire Protection Association, or they
don't know what risk information means?
MR. COLLINS: Let me ask Gary to respond
to that.
MR. HOLAHAN: Gary Holahan, NRR. First,
George, I think - we've had some of these discussions
before. The first point is that NFPA 805 is the Fire
Protection Agency's - I mean, it is the consensus
among the fire protection experts of the country as to
how to approach these issues. And I think it's not a
proper characterization to say that the approach is to
have deterministic requirements, and on top of it do
a PRA.
CHAIRMAN APOSTOLAKIS: At some point it
was, now maybe not now.
MR. HOLAHAN: I think at this point - in
fact, 805 has been - was published last year. The
way 805 works is to have identified those areas in
which the technology exists to risk inform specific
parts of fire protection requirements. Okay. And
those are specifically identified, and then what the
fire protection community felt, there was no
technology available to risk inform it. They
identified what were, in effect, deterministic
requirements. Okay. So the document has areas that
can and should be risk informed, and those for which
there ought to be deterministic requirements.
In the NRC's endorsement of this rule, it
is likely that we will accept that as the current
state of the art, but then also recognize that where
that state of the art is changing, where a licensee or
an industry group wants to come in and say some of
those things which 805 is not risk informed, we now
have developed the technology to do that. We will be
receptive to those, as well. And the rule that we
write, which is still in the formative stage, but the
rule that we write will be receptive to using the risk
informed pieces of 805, but also be receptive to
additional risk informed activities in some areas that
are not currently developed. But it will send a
signal that we're receptive to those, and it will give
a rule that will allow us to do those things without
going through an exemption process. So I view it as
a step forward within the existing technology. Okay.
But an invitation to stretch the technology, as well.
And I think the previous comments about usability and
who will use this, frankly, we don't have a pilot
plant who wants to try this out. And it may be, at
least early on, that licensees will only want to pick
and choose pieces of 805 and the new rule, you know,
as it suits them, or as they have changes to their
programs, they will pick up pieces. But the fire
protection community, you know, was highly involved,
something like 60,000 fire protection engineers around
the country, you know, voted on this thing, so I think
they think it's useful. But I think utilities will
pick and choose the pieces of this that they find
appropriate.
CHAIRMAN APOSTOLAKIS: I think that's a
clear case where you see how this agency differs from
everybody else. I mean, there is a much higher
appreciation of risk information within this agency
than in other places. And I think that was part of
why 805 was not -
MR. THADANI: George, to round out this
discussion, I think - I suspect you know, but in case
the Committee doesn't know, there's been healthy
debate about to what extent 805 really does consider
risk information, provides guidance and how to go
forward.
The American Nuclear Society has got
effort - they're initiating an effort to develop -
I believe they have decided - I may be wrong on that.
I need to confirm - to initiate an effort to develop
a standard for conducting fire risk analysis.
CHAIRMAN APOSTOLAKIS: So that says a lot.
MR. COLLINS: We would view any effort in
this area that's a consensus agreement as a better
place than the existing requirements for the fire
protection that we have, so that's one of the impetus
that we have in this area.
MR. TRAVERS: Sam, could I make one
comment?
MR. COLLINS: Yes.
MR. TRAVERS: While we're on the rule
making slide here, I thought I might make note of the
fact, and it's probably not surprising to you that
we've gotten just recently a rule making petition on
50.46 from NEI that focuses on just the size of the
break. And that's clearly been the focus of industry
stakeholders up until now, so we'll be treating that
in the process that we use for considering those
things. Of course, our effort has been broader in
context.
MR. THADANI: Yeah, let me just - I was
going to get into it later on, so maybe this is the
right place. The scope of 50.46 in the paper that Sam
referred to that's in front of the Commission, issued
in July, had certain options in it. The first piece
had to do with models, the Appendix K models versus
should we look at the decay heat curve, clad water
interactions and so on. The break size, to what
extent we can rely on single failure criterion versus
functional reliability systems, so we broke it down in
three pieces. And the first piece had to do with sort
of models, analysis, functionality. There is actually
a petition even on that. Industry had a petition to
modify the - get rid of 1971 decay heat curve and use
1994 decay heat curve, which is clearly more
realistic, and the `71 curve is conservative. But
there are issues regarding Appendix K models, and we
were trying to make sure we took an equal look at
that.
The second piece was some of the
assumptions that go in the analysis, such as large
break, along with loss of off-site power and so on.
We thought that was an area we could handle based on
what we have in the near term, so we had hoped to
complete that work next summer. There's been a delay
of two to three months because of the September 11th
follow-up activities. But the final issue, which is
defining what's an appropriate break size, we
identified a number of areas that we need to work on,
and we felt it was going to take some time, so in the
paper to the Commission we said it probably will take
us on the order of three years to get to a new, what
I call design base accident for current plans. We now
have a petition which would clearly require that we
take another look and see where we can go.
MR. COLLINS: Okay. Thank you. As you
know, these are integrated activities, as demonstrated
by the discussion here today, and they are coordinated
to the Risk Informed Regulatory Implementation Plan,
latest update of that was December 5th, 2001.
MEMBER POWERS: Sam, if I were you - had
your job, I'd be a very frustrated individual. You've
got a list of rule making activities here. I bet
you've got another list that you fear on the horizon,
and every single one of them has all this risk
information that you're supposed to take into account.
And you've got a Commission that's telling you to be
more risk informed. But the way your staff goes about
getting risk information strikes me as clumsy.
As I understand it, if they want risk
information, they go down to Rich Barret and talk to
him about risk information. It's always useful to
talk to Rich Barret. I know, I always learn something
from him. But, you know, Rich gets busy. And he's
got a lot of other things, and you've got to kind of
wait for him to deliver the risk information that you
need. He might have to go over to research because he
- it's a question beyond what the capabilities he
has.
Your staff can't dial up risk information
on a particular plant at any time they want one. Have
you thought about what your staff needs to really
facilitate this risk information move to make it
convenient to address all these risk informed things
that are coming down the pike at you?
MR. COLLINS: I guess you put a challenge
in front of me that I didn't realize existed to the
extent that you express it. We have a risk informed
group within the Leadership Organization that cuts
across all divisions, and the risk informed activities
themselves are really centered in David Matthews'
organization. Rich Barret is an expert in that area,
but most of our risk application is really in the
process area, how do you consider risk.
Plant specific information, we rely on the
tool that in large part are provided by research,
whether they be the SPAR models, or the SAPHIRE
models, or the different phases of the workbooks for
the significant determination process. But ultimately
to make regulatory decisions, we need the input from
the licensee if we're going to make what I would call
regulatory decision, as opposed to a programmatic
process definition, if that makes sense to you.
MEMBER POWERS: I think what you're
telling me is that you're comfortable relying on
getting all your risk information from outside
sources, and never playing with it yourself. I mean,
just taking this as truth, and having all answered,
and he's characterized for you satisfactorily all the
contingency plans that - for you by the outside. And
that the individual in your organization that's
actually formulating them, and it goes through a lot
of processes and consideration, but there is somebody
making the initial determination of actions to take,
that he can rely on this, and he doesn't need direct
access to risk information.
MR. COLLINS: Well, we have risk
information that's available to us in the form of the
tools that's been provided by research. We make
Rich's decisions day to day in the Operating Events
Analysis Group, where we get the event reports from
the licensees. We do a rough cut of the event
significance to determine what's our response. The
region does the same based on their input to the
plants themselves. That's a very different decision,
and uses a different tool, than a long term, long term
being up to a year approval of a risk informed license
amendment would be. In which case, we would rely on
the licensee to submit the portion of the PRA that's
most applicable to the area of the license that they
in fact want to change, so it's a graded approach.
And I think the graded approach is the tool we have
today. Now can we improve that? I think the answer
to that is yes, and it will be improved by the PRA
standard being applied by the working agreements that
we have with research to upgrade the tools for the
field, as well as for those decision makers in NRR.
But I'm trying to contrast a difference, Dr. Powers,
in the dynamic decision making that's done as a result
of events so we know how to respond in the short term,
as a result of the more programmatic reviews which are
done with the aid of the licensee's information.
Ultimately, when we make a regulatory
decision in enforcement space, for example, or in the
significance determination process, we use our
internal tools, and we rely on the licensee to bring
their information to the table, and then we reconcile
that. And that's ultimately what prevails.
MR. THADANI: Let me just comment on what
was just said. Coming from the culture I come from,
I think what you described I will characterize as -
MEMBER POWERS: Well, the culture you come
from is his culture.
MR. THADANI: And we - I don't think we
have reached that stage at the agency, obviously, but
I don't - I want to make sure you know the effort
that's ongoing in developing SPAR models that cover
all 70 sites, and the process we're going through to
make sure they're technically appropriate and can be
used by all the staff at NRR and regions, and so on.
So I think that's an important step towards I believe
what you're describing.
MEMBER POWERS: Well, I guess it is and,
you know, we discuss these SPAR models and argue over
well, are they good enough and things like that. And
quite frankly, I think we both believe that perfect is
the enemy of accomplishing anything here. And first
steps are worthwhile, and the SPAR models are - it's
just that my comment is driven by if I were doing -
working for Sam, I - and I knew he was being hammered
with all these people saying take risk information and
do more with it, and things like that, you know, I'd
want that SPAR model yesterday, and I'd want it
continuously better so that I could play with it and
understand risk, rather than having to rely on
somebody else because Sam is a very good
Administrative Assistant to hold me accountable as
somebody working for him for what I produce. And
maybe it's effectively taken care of, but like you
say, maybe we can improve it, and maybe the SPAR
models help.
CHAIRMAN APOSTOLAKIS: I wonder whether we
should look at the clock every now and then. We want
Mr. Thadani to have enough time, as well. Although,
I'm sure that -
MR. COLLINS: Well, he's taken some time
already.
CHAIRMAN APOSTOLAKIS: There will not be
as many questions for him.
MEMBER POWERS: The program is in such
good shape you can't have any questions then.
MR. THADANI: I will be very efficient.
MEMBER POWERS: Unfortunately, we won't.
MR. THADANI: I didn't say that.
CHAIRMAN APOSTOLAKIS: You are risk
informed.
MR. THADANI: Yes.
MR. COLLINS: I'll move quickly so you can
get -
CHAIRMAN APOSTOLAKIS: Okay. I don't want
to rush you, but I will do it any way.
MR. COLLINS: New reactor licensing, in
response to the Commission's request, we provided the
Commission in October of last year, future licensing
and inspection readiness assessment as SECY 0188. And
that assessment identified several areas that may need
to be performed in support of new reactor licensing.
Now I would have to indicate to you, as
you well realize, that one of our challenges in this
area is the uncertainty in the plans of the potential
applicant. And having said that, there are areas that
the Office of Research, and the Office of NRR are
focusing on as we move forward in providing products
to the stakeholders. One of those, I believe, is
within the purview of the ACRS, as you well realize,
is the pre-certification and the certification review
of the AP1000.
During the week of January 28th, we
completed the Phase Two evaluation, having to do with
the scaling analysis, and we met with Westinghouse on
the 23rd of January on two issues that Westinghouse
has agreed to respond to as a result of that Phase Two
review. The report is in preparation as a result of
the review at this time.
On the 15th of this month, Jim Lyons'
organization, the new reactor licensing project office
is scheduled to brief a sub-committee on the
applicability of the AP600 standard design, analysis
code and test program as it applies to the AP1000
standard design. WE're expected to brief the full
Committee on these issues during March 14th and 15th,
I believe.
We also have a challenge in defining Part
52, and those areas that need policy guidance from the
Commission. WE have a tentative date right now for
ACRS briefing on November 2002, and that timing of the
ACRS interaction is dependent on when an SRM is
received on the proposed rule itself.
The ACRS interaction on NEI proposed
alternative regulatory framework for advanced reactor
designs will be determined after staff has received
the NEI white paper, and is currently expected in the
second quarter of 2002. ACRS interaction will
probably be late 2002 or 2003. Research plays a key
role in that area.
The application reviews for the Pebble Bed
marginal reactor, the AP1000, the GTMHR are expected
to involve several policy issues that most likely will
require ACRS interaction. And again, these schedules
and policy issues are dependent on the application
schedules and the applicant's willingness to support
the work from the staff.
The PBMR application review is expected
approximately in 2004. The AP1000 standard design
certification review is expected this year in 2002,
and the gas turbine modular Helium reactor, GTMHR
combined license application is expected in late 2004.
Also within our budget assumptions, we
would be receiving requests for early site permits
this year and in 2003, and all of those are in the
budget assumptions that have been provided to the
Commission itself.
CHAIRMAN APOSTOLAKIS: Are you going to
get into this shop in your presentation as well?
MR. THADANI: Yes. I'm going to talk
about what the research plans are.
CHAIRMAN APOSTOLAKIS: Okay. So we
shouldn't be asking those questions now.
MR. COLLINS: And I can chime in, if
necessary, if you want to just consolidate the
questions in one area. That would be sufficient.
Licensing issues, there are three broad
areas I'd like to bring to your attention. One, of
course, is familiar to you, and that's power uprights,
and we've had presentations in front of the Committee
for power uprights. We think the first presentation,
the staff had a few missteps in support for ACRS in
that area, and we did recover in our subsequent
presentation. We are in receipt of the ACRS
recommendation to consider a standard review plan in
this area. We have it under advisement.
I would want to relay to you that that's
a balancing of needs within the office now. We have
taken a rough look at what it would take to formulate
a standard review plan, and the time frame necessary,
which essentially would be this year given the amount
of power uprights that are coming in in 2002, 2003,
that's when really the application would be. Against
the use of the revised topical, which is currently
under review by the NRR staff, as well as the
initiative to use an already approved power upright as
a template for the SER, and we will balance those and
come to a weighted decision and provide that back, not
only to the ACRS, but the Commission has asked us for
that consideration also.
MEMBER WALLIS: While we're on power
uprights, you're aware we've had some correspondence
about the safety evaluation report, and this is, of
course, the written technical justification for
decisions made by the Commission. And I think both of
us, or all of us are very concerned, are very
interested in it being as clear a document as
possible, giving all the reasons for the decisions
that are relevant.
MR. COLLINS: We agree, and I think we've
made some revisions to the SER to provide for that
standard, as indicated by the ACRS.
MEMBER WALLIS: Okay.
MR. TRAVERS: There's been a similar
comment, Graham, that we got from the IG, noting that
while we're getting to the right technical conclusion,
very similar to your recommendation that there is a
need to revisit, and NRR is developing an initiative
to do that, looking at the characterization of those
conclusions and the basis for them in SERs that we -
we agree with you that they're fundamental in a whole
host of ways to the products that we put out. And I
know Sam and his folks are working very hard on that.
MR. COLLINS: Right. The issue there is
the basis for the decision, as you well indicated.
MR. TRAVERS: Not the bottom line.
MR. COLLINS: Right. We currently have
eleven power uprights under staff review, including
four extended power uprights, so this is a business
line of our's which is very active. WE've completed
22 uprights during our review processes, including
five extended power uprights. That total for all
power uprights is about 3,200 megawatts of
electricity.
License renewals, you're well familiar
with license renewals. Again, that's a premiere
product line. The Commission, as well as the ACRS, is
very interested in that. Under review we have Turkey
Point, North Inniserry (phonetic), Catawba, McGuire,
Peach Bottom, St. Louis, and Fort Calhoun. Next
decision on license renewal is Turkey Point.
The generic guidance has been issued to
assist the NRC and the future applicants in improving
the effectiveness and efficiency of the reviews. And
the first applicant to use this guidance fully is Fort
Calhoun, we expect.
You should be aware that we're assuming a
33 percent efficiency in this area. It's part of the
staff initiatives, and Marty mentioned the program
office goal of NMSS of efficiency and effectiveness,
and this is one of the assumptions that we have made
in refining our processes, going from about eighteen
and a half to about twelve and a half FDE per license
renewal.
We're also looking at potential plant
reactivations. Brown's Ferry is under consideration,
and to a lesser extent, WMP1 completion study is
underway. These initiatives are not new licensing
organization responsibilities under the project
office. Those will come under John Zwalinsky in the
licensing area, but those are challenges at this point
that we have not budgeted for, although we do have a
general knowledge of what it would take, particularly
in the case of Brown's Ferry, for the recovery given
the other two units that have been recovered. There
would be additional challenges to the staff.
CHAIRMAN APOSTOLAKIS: But you have had
some indications that somebody is thinking about it,
but you may get some application?
MR. COLLINS: Our understanding, and I
think it's been announced that a decision on Brown's
Ferry would be expected in the early spring, in the
April time frame.
CHAIRMAN APOSTOLAKIS: Okay.
MR. COLLINS: And there have been ongoing
technical reviews for approximately the past year at
that site.
MEMBER ROSEN: Have you heard anything
about Zion?
MR. COLLINS: I've heard, Steve, probably
what you have about Zion, and that's only in the trade
press. Let me move on to the summary. We have a
great deal of work on our plate, as does the ACRS.
I'd like to acknowledge the benefit in the exchange of
staff with the ACRS. We thank you for Noel Dudley.
He will be a great addition to our staff. We welcome
him back.
MEMBER POWERS: We hate you for this.
MR. COLLINS: Well, we'll donate it in
other areas. John and I have a healthy exchange on
the support for ACRS, and we believe that it is a
worthy rotation for our staff to understand not only
the issues, but the perspectives that the ACRS brings
to those, and we're working on rotational development
opportunities.
We are meeting our commitments. We were
very successful last year in our performance goals.
And again, that's indicative of the support between
the offices. One of our challenges is to find a way
for the ACRS to understand some of the resource
constraints that are facing the staff, and I think
that is an area of a challenge for us to exchange.
And that's a little bit of realism into the
application of the answers. It's always good to think
broadly, and to deal conceptually. And we welcome
those inputs.
The Offices of NMSS and NRR are really
where we have to apply the issues, and we're under
fairly strict time constraints as mandated by the
Commission in many of these areas, so I would believe
that there's room for us to enter into a constructive
conversation in that area, particularly when we're
talking about refinement of programs.
MEMBER POWERS: It's one of - I mean,
it's a challenge that I don't know that the ACRS can
ever meet on this. You know, we just don't know what
your constraints are, and we can't evaluate it, so I
think that's the most valuable when you respond to
some of our suggestions. And you say well, look, you
know, this is - it's just a constraint of time, or
manpower or things like that, you know, that's one we
just have to accept from you, because there's no - we
just don't have the time or expertise, it seems to me,
to go into that.
MR. COLLINS: I understand.
MEMBER POWERS: And I think it would be
unwise for the ACRS to temper its comments by mentally
-- suppressing them by mentally taking that into
account.
MR. COLLINS: Right.
MEMBER POWERS: I mean, it's far better
for you to come back and say well, I just can't - I
just don't have the manpower to do this.
MR. COLLINS: That's a fair observation.
I think it's information that we owe to the ACRS, and
that is put in perspective in -
MEMBER POWERS: Yeah, I mean it's -
everybody has to somehow live within their budget
plan.
MR. COLLINS: Right. And I do believe
additionally there is a role, and it may be a by-
product but its one that we would want to be sure
you're aware of, of the ability of the ACRS in your
decision making, in your input to help us with our
public confidence goal. And again, that public
confidence is not directly in the context of promoting
nuclear power. It's in the context of the NRC as a
strong credible regulator. And the role of the ACRS
in looking at the products, and challenging the staff,
and in providing for that independence, I think is
important. It may not be well known, but - and I
think to some extent the staff ourselves can work with
you on the right context to place that in. We don't
want to leverage it overly, but I do believe it's
valuable.
MR. GARRICK: Sam, a while ago you
referred to the collaboration between NRR and NMSS.
One thing that occurred to me is that over the past
four or five years in particular, the two Committees
have written numerous letters to the Commission on
this whole issue of risk informing the regulations,
and risk informing the process.
You also mentioned a little earlier about
a cross-cutting group in the risk area. I'm curious
if somebody is looking at these letters from the
standpoint of consistency of advice, consistency of
application. In particular, our Committee has been
very focused on the issue, some very fundamental and
philosophical issues associated with what constitutes
risk assessment, having to do with transitioning from
assumption based to evidence based analyses, having to
do with reasonable, as opposed to conservative,
unnecessarily conservative.
The reason we want to do uncertainty
analysis is because we don't have to make the choice
of being conservative or non-conservative. We put
forth our best shot at what we think the risk is.
These are very fundamental ideas and
issues, and I just wonder if somebody is looking at
that database, if you wish, and tracking the
consistency of advice that the Commission is
receiving. That would seem to me to be a very useful
platform of collaboration in this rather important
concept.
The other thing I just wanted to mention
before we got into the research area, is that Dana had
referred to earlier, alluded to the difficulties of
getting your arms around all of the risk informing
activities that are going on. One of the things that
this Committee found very useful was the Commission
white paper of three or four years ago on risk
performance based regulation. That paper was
refreshing in that it reached out more than any I had
ever seen to deal with the issues of what was meant by
risk assessment, what was meant by performance based,
and what was meant by defense in depth, and
precipitated a whole list of very important points
that we found very useful in using as a kind of a
starting point for subsequent advice on risk and what
have you, so those two things.
One, is somebody looking at the advice in
terms of the consistency on these rather critical
issues. And second, what prospect is there for maybe
a sequel to the white paper, an update on the white
paper that went beyond the rigid and formal structure
of rules and regulations, and indicated some sense of
how the Committee - how the Commission was thinking
about these extremely important issues.
MR. COLLINS: The - I'll defer to Ashok
here in just a moment, but the forum we have to
integrate our risk informed activities is the Risk
Informed Implementation Panel and the Steering
Committee, which Ashok is the Chairman of, and which
Marty and I are also members.
Now in direct answer to your question, do
we take the ACRS letters and look at those, and
compare those? I think the answer to that is no.
Each office does that individually.
MR. GARRICK: Uh-huh.
MR. COLLINS: I think what we would have
to do is take that under advisement, and it could
easily be done by the panel as an order of business.
MR. GARRICK: Well, what triggered it was
your reference to a cross-cutting group. It just
seemed to me this might be a useful exercise.
MR. COLLINS: I think it's the right forum
for that.
MR. GARRICK: Yeah. Right.
MR. COLLINS: Right. Yeah.
MR. THADANI: John, just to add to what
Sam was saying, we - today we do not really have a
systematic process in place to look if various
decisions are consistent in terms of risk information.
We are applying at least areas where we in
research are involved, and for example, work on the
cask. We're trying to make sure that there's
consistent application of risk informed thinking as we
go forward. Marty indicated that we're starting out
on a number of areas just now, within NMSS activities,
and research is engaged in that to make sure that
again, if there are going to be differences in
applications and decisions, we understand what those
differences are, and able to account for those
differences.
In addition to that, actually Marty's
initiative, there is a PRS Steering Committee that I
Chair, of the Program Officers who are members of the
Committee. Louise Reyes from Region Two is a member.
OTC is a member and so on.
Marty has actually brought to table a
number of initiatives within NMSS just for that
purpose, to share with the Committee, to see if there
are some inconsistencies, there may be some
inconsistences. At least offer an opportunity for
discussion of those, but I could tell you, we're just
barely starting. Marty may want to add to that, but
I think it's just an initial stage where we are.
MR. COLLINS: Yeah. Let me just finish by
clarifying a comment that Ashok made, and that is that
in the Office of NRR, when we make risk informed
decision making, we do attempt to go back and do a
quality check. For example, in the revised oversight
process, or the significance determination process, we
make those decisions as provided for by the ROP, the
Revised Oversight Process. Research does an
independent check of those after the fact, and
provides us an input into whether that consideration
was appropriate or not.
MR. THADANI: That's through our accident
sequence precursor program.
MR. COLLINS: Right. So that's kind of -
that's how we're trying to balance that, but your two
points are still appropriate for us to take away.
MR. THADANI: Yeah.
CHAIRMAN APOSTOLAKIS: One last question
on the letters, since the issue of the letters came
up. We are discussing among ourselves what, you know,
the best way would be to communicate with you. And
there are several ideas regarding the structure of the
letters. And in fact, in the last several years we've
been using that structure that has the recommendations
up front, or the conclusions and recommendations, and
then a discussion, which we believe is an improvement
over past practices, where you really had to look all
over the place to find the recommendations.
But one of the things that's happening is
that in some issues where there are disagreements
among committee members, the letter has to be written,
you know, at some point, so the easy way out is to
eliminate as much as we can the controversial issues,
and come up with the lowest common denominator and say
this is a conclusion now. And some members feel that,
you know, there is a lot of useful - there are a lot
of useful ideas and maybe possible recommendations
that are eliminated that way, so you guys never know
that some members felt that way, and others countered
with counter arguments.
So what - do you think that the letters
are better if they have clear recommendations without
any controversial debates, or they would be improved
if some of the issues that the members have been
discussing in the open forum here were actually
reflected in the letter, but diluting the
recommendations?
MR. TRAVERS: Yeah. Maybe I can start,
and I'll be happy to -
CHAIRMAN APOSTOLAKIS: Sure. Sure.
MR. TRAVERS: - hear conflicting views.
But from my vantage, it's helpful to have a clearly
articulated recommendation. It - from a number of
perspectives, not the least of which is being able to
track our responsiveness to that. You know, clearly
identify where, in a consensus sense, the Committee
has come down on an issue.
Having said that, I don't think we'd look
askance at additional information that bore on your
deliberation on any particular issue, including views
by members, as long as, you know, it was sort of
illustrative of the debate or discussion that sort of
was carried out in connection with that. But I -
from our - you know, from a management standpoint,
it's very helpful to be able to start with a consensus
or majority view, however it's done here. I believe
it's majority, and be able to work that top level view
as we deem appropriate, and certainly be able to
respond to you in the context of how we're doing that.
Or if we disagree, why and what the constraints may be
that force that view on our part.
MEMBER KRESS: From that standpoint, what
do you do when you get a letter from us that has
additional comments from particular members, that may
be either giving additional information, or may be
contrary to what's in the main body of the letter?
What do you do with that sort of information?
MR. TRAVERS: I think it's a matter of us
considering that in an informing sense as we go about
our duties and responsibilities. Any information of
that sort, I think, can be helpful, and it may shed
some additional light on the recommendation and
conclusion that you reached. You know, it's
illustrative, I think, of a healthy discussion/debate
here on the part of the committees.
MR. COLLINS: In summary, I've covered the
major areas I believe are of interest, although I've
been very specific in the topics. Other areas that
might be of interest for future discussions, one might
be a presentation by the Steering Committee on Risk,
for example, if that would be beneficial in response
to some of the lines of questioning today.
Revised Oversight Process is clearly of
interest to the ACRS Committee, including the use of
PIs, the Significance Determination Process and the
Colors. We do have materials engineering challenges
with the cracking in the control rod drive mechanism.
We have other areas that we're looking at, stress
corrosion cracking in general. Do appreciate the
ACRS' involvement in the proposed orders that were
formulated for the CRDM cracking.
License renewal I talked about a little
bit. In response to the September 11th attack, and
the resources and the focus of the program offices, I
think will cascade down to some of our products, and
we'll be talking to the ACRS, if necessary, on
schedules.
Managing human capital is not necessarily
an area of ACRS purview. I think it is a challenge
for the agency right now to staff up to the levels
that are necessary to support the new work in response
to the event of 9/11, as well as new reactors. And
improving our business practices, including defining
performance goals is an area that I believe the ACRS,
at least, should be aware of to know that we are
accountable for our products, and our outputs, and our
outcomes, and ACRS in many cases is an integral part
of that. As you are in our definition of work as far
as our work flow diagrams, ACRS is integrated into
that processes, and we need to be coordinated to be
successful. So that concludes my remarks, and I would
leave the remaining five minutes to -
MR. THADANI: Okay. Well -
MEMBER POWERS: Never draws any
controversy.
MR. THADANI: As always. Let me also
start out with some positive thoughts here. Every
time I meet with you, the committees always walk away
learning things, and recognizing there are areas that
maybe I need to pay a little more attention to, so I
find these dialogues extremely valuable and, I think,
important.
I particularly want to acknowledge what I
think was a tremendous amount of effort on your part,
look at research programs and the document that you
produce. It's, in my view, a masterpiece. It is
extremely well thought out. As I have indicated to
you before, I think we were in agreement in most of
the areas. There were some small differences of
views, but they were not fundamental in nature. It
was just a matter of relative timing of what we do
with some of the recommendations that you had in your
report.
Similarly, this, of course, has some
aspects of waste in it, as well. I also wanted to
note that the February 5th letter, in particular, from
the ACNW had a number of recommendations about
research that were taking to heart there are things we
can do, and there are things we cannot do. So to the
extent we, as an office, can address those, we are
doing that. And, in fact, we've made progress in some
of those areas.
What I will do since I only have about ten
minutes, I think I have eight or nine charts. I'm not
going to talk about everything. Let me briefly take
each chart and see if I can't make what the key points
may be, and we can move on. And I'll certainly try to
answer whatever questions you may have on the topics
that you will see. Let me go to the first - may I
have the first chart, please. Okay.
This is a list of - I mean, this is not
a complete list, but what I call major issues that we
have been working with the committees, and we expect
to continue to work on, different specific areas, but
for the next two to three years, this is going to take
a lot of attention and our time, and I expect a lot of
interactions with you.
I'm not going to say anything about the
follow-up to 9/11 activities, except that there is
significant ongoing effort in the Office of Research.
Much of it is classified, and I do expect down the
road that there will be some interaction with the
Committee in terms of what we're doing here.
Let me go on to the next chart. Advance
reactors, the two parts that I want to highlight first
in terms of the importance of trying to make sure we
have a reasonable framework in place, which would
guide our activities, would define how far we go in
certain areas, what the boundaries would be. And this
is clearly - this would include the role of safety
goals. Obviously, safety goals alone would not be
sufficient. The discussion is ongoing, and I know
you're looking at issues of frequency consequences
approach, and what's the role of deterministic
thinking in this process. And then how to account,
when you go to designs that are pretty unique and new
to us, how are we going to account for lack of data,
inexperience, and that would impact the quality of
PRAs and the role of PRAs, and whatever decisions have
to be made.
We're pulling together a research plan,
and the scope of the research plan is going to cover
PBMR, the GTMHR, AP1000, and IRIS. What are some of
the key technical issues? How would we go about
making sure we have appropriate tools to help us make
those independent decisions as an agency? What sort
of resources would be necessary, schedules? And this
would be a living plan. It's one that we would hope
to have a draft later this month. We have sent out
the first version for comments, but we expect to have
our initial discussions with you on the plan in April.
That means we'll get you a draft some time in March so
you would have an opportunity to have looked at it.
Just to give you a sense of what's in it,
if I can be brief. This is a snapshot. AS I said,
this is a living plan. And I think you will recognize
some similarity cornerstones approach that you're very
familiar with. The idea here was that we want to take
what I would call a systems approach to defining what
we need to do in terms of research. And you will note
that this includes also the fuel cycle issues, because
we need to look up front from beginning to end, and
not necessarily continue with the ways of the past, so
to speak.
This - the whole idea here is the idea of
completeness. We want to make sure we lay out all the
issues that may be important, and some of which may
require research effort. Our intention is to go
through this process in a very systematic way, the
kind of thinking that I talked about, part type
thinking has to be applied to each of these issues as
we go forward.
CHAIRMAN APOSTOLAKIS: So you worry about
the aging for future reactors?
MR. THADANI: Absolutely. Sixty years.
MEMBER POWERS: Well, if you're radiating
graphite, you're better to learn about aging real
quick.
MR. THADANI: Exactly. I would use the
same example, and there are some very interesting
technical issues from aging of graphite which could
have very significant bearing on the design, I think.
I'm not going to go through this. We will be
discussing a lot of this with you in April. Let me go
to the next chart.
On risk informed initiatives, obviously we
can spend an awful lot of time, but let me make just
a few comments. This is an area where we have very
extensive interactions with you, and I anticipate will
continue for the next three years, so - but just to
give - to bring to your attention that we have a Risk
Informed Regulation Implementation Plan. We updated
it, and Sam mentioned that last version went to the
Commission December 5th.
It includes everything that we at the
agency are doing in terms of applications, risk
informed thinking. I think that's a very good thing
to do, because it does bring us together in terms of
communication and so on. It includes prioritizing
activities, identifying necessary tools, resources,
and integration activities, as well.
We're trying to identify in that plan what
are some of the most critical milestones that need to
be completed before one can go on to some place else,
and what some of the cross-cutting issues are. I
think we need to do better than what we've done up to
now, but it is a good start, I think. It does
identify what some of the cross-cutting issues are.
Another effort that's good in this
document, we often talk about risk informed
regulations, and performance based to the extent
practical. WE've always said that. Here's a document
we're trying to make sure as we go forward, and with
any future rule makings, we systematically consider
factors, but we can, in fact, be performance based in
the articulation of our regulation. So that's - I
think that's a good move. It's - to me, I look at it
like a handbook. It's a good handbook, I think. And
on individual pieces, of course, you may have to go
elsewhere.
In terms of the PRA quality, you know
about the standards. I won't dwell on it. WE're
looking at all the standards, the ASME/ANS, as well as
looking at the NEI peer review document. We're
planning to pull together a guide that will help us
integrate the role of the standards, role of the peer
review, so there's one place one can see how these
things are going to be utilized. And our intention is
to pull that guide together, and we'll be meeting with
you on that guide.
I won't say anything about PTS because I
know you're up to speed, 50.44, 50.46. We had the
discussions. Human reliability analysis, I guess I
just want to make sure you know that we're sunsetting
ATHENA this year.
CHAIRMAN APOSTOLAKIS: You're what?
MR. THADANI: We're going to sunset
developmental activities within ATHENA. We're
applying it, as you know, in the area of -
CHAIRMAN APOSTOLAKIS: Speaking of cross-
cutting and working with other offices, I saw the
slides the two gentlemen from NMSS were using
recently, two, three weeks ago in a presentation to
the staff on human performance for NMSS.
MR. THADANI: NMSS, yes.
CHAIRMAN APOSTOLAKIS: They went back to
THERP.
MR. THADANI: They went back -
CHAIRMAN APOSTOLAKIS: No mention of
ATHENA, no mention of -
MR. THADANI: Uh-huh.
CHAIRMAN APOSTOLAKIS: THERP, way back.
MR. THADANI: That went to THERP. Okay.
That's good input. I told you, I always learn things
here.
CHAIRMAN APOSTOLAKIS: I know that you
guys have been doing this, or they studied it and they
said this is not helpful.
MR. THADANI: Yeah. Yeah.
CHAIRMAN APOSTOLAKIS: Okay.
MR. THADANI: Yeah. Good thinking.
CHAIRMAN APOSTOLAKIS: This is always a
question that comes to mind.
MR. THADANI: Yes. Thank you. It's a
message that's well received, and will be followed up
on. The only point I wanted to make sure you knew
under human reliability analysis reach plan, that
there are some boundaries that we're not crossing.
We're not looking at the issues of safety culture.
We're not looking at the issues of organization and
management, but we are monitoring what's happening in
the international community. And I think you know the
next piece very well. We've got significant efforts
going -
MEMBER POWERS: Let me just inject, Ashok.
We've been holding off on looking at this, your
current human reliability analysis research program
because we had the perception that your staff was a
little busy to come talk to us about this with other
activities, and I still want to do that, if not to
impose on them, but I don't want to hold you up
either.
MR. THADANI: Yeah. In fact, I'm glad you
raised that. I think we sent you a draft plan some
time ago.
MEMBER POWERS: Yes.
MR. THADANI: It's got to be revised, I
think.
MEMBER POWERS: Okay.
MR. THADANI: And that's the issue. I
think we need to revise it, and then get it to you,
and then have meetings with you.
CHAIRMAN APOSTOLAKIS: Ashok, we spent so
much time reading it, now you're revising it?
MR. THADANI: When I say revised, certain
elements. So you haven't read it as -
MEMBER POWERS: Tell him completely from
top to bottom, throw away all those comments that he
has.
MR. THADANI: I just wanted to note that
I think you know all the work that research is doing
to support NRR, in terms of looking at operating
experience, and how it can be utilized, the analysis
of this experience on a number of initiatives that NRR
has ongoing.
CHAIRMAN APOSTOLAKIS: I have a comment on
this. The rest of the slides deal with engineering
issues and so on, so maybe this is the best place to
raise it.
MR. THADANI: Sure.
CHAIRMAN APOSTOLAKIS: We made the
recommendation - well, I'm not going to raise
decision theory when we talk about fuels.
MR. THADANI: Right. Fine.
CHAIRMAN APOSTOLAKIS: When you talk about
risk informing initiatives, making decisions -
MR. THADANI: Uh-huh.
CHAIRMAN APOSTOLAKIS: - and we had a
recommendation in the research report -
MR. THADANI: Yes.
CHAIRMAN APOSTOLAKIS: - that formal
methods have existed for quite a while now, and we
recommended that your staff investigate the
possibility of taking advantage of this work that
people have done.
MR. THADANI: Uh-huh.
CHAIRMAN APOSTOLAKIS: And we don't see
anything here. And I want to make it clear, that we
are not really proposing that you use formal decision
theory in all your decision making activities. I
mean, that would be absurd, but there should be, I
think, some appreciation of what these methods can do
within the Office of Research, at least.
MR. THADANI: Yeah.
CHAIRMAN APOSTOLAKIS: And for example, in
the last bullet with the performance indicators, you
would have found this expertise useful. It would have
helped you do certain things better than they were
done in the documents we saw. And there may be other
places where you may also take advantage.
MR. THADANI: Sure.
CHAIRMAN APOSTOLAKIS: In other words,
here is a decision making agency that's using risk
information, which means uncertainty estimates all the
time.
MR. THADANI: Uh-huh.
CHAIRMAN APOSTOLAKIS: And it's trying to
risk inform its regulations, and there is this large
body of knowledge that tells you how to use these
uncertainty estimates in a rational way. And we are
not really taking advantage of that. I mean, this is
really where we're coming from. It's not that we want
you to say well, gee, you know, Sam has a problem
tomorrow. He has to make a decision, oh formal
decision. No, not at all. But there is a lot of
information there that could be useful.
MR. THADANI: Yeah.
CHAIRMAN APOSTOLAKIS: In fact, it would
be. I know it would be, and we are not really using
that. Now I know you have asked one of your staff
members to do some investigation. That's not good
enough, in my view.
MR. THADANI: Well, I -
CHAIRMAN APOSTOLAKIS: We should take it
a bit more seriously in the sense of at least there
should be a bullet there saying that you're thinking
about it.
MR. THADANI: Well, let me say that we're
thinking about.
CHAIRMAN APOSTOLAKIS: Okay.
MR. THADANI: If that satisfies you.
CHAIRMAN APOSTOLAKIS: You will take
appropriate action in the future.
MR. THADANI: Yes. In addition to that,
I have asked, and Dr. Johnson happens to be sitting
here.
CHAIRMAN APOSTOLAKIS: I know.
MR. THADANI: He is looking at what I
would call looking at the state of the art, to what
extent we can utilize these methods, which approach is
maybe better, and so on. I can tell you that we're
not embarked on an extensive evaluation. Until we do
an initial assessment, then we'll have to decide how
far we can go, or can't go in certain directions, but
it is a first step. And once we get done with that
first step, maybe I can come back and tell you more
about where we are planning to go.
I just wanted to make a note here, you
have in the past raised some concerns about the need
to do a peer review of SAPHIRE, and we have - I mean,
you had this discussion just two weeks ago. And we've
looked at all the information, and we think that we
will go forward with peer review this year. The scope
and so on is yet to be decided, but I've asked the
staff to come and meet with you before we initiate any
peer review, because I want to be sure that we are, in
fact, properly focused on whatever issues there may be
in terms of applying -
CHAIRMAN APOSTOLAKIS: You are aware of
the fact that some NASA Manager have volunteered -
MR. THADANI: Yes.
CHAIRMAN APOSTOLAKIS: - to participate
in this.
MR. THADANI: Yes, I am. Yes.
CHAIRMAN APOSTOLAKIS: Okay.
MR. THADANI: But I just wanted to let you
know, because this is just recently we decided we'll
go forward.
CHAIRMAN APOSTOLAKIS: Yeah.
MR. THADANI: Let me go to the next chart,
and I think many of you are so familiar with this,
that let me just say that much of - in terms of the
MOX fuel, the high burnup fuel, we've had a number of
PIRT meetings. WE're going to have a draft report
next month, and we'll be assessing various models.
We'll be looking at taking advantage of data from
different countries in terms of where we go. And this
is one of those ongoing interactions with the ACRS.
MEMBER POWERS: Are we ever going to get
the French data on MOX, and high burnup fuel?
MR. THADANI: Well, we - I hesitate to
say that we have reached that agreement, but I'm
hoping we're there. I'll have to - I don't think we
have yet reached that end point, but I'm hoping we'll
get there. And we have to - we're discussing options
of how can we provide the necessary resources to get
the data from the French.
MEMBER POWERS: Well, wouldn't - and you
might want to be aware that we've invited the - some
of the investigators from the PHEBUS Program to come
speak to the Committee in May.
MR. THADANI: In May? Okay. I didn't
know.
MEMBER POWERS: About both the current
PHEBUS Program and some of their plans for follow on
programs. Not that the Committee is taking any
action, just for information purposes.
MR. THADANI: Yeah. We're hoping,
depending on how `03 budget comes out in the end.
We're hoping to continue our relationship with the
French on PHEBUS Program.
MEMBER POWERS: That raises one of the
questions. We've been very supportive of your efforts
in the high burnup fuel area, and I note that you've
even expanded those activities beyond what they
originally convened, and they seem to be progressing
well.
MR. THADANI: Yeah.
MEMBER POWERS: I mean, there are some
hiccups in just getting available fuel but, you know,
those things happen in research, and that's why you
have lots of white hair, undoubtedly.
One of the questions that's going to come
up is that you're doing a lot of experiments on single
rod, specialized experiments, and they look very
useful. The question is, is that going to be enough?
Do we have to go to multi rod experiments to
understand things?
MR. THADANI: Let me - because of my, you
know, limited time, let me take that as a question to
get back to.
MEMBER POWERS: Yeah. I mean, you don't
have to give me an answer now. It's one of those -
CHAIRMAN APOSTOLAKIS: Speaking of that,
Bill, is it okay to go to fifteen -
MR. TRAVERS: Sure.
CHAIRMAN APOSTOLAKIS: Okay. Great.
MEMBER KRESS: Ashok.
MR. THADANI: Yes, Tom.
MEMBER KRESS: Before we get off of the
question of the French data, if you ever get the
VERCORS data -
MR. THADANI: VERCORS -
MEMBER KRESS: - I would be very
interested in getting it as soon as - letting me
know as soon as I could.
MR. THADANI: Okay. All right.
MEMBER KRESS: Because I could use it
right now.
MR. THADANI: We're also talking to the
Japanese, of course. I think you know about the VEGA.
Okay. Another thought here that I just want to make
sure and capture, is that we are looking at the burnup
and correlation affects on cladding, cladding of
various designs, Zircaloy 2, 4, Zirlo, M-5 and so on,
so this is something that various types of testing,
and trying to make sure that there are no surprises
there for us. This talks a little about HTGR. I just
wanted to make sure that you knew that this is going
to be captured in the plan.
The next chart, in fact, I'm not sure that
I need to say any more. We've had very extensive
discussions with the Thermal Hydraulics Subcommittee,
and we - I think you're up to speed on where we stand
in terms of assessment of TRACM and so on, and what
some of our continuing efforts are going to be.
A point I'd like to highlight here is that
I think this is an area where I'd like for us to be
proud that we've got pretty good in-house capability,
and I hope it's coming through to you, as well. We've
got a pretty strong group of people in this area.
MEMBER KRESS: Are you doing anything in
severe accident area -
MR. THADANI: Yes.
MEMBER KRESS: - on air ingression
accidents?
MR. THADANI: Yes. We are now discussing
with an Eastern European country to see what kind of
test could be done at very reasonable cost to get some
data.
MEMBER KRESS: You have good ideas.
MR. THADANI: Well, I am assuming you know
we're looking at similar approach for severe accident
codes, such as MELCORS, as we were with the DPA type
activities. Next chart, please.
I mean, this is an area where NRR and
research are very tightly integrated, I would say.
Very significant information that's coming out of
operating reactors, examples that Sam used, that
clearly require that we be prepared to support NRR in
dealing with some of those issues. And the work at
Argonne and some international arenas has been very,
very valuable, and I want to thank you for the support
that you've given us.
Our focus now is largely on a radiation
assisted stress corrosion cracking, crevice chemistry,
and some of those fundamental mechanisms that may be
involved, and trying to see how we can make sure we
are prepared, that we can do appropriate - we, as an
agency, can do appropriate inspections, and know
what's really going on.
The - again I'll not say much about
advanced reactors in this area, other than to just
note that high temperature metals, Graphite and the
containment confinement are going to be significant
challenges that we're going to have to deal with this
perspective as well.
Steam generators, we're following the plan
that, Dana, you know about, and we're on course.
Digital I and C, we've got a plan that's been
discussed with you, so let me just move on to the next
chart, unless you have questions.
Waste issues is sort of - some of the
issues that we're involved in. You touched upon dry
cask. You raised a question earlier, Dana.
Obviously, we're looking at structural material
systems and probabalistic considerations in an
integrated fashion. And we're doing a PRA. I'm happy
to note much of the work is actually being done in-
house by our staff, so it's very important.
We do have some cooperative efforts with
the Electric Power Research Institute, and Department
of Energy to looking at fuels, for example, various
burnup levels, trying to get the right data. The idea
behind this, of course, is to support NMSS and their
decisions on license extensions from 20 years, to 40,
to 60 and so on, so this is sort of a time dependent
element of this, how fast we can go.
Very quickly, package performance studies
is related to transportation cask beyond - looking at
beyond design basis types of events. Marty indicated
that, and we're looking at things like high speed
impact, and fires and so on, as part of the - when I
say we're looking at, I'm sorry. Let me clear it -
clarify it. We developed a plan, and we need to get
international support to - we can elaborate our
resources, and a number of countries have indicated
very strong interest, and I don't anticipate a big
problem in moving forward on that.
Radionuclide transport, John, there was
very extension discussion, workshop. Let me not -
I'm looking forward to your report on that workshop
because it would be very useful to get your insights
on where you see things are going.
I already touched on the whole issue of
looking at the full cycle for these new reactors.
MEMBER KRESS: Are you doing anything more
on spent fuel pools?
MR. THADANI: Spent fuel pools, some of
the work we're doing can be - let me say, I can't
talk about what we're doing as a following to the 9/11
event. We're doing some work there, but in terms of
the source term, some of the other work we're doing
can be of value to spent pool fuel issues.
Let me just quickly say that both - you
know, the expert panel which was headed by
Commissioner Rogers and others, have clearly indicated
need that we in the Office of Research need to do a
better job of telling what we're doing, why, and how
it can be of value to the agency and the public at
large.
There's sort of a synopsis. I think you
know of the paper that we prepared, lay out, vision,
mission, role and responsibilities of research. And
I think you know what is confirmatory, and what is
anticipatory. Let me highlight two or three points.
We're going to - we are embarked on a
plan to do a much better job of communication,
internal and external to the agency. And a number of
initiatives that we have, planned, some of them
planned, others are ongoing. But in terms of the
internal communication, we have - I think the biggest
step that has helped us in terms of better
communication has been what I call - what I guess we
call leadership team meetings. At the division
levels, there's very frequent contact between the
offices, NRR and research in particular. And I know
Cheryl meets with NMSS folks on a weekly basis to make
sure there is proper communication taking place.
I think that has improved our
understanding of the challenges NRR has, and NRR's
understanding of how the work we do fits in in terms
of the decisions they have to make.
The other part that's I think helping us,
and that we still have to go further is the Research
Effectiveness Review Board. I think you know
generally what the plans are. They've been focusing
in largely on the user needs, and the process of user
needs and follow through. And that's going to
continue, and I think it can only be of value to us.
I want to highlight one item under
anticipatory research. You criticized us, and I think
it was George, you in particular. And I notice that
- I wanted to get your attention, and actually it was
Rogers Expert Panel also, that in anticipatory
research, we were too inward looking, that we didn't
seek ideas and concepts from a broader spectrum of
audience, so I have just this week sent letters, first
internally to NRR/NMSS regions, to research staff,
briefly giving background, and seeking ideas from
them.
I've also sent letters to Nuclear Energy
- Nuclear Engineering Department Heads Organization,
to NEI, to EPRI, to UCS, Nuclear Control Institute,
and also in Federal Register Notice, making sure that
we're not interested in just a lot of ideas. We're
interested in ideas which are focused on the areas
we've identified. And also, ideas that have a
reasonable chance of success in terms of getting
there.
We may well get some good ideas. Clearly,
we're not going to be able to consider them for this
budget cycle, because we have to provide our input in
a matter of six weeks or so, so it can't be done. And
I wanted to make sure that various organizations had
enough time to really think about these things. I'm
seeking input by June 1st, and we will then set up a
group that will evaluate the recommendations and
ideas, and then we will consider them for the
following budget cycle. It's taken a year longer than
I had hoped, but nevertheless, we are moving in that
arena.
The other point that I want to make that
I think is helping in coordination is that we're now
giving a number of briefings, for example, program
review committee briefing and our research did it
jointly, NMSS and research did it jointly. I think
these are good steps forward. It's working better.
I think coordination is improved. We need to go
further, but it's going in the right direction. With
that I will stop.
MEMBER SHACK: There's one exercise here
that I happen to be the guinea pig that you did once
upon a time, and it's never been repeated, but I
thought it was quite successful, where you had a
research program with an open meeting with the public.
MR. THADANI: Yes.
MEMBER SHACK: And then we had people from
the NRC -
MR. THADANI: Yes.
MEMBER SHACK: - utilities, owners
groups, and intervenor groups. And I thought it was
very successful. I think the intervenor groups went
away with a much better feeling that the NRC was
getting some independent information. They had a
chance to ask questions, which you never can ask
questions of a report, and all in all, it was a pretty
successful thing. And then, you know, it sort of -
we tried it once, and -
MR. THADANI: Yeah. I think you're
exactly right. I know when that meeting was held, and
it went very, very well. We have done a little bit
more of that, but I think it's clear, and I'm glad you
brought it up. We need to maybe expand in that area,
as well.
MR. TRAVERS: That is a good point, as you
probably. I'm sure you know, that increasing public
confidence is one of our four agency strategic goals,
so we're always looking for some good ideas on how to
achieve that. And frankly, it's a difficult thing to
get your arms around in any particular forum.
That does complete our presentation. I
wanted to respond though. I didn't mean to keep you
waiting, Dana, on any negatives that we had, hold you
in any suspense. But the answer quite simply from a
significant negative perspective, at least on our
part, is that we don't have any. We think we enjoy
today the kind of professional relationship that is
advantageous to our separate, but very related
responsibilities, under statute. Can we do better?
We can always do better. Can we do better from our
standpoint in providing documents in a timely way,
getting input from you in a timely way to support some
of our activities? Of course. But I think what I
would like to send as a final message is that we very
much appreciate the relationship that we do have. We
think its value added, frankly, from our perspective.
Do we agree with everything you recommend? No, but we
try to get back to you with reasoned basis for not
agreeing in all instances. But I would say that in
many, if not most instances, we take and implement
much of the recommendations and -
MEMBER POWERS: Well, I think the message
I give you here is that I, certainly, and I think the
Committee as a whole is not adverse to you coming back
and saying something was not helpful. Okay. Some
direction that we're taking, some approach that we
take, something like that. I mean, we can disagree
with you too, but it's useful for us to know the
downside, as well as the bright side. And, I mean,
some mechanism to just indicate something is not
helpful.
MR. TRAVERS: Okay. That's fair. I think
we do have mechanisms for doing that, and we'll take
that as a -
MEMBER POWERS: Sure.
MR. TRAVERS: Thank you.
CHAIRMAN APOSTOLAKIS: Okay? Well, any
members have any comments, observations they would
like to make?
MEMBER POWERS: Ashok, you mentioned
letters and a Federal Register Notice that you were
sending.
MR. THADANI: Yes.
MEMBER POWERS: Could we get a copy of
that?
MR. THADANI: Absolutely, yes.
MEMBER ROSEN: I have one, George.
CHAIRMAN APOSTOLAKIS: Sure.
MEMBER ROSEN: I was struck by the fact
that you did not ask the stakeholders directly, the
licensees for input on the research program. You did
it through NEI. I think it might be valuable for you
to think about that.
MR. TRAVERS: Thanks, Steve. Yeah, I
will, and I think I understand the message there.
CHAIRMAN APOSTOLAKIS: Any other comments?
All right, gentlemen. Thank you very much. WE really
appreciate your taking the time to come to talk to us,
and let's hope that we'll do this again, maybe in the
not too distant future. Okay? Thank you very much.
We're recessing until 1:30.
(Whereupon, the proceedings went off the
record for a lunch break at 12:35 p.m.)
. A-F-T-E-R-N-O-O-N P-R-O-C-E-E-D-I-N-G-S
(1:35 p.m.)
CHAIRMAN APOSTOLAKIS: Do we have a
quorum? One, two, three, four, five, six. All right.
We're back in session. The next item on
the agenda is the Proposed Final Revision to
Regulatory Guide 1.174 and SRP Chapter 19.
Ms. Drouin?
MS. DROUIN: Okay.
My name is Mary Drouin with the Office of
Research, the Probable Risk Analysis Branch.
CHAIRMAN APOSTOLAKIS: Excuse me, Mary.
I was suppose to make an announcement.
MS. DROUIN: Sorry.
CHAIRMAN APOSTOLAKIS: You have this
hand-out number 10 in front of you, members? Number
10? You all have this? It's very thick.
Reconciliation of ACRS Comments and Recommendations.
Please take a few minutes to read it today. We're
going to discuss it tomorrow. It includes the EDO's
response to our letter on the reactive oversight
process, steam-generated, tube integrity. There's a
lot of good stuff here. So please do that.
Okay, Mary. Sorry.
MS. DROUIN: Are we ready?
CHAIRMAN APOSTOLAKIS: Yeah.
MS. DROUIN: Okay. I'm Mary Drouin with
the Office of Research. And today I'm here to give
the committee a status of where we are in terms of our
update on Regulatory Guide 1.174. Part of that is
also a status in the plan for endorsing the PRA
standards. For example, the ASME PRA standard, which
is going to be coming out public very imminently and
the industry program, particularly in the NEI-00-02,
which is the Peer Review Certification Program.
We're here today not just to, of course,
give you a status, but, as always, we solicit and
welcome the committee's input on our work. Because as
we go through, we're going to go into detail what our
plan is beyond updating Reg Guide 1.174. And so we're
in the midst of formulating this plan right now.
A lot of the basis for a plan -- before we
get into it and where we are on the Reg Guide -- is
that I think it's very important that we spend just a
few minutes going back historically and why we are
where we are today. And this had a lot of input into
some of the decisions we made in terms of our update
and how we plan to endorse the standard.
If we go back to April 18th of 2000,
that's when we received direction from the Commission
that said the staff should provide its recommendations
to the Commission for addressing the issue of PRA
quality until the ASME and ANS standards have been
completed, including the potential role of an industry
peer certification process.
So a lot of that got started. And we went
through a very intensive effort. And we issued
SECY-0162, that I know the ACRS looked at. And we
actually received comments from the ACRS on that
document.
But if you look at 162, what I've put here
are four of the major things that the staff committed
to in that SECY document, which are very important in
terms of how we decided to update Reg 1.174 in our
plan for endorsement of the standards.
First of all, one of the things that was
in SECY-0162 is that we went through and we identified
the scope and the minimal technical attributes at a
functional level that were necessary to ensure that
the PRA is capable of providing certain results. And
we listed those results, like core-damage frequency
LERF, what the dominant contributors are out of a PRA
on both a relative and absolute basis. And all of
that information was contained in that Attachment 1 to
SECY-0162.
Also in that document we had indicated to
the Commission that, if appropriate, the staff would
endorse them, meaning the standards, in an updated
Regulatory Guide 1.174 -- but what's also important --
or elsewhere to support other risk-informed
activities. Also, we indicated in that document that
during our endorsement we may take exception of
provide additional criteria in those places where we
thought the standard or the industry program was not
sufficient.
And lastly, that the information that was
contained in SECY-0162, particularly the two
attachments, that we would include those in the update
of Reg Guide 1.174. And the basis for including that
in there, because at the time I thought was to endorse
the standards in the updated Reg Guide 1.174.
We did receive an SRM from that SECY. And
the main points that came out on this SRM is the
Commission did tell us to proceed forward. They
wanted timely resolution to this. And that we should
expand our discussion on further examples. In fact,
we received a similar comment from the ACRS, the
examples that were in Attachment 2, that if we could
do more in that area, that would be good.
But that's all I was going to go in terms
of background, and then I'll get right into where are
we in terms of our status of Reg Guide 1.174.
We did issue a draft update of 1.174 and
SRP Chapter 19 in June of 200, and we went out for
public review and comment. That update, there were
four major changes that were included in that update.
The first one was, where we had new,
unforeseen hazards, we would potentially ask for more
information in that area. Also, in regards to
increasing in power level and fuel/burnup, there was
some concern in terms of their impact on LERF, and
that we may need to provide some additional guidelines
there.
MEMBER ROSEN: Is there something missing
from that Number 2? "Levels above megawatt thermal,"
it says.
MS. DROUIN: Oh, sorry; 3,800 I believe
was --
MR. CUNNINGHAM: Okay.
MS. DROUIN: Also included in the update
is, we took that information from Attachment 1, which
was identification, a description of the minimal
technical attributes. That was added to the Reg Guide
and to the SRP. And also, the information from
Attachment 2, which was a discussion on example
applications using risk insights in the
decision-making process.
MEMBER ROSEN: Can you say something quick
about how you picked that number? 3,800 I mean.
MS. DROUIN: No. This is not an area that
I was involved in. But let Mark answer that.
MR. CUNNINGHAM: This is Mark Cunningham.
Traditionally, there's been an implicit,
unofficial limit to the allowable reactor power for
lightwater reactors that's been floated around the
Commission. And that was 3,800 megawatts. It's not
a legal restraint or anything like that. But it's
been the presumed maximum legal power that --
MEMBER ROSEN: So it's a legal thing,
really, not so much impact on LERF. I mean, because
nothing happens between 3,800 and 3,801 to LERF.
MR. CUNNINGHAM: That's correct. The
interest was, if you wanted to go to 4,000 or
something like that. At some point, the LERF
criterion would be affected. And this was putting a
proviso in for public comment that we may need to
revisit that if you went substantially above 3,800.
You're right. If it went to 3,801 or 3,860 or
something like that, we don't expect it would change.
MS. DROUIN: If you look at the second
sub-bullet there, it does indicate that it's not
expected to have an effect. But there is some -- from
my understanding, there are some experts out there
that aren't necessarily in agreement, so this is still
a topic under discussion.
MEMBER ROSEN: Well, that's not what the
second bullet says, Mary. I beg to differ. It says,
"increases in fuel/burnup beyond 40,000 megawatts."
So it doesn't talk about the power level.
MS. DROUIN: You're right. You're
correct. You're correct.
MR. CUNNINGHAM: The concern was -- in
effect, the original concern was the combination of
perhaps going to higher powers, plus burnup, plus the
potential in there for use of mixed-oxide fuel. So a
combination of several things that led to the concern
about LERF.
MEMBER ROSEN: Okay. Thank you.
CHAIRMAN APOSTOLAKIS: Now, in the
discussion of example applications using these
insights in the decision-making process, are you the
staff prepared to ask the industry to do an
uncertainty analysis, the standard uncertainty
analysis, in all applications, considering that as
part of the risk insight?
MS. DROUIN: If you look at Reg Guide
1.174, there is a discussion there on certainty, and
they are suppose to address uncertainties. The level
at which you're going to address the uncertainty is
not going to be the same on every application.
CHAIRMAN APOSTOLAKIS: But the propagation
of uncertainty is due to uncertainties in failure
rates and frequencies of initiating events. It's sort
of a routine kind of thing now with the computer
programs that are available.
Why don't we ask everybody to do that?
And then the rest of it can be addressed as
appropriate.
CHAIRMAN APOSTOLAKIS: When it comes to
your parameter uncertainties, the standard right now
does ask for them to do that.
CHAIRMAN APOSTOLAKIS: In all
applications?
MS. DROUIN: Well, the standard doesn't
get into applications. It's application-independent.
CHAIRMAN APOSTOLAKIS: But I remember
there were three columns. In one column they said,
mean values are good enough. That's not true anymore?
MS. DROUIN: The middle column of the
standard, which is your best practice, does ask them
to do that.
CHAIRMAN APOSTOLAKIS: To do uncertainty?
MS. DROUIN: Yes.
CHAIRMAN APOSTOLAKIS: Okay. Wonderful.
MEMBER KRESS: Mary, before we leave this
slide, I just now got through reading Number 2 under
your second bullet. And the second sub-bullet under
there, what's the basis for saying that "increases in
fuel/burnup are not expected to have an appreciable
effect on LERF guidelines"?
MS. DROUIN: Again, I'm going to turn this
back over to Mark. You're out of my area of expertise
on this one.
MR. CUNNINGHAM: In effect, what the
committee's getting at is the same issue that was
raised in public comment, that this is a rather vague
type of -- perhaps we need to do this if you go above
this. And it wasn't very well founded as to why
3,800, why 40,000, and that sort of thing. And Mary
will come back later to how we're dealing with that
public comment.
We got the same comments. And I think
we've decided we're going to take an action to be much
more definitive about this, that say we'll do some
work to say, is there any reason to believe that the
LERF criteria in 1.174 would be compromised if we went
substantially above these things. We're putting the
burden back on us in effect.
MEMBER KRESS: I don't even like the way
it's stated, because, of course, it's going to be an
effect if you go beyond -- you meant to say
substantially beyond 70,000 or something?
MR. CUNNINGHAM: Yes. Yes, exactly.
MEMBER KRESS: Well, you'll need to come
up with some level of burnup where you think the LERF
criteria ought to be corrective.
MR. CUNNINGHAM: Yeah, that's right.
MEMBER KRESS: That's the idea.
MR. CUNNINGHAM: And in effect, that's
what we're doing. We're going to take the
responsibility back on ourselves to do that sort of
calculation, rather than have this sort of vague
guidance out there that licensees, perhaps, wouldn't
know what to do with, or probably would not know what
to do with.
MEMBER KRESS: Are we going to take
advantage of the PIRT that's undergoing?
MR. CUNNINGHAM: Yes. That's the expert
panels, the rest of that --
MEMBER KRESS: Oh, that's what you mean.
That's what you're talking about.
MR. CUNNINGHAM: That's right. That's a
veiled reference to the PIRTs that have been going on.
Yes, sir.
MEMBER KRESS: Okay. I appreciate that.
MEMBER ROSEN: I think you've said all the
right words now. But licensees who are at 3,800
megawatts -- and I know one in particular -- will need
to know ahead of time what they're about to run into
if they want to increase their power level. What are
the additional requirements implied by these bullets?
MR. CUNNINGHAM: Yes. And, again, that
was the type of comment we got, and it's a legitimate
comment. And we're going to have to resolve that
separately. And again, given what we know about LERF,
we don't expect if they went from 38 to 39 --
MEMBER KRESS: Before you leave this slide
--
MS. DROUIN: It's the same slide.
MR. CUNNINGHAM: But we need to clarify
that.
MEMBER KRESS: I have one more comment on
this business of LERF and 1.174. 1.174 doesn't seem
to deal with sites that have multiple units on it,
more than one plant. It's my view that LERF is a site
criteria. And that 1.174 ought to recognize that, and
talk about the effect of multiple units on a site in
terms of your LERF criteria that you're going to use
in 1.174. That's a comment you need to think about.
MR. CUNNINGHAM: You're right --
MEMBER KRESS: In multi-unit sites, it
ought to have some influence on your LERF criteria.
CHAIRMAN APOSTOLAKIS: Which is really
relevant to PBMR.
MR. CUNNINGHAM: Understood -- advance
reactors are --
CHAIRMAN APOSTOLAKIS: No, but it's the
same idea.
MR. CUNNINGHAM: Oh, I understand. Yes.
CHAIRMAN APOSTOLAKIS: You have 10 modules
there, I mean, you can't expect the criteria to be the
same as if you had one.
MR. CUNNINGHAM: Yes.
MEMBER KRESS: And if you had more than
one reactor on the site, you couldn't expect the same
LERF criteria.
CHAIRMAN APOSTOLAKIS: Yes.
MS. DROUIN: Continuing with the same
discussion, all I've shown on the next slide is the
comments we did receive in these four areas in terms
of the unforeseen hazards, we didn't receive any
comments from the public. And as the discussion's
been going on, you aren't surprised by the comments we
got. We know that there was no justification for it.
We know we needed to provide -- if we have that in
there, to provide additional step guidance.
And then we also, in terms of the
information from 162, the two major comments we got is
that there was inadequate explanation of discussion
for the purpose and use of the information,
particularly, as it pertained to the information from
Attachment 1, which got into the minimal technical
attributes, and also a lot of comments we got that
this was the inappropriate document to include that
information in.
So based on that, where we are in terms of
the update for Reg Guide 1.174 in those four areas, is
that in terms of endorsing the PRA standards and the
Peer Review Program is that we would endorse that
elsewhere in a different document and not in Reg Guide
1.174. So our plan at the moment is to issue Reg
Guide 1.174 with three of those changes. And one of
them is still iffy. We have some work to do to
determine whether or if we're going to be doing it.
First, in terms of the unforeseen hazard,
as I said, we've received no comments on that, so that
part of the updates to Reg Guide 1.174 would remain.
In terms of SECY-0162, the information from Attachment
1, which is the minimal technical attributes, that
information has been stripped out. But we are leaving
in the information in terms of the examples.
In terms of the increase in the power
level and fuel/burnup, as Mark indicated, we still
have some more discussion ongoing to determine if we
will keep that in there; and if we do, to what extent
to how we will have it in the document. So the
decision hasn't been made yet on that particular item.
MEMBER ROSEN: Mary, your first bullet, I
would have expected that to say, "Endorse industry
standards/Peer Review Program," rather than "or,"
since the Peer Review Program is embedded in the
standards.
MS. DROUIN: I'm sorry. That is a typo.
That should not be an "or," it's an "an."
MEMBER ROSEN: It should be?
MS. DROUIN: That's an "an." Thank you.
MEMBER ROSEN: Okay. Now I'm happy.
MS. DROUIN: It's not an "or."
MEMBER ROSEN: Right. The Peer Review
Program is part of the industry standard, right?
MS. DROUIN: Yes. Thank you.
MEMBER KRESS: Your first sub-bullet under
the second bullet, when you may call for more
risk-related information, do you see that as a
sufficient handle for you to get the information when
you need it? If you -- I don't know. If I had a PRA
that told me that the increase in risk were something
beyond which I'm comfortable with, then I could say I
need a PRA. It's a circular argument.
How do you make these judgments without
having a PRA in the first place?
CHAIRMAN APOSTOLAKIS: You mean the first
sub-bullet?
MEMBER KRESS: The first sub-bullet under
the second bullet.
CHAIRMAN APOSTOLAKIS: Yeah, risk-related
information.
MEMBER KRESS: Yeah. You need to know
what the risk implications are before you ask for the
risk information. I'm having trouble figuring how you
do this circularity.
CHAIRMAN APOSTOLAKIS: I'm troubled for
this, but for an additional reason. 1.174 is intended
to apply to risk-informed applications, not to general
applications. So I don't understand what that means.
MR. CUNNINGHAM: If I might, we might go
back a year or two in history on this.
This issue was subject to committee
meetings and a commission paper, and an SRM probably
a year-and-a-half or so ago. The starting point of
the potentially circular argument is, a licensee
chooses to come in with a proposed license amendment,
and the licensee does not choose to use risk
information.
CHAIRMAN APOSTOLAKIS: And that's what's
confusing me.
MR. CUNNINGHAM: Yes.
CHAIRMAN APOSTOLAKIS: Because then 1.174
does not apply.
MR. CUNNINGHAM: However, this is a
statement in 1.174 that said, "if the licensee does
not submit the information, the staff has the
authority to request the information on the risk" --
pardon?
CHAIRMAN APOSTOLAKIS: It's already
something in the books that says, if it's an issue of
article protection, you can do that. Does this go
beyond that?
MR. CUNNINGHAM: This is a clarification
of that in effect.
CHAIRMAN APOSTOLAKIS: So it goes beyond,
because this doesn't have to be article protection.
MR. CUNNINGHAM: I believe so.
CHAIRMAN APOSTOLAKIS: Oh.
MEMBER SHACK: So it's not new policy.
MR. CUNNINGHAM: This is not new policy.
CHAIRMAN APOSTOLAKIS: But the original
policy was you had to justify it in terms of article
protection.
MEMBER SHACK: Undue -- what's the word?
CHAIRMAN APOSTOLAKIS: Undue risk is the
same thing.
MR. CUNNINGHAM: That's right. And we
didn't come in prepared to go back to the discussions
from a year or two ago. But all of that was discussed
in the commission paper of a year or two ago that --
CHAIRMAN APOSTOLAKIS: It seems to me this
--
MR. CUNNINGHAM: You may remember Bob
Halep was the staff contact on that.
CHAIRMAN APOSTOLAKIS: I remember that.
And the whole thing had to do with undue risk or,
equivalently, article protection.
MR. CUNNINGHAM: Yeah.
CHAIRMAN APOSTOLAKIS: And now it seems
it's going beyond that, which is fine with me. It's
going in the right direction.
MEMBER ROSEN: I was wondering if you were
objecting.
CHAIRMAN APOSTOLAKIS: No, I'm not
objecting.
MR. CUNNINGHAM: This is an incorporation
of what the Commission approved about a
year-and-a-half or so ago into this regulatory guide.
CHAIRMAN APOSTOLAKIS: I'm just wondering
why it's here and not somewhere else. It should be in
the deterministic guides, regulatory guides.
MR. CUNNINGHAM: It may go there as well.
CHAIRMAN APOSTOLAKIS: You're coming for
this. But because of this, we want you to also do
this.
MR. CUNNINGHAM: This is not the only
place that this comes in.
CHAIRMAN APOSTOLAKIS: Ah. It's a
multiplicity of things.
MR. CUNNINGHAM: There are a multiplicity
of places that this guidance shows up. I believe the
principal place is to go over to the people who would
-- guidance to those people.
CHAIRMAN APOSTOLAKIS: You're going in the
right direction anyway.
This business of integrated
decision-making process, where the analysts -- not the
analysts -- the staff has to take into account risk
information, difference in depth philosophy, margins
and so on, we never really told them how to do that.
And I understand that in the revised version it's also
-- I mean, that you have not touched.
Are people comfortable about this? I
mean, it's so subjective it seems to me. Not that I
know how to do it better, but should we be trying to
understand a little better the whole process and go a
little beyond? I mean, we're just giving five boxes
now with arrows. And we're saying there's going to be
this integrated decision-making in the middle, somehow
taking into account the margins, somehow taking into
account difference in depth, and somehow taking into
account the risk insights.
Can we be a little more explicit some time
in the future?
MR. CUNNINGHAM: I think we could.
CHAIRMAN APOSTOLAKIS: And there is a
beautiful discussion on uncertainty. And then, of
course, the staff was not requesting uncertainty
estimates, but now you're saying you will. So that's
a good thing.
MR. CUNNINGHAM: The ASME standard does.
But, yes.
CHAIRMAN APOSTOLAKIS: Which is good.
MR. CUNNINGHAM: Yes, which is good.
CHAIRMAN APOSTOLAKIS: Not because I can't
see you disagreeing with -- let's go to a subcommittee
meeting if you do. But this issue of decision-making
--
MR. CUNNINGHAM: Yes. One of the issues
that has kind of -- well, the ACRS raised it last year
in their report on research, and we've been mulling it
around, of kind of lessons learned from these
decisions or similar types of decisions-- is there
something that could be done better or clear guidance
provided and that sort of thing. It's a fair
question. And we're formulating our research budget
for next year and things now. And that's one of the
things that's on the table-- is there value to that
work in that area.
MEMBER ROSEN: I happen to have a problem
with that. I think, George, what you're asking for is
some guidance on how to think.
MEMBER POWERS: I definitely need that.
MEMBER ROSEN: I serve on one of these
expert panels still, and have for a long time. And I
think what it is, is that it's a process of
integrating all of these issues --
CHAIRMAN APOSTOLAKIS: Right.
MEMBER ROSEN: -- with a lot of experience
and knowledge of a particular subject matter.
CHAIRMAN APOSTOLAKIS: But Steve, it was
the same thing with the senior management.
MEMBER ROSEN: But I wouldn't know how to
follow a procedure to do that.
CHAIRMAN APOSTOLAKIS: Well, it doesn't
have to be a procedure. The senior management was
doing the same thing, and the industry objected as it
being too opaque.
MEMBER ROSEN: Yeah. But that was in the
regulatory process.
MEMBER KRESS: Now the public's going to
object.
CHAIRMAN APOSTOLAKIS: So the more
transparent you make the process, the better off you
are.
I repeat, I don't know of any methodology
that will do this, but it seems to me that it's
something that we ought to be thinking about.
For example, all this debate we've had
among ourselves regarding the rationalist approach and
so on. That's part of what I have in mind-- let's
understand better the margins. Maybe one of these
days it will dawn upon us that we should quantify
those margins in terms of probabilities. And then
difference in depth and margins will not be so
different. But that's not something for tomorrow.
VICE CHAIRMAN BONACA: But there are some
rules, I guess, that come in as the stuff -- looks at
these things coming in, I guess. I tend to think the
way that Steve does.
CHAIRMAN APOSTOLAKIS: I'm not saying
these are wrong -- this is the way to do it.
MEMBER ROSEN: No. There are clearly some
rules. And you, one of the authors of the papers that
we've used -- hate to flatter you -- no, that's okay.
MEMBER KRESS: Go ahead.
CHAIRMAN APOSTOLAKIS: I'm above that.
MEMBER ROSEN: We use papers by you and
others on expert elicitation techniques in training of
the members of the panel so they know what's going on
when the chairman says, stop, I haven't heard from
you, Tom, pooling out the quiet guy.
CHAIRMAN APOSTOLAKIS: That's part of
structuring the process. Yes, all these things are
important.
MEMBER ROSEN: So it's not like these
panels are operating without any guidance.
CHAIRMAN APOSTOLAKIS: And I never said
that.
The other issue -- there's a beautiful
discussion -- in the original 1.174. The model
uncertainty, uncertainty and incompleteness. And then
what? I mean, it's a demonstration that somebody
understands these issues, but was it ever used by the
staff. And that's where these case studies of large
measurement would be enlightening.
Did anybody open it up and say, well, when
you Mr. Utility come here, look what we're saying
here, have you tried to do this, say something about
model uncertainty or incompleteness and all that? Is
it being used, or was it just something that was
placed there to show that we have good people on the
staff?
Mary, why don't you continue?
MS. DROUIN: Okay. Well, if we aren't
going to endorse --
CHAIRMAN APOSTOLAKIS: You are going to
endorse? Whoa.
MS. DROUIN: Let me finish my sentence.
If we are not going to endorse it in Reg Guide 1.174,
what is our plan for endorsement? And so that's what
we're going to talk about now.
CHAIRMAN APOSTOLAKIS: The standard.
MS. DROUIN: The standards and NEI-00-02.
Well, if you look at the various
risk-informed activities that we're involved in, you
will see a lot of them use PRA information in their
decision-making process. And that leads you that
there is a common element dependency. And it leads
into the question that the Commission keeps raising on
PRA quality and trying to understand or get to know
what is the level of confidence of the PRA results
that you're going to use for your insights.
So this is a common thread, whether you're
looking at Option 3, whether you're looking at
licensing actions using Reg Guide 1.174, tech spec
changes, your categorization. They all have this
common dependency there. And we do feel very strongly
that the PRA standards and that the Industry Peer
Review Program can help in this area in terms of
identifying your strengths and weaknesses.
So where we finally came to is that we
felt there ought to be this single supporting reg
guide and SRP to bridge that gap and answer that
question that would kind of cross and support all
these activities.
MEMBER KRESS: Do you envision the reg
guide to be binding on the SPAR models?
MS. DROUIN: I'm sorry?
MEMBER KRESS: Do you envision such a reg
guide that would dictate the quality of the SPAR
models?
MS. DROUIN: I do know that when we sent
out internally the Reg 14A. And it didn't just start
then. But I know that Pat did look at the SPAR models
to see how did they match up against the standard.
And there were some places that wasn't a match. And
my understanding is that it's his intent to correct
that.
MEMBER POWERS: Mary, the industry-review
approach kind of goes through and looks at a
licensee's PRA. And comes back and says, well, you
get an A here and a B here and a C here. Then it
says, this is good for this class of applications. Is
my characterization correct?
MS. DROUIN: They don't give an overall
classification for the entire PRA. They give it on a
-- I don't think they call it an element. They call
it -- I can't remember the word they use, but I'll use
the word "element." But on an element by element.
And they have like 110 elements in there that they
look at. And they grade it on an element-by-element
basis. But at the end they don't come in and say,
this PRA can support; they do it on the individual
pieces.
MEMBER KRESS: If you were to endorse that
and want to use, then, such a PRA that had these
grades and the various elements, and you're wanting to
say, now I want to use this PRA to do this with, how
would you make that jump?
MS. DROUIN: Well, that's no different --
that challenge there is no different than the
challenge we have on the ASME standard because the
ASME standard also has the three categories.
MEMBER KRESS: The three categories.
That's right.
MS. DROUIN: For any given application,
you do not need to have the same level of detail for
every element. You might have an application that's
just dealing with, say, ATWS. That doesn't mean,
then, that you necessarily have to have a very
detailed, robust LOCA analysis. Maybe but for some
other application you might need the LOCA, and you
don't need the ATWS.
So for any given application, to have that
vertical slice and say you need it to the same level
of degree is not -- so you have to look at it on an
element-by-element basis.
MEMBER KRESS: So you identify which of
those elements are important to your application, and
see what grade you got in those. Now, suppose four of
them were important. And one of them got an A, one of
them got a B, and two of them got C's? Do you take
the lowest, and say, since we got C's in two of them,
and they're important, I can't use it for this? Or do
you have a process for doing that?
MEMBER ROSEN: Let me try to answer your
question. Mary, just step back for a minute.
MS. DROUIN: Okay.
MEMBER ROSEN: The persons who are
submitting this application to the staff is the
licensee. He very well knows the grades he has in
those areas, because he's already been peer certified.
And what he's probably done is made some priority cut
on his certification and said, these five things I
better fix right away because I've got applications
that will be affected by them. And the rest of them
I don't care about for a while. So he's made that
judgment, and then improved his PRA in those areas,
presumably, to support the applications he has in
mind.
MEMBER KRESS: Does he have to go in and
get another peer review?
MEMBER ROSEN: No. No. But he does have
to come to the staff. At that point, he comes to the
staff and says, okay, I want to this, risk-informed
application. The staff says, well, let me look at
your peer certification. He shows them that. And he
shows them, presumably, what he's done to fix the
areas if they were in the area that have an effect on
his pending application. But that's the way I would
see the process working.
MEMBER KRESS: That could probably work.
MR. CUNNINGHAM: So the fundamental tenet
here is that a PRA doesn't have to be perfect in every
area in order to do some small thing, to do something.
It's very application-specific and site-specific and
application-dependent. And that was where the
industry and the staff could come together and agree.
Now we'll see how that plays out in implementation.
I think it will play out along the lines that I just
outlined.
CHAIRMAN APOSTOLAKIS: But they will not
come to a staff and say, you don't need to review this
element because we've got an A.
MEMBER ROSEN: No. Of course not. I
mean, they can say that, but I don't think they will.
CHAIRMAN APOSTOLAKIS: The staff may
choose not to pay much attention because of the A, but
having an A doesn't mean much.
MEMBER POWERS: If I'm an A and a 1, and
I want a power uprate, and I want to do that on a
risk-informed basis; and I come in and I say, okay,
well, here's my analysis of operational events with my
PRA with and without this power uprate, and I don't
have a fire PRA, that's going to be okay?
MEMBER ROSEN: I don't know how to answer
your question.
MEMBER POWERS: Well, say yes.
MEMBER ROSEN: That's a heuristic device
so you can say that's ridiculous?
MEMBER POWERS: So I can go, aha, put the
risk down in a term, and the risk assessment for N01
is fire.
CHAIRMAN APOSTOLAKIS: All because you
don't know it's the risk because you haven't done the
PRA.
VICE CHAIRMAN BONACA: You are getting
ahead of me. The application you are quoting as an
example, that's pretty broad. I would expect that you
would want to have a pretty complete PRA, if you want
to make statements on risk information, because it's
really a full licensing of the plant at the higher
power level.
CHAIRMAN APOSTOLAKIS: The point is that
the A, B, C's are for the licensee.
MEMBER ROSEN: 1, 2, 3's.
CHAIRMAN APOSTOLAKIS: The licensee gets
an A from his peers. He knows that when this element
is done, the state-of-the-art work, the staff is not
constrained. The staff can review it and say we don't
like it, right?
MEMBER KRESS: So there's no need for the
staff to endorse the peer review process then.
CHAIRMAN APOSTOLAKIS: No. Well, in
principle, really, no. I guess the industry wants to
have some feeling that whenever they're doing all this
effort, there is some receptivity.
MEMBER POWERS: Here's what I'm concerned
about, George. We come in and we say, those parts of
your scope that you've done and you've got an A on,
but the real issue is whether the scope is right.
MEMBER KRESS: You get an F on that.
MEMBER POWERS: Sure.
CHAIRMAN APOSTOLAKIS: I assume they
thought about that.
MEMBER ROSEN: That's right. Was your PRA
-- is it good in the areas that are important to the
application you're proposing. Staff will judge that.
And certain members of the staff -- and particularly
sitting in front of us -- were active in the ASME PRA
standard development; were you not?
CHAIRMAN APOSTOLAKIS: Many was active.
MEMBER ROSEN: Yes.
MS. DROUIN: I was active.
MEMBER ROSEN: So she has a good knowledge
of the PRA standard --
MS. DROUIN: I thought I might just have
one meeting where I wouldn't hear those words.
MEMBER ROSEN: I chose to use the words
"active" and "lady."
CHAIRMAN APOSTOLAKIS: She was the easiest
person on the staff --
MEMBER KRESS: We're not debating the lady
part. I mean, that goes without saying.
CHAIRMAN APOSTOLAKIS: But don't forget
those, Steve; that even if the staff along,
ultimately, some of these things will come before us.
MEMBER ROSEN: Yes.
CHAIRMAN APOSTOLAKIS: And we're not bound
by A's and B's and C's.
MEMBER ROSEN: No. You're not bound by
anything.
CHAIRMAN APOSTOLAKIS: The meeting is
getting out of hand.
MEMBER KRESS: It's deteriorating.
CHAIRMAN APOSTOLAKIS: Let's come back.
It's rapidly degrading.
MS. DROUIN: Okay. Well, let's try and
explain a little bit of what our vision is for this
new regulatory guide that we plan on putting together.
CHAIRMAN APOSTOLAKIS: And which we will
have an opportunity to review at some point?
MS. DROUIN: Absolutely.
CHAIRMAN APOSTOLAKIS: Good.
MS. DROUIN: When we get to the schedule,
you're going to see places where we've noted where
we'd like to come back to the ACRS.
CHAIRMAN APOSTOLAKIS: What is the status
now regarding the 1.174, Mike? Are we going to look
at it again, or it's out now?
MR. MARKLEY: Well, George, we don't have
a document to review right now. But the last letter
we wrote was July 20, 2000, and it was a Larkins
Graham. And the committee has no additional comments
or concerns on the proposed final revisions. That's
where we were. You basically agreed with the license
amendment process and using risk information when it
was deterministic submittal if the staff felt the need
to do more in terms of risk analysis.
CHAIRMAN APOSTOLAKIS: So what we
anticipate is that we may have to write another
Larkins Graham or something?
MR. MARKLEY: If and when the document --
CHAIRMAN APOSTOLAKIS: If and when the
document. Okay. Thank you.
Okay.
MS. DROUIN: Okay. In the reg guide, the
main body -- because we are proposing appendices with
this reg guide. But the main body is going to provide
the guidance to the licensees on how to use the
standards or the NEI-00-02 document, such that we
would have the level of confidence of the results for
that particular application. So it's going to go
through a lot of the discussion that ya'll just had in
trying to answer those kind of questions.
But given that it's telling you how to use
the standard -- and still at some point we need to say
whether we agree technically with the standard or with
the NEI-00-02. If, for example, the standard came out
and said on their initiating event analysis only look
at transients for your initiators, well, we would
disagree with that. You can't just look at
transients; you've got to consider all your
initiators.
So that would be an example of something
that would go into the appendix. And that's where we
would provide clarification or additional criteria so
that we think that the document itself is technically
sound.
So our envision is that -- for example,
Appendix A would address the ASME standard; Appendix
B, NEI-00-02. And as you know, ASME is not covering
the full scope. So as each of the different pieces of
the standards would come out, those would be added on
into another appendix.
MEMBER ROSEN: Now, Mary, I'm sure you
know of the OMB circular that talks about the desire
of the federal government to use industry consensus
standards.
CHAIRMAN APOSTOLAKIS: Yes.
MEMBER ROSEN: Now, here you're talking
about a process to not use industry-consensus
standards, to go beyond them.
CHAIRMAN APOSTOLAKIS: No. The law is
that industry-consensus standards should be used as
appropriate.
MS. DROUIN: As appropriate.
CHAIRMAN APOSTOLAKIS: And she's
elaborating on the "as appropriate."
MEMBER ROSEN: Yeah. And I'm elaborating
on that by saying that, certainly, the NRC staff's
participation was extensive and welcome, most of the
time. Even when it wasn't, it was still given, and
used, and ultimately reconciled. And I will look very
hard at anything that's in these appendices beyond
what's in the standards.
Now, I'm talking about the standards, the
industry-consensus standards, not NEI-00-02. That's
in a different class. But anything beyond the
standards -- where the staff feels it absolutely must
take exception to or elaborate of the standards -- we
really ought to be given a very hard look by NRC
management.
MS. DROUIN: And it is. You go back to
the letter on Reference 14A. You might have seen
Scott Newberry's signature, but trust me. A lot of
people beyond Mr. Newberry reviewed that letter.
Because we were transmitting here's where our concerns
are on this final version. So this is taken very
seriously in terms of if we take exception.
VICE CHAIRMAN BONACA: And it seems to me
it is a little bit of a problem, because if I remember
some of the early applications we saw with 1.174, some
of the SERs were going at length to explain why
something in the application was deficient, but you
could endorse it because of consideration of this and
that. It seems as if the burden was on the staff to
provide the reasons why the risk information coming
from the application was justifiable; it was
sufficient. So I think it would be good to at least
set the minimum standard of what you expect so that
the reviewers don't have to -- are not forced to have
that kind of degree of individual support of the
application when they come through.
I understand your concern, Steve. And I'm
only saying, some of the early applications to me were
not very satisfactory, the SERs.
MEMBER ROSEN: Well, we're all learning.
But I think the best possible result would be to have
a standard that the NRC was able to endorse without
qualification. And we may not get there, but we
really need to minimize, if we can, somehow the degree
to which the staff needs to put on additional
requirements beyond the industry-consensus standard.
CHAIRMAN APOSTOLAKIS: Actually, I think
this someone had mentioned earlier. But it's not
clear to me, really, what the word "endorsement" means
in this case. And I don't know what you guys are
going to do if in the future somebody comes before you
and says, I have done what the standard says. You
have endorsed it. I expect an answer from you in a
week. You don't have to review it. What are you
going to do? Are you going to say no?
MEMBER POWERS: That's a good answer.
CHAIRMAN APOSTOLAKIS: No?
MEMBER POWERS: The answer is no.
CHAIRMAN APOSTOLAKIS: So why all this
fight, then, about endorsement? I mean, if that's the
case. It's really internal to the industry document
telling them, look, this is what the staff expects of
you. But you prefer to go through a review if they
choose to do that.
MEMBER POWERS: This is an acceptable
method.
CHAIRMAN APOSTOLAKIS: But that's what I'm
saying. If they tell you I met all your requirements,
therefore, it's acceptable. Don't review it.
MEMBER POWERS: Well, you've got to find
out if they, in fact, carried out the method properly.
MEMBER SIEBER: Or used some other ones.
CHAIRMAN APOSTOLAKIS: Then you come back
and you say, look, for common-cause failures, you
really did this and should have done something else.
The standard doesn't say anything about that. The
standard says, do a common-cause failure analysis.
Isn't that what it says?
MS. DROUIN: Well, it says a little bit
more than that.
CHAIRMAN APOSTOLAKIS: It goes beyond
that?
MR. DUDLEY: Just as a matter of public
record, in case someone wants to look up what the
circular says, it's OMB Circular, A-119. But what
Mary's describing is really somewhat consistent with
how we endorse industrial standards and have for year.
If they're exceptions and clarifications, we usually
note it in the body of the supporting document.
MEMBER POWERS: And Steve is going to
scrutinize that, for these exceptions and
clarifications.
CHAIRMAN APOSTOLAKIS: Anyway, we'll have
to wait and see the standard itself.
MS. DROUIN: I mean, at this point in
time, whether or not we have exceptions, I can't
answer that. We're in the midst of reviewing the
final version.
All we are trying to show on this figure
is looking at it and trying to show the relationship
of all these different documents and activities that
we're talking about. If you start there at the top
level where you have some regulatory activity -- this
is not meant to be complete; this is just an
illustration with some examples shown there. But if
you have some examples -- like licensing actions,
categorization, Option 3. Those were the three
examples I picked. And if you come there across the
right, if I look at Option 3, 5046 right now, we're in
the midst of risk-informing the ECCS reliability
requirements. Underneath that you'll see inputting
into it, there will be some regulatory guide that will
go with that rule.
If you look at your licensing action,
somebody does something where they want to come and
make a change to their licensing basis, there's a
regulatory guide supporting that, Reg Guide 1.174. On
your tech specs, say they're going to do a
risk-informed techs spec change, there's a regulatory
guide.
So the point is, is that you got these
things here, but also in the decision-making process
for all of this is that they're using PRA information
as one of the inputs. And so that's where we talked
about there's this common element for all of these.
And that's why in our mind, to endorse the standards
in this document didn't make sense, because the
questions that you're dealing with are the same kind
of questions here and here, and potentially elsewhere.
And that's why we thought coming up with a single
regulatory guide that would feed into all of these
just make practically a lot more sense to us.
And that's where we divvied it up. This
is all the same guide. This is just the appendices
that supports that. And this would provide the
guidance on how to use the standards in the Peer
Review Program. And then if we wanted to take
exception or add clarification on a particular
standard, that would be in the appendix.
CHAIRMAN APOSTOLAKIS: Mary, the box on
the left, "Determine the confidence with which PRA
results demonstrate," and so on. Now, if the PRA has
done parametric uncertainty analysis -- parameter
uncertainty analysis -- essentially, what you're
saying there is, you're asking them to think in terms
of model uncertainty. How good is the PRA in other
words? Right? That's what you're asking them to do,
but not in so many words. Is that the level of, when
we say, determine the confidence with which the PRA
results demonstrate the technical attributes have been
met?
MS. DROUIN: Were needed.
CHAIRMAN APOSTOLAKIS: Were needed. Yes,
were needed. But this is really a question of model;
isn't it? Model adequacy?
MS. DROUIN: Not necessarily. I mean, to
me, that would also be a question of data.
CHAIRMAN APOSTOLAKIS: If I do a parameter
uncertainty analysis, and I have a scarce data, I'll
have a wide distribution. But I have quantified. Is
that sufficient to answer this requirement, determine
the confidence? Is this my confidence, or I have to
go beyond that and say, well gee, now they have used
an HRA model which I know is kind of soft, and maybe
I should be a little more cautious, without really
saying how cautious you're going to be?
MS. DROUIN: When we use the word
"confidence" here we aren't using this confidence in
the sense of your uncertainty analysis that would get
into your levels of confidence. That's not what that
--
CHAIRMAN APOSTOLAKIS: Well, confidence in
general means that. It means how uncertain are you,
how much do you believe, right?
MS. DROUIN: Yes.
CHAIRMAN APOSTOLAKIS: And what I'm saying
is that you're asking them there to reflect on the
whole thing. Is there a need for a structuralist
defense, in-depth approach? And I wonder why we don't
say that.
MR. PARRY: George, can I --
CHAIRMAN APOSTOLAKIS: Because that's
really what you're asking them to do.
MR. PARRY: -- can I put my two-cents
worth in here? This is Gary Parry from the staff.
I think the way to think about that box,
rather than to worry about the technical attributes is
to say, basically, it's to determine the confidence of
which the PRA results have shown compliance with the
acceptance criteria or guidelines that you're going to
use in the decision-making.
CHAIRMAN APOSTOLAKIS: It's either way.
MR. PARRY: And it's specific -- but it's
specific, though, to the PRA aspects of it.
CHAIRMAN APOSTOLAKIS: Right.
MR. PARRY: So I think you're right in
saying that what you're requiring here is actually
what we describe in Reg Guide 1.174, where we say,
show that the results that you have -- the conclusions
that you're drawing are robust with respect to the all
the sources of uncertainty that you can identify. And
that's really what that statement means.
CHAIRMAN APOSTOLAKIS: But it's model --
MR. PARRY: It's model uncertainty, and to
some extent incompleteness as well.
CHAIRMAN APOSTOLAKIS: Well, for me that's
model -- but you are asking to reflect on that aspect.
MR. PARRY: Absolutely.
CHAIRMAN APOSTOLAKIS: And what I'm saying
is that maybe it's time to start naming it, so people
will know that you're interested in that. And maybe
six years from now people will start doing something
about it.
MR. PARRY: I suspect --
CHAIRMAN APOSTOLAKIS: As long as you
don't name it, people will not be aware of it. That's
what you're doing.
MR. PARRY: I suspect the main body of the
reg guide will actually address this.
CHAIRMAN APOSTOLAKIS: Or in a footnote.
MR. PARRY: No. No. I think in the
model.
CHAIRMAN APOSTOLAKIS: I mean, if I do a
fire risk assessment -- I know some members here feel
that this is the perfect PRA, but even there are
weaknesses. Right? Dana, you agree?
MEMBER POWERS: There are weaknesses in
fire risk assessments. Sure.
CHAIRMAN APOSTOLAKIS: So they're doing
all these funny things. The domain of influence of
fire down there is a cone with a 35-degree angle and
all that. Now, nature doesn't work that way. It
doesn't say I'm going to burn things that are inside
the cone but not the ones that are outside. That's an
approximation. That's a model approximation, right?
Now, if you ask me to quantify, I don't
think anybody knows how to quantify that. But it
shakes my confidence in the results. Okay? So you're
asking me to think about it.
MR. PARRY: Right.
CHAIRMAN APOSTOLAKIS: So that's a model
uncertainty. So please name it so people will know
that you worry about it.
MEMBER KRESS: Now, which word do you want
to you use? "Model"?
CHAIRMAN APOSTOLAKIS: "Model
uncertainty." Not "modeling." "Model uncertainty."
MEMBER KRESS: You want to use "model
uncertainty."
CHAIRMAN APOSTOLAKIS: Model uncertainty.
MEMBER KRESS: Not some other name.
MEMBER ROSEN: George?
CHAIRMAN APOSTOLAKIS: No. Model is part
of our understanding.
MEMBER POWERS: Come up with a good Greek
word, would you?
CHAIRMAN APOSTOLAKIS: We have enough
problems with understanding.
MEMBER ROSEN: Beyond this, just quickly,
do you understand regulatory guides are not
regulations?
MS. DROUIN: Absolutely.
MEMBER ROSEN: And the standard paragraph
is going to go in the front of this regulatory guide.
It says, "Anyone caught using this will" --
MEMBER POWERS: Something else.
MEMBER ROSEN: Sure.
MEMBER POWERS: And as a lot of times,
money can come in --
CHAIRMAN APOSTOLAKIS: This is an
acceptable way.
MEMBER ROSEN: An acceptable way.
CHAIRMAN APOSTOLAKIS: Oh, you don't know
how many times I was told that when we were developing
1.174. It's an acceptable way.
MEMBER ROSEN: Well, this is the same --
I understand how difficult that is.
CHAIRMAN APOSTOLAKIS: It is very
difficult.
MEMBER ROSEN: But I'm thinking about --
I'm really thinking of an advanced reactive area.
Some of this may not work so well. So I don't know
how general this is.
CHAIRMAN APOSTOLAKIS: I think they have
LWRs in mind when they do this. Let's not try to be
as general as we can in everything. It's too much.
It's too hard. But this is really LWR.
MEMBER ROSEN: But my point is, that out
is what made -- the advance reactive people may need
to use them at those outs. They may not be able to
comply with a regulatory guide.
MS. DROUIN: I can tell you right now, in
terms of the ASME standards -- I certainly wasn't
involved in any discussion with ANS, and I can't speak
for NEA-00-02. But as part as the writing team on the
ASME, it was decided very clearly that this would only
apply to LWRs. That was a very deliberate decision
made up front.
MEMBER ROSEN: Okay. Then I'm right. I
guess what I'm saying is that, if the ASME standard
only applies to LWR, then this regulatory guide, which
refers to the standard, obviously, can't apply to the
advance reactive.
CHAIRMAN APOSTOLAKIS: Although the
standard itself, I don't see why you could not apply,
or most of it anyway.
MEMBER SIEBER: Does it say that
specifically in the standard?
MS. DROUIN: I know one version did.
CHAIRMAN APOSTOLAKIS: Well, the reason
why they might say it, Jack, is to avoid debate. You
know, stop thinking about future reactors; have we
covered everything. And I think they did the right
thing if they say so.
MEMBER SIEBER: If somebody was going to
buy a future reactor, and didn't know what the rules
were, though, I think that would put another bundle of
uncertainty into whether that would be a good
investment or not.
CHAIRMAN APOSTOLAKIS: I doubt that the
availability of an ASME standard to do a PRA would --
would change their purpose.
MEMBER SIEBER: Well, it's really a
philosophical question.
CHAIRMAN APOSTOLAKIS: No. But you see
why I'm saying that. They've had a lot of problems
proving this one for LWRs. Now, if you ask them to
think about future reactor applicability, they'll
never finish. And there is some urgency to this.
MEMBER SIEBER: Well --
CHAIRMAN APOSTOLAKIS: I mean, somebody
else might want to --
MEMBER SIEBER: I consider that a lot of
conceptual elements, or a conceptual design of a new
type reactor, would rely on some of the aspects of PRA
as applied to that, and whether you have to contain it
or not, for example. They can almost add 30 percent
on to the cost of the plant if you decide that that's
what you're going to have to do to get the PRA spec.
So it should be versatile enough to be
able to at least have some kind of an idea as to what
kind of regulatory space you're in when you propose a
new concept, it seems to me.
CHAIRMAN APOSTOLAKIS: I would tie that to
having a risk-informed licensing process. Think about
all that stuff there and go revisit the ASME standard,
and say, yeah, most of it applies. But I would think
that as a separate activity because of its importance.
MEMBER SIEBER: Okay. Thank you.
CHAIRMAN APOSTOLAKIS: That's what I would
-- I don't know. You want to think about future
reactors in this context?
MS. DROUIN: Well, I mean, I can just --
I can give you my personal opinion. I can also share
with you some of the decisions that went into the
standard. I mean, the decision was made that this
would not cover future reactors. One of the reasons
-- there were a lot of reasons. But one of the main
reasons is that this was not a standard to do a PRA.
This was a standard to support risk-informed
applications.
So if this were to be written as a
standard that you wanted to pick up, and you were
starting with a blank piece of paper and wanted to do
your PRA, a lot of this would have looked a lot
different. It would have --
CHAIRMAN APOSTOLAKIS: If the licensee was
to come with an application, say for the PBMR, and
they say, we did a PRA using the ASME standard. Even
if the standard says --
MEMBER SIEBER: It would require
additional review --
CHAIRMAN APOSTOLAKIS: Some additional
review.
MEMBER SIEBER: -- to assure its validity.
Exactly.
CHAIRMAN APOSTOLAKIS: But it would carry
some weight. So they did comply.
MS. DROUIN: Well, they could comply with
some of it, but some of it's just not going to be
applicable.
CHAIRMAN APOSTOLAKIS: Exactly. And they
will decide.
MS. DROUIN: Right. Okay.
Now, given that we have this new
regulatory guide, we wanted to just talk a little bit
about how does this fit in, in terms of -- because,
again, the ASME standard is just part of it. What
we're trying to show you is all the different pieces
ultimately of where we'll be down the road.
As you can see, the ASME was only covering
Level 1 and only part of a Level 2, full-power,
internal events, excluding fire.
MEMBER KRESS: Is any thought given to
adding a late-containment failure criteria in 1.174?
You know, we had that big debate over whether late
containment is a regulatory objective that ought to be
part of the licensing changes.
MS. DROUIN: Well, we're certainly going
to deal with some aspects of late-containment failure,
because -- we did have a meeting with NEI. And they
came in and was showing their interpretation of some
stuff out of 1150, showing that late-containment
failure was less than zero to your LERF.
MEMBER KRESS: Of course. But the point
is, you don't want late-containment failures either.
And it seems to me like there ought to be some
guidance in 1.174 related to late-containment
failures. If somebody comes in with a licensing
change request that significantly affects that part of
their protection, it seems to me that you guys would
think that was important.
MS. DROUIN: Absolutely. I don't
disagree. I think this is an issue --
MEMBER KRESS: Yeah, well, this seems to
me like something that ought to be discussed in 1.174
somewhere.
MR. PARRY: It's sort of there. If you
look closely, it's there in the concept, difference in
depth, from maintaining the containment function. And
that's the approach that's being used. And that's the
approach I believe is going to be incorporated in
NEI-00-04 to address things to like containment
systems.
MEMBER KRESS: Yeah. I would have rather
seen some -- we've got something called LERF, large
early. I would like to see something called
late-containment failure or LCF, and have a guideline
on it that's numerical and deltas on that. Another
one of your created darkness charts.
MEMBER ROSEN: Now, what distinguished
early from late is late means after the implementation
of moving people out of the way.
MEMBER KRESS: Sure.
CHAIRMAN APOSTOLAKIS: Basically.
MEMBER ROSEN: Basically. So when you
crack to the adequate-protection standard, protection
of the public's health and safety, the public are
already gone. So now you've got a whole different set
of issues.
MEMBER KRESS: You've got two safety
goals. One of them deals with prompt fatalities and
one of them deals with latent fatalities. This is a
latent fatality issue. It's also a land contamination
issue. And in NRC has that within their regulatory
objectives.
MEMBER ROSEN: I agree with you. With
respect to the land contamination issue, clearly we're
on the same page.
MEMBER KRESS: Yeah. And there ought to
be some criteria in 1.174 that deals with that.
MEMBER ROSEN: We're not arguing. All I'm
saying is that there's a whole different set of
considerations in order to develop --
MEMBER KRESS: Oh, yeah. You've got to
have new guidelines, new deltas, new metrics.
MEMBER ROSEN: New metrics, new
guidelines, thresholds, et cetera.
MEMBER KRESS: But it seems to me like
1.174 is where they belong. It deals with changes to
the licensee basis. And these are things you're
interested in preserving, or preserving adequately.
And it's always bothered me that there's nothing much
in there with respect to that, other than words like
"you will also meet all the other regulations," and
things like that. You can construe that to have some
meaning. It seems to me that it ought to be in there
explicitly. But that's just my opinion.
MEMBER SHACK: It's included more
explicitly in the framework document.
MEMBER KRESS: Oh yes.
MEMBER SHACK: I don't see why you
couldn't bring that right into --
MEMBER KRESS: Yeah, the framework
document definitely has it. And that would be a good
place to look to see how maybe to incorporate it in
here.
CHAIRMAN APOSTOLAKIS: And there's no need
for a standard for Level 3 PRA because it's done so
well, right?
Okay. Shall we go on, Mary? Twelve?
MS. DROUIN: This one just shows right now
what the status of these documents are. ANSI has
voted for the standard for ASME. So that should be
coming out imminently in the next couple -- they've
got to go through a check-editing process, but it's on
the way out the door is the point. They're on
schedule.
NEI-00-02. We met with them because our
initial review, they had asked us to review it. But
in the context of Option 2 applications only, we
received a letter in December from NEI asking us to
expand our review to all applications. We are in the
midst of doing that. We plan to have preliminary
comments back to them in March and final comments in
May. But that's still a preliminary schedule.
ANS. Their schedule right now is to have
a fire published standard by the end of December of
this year for low-power shut down. That's a year
away. In terms of the fire, they're only right now
putting together their writing team for the standard.
So there's no schedule right now that's established
for the fire part of it.
MEMBER ROSEN: Mary, could I ask a
question of our staff? I don't remember seeing
NEI-00-02. Has it been distributed to the ACRS?
MR. MARKLEY: We reviewed it about a
year-and-a-half ago.
MEMBER KRESS: It wasn't 02 then, was it?
MR. MARKLEY: Yes, it was.
MEMBER KRESS: Was it?
MR. MARKLEY: Yeah.
MEMBER KRESS: I thought it was 99-02.
MEMBER ROSEN: I guess I wasn't in the
loop at that time. Would you see about getting me a
copy?
MEMBER POWERS: You may be right, Tom.
But it's a 0002. That I know. And all members of the
Fire Protection subcommittee have several copies of
it.
CHAIRMAN APOSTOLAKIS: And the one that
applies to Option 2 is 0004.
MEMBER ROSEN: Right.
CHAIRMAN APOSTOLAKIS: What happened to 3?
MEMBER ROSEN: They only do even numbers.
MEMBER POWERS: Let me ask you a question,
Mary. You're a knowledgeable student of
risk-assessment culture in this country.
Have there been enough prior PRAs done?
Has there been enough debate within the academic
community to write something like a standard?
MS. DROUIN: Personally, I think there is.
Is it going to be the ultimate standard? No. But I
don't think you have to have every "i" dotted and "t"
crossed before you can start writing a standard, in my
opinion.
CHAIRMAN APOSTOLAKIS: I guess the problem
is that we don't even know where the i's are; let
alone dot them.
MEMBER POWERS: It seems to me that -- I
think I'm not in an unusual situation of planting
myself in total agreement with the chairman here, that
whereas in PRA for normal operational events, lots of
people do them. And there are big conferences held
often in lovely hotels. One of the best, of course,
is in Beverly Hills, I believe, and set a new standard
by which such conferences were held in which people
debate various aspects --
CHAIRMAN APOSTOLAKIS: That was a Beverly
Hills standard.
MEMBER POWERS: Beverly Hills standard,
right. And it's been through a crucible of technical
discussion of some magnitude. Whereas, in the area of
fire PRA, I find, first of all, less, and that that
debate has largely been in the hands of those with a
probablistic bent, and that my friends and
acquaintances in the fire protection community have
not weighed in to this discussion, whether
phenomenologically things are about right; whether
this cone that Dr. Apostolakis, should be 35 or 37
degrees.
That kind of interchange -- I question
whether that kind of technical debate, discussion,
what not, has taken place to the point that you could
write a standard with any confidence, that it would
put useful constraints and guidance on the writers of
a fire PRA.
MS. DROUIN: Well, I think when you write
any standard, I mean, I think it's obvious you've got
to have the right people in there to get a decent
standard. And you want to make sure that you get a
good cross-cut of all the different aspects with the
different knowledge and the different variance and
expertise involved in the process.
MEMBER POWERS: So will this writing group
have probabilists or will it have fire protection
people, or will it have it both?
MS. DROUIN: I hope it would have both, in
my opinion.
CHAIRMAN APOSTOLAKIS: I think it should
have also people who understand fire phenomena. I
mean, there is a whole community out there. I don't
think they call themselves fire protection people.
MEMBER POWERS: That's right.
CHAIRMAN APOSTOLAKIS: It's fire safety
scientists or engineers, familiar with all the
experiments that have been done, the correlations that
have been developed. These are extremely important
people to be represented in an effort like this.
MS. DROUIN: I don't disagree. I have no
say so who's on this writing team.
CHAIRMAN APOSTOLAKIS: We understand.
MEMBER ROSEN: I do.
CHAIRMAN APOSTOLAKIS: We understand.
MEMBER ROSEN: I'm on the Steering
Committee.
CHAIRMAN APOSTOLAKIS: Oh, okay.
MEMBER ROSEN: So I'm listening.
MS. DROUIN: But I think it's critical
that you get all these --
CHAIRMAN APOSTOLAKIS: I mean, it would be
a big mistake if you have only the PRA types and fire
protection types.
MEMBER ROSEN: Well, we're just getting
started. I'm speaking now for the ANS-RISC Committee,
Risk-Informed Standards Committee.
CHAIRMAN APOSTOLAKIS: Oh, just like the
Option 2 risk?
MEMBER ROSEN: Yeah. It's called the
RISC, which I'm on.
CHAIRMAN APOSTOLAKIS: So you were
categorized.
MEMBER ROSEN: And what we're trying to do
is put together a team now to write this. So it's
helpful to have these thoughts.
MEMBER KRESS: I would spell it R-I-S-K.
MEMBER ROSEN: It's a committee, and you
don't spell committee with a "K."
CHAIRMAN APOSTOLAKIS: Ms. Drouin, let's
move on.
MS. DROUIN: Okay. We're very close to
being done. I just have one more slide after this
one.
Some other related efforts that are
ongoing, but they're a part of all of this. You've
probably heard about it. I know you haven't seen
anything. But we have been working on what we call a
data handbook that is to compliment the ASME, the
standards that are ongoing. And this handbook is
getting more into -- it's a methods book, covering the
sources and methods for doing your data, covering
initiating events, components, failure rates, and
equipment, non-recovery probabilities. The thing I
want to point out is it's not covering common cause,
because there is a separate document that the NRC has
produced. So we aren't trying to replicate work
there.
MEMBER KRESS: I think it's a great idea,
by the way.
MS. DROUIN: Thank you.
The date there, though, for July. Sorry,
that's a typo. That should have been August.
Another effort that's also ongoing to
support the standards effort, is when you go and you
look at Reg 1.174, also the standard. The ASME
standard refers to NUREG 6595 as an acceptable
approach for LERF. It's the document that gets them
-- how to do a simplified LERF. But it only covers
full power. We're in the midst of updating that to
also cover low-power shut-down conditions to support
the standards effort. And we plan for that to go out
in August of this year.
MEMBER KRESS: Can we get it before it
goes for public review?
MS. DROUIN: You will receive it when it
goes out for public review and comment. Yes.
MEMBER KRESS: At the same time? Yeah.
Okay.
MS. DROUIN: So in wrapping up, this is a
tentative, preliminary schedule we've put together.
We did have a public meeting on Tuesday. In fact, the
schedule -- because of information we got from the
public meeting, we revised the schedule as you see it
today. It might go through some more revisions.
That's why we're just saying it's preliminary. We've
met with ACRS today.
But right now what we're thinking about is
that in March we will have a SECY that goes forward to
the Commission. Now, that date, though, is firm in
terms of the SECY for transmitting the updated Reg
Guide 1.174. That will go forward. Also attached to
that SECY would be the plan for our endorsement of the
standards in the industry peer review, which we've
talked about today.
We also were thinking at that time -- we
have been reviewing NEI-00-02 for all applications in
submitting preliminary comments to NEI at that time.
One of the reasons that the comments wouldn't be final
is one of the things we did learn at the public
meeting is that NEI is working on a document that
would show the differences between NEI-00-02 and the
standard. And we felt having that information, and in
terms of our review of it, would be very important.
So we were changing the schedule to try and integrate
that in there.
So we were saying preliminary comments in
March, and then going final comments to NEI-00-02 in
May. Give them time to come back to us with their
response. And then in July we would go forward to
release the draft reg guide, plus the appendices. So
we would have two appendices at that time-- Appendix
A which would address the ASME standard, and Appendix
B which would address NEI-00-02, and, of course, our
SRP.
At the same time, we would go out with our
draft version of the data handbook and a draft of
NUREG-CR-6595, Rev. 1.
MEMBER KRESS: What's the status of that?
Has it been started?
MS. DROUIN: The status of what?
MEMBER KRESS: The NUREG CR-6595, Rev. 1.
MS. DROUIN: We started it about six
months ago. So it's been in progress.
MEMBER KRESS: Is there some sort of a
rough draft available?
MS. DROUIN: Not at this point, not enough
to --
MEMBER KRESS: Would there be one
available by, say, April? I would really love to see
that, even though it's draft and rough and not been
approved, et cetera. If I could get one. But it
would have to be in April. It would be too late to
get it after that. I have a reason for wanting to
look at it.
MS. DROUIN: I don't know what our
position is. I mean, in terms of coming, to have it
to a form to show the ACRS, I hesitate to say --
MEMBER KRESS: No, I don't want it for --
this is not for the ACRS; this is for me. Just for
personal use. For my edification.
MS. DROUIN: Well, we can talk off line on
that.
MS. DROUIN: He promises not to share with
us.
MEMBER KRESS: Yeah. I promise not to
even let them know what it is.
MS. DROUIN: Okay.
MEMBER ROSEN: Mary, what is the title of
NEI-00-02?
CHAIRMAN APOSTOLAKIS: It's the peer
review document; isn't it?
MEMBER KRESS: Yes, the peer review --
CHAIRMAN APOSTOLAKIS: The peer review
process.
MS. DROUIN: It's a long title. You know,
Steve, I apologize. I just don't remember.
MEMBER ROSEN: But it's a --
CHAIRMAN APOSTOLAKIS: Something like
that. Industry guidelines for peer review.
MEMBER ROSEN: It's the one that lays out
the peer review process.
MS. DROUIN: Yeah. It's the Peer Review
Certification Program.
CHAIRMAN APOSTOLAKIS: And up until now,
it has been the only document that addresses the issue
of quality PRA, right? No, ASME had not been
approved. The only document out there -- because I
know the foreign organizations are using N-00-02 to do
peer reviews, because that's the only thing that's
available.
MEMBER ROSEN: Right. But the substance
of it, before it was called NEI-00-02, was part of the
BWR Owners Group.
MS. DROUIN: Is the BWR Owners Group.
CHAIRMAN APOSTOLAKIS: That's right.
Well, that was a long time ago.
MS. DROUIN: They're essentially the same.
Okay. In August we were planning on
coming back to the ACRS --
MEMBER POWERS: You'll come alone.
CHAIRMAN APOSTOLAKIS: Well, we aren't
really meeting in August.
MEMBER KRESS: We won't be here.
CHAIRMAN APOSTOLAKIS: You can come.
MEMBER POWERS: It's a terrific meeting.
You'll get no objections, and your presentation will
go very quick.
MS. DROUIN: Well, then, I love that idea.
I know when he had talked, the schedule
looked differently then.
MEMBER ROSEN: Yeah, that's June.
CHAIRMAN APOSTOLAKIS: June?
MR. MARKLEY: Yes, she's changed it to
August.
CHAIRMAN APOSTOLAKIS: Now when you say
ACRS, you mean the full committee, right? You don't
mean the subcommittee?
MEMBER KRESS: We could have a
subcommittee.
CHAIRMAN APOSTOLAKIS: PRA subcommittee?
MS. DROUIN: No. I think in August we
were talking subcommittee.
CHAIRMAN APOSTOLAKIS: Subcommittee?
MEMBER KRESS: Oh. Well, that's possible
then.
MR. MARKLEY: She's sliding the June
subcommittee back to August.
MS. DROUIN: Yes.
CHAIRMAN APOSTOLAKIS: We have
subcommittee, if that's what you'd like. We're trying
to avoid it.
MEMBER KRESS: We're trying to avoid it,
but we do.
MEMBER POWERS: We only have subcommittee
meetings in August when we know that Professor
Apostolakis is in Greece.
CHAIRMAN APOSTOLAKIS: Yes.
MEMBER ROSEN: But he's not. He's in
Crete.
MR. MARKLEY: Would late July do okay?
MS. DROUIN: We'll get back with you.
We'll work something out.
Anyway, in our October -- sorry. Back to
August.
CHAIRMAN APOSTOLAKIS: Yes?
MS. DROUIN: We're also planning very
quickly after releasing these documents for public
review, is having a workshop to walk people through
the documents. This is a lot of information we're
going to be putting out. If it was just the reg
guide, that would be -- but we have the reg guide.
Just to show you. Here's the data handbook; it's a
thick document. You know, we're going to have NUREG
CR-6595. So a lot of information. So we really
thought having a workshop and walking everyone through
these documents was going to be very important.
MEMBER POWERS: Why don't we instead of
having the subcommittee meeting in August just go to
their workshop?
CHAIRMAN APOSTOLAKIS: That's an idea.
MS. DROUIN: That's --
MEMBER ROSEN: It probably be more useful.
CHAIRMAN APOSTOLAKIS: It also increases
the presence of ACRS members to meetings.
Is the workshop going to be in Washington?
MS. DROUIN: No decisions have been made,
but I would like to think so.
MEMBER POWERS: Or Las Vegas, someplace
like that.
MEMBER ROSEN: In August?
MEMBER POWERS: Yeah.
MEMBER ROSEN: 412 degrees?
MEMBER POWERS: It's 112. What's wrong
with that? Relatively cool.
CHAIRMAN APOSTOLAKIS: There's a
suggestion by your vice chairman here.
VICE CHAIRMAN BONACA: What about Crete?
MS. DROUIN: Anyway, in August, we're
talking about a 60-day public review and comment
period. And then hopefully come October, we would
have all the comments back and hopefully time to go
through them and understand them, and come to a
resolution, and have another public meeting at that
point to share what our position is on the comments.
And then come December, have a SECY that would
transmit this to the Commission for release.
CHAIRMAN APOSTOLAKIS: Very good.
MEMBER KRESS: Very good.
MS. DROUIN: So that's --
CHAIRMAN APOSTOLAKIS: Any questions for
Ms. Drouin?
Well, thank you very much, Mary.
MEMBER KRESS: Thank you.
MS. DROUIN: Thank you very much.
CHAIRMAN APOSTOLAKIS: We'll see you in
August.
MS. DROUIN: Right here.
CHAIRMAN APOSTOLAKIS: We'll be back at
3:10.
(Whereupon, the foregoing matter went off
the record at 2:53 p.m.)
CHAIRMAN APOSTOLAKIS: Okay, we're back in
session.
The next item on the agenda is the PTS
Technical Bases Reevaluation Project. Dr. Ford, would
you be kind enough to lead us through this?
MEMBER FORD: George, this is a good
engineering project. It was --
MEMBER POWERS: How can it be a good
engineering project, when it's got epistemic and
aleatory --
(Laughter.)
MEMBER POWERS: -- ain't no fire.
MEMBER FORD: The last time we heard --
the full committee heard about this project was about
16 months ago. Last month, the materials subcommittee
had a full 1-1/2 days briefing on the progress on
this, and you will find it very exciting.
However, we only got two hours, and
therefore we're going to restrict comments to
technical basis for changing the frequency of through-
wall quick cracking. Discussions on things such as
acceptance criteria, that will be covered in a meeting
in May. So --
MEMBER POWERS: What is that --
MEMBER FORD: -- I'll pass it on to Ed at
this point, Ed Hackett.
CHAIRMAN APOSTOLAKIS: I would like to --
I was -- oh.
MEMBER POWERS: You said, what part was
going to be deferred to --
MEMBER FORD: The discussions on, for
instance, changing acceptance criteria due to
containment failure, source terms, etc. That will be
put off until May.
CHAIRMAN APOSTOLAKIS: And I would like to
see a recent example of model uncertainty
quantification in an HRA. I looked at the viewgraphs
from the subcommittee, and I'm not sure I found the
answer.
MR. HACKETT: Why don't I go ahead and
read off a few things while we're working through
this.
First off, just to introduce the table,
I'm Ed Hackett, and I'm Assistant Chief of the
Materials Branch in the Office of Research. To my
right is Alan Kolaczkowski, and he's an expert in
probabilistic risk assessment from Sandia National
Laboratories.
To my left immediately is Mark Kirk, who's
also in the Materials Engineering Branch. And his
expertise is in probabilistic fracture mechanics.
And also, Dave Bissett from the Simsab
Branch in Research is with us also, and his expertise
is thermal hydraulics. So that's trying to represent
-- you know, get the key disciplines aligned here for
you so we can go through this.
It looks like we're close on slides here.
In addition to that, I can reference that
there's -- these are the front people, particularly
speaking for myself, for what's a fairly large team of
people at NRC in general, including NRR, including our
contractors, and also including the industry.
This has been kind of a first for us to do
this on as participatory a basis as we have with the
industry, and I think that's something that's really
worked very well for us.
The other thing I just mentioned that's
been very key to this has been the backing. We've
had, especially from our senior management in the
Office of Research, particularly my division director
Mike Mayfield, who's taken this on with a sort of
personal vengeance -- those of you know Mike, you know
what that means it brings to a project. And also, a
shift tonight. So, we've had a lot of support, is the
bottom line.
I guess we'll go ahead and go to the next
slide.
I really wanted to do three things for you
today, and maybe I'll start by addressing a comment
George made in the opening. This is a very condensed
version of a presentation or series of presentations
that we did about a month ago for the subcommittee,
where we tried to walk through the uncertainty
treatment in the project that I believe was almost 180
slides. So, you'll see a much condensed version of
that today.
I don't know if we'll get to exactly what
Professor Apostolakis was looking for. If we don't,
we'll try and cover that some way. But we'll make a
good stab at that, and Alan especially will be here to
address the human reliability aspects.
But, just a couple of quick things because
I don't want to belabor, you know, my opening here.
The overall project status -- we started this, as some
of you remember, in 1999. There have been a lot of
challenges along the way, both technical and
otherwise, and still continue to be. Also, schedule
challenges.
I mentioned the industry participation
that's been key to this thing. We expect right now
that the tech basis will complete this year in 2002,
and we will do sort of a parallel handoff with NRR to
initiate rulemaking activities. So hopefully later
this year is -- that's part of the plan right now.
The second bullet there is, we're going to
be, in this presentation, giving you results and
insights from analysis of Oconee Unit 1, which was the
first of four plants that are being evaluated in the
process. I guess I'll emphasize up front that these
results are preliminary. They also look, in our
opinion, pretty good at this point, but we'll get to
that in a minute. But they are preliminary. We're
still working through finalizing that.
And then, a key part of this is describing
the modeling process and the uncertainty
quantification, particularly in the three areas of
PRA, thermal hydraulics and probabilistic fracture
mechanics. I appreciate the help we got previously
with the subcommittee. There was some good dialogue
there and some good comments, and it probably glazed
over the members by the end of a day and a half worth
of PTS.
One of the things we talked about in the
way of some business aspect -- with Dr. Ford, we
talked about the last time that we would probably be
looking for a letter from the committee this time
around. If nothing else -- it's been a long time. I
think Dr. Ford, when he led off the subcommittee
meeting, had indicated that the last time there had
been formal correspondence I think was about a year or
a year and a half.
MEMBER FORD: October 20, 2000.
MR. HACKETT: Okay. So probably at this
point, we're -- it would be appreciated to have a
letter from the committee.
I sat in this morning when you guys were
grilling Shook and Sam and company. And Shook said you
guys were up to speed on this, so I'm hoping that's
the case.
We'll go to the next slide, which is --
what I wanted to do here -- we had a day and a half
the last time with 180 slides, so we came up with our
emergency summary slide, just in case. It's not quite
an emergency here yet today.
MEMBER ROSEN: I hope you change the
title.
MR. BESSETTE: We did.
MR. HACKETT: Oh, you did. Well - never
mind. I still had "emergency summary slide".
So, also we translated to try and give it
to you guys up front, where this is heading in plain
language space. The transients are obviously --
MEMBER WALLIS: That's a wonderful
development -- plain language space. Please do it
some more.
MEMBER POWERS: But Graham, if you look,
the end sentences were prepositions -- God help us.
This is all --
MEMBER WALLIS: This is all one sentence.
It's all one sentence.
MEMBER POWERS: If this is plain language,
we're in trouble.
MEMBER WALLIS: That's what plain language
is. Plain language does end sentences with a
preposition.
MR. HACKETT: You should have seen the
other slide.
MEMBER WALLIS: I think it's the only
thing I can listen to.
(Laughter.)
MEMBER KRESS: I think it's written in
Tennessee.
(Laughter.)
MR. HACKETT: But we do have Terry Dixon
on the project.
MEMBER KRESS: Oh, that's where it came
from.
MR. HACKETT: I don't know if we have
Terry on the phone today; probably not. But Terry's
been another key member of this project at the
Oakridge laboratory. But I guess these kind of speak
for themselves. I won't necessarily go over those
point by point.
The last point, though, on the cracks --
our colleagues in the industry wanted me to make sure
to say we are actually seeing more cracks as a result
of -- or, I should say defects by virtue of the
analyses that we've done, but they are more benign and
not as large as we'd assumed before with the Marshall
distribution. You'll hear some more about that.
The four bullets that you see there are
fairly consistent with what we figured would be the
expectations on this project when we went into it, so
we're happy to see the vector's in the right
direction.
MEMBER POWERS: I haven't got a clue what
the second one means. Round operators --
MEMBER WALLIS: It's all one big sentence
with commas in it.
MR. HACKETT: It is all kind of -- I'll
apologize for it. The English is not precise, I
suppose. But it's --
MEMBER WALLIS: That's because it's plain.
(Laughter.)
MR. HACKETT: Because it's plain. It's
colloquial and plain.
MEMBER POWERS: Well, it's confusing. A
round operator --
MR. HACKETT: We were paraphrasing from --
we were paraphrasing from a summary that Bob Hardees,
our industry colleague, did to the Water Reactor
Safety Meeting about two years ago. Really, it's in
the presence or having operators that performed better
than we gave them credit for because we didn't give
them any credit previously.
MEMBER ROSEN: In plants with operators,
though.
MR. HACKETT: There are a lot of ways to
make the English better, yes.
CHAIRMAN APOSTOLAKIS: Can we move on to
the second or third slide. This one creates more
discussion than it deserves.
(Laughter.)
MR. HACKETT: Let's go to "Project
Status". This is the one you'll see replicated
throughout the presentation. The approach -- and this
has been discussed with the committee several times
now. The approach is really broken into the three
primary technical areas: PRA and event sequence
analysis, thermal hydraulics and probabilistic
fracture mechanics.
What we've tried to capture down in the
matrix there is to show you where we are, and we're
about done with Oconee. It's -- in all the boxes,
it's listed as "draft", but that should be about to go
final this month. The other ones are pretty much
where you see them. There's work beginning in all the
other areas for the other plants. And again, the
schedule challenges are to have a lot of that done by
basically the middle of this year --
MEMBER ROSEN: I'm going to interrupt you.
I have read all this and all that. And the main
question I come up with is, what are you doing here?
Where are you going with all of this?
MR. HACKETT: We'll get to that today.
And I guess, depending on the information that's
you've had, where you are going -- to cut to the chase
is a revised 10 C.F.R. 50.61 that will be less
restrictive than it is now, hopefully, is where the
thing's heading at the moment.
I might as well jump to the next slide.
That said, there's plenty of work
remaining to get there -- to Dr. Rosen's question.
One of the things that Dr. Ford focused on the last
time in terms of work remaining is QA, quality
assurance, for the project. We have work going on in
QA, in all three key areas, and there were some
comments from the subcommittee meeting on that.
The internal events analyses are really
the thing that's going to create a lot of the schedule
challenges for us in getting those done on time, and
that's for the remaining plants -- primarily,
Palisades, Beaver Valley and Calvert Cliffs.
Another issue came up -- and we'll get
into this to some greater degree, too. Because the
initial risk --
MEMBER WALLIS: Remind me -- these are
only CE plants, is that it?
MR. HACKETT: They're not. Actually, they
represent all the --
MEMBER WALLIS: So, why just these plants?
Why not all plants?
MR. HACKETT: A couple of reasons, and you
get to one of the key questions in terms of
generalizing this thing. We attempted to replicate
the work that was done before in the 1980s on the --
MEMBER WALLIS: Because these are typical
plants for covering all the major categories in the
plants.
MR. HACKETT: Covering the major and
triple-S vendors -- BNW Combustion, Westinghouse --
MEMBER WALLIS: And if these are okay, you
decide the others are okay, is that the idea?
MR. HACKETT: We're going to try and make
that argument. And that's not going to be easy,
either. But that's -- resource-wise, that's going to
be one of our constraints.
MEMBER FORD: But Ed, you might also point
out that two of them, Oconee and Calvert Cliffs, are,
according to the (inaudible) regulations, fairly safe.
MR. HACKETT: They are.
MEMBER FORD: Whereas, Palisades and
Beaver Valley are not.
MR. HACKETT: Palisades and Beaver Valley
-- just to get into that, by the current criteria in
10 C.F.R. 50.61, Palisades and Beaver Valley are okay
to the end of their current licenses, but they're just
right there at the end of those licenses. In those
cases, it's 2011 and 2016, if I remember right, so we
do have some time here before anyone is impacted, even
in the current license period.
With the extended license period, we could
get into probably a bit of debate or discussion over
which plants may or may not be impacted. But the
bottom line is, we have some time to deal with this.
MEMBER KRESS: The previous study just had
Oconee and Calvert Cliffs and Monticello?
MR. HACKETT: The previous study had
Oconee, Calvert Cliffs and H. B. Robinson.
Unfortunately, we were not able to get -- the Robinson
plant had some difficulties with participating, so we
picked up Beaver Valley instead.
The external risk piece is an interesting
piece for us. We are just getting to addressing that
now because we have -- right now with Oconee, one of
the things you'll see is, we think we're a couple of
orders magnitude down in the overall risk from where
we thought we were, which is a good outcome. But that
also necessitates that we consider the impact of
external events on the risk contribution.
The good news is, we have a funded task
underway with Alan at Sandia that we're hoping is
going to have some results probably this spring
timeframe somewhere. So, that work will be under way.
The last piece is the integration of this
whole thing with regard to the risk criteria. And
there, we also have -- we're not prepared to discuss
that in detail today, but we have some work in
progress that's to result in a SECY paper, again,
about the March, April time frame that's going to lay
out the staff views on the risk criterion for the --
MEMBER WALLIS: Is this going to lead to
a reg. guide or something?
MR. HACKETT: There are two pieces.
There's -- the main piece, of course, would be a
revision to the rule itself, to 10 C.F.R. 50.61.
There's an associated regulatory guide, which is
1.154, which takes you into the plant-specific case.
Right now, if it goes the way we're seeing
things are going right now, we may end up with a
situation where it might not be worth the resources to
re-do the reg. guide. If the screening criteria is
relaxed to a great enough degree, it might be that
nobody would ever engage with that reg. guide. They
haven't yet either, so I think that's one of the
things we're going to have to look at in resource
space.
MEMBER KRESS: Let me ask you about the
second bullet; I wasn't at the second subcommittee
meeting. We have these plants that are being
irradiated, the vessels, and going on with time. And
they have these transients occasionally, but not very
often, and thermal shocks, for example. And so, your
answer (inaudible) was, the vessel's going to survive
at a certain level of risk over its lifetime. It
would seem to me like external events don't enter into
that picture.
MR. HACKETT: We're hoping it's a very
small contribution. I think --
MEMBER KRESS: Can't you almost rule it
out before you start by looking at frequencies?
MR. HACKETT: -- I'll defer to Alan on
that one.
MR. KOLACZKOWSKI: I think it's a little
difficult to do, the reason being that external
events, as we all know, can have common cause types of
effects, wherein in the internal -- as a 'for
instance' -- where maybe you're worried about a
turbine bypass valve staying open and therefore
starting to cool the plant, and then the operator does
or does not throttle, etc., they're pretty much
independent actions.
But you could imagine -- take a fire, for
instance causing hutch shorts in such a way that not
only does it force the TBV to stick open but it also
shuts down the reactor coolant pumps. And now you,
first of all, with one single event -- granted at a
lower frequency -- but now one single event has caused
multiple things to occur. A cool-down effect -- and
you've shut down the reactor coolant pumps because you
affected the circuits nearby, and now you're also
introducing the possibility of stagnation on top of
the event that normally would have been another
independent failure.
So, I think just on the surface, you can't
just say, well, external events, we don't have to
worry about.
MEMBER KRESS: Okay. I can see the
possibility of the fire being one, but I think I would
rule out earthquakes and floods and tornadoes, just
right off the bat.
MR. KOLACZKOWSKI: We certainly have some
insights. For instance, Calvert Cliffs has a current
PRA that has PTS involved along with the rest of the
PRA, the normal station blackout and all these other
sequences.
If their analysis is any indication, it
looks to them that, from an external event point of
view, the two that you need to worry about are fires
and internal floods. Now their results, just taken on
the surface, do suggest that external events do not
contribute as much as internal, but that's something
we want to check out and verify.
But it's not necessarily, should I say, so
insignificant that you can just rule it out right off
the bat.
MEMBER KRESS: Thank you. That helps.
MR. HACKETT: I guess what I'd say in
closing this opening piece, I guess what I'd like to
focus on -- we're going to go into giving you the
results and insights from Oconee. But I think the
focus really needs to be on the modeling process and
quantification on the uncertainty, which is really
what we tried to do in detail at the day-and-a-half
meeting.
So, we're going to try and keep that
focus. If we don't, let us know. We may need more
time to do that. If we need to come back, we'll do
that. But that's going to be the focus.
With that, I'll hand it off to Mark Kirk
to --
MR. KIRK: Do you want to do the Oconee
part now?
MR. HACKETT: Yeah. We might as well do
the Oconee part now.
MR. KIRK: Okay. You've seen this graphic
before. This just orients you in terms of the various
steps in the process that PRA defines the sequences
and the sequence frequencies. That feeds into TH,
which gives us pressures and temperatures to run
through the PFM model, which generates conditional
probabilities of vessel failure that are then
mathematically in mind with the sequence frequencies,
to give us an estimate of the yearly frequency of
through-wall cracking, complete with uncertainties.
MEMBER KRESS: The PFM analysis has inputs
that are like fluence --
MR. KIRK: Oh, yeah. Yes. Yes.
MEMBER KRESS: -- and crack size and
growth rates and things like that?
MR. KIRK: That's right. There's a whole
host of details under this, and we'll get to expanding
that later. That's absolutely right.
Just in terms of -- and I think this will
get to that -- some of the higher level things in each
of these analyses. In PRA, we start off with
something on the order of 10,000 sequences, just for
reasons of --
MEMBER WALLIS: It says there are 100,000.
MR. KIRK: 100,000.
MEMBER KRESS: That has to do with the
number of crack sizes you have?
MR. KIRK: Number of what?
MEMBER KRESS: Crack sizes.
MR. KIRK: No. Number of possible
initiating events in the plant.
MR. KOLACZKOWSKI: Number of scenarios --
just various combinations of valve sticking open/not
sticking open, both on the primary/on the secondary,
operators doing things/operators not doing things.
You look at all the possible combinations. Right now,
we've modeled something like 100,000 different
scenarios that could lead to overcooling.
MEMBER KRESS: Okay.
MR. KIRK: But in order to make the
computation tractable, that has to get partitioned
down into something on the order of a hundred
different bins that somehow, by the process that we go
through in PRA and in SAPHIRE, that represent these
hundred-thousand sequences.
MR. KOLACZKOWSKI: Dave over there doesn't
want to have to run 100,000 RELAP runs.
SPEAKER: I can't imagine why not.
MEMBER KRESS: When you -- you bin these
by similar consequences and add up the frequencies?
MR. KIRK: Basically, yes.
MEMBER KRESS: Okay.
MR. KIRK: And when we get into more of
the details about how the computation's being done.
Alan will be describing these more. But just, again,
to get the high points, in this analysis we use the
most recent data coming out of the plants, give credit
where credit is due for recent training and
procedures, and as I said, human reliability credit,
as appropriate.
That then feeds into the thermal hydraulic
analysis where, for the particular case of Oconee, we
had approximately 150 transients for which we actually
performed RELAP runs. Of those, approximately 50 of
them just got based on an inspection of the --
pressures and temperatures got screened out, meaning
they never got passed on to PFM. You could just look
at the transient and tell that it was adequately
benign, that it was really not a PTS precursor.
Of the remaining 100, approximately half
of them then fell into the base case and half of them
got used to assess --
MEMBER FORD: Sorry. Did you -- I know
more than you just were saying. I just want people to
be aware that there were quantitative criteria as to
which ones you threw out.
MR. KIRK: Yes. Yes, certainly.
MEMBER FORD: I did have more on a certain
Delta T in a certain time.
MR. KIRK: That's right.
MEMBER FORD: Anything above that was
chucked out.
MR. KIRK: The criteria for falling into
screened, not further considered, was based on some
very conservative, deterministic, probable -- I'm
sorry -- fracture mechanics calculations. And we, in
a sense validated those criteria because, of course,
having screened things out, some things weren't
screened that were right on the line. When we
actually ran the Oconee calculations, we found that
those had absolutely zero contribution to the
conditional probabilities.
MEMBER POWERS: As I look at this chart,
you have this PRA event sequence, and that's
metaphysics. You can't experimentally validate that.
You have thermal hydraulics. That's a
well-established discipline. You can have models that
are experimentally validated for the thermal
hydraulics.
Probabilistic fracture mechanics is
metaphysics that has some hope of being experimentally
validated.
MEMBER ROSEN: You're just fantastic.
It's the same old PRA.
MEMBER POWERS: The interface between
thermal hydraulics and the probabilistic fracture
mechanics -- you have some model. There's some model
there.
I guess what I'm struggling with is, has
there ever been an undercooling event at any pressure
vessel, thick-walled pressure vessel that led to a
failure that gives you some confidence that that model
you had between the thermal hydraulics and the
probabilistic fracture mechanics was -- it was a valid
say?
MR. KIRK: In the in-service, no.
MR. HACKETT: In the in-service, no.
MEMBER POWERS: Well, experimentally.
MR. KIRK: Experimentally, yes.
MR. HACKETT: Experimentally, yes, we
could say a few things because you're going to the
issue with the deterministic aspect of this thing.
And there, we have the benefit of a lot of historical
tests that were run at the Oak Ridge Laboratory for
the NRC research program, where we did in fact do
that. And the state of the art at the time predicted
with varying degrees of success.
But most recently, we've had the NRC staff
and the contingent from the Oak Ridge National
Laboratory has been very successful in predicting the
behavior of these large-scale reference experiments
that have been run under the NESC program, the
European Network on Evaluation of Steel Components.
In fact, I think Richard Bass is the
principal of the HSST program for us, and the staff
predicted the crack behavior of the NESC One spinning
cylinder test in the UK, I think just about dead on.
And again, that's a deterministic problem, so that the
probabilistics get you off into the other --
MEMBER POWERS: Metaphysics, yes.
Okay. I mean, that's something that
people don't always recognize is that the interfaces
between models as a model. That has to be validated,
and it seems like you've taken care of that.
MEMBER FORD: Well, Dana, you bring up a
very good point, could deterministic validation of the
failure could, as it existed before you put all the
probabilistic stuff in. You mentioned one point, the
spinning test in the United Kingdom. What about the
70 other odd tests that have been done around the
world?
MR. KIRK: Recently -- the answer is, you
know, if you ask me to pull out a report and give it
to you, I can't do that right now. It certainly is
possible to be done and I think we need to look into
that because approximately a year ago, under the HSST
program, we asked them to summarize, like you said,
the 70-odd reference experiments that are well
documented in structural scale.
Certainly, it would be possible, and I
think necessary here, to go back and use the --
there's a switch in FAVOR, where you can turn off the
probabilistic aspects and just use it as a fancy
deterministic calculator and use it to predict the
results of some of these well-known, well documented
reference experiments.
As Ed pointed out, we've been through that
exercise a number of time in the past, dating back to
the 1980s, using FAVOR precursors, which did well
using technology of the time. We haven't done that
with the current version of FAVOR.
MEMBER KRESS: As I recall, Dana, there
was almost a museum of huge thick-walled vessels at
Oak Ridge with impressive cracks. The thing I
remember though -- I may have a faulty memory and you
guys can correct me here -- it was extremely hard to
create those vessels with the embrittlement
characteristics that you get from the radiation. And
it's also kind of hard to reproduce the thermal
transients that you get. You had to do some things
with both of them.
MR. KIRK: There were some things that you
clearly had to do in the laboratory environment that
you couldn't entirely mock up to simulate the full
scale exactly.
And in the case of the embrittlement, what
was done was actually a long series of studies that
were done at Oak Ridge and elsewhere to try and
simulate the radiation embrittlement damage with
changes in the treatment of the steel so that there
was work done there.
In terms of simulating thermal transients,
in a lot of cases, you know, things like liquid
nitrogen had to be used to get some of the type of
Delta Ts that you think you would see or had seen in
the operational experience.
MEMBER KRESS: And they had to introduce
artificial cracks.
MR. HACKETT: Absolutely.
MEMBER FORD: But in the spinning tests in
the United Kingdom, though, they didn't simulate a
real PTS event.
MR. HACKETT: That one was very similar
because -- the spinning cylinder test, for those who
aren't familiar, they generate the stresses in the
wall by spinning a fairly large cylinder that had a
flaw embedded in the wall.
In that case, they induced the thermal
shock with the water spray that's fairly severe. And
that crack did initiate and run, and the deterministic
version, as Mark said, of the FAVOR code used at that
time, which is not the current version of FAVOR,
pretty much predicted that event right on.
There have also been the predecessor to
the NESC program. NESC evaluations are still ongoing,
or actually there's a son of NECS. Shah's here; he
knows what the name of that program is. But the
predecessor to that was something called FALSIRE --
Fracture Analysis of Large-Scale International
Reference Experiments -- if you like acronyms. And
there was actually a FALSIRE 1 and a FALSIRE 2
program.
Those programs looked at these -- I don't
know what the number is -- 50, 60, 70 experiments that
were fairly large-scale, some done at Oak Ridge, some
done in the U.K., Japan, also Germany, I know at least
were participants in that study, and then, looking at
the international contribution, their approaches to
taking this thing on, too.
Our approach is, as Mark said, would be to
have used FAVOR when we had it, or the FAVOR
predecessors, which were VISA and VISA 2D and OCA and
OCA-P, which were developed at Oak Ridge. So I guess
the bottom line of all of this is that there has been
a lot of work done in that area.
MR. MALIK: Yes, I would like to add to
Ed's comments. We have a report recently put out
about -- as (inaudible) explained -- we have combined
together in a single report. And that would be aspect
to go over and see how many we can predict with the
new code. We're going to look into that. But the
comprehensive report gives a little bit of detail over
all 70. Each individually would report on it's own
self, and we'll be trying to see how much it can do
under the new code.
During that earlier work, we have gone
through with all the (inaudible) as well.
MEMBER WALLIS: Now, Dana asked a question
about whether there were experiments to back up this
pressure-temperature-embrittlement-cracking
interrelationship. And he implied that we knew
everything we needed to know about thermal hydraulics.
But are we going to hear later about the experiments
that back up the analysis that goes into the thermal
hydraulics?
MR. BESSETTE: Well, later, yes. Not
today, though.
MEMBER WALLIS: Not today?
MR. BESSETTE: Well, there's not time,
really, to go into --
MEMBER WALLIS: Is that because you don't
have the evidence or you just don't want to show it
today?
MR. BESSETTE: No. I think we have enough
evidence now.
MEMBER WALLIS: So you can reassure us
that you do know what you're doing with thermal
hydraulics?
MR. BESSETTE: We can do that, yes. I
think we're capable of doing that. We are able to do
that. We're in the position to do that.
(Laughter.)
MEMBER WALLIS: You're waiting for the
right moment.
(Laughter.)
MEMBER FORD: But just following on from
that, do we have any back-up -- maybe we're jumping
into what you talk about later one. But the RELAP
code will give you the temperature, pressure and heat
coefficient variations as a function of time of the
liquid, which you say can be corroborated with data.
What about the similar corroboration of
the material temperature, stress, biaxial and triaxial
stresses, as a function of time and distance? Do we
have that data?
MR. KIRK: You mean simply a validation of
the thermal stress analysis?
MEMBER FORD: Correct, as a function of
variations of residual stress?
(Comments off-mic.)
(Laughter.)
MEMBER FORD: Well, I know, but everybody
keeps on saying that's an easy kind of exercise to do,
but has anyone validated it?
MEMBER SHACK: Well, there've been a lot
of thermal analysis tests that tell you know how to do
thermal analysis. You convert them into stress.
MEMBER KRESS: I don't think that's -- but
the question about the thermal hydraulics and its
uncertainties I think is a real legitimate one because
I don't think you can calculate these things as RELAP.
You use RELAP to set up -- to tell you
when, in a thermal hydraulic sequence, you want to end
up being stagnant. And then you use something else to
calculate the thing, don't you? You don't use RELAP
from there on, do you?
MR. KIRK: Well, FAVOR calculates its own
stress field, and we don't take that from RELAP. I
mean, RALAP will give you a temperature -- wall
temperature, but we don't use that wall temperature.
MEMBER SHACK: Yeah -- but you're right.
If you've got temperature and pressure, you can feed
it in. But to get that temperature and pressure, you
don't use RELAP, do you? That was my question.
MR. BESSETTE: We provide the boundary
condition at the vessel inner wall, but not the
conditions within the vessel.
MEMBER WALLIS: No, you give an H in a
heat transfer coefficient temperature, and the
pressure is the pressure of the volume.
MEMBER KRESS: And that comes from RELAP?
MR. BESSETTE: Yes.
MEMBER KRESS: What ever happened to
REMIX?
MR. BESSETTE: Remix --well, we used RELAP
-- you've heard about 150 calculations. Those were
all done by RELAP. We have some supplemental
calculations done with REMIX, but basically REMIX was
created because of the concern about the plume. And
since we've shown that the plume is gone --
MEMBER KRESS: But you don't that.
MR. BESSETTE: We don't the supplemental
calculations to provide the boundary conditions.
MEMBER WALLIS: I think what you're saying
is that RELAP makes various assumptions about how well
mixed the nodes are, and so on. And you have done
separate tests to verify that those assumptions are
okay.
MR. BESSETTE: That's correct because RELAP
cannot model plumes.
MEMBER KRESS: That was basically my
question, how did you deal with it. And I understand
it.
MEMBER WALLIS: Then the question about
how well will that model the whole transient and
whether or not it predicts its stagnation right is a
big question, isn't it?
MR. BESSETTE: That is the big question.
That -- you are correct.
MEMBER WALLIS: And that's the question
you're not proposing to answer today.
MR. BESSETTE: We don't -- I don't have --
I can't show you all the evidence for that --
MEMBER WALLIS: There's no way you can win
a prize on this show today, then.
(Laughter.)
VICE CHAIRMAN BONACA: Although there have
been a lot of applications or views of RELAP 5 by
vendors in those specific applications, and they have
submitted a lot of comparisons to experiments to
justify the use, and it's been approved by the staff,
so there should be available information by the staff
on the different applications done with RELAP 5.
MR. BESSETTE: All the risk-significant
transients that have shown up in the Oconee study have
been LOCAS. Speaking generally, we do have quite a
substantial validation base to speak about with RELAP
for LOCAS that we can reference.
MR. KIRK: And I think the other piece, in
response to Dr. Wallis's question, is that there are
experiments that David had run at APEX. And I know
the committee got to visit out there. And I think the
evaluation of that in detail is still in process, and
there will be information obviously available.
MEMBER WALLIS: I was just wondering about
what you're going to get in our letter about thermal
hydraulics if you didn't tell us anything.
MEMBER KRESS: We may have to just -- we
may have to punt on that until later.
MR. HACKETT: I guess we'll have to wait
and see how David's piece of this goes.
MEMBER WALLIS: I don't know how you'd
punt thermal hydraulics.
MEMBER KRESS: You say, we didn't discuss
thermal hydraulics and we'll comment on that later.
MR. HACKETT: I guess in the interest of
getting to the key part of this thing on the
uncertainty evaluation, we'll try and forge ahead
here.
MR. KIRK: I think the intent here was
just to provide a sense of what goes in the boxes, and
we're going to do that in more detail later, so we'll
go on.
One question that invariably gets asked --
and this of course can be gone into in any level of
detail one would desire -- is what's changed since the
analyses that established 10 C.F.R. 50.61 in the mid
1980s. So here, we've summarized the major
contributions and shown at least qualitatively by way
of the arrows what are the things that are driving the
risk of vessel cracking down, shown by the green
arrows, and what's driving the risk of vessel cracking
up.
Again, we'll be going into more detail,
but just to provide a sense of -- one of the biggest
things in PRA is the fact that we're now discretizing
these hundred-thousand events into many more bins than
we used to. Before, there was something like a
countable number of bins on two hands, whereas now
we've got something more on the order of hundreds. A
hundred still sounds pretty coarse when you looking at
an overall population of 100,000, but in getting them
into those bins, we have to make a lot fewer
conservative assumptions, and that has a very positive
impact on driving the risk numbers down.
MEMBER WALLIS: Isn't "PRA data" sort of
oxy-moron?
(Laughter.)
MR. KIRK: Or perhaps a metaphysical oxy-
moron. I'll pass on that one.
We have done some things --
CHAIRMAN APOSTOLAKIS: What do the arrows
mean?
MR. KIRK: The arrows mean that generally,
considering that factor or what we've done there, the
green arrows indicate that it would be driving the
risk down, whereas red arrows indicate it's driving
the risk up. So, for example, in PRA we've considered
acts of commission. You did the wrong thing that we
never considered.
CHAIRMAN APOSTOLAKIS: I think they don't
necessarily indicate that that's what happened. This
is what you would expect --
MR. KIRK: In some cases -- for example,
external events. We haven't considered external
events. We would expect that it would drive it up. In
fact, you know a priori it has to drive it up because
you didn't consider it before. You can't drive it
down if you didn't consider it.
CHAIRMAN APOSTOLAKIS: But it may not
drive it up significantly.
MR. KIRK: It may not drive it up
significantly, that's correct.
CHAIRMAN APOSTOLAKIS: This is what you
would expect.
MR. KIRK: In some cases, they are
results. For example, if you look in PFM, the first
arrow -- these aren't scaled, also. The first arrow
in PFM should be huge. We've removed a significant
conservative bias in the toughness model in the index
temperature. That's resulted by itself in at least an
order of magnitude reduction.
We've recognized the spatial variation in
fluence, whereas before the entire vessel was burdened
with the peak fluence. That's a major drop. Most
flaws are now embedded, rather than being on the
surface; again, a huge drop. And those are all
quantifiable.
So, in some cases, there are expectations.
In other cases like external events, which we haven't
considered, it's an expectation.
MR. KOLACZKOWSKI: Mark, let me even
clarify further for Dr. Apostolakis. With the
exception of the external events arrow, every one of
those other ones are results.
MR. KIRK: That's true.
CHAIRMAN APOSTOLAKIS: During the
subcommittee, I asked a question regarding, at some
point, getting information for -- quantitatively a
(inaudible) to get a sense of what comes from the PRA.
Because much of that has to do with eliminating a
number of limiting sequences, based on critical
(inaudible) action and getting a sense of how much
you're getting from PFM, because that's more
thermalistic in part. And I wonder if you have --
MR. KIRK: We haven't got that
quantification at this time, but that's something
that's definitely a stay-tuned document, and that
question is clearly on everybody's mind. What our
current goal is, is to get through -- Ed presented the
status -- turn the draft on Oconee into final. And
then a number of questions along those lines have come
up that I think represent the definition of
sensitivity studies that we're going to be needing to
do to better flesh that out.
MEMBER POWERS: Before you leave that
previous slide, or maybe after you're -- if I have a
PWR and I operate it with higher burn-up to fuel,
higher burn-up, and I sustain an axial offset anomaly,
do I change my fluence distribution in a way that you
have not taken into consideration?
MR. KIRK: The fluence distributions -- I
believe the short answer is yes. The fluence
distributions right now -- we have fluence maps that
were calculated at Brookhaven, at two different
fluences, I believe 32 and 40 years. Those are then
-- the other fluences that we analyze are extrapolated
from those, which is to say that we're assuming that
the fuel loading remains the same.
MR. HACKETT: Yes. If the burn-up -- of
if we were to address mocks or something like that and
it was to significantly change the neutron spectrum,
this project doesn't address those factors at this
point.
MEMBER ROSEN: But Dana, an axial
significantly skewed profile due to an axial offset
anomaly is not something that's going to be sustained
for 20, 30, 40 cycles. It's something that happens
potentially, but then the plant says "whoops" and
corrects it in the next cycle.
So, I don't know that it has an integrated
effect on fluence that's -- it doesn't have a
substantial effect on the integrated fluence because
it's corrected. It's not something that's sustained
over the life of the plant.
Am I correct on that, Ed?
MR. HACKETT: I would think that would be
correct.
MEMBER POWERS: I'm gratified at the
assurances but I'd like to know what it does do to
your mark.
MEMBER FORD: I think a much bigger fate
would be the fluence attenuation in the pressure
vessel vault, which has not been qualified, as I
understand it.
MR. HACKETT: That was one of our due
dils. from the last meeting. It hasn't been
quantified in an experimental sense by taking section
through a wall. And indeed, it would be very
difficult to do that in a statistical sense to get a
valid comparison. But there has been a lot of
modeling work that went into that. It's largely the
work of, I guess, Professor Odet and others.
But you're right. That's an area that's
also ripe for some more work in the future because
we're not proposing this project changing that
attenuation function.
MEMBER SHACK: But I think there's general
agreement that your attenuation function is a
conservative one.
MR. HACKETT: Yes.
MR. KIRK: Yes.
MEMBER SHACK: But how can you say that,
Bill, if you haven't done any experiments?
MR. HACKETT: People have done
experiments. They haven't done those experiments but
people have measured attenuation in steel. You know,
there are models for this and, you know, the model
that they've used is I think generally considered to
be a conservative one.
MEMBER FORD: Well, I know they've done it
-- they've measured in stainless steels, but have they
done it in ferrittic?
MR. HACKETT: Like Bill said, it's been
done in a modeling sense.
MEMBER FORD: I guess I'm being nitpicky,
but the point is that it does have a big impact on the
rest of the crack -- the specific fluence attenuation.
You know, that's a big output from this overall model.
MEMBER WALLIS: Isn't this just simple
physics about how neutrons behave in steel, and it's
all well understood --
(Laughter.)
MR. HACKETT: It's certainly physics. I
think there is probably a lot of contention about the
simple part I guess.
MEMBER POWERS: The neutron, we
understand. It's the steel that we don't understand.
MR. KIRK: Well, I think to follow up on
Dr. Shack's point, I think -- I mean, clearly it's not
within the scope of this project, although many people
would like to, myself included, whack up a vessel wall
into sharpies and test it through the thickness. That
would be very satisfying. Unfortunately, it would,
you know, consume the entire RES budget.
But, I do think it's incumbent upon us to
do what Dr. Shack suggested, which is to better
document the common engineering assumption that the
Reg. Guide 199 rev. 2 fluence function is
conservative. Again, ask me to pull out the document,
I'll say, no, I can't do it. But that's something
we're looking into putting together because I think,
you know, absent the experimental proof or the right
answer --
SPEAKER: Where's Art Lowe when you need
him.
MR. KIRK: I've got a Ouija board, if you
need it.
MEMBER WALLIS: Well, maybe -- the fluence
to me is just what the neutrons are doing. Their
effect on the steel is another question.
MR. HACKETT: I think Dr. Lewis from NRR
has some comments.
MR. LEWIS: Yes. I think we do know a
little -- my name is Lambrose Lewis. I'm from the
(inaudible) branch. We do know a bit more than what
you assumed previously.
We've had dosimetry inside and outside the
pressure vessel, so we know how many neutrons go in
and how many get out on the other side. What does
change, however, is the spectrum -- namely, those that
are left behind and exit the other side are much, much
harder, as they are a much harder spectrum.
Therefore, they cause percolision much greater damage.
However, there is several DPA -- the
function -- which one can quantify the damage per
neutron as it goes through. Therefore, we do have a
function which tells us how much damage has been done
from inside out; not only the number of neutrons but
also how much damage they caused in that. So, there
is a quantifiable function that we have.
Now, there are some measurements on the
cross-sections of steels that have been exposed -- not
pressure vessels but other steels -- that only refer
to, as far as I'm aware of, to hardness. People
measure hardness and say, well, yeah, roughly they
correspond to each other. So, to that extent there is
some rudimentary experimental evidence.
MR. HACKETT: Thanks, Lambrose. That's
what you get when you ask some metallurgist to try and
answer about dosimetry, it seems, and it's helpful.
MR. KIRK: We just had two fairly high-
level slides on the current results from Oconee, and
I should point out that these were results that were
generated back in December. As with any good
engineering calculation -- or maybe I should just say
any engineering calculation -- you go through and you
find that there are certain things that you should
have done better or maybe shouldn't ought to have
done. So these calculations will be redone next week.
Having said that, I'm not sure that any of
the overall results will change. This takes, now, yet
a higher level view of the 100,000 sequences that were
reduced into 150 thermal hydraulic bins and gathers
them together yet further, because we find as a result
of the analysis that approximately nine or ten of the
individual bins contributed over 90 percent of the
probability of through-wall cracking.
Just as a class, LOCAs are contributing by
far the largest part. They're responsible for well
over 50 percent of the crack initiations. However,
they're not as dominant in vessel failure simply
because the larger the LOCA break gets, the pressure
drops and the crack can't go through the wall.
The red section and the green section
provide an interesting example of the effect of human
reliability credit. Over on crack initiation, the red
and the green essentially all represent the same
starting sequence -- I'm sorry. The red and the green
all represent the same starting sequence. The
difference between the red and the green is that in
the case of the red we've got a stuck open pressurizer
safety valve that the operator, for whatever reason,
didn't intervene at all, and the valve simply re-
closes automatically. Whereas, with the green
section, we've got two variants of that transient,
where the operator throttled HPI at different times in
to the transient.
So, in this case, in the red you've got
one initiating event that produces a few initiations
but produces a whole lot more failures because HPI was
never throttled, whereas here you've got two of the
same starting events, which produced, of course, more
initiations. But because the operator was of course
doing the right thing, it didn't get as bad in terms
of punching through the wall.
This, of course, is just one snapshot --
CHAIRMAN APOSTOLAKIS: So this is -- this
would be a matter of omission, would it not?
MR. KIRK: This would be -- the red would
be an error of omission, yes.
CHAIRMAN APOSTOLAKIS: That could have
been done reasonably well in the '80s.
MR. KIRK: Yeah.
CHAIRMAN APOSTOLAKIS: It was not? It
didn't included human error in those studies? Pure
human action?
MR. KOLACZKOWSKI: The general answer --
if you look at the 1980s work, you will find that with
a few exceptions, a little to no credit for the human
doing anything at all was in those models.
CHAIRMAN APOSTOLAKIS: So in your previous
slide with the summary of changes --
MR. KOLACZKOWSKI: Correct.
CHAIRMAN APOSTOLAKIS: -- you should have
-- oh, no. I'm sorry. You have it. Good.
MR. KOLACZKOWSKI: Generally, as you give
operator credit, it makes the risks go down.
CHAIRMAN APOSTOLAKIS: So what is the
number now that you use for that? Will you talk to us
about it later?
MR. KIRK: The number for what? For human
credit?
CHAIRMAN APOSTOLAKIS: Yes. That's coming
up?
MR. KOLACZKOWSKI: Yeah. I mean, we can
show you some examples, if that's what you want to
see.
CHAIRMAN APOSTOLAKIS: Yeah, that's what
I want to see.
MR. KIRK: All right.
MR. KOLACZKOWSKI: By the way, Mark, those
will be on some back-up slides.
MR. KIRK: Okay.
These are, of course, results at just one
embrittlement level. We've run the calculations for
a number of embrittlement levels, and what you see on
this plot --
CHAIRMAN APOSTOLAKIS: You should add more
information to this. It's too simple.
MR. KIRK: We've already removed a lot of
information.
(Laughter.)
CHAIRMAN APOSTOLAKIS: Ah, it depends on
your starting point, right?
MEMBER SHACK: It's only the distribution
of epistemic frequencies.
CHAIRMAN APOSTOLAKIS: He's got only the
95th percentile. We need superimposed upon this an
epistemic distribution of --
CHAIRMAN APOSTOLAKIS: I'm looking.
MR. KOLACZKOWSKI: We have it.
MR. KIRK: The curves here represent the
95th percentile of the vessel failure distribution. So
that's some measure of an upper bound.
CHAIRMAN APOSTOLAKIS: The frequency of
vessel failure?
MR. KIRK: Yes. Yes.
CHAIRMAN APOSTOLAKIS: Due to any -- all
causes?
MEMBER ROSEN: No, no. PTS.
MR. KIRK: PTS. All PTS causes, yes.
CHAIRMAN APOSTOLAKIS: PTS.
MR. KIRK: Everything. The points to take
away here -- as we started off by performing a
calculation of the Oconee Plant at -- I used to say at
the end of their current license period, but that's
not correct because NRR's granted them an extension.
So, at the end of their original operating license,
this analysis was estimating their 95th percentile
through wall-cracking frequencies, down on the order
of 1E minus ten, 1E minus 9. Relative to our current
risk criteria, that's about four orders of magnitude
down.
If we take that same plant and crank up
the fluence so that the most embrittled axial weld and
the most embrittled circ weld cross our current 10
C.F.R. 50.61 screening limits, we find out that even
at that level of embrittlement -- and this is perhaps
the more important point for purposes of revision of
the rule. Previously we associated -- we would have
thought that the plant would have gone through these
screening limits at 5E minus 6, whereas for this
plant-specific analysis at Oconee, they go through it
more like 5E minus 8. So, we're again two orders of
magnitude.
MEMBER SHACK: Oh, you thought your
scoping criteria would limit to you to your acceptance
criteria? That's an interesting concept. I was going
to say, how much margin did I expect my scoping
criteria to have, and the answer is none, right?
MR. KIRK: That would have been my answer.
But that's a personal view.
MEMBER SHACK: Oh, okay. A different
answer -- or a different notion of a screening
criteria than I normally would accept.
MEMBER FORD: Could you point out where
the 60 --
MR. KIRK: Sixty is just the --
MEMBER FORD: Next one up?
MR. KIRK: -- yeah.
MEMBER WALLIS: So, in plain language, the
real story is down in the orange or whatever color
that is --
MR. KIRK: Yes.
MEMBER SHACK: The real story is up in the
blue.
MR. HACKETT: The real story is where you
force it up to where the --
MEMBER WALLIS: But you have to force it
up there.
MR. HACKETT: Yeah. Well, that's where it
would be.
MEMBER SHACK: If the vessel is
embrittled. Everybody agrees that if you don't have
an embrittled vessel, PTS is not going to give you
much of a failure. The real crit is, do we have the
criterion to describe embrittlement.
MR. KIRK: That's right.
MEMBER SHACK: And the answer seems to be
that our current criterion give you a conservative
answer. Just how conservative it is seems to be a
matter of --
MR. KIRK: Or how conservative it needs to
be.
MEMBER WALLIS: Well, the criterion is
conservative by a factor of 100, but the actual
operation is even more conservative because you never
get near the criterion.
MR. KIRK: That's right. And the actual
operation for Oconee is down here. Now, we might see
some differences, as Dr. Ford pointed out, in the four
plants that we're analyzing, if you walked up to your
random materials person in the street and said
Palisades, Calvert Cliffs, Beaver Valley I and Oconee
I, which ones are closer to limit now? You'd say
Beaver Valley I and Palisades. So when we do those
analyses, it will be interesting to see how they
compare with these.
MEMBER FORD: Now, there are a lot of
assumptions in this whole process. What are you going
to do if you find that is not the case?
MR. KIRK: Well --
MEMBER FORD: In other words, what do you
throw out?
MR. KIRK: Well, if you made me bet some
of my own money right now, I would conjecture, at
least from a materials viewpoint, they're all going to
be coming in pretty close to each other because of the
changes that we've made. The change -- right now, we
think about plants relative to the PTS screening
criteria using the embrittlement correlation in the
current reg guide.
In these calculations, we're using, of
course, the updated embrittlement correlation -- and
I see you have that report open, so you know about
that. That reshuffles the deck chairs, to the point
that now -- I mean, certainly it doesn't make Beaver
Valley and Palisades, you just say, well, why do you
ever worry about them? But the plants move around and
some of them become closer to whatever the screening
criteria might be than others.
And when we look at the distribution of RT
NDT values in the four plants that we're looking at,
they're much more similar than we would have
previously thought them to be.
MEMBER FORD: Okay.
MEMBER SHACK: Now, where would the mean
be? Here's the 95th percentile. Where would the mean
be? Is it a decade lower, or --
MR. KIRK: Not too much -- it's a very
skewed distribution, so not too much further down.
MEMBER ROSEN: Are you going to let us in
on what magical change -- shuffle the deck chairs, as
you say.
MR. KIRK: We could go through that again
in an immense degree of detail, but the differences
between the Reg. Guide 199 rev. 2 Embrittlement
Correlation and what we have now are, well,
considerable.
The old correlation was essentially just
an empirical curve fit to about 150 surveillance
points made in the mid 1980s. Since that time, we've
evolved to what I would call a physically motivated,
empirically calibrated model, where we use our
understanding in the physics of irradiation damage to
set the functional forms and use the surveillance data
to calibrate it.
So, I think the opinion of the technical
community would be that we've got a more robust
correlation now, and that it's tied to physics. And
what you find out is that because the old
embrittlement function was simply a curve fit, and a
very simple gross curve fit to all the data, some
situations were assessed more optimistically than they
should have been, and some were assessed more
pessimistically than they should have been.
MR. HACKETT: I guess I'd add in the
interest of moving on, Mark has gotten into some of
the PFM details. We do have at least a little bit of
the segment of the presentation here today on that, so
hopefully we'll get into some of those.
MR. KIRK: I think we've commented on most
of these. These were just a few observations that we
wanted to share coming out of the Oconee analysis.
We've already pointed out that the dominant scenarios
are all primary system LOCAs. The second point
follows from that, in that realistic accounting of
operator action has significantly mitigated the
influence of secondary system events on the total
failure probability in this plant. I've learned a lot
about PRA practice in going through this, and it's
been commented that we may not find this to be true in
the other plants.
One thing that I think came as a surprise
to many of us, but we've been able to understand, is
that main steam line break and steam generator tube
rupture, both of which were very dominant events in
early analyses, no longer occur, or no longer show up
as being dominant. That's occurring for a number of
reasons, perhaps most primarily being the fact that
those scenarios were dealt with very conservatively
before, and so their influence was artificially
elevated.
MEMBER WALLIS: Within the LOCAs, are
there big variations in the effects of different kinds
of LOCAs?
MR. KIRK: Certainly break size is a
factor.
MEMBER WALLIS: Is it a big actor?
MR. KIRK: Yes.
MEMBER WALLIS: So which LOCAs are the
worst?
MR. KIRK: Small to medium.
MEMBER WALLIS: Small --
MR. HACKETT: Small to medium.
MR. KOLACZKOWSKI: Yeah, small to medium.
MEMBER WALLIS: So, if someone came along
and said we should forget about large breaks, that
wouldn't be significant as far as PTS is concerned?
(Laughter.)
MR. HACKETT: I think it's probably
premature for us to say that, since I know, as Shook
mentioned to the committee, we were embarking on maybe
a couple of years worth of effort to look at
redefining large break LOCA. And I concur with that.
In this particular case, it goes fairly
obviously to the -- small to medium break LOCAs are
the ones that you're going to focus on because you
have the possibility for the repressurization with
HIPSI. And also, not to mention the stuck-open valve
scenarios, too -- safety relief, stuck open valve
scenarios.
MEMBER ROSEN: Did the overly conservative
treatment of main steam line break and steam generator
tube rupture lie mainly in what the operators were
credited with and what they were not credited with?
MR. KOLACZKOWSKI: There were actually two
effects. And I know we have some back-up slides on
this, but I'll try to answer it quickly, and if that's
sufficient, great.
There were two effects -- first of all,
you're right. They basically gave the operator no
credit in the '80s work, whereas, if you account at
all for the fact that once a severe depressurization
on the secondary is recognized, that once the operator
stops the feed to the bad steam generator, it's over.
`And we're finding out from the thermal
hydraulics that 10, 15 minutes into the accident is
how much time the operator has before it would even
begin to be what we would categorize as a challenging
event from a PTS standpoint. And believe me, when
operators have 15 minutes to do something, that's a
lot of time.
MEMBER ROSEN: And what they have to do is
figure out which steam generator is faulted --
MR. KOLACZKOWSKI: Correct.
MEMBER ROSEN: -- which is practically
screaming at them.
MR. KOLACZKOWSKI: Oconee also has some
unique indications to help them.
MEMBER ROSEN: -- and then trip the feed
to that steam generator.
MR. KOLACZKOWSKI: Yeah. All they've got
to do is recognize which one it is, stop the feed to
the generator, the event's over. He basically is
performing the same function of a main steam isolation
valve, if the break is downstream of the valve. But
of course, he's doing it a different way. He's
stopping the feed, rather than stopping the steam exit
sign.
MEMBER ROSEN: So he reaches up and trips
the feed to that steam generator.
MR. KOLACZKOWSKI: Yeah. I mean, it's --
in the case of this plant, it's -- there's a couple of
things. The main steam line break circuitry will
probably stop main feed already, automatically. But
even if you allow for that failure, all he's got to
trip the pumps. And, then if ox-feed comes on, it's
a one-handle action, and ox-feed is off. So, once he
recognizes to do it, it takes no time to do it.
MEMBER ROSEN: And we'll assume that he's
trained in that.
MR. KOLACZKOWSKI: Absolutely. The BNW
plants in particular are very sensitive to overcooling
because they know they've got, you know, as much more
sensitive plant than perhaps Westinghouse and CE,
where --
MEMBER ROSEN: Trained and drilled in the
simulator on it.
MR. KOLACZKOWSKI: Yeah. We watched -- I
don't know if we can get into some of this other
stuff, but as part of all this work and coming up with
the human reliability numbers, etc., we had the
licensee involved in coming up with these estimates.
we observed four to five simulated events on their
simulator at Oconee of overcooling events.
So, we got a feeling for operators, have
fast it can respond to certain things, what is the
lay-out, what are some of the problems with the
layout, etc. It was not just a bunch of PRA geeks
sitting around in an office somewhere and saying, I
think the human reliability number's 10(-2). We did
a lot more than that.
VICE CHAIRMAN BONACA: Why would you think
it would be so much plant-dependent? I mean --
MR. KOLACZKOWSKI: Well, in terms of this
particular kind of accident, as has been pointed out,
the key is recognizing that you've got a
depressurization of which steam generator it is, and
then just how many actions does it take to isolate the
steam generator.
VICE CHAIRMAN BONACA: Sure.
MR. KOLACZKOWSKI: I would say the simpler
those are, the more likely of success or the less
probability of failure. And as part of the
generalization step, what we're going to have to look
at is what are the training and typical layouts of
those kinds of indications at other plants? If
they're similar to Oconee, we would expect similar
results. If they're vastly different, then we'll have
to account for that.
VICE CHAIRMAN BONACA: Yeah, the reason
why I brought it ups is the susceptibility of the
steam generators at Oconee is, you know, much higher
than other steam generators, where you have a pot of
water. And so that's why. But I'm saying that the
expectation would be that you would find similar
trends in other NPRs.
MR. KOLACZKOWSKI: Anyway, to finish
answering your question, giving human reliability
credit is one reason why these accidents went away,
generally speaking. They don't completely go away,
but they're not among the dominant anymore.
And the other reason is, again, remember
that we mentioned that we're doing 100 to 150 bins,
whereas before in the early work they were using 10
bins. So whether it was a small steam line break or
a large steam line break, it got treated as a large
steam line break, it got treated as a large steam line
break in the early work.
So, the cooling transient was much worse,
etc. Just by virtue of separating those two and
making a small class and a large class, that by itself
made some of it not be as dominant because now you're
treating it more realistically, rather than treating
all breaks very conservatively. So, just the nature
of adding more bins made some of the conservatism go
away and made the main steam line breaks go away, for
instance. So there are actually two things that are
causing that to happen.
MEMBER POWERS: Let me return to your main
steam line break conclusion. In the not-too-distant
past, the committee has been entertained by the
description of a main steam line rupture. And in that
description, the speaker decried the ability to model
those events because he describe shockwaves, motions
of equipment and piping systems of a substantial
nature, if the thing really moved around on you.
Do you take those into account when you
come to this conclusion that the main steam line break
is not a contributor to this?
MR. HACKETT: I think -- let me try and
answer for Alan. He could probably give a more
precise answer because I think know what Dr. Powers is
referring to, which was an incident, as I recall, at
Turkey Point Plant, and I think the answer is now.
And I know, when that came up I think it
was presented as part of the steam generator action
plan, resolution of the DPO. In that case, there were
some potentially very significant distractions to the
operators that I don't believe we accounted for.
MR. KOLACZKOWSKI: Actually, if that's
what we're really getting at, I know when we came up
with the human reliability numbers that we decided to
use for failure of operators to, let's say, isolate
the faulty steam generator, that was one of the
contexts that we considered.
We said, you know what? There could be a
major distraction out there; maybe they're finding out
their best buddy was just killed in the accident, etc.
Wouldn't that affect, potentially, the human
probability of isolating because he's at least
momentarily distracted with, let's say, other issues
that he finds equally important. In fact, we tried to
consider that in our quantitative assessment for human
reliability.
MEMBER POWERS: Well, then how about just
the mechanical loads on the vessel?
MR. HACKETT: Yes. In that case, I guess
I'd say a couple of things. It's interesting. I
didn't know the points that Alan was making prior to
that. In the case that I think that Dr. Powers was
referring to, there were significant motions of
associated structures, and then I believe sonic booms,
things of that nature, referred to.
And it sounds like your opinion would be
that that is something that, despite that, these
operators and training that they received would still
enable them to do the isolation and mitigate the
situation effectively.
MR. KOLACZKOWSKI: Yes, if the question
you're getting at is, are there new things added to
this scenario now that create some kind of new LOCAs
out in the secondary side that can't be isolated or
something like that, well, then clearly the answer
would be no. But again, if the major crux of your
concern is, did we consider it at least in the
operator response, I would say, yeah, we tried to do
so.
MEMBER POWERS: I'm more interested in the
mechanical response.
MR. KOLACZKOWSKI: Yeah, if you're worried
about pipe whipping and the effect on the other
equipment, etc., obviously I would have to say that we
didn't think about that on the PRA side.
MR. HACKETT: I think as I understand the
concern that was voiced as -- again, this is a mixture
of things in the steam generator area.
I don't believe that that would result --
some of the concerns that were voiced were things like
oscillation of the steam generator tubes and potential
for fatigue damage and so on. I don't believe any of
that would result in any additional overcooling that
would impact the reactor vessel.
MEMBER POWERS: How about shockwaves
traveling through the steel affecting the crack
propagation?
MR. HACKETT: Again, as far as -- I would
see that that could affect the steam generator and the
steam generator shell could affect and probably would
affect the steam generator. I would not think there
would be a huge impact -- no pun intended -- on the
reactor vessel. And again, this is intuitive on my
part. I don't have any analysis to back that up, nor
have any been conducted that I'm aware of.
MEMBER POWERS: I guess that until you
have some substantiation, I'd be cautious about the
conclusion.
I think you're probably on better grounds
on your steam generator tube rupture, but until we
understand better what these dynamic effects -- sonic
booms, things like that -- are on the main steam line
break, I'd be cautious about the conclusion.
VICE CHAIRMAN BONACA: That's why I was
asking the relative contribution because there is a
question -- again, I mean, from the defense in-depth
set point the reliance on operator action has to be
examined because you're going to give a lot of credit
to the vessels.
MR. HACKETT: Good point.
MEMBER FORD: Ed, just looking at time, I
guess we'd better move ahead.
MR. HACKETT: Yes. I guess what I was
going to say, we've already effectively handed off to
Alan, so we'll just let Alan continue with the main
topic here, which is the uncertainty treatment with
PRA HR --
MEMBER POWERS: Possibly being a
controversy associated with --
MR. KOLACZKOWSKI: No. There couldn't be
any controversy.
First of all, to start this subject off,
let's keep in mind what the PRA part of this is trying
to do and what its contribution is to the overall
analysis. I tried to succinctly describe that here
with this bullet.
What the PRA is really trying to come up
with is the frequencies of a wide range of
representative plant responses, which we'll call
scenarios, that are each described by some set of T-H
curves, which will describe the pressure, temperature,
heat transfer coefficients of the resulting scenario
-- as the result of mitigating equipment successes and
failure, as well as operator actions that result in
various degrees of overcooling of the reactor vessel
downcomer (phonetic) wall.
So, really, if you think about it, there's
just two things that we're really trying to -- that
the PRA part is trying to address. What are the
scenarios? And, what are the frequencies of those
scenarios? When you start getting into that, there
are really two sources of uncertainty. There are
modeling uncertainties in trying to represent what are
the plant scenarios of interest, and then ultimately
there's uncertainty associated with the frequency of
each one of those modeled scenarios.
Let's go to the next slide.
From the modeling standpoint, again, just
think of the fact that each scenario is really a
collection of events. It starts with initiating
event, then the systems do certain things. Certain
ones work; certain ones don't. Operators do certain
things right or wrong. That collection of events ends
up resulting in some sort of a scenario.
As I told you, we have something like
100,000 scenarios or something modeled in this PTS
model. One of the things we had to introduce, because
the operator plays such a key role in when the
challenge ends or at least get mitigated to some
extent, etc., that we had to look at various timings
of operator action. Obviously, as the event starts
and the downcomer (phonetic) wall begins to cool down,
if the operator takes an action five minutes into the
event, that leads to a very different challenge than
if he waits to take that action 20 minutes into the
event.
So, we actually had to take many of the
actions and put into the model -- and we picked
certain discrete points to look at, of operators
taking actions ten minutes into the event, 20 minutes
into the event, etc., and therefore had to come up
with probabilities of each one of those.
CHAIRMAN APOSTOLAKIS: So these times are
sequence dependent?
MR. KOLACZKOWSKI: yes.
CHAIRMAN APOSTOLAKIS: I mean, you don't
have a 10-, 20-minute --
MR. KOLACZKOWSKI: Necessarily over
everything, no.
CHAIRMAN APOSTOLAKIS: Could you give us
an example of this now?
MR. KOLACZKOWSKI: Yes.
CHAIRMAN APOSTOLAKIS: How did you get the
-- pick any example you want -- the probabilities for
this?
MR. KOLACZKOWSKI: Probably what we ought
to do -- now the one that is dominant --
CHAIRMAN APOSTOLAKIS: It's not in your
main presentation.
MR. KOLACZKOWSKI: It won't be here,
George, but to try to answer your question, one of the
most dominant scenarios is a transient where by a
safety relief valve when the pressurizer gets demanded
-- it sticks open.
CHAIRMAN APOSTOLAKIS: Yes.
MR. KOLACZKOWSKI: It stays open for a
while, and as the pressure comes down it finally re-
closes.
CHAIRMAN APOSTOLAKIS: Yes.
MR. KOLACZKOWSKI: That's an extremely
challenging event for the operator because once the
SRV re-closes -- you've got to remember, a high-
pressure injection's probably on full at this point
because the SRV at Oconee is big enough that he needs
about all the HPI that he can put in. He's not
necessarily throttling it yet. And suddenly the SRV
re-closes, so the system begins to refill, of course,
right away and the pressure goes up through the roof.
And we're talking over minutes of time. It's not very
long.
In talking with the Oconee operators and
training staff, they acknowledge this is an extremely
challenging event for the operator to catch because
he's also not allowed to throttle HPI until he meets
certain criteria -- five degrees sub-cooling and
roughly a hundred inches in the pressurizer.
Well, a the sub-cooling comes back up
after the SRV is re-closed, and the level starts
coming back up, etc., believe me, you go through that
sub-cooling and that level real, real fast. And
before he knows it, he's got 50 degrees sub-cooling
and the pressurizer level's up through the top of the
pressurizer.
We recognized that that event was
happening so fast, George, was that we did was we
said, okay, this is an action where we ought to come
up with a probability for the operator failing to HPI,
one minute after reaching the throttling criteria; ten
minutes after reaching the throttling criteria; and
then we said, or, we're going to assume that if he
doesn't do it by ten minutes, he never does it.
And so, we came up then with recognizing
what indications would there be? What did the
procedures tell him to do? How is he trained? Etc.
We came up with probabilities of the operator failing
to throttle at one minute after reaching the
throttling criteria, ten minutes, or then we assume
never, as opposed to a slower transient, or maybe, as
I mentioned, a main steam line break -- w e don't even
get concerned until ten minutes. So we didn't even
choose our first time until ten minutes into the
event.
CHAIRMAN APOSTOLAKIS: Right.
MR. KOLACZKOWSKI: So the times were
dependent on which scenarios we were looking at.
CHAIRMAN APOSTOLAKIS: As they should be.
Now, this evaluation of the probability,
this was done by asking experts, describing the whole
thing -- that's what you say in one of the slides on
the separate meeting.
MR. KOLACZKOWSKI: Yes. Oconee was done
in the following way. NRC contractors got together.
We talked about the context of each scenario -- I know
you've heard this from another project that I won't
mention.
CHAIRMAN APOSTOLAKIS: That was my next
question.
MR. KOLACZKOWSKI: We talked about the
qualitative context, etc. What's going on? What are
the indications doing? Where are we in the step?
Remember, we had observed certain simulations, so we
had some idea, how long does it take operators to get
to certain steps, and things of that nature. What if
an instrument has failed? How much would that add as
a confusion factor? We may want to put that on the
tail-end of our distribution. We talked about a
number of contexts.
Contractors then get together and made an
estimate with regard to a mean and an uncertainty
bound on the failure probability.
CHAIRMAN APOSTOLAKIS: (inaudible).
MR. KOLACZKOWSKI: I'm sorry?
CHAIRMAN APOSTOLAKIS: All (inaudible).
MR. KOLACZKOWSKI: Yes. And there, we
limited it to four different values. We said .5, .1,
.01 or .001. That had some quantitative definitions
associated with it.
MEMBER ROSEN: What were those numbers?
MR. KOLACZKOWSKI: .05, .1 --
MEMBER ROSEN: No, no. What are those
frequency --
MR. KOLACZKOWSKI: Those were the failure
probability of the human to take that action at
whatever time we were looking at.
MEMBER ROSEN: Yes.
MR. KOLACZKOWSKI: We then took that
information -- after we quantified all the human error
events in the model, we then provided that to Oconee
and actually went back down to Oconee and spent the
day with their training and operator staff, and we
said what do you think? We tried to calibrate them in
probability and so forth, because they also
understand, this is likely to happen or this is not
very likely to happen.
MEMBER ROSEN: right.
MR. KOLACZKOWSKI: And you have to ask
them, well, what does not very likely mean? Does that
mean one in a thousand crews would fail, or one in ten
thousand, or one in a hundred. So there was a little
bit of a calibration that had to go on.
But nevertheless, we got through that and
then we asked Oconee for their comments. They
provided comments on our human error probabilities.
I would say that they agreed with probably 50 percent
of them. The other 50 percent -- generally, they did
not have too much in the way of major problem with
most of our means. In some cases, they thought our
upper bound was too high.
MEMBER ROSEN: They claimed to be not as
good as you thought they were?
CHAIRMAN APOSTOLAKIS: No.
MR. KOLACZKOWSKI: No. They generally
claimed to be a little -- to be better. The worst
crew would be better under the worst conditions.
Quite frankly, we agreed with them on a
few cases but in other places we held our ground and
said, no, we still think that the range is this wide
because of the following reason.
In other words, Oconee had a change to
basically review our human error probabilities,
provide their input to that. And to some extent we
changed things; to some extent, we didn't.
CHAIRMAN APOSTOLAKIS: I think that what
you describe is reasonable.
Now, maybe it's not related to this
project, but we have been told several times that
ATHENA is being implemented this year in a very
successful way in the PTS project.
MR. KOLACZKOWSKI: That's in Palisades.
CHAIRMAN APOSTOLAKIS: What -- so you
didn't do it for Oconee?
MR. KOLACZKOWSKI: Well, I explained what
we did for Oconee, which is kind of going along the
lines of where ATHENA was trying to get to. Remember
now, it still relies on expert elicitation. In this
case, the experts were the NRC contractors, but then
Oconee did a review.
In the case of Palisades, we've already
done our first cut at the human error probability of
the Palisades PTS model. In that case, we did it
differently. We actually went up to the plant. In
three to four days, we had three licensee trainers and
one operator crew member as experts, and then we had
two NRC contractors who actually had five experts, and
we together went through contacts, etc., developed
probability, got them calibrated, developed
probabilities.
And in that case, we did not limit it to
the .5, .1, etc. We actually asked for a complete
distribution -- a one percentile, a 25 percentile, a
50 percentile, etc.
So I guess I would say, as we keep trying
to apply ATHENA and where it is going better and
better each time, we're getting more rigorous, and in
the case of Palisades, we actually wrote down
everything about the context -- which ones are
aleatory uncertainties, which ones are epistemic?
We're trying to keep doing this better and better as
it goes, but again, the underlying quantification
mechanism is an expert elicitation process.
CHAIRMAN APOSTOLAKIS: So in the case of
Oconee, then, you did not use ATHENA.
MR. KOLACZKOWSKI: I would you say we kind
of use it.
CHAIRMAN APOSTOLAKIS: You didn't. What
you just described is what competent HRA analysts or
anyone in the world would do. They understood the
situation and then they said, this is our best
judgment. Now we try -- you know, what do you think,
and so on. I mean, there's nothing --
MR. KOLACZKOWSKI: That's true.
CHAIRMAN APOSTOLAKIS: -- I'm not trying
to put it down. It's just that it's --
MR. KOLACZKOWSKI: No. That's true.
CHAIRMAN APOSTOLAKIS: I understand that
now.
So, the other thing I have to understand
is modeling. Now, by the way, this is human error of
omission because they don't throttle.
MR. KOLACZKOWSKI: For the most part --
oh, yeah. The ones I've been talking about are like,
failed to isolate a steam generator. That's an
omission. Failure to throttle is an omission.
CHAIRMAN APOSTOLAKIS: Is that a
commission error or something?
MR. KOLACZKOWSKI: We did find one or two
commission errors that we modeled, such as
inadvertently tripping reactor coolant pumps when you
don't want to.
CHAIRMAN APOSTOLAKIS: Okay.
MR. KOLACZKOWSKI: That tends to make
things more stagnant. And we did look at conditions
that operators might get fooled into thinking that
they should do that, and we tried to quantify those.
CHAIRMAN APOSTOLAKIS: And you -- through
the same process of evaluating it and so on?
MR. KOLACZKOWSKI: The same process. Now,
you also find that the procedures do cause acts --
I'll call them acts of commission that will worsen a
PTS event by procedure. But it's because they're
trying to prevent core damage. They're trying to
prevent undercooling.
As an example, say you have no feed to the
steam generators at all. Well, what they're supposed
to do is depressurize the plant in sort of an
emergency mode, open up all the turbine bypass valves
and try to get condensate into the steam generator.
Well, when you do that, it cools down the plant.
That's a procedure-driven commission. It's not an
error in the sense --
CHAIRMAN APOSTOLAKIS: Right. So how did
you handle that?
MR. KOLACZKOWSKI: We handled that and
basically said, well, he's going to do that. So, for
all practical purposes, that's close enough to a 1.0
probability that that's what's in the model. He will
do what his procedures tell him to do. It turns out,
it's worse for PTS, but that's handled in the model.
CHAIRMAN APOSTOLAKIS: Okay. So now we
have to find another example of modern uncertainty
that's not related to human errors.
MR. KOLACZKOWSKI: But there are some.
CHAIRMAN APOSTOLAKIS: Yeah, and I'd like
to see how you quantify those.
MEMBER WALLIS: Do the thermal hydraulics
people give you ways to predict uncertainties,
modeling of thermal hydraulics? Oh, you've -- there
must a more logical way to do it than that.
CHAIRMAN APOSTOLAKIS: There are
uncertainties.
MR. KOLACZKOWSKI: Oh, yes. I mean, the
T-H uncertainties are handled in a different way.
Now, they end up potentially affecting our model.
Let me give you an example. Actually, it
will address some of the things that are in the
presentation anyway, so -- again, the SRV's stuck open
and re-closes. We don't know when it's going to re-
close. If it re-closes very quickly into the
accident, again, it's relatively benign from a PTS
standpoint. If it re-closes very late in the accident
and then this sudden repressurization happens, much,
much worse.
So what we did was, we took the possible
continuum of when the valve could re-close. And I'll
grant you, this is very crude, but we basically said
we're going to discretize out into two points. It re-
closes at 6,000 seconds in to the event or it re-
closes 3,000 seconds into the event. And we can get
a little bit into why we picked those particular
times.
CHAIRMAN APOSTOLAKIS: The reason why you
use seconds is because it's thermal hydraulics?
MR. KOLACZKOWSKI: That's right. As soon
as you go to RELAP, everything's in seconds, rather
than minutes or hours.
MEMBER ROSEN: If it says minutes, you're
in HRA.
(Laughter.)
MR. KOLACZKOWSKI: Or, HRA is in minutes.
That's correct.
MEMBER WALLIS: These are uncertainties in
the inputs to RELAPs. They're uncertainties in the
RELAP itself.
MR. KOLACZKOWSKI: Yes. That's a whole
other issue that Dave can address to some point here.
But I'll just point out that what we did was we then
did was discretized the time into two times and ran
thermal hydraulic runs for the 3,000-second case and
the 6,000-second case, and then ran those through the
fracture mechanics code to see how different the
results would be, and treated that as basically an
aleatory uncertainty.
We don't know when it's going to re-close.
We're going to say there's a 50-percent chance it
closes around 3,000 seconds and there's a 50-percent
chance it closes around 6,000 seconds. That's trying,
if you will, capture the aleatory uncertainty in when
is it going to close.
CHAIRMAN APOSTOLAKIS: Then you have a
epistemic on the 30 percent.
MR. KOLACZKOWSKI: But then you have
epistemic on the 30 percent as to what the frequency
of that --
CHAIRMAN APOSTOLAKIS: You did that, too?
MR. KOLACZKOWSKI: That is correct. Yes.
CHAIRMAN APOSTOLAKIS: And that was,
again, by calling expert people?
MR. KOLACZKOWSKI: Yeah. You had the HRA
uncertainty, which I described. You had the
uncertainty of initiating event frequency, the
uncertainty of failure probability --
CHAIRMAN APOSTOLAKIS: -- But I meant, for
the timing of vessel re-closure.
MR. KOLACZKOWSKI: Yes. You called people
to get those distributions. I mean, it was an expert
opinion kind of --
MR. KOLACZKOWSKI: Yes, but in this case,
as I said, it was a pretty crude one. We just said,
we're going to discretize all the times into two
times; 3,000 seconds and 6,000 seconds.
MEMBER ROSEN: The fact of the matter is,
the only thing that knows when it's going to re-close
is the valve.
MR. KOLACZKOWSKI: That's correct. And
it's purely random. It's not an epistemic
uncertainty.
MEMBER ROSEN: It depends upon how it was
set up and how old it is and what it's commission is,
and maybe 20 other things.
MEMBER FORD: Could I again -- I assume
you don't want to go much beyond five.
MEMBER ROSEN: We're having a lot of fun.
CHAIRMAN APOSTOLAKIS: This is stuff we
have to understand. I mean, there's --
MEMBER FORD: I know, but there's also a
whole lot on thermal hydraulic uncertainties, too.
CHAIRMAN APOSTOLAKIS: But I thought they
were not going to discuss thermal hydraulic
uncertainties.
MR. HACKETT: That's one of the things we
may need some guidance on, given the time, because we
were shooting towards five, too, for everybody's sake,
I guess, and Alan has --
MR. KOLACZKOWSKI: Let me go to the last
line.
CHAIRMAN APOSTOLAKIS: That's an excellent
idea.
MR. KOLACZKOWSKI: Let me just go to the
last line -- the one after that.
MEMBER POWERS: We'll make sure that you
don't make your time limit. You have listed down here
operator actions were treated as random events, but
they would seem to have a strong covariance with the
initiating event and the mitigating successes or
failures. Is that true?
MR. KOLACZKOWSKI: Yes, that's true. I
mean, I guess to say that they're purely random,
that's not the case. Obviously, training and
procedures direct certain kinds of actions, and so you
would expect to see them as more prevalent, that is
true. But, yes. You're right.
MEMBER POWERS: And later on in the slide,
you say you used Latin hypercube sampling, and that
means, ipso facto, you have a variance there.
MR. KOLACZKOWSKI: Yes.
MEMBER POWERS: If you're going to turn
around and use the 95-percent confluence bound on
these.
MR. KOLACZKOWSKI: No.
MEMBER POWERS: That means you much have
-- well, I mean, you present everything in 85
percentiles that I've seen.
MR. KOLACZKOWSKI: Yeah, but we gave the
entire resulting uncertainty distribution to the PFM
folks.
MEMBER POWERS: Yeah, but in the end
they're going to come up with a 95 percentile that
they're going to show us, right?
MR. KOLACZKOWSKI: Well, that's what
they're shown you but they can show you the whole
distribution if we so desire.
MEMBER POWERS: But you haven't got tails
in the distribution right, because you used Latin
hypercube sampling. Your details are all short
compared to the real details.
MR. KOLACZKOWSKI: Yes --
MEMBER POWERS: Your distributions are all
peaked. That's what Latin hypercube sampling does for
you.
MR. KOLACZKOWSKI: Yeah. It makes a
process more efficient. I don't know if I agree --
MEMBER POWERS: -- you get a narrowing of
the distribution any time you use a Latin hypercube
sampling.
MR. KOLACZKOWSKI: Not being a
statistician expert, I've got to think about it. Yes
-- but, okay, if you're suggesting that we ought to go
and also use Monte Carlo or something?
MEMBER POWERS: Monte Carlo will do the
same thing for you; it just doesn't it as bad.
(Laughter.)
MR. KOLACZKOWSKI: Well, we can certainly
use other sampling techniques.
MEMBER POWERS: You can sample until you
turn blind. You'll still end up with narrow
distributions relative to what you'd have if you went
to an infinite sample distribution. You've got to
have some measure of how bad you are.
MEMBER FORD: And to help you with your
timing, Bill, are you agreed that you have no
questions on the PFM?
MEMBER SHACK: No. I think that's rocket
science compared to the rest of this.
(Laughter.)
MEMBER FORD: So you don't have to do
anything on PFM?
CHAIRMAN APOSTOLAKIS: Is PFM an accepted
approach by the experts? Is that something that's
been validated?
MR. KOLACZKOWSKI: Yes.
MR. KIRK: Yes.
MEMBER FORD: There's a whole lot of
questions I have, which will be --
MEMBER SHACK: Well, there's PFM and
there's PFM. I mean, what part of PFM are we talking
about? The probability or the fracture mechanics?
MEMBER POWERS: But as a discipline, not
only is it relatively well established, some of the
gentlemen involved in this study were part of those
that established it.
MEMBER FORD: Really? So, why don't we
concentrate on this last slide and on then on the
thermal hydraulic --
MR. HACKETT: All right.
MR. KOLACZKOWSKI: I just want to point
out that ultimately, what is being provided to the
fracture mechanics part, the T-H part, essentially the
output of the PRA part is -- and actually it's coupled
with the T-H.
Imagine that what's being provided as a
set of T-H curves, which of course are developed from
the thermal hydraulics portion of this, that
correspond some scenarios that we say -- okay, these
scenarios are indeed representative of this particular
set of T-H curves. And then, what's the frequency of
that bin when you add up all the scenarios that are in
that bin? What does the frequency of that bin look
like?
We described that frequency via histogram
that then went on to the fracture mechanics people,
who then sampled this histogram for a given sequence,
and then of course sampled across all the sequences
that cause overcooling.
When you're sampling across all the
sequences, you're essentially addressing, if you will,
the aleatory uncertainty and all the ways overcooling
an occur. Then when you sample from the individual
frequency histogram, you're trying to address the
epistemic uncertainty in that frequency, when you go
and ultimately develop the uncertainty in the PFM
results.
So again, what's being provided here as a
bin, a frequency, and that frequency for that bin is
being described by a histogram, and of course you're
doing that for some number of scenarios.
CHAIRMAN APOSTOLAKIS: Now, there's an
interesting comment in one of the viewgraphs from the
subcommittee meeting. It said that there were
situations where uncertainty types were mixed. And I
was wondering how you guys handle that. That was
slide 64, Current Toughness and Embrittlement models.
MR. KIRK: We fixed that.
CHAIRMAN APOSTOLAKIS: You fixed that.
You separate it.
MR. KIRK: We separated it.
CHAIRMAN APOSTOLAKIS: And how did you do
that? I mean, was there --
MR. KIRK: I forgot my wizard's hat.
(Laughter.)
MR. KIRK: We want to do that? I mean, we
may as well --
CHAIRMAN APOSTOLAKIS: They were in the
embrittlement model?
MR. KIRK: Yes.
CHAIRMAN APOSTOLAKIS: I would expect them
to be there.
MR. KIRK: What we did was we used a
process, which is shown here -- apparently, I can't
get my slide show working any more -- whereby we use
physical models to help to identify -- well, we took
apart the process to identify where the uncertainties
are, and then we referenced our non-toughness data,
which is what mixed the uncertainties, to real
toughness data to help us to separate out the aleatory
and epistemic parts so that we could treat them
separately. So then, each of the major variables in
the toughness model -- the index temperatures, the
toughest variables themselves, were individually
identified as being aleatory or epistemic.
CHAIRMAN APOSTOLAKIS: When you say either
one or the other, you mean situations where one
dominated the other?
MR. KIRK: Yes. And in reality, one
dominated. For example, in the toughness variables
themselves, the K1-c and K1-a, were dominantly
aleatory, and so we propagate those uncertainties as
distributions of the models, whereas the index
temperatures were predominantly epistemic. So those
are sampled.
MEMBER KRESS: Let me see if I understand
your uncertainty on the frequency again. You don't
have to go back to the slides; I'll just ask the
question.
You have identified something like
several, a hundred thousand sequences. You've binned
it into 150. So each of those bins has a lot of
sequences in it, and somehow each of those sequences
in that bin has associated with it a frequency, and
you convert that into a histogram. Where did the
frequencies come from in the first place?
MR. KOLACZKOWSKI: Because a scenario is
a combination of events, they're essentially a
multiplication of initiating event frequency, time-
certain failure probabilities of certain equipment,
time-certain failure probabilities of operator
actions, each of which has a mean and uncertainty,
etc., so you end up --
MEMBER KRESS: So you took all hundred
thousand of these and got their --
MR. KOLACZKOWSKI: Well, no, because of
course what you find out is that among a hundred
thousand sequences, there are 500 that really matter.
You've got to remember that we threw some sequences
because --
MEMBER KRESS: Okay, so that histogram may
have -- for each bin, they have several hundred
sequences in it?
MR. KOLACZKOWSKI: Some may be as many as
that. Some may be as few as just a few sequences. It
just depends.
MEMBER KRESS: Now, the frequency of this
sequence is -- that histogram you had.
MR. KOLACZKOWSKI: Yes.
MEMBER KRESS: Was that just the
initiating event frequency, or --
MR. KOLACZKOWSKI: No. That's ultimately
-- imagine that, if you will, as to all the sequences
that are in that bin mushed together to get what is
the frequency of that bin and the uncertainty in that
bin.
MEMBER SHACK: It's sort of the frequency
of the thermal hydraulic event associated with that
bin.
MR. KOLACZKOWSKI: Yes.
VICE CHAIRMAN BONACA: So there's specific
trend, pressure temperature.
MEMBER SHACK: The depreciating
frequencies have all been narrowed down, and you're
coming up with eight. What you're really getting is
a frequency of a thermal hydraulic event.
VICE CHAIRMAN BONACA: Right. All the
bunches of sequences in the bin will have a pressure
temperature behavior that resembles the one that
you're --
MEMBER SHACK: And he's got an epistemic
distribution because all of his frequencies -- in
fact, he doesn't have any number. He's got a
distribution for each of his frequencies. He's got a
probability that something will happen, but each of
those probabilities has an epistemic distribution.
MEMBER KRESS: And he takes the mean of
those?
MR. KOLACZKOWSKI: Well, ultimately it
calculates the mean. We don't really pass the mean
necessarily along, although we could. We provide --
the resulting distribution goes to the PFM folks. If
they want to know what the mean is, they can figure it
out, but we give them the whole distribution.
MEMBER POWERS: What you send to them is,
in fact, a histogram, right?
MR. KOLACZKOWSKI: That is correct. We
send them the histogram.
MEMBER POWERS: And I think that because
you send them a histogram, what they get is an
artificially narrowed distribution.
MR. KOLACZKOWSKI: I've taken your note on
that. We will do some other sampling techniques and
see how much it changes.
CHAIRMAN APOSTOLAKIS: Now I'm curious.
You have all these -- I mean, you said earlier that
the dominant model uncertainty was quantified using
expert judgment and that was one of the most --
MR. KOLACZKOWSKI: For example, when the
SRV re-closes.
CHAIRMAN APOSTOLAKIS: -- and so on. So
if I go now to the early slide that showed -- yes, the
preliminary results for Oconee 1, and you recognize,
or you conclude, that there are orders of magnitude --
the current results are four orders of magnitude lower
after 40 years of operation, two orders at the current
screening -- yes, exactly.
And knowing that a lot of this stuff
really depends on expert judgment, if I were a
structuralist, I would ask myself, what if I'm wrong.
And I would leave Regulatory Guide 1.154's criteria
alone because if I'm wrong, I'm wrong -- right? So,
would Dr. Powers agree with this kind of thinking?
MEMBER POWERS: No. This is not one where
I'll go along with you.
CHAIRMAN APOSTOLAKIS: I'm a structuralist
and I recognize that they've done an excellent job,
but my goodness, they depend on expert judgment a lot,
and experts have been wrong.
MEMBER POWERS: They certainly have a
history of being wrong more often than they are right.
That is absolutely true.
CHAIRMAN APOSTOLAKIS: So I'm scared now.
I want to be a structuralist.
MEMBER POWERS: Okay. You're going to be
a very conservative structuralist.
CHAIRMAN APOSTOLAKIS: Oh, now we have a
new classification.
MEMBER FORD: But George, surely, now
you're arguing about what that risk criteria should
be.
CHAIRMAN APOSTOLAKIS: No, I'm not arguing
with them. I'm arguing with him.
MEMBER FORD: Oh, okay.
MEMBER SHACK: You're still arguing over
the frequency. We haven't even gotten to the risk
criteria.
MEMBER FORD: That's my point. Let's my
point.
MR. KOLACZKOWSKI: George, we're trying to
be risk-informed.
CHAIRMAN APOSTOLAKIS: But if I'm a
structuralist defense in-depth (inaudible), then I'm
not risk-informed.
VICE CHAIRMAN BONACA: Well, no but if a
structuralist wants to know what the consequences are,
then --
CHAIRMAN APOSTOLAKIS: And then say, what
if you're wrong and goes back --
VICE CHAIRMAN BONACA: Well, no, he tries
to be reasonable.
CHAIRMAN APOSTOLAKIS: Oh.
MEMBER POWERS: That's why we have a
containment building.
(Laughter.)
MEMBER FORD: I'm trying to get through by
five o'clock here, George, and we've got all the
thermal hydraulic stuff here.
CHAIRMAN APOSTOLAKIS: That would mean
it's just too much work --
(Laughter.)
MEMBER POWERS: It's a well-established
science.
CHAIRMAN APOSTOLAKIS: It's a well-
established science.
VICE CHAIRMAN BONACA: Oh, yes. Cracks
propagate, then it could be trouble.
MR. BESSETTE: I'll try to go quickly
through thermal hydraulics.
MR. KOLACZKOWSKI: Yeah, you're going to
try.
CHAIRMAN APOSTOLAKIS: Let me understand
this. You said earlier in answering a question that
we will not discuss thermal hydraulics today. What is
--
MEMBER FORD: I said we wouldn't discuss
the criteria today.
CHAIRMAN APOSTOLAKIS: No, no, no.
MEMBER SHACK: Just the fundamental
verification of the RELAP code today.
MR. BESSETTE: Yes. We wouldn't talk about
the general validation of RELAP, overall assessment
work.
MEMBER SHACK: This now assumes that RELAP
works.
MR. BESSETTE: Yes. This assumes that
RELAP has some connection with reality.
(Laughter.)
CHAIRMAN APOSTOLAKIS: It says, RELAP 5
Internal Model Uncertainty.
(Laughter.)
CHAIRMAN APOSTOLAKIS: So you recognize
the uncertainties.
MR. BESSETTE: So we were faced with --
first of all, we tried to classify what we were
dealing with. And so, we categorize things into
events, sequence modeling and mapping, just basically
trying to determine what bins we're dealing with.
We realized early on that what was
dominating the outcome, or the uncertainties, in these
events was not the physical remodeling of RELAP but
rather the definition of the sequences. And there, we
worked a great deal with the PRA people.
So, my breaking down the sequences into
finer bins, we were getting much better -- you know,
a greater return on our --
CHAIRMAN APOSTOLAKIS: I don't understand.
I'm confused now.
MEMBER WALLIS: I think what you're saying
is there's much more uncertainty in the PRA modeling
than there is in the RELAP.
MR. BESSETTE: That's right. If you
strain the sequence a little bit, you can get
significantly different outcome, in terms of does the
operator act or not act, does the valve open or not
open. It's much greater determinant to these key
factors of temperature and pressure than the physical
models within RELAP. So, we put a great deal of effort
into expanding and defining the bins. That's one thing
we spent a great deal of time on.
Then you get into the use of RELAP itself,
and again, we can categorize that according to the
uncertainty in the hundreds of physical and empirical
models that comprise RELAP and basically in the
preparation of the input deck, how you model the
plant.
The first category of this thing is, we
treated -- we tried to decide what are the dominant
modeling features of RELAP that would affect the
answer. With regard to the latter, we simply tried to
model the plant as best we could using standard
practice of globalization --
MEMBER WALLIS: How do you put numbers on
that say RELAP 5 Internal Model Uncertainties?
MR. BESSETTE: Well, what we do is we go
through it like a PIRT process. We go through a PIRT
process to try to identify the most important physical
models and the most important boundary conditions that
would affect the things we're concerned with, which is
temperature and pressure. So, we tried to decide that
we have a proper list to do our sensitivity studies.
CHAIRMAN APOSTOLAKIS: Didn't you get some
expert opinion or recitation by (inaudible)
MEMBER WALLIS: That's what it is.
MR. BESSETTE: We did that. We had
available -- two prior PIRTs that were done for
perforized thermal shock -- one at the time of Yankee-
Rowe and one for H.B. Robinson.
CHAIRMAN APOSTOLAKIS: But the PIRTs do
not quantify uncertainty, do they?
MEMBER WALLIS: No.
MR. BESSETTE: The PIRTs, no. The PIRTs
simply list things that are thought to be most
important.
CHAIRMAN APOSTOLAKIS: So you went to
experts after that to get probabilities, to do what
Alan did, or you did not do that?
MR. BESSETTE: So, for each thing you think
is important, then you decide how much you want to
arrange it.
MEMBER WALLIS: So you arrange things like
correlations? Do you put a fudge factor on heat
transfer correlation or something?
MR. BESSETTE: That's right.
MEMBER WALLIS: Well, I can understand
doing that work, but I don't know how you get this
Internal Model Uncertainties in RELAP because of the
way it, itself, is structure. It has certain
conservation laws that no one ever seems to question
the way they're formulated. But they lead to
uncertainties because they are great simplifications
of what's actually happening.
MR. BESSETTE: Well --
MEMBER SHACK: One example might be this
break flow calculation. That's sort of one explicit
way you treated a model uncertainty, isn't it? You
got that one?
MR. BESSETTE: We lost you.
MR. BESSETTE: This is the list we came up
with. On the left-hand side, we have the
uncertainties -- the dominant uncertainties in terms
of the boundary conditions of the analysis. That's
things like break size, break location, decay heat,
the temperatures of the --
CHAIRMAN APOSTOLAKIS: So these were
treated --
MR. BESSETTE: These were each --
everything that you see on this list here was treated.
MEMBER WALLIS: But there's nothing here
about the code itself. Well, the right-hand side
related --
MEMBER WALLIS: If you put a heat transfer
coefficient, you put it into this big machine called
RELAP, there's nothing here about how uncertain RELAP
itself is because of the way it's built.
MR. BESSETTE: Well, you know, I think
there you're going back to the question of the
relationship between RELAP and reality -- things like
flow regime maps and conservation equations, which
rely on a much broader --
MEMBER WALLIS: But you can quantify that,
too, by comparison with date. That's what we make
people do at CSAU.
MR. HACKETT: I think Dr. Wallis is
exactly right. Jack Rosenthal was here at the last
meeting with the subcommittee, and I'm certainly no
expert in this area, but I think he assured us that
the RELAP -- I'm sure Dr. Wallis knows this -- has
been extensively benchmarked by experiments, and --
MEMBER WALLIS: I'm sorry, but that sort
of benchmarking usually consists of people looking at
data and saying, well, the line isn't too far from the
data. We're in a more sophisticated world now. We
actually want quantitative measures of those
uncertainties.
MR. HACKETT: That's correct.
MEMBER SHACK: One thing they did when
they made the presentation to the subcommittee was --
just take the break flow as one example. There are
various models for the break flow.
CHAIRMAN APOSTOLAKIS: Where is that?
What are you looking at?
MEMBER SHACK: It would be page 51 of the
big viewgraph package that you have.
CHAIRMAN APOSTOLAKIS: The big viewgraph,
from the subcommittee meeting.
MEMBER SHACK: From the subcommittee. And
so the different break flow models gave you different
flow rates. The way they handled that was essentially
to vary the break flow area. You know, you can either
have a different -- for a given break flow area, you
can have a different break flow model but instead,
because you can't do that all that handily, they took
one model and varied the break flow rate.
MEMBER WALLIS: I understand all that, but
there's this -- the assumption is that the machinery
of RELAP is deterministic and exact once you put these
various uncertainties and --
CHAIRMAN APOSTOLAKIS: But the question
is, if they do that -- they say they varied the break
area by 30 percent to account for model differences.
Was there a probability assigned, then, to the
different curves?
MR. KOLACZKOWSKI: George --
CHAIRMAN APOSTOLAKIS: That's the issue
here.
MR. KOLACZKOWSKI: The answer is, we
varied those by the amounts that we decided.
CHAIRMAN APOSTOLAKIS: Right.
MR. KOLACZKOWSKI: And then basically we
did a run -- well, suppose it's 30 percent greater;
suppose it's 30 percent less.
CHAIRMAN APOSTOLAKIS: Right.
MR. KOLACZKOWSKI: If we found that it
wasn't going to make much difference in the
temperature or pressure or heat transfer coefficient
response, in fact it was entirely dropped.
CHAIRMAN APOSTOLAKIS: Okay.
MR. KOLACZKOWSKI: Only if it was making
a big change, then we went through this more -- as you
say, a process of saying, okay, let's capture that
uncertainty, let's weight it by some expert opinion,
etc. And that was done on a few variables that seemed
to really matter.
CHAIRMAN APOSTOLAKIS: Okay. All right.
All right. But that's what --
MR. KOLACZKOWSKI: I think that one of the
things that Dave was trying to get at, which is fine,
is that the things that really matter are more when
the operator does the action; probably not
uncertainties in the actual RELAP code.
MEMBER WALLIS: These aren't internal
modeling uncertainties in RELAP, though.
MR. KOLACZKOWSKI: Sorry.
MR> WALLIS: These are not internal
modeling uncertainties in RELAP itself.
MR. KOLACZKOWSKI: No, although -- and
again, these are surrogates in some cases to try to
get at what those uncertainties might be.
CHAIRMAN APOSTOLAKIS: but you did try to
quantify some of the internal model uncertainties --
MR. KOLACZKOWSKI: Yes.
CHAIRMAN APOSTOLAKIS: -- even though the
operator actions dominate.
MR. KOLACZKOWSKI: Yes.
CHAIRMAN APOSTOLAKIS: Okay.
MR. BESSETTE: So we go through these
multi-step -- we go through the operator uncertainty
and --
CHAIRMAN APOSTOLAKIS: So the general
approach was, you know, identify -- not only here but
in the full project -- identify what's important and
then subjectively, through expert judgment and
dissertation, assign --
MR. KOLACZKOWSKI: Do the uncertainties on
the things that were important enough to do it on, and
not on everything.
MR. HACKETT: I think that one of the
things we mentioned at the subcommittee meeting that
we didn't mention at the start of this -- of course,
that treatment is not equal across the disciplines.
In some cases, we had the benefit of a lot more rigor
and a lot more data; in other cases, we didn't.
CHAIRMAN APOSTOLAKIS: Well, presumably
that's reflected on the uncertainties.
MR. HACKETT: Yes.
CHAIRMAN APOSTOLAKIS: okay. I think we
understand. At least I understand.
MEMBER WALLIS: Well, I don't understand,
though.
(Laughter.)
CHAIRMAN APOSTOLAKIS: All right. Ask
away.
MR. RANSOM: Just one quick question --
CHAIRMAN APOSTOLAKIS: Please identify
yourself.
MR. RANSOM: Victor Ransom.
When you plot the temperature versus time,
is that the code link temperature in the affected
loop?
MR. HACKETT: It should be the downcomer
temperature.
MR. RANSOM: Well, then I'm wondering how
you have four loops presumably mixing together, and
then in RELAP 5 these are all mixed into a one-
dimensional downcomer. That wouldn't give you the
worst-case temperature that you might see at the
entrance of the code link to the downcomer. I'm
wondering, how did you account for that effect?
MR. BESSETTE: Well, we -- our focus is on
the temperature in the downcomer in the region
adjacent to the core. What we found was that by the
time you get down there, things are pretty well mixed.
We don't see a lot of circumf- --
MEMBER WALLIS: Azirmuthal.
MR. BESSETTE: -- azirmuthal variation.
MR. RANSOM: That would surprise me. I
would think you'd find the maximum temperature and
overcooling near the entrance of the *code link that's
affected.
MR. BESSETTE: Yes, but we don't care about
that so much. It's not that temperature we're focused
on. Our focus is on the downcomer temperature.
MEMBER SHACK: It's the worst temperature,
but the vessel is tough up there, so you have to look
where the vessel is brittle.
MR. RANSOM: Okay.
MR. KOLACZKOWSKI: You have to go down
significantly lower, where the fluence is high and the
embrittlement is happening.
MR. RANSOM: So you attempt to use the
mass mean temperature, I guess, at that point, huh, in
the downcomer?
MR. BESSETTE: Yes. We supply a single
temperature boundary condition to the fracture code.
MEMBER WALLIS: Do you give it
uncertainties?
MR. BESSETTE: That's the -- that was the
uncertainties --
MEMBER WALLIS: But you don't give
uncertainties based on sort of input parameters, not
the way the code works itself?
MR. BESSETTE: Well, we try to do both. We
try to take into account both things like temperatures
of the safety ejection water.
MEMBER WALLIS: Well, there's nothing
there that says that modeling of conservation laws in
RELAP contributes uncertainty.
MR. BESSETTE: Other than when the
treatment of the --
MEMBER WALLIS: That leads to
uncertainties.
MR. BESSETTE: Yes. So we did consider
that.
MEMBER WALLIS: Well, you say they were
validated. That's not a measure of uncertainty.
MR. BESSETTE: What we validated is when
the assumption -- did not contribute significantly to
--
MEMBER WALLIS: Well, you've got to say
how much uncertainty it contributes, and the
validation should end up in a measure of uncertainty,
which is quantitative, not --
MR. KOLACZKOWSKI: Dave, go to the slide
that has six, seven and eight on it -- Uncertainty,
two out of three.
I mean, what they did was, they ran a
whole lot of cases, which is what's trying to be
illustrated on the left-hand side. Different break
flows, different -- summer or winter -- whatever,
which, granted, is not directly getting at some of the
code on certainties but is trying, in a surrogate way,
address some of that.
MEMBER WALLIS: I understand what's you've
done, but if you'd run a different code, you know, a
different structure, a different way of handling
conservation laws, one-dimensionality -- on these,
you've got different answers, irrespective of all this
stuff. That's sort of what I'm getting at. It's
something which I don't think the Agency knows how to
handle yet.
MEMBER KRESS: One of the things they do,
Graham, is they take situations which they know
analytical solutions, and then they run the code to
see how well it does it. And it gives you a little
bit of confidence that the code, as a structure put
together, is doing something correctly. That doesn't
lead to quantification of what you're talking about,
but it gives you a little more confidence, if they do
enough of that with enough known results.
But I agree with you. If you used a
different code to do all this, you'd get a different
result, but it may not be that much different.
MR. BESSETTE: Of these 150 calculations,
we did run five -- the selected five of them were ran
on TRAC. Well, you can say, well, TRAC is not a
different code, but we got similar results.
MR. HACKETT: I guess I'd interject at
this point and --
CHAIRMAN APOSTOLAKIS: Regarding the
transients you use, I saw LOCAs and steam line break
becoming -- I know the French have had a concern about
cooldown in a low-power and shutdown condition, where
you have repressurization. And they. in fact, are
making planned changes for that. Have you looked at
those kind of scenarios?
MR. KOLACZKOWSKI: Yes. And in fact,
among the low power, that same SRV re-closure event
is, again, a pretty bad event.
CHAIRMAN APOSTOLAKIS: Okay.
MR. KOLACZKOWSKI: Of course, you're being
helped by the fact that the portion of the year that
you're at low power is a lot less, so it tends to
drive the frequencies down. But nevertheless, yes,
it's included in the model, and in fact it's among our
results.
MR. HACKETT: I guess what I was going to
say is, I guess we'd need, at this point, maybe to
take an action that -- we're suffering, obviously,
here from trying to condense an awful lot into too
short a time frame. It's probably overly ambitious.
I apologize for that. I think we need to come back
and at least address the thermal hydraulics area in
greater detail in a separate meeting, if we don't want
to hold people over who have other obligations.
MEMBER FORD: I agree. The purpose of
this meeting was to just get a feeling as to whether
you're going the right direction and whether there are
any fundamental errors in the way that you're
proceeding. And I've got a draft letter going on
those lines.
If we do have outstanding question that
seem to be relating on the uncertainty aspect, maybe
we've got a different meeting to address those
specific issues.
MR. HACKETT: We'll be glad to so that.
MEMBER FORD: Do you have a summary slide
to --
MR. HACKETT: We have a plain-language
summary.
CHAIRMAN APOSTOLAKIS: Oh, the plan
language.
MR. HACKETT: Or the lack of plain
language.
CHAIRMAN APOSTOLAKIS: We've seen that.
(Laughter.)
MR. HACKETT: You've seen that. In case
we had had an emergency in the day and a half, we had
this prepared. To go to Dr. Rosen's point, where is
this all going, where we're hoping this is going at
this point is, we have a lot of work to do, is the
bottom line. We have three more plants to analyze, an
awful lot of runs of a lot of codes to do, a lot of
integration.
I think that one of the questions came up
is, once we get all these four plants, we need to make
the sort of quantum jump to generalizing that result
to the whole PWR fleet. So we have a fair bit of work
ahead of us, I think it's fair to say. and probably
at the -- I'd be willing to go so far as to say, end
of the summer time frame, we might be in a position to
have a lot of that done and be ready to speak to that
in a more generalized sense. In the meantime, we're
happy to come back and hit any one of these in more
detail.
MEMBER FORD: If I could make a suggestion
--
MEMBER ROSEN: You didn't answer my
question, exactly. My question was, when you're done
with all of that, what do you do? Do you propose a
Notice of Proposed Rulemaking to amend 50.61.
CHAIRMAN APOSTOLAKIS: Your microphone --
MEMBER ROSEN: My question was, at the end
of all that, do you go to a proposed rulemaking to
address revisions to 50.61?
MR. HACKETT: Yes. I'm sorry. I started
rambling there and didn't close that loop. The
schedule that we're proposing right now, that we're
still in some debate with amongst the management, is
that we would have a technical product that's the
technical basis that's the result of all this work
we're talking to you about here in about the November
time frame, at which point we would that hand that
product off to NRR.
NRR has already budgeted resources to
initiate rulemaking on a revised 50.61. We would
continue some of the technical work to finalize that
in parallel, and we would be working closely with NRR.
That, in and of itself, as you know, is probably two-
to three-year process in and of itself.
MEMBER FORD: If I could make a
suggestion. In the shorter term -- i.e., the next
three months --
MEMBER POWERS: Peter, you have to use
your microphone, just like everybody else.
MEMBER FORD: Aye, but he understands me.
He comes from the right country.
MEMBER POWERS: And you come from the
North, which is considered somewhat less than
civilized.
MEMBER FORD: If I could make a
suggestion. I think the subcommittee -- we had a full
one and a half days and we went into a lot of detail;
maybe not quite so much detail in some of the thermal
hydraulics and the PRA. But I think that we've got
enough of a background to write an informative letter
for EDO tonight, or in the next couple of days.
But I do suggest that within the next two
months, we do have a subcommittee meeting with the PRA
and thermal hydraulics people to go over the
outstanding questions. I don't think, quite honestly,
they're show stoppers, but I do think that it could
give us some useful advice.
MEMBER WALLIS: Could I say something
about thermal hydraulics here?
MEMBER FORD: Yes.
MEMBER WALLIS: I think that these codes
like RELAP, historically, have been validated by
looking how assumption and so on affect figures of
merit for court damage, such as peak clad temperature.
And we had all these justifications that RELAP's okay
because you can change all these assumptions and peak
clad temperature doesn't change.
But we're not talking about peak clad
temperature. We're talking about temperature in a
downcomer, and I don't quite know how well RELAP's
been validated for that, so that's the sort of thing
I will ask at the next meeting when we talk about
thermal hydraulics.
MEMBER FORD: Okay. I'll pass it back to
you, George.
CHAIRMAN APOSTOLAKIS: Thank you, Peter.
When I see a slide like this, I'm
wondering if we're doing this again 20 years from now.
Where do you think your successors will say, this and
this and this is better or worse than we thought it
was? "Thought" meaning what you think now. Where is
your major uncertainty here that might be revised? I
mean, your colleagues of 20 years ago were doing also
a good job at the time, right?
MR. HACKETT: Actually, I guess maybe some
combination of us can answer that. I can speak to a
few pieces at least, and there I'll give credit to our
predecessors. I think 20 years ago, it was recognized
that the flaw density and distribution was a major
player in this thing, with the work that came out of
the U.K. with the Marshall Committee. They did a
tremendous body of work on that. They did as good as
they could at the time.
They did as good as they could at the
time. We've had a lot of resources brought to bear on
that, since -- where we've cut up reactor vessel
welds, for instance -- and we've seen what flaws are
there that's helped refine this thing and make it more
accurate.
But that's been a huge impact on this
project because previously we were assuming from the
Marshall Distribution there were some very large
flaws, and we didn't know what else to do at the time
in fracture mechanics, so we put them all on the
surface. A huge conservatism in the project -- that's
just --
CHAIRMAN APOSTOLAKIS: But they knew that
it was a major concern?
MR. HACKETT: They did. That was one that
I think that the folks sitting around back in the
early '80s would have said -- in fact, when I read the
Marshall Report, they said, don't use this for very
long. Unfortunately, we did. We didn't have the
wherewithal to get the research done in as timely a
fashion as we like.
But one of their cautions was, you know,
we're pretty far out on a limb here, so don't be
utilizing this ten years from now, and we were. So,
we're only getting to that point now, so we're late
but I think we've finally gotten there in the right
way, and that's just one example. I think there are
many in the --
CHAIRMAN APOSTOLAKIS: But the cooldown
rates were the same. I mean, everybody knew 20 years
ago that they were excessive and I don't know why, in
fact, it wasn't questioned further. I mean, there was
no operator action to intervene. there was nothing
that --
VICE CHAIRMAN BONACA: -- going forever
until simply the system couldn't cool down anymore.
MR. HACKETT: Good point, too. The other
ones we're familiar with, at least speaking in the PFM
area, the material -- we've made some assumptions that
the material I think frankly was probably a lot more
embrittled than it really is, in general terms.
There were some conservatisms that came
out of the foundations of ASME code, Sections 3 and
11, and I think we're find out that they were indeed
conservative, that these materials are, on balance,
significantly tougher than we've given them credit
for, and we now have the science and the refinements
in fracture mechanics to bring some more accuracy to
that area. And that's not the only area.
MEMBER POWERS: And it seems to me that
what you're saying -- and correct me if I'm wrong --
is that we've researched these vessel materials and
their irradiation to the point that we have a
satisfactory understanding of that material?
MR. HACKETT: I think that's always a good
question because I don't want to shoot our research
program in the foot.
(Laughter.)
MR. HACKETT: We have been challenged --
Dr. Powers knows this, I'm sure -- we have been
challenged as to why we're still looking at vessel
materials 30 years after we initiated the project at
Oak Ridge. And the answer to that is, of course, it's
a dynamic situation. We didn't know at the time that
project was initiated what a bad actor copper was in
welds. We didn't know what synergisms of manganese
and nickel. So, there are things that have changed
over time and valid reasons to continue to work.
I think, you know, for those of you
familiar with this technology S curve, we're way up on
the curve when it comes to understanding the material
behavior, if not on a plateau, very close to it. So,
we are, I think, at some point reaching a level of
diminishing --
MEMBER POWERS: A point of diminishing
returns where maybe that -- resources could be
redirected toward more 00
MEMBER ROSEN: This is a conclusion about
the heavy section steel technology program?
(Laughter.)
MEMBER KRESS: All right.
MEMBER ROSEN: Or are you moving towards
this?
MEMBER POWERS: Something -- redirect
those resources to something more useful, like --
MEMBER ROSEN: ATHENA --
MEMBER POWERS: understanding ATHENA --
right.
(Laughter.)
MR. HACKETT: That said, I feel it
incumbent on my position to make a plug for the HSST
program, or should I say what remains of it, in
reflecting that in the RES budget. The HSST program
funding has declined significantly, I think exactly
for that reason.
At one point, the committee will probably
remember that was a major multi-million dollar effort
where we were doing a lot of testing of large-scale
specimens. The effort now is on the order of $4- to
$500,000 per year that's mainly aimed at analytical
and advanced fracture mechanics. so, I think that has
been reflected in the RES budget, and in our budget
assumptions. It's a lesser effort than it used to be.
MEMBER KRESS: Yeah, but you didn't really
answer George's questions. What you said that I heard
was that these are the things in the original --
CHAIRMAN APOSTOLAKIS: Right.
MEMBER KRESS: -- that we knew were bad,
and now we've improved that. Twenty years from now,
what is going to be the theme?
MR. KIRK: I think you'll probably many of
the same things. For example, we've, you know --
MEMBER KRESS: You can improve these
further.
MR. KIRK: Further improve -- I mean, Ed
correctly pointed out that we've gotten significant
benefit from our advanced understanding of fracture
mechanics.
Having said that, you know, I know there's
a factor in the simulation right now, where we've got
an uncertainty programmed in -- I can show you the
equation -- of 180 degree Fahrenheit on the transition
temperature, that if you made a more direct
measurement of the vessel, which is -- it's possible
to do it today but it's simply not practical to
require it of the licensees.
But if you made a more direct measurement
and had the stated knowledge that you could ensure
that that more direct measurement adequately
represents the material in the vessel, you could take
that 180-degree uncertainty and turn it off.
So, there's a fairly substantial and I
think quantifiable benefit that one could derive from
going from what is certainly a more state of knowledge
in fracture mechanics, but still predominantly
empirically based to one that's more driven by the
physics, so that we know the origin of these
uncertainties. And I'm sure there are examples in
other areas. But I suspect that the answer is, again,
you know, more of the same.
MR. HACKETT: I think also, to elaborate
on that, based on these advancements in all three of
these areas, one of the things to look forward to with
the next generation with the advanced plants is, I
don't think we need a PTS rule for those plants. So
probably -- I mean, and there are our colleagues --
MEMBER KRESS: They're going to control
the chemistry in the vessel and stop the fluids.
MR. HACKETT: -- our colleagues in the
Navy who have been doing this and had the advantage to
build new machines, they know how to do it.
CHAIRMAN APOSTOLAKIS: Are you requesting
a letter?
MR. HACKETT: Yes. I think what we talked
about with Dr. Ford was a letter at least indicating
generally, you know, how the committee felt about the
key areas would be indicated.
CHAIRMAN APOSTOLAKIS: Okay. Any other
comments from the members?
(No response.)
CHAIRMAN APOSTOLAKIS: Thank you very
much, gentlemen.
MR. HACKETT: Thank you.
MR. KIRK: Thank you.
CHAIRMAN APOSTOLAKIS: I suggest we take
a short break this time and then come back and discuss
a few things of great importance. How about 5:30,
we'll be back.
(Off the record at 5:21 p.m.)
Page Last Reviewed/Updated Monday, July 18, 2016