472nd Advisory Committee on Reactor Safeguards - May 11, 2000
UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
***
MEETING: 472ND ADVISORY COMMITTEE ON
REACTOR SAFEGUARDS
U.S. NRC
Two White Flint North, Room T2-B3
11545 Rockville Pike
Rockville, MD
Thursday, May 11, 2000
The committee met, pursuant to notice, at 8:30
a.m.
MEMBERS PRESENT:
DANA A. POWERS, Chairman
GEORGE APOSTOLAKIS, Vice-Chairman
JOHN J. BARTON, Member
MARIO V. BONACA, Member
THOMAS S. KRESS, Member
ROBERT L. SEALE, Member
WILLIAM J. SHACK, Member
JOHN D. SIEBER, Member
ROBERT E. UHRIG, Member
GRAHAM B. WALLIS, Member. C O N T E N T S
ATTACHMENT PAGE
SCHEDULE AND OUTLINE FOR DISCUSSION 3
OPENING STATEMENT 3
RISK-INFORMED TECHNICAL SPECIFICATIONS 6
TSTF 358 - MISSED SURVEILLANCES 13
INITIATIVE 2 - MISSED SURVEILLANCES 13
POTENTIAL REVISIONS TO PTS SCREENING
CRITERION 46
GUIDELINES FOR USING RISK INFORMATON REGULATORY
DECISIONMAKING 132
SRP DEVELOPMENT FOR T/H CODE REVIEWS 203
DRAFT REGULATORY GUIDE DG-1096 204
. P R O C E E D I N G S
[8:30 a.m.]
DR. POWERS: The meeting will come to order. This
is the first day of the 472nd meeting of the Advisory
Committee on Reactor Safeguards. At today's meeting, the
committee will consider the following initiatives related to
risk-informed technical specifications: potential revisions
to the pressurized thermal shock acceptance criteria;
proposed revision to Regulatory Guide 1.174, an approach for
using probabilistic risk assessment in risk-informed
decisions on plant-specific changes to the licensing basis;
proposed Regulatory Guide and Standard Review Plan sections
associated with NRC code reviews; proposed ACRS reports -- I
am anxious in our work on the reports to get through the
three letters listed on the board today.
This meeting is being conducted in accordance with
provisions of the Federal Advisory Committee Act. Dr. John
T. Larkins is the designated Federal official for the
initial portion of the meeting. We have received no written
statements or requests for time to make oral statements for
members of the public regarding today's session. A
transcript of portions of the meeting is being kept, and it
is requested that speakers use one of the microphones;
identify themselves and speak with sufficient clarity and
volume so they can be readily heard.
We begin with some items of current interest.
First, I will note that Dr. Kress and Professor Seale have
been attending the CSUP meeting, and some members may want
to interrogate them closely on any revelations concerning
severe accidents that came from that meeting.
Next, I wanted to acquaint members with some
additions to our staff. Joining our staff engineers is Meg
Weston. Meg has a B.S. degree in physics and 20 years of
experience in the NRC. She has recently been working with
Sam Cullins. We've looked at her impressive resume and
maybe during one of the breaks, she'd even talk to me about
the impressive recommendations she has. In particular, she
has experience in the area of electronics, and so, Bob, you
may find her a very useful addition to your team working on
digital electronics.
Another addition is Jenny Gallo, in the operations
support branch. Jenny? Jenny has a bachelor of science
degree in sociology from Northeastern University. She has
promised me she is not using this time to develop research
for a master's degree.
[Laughter.]
DR. POWERS: She'll be working on budgetary
activities, trying to keep us square with the law, right,
Jenny?
DR. SEALE: Is she writing a tell-all book?
[Laughter.]
DR. POWERS: Each of the members should have
detailed resumes on these individuals. I encourage you to
introduce yourself to them, because they are part of our
permanent team. So welcome aboard to both of you.
Do members have any comments they would care to
make before we open the session today?
[No response.]
DR. POWERS: Seeing none, then, we'll turn to our
first item, which are initiatives related to risk-informed
technical specifications, and Jack, I think you're going to
start us on this subject.
DR. SIEBER: Yes, sir; thank you, Mr. Chairman.
Two weeks ago today, actually, there was a joint meeting of
the subcommittees on reliability and probabilistic risk
assessment and plant operations to discuss two of seven
initiatives in risk-informing the topic of specifications.
Those two items or two initiatives were initiatives two and
three, one of which involved missed technical surveillances
and how those would be treated and also the mode preference
for action statements when plant maneuvering is required
because of LCOs.
At the conclusion of the subcommittee meeting,
individual members, of which seven of the 10 members of the
full committee were there to provide their views and
comments related to the reservations, and the staff is
prepared to address the more important of those today that I
think will help us all understand basically how these items
will be resolved.
So without further discussion, I'd like to
introduce Mr. Robert Denning technical specifications branch
of the Office of Nuclear Reactor Regulatory who will
introduce the speakers and conduct this session.
MR. DENNIG: Thank you, Mr. Sieber.
We're pleased to be back here again before the
ACRS to discuss this important subject. Today's
presentation, the way we structured it, we're going to have
Mr. Biff Bradley from NEI present the industry's
presentation and discussion on initiatives two and three and
any feedback they have on our last meeting, and the staff
will follow. And with me today from the staff is Scott
Newberry, deputy director of my division; Rich Barrett,
branch chief from probabilistic safety assessment branch;
Mark Reinhart, who is a section chief in the probabilistic
safety assessment branch and Nanette Gilles, who will also
help make some presentations.
Before I turn things over to Mr. Bradley, Scott or
Rich, do you have anything that you'd like to say?
[No response.]
MR. DENNIG: Okay; it looks like we can just go
ahead and get into the presentation, Biff?
DR. BRADLEY: Thank you; I'm Biff Bradley of NEI.
I actually just found out I was going first, so I didn't
have an opportunity to give you copies of my material, but I
will pass it out as quickly as I can.
The industry views these initiatives on tech specs
as a first step in an important evolution -- actually, the
next step in an important evolution in plant configuration
control, and this is a slide I used at the subcommittee
meetings, but this is just a brief background of the history
of how we've dealt with plant configuration, starting out
with custom tech specs; moving on to standard tech specs;
NUMARC 91-06, which provides control of shutdown outage
maneuvering and improves standard tech specs which will come
about and are widely adopted by the industry now over the
last several years.
And, of course, in the last couple of years, we've
had some success finally with risk-informed applications on
AOTs, and a fair number of those have actually made their
way through the system in the last year or so.
The next and important step is coming up in the
fall of this year. The Commission just approved issuance of
the maintenance rule (a)(4) implementation guidance with
180-day implementation period, so in around November of this
year, the industry will have a new regulatory requirement
for a risk-informed plant configuration control approach
through 50.65(a)(4). And one of the conditions that the
staff has put on the reg guide that endorses the industry's
guidance on (a)(4) is that you still have to meet your
technical specifications. So we're left now with both a
tech spec and a maintenance rule (a)(4) provision
controlling plant configuration.
DR. APOSTOLAKIS: Biff?
DR. BRADLEY: Yes.
DR. APOSTOLAKIS: You have custom tech specs and
standard tech specs? Could you explain a little bit?
DR. BRADLEY: Sure; well, I think this really
predates my involvement in the industry, but back in the
early licensing days, plants pretty much -- every plant had
their own unique tech specs, starting off with some very
small tech specs for the early units, and there was an
evolution -- and there really was no standardization. Every
plant -- I used to work at Plant Hatch back in the early
eighties. We had custom tech specs on unit one and standard
tech specs on unit two; two identical units and, you know,
one set of operators, so that made life interesting.
But the standard tech specs evolved -- I think
there was a recognition as more units got licensed that the
tech specs ought to be standardized, so it started, I guess,
in the early to mideighties, standard tech specs came about.
And then, the ITS improved standard tech specs came about as
a result of a 50.36 revision that put the four criteria in
and then allowed tech specs to be scoped accordingly to
those four criteria, which actually resulted in some
reduction in scoping.
DR. APOSTOLAKIS: But we still have custom tech
specs.
DR. BRADLEY: Yes; there are some -- a fair number
of plants -- a small number, I should say -- that still have
custom tech specs.
DR. APOSTOLAKIS: But isn't this contrary to the
risk-informed regulatory approach? I mean, you have allowed
outer -- and all of that, and these are specified for all
plants without regard to how many diesels you have or, you
know, other things?
DR. BRADLEY: I wouldn't say so much -- it is now
contrary to the risk-informed approach. I think at the time
it was evolved, it was, you know, deterministic --
DR. APOSTOLAKIS: Sure.
DR. BRADLEY: -- and it was all we had at the
time. At the time the custom tech specs first came out,
PRAs were just starting to be developed. That is correct; I
think maybe my second slide -- the scope and the process of
(a)(4) is a risk-informed process. I'm just getting at what
you're talking about, looking at all of the equipment in the
plant that may be out of service and looking at the risk
impact.
DR. APOSTOLAKIS: So we are going back, then, to
customized tech specs but through the risk route.
DR. BRADLEY: Yes; well, I'm not suggesting we go
back to custom tech specs. What we want to suggest is that
these initiatives are the first step in a process that will
bring the current tech specs into -- so that they are --
comport with (a)(4), and there's no longer a potential to
have a dichotomy between what (a)(4) tells you to do and
what the tech specs might tell you to do.
With (a)(4) becoming a requirement later this year
and, of course, obviously, you still have to meet your tech
specs, there is potential for a conflict and, you know,
(a)(4), because of its risk-informed scope, it's looking at
the conditions of much equipment that's not even covered by
tech specs.
DR. APOSTOLAKIS: Another question that puzzles me
a little bit: a lot of these things depend on risk
assessments, don't they, and risk information?
DR. BRADLEY: Yes, yes.
DR. APOSTOLAKIS: And they will be mandatory,
correct?
DR. BRADLEY: That is correct.
DR. APOSTOLAKIS: So why are we allowed to do this
here, but when the staff wants to ask for risk information,
immediately, we have legal problems? What's the difference?
DR. BRADLEY: I don't think that when we get to
that issue later this morning or this afternoon that we
would necessarily suggest that there are significant legal
problems with what the staff is proposing on the request for
risk information. That's my later presentation.
DR. APOSTOLAKIS: I understand.
DR. BRADLEY: I don't want to go into that now.
But I think, you know, that one of our first bullets, which
you'll see later, is we agree that there are instances where
the staff should look at risk issues even though you're
meeting the deterministic licensing basis so --
DR. APOSTOLAKIS: Yes, but the thing that strikes
me is that -- I mean, we are talking about another subject
now, but the staff had to invoke issues of adequate
protection to be allowed to ask for risk information there,
and yet here, both the industry and the staff seem to be
working together to develop regulations that are
risk-informed. So I'm trying to understand what the
difference is. Maybe we can discuss it later.
DR. BRADLEY: I guess my simplistic view of it is
there is no conflict or difference in the approach we're
suggesting. There -- maybe we can talk about that in more
detail when we get to --
DR. APOSTOLAKIS: Fine.
DR. BRADLEY: -- that consideration.
DR. APOSTOLAKIS: But I'm just interested in my --
DR. BRADLEY: Yes.
DR. APOSTOLAKIS: -- questions.
DR. BRADLEY: Yes; we're not suggesting that the
staff should not be raising those risk issues. As we'll
talk, we think there does need to be a clear definition of
the threshold, and it may not be the same as some of the
other thresholds we've been using in, say, Reg Guide 1.174,
but I don't see a conflict there.
DR. APOSTOLAKIS: Okay.
DR. BRADLEY: Okay?
You got me off track; I've got to refocus here.
[Laughter.]
DR. POWERS: It's a plot.
[Laughter.]
DR. BRADLEY: Very clever. We are --
DR. APOSTOLAKIS: Nothing less is expected from
us.
DR. BRADLEY: Of course.
We are -- these two initiatives that we're going
to discuss today really represent the first efforts, I
think, to use risk insights and adjust principles and
technical specifications that go beyond just AOTs, which is
typically what we've looked at before, but when you look at
tech specs, you have action requirements, mode change
restrictions, issues like missed surveillances, 303, 304.
All of those are really amenable to risk-information, and
really, all of those types of situations will have to be
addressed through (a)(4), so what we're proposing to do
today is to make the first step toward bringing the tech
specs in (a)(4) such that they'll give you the same result.
And this is the first step in a lengthy process.
Ultimately, you know, we'd like to see the tech specs in
(a)(4) be -- I don't know if they'll ever be equivalent but
at least remove any conflicts.
For instance, today, you may have a seven-day AOT
in tech specs; (a)(4), because of other things you have out
of service that may not even be covered in tech specs will
tell you a four-day AOT is probably all you should do before
you hit the risk thresholds. That's the example of a
conflict, and that's a regulatory requirement where you're
pretty much going to have to meet (a)(4). And then, of
course, you're going to have the inverse, where the tech
specs will allow you seven days, but in-force says you can
keep the equipment out of service for much longer.
And so, those are the kinds of conflicts that
we're going to be working, and this is the first step to get
there.
We'll talk specifically about the first initiative
on missed surveillances. The requirement is that enter the
LCO, which is typically a shut-down requirement. If you've
missed a surveillance, and it can't be performed within 24
hours plus the existing AOT; what we're proposing is that
the missed surveillance, when you discover that you've
missed a surveillance, could be performed at the next
reasonable opportunity up to the total surveillance
interval, and to do that would require a risk evaluation.
There was considerable discussion at the
subcommittee meeting about the potential for mischief or
misuse of this provision, and I wanted to reemphasize today
that there is no intent whatsoever to establish a provision
in tech specs that would allow surveillances to be willfully
missed. This is intended to address situations where you
discover after the fact or the surveillance has been missed,
and whereas tech specs right now would tell you to shut down
and go through a plant transient, with the vast majority of
the equipment in the plant, the risk impact of delaying a
surveillance is dwarfed by the risk impact of a plant
transient to shut it down and take it back up to power to
perform a surveillance.
Another thing I would mention is that the vast
majority of surveillances in tech specs can be performed
online, and really, this provision wouldn't even come into
play, but there are some tech spec surveillances where you
would have to change the plant condition in a containment,
what have you, to make a surveillance, and those are the
scope of things that would be impacted here.
DR. POWERS: Could you explain to me how I do a
risk evaluation on a missed surveillance?
DR. BRADLEY: Yes; I'll get to that in just a
moment. I guess I'll do that now and come back. We've made
some changes to the -- we're in the process of making some
changes to this tech spec on the TSTF 3.58, which is
initiative two as a result of some of the discussions we had
with the subcommittee as well as the meeting we had with the
staff and industry several weeks ago, and the way that you
would do what you're asking, Dr. Powers, is that basically,
the missed surveillance is like any other emergent condition
that can be treated under 50.65(a)(4).
50.65(a)(4) provides most of the foundation that
you need to address this issue, and it contains provisions
on temporary and aggregate risk guidelines; that is, as well
as the risk management actions that you could consider based
on the result of that risk evaluation and the temporary and
aggregate impacts and also contains a discussion of how you
control additional emergent work that's going -- that may
emerge once you've performed the assessment. Now, how do
you actually do the risk evaluation?
It's fairly simple for a missed surveillance. You
can screen using RAW basically on the assumption that the
equipment is available, and for a large amount of the
equipment in the plant with this type of screen, you would
still be able to demonstrate a relatively small risk impact,
generally within the guidelines of Reg Guide 1.182, which is
the (a)(4) guidance on a temporary CDP and LERF.
DR. POWERS: So I go in, and I say I want to do
this and this for surveillance, and I go to the RAW value.
And I make my decision based on that.
DR. BRADLEY: You could make the decision based on
the RAW value. It's a screen; I mean, clearly, in most
cases, missing a surveillance isn't equivalent to having the
equipment unavailable, but there's a simplified approach for
it, and for the plants that are implementing this in, like
Southern Cal Edison which would look at this and run the PRA
models, for most of their components and tech specs, they
can just use a RAW approach.
And the way this would work is if you would have
to make the assumption that that equipment is unavailable
and roll that into your (a)(4) program, okay? So, from now
on, all the plant configuration evolutions you're
undertaking would be with the assumption that that equipment
is unavailable, and then, you'd have to factor in the impact
of everything else you may be taking out of service. So if
you're going to take out the other train of that safety
function, it probably would change that plan.
Now, if you want to go beyond the screening
approach and do something that uses the model -- maybe
you've failed the screen, but you still believe that the
condition is not significant, then, you can adjust the
failure rate in the PRA based on the fact that you've missed
the surveillance and run the modeling and get a --
DR. POWERS: I guess what I'm interested in is if
you screen something out, and there have been any
precipitous surveillances that have been missed are going to
be ones that are really going to have really little impact.
DR. BRADLEY: Right; that is correct.
DR. POWERS: And you're going to say, okay, I
missed this thing, and based on my PRA, what confidence
level do you have to have on this?
DR. BRADLEY: You mean quantitatively, what
confidence level?
DR. POWERS: No, I'm looking, you know, am I 99
percent certain that it's below this screening criteria? Or
am I 50 percent certain? With that kind of quantitative
accuracy.
DR. BRADLEY: I guess -- and I can't really answer
that question. We generally view this for the vast majority
of surveillances you can miss as a fairly simple risk
assessment.
DR. POWERS: What you're saying is that -- I think
what you're saying --
DR. BRADLEY: Yes.
DR. POWERS: -- is that most of these
surveillances are going to have such a low -- that it's
going to be able to say it doesn't matter squat.
DR. KRESS: It raises another question, which is
why are they in a surveillance program in the first place.
DR. POWERS: Of course.
[Laughter.]
DR. BRADLEY: Yes, and that's why initiative five
of this effort, which we will I am sure be discussing at
some point later this year looks at optimizing the
surveillances that are in there now.
DR. APOSTOLAKIS: Is it possible that if this is
allowed, at some point, there will be a de facto extension
of the interval between surveillances? In other words, you
know, for components whose RAW is very low, or the failure
rate changes do not affect anything?
DR. BRADLEY: That is not the intent. Whether
it's possible or not, I guess, is a theoretical question,
but I don't believe that would be the case because of the --
and that is that de facto, making this a de facto mechanism
for surveillances would result in routinely missing
surveillances, and as I mentioned earlier, that's not the
intent.
I think when we move to initiative five and look
at optimizing intervals as well as the actual tests
themselves, that is the mechanism for a, you know, de facto
permanent kind of change to the approach, and that's not
what we're trying to do here. This is just an initial step
in that direction.
DR. APOSTOLAKIS: Is the number of missed
surveillances part of the new oversight program?
DR. BRADLEY: Yes; the corrective action program
will track that, and the corrective action program is one of
the major elements of the new oversight process, and if you
miss a surveillance, you would be under the CAP program. I
think that is an important element of what we're proposing
to do here.
Personally, I don't think that this change would
cause any increase in the number of missed surveillances.
It's just to allow us to have a smarter approach when we do
discover that we've missed one.
DR. KRESS: When you miss a surveillance and go
ahead with operation to the next surveillance, you've
essentially handed what I would call a temporary spike to
the baseline risk, and this committee has advocated in the
past that there should be some sort of criteria on the spike
that's allowed and with how to deal with some interval over
time.
Are you advocating some sort of criteria on this?
DR. BRADLEY: Essentially, the answer to that is
yes. As I mentioned, the way this would work, it would be
rolled into your (a)(4) program, and as you remember from
reviewing the (a)(4) guidance, we have guidelines in (a)(4)
on the spike, and that's a 10-3 CDF. We have guidance on
the CDP, which is a 10-6. And then, we have guidance -- we
don't have quantitative guidance, but there's a discussion
in there of the need to track and control the aggregate risk
impact and look at this in light of where you've been
previously.
So I think the answer to your question is yes.
That's one of the reasons we advocate directly referencing
1.182 in this change.
DR. KRESS: So there is a -- existing now in the
rules, there is a spike criterion, in essence. Do you know
what the technical basis for that is, or would you explain
it to us?
DR. BRADLEY: Well, as you know, that's a subject
that's been discussed at great length for a number of years,
and what we used was out of the EPRI PSA applications
guideline, and that's the best -- we picked the best number
that we knew based on all of the information that we had,
and the applications guideline, I think, is what we chose,
because that's the best information we have. Beyond that,
it's pragmatic in that it allows you to -- it's typically
equivalent to taking something like a diesel generator out
for several days. I mean, that's typically what's going to
give you that kind of CDP. So it's sort of a bounding thing
for the single most significant component in the plant.
That's, you know, a loose technical basis for it.
Yes?
DR. SIEBER: A quick question that's two slides
back, the second bullet, where it says change failure rates
to address missed surveillance. Assuming that the item that
was to have been surveilled is actually in your PSA, which I
think a lot are not --
DR. BRADLEY: Right.
DR. SIEBER: -- how are you going to change the
failure rates?
DR. BRADLEY: If it's not in the PSA?
DR. SIEBER: If it is in the PSA. If it's not --
DR. BRADLEY: Right.
DR. SIEBER: -- there's probably no raw value.
DR. BRADLEY: We are not prescriptive in the tech
spec itself about how you would do that. I think that would
be a function of the specific component and the surveillance
and what you know about its history. Some components -- for
instance, you have a raw water system or something where,
you know, there is a degradation over time, you have to take
those things into account. But we're not -- there is no
single number that we're proposing. I think typically, you
might double the failure rate, you know.
DR. APOSTOLAKIS: Actually, I'm not sure that you
should do that. I mean, you would take the unavailability
that's calculated using the standard interval between
surveillances, and if you missed one, then, you'd double
that.
DR. BRADLEY: Right.
DR. APOSTOLAKIS: But the failure rate really
shouldn't change.
DR. SIEBER: Well, it's not clear to me that
that's the correct solution either, but having it as sort of
an ad hoc thing where, well, you have to consider this, and
you have to consider that, and to me, that seems a really
vague technical basis for doing that. On the other hand,
the decision as to how long you can go or whether you should
shut down or take some other action because of the missed
surveillance depends on the robustness of how you treat this
issue, and it just seems to me that, you know, it's not
there right now.
DR. KRESS: I think George is right, though. To
try to develop some sort of correlation for the missed
surveillance versus the change in reliability just is not
going to be able to do that. So you just work on the
surveillance time, and whatever the change in the
surveillance time, you put that into your thing. I think
that's the only practical way you could approach it. But
that would tend to make it -- that would tend to make it a
very small impact, though.
DR. APOSTOLAKIS: All of these impacts, I think,
are very small.
DR. SIEBER: Yes, well, there are some pieces of
equipment that at least in my experience that have sort of a
lifetime to them, you know, a piece of rotating equipment is
going to perform its function for a certain period of time,
and during that time, it's wearing out; things are happening
to it, and the longer you run it, the higher the likelihood
probably on an exponential basis is that it's going to
ill-perform properly or fail altogether.
DR. KRESS: Well, that concept is not built into
PRAs. It would be nice to have that. George is working on
that.
DR. APOSTOLAKIS: Well, although this is not
aging, because you talk about shorter time scales, but in
general, the failure rate in the PRA -- in fact, not in
general; 100 percent of the time is assumed to be constant.
DR. SIEBER: Yes.
DR. APOSTOLAKIS: So this particular phenomenon is
not modeled. So these components would have to be treated
separately if indeed you have this increase in the failure
rate.
DR. BRADLEY: And that's the same comment I was
making about a raw water system where you may be getting,
you know, like a heat exchanger, where you may, over time,
have clogging or whatever.
DR. APOSTOLAKIS: Yes.
DR. BRADLEY: And the same thing is true of
rotating equipment, where if you know there's a degradation
that's going to increase with time, you have to -- you know,
you have to take that into account.
DR. SIEBER: Well, it seems to me that this is
sort of a weakness, in my view.
DR. APOSTOLAKIS: Maybe those equipment, Jack,
should be identified in advance.
DR. SEALE: You know, it's one thing to use a
constant failure rate when you have fairly well documented
defense-in-depth in a wide number of areas to protect you.
On the other hand, when you start taking things out of
service, you don't have all of those defense-in-depth things
and so on, and, you know, we all know that light rub curves
exist, and those aren't constant failure rates either.
DR. APOSTOLAKIS: No, but, you know, very often,
in fact, if the failure rate exhibits an increasing
behavior, the constant value we use is sort of an upper
bound.
DR. KRESS: It's an average.
DR. APOSTOLAKIS: It could be an average; it could
be a slight --
DR. SEALE: It could be an upper bound.
DR. KRESS: So it's accounting for it, basically
from the database.
DR. APOSTOLAKIS: Yes so --
DR. SEALE: If it were an upper bound, that might
be appropriate in the case of equipment out of service,
whereas, the average may be more appropriate if you're
talking about, you know --
DR. APOSTOLAKIS: I think we're getting into a
detail now that -- other questions that --
DR. SEALE: What kind of details that this kind of
question may bring up.
DR. APOSTOLAKIS: You really have to understand
how the particular failure rate of that component was
calculated, and I think that in the vast majority of cases,
it's a conservative bound, because you go with the
statistics of failures and so on.
DR. KRESS: Well, of course, if he uses the RAW,
why, that's conservative enough.
DR. APOSTOLAKIS: The RAW is overly conservative.
Maybe for those distinct components, they should not be
allowed to do the second bullet; maybe they could do just
the first for some of these components if indeed there is
evidence that if you miss one surveillance, things get much
worse.
DR. BRADLEY: And don't forget that you're still
under the other provisions of the maintenance rule to
balance availability and reliability and take corrective
action if the performance of the equipment is outside your
goals that you set through the maintenance rule, the other
elements of the maintenance rule.
Are there any other questions on the missed
surveillance?
DR. APOSTOLAKIS: Just a note that I really think
what is happening now is something that PRA guides, you
know, have been expecting for a long time. The actual
applications now are forcing us to go back and look at the
assumptions behind the individual calculations in the PRA,
so this is the most useful feedback. Because up until now,
if you told anyone that you would do something to the
failure rates, it's good enough; what we're doing is good
enough. But now that we have this kind of feedback, you may
go back and really examine those.
DR. BONACA: I have a question to do with -- until
now, when you missed the surveillance in the tech spec, you
declared equipment inoperable, and so, this proposal has an
effect also on the definition of repairability at large; I
mean, how does it apply to other definitions of
repairability within the licensing domain? And maybe that's
a question I should ask of the staff as it relates to other
equipment or other actions, not to do necessarily with
surveillances.
Here, we have a risk basis to say that the
equipment is not any more inoperable. It will be operable
until the next surveillance.
DR. KRESS: Same principle applies to appendix J
on the leak rate testing and containment. When you --
DR. BONACA: I just would like to hear from the
staff maybe during the presentation that they're having, you
know, this has implications that go beyond the tech specs as
a principle.
DR. BRADLEY: Yes, and that's a good point, and as
we move more and more toward a risk-informed configuration
control approach, the conflict between the definition of
operability versus, say, functionality or availability
becomes complicated, and that's another of our initiatives
that -- I think that's number seven, and the current
definition of operability basically states that you've met
your surveillances. I guess the way I would interpret that
is that through this provision, you can continue to assume
that you've met that as long as you've done the risk
evaluation and, you know, will take the appropriate actions.
DR. BONACA: But we're still talking about within
the confines of the tech specs. I can think of -- you
discover that a pump doesn't meet some code requirements,
but it is fully functional. I mean, you know, until now,
you called it inoperable, because it did not meet the code
requirements. Now, this is a true precedent for saying that
the system should be called operable if, in fact, it's
possible. It would be important, maybe, that the staff will
tell us a little bit if they are viewing this as a
redirection for the whole concept of operability.
DR. REINHART: This is Mark Reinhart in the
probabilistic safety assessment branch. We will address
that point, and really, I think you've hit it: it's a
matter of degree and how that degree is evaluated and
addressed, but we'll talk about that.
DR. BONACA: Again, I'm not against it. I'm just
looking at this as a precedent that will affect the whole
definition of operability.
DR. REINHART: We agree, and that -- like this, at
issue number seven, which is a separate issue from what
we're really focusing on today really, ultimately gets to
the heart of that issue, and I think that's important.
DR. BONACA: But that's only tech specs.
Operability goes much beyond --
DR. REINHART: Agree, agree.
DR. BONACA: -- tech specs, and it seems to me
that a lot of events that we've had in the past where we had
LERs ended up with equipment inoperable and were
characterized by there wasn't a piece of equipment
functioning at a plant are going to change.
DR. REINHART: When you get into tech specs, that
definition becomes more legal.
DR. BONACA: Oh, yes, I understand.
DR. REINHART: Where a piece of equipment that's
not in tech specs, you can, in an outside the tech spec
definition say does this work good enough to satisfy its
intended safety function?
DR. BRADLEY: And there are other instances where
their (a)(4), for instance, will allow you to make a
temporary plant alteration to support a maintenance
activity. That might involve something like removal of a
barrier or opening a door. And (a)(4) will allow you, for
up to 90 days, to control that under (a)(4) versus 50.59.
However, you're still presented with a conflict with tech
specs, because with that barrier off, the equipment is not
technically operable, so if that's a tech spec piece of
equipment, those kinds of conflicts arise.
Any other questions on the missed surveillance
issue?
DR. APOSTOLAKIS: How many can they miss in a row
for a particular component? I mean, is it just a random
thing, we missed one, or can they miss three of them?
DR. BRADLEY: I would certainly -- some of the
tech spec branch here may have more of a history of this,
but, you know, my view is that missing more than one in a
row would be -- I mean, giving that missing any surveillance
is a fairly unusual event, missing two in a row or more than
one in a row would be very, very unusual. There's no
specific limit, but again, you know, the intent here is not
to -- remember, you're going to have your corrective action
program. I assume if you miss the same surveillance twice,
you would get hammered pretty hard in the CAP for that,
because you've failed to correct it the first time around.
So I view that as a very unlikely situation.
DR. APOSTOLAKIS: Do we what the causes for
missing surveillances are?
DR. BRADLEY: Yes; I think the staff has spoken to
some of those before. You may have procedure changes that
may change a surveillance, or you may have missed some
portion of a surveillance, not completed it. I don't know
if there are others that you guys are familiar with.
DR. REINHART: This is Mark Reinhart again. I
think I would address -- our expectation is the proposal is
going to be for just one. If you miss the surveillance,
then, we get into this process, and there's not a provision
to miss it again.
DR. BONACA: But what about to miss another
surveillance?
DR. REINHART: Okay; another surveillance is, you
know, a separate situation. But say we miss surveillance SR
whatever it is on pump X. That same SR, there's not a
provision to the next time say, oops, I missed it. No way;
the focus on that has to be so great that we achieve it.
Now, if you have valve Q over here, that's a
separate issue, but I think what the industry has presented
to us, their survey of data is that missed surveillances
are, is it fair to say, extremely rare?
DR. BRADLEY: I think there were 10 NOEDs in the
last 5 years. I think that's some data like that.
DR. BONACA: Yes, but that's because in the past
you got hammered if you missed it.
DR. BRADLEY: Yes.
DR. BONACA: In the future, you won't. So, you
know, I'm only pointing out that if you go in an outage,
oftentimes, surveillances are in critical path.
DR. BRADLEY: Right.
DR. BONACA: And if you get pressure to restart,
you say oops, you know, so, we have one. And I'm only
saying that technically, there should be a monitoring of the
part of how this rate is going to increase, because the past
is not going to reflect --
DR. BRADLEY: If you're talking about a willful
violation, you know, that's something that --
DR. BONACA: No, no, I'm not talking about
willful; I'm only talking about conditions that at times are
plenty legitimate, and they are conditions under which you
may discover that you have the surveillance done may delay a
restart, for example, by a significant time, and there are
-- and it may not be the best decision, in fact, and now,
you have an opportunity for saying it doesn't make sense,
and the staff may agree with you. In fact, it may make
sense in that particular case to put it off.
DR. BRADLEY: I think what would happen in today's
world is you would go in for an NOED enforcement discretion.
DR. BONACA: Yes.
DR. BRADLEY: And with a good case, a good risk
case, it would probably be granted. So the net effect is no
different; it's just that the process is built in.
DR. BONACA: I agree with you.
DR. BRADLEY: Yes.
DR. BONACA: I'm only saying that -- I was talking
more in terms of oversight and --
DR. BRADLEY: Right.
DR. BONACA: -- you know, I mean, there could be
almost a certain post-PI on the part of any plant to see how
you perform on this.
DR. BRADLEY: And the correct -- I know the
corrective action program is a key element of the new
oversight process, and I believe that the disincentive to do
this remains strong. It's just a different disincentive.
It's not you're going to shut the plant down. It's you're
going to get hit on the corrective action program element.
DR. REINHART: This is Mark Reinhart again. We're
well aware that the oversight program has to mesh with this
program and be aware of the kind of potentials you're
talking about and pick up on them, but clearly, the intent
is not that a licensee would use this as a convenience to
extend or shorten an outage or extend some frequency just
because they need to start up.
DR. SIEBER: I would doubt that that condition
would exist in any plant where people would intentionally
miss surveillances to pressure start-up.
MR. MARKLEY: But, Jack, I don't think they'd miss
surveillances, but I don't think they'd hesitate to use this
at all every startup, every outage if they needed to.
DR. SIEBER: If they missed a surveillance, and it
would mean going back, changing a plant condition to go back
to pick it up, I imagine that if it was a legitimate miss
that they would use this mechanism.
DR. BRADLEY: Yes.
DR. SIEBER: And that's what it's intended to do.
DR. BRADLEY: That is correct; that is correct.
DR. SEALE: And you tell me that these events are
so rare that there's really no reason to make any provision
for what you would do if you had two of these independent,
now, in the bank at one time.
DR. BRADLEY: I don't believe there is a need to
do that; that is correct.
DR. SIEBER: It would be a rare situation to have
two at one time in any given year in one plant.
DR. BRADLEY: Yes.
DR. SIEBER: Okay.
DR. BONACA: I can see outages in the middle of
the summer where, you know, you have the great strain where
it doesn't make any sense to stay down for a day and a half
because you have to complete the surveillance, and you would
go back for --
DR. SIEBER: NOED.
DR. BONACA: -- for discretion --
DR. SIEBER: Right.
DR. BONACA: -- today. But for a utility to do
that, it's actually one event; that they don't do it again
for two years maybe. You know, if you have this, it's going
to be more. I mean, I think monitoring is an important
thing; just monitoring that this is not the grading fact,
the past good performance that is supporting, you know,
going in this direction so --
DR. REINHART: This is Mark Reinhart again. A
thought on the missing two surveillances, again, that that
would be picked up in the risk assessment. I mean, it
wouldn't be missed. Another point is when we were
developing the improved standard tech specs, we tried to
think of a lot of what if scenarios early on but did
conclude that the bookkeeping associated with trying to
track such rare situations would be more burdensome than
keeping our eye on the risk-significant aspects of operating
the plant.
DR. POWERS: Okay.
DR. BRADLEY: Okay; well, last call on the missed
surveillances?
[Laughter.]
DR. BRADLEY: All right; I want to talk a little
bit about the initiative three, and that's on mode change
restraints. I don't have as many slides on this one. Let
me just talk about it a little bit. The existing tech spec
3.0.4 in the standard tech specs precludes making plant mode
changes up in mode when you are in an LCO that would be
applicable or when you have a piece of equipment whose LCO
would be applicable in the mode that you are preparing to
enter.
Currently, a large number of those 3.0.4 -- there
are 3.0.4 exceptions provided in the standard tech specs for
a fair number of the systems and components in tech specs
right now. Those 3.0.4 exceptions have evolved over time
and are not what I would call risk-informed exceptions.
They may be qualitative or obvious types of issues where
it's clear that the -- going up in mode that this particular
component would not have an impact. And what we're
proposing to do with initiative three is to essentially make
the 3.0.4 exception a risk-informed process rather than what
we have now, which is just a mixed bag of exceptions in the
tech specs.
This one involves -- what it basically would allow
you to do is to -- is require each -- the way this would
work is that each owners' group will have to develop a
model, a transition risk study that will look at the impact
of various components as a function of plant mode and
various initiators that you could have in those modes and
what are the accident sequences that are important and what
areas of those components contribute to.
And this has been done for one of the four owners'
groups already, CE. What that showed for their line of
plants was that there were about four systems in the plant
that would generally require a more rigorous risk evaluation
before you could remove the mode change, existing mode
change constraints. However, for the remainder of the
systems in the plant, they were basically able to show
generically through this CE model that the impact was very
small.
And what this means is that when you move up in
mode, you would still have to return the equipment to
service within its AOT time, so the difference from where
you are today is that you couldn't move up in mode at all
now with this change, unless you have the 3.0.4 exception,
which many systems do. This would allow you to go on out,
but you still have to get the component back in service by
the end of the AOT time.
This, again, is -- the types of things you have to
consider here are very similar, again, to what we're doing
in (a)(4). (A)(4), as a matter of fact, talks about mode
changes and the impact -- you have to evaluate the impact of
mode changes on your plant configuration control program, so
again, here, we're trying to develop some symmetry with what
(a)(4) would require.
DR. KRESS: Would you make the allowed outage time
then consistent with the criteria that's in (a)(4), the
spike criteria?
DR. BRADLEY: Under (a)(4), you would have to do
that. Now, ultimately, as we go to initiative four, we will
try to globally make the AOTs and tech specs compatible with
what (a)(4) would do. I mean, this particular initiative
won't do that on its own, so we'll still be in the
existing --
DR. KRESS: What do the tech specs say on this?
DR. BRADLEY: The existing AOT; but ultimately,
when initiative four goes through, that AOT would be a
risk-informed AOT. But (a)(4) would still control you to
the risk-right element of this.
This one requires a little more analysis in
initiative two. It does require that each of the -- and the
other three owners' groups need, in order to get this change
into the ITS, do have to go develop a similar work to what
CE has done. But once the plants -- the owners' groups have
done that, it will be a list of systems that basically, you
couldn't do this -- make this exception unless you do a
plant -- if you have a plant-specific model that would allow
you to make a finer resolution on that risk impact and show
it was okay.
DR. KRESS: But we haven't seen this transition
risk model.
DR. BRADLEY: It was discussed at the subcommittee
meeting. I forget; was it supposed to be performed?
MR. MARKLEY: The CEO's group was supposed to be
providing it.
DR. BRADLEY: And they have not provided it?
DR. APOSTOLAKIS: Yes; and also, the point about
the supplementary meeting for a full transition.
DR. KRESS: Yes; it's an outshot of --
DR. APOSTOLAKIS: Right.
DR. BRADLEY: It's a simplified --
DR. APOSTOLAKIS: Yes.
DR. BRADLEY: I will make sure that you guys get
that.
DR. UHRIG: When does the clock start on the AOT?
Is it the time the mode changes or when the equipment went
out of service initially?
DR. BRADLEY: Typically, the equipment is not --
since you're going up in mode, it's generally going to be a
case where the equipment may not have been in service
previously, so that the clock would start when you enter the
modal applicability for the LCO.
DR. SIEBER: For the other reactor vendors, when
they do this analysis, some vendors have differences in --
within a category, for example; General Electric, just to
pick -- different types of plants, they are going to have to
perform a transformation model for each type?
DR. BRADLEY: What the owners' groups would do --
and the BWR owners' group, I presume, will have to look at
the different BWR types, containment types, and if there are
differences in how this would be impacted for the BWR Four
Mark One or whatever, Mark Two, what have you, then, yes;
the answer to your question is yes. It may not be purely
generic, you know, how here's all BWRs do it. There may be
deltas based on specifics of the plant.
DR. SIEBER: So for some types of reactors, there
would be more involved than for other types.
DR. BRADLEY: Right; what these studies are trying
to do is basically show which components can't be
dispositioned generically that, you know, here are the four
things that you can't remove the exception from unless you
can do a plant-specific study and show that you're meeting
the 1.182 type threshold.
DR. SIEBER: Than you.
DR. BRADLEY: Sure.
Any questions, further questions?
DR. APOSTOLAKIS: A more general question; it will
probably be addressed more later. Are we really
extraordinizing our -- if it doesn't look any different,
you're missing one surveillance; as long as they remain
rare. From the risk perspective, it seems to me --
DR. KRESS: I think you're right, George.
DR. APOSTOLAKIS: We're doing this because we are
used to the existing regulatory approach as being, you know,
very prescriptive and very deterministic. In the larger
scheme of things, does it really matter? Maybe there's one
of the components that, as Jack mentioned, that you really
don't want to do that.
DR. KRESS: Somebody brought up the issue of
slower motion -- but that's the only place I can see where
-- but you're right; this has an essentially insignificant
impact on risk.
DR. BRADLEY: To use our overused term, this is
the low-hanging fruit of the tech spec risk-informing, and
these are the -- the reason we did these first was to get
our toe in the water and understand how to make this work
for a simple application.
DR. KRESS: What I'm interested in, George, is
this is sort of a precedent for things to come, maybe, and
we're scrutinizing it, I think, because we want to be sure
it's spelled right even though it has no real risk impact.
And the things you want to ask are the things we're asking,
like what is your criteria for a spiking risk, and how do
you apply it? What's the role of the PRA? How does it fit
in there? How do you do it? And so, I think those are the
things that we're trying to establish, even though this
particular issue, I think, we can almost set off and say
never mind.
DR. SEALE: You know, the comment is made there on
the side that most plant systems have insignificant risk
effects. That's most; it's not all. I would be curious to
have some idea of what the exceptions are. I have a sort of
an intuition that tells me what I think they are, but I'd be
interested to know what the exceptions are, if nothing else
than to reinforce my wrongheadedness, maybe.
DR. KRESS: Prejudices?
DR. SEALE: My prejudices, yes.
DR. BRADLEY: I believe there were four systems.
Can the staff help me here? I don't recall the exact
systems in the CE.
DR. REINHART: Well, this is Mark Reinhart again.
It was HPCI, aux feed water and diesels.
DR. SEALE: I feel smarter already.
[Laughter.]
DR. APOSTOLAKIS: Would you tell me about the
risk-significance of the systems, the four systems?
DR. BRADLEY: Those are the ones that you -- that
absent some plant-specific evaluation, you couldn't
generically make a 3.0.4 exception; you couldn't come up in
mode with those systems out of service.
DR. REINHART: A train; we're looking at a train.
DR. BRADLEY: Yes, where the tech specs were for
some of the -- right.
DR. SEALE: It's nice to see that you think that
-- discrimination, really.
DR. APOSTOLAKIS: Is the staff going to review
these transition list models and see if they are acceptable
for risk orientation?
DR. REINHART: Review in the sense that we'll look
at them; we'll make an evaluation. We don't -- we're not
going to go into a minutely-detailed review.
DR. APOSTOLAKIS: Well, you will make a decision
as to whether they are good enough for this application.
DR. REINHART: Yes.
DR. APOSTOLAKIS: Not whether they are -- yes, you
will do that, and we will get a chance to see your
evaluation, the ACRS?
DR. REINHART: We can make that available, sure.
DR. APOSTOLAKIS: Yes; I would like to see it,
because I see this as a great way to improve the PRAs, the
applications.
DR. KRESS: Yes, I agree.
DR. SIEBER: Well, individual licensing actions
will be made licensee by licensee.
DR. BRADLEY: Yes.
DR. SIEBER: So no generic program.
MR. MARKLEY: Will we be issuing SERs based on
that?
DR. REINHART: Bob Denning -- he's going to cover
how we're looking at the application, how the TSTF will be
reviewed and how a plant-specific application will be
reviewed.
DR. APOSTOLAKIS: It's interesting that if you
look at the workshop proceedings for the staff that there
was almost universal agreement that transition risk was not
more than -- and people were very -- what is it here that
makes us comfortable with the transition modeling? You
informed me in a statement from industry, for example, that
-- the staff -- and you would not have that concern?
DR. REINHART: We could add the thought that
between then and now, industry has done on a lot of work on
your transition models. We've had staff look at San
Onofre's AFLPs, so I guess my overall answer is that over
time, it has improved, and we're looking to improving. I
know that the CE owners' group -- San Onofre is a CE plant
-- is generously sharing their information or offering to
share their information with the others so they can get not
only the information out there but obviously some peer
feedback.
DR. APOSTOLAKIS: It seems to me that this is an
important enough issue that perhaps we should get a
supplement on this, not on this issue of mode change but the
transition risk models. And I haven't even seen anything
like that so --
DR. KRESS: Good issue.
DR. APOSTOLAKIS: And, you know, in the broader
context, what do these models give you, and how do they use
it?
DR. KRESS: Yes; I think that would be a good
issue.
DR. SIEBER: It looks like at the beginning, if
you get enough discrimination in their application that it's
worthwhile to look at what they're doing now.
DR. APOSTOLAKIS: Yes.
DR. SIEBER: Before they had that, it wasn't --
DR. APOSTOLAKIS: Yes.
DR. SIEBER: The kind of models that you're
looking at, for example, with combustion are not true
transition models. They're snapshots after a mode change.
So you aren't modeling in the issues.
DR. APOSTOLAKIS: That's why we need a supplement.
DR. SIEBER: This is not, to me, transition risk.
DR. APOSTOLAKIS: It's better than nothing but --
DR. SIEBER: Well, I think that it's adequate to
do this kind of work.
DR. SEALE: It's a first step.
DR. SIEBER: Well, it supports what they're doing.
DR. SEALE: Yes.
DR. APOSTOLAKIS: Who on the staff would be the
person to comment if we wanted to establish a supplement?
Is it Cunningham or --
MR. BARRETT: Yes; I would think that this is
something the Office of Research would take the lead for,
you know, if you're talking about the accuracy of methods,
state-of-the-art as far as PRA methodology is concerned.
DR. SIEBER: In order to avoid shortchanging the
staff on their presentation time, I might ask if anybody has
any more questions for Mr. Bradley or if he has anything
more to present? You do not?
DR. BRADLEY: I do not. Just in closing, I would
just say that we appreciate the efforts of the staff and
this committee in helping us move forward with these
prototypical changes to tech specs, and as someone said
here, I thought it was the best comment: these are the
kinds of things we need to incentivize the industry and get
that movement toward a more risk-informed approach across
the board, and that's why we think this is important.
DR. SIEBER: Thank you very much.
DR. SEALE: I think we ought to let Biff lead off
all the time.
[Laughter.]
DR. APOSTOLAKIS: Now, speaking of that, by the
way, this was a highly unusual order of presentations. Why
did the staff feel that NEI should go first?
MR. DENNIG: Well, I that part of our reasoning
was that in our last discussions, there was a lot of
interest in some details of the industry proposals, and
those details are -- as far as the staff is concerned,
they're now being worked into things that we haven't got
yet. We haven't seen the revision of the things that we
have under review that address a lot of the things that were
brought up, and I thought Biff did a good job of bridging
the gap between what the staff has seen and what we're about
to see as far as those kinds of details.
So we figured that would be the most productive
way to go and then come behind and then anything that was a
concern from the last meeting that hadn't been adequately
addressed or was something that was peculiar to us, then, we
would just take that on. So that was the extent of the
reasoning.
I'm Bob Denning again, and I have Mark Reinhart to
my left and Ann Gilles. We're going to coordinate our
presentation. Again, let me say at the beginning, I think
our major message: we hear you. The focus of our
presentation is to tell you what we heard at the last
meeting by way of your concerns and to give you a response
to those concerns as to how they've already been addressed
or they will be addressed.
What we have up initially is basically a punch
list of what we thought were the major categories of how we
might bin your comments to us, and we will -- we'll go
through these in some detail. Let me go to the next one.
The first item is the details of plant risk decision
process, and again, Biff has done, I think, a very good job
of filling in the blanks about how the specifics of the
process that would be used in initiatives two and three is
being fleshed out in view of the approval of the (a)(4) reg
guide and some thinking that the industry has done. We
reiterate that we will be getting more detailed proposals in
response to comments that we've provided, comments that
you've provided and some thought that the industry has
given; that we will be focusing on the kinds of issues that
Mr. Sieber brings up: how do you do this calculation where
the options for doing this calculation, where the
quantitative and qualitative aspects of that.
We're aware of the concern about risk-informing by
committee, if you will. That lets us all get together and
see what we think and then decide that that's risk-informed.
We will, in our safety evaluation, outline what we believe
-- what we will require those decision processes for
initiatives two and three to contain, and individual plant
amendment requests will include more detail if necessary.
And I'm going to come back to that a little bit later when
we described a process for -- that's recently been put into
place for adopting generic changes to the standard. And I
think now, I turn this over to Mark Reinhart.
DR. REINHART: I just wanted to cover a couple of
aspects on the slide Bob just talked about, and one relates
to the question that was raised earlier on missed
surveillance. And I guess my thought is what do we really
mean by a missed surveillance? If you just say that phrase,
oh, they missed a surveillance, that brings up one picture.
What often really happens is a licensee has a program that
tracks their surveillances, so it's very rare that they just
miss their schedule.
What most often happens is there will be a
procedure change, a change in the way the surveillance is
performed, and in review, they find out they missed some
technical detail that officially says they missed a
surveillance, but the large majority of it was actually
performed.
So in the missed surveillance arena, one, the
licensee definitely needs a program or a process to go in
and look at whether they used the RAW; look at their failure
rate; whether they have some qualitative evaluation, but
they can say what parts of the surveillance have we
performed? What does that tell us? What is the history of
this surveillance? What are the pieces we missed? What can
we do with the physical plant to go in and give us some
level of confidence that the piece we can't get in whatever
condition we're in -- what is our feeling of that status?
So, by and large, they're going to be able to go a long way
to say whether that surveillance was actually completely
missed or just some portion of it. And then, they decide
whether they look at RAW, the failure rate, and come up with
a quantitative but also qualitative risk-informed decision
as to what to do with that plant and that surveillance.
So all of that is, in our thinking, hoping we see
this in whatever the licensee proposes in these TSTFs, some
sort of process that gives us that flavor.
DR. SIEBER: There will be some kind of
description someplace as to how to do this, though.
DR. REINHART: My view is the licensee is going to
have a program that would describe that, yes.
DR. SIEBER: And you will look at that program and
say yes, this is adequate or not, somebody will.
DR. REINHART: The program is going to be
available at the plant site. I'm sure that the regional and
the resident, in their inspection process, will have a look
at that.
DR. SIEBER: Okay; thank you.
DR. REINHART: We likely, in our -- maybe our
safety evaluation of looking at their submittal are going to
see what aspects they propose to have in the program.
DR. SIEBER: Right.
DR. REINHART: Then, if we go to the next one, one
of the things we're looking at is right now, we're saying
what is the quality of the PSA that licensee has, the
quality that backs the submittal for the TSTF or the tech
spec task force change document that makes the change but
also the individual licensee's program that tells the
quality of their program. Now, today, we look at what
they're doing. A licensee performs a PSA. Then, they come
back, and they say okay, I had a certain independent review
internally; maybe a contractor; maybe a peer review; maybe
some certification process. The staff goes and does some
looking at that PSA, whether it's the exact application or a
little bit broader for what we think is coming down the
road.
And so, we're looking at a de facto, if you will,
compilation of how good is that standard today? We're
looking eventually to have some agreement, some standard,
that we all agree with that, yes, we meet this structure.
DR. APOSTOLAKIS: I'm a bit confused now. My
understanding is that the agency has been very cautious not
to demand the PSA from the licensees. Now, what's happening
here now? Do they have to -- if a licensee says I was
licensed 25 years ago; I didn't have to do a PRA; I don't
want to do it now; I'm going with the old system --
DR. REINHART: They may do that.
DR. APOSTOLAKIS: How do they do this?
DR. REINHART: They can't. This is a
risk-informed application. They're saying I'm requesting,
staff, this program.
DR. APOSTOLAKIS: They may choose not to implement
this?
DR. REINHART: They don't have to apply if they
don't want -- they can stay with the tech specs they have.
DR. APOSTOLAKIS: Okay.
DR. REINHART: It's voluntary.
DR. APOSTOLAKIS: All these are voluntary.
DR. REINHART: Yes.
What the industry --
DR. APOSTOLAKIS: So if they miss a surveillance,
then, the staff cannot ask for the risk-significance of that
missed surveillance, because it's not an issue of public
protection.
DR. SIEBER: Well, you answer the question.
DR. REINHART: We've gone through a process to get
in place staff authority to ask for risk information, and
we're kind of into what if space, but if the situation
arose, and we felt that it was necessary to have risk
information, we would have the authority to ask for it.
DR. APOSTOLAKIS: But you would have to say
something about adequate protection, the way I understand
it.
MR. BARRETT: Could I add a word? I think we'll
be talking about this this afternoon.
DR. APOSTOLAKIS: Yes.
MR. BARRETT: The question of when we can ask for
risk information, but in a case like this, you know, I think
the main distinction is asking for information that's
relevant to compliance with the regulation, which is what we
might be talking about here, for instance; if someone wanted
to miss a legally-binding surveillance requirement, they
might be asking for us to give them relief. The situation
we're having this afternoon is a situation where we want to
bring risk into a licensing application in a situation where
risk is not required under the legally-binding requirements
and where a licensee is perfectly within their rights to
meet the regulations without reference to risk. So that's
the distinction between what we're talking about here and
what we're talking about this afternoon.
MR. MARKLEY: George, there are really kind of
three options, you know. If they have a problem there, we
would expect them to comply with the technical specification
requirements. Their first alternative would be enforcement
discretion, which is what they currently do now, and then,
the new option is the risk-informed approach. So that's
kind of where they fall together.
DR. SIEBER: But they would be in one category or
the other. They would be in a deterministic set of tech
specs or a risk-informed set of tech specs.
DR. REINHART: Yes.
DR. SIEBER: And you follow the tech specs that
you have.
DR. SEALE: Could a licensee go to a set -- to the
standard set of risk-informed tech specs but not play this
game with regard to missed surveillances and so on?
DR. REINHART: For a licensee to get the piece
they're asking for, if it's risk-informed, they need a PSA
and a program --
DR. SEALE: Yes.
DR. REINHART: -- in place sufficient to that
application.
DR. SEALE: To invoke these processes.
DR. REINHART: Yes.
DR. SEALE: But what I'm saying is you could have
a -- you could adopt the risk-informed tech spec format and
still play the old game.
DR. REINHART: There's an improved standard
format --
DR. SEALE: Yes.
DR. REINHART: -- that all licensees have the
option to go to. The risk-informed improvements would
maintain that format.
DR. SEALE: Yes.
DR. REINHART: There might be some slight nuances
programmatically on how they would invoke their particular
regulations, their particular pieces, but they couldn't pick
up what's proposed in these TSTFs without that information.
The only exception might be if industry successfully makes a
case, proposal X, whatever that might be, we have enough
generic information to show that really, the risk puts it
over here rather than over there, and that's applicable to
everybody, and we accept that. Then, they could go that
way.
For these two issues, two and three, the industry
is proposing that the maintenance rule (a)(4) guidance gives
them an adequate place to start with the requirements
they're looking for. Now, with that in mind, if we back up
a little bit, first, there was the old maintenance rule, if
you will, the (a)(3). Then, the staff had the configuration
risk management program that it worked out with the industry
for risk-informed tech spec applications. The thought all
along was that (a)(4) was going to pick that up, and the
thought was rather than have a different program for this
application versus that application, we come up with one
program that we can apply without thinking in mind the
program the licensee has to support this application ought
to support also a four and vice versa.
So for these areas, particularly two, this may be
the right level. For area three, I'm not sure. I think we
need to see exactly what they propose there. But in any
case, when we're done, we want to make sure that the safety
evaluation -- and again, Bob is going to explain a little
more about this in a minute -- but the safety evaluation
outlines the requirements we're looking for to support the
particular application under discussion.
DR. APOSTOLAKIS: We have 11 minutes left.
DR. REINHART: With that in mind, I'm going to
turn over to Nanette Gilles.
DR. APOSTOLAKIS: Good.
MS. GILLES: I'm Nanette Gilles from the technical
specifications branch, and I'm going to very quickly talk a
little bit about the oversight process rule. This was a
concern that was brought up at the subcommittee meeting.
And to make just a few points, one, we have some ongoing
discussions with the NRR staff involved in the oversight and
inspection process. We have met with Mr. Dean and explained
the initiatives to him, and we are working with them to try
to both give them some understanding of the changes we're
making in tech specs and how that might affect the oversight
process and gain some understanding ourselves of how the
oversight process might help us in monitoring such changes.
And we have identified some specific inspection
procedures which would be useful in inspecting the areas in
which the changes are being made. Those are inspection
procedures in the areas of surveillance testing, maintenance
risk assessments and emergent work control which would help
you in initiative three, and then, the identification and
resolution of problems which is the inspection procedure
that covers licensees' corrective action programs, and since
all missed surveillances will be entered into the corrective
action program, that would be an important procedure from
our point of view.
We also have accepted an invitation to become
involved in the research risk space performance indicator
program, which will help us feed back to them the usefulness
of the current indicators and perhaps the need for future
indicators as we proceed with risk-informed changes to tech
specs. I'll also point out that in the reporting area, any
missed surveillance that is subsequently failed once it is
finally performed will still be a reportable issue.
In the area of public involvement and support, we
did have some early dialogue with concerned groups on this
issue. You may recall at the first subcommittee meeting in
December, Public Citizen, Jim Riccio, presented a letter to
the ACRS subcommittee, and we did follow up with Mr. Riccio
and respond to that letter and have some discussions with
him, and we have been inviting him to our meetings and
keeping him informed of our activities.
In addition, we do plan to implement a process
called the consolidated line item improvement process for
these changes to the technical specifications. That is a
new process that will involve publishing the draft safety
evaluation for public comment, and this is the safety
evaluation on the generic request for the change to the
standards. So we will give the public some time to comment
on that, and then, we will take their comments into
consideration before we put that change up for use by
individual licensees.
But again, the public has a second opportunity to
get involved on the individual license amendment request
because, of course, those will be noticed as all license
amendment requests are.
Finally, we have identified a need for a
communications plan on these tech spec initiatives that
would help us to reach out to more external stakeholders as
far as communicating what the staff is contemplating, what
the industry is proposing.
DR. SEALE: You mentioned that Mr. Riccio, I
believe, you followed up with him and so on. Has he taken
advantage, or has he exercised your invitation and
participated further in discussions on this issue?
MS. GILLES: He has not come to any of our
subsequent meetings, but he has indicated to us that he has
become rather overwhelmed with work in the risk-informed
regulation area, and I think that is not from a lack of
interest. It's just that there is so much going on in this
area that he's trying to cover that he hasn't been able to
make it to subsequent meetings, but we are providing us with
our meeting notices and our meeting summaries.
DR. SEALE: That's all he came to?
MS. GILLES: And now, I'm going to turn it back
over to Mark Reinhart, who is going to discuss effects on
safety culture.
DR. REINHART: Just briefly, well -- very
delicately --
DR. APOSTOLAKIS: This is not the term the agency
is using, is it?
DR. REINHART: This was an issue that the
subcommittee brought up, right, and we were assured that
this isn't a technical issue but an issue that you all are
aware that it could have an impact, so what we were going to
try and do in the way of addressing it is say that our
premise, and we understand the licensee's premise, is that
they will operate their plant safely and that -- I'll jump
in under here -- really, by taking a look at risk
information, looking more at the risk, the safety of the
configuration, where it is and where they want it to go
moreso in the past because of what the PSA is doing for them
is going to actually increase that safety culture by
focusing the licensee on the safety significant aspects of
the plant.
At the same time, we're going to have our
inspection process looking at the programmatic situation;
making sure there's no breakdowns there and not only an
issue by issue look but an annual, overall problem
identification, and as Nanette mentioned, also working for
some performance indicators that would give us some trending
on the situation, so however we want to define it, we
believe that this is making licensees more aware of the risk
of their plant configurations, which is an improvement.
So with that in mind, I'll turn it back to Nan.
MS. GILLES: The last concern that we had to
discuss was a concern about some NRC staff programmatic
aspects and perhaps how some of these changes might be
communicated to the rest of the staff, and we had mentioned
previously that we are going to become involved with
research and their performance, risk-based performance
indicator program and have been working with the oversight
process staff, and we also, again, have identified the need
for a communications plan to help us reach the remaining
parts of the agency staff.
I believe that concludes our presentation. I'll
take any questions.
DR. BONACA: Well, I had a question for general
identification of this change, where by now, this definition
of repairability in tech specs has changed, and it has
applications for all of Part 50, and clearly, you know, it
has that kind of implication that Dr. Kress put before, I
mean, that implication is large from limited changes we make
here, and do you have any --
DR. REINHART: I'll add for a long time, industry
and the staff has been aware that particularly in tech
specs, the legal definition of what is operable is not
always coincident with the equipment works well enough right
now to perform its safety function, and how to solve that
problem is still a piece we're working with. Issue number
seven of the seven issues is to address that more head on.
This obviously is a precursor. It's not the final answer.
MR. BECKER: This is Bill Beckner. I'm chief of
tech specs. I'm sure we'll address that a little bit more.
Throughout regulations, we have the ability to, in fact,
address the degraded but operable. It isn't quite up to
pedigree, but we know it's going to work. In tech specs, we
also have that ability to take a look, and tech spec
definition is basically can it perform its function?
However, there's one thing that trumps that, and that's if
it doesn't meet the surveillance, regardless of what the
equipment is, it's inoperable, even if we know it's fully
functional, and I think that's the issue here is if we know
it can perform its function or if we know there's no safety
concern to basically get rid of what I call that failed
surveillance trump, and that's, I think, what we're trying
to address here.
DR. BONACA: The only point I was making is that
there is Part 50, and then, there are the tech specs.
MR. BECKNER: Correct.
DR. BONACA: The importance of tech specs is such
that it is -- the issue of the operability of a component to
tech specs, okay, is almost in my mind of higher importance
than either a definition for components which are not in
tech specs.
MR. BECKNER: Correct.
DR. BONACA: For which now we are going to relax
the definition of operability for components in tech specs,
okay? I think it begs a revising of the definition of
operability within Part 50, I mean, because -- do you see
what I'm trying to say? Here, we are taking the most
important components, which are in tech specs, and changing
the operability definition.
MR. BECKNER: No, let me say I'm not sure I'm
totally following you. If a piece of equipment cannot
perform its function --
DR. BONACA: No, let me give you an example now.
Here, you have a component which is because it was not --
did not have the surveillance done on time, we recognize
it's still operable.
MR. BECKNER: Correct.
DR. BONACA: Okay; so one part of its attributes
that define operability wasn't met. But the functionality
is there; therefore, we call it other. Now, there is a
component there you find in the plant that doesn't really --
we discover it doesn't meet the requirements of, I don't
know, code requirements on the temperature or pressure, but
you know that in the regulation that it would perform its
function. So one attribute is found not to meet the
requirements of operability for a lesser component, because
it's not in tech specs.
And I think that then, the same should continue to
be applied to the component for recognizing functionality.
I'm not pushing the issue here but showing how I'm making
minor changes here in a very important group of documents,
which is the tech specs, begs the question for similar
initiatives, you know, for lesser components of the same
principle.
MR. BECKNER: I think I understand your concern,
and we'll have to do some thinking about it. I think the
fundamental can the equipment do its function is still the
operability definition, and I think it's just a matter of,
like I said, of making sure that a surveillance -- not a
surveillance failure but just not performance bounds doesn't
trump that in my terminology, but we'll think about it and
see if we can -- see if there's any obvious need for other
changes.
MS. GILLES: I will point out that initiative
number seven is number seven on the list of seven just for
this issue that you've raised, that we all recognize that it
is tied intimately to so many other parts of the regulation
that it's one that we have to pursue cautiously, and I think
that we've even, you know, seen some of these concerns rise
up in the risk-informing Part 50 Option II area, where they
are trying to separate, you know, the definition of
safety-related from some of these special treatment
requirements, which is very similar to the operability
concern where you're also talking about seismic concerns and
equipment qualifications, which all come into play, and how
do you separate those from functionality?
DR. REINHART: Any further questions?
DR. SEALE: Well, at the bottom of the comment was
the question whether or not we think we need to write a
letter on this.
DR. APOSTOLAKIS: I think that's to be discussed
later.
DR. SEALE: All right.
DR. APOSTOLAKIS: Yes; yes, we'll discuss it
later.
DR. SEALE: Okay.
DR. SIEBER: Well, I'd like to thank Mr. Bradley
and the staff for their presentations, and I think it
increased our knowledge, and so, Mr. Chairman, I'll turn it
back to you.
DR. APOSTOLAKIS: Thank you.
We'll recess until 10:18.
[Recess.]
DR. APOSTOLAKIS: We're back in session; okay, the
next item on the agenda is potential revisions to the
pressurized thermal shock acceptance criteria, and the
cognizant manager is not here, so we'll go ahead anyway.
Mr. Cunningham, you're in charge.
DR. CUNNINGHAM: Thank you, sir.
My name is Mark Cunningham. I'm the chief of the
PRA branch in the Office of Research. I'll be making part
of the presentation this morning. The other part will be
made by Ed Hackett, who's the acting chief of the materials
engineering branch in the Office of Research.
As Professor Apostolakis said, we are here to talk
about the revisions to one aspect of the screening criterion
that's used in the pressurized thermal shock pool. A few
weeks back, we provided the committee and the subcommittee a
draft Commission paper on the subject of potential revisions
to the PTS screening criteria. Today, we're going to
summarize some background on the purpose of that paper and
some information on the screening criteria itself and PTS
accidents. I want to provide some more information that's
relevant to potential revisions; Ed will talk about work
that's been going on over the last few years or number of
years in materials research with respect to better
understanding of reactor vessel flaws, and I'll talk some
more about Commission guidance in the context of PRA, and
we'll talk about some approaches, some potential approaches
for revisiting the screening criteria and where the staff
intends to go with this.
I should note that when we talked to the
subcommittee, we talked in the context of a notation vote to
the Commission with options and a recommendation for an
option. Based on our conversations with the subcommittee,
we've revised our approach so it's more of an information
paper at this point, informing the Commission of where we
intend to go with this and laying out a couple of issues
that the staff is going to have to work on. We'd like to
get some input from the full committee today on where we
think we're going.
At this point, based on the fact that we're
changing this to an information paper, we're not formally
asking for a letter from the committee at this time. Again,
that's somewhat different than what we talked about at the
subcommittee meeting, but we don't think the nature of the
paper as you'll hear it today is such that we don't think
it's necessary to have -- and it may be a bit premature to
have the committee write a letter at this point.
By the way of background, the last 9 months or so,
the staff has had work underway to revise the technical
basis for the pressurized thermal shock rule. We're doing
that because there are some indications that the work that's
been done in the previous years in materials research and
our experiences with implementation of the rule say that we
may want to revise the rule and provide a more realistic
basis, technical basis, for the rule.
The purpose of the Commission paper that we
provided the subcommittee was to provide -- basically to get
some early decisions on a couple of elements or one
particular element of the rule revision or the technical
basis revision. That deals with one part of the screening
criteria that's used in the PTS rule, and I'll discuss that
in more detail later. Basically, the paper would be, then,
to inform the Commission of what we're up to and where we're
going with this and then telling them about some issues that
we're going to have to work on.
Again, by way of background, the impetus for the
rule in the early 1980s was some experience with overcooling
transients in some reactor vessel operating experience.
Basically, PTS accidents are initiated by small locas or
transients in the plant which lead to a situation where you
have an overcooling of the reactor vessel with either -- at
pressure or with a subsequent repressurization of the
reactor coolant system. This has the potential to impact
some of the preexisting flaws that are a result of the
fabrication of the reactor vessel. They have the potential,
these flaws, with a pressure transient and overcooling
transient have the potential to lengthen and deepen to the
extent that some, if they're at an initial critical length,
if you will, have the potential to go all the way through to
the reactor vessel wall.
If such a flaw proceeds through the vessel wall,
it has a good likelihood, if you will, of opening into a
very large opening in a reactor vessel and basically
seriously compromises the ability to core the cool -- cool
the core.
DR. APOSTOLAKIS: Is it the other cooling that
really is supervisor for the flaws going?
DR. HACKETT: If I could address that, it's
actually both, because we have looked at the situations for
BWRs where we just looked at the thermal shock portion. The
thermal shock without the repressurization is a much less
critical situation. The repressurization is where you're
going to drive the flaw through the wall and end up with
this big opening that Mark was describing, so you really
have to have both, but the initiator is the thermal shock.
And then, the subsequent significant driver is the
repressurization if that occurs.
DR. APOSTOLAKIS: And what time scale are we
talking about on there?
DR. HACKETT: This would be many seconds,
generally, that type of time scale.
DR. APOSTOLAKIS: Pretty fast.
DR. CUNNINGHAM: The repressurization.
DR. HACKETT: I probably would defer to someone
else on the --
DR. APOSTOLAKIS: If it's another cooling event
and repressurization; it's a matter of seconds.
DR. HACKETT: For the materials response. What I
was referring to was the materials response, and actually,
if you're going to get one of these large flaws that's
growing like Mark talked about, that's going to be even
milliseconds. If you get something initiated and the
repressurization occurs, it's going to pop through very
rapidly. It will be a catastrophic fracture from a
materials perspective. From the operational assessment, you
know, what drives that, I'm sure, is, again, much slower.
It may be more in the range of minutes or many minutes.
DR. CUNNINGHAM: Okay; if you were to have one of
these through-wall cracks propagate to a large opening in
the reactor vessel, one of the key issues then becomes the
containment response. You have a dynamic loadings, just the
fact that you're putting a severe load on the vessel, and
that could cause the motion of the vessel, which could
impact what's happening inside on the core itself and could
have an impact on penetrations and that sort of thing on the
lines that are attached to the reactor coolant system.
You're also obviously going to blow down the reactor vessel,
so you'll have pressure loadings. You have some potential
for dispersal of the core in this -- during the blowdown of
the reactor vessel.
On the other hand, this is an accident that's
driven by large amounts of water, so you have core -- you
have cooling capability from the water, and you also have
the potential availability of containment ESS. So the
containment situation is complex, to say the least, and it's
not very well understood at this point. And we'll come back
to that later in terms of what approach we want to take in
terms of reassessing the technical basis for the rule.
The question came up in the subcommittee of what
do PRAs tell us about these events? Typically, most PRAs
you see today will not have a PTS set of scenarios in there
with a frequency. There's a couple of reasons for that.
One is for most reactor vessels, PTS is a very low frequency
event, because the number of flaws that they would have are
not such that you would expect any substantial challenge to
the vessel from flaw growth. There are some special
characteristics of a few reactor vessels that make them more
susceptible to PTS accidents.
DR. APOSTOLAKIS: But is it just the number of
flaws or also the fact that you don't have this reactor --
DR. HACKETT: I think it's a number of factors. I
think Mark was correct that the principal materials driver
is the number of flaws that have been assumed in these
analyses before when we last talked to the joint
subcommittee was based on a distribution attributed to
Marshall from quite a ways back now. What we've found since
is that we don't tend to see things as severe from what
we're actually examining with vessels.
Another factor is the materials themselves are
generally significantly tougher than we give them credit for
in these analyses or have given in the past, because we
basically use lower bound approaches on materials toughness.
Another element would be -- and when we discuss these, I
guess we're probably getting a little bit ahead here, the
embrittlement predictive methodologies have been refined,
too, such that that piece is changing, and at least it's
still iterating, so I can't say for sure exactly where this
is going to end up, but at least preliminary versions of
that were predicting less embrittlement than what we had
seen before with -- and here, we're talking Regulatory Guide
199 revision two is what was used before. We're working on
a database that will go into revision three that at least
previously has indicated we might be seeing less
embrittlement. So there are a number of materials factors
that add up to this being a better situation.
George, from a PRA standpoint, I think basically
what we have is a conservative screening criteria on the
transition temperature so that if the screening criteria is
met, then, the contribution of PTS to risk for that given
plant would meet your screening criteria for not including
it in your cut sets, not including it in your PRA.
DR. APOSTOLAKIS: I don't know if there's a
presumption, but that's if you can meet the criteria, then,
your weight is 5x7-6.
DR. KRESS: It's lower than that actually?
DR. APOSTOLAKIS: Lower?
DR. KRESS: You would screen it out in your PRA
because of that.
DR. APOSTOLAKIS: Is it 1160 grams here?
DR. CUNNINGHAM: Not really; not really. We
discussed it but not in any serious way, if you will.
There are some plants that are more vulnerable, if
you will, to PTS accidents where you've seen PRAs done for
those -- that set of initiators, and the program that we're
running here in the office is taking advantage of that.
There's at least a couple of PRAs that licensees have done
to assess their PTS risk.
DR. KRESS: Those are the ones that are projecting
to exceed the screening criteria; is that right?
DR. CUNNINGHAM: Not necessarily exceeding it but
approach it at the end of life. And we'll come back to the
screening criteria, but it's a projection out to the end of
life, so many different PRAs, a snapshot of today, are going
to begin to reach their margin of what it will be at the end
of life.
DR. UHRIG: Presumably, this could be extended
further if a plant came in for licensing?
DR. CUNNINGHAM: Yes; that is correct. It's
whether it's a 40-year end of life or 60-year end of life;
that's right. In fact, there are a number of plants that
would not experience -- have an issue with PTS in a 40-year
life but could in a 60, so that's one impetus for doing this
work.
I'm going to turn back and talk about the
acceptance criteria a little bit now. PTS rule was
established in 1983, so it's been awhile since this was put
together. It was put together as an adequate protection
rule. We've been kind of dancing around the issue of
talking about the issue of the acceptance criteria.
Basically, there's an embrittlement screening criterion
that's built into the rule in the form of what they call an
RTPTS, and that specifies an acceptable temperature for
reactor vessels and for the inside wall of the reactor
vessels in these PTS events. There's a -- if the plant is
projected to exceed that temperature at the end of its life,
then, they have to do additional analyses to demonstrate
that the vessel can operate safely. Reg Guide 1.154
provides one method of performing that safety analysis.
Associated with that screening criteria, the
RTPTS, is a frequency of what was perceived at the time to
be an acceptable through-wall crack frequency.
DR. APOSTOLAKIS: I understand how that was
established.
DR. KRESS: With great difficulty.
DR. CUNNINGHAM: It was very complex.
DR. APOSTOLAKIS: But there is a calculation?
DR. KRESS: You have to go through the Monte Carlo
calculation to get the --
DR. CUNNINGHAM: Maybe I can come back to that in
a minute.
DR. APOSTOLAKIS: Okay.
DR. CUNNINGHAM: So it shows up in two places,
that the frequency of 5x7-6 shows up in a couple of places.
One is that it was one factor in determining what the
acceptable RTPTS was that's used -- that's specified in the
rule. I'll come back to that in a minute. It's also noted
in Reg Guide 1.154, if you do this additional safety
analysis, in effect, you're doing a PRA for PTS accidents,
and you calculate a frequency equal to or less than 5x10-6,
then, Reg Guide 1.154 says that that's an acceptable
frequency, and so, it's acceptable for the two-year
operation of the plant.
DR. KRESS: A question: is that a CDF, or is it a
LERF, or is it safety goals, or is it adequate protection?
Because those are the -- and how do you deal with it instead
of this is one sequence versus the whole set of sequences,
and how do you deal with defense-in-depth in this context?
DR. APOSTOLAKIS: So 5x10-6 is the frequency of
what? Of the pressure vessel failing?
DR. KRESS: This is a through-wall crack.
DR. CUNNINGHAM: Through-wall crack.
DR. KRESS: I mean, you could say that when you
fail the vessel in this way, you also fail containment, so
it could be a LERF.
DR. HACKETT: We are equating a through-wall to
core damage, basically, is the assumption.
DR. CUNNINGHAM: The current rule does that. If
you have a through-wall crack, it's equivalent basically to
a large opening in the reactor vessel and equivalent to core
damage.
DR. APOSTOLAKIS: Is it equivalent to LERF?
DR. CUNNINGHAM: Not in the rules.
DR. APOSTOLAKIS: But is there a possibility of
that?
DR. KRESS: That's one of the issues.
DR. CUNNINGHAM: It's an issue that's before us
today. It's not an issue in the terms of the rule.
I wanted to use this figure, to get back to
Professor Apostolakis' issue of how does the 5x10-6 fit in?
What's right here is -- all this is from an analysis -- this
figure is drawn from a Commission paper in 1982 where they
were trying to develop a technical basis for the original
rule, and what they did, in effect, was come up with a --
did a PRA to looking at all types of initiators, all kinds
of initiators, from an operating experience and from more
PRA type of projected initiators.
We looked at the frequencies of the initiators; we
looked at the thermohydraulic response to those initiators
and looked at the vessel response to the thermohydraulic
transfer. So this is built upon basically different
frequencies of different challenges combined with the
fracture mechanics information. The fracture mechanics
information was in the form of what we call probabilistic
fracture mechanics, so that the key fracture mechanics
parameters had uncertainties associated with them or
distributions with them.
So using all that information and creating a PRA,
they created this figure of basically frequency versus the
mean surface temperature RTNDT. The 5x10-6 is down here in
the corner. But you see it there, and that -- this is an
idea that that was an acceptable value of frequency; that
leads you to say that an acceptable RTNDT would be 210
degrees. However, they recognize that there is variability
in this curve, and the estimate of that variability was that
one standard deviation was 30 degrees in this direction.
They said that because we want to make sure that we want to
have high confidence that we don't have trouble with these
vessels, we're going to add two sigma to the 210 degrees, so
that makes that 270 degrees. So that's what's specified in
the rule is 270 degrees.
So it's -- there's a lot of pieces of information
that went into the specification of 270: 5x7-6 was one
piece of the information base, if you will, so that came
into play there.
It also came into play in the Reg Guide 1.154
space, where they basically said if you have 5x7-6, it's
acceptable.
DR. KRESS: In a normal distribution, two sigma
correlates to what? Ninety-seven and a half pockets, right?
DR. HACKETT: Yes.
DR. CUNNINGHAM: The bulk of the work that's going
on in the Office of Research today to revisit the technical
basis for the rule is to reformulate that curve. We're
reconsidering the types of initiators that could cause PTS
accidents; looking at operating experience since the
eighties and modern PRA information; reassessing the
frequencies of those challenges; redoing the
thermohydraulics using what we have today, our state of
technology thermohydraulics capability, and we're redoing
the fracture mechanics, the probabilistic fracture mechanics
reflecting a lot of different things, including the
materials research that Ed will talk about in a couple of
minutes.
So most of that work is the reformulation of that
curve, and this is going to be going on for the next year,
year and a half basically. One of the factors that we
wanted to get an early decision on was, okay, when you
reformulate that, is the use of 5x10-6 still the thing that
you ought to be using, or should you be using a different
value? A lot has gone on in PRAs since 1982 or 1983, and
would that lead you to decide that that value is the right
one or not the right one or what? And that's the purpose of
the paper today, the paper that we're sending up to the
Commission this month.
So anyway, basically, if you go back to the rule
and the underlying technical basis, there's a few key
assumptions that in the context of revisiting the screening
criteria. One is that basically, there's a through-wall
crack frequency of 5x10-6 was, in effect, acceptable in Reg
Guide 1.154. The through-wall crack was presumed to be
equivalent to a large opening and was presumed to be core
damage.
DR. KRESS: Along with that key assumption should
be a statement, because your actual criteria is the
transition temperature.
DR. CUNNINGHAM: Yes.
DR. KRESS: There should be a statement of added
two sigma confidence level.
DR. CUNNINGHAM: Yes.
DR. KRESS: I mean, in order to complete that
statement.
DR. CUNNINGHAM: Yes.
DR. KRESS: It meets that.
DR. CUNNINGHAM: That's a good point.
DR. KRESS: Okay, because if you don't say that,
it could be interpreted as a mean.
DR. CUNNINGHAM: You're absolutely right.
DR. KRESS: Okay.
DR. CUNNINGHAM: Another key underlying assumption
if you go back to the 1980s was that containment performance
wasn't really impacted by the pressurized thermal shock
event, even with a large opening in the vessel, and as we
were discussing a little earlier, given today's thinking
about how we do this PRA, this comes to the issue of LERF
and how we consider LERF here.
A key factor in the early 1980s, as I mentioned
earlier, was that this is an accident that occurs with a
great deal of water. There's water in the reactor coolant
system when you start; the core is intact; when you start,
it's not melted, and you've got -- probably, you have the
containment engineering safety features available, and that
has -- could have considerable impact on your potential for
a large early release.
I'm going to talk now about a couple of things
where basically what's changed since 1982, 1983 in terms of
the technology. Ed Hackett is going to talk for a few
minutes about what they've been doing in terms of materials
research over the last few years to give us a much better
understanding of the flaws that are in the reactor vessel
and the potential for vessel failure given a challenge.
MR. ELTAWILA: This is Farouk Eltawila from
research. I'd like to make a comment on the last remark by
Mark Cunningham that the scenario has adequate supply of
water. That's the assumption. I think we cannot make that
as an assertion right now; that if you have all the water,
but you have no guarantee that this water is going to go to
the core. You know, if you open the vessel on the side, the
water is going to go outside of the break and is not going
to cool the core. So we don't know how the core is going to
stay during that thermal shock. You can get the fuel and
the movement of the fuel itself can use the coolable
geometry of the core.
So I think the correct -- it is a question right
now; not a statement.
DR. CUNNINGHAM: Just to be clear, I was talking
about what was the -- in 1982, what was basically how they
established the basis of the rule. Farouk is right that
that's a question.
DR. KRESS: It may be reevaluated.
DR. CUNNINGHAM: Yes; I think we're planning to do
some re-evaluation, and we think some re-evaluation will be
needed in that area.
DR. KRESS: It seems like it's also a good
sequence that would be considered in our regression.
MR. ELTAWILA: That's exactly another point. I'm
concerned about it from the other discussion we had with
you.
DR. HACKETT: From the material perspective, I
guess we touched on a few of them briefly in response to
Professor Apostolakis' questions. One of the key drivers in
these analyses from a materials perspective has been the
core size density and location. We've made a lot of
progress on that over the last -- I guess it's 15 years now,
and a lot of that is summarized here. I guess I don't need
to go through this in detail, but we've done detailed
nondestructive and destructive examinations of actual
reactor vessel welds over that time frame, and by and large,
what that has indicated is in many cases, a lot of flaws at
places where you'd expect them like the clad based metal
interface over the welds. But these flaws are very small.
By and large, you're talking about things that are two
millimeters in critical dimensions or two millimeters to
four millimeters.
What the probabilistic fracture mechanics would
tell you from that is these are flaws that don't participate
significantly in contributing to the failure frequency
overall, so this has largely been -- to summarize, largely
been a good news story.
What we're focusing on is development of a
generalized statistical distribution, if we can do it, for
all US LWRs; if not that, at least hopefully something
that's separable by fabricator, like combustion engineering
or Chicago Bridge and Iron, B&W, that type of thing. And we
hope to be able to come back to the committee -- I think we
are scheduling a September type time frame. We have an
expert elicitation process underway right now to try to
finalize some of that, and I believe we'll have that largely
sorted out within the next month, so we'll be able to
present you with a lot more information on that in the near
future, but that one is a key driver.
On the next slide, the next one is irradiation
embrittlement correlations, and again, I mentioned that one
previously. We have a much larger and more robust database
to draw from now on predicting irradiation embrittlement
trends, which at least previously seems to be indicating
that we would be expecting by and large less embrittlement
than is predicted by the current Regulatory Guide 199 Rev 2
for PWRs in general. This, as I mentioned earlier, there is
still some iteration going on on the exact form of what this
predictive equation is going to be, but there's no question
it will be a better prediction in terms of, you know,
accuracy and robustness.
We have a database now that's expanded on the
order of four times over what we had available to draw from
when Reg Guide 199 Rev 2 was done. So that that was more
cause for optimism. For the first time, this project is
considering statistical distributions on the material
fracture toughness. What's been done before is largely to
draw from the ASME fracture databases that were developed
for K-1-C and K-1-A, which were lower bounding curves. That
was dealt with computationally in the probabilistic fracture
mechanics code but was never addressed statistically before.
By and large, there's an expectation that this will be an
improvement. I think we have felt all along that the
material fracture toughness is greater than, you know, we
were giving credit for in the analysis, so we anticipate
there is some improvement, some gains to be made there.
Statistical distributions; the next one down on
material chemistry and initial RTNDT; these are some things
we've done on plant-specific bases. I was involved with an
evaluation we did for Palisades when I worked in NRR, and
this type of thing was done for Palisades when we did that
review but has not generally been applied on a generic
basis, so we're addressing that, again, really rigorously or
as rigorously as we can for the first time in this project
on a more generic basis, and there are improvements to be
gained or strides to be made there, too, in terms of the
understanding.
We have a lot better understanding of variability
of plate and weld material. It isn't always good news in a
lot of cases, but we think we understand it better.
DR. POWERS: Let me come back to the K-1-C K-1-A
curve. Additional engineering uses these lower bound curves
to come out of the ASME code, and they're used for a variety
of our purposes, not just for pressure vessel codes, which
is everything else that's designed with steel --
DR. HACKETT: Right.
DR. POWERS: -- is based on these kinds of curves.
How are you portraying the use of a curve different from
those and its technical justification here?
DR. HACKETT: Yes; first off, that's a really good
point, because it is true in a lot of things we do; if we
can afford to deal with the problem deterministically and
bound it, and we get an acceptable answer out of that, we
try to go that route, sometimes even just from the
standpoint of resources alone. And increasingly, though,
obviously, these days, we're trying to extend the lives of
the plants and so on. We're trying to get more accurate.
Where we're actually going with that in this case -- maybe I
should back up and say what we've done. In the
probabilistic fracture mechanics code which I'll come to on
the next slide, it's called favor. What's been done
previously, since this is supposed to be a best estimate
analysis, we would take the ASME K-1-C and K-1-A curves and
then do a -- basically, a two-sigma variation on those
statistically and then derive a mean curve out of that is
what has been done previously in favor.
What we're doing now is actually taking advantage
of the availability of some -- I don't know the exact number
-- but many hundreds more data points that have been
generated for K-1-C K-1-A on these materials since then and
actually doing a statistical and mechanistic evaluation of
that for, you know, the curve we would use. But I think to
get exactly to the point of your question, what we would do
is use a mean curve, a best estimate curve of all of that
data which, before this, has not been done for these types
of applications.
There's also another technology that you may have
heard about that we're not using specifically on this task
right now called the master curve, which is a different way
of coming at this thing, but, you know, we're using I guess
what I'd say is a similar approach to try to get there.
We're trying to give the proper credit this time around for
the material fracture toughness as opposed to any kind of
bounds. I don't know if that quite gets at your question or
not.
DR. POWERS: I think I understand exactly what
you're doing, and I'm trying to think of what the skeptic is
going to say. The skeptic is going to come in and say, yes,
you've got all these data points. These were laboratory
specimens that people took a great deal of care in preparing
and what not; put them into their Instron machines and did
their whatever it is that they did to test these things.
Now, what about the material that you actually
have making up the vessel? Okay? That's an engineering
material. I mean, it got treated in ways that are somewhat
mysterious to us, and it was produced at a plant, a steel
mill that was different from where these specimens are.
DR. HACKETT: Right.
DR. POWERS: How do I know that this mean of all
these data points here is the mean for this material here?
DR. HACKETT: I think the short answer is you
don't, and it also is one of those things that -- we're
really wrestling with this hard with regard to this master
curve issue, which is another iteration beyond this. But
you're getting exactly to the point of, you know, you're
trying to -- we're trying to eliminate margin here to the
extent that it's justifiable, because we've believed that
there are many conservative elements to this. One of the
nonconservative elements is what you've just pointed out:
how well do you know this material? You've tested these
hundreds, and I'm going to argue here to tell me that this
fairly represents every vessel out there.
I think my answer would be it probably captures
that to a great degree, but I don't think you can sit up
here and say that you've got everything covered. There's
variability; there are issues that get to be plant-specific
with regard to exactly how the fabrication was done;
copper-coated weld wire, things we've talked to the
committee about before that I guess were addressed on the
previous slide, too, in terms of the variability.
I think the short answer is it's always going to
need -- there will always be a need to add some sort of
margin for covering that type of situation.
DR. KRESS: Since we have Dr. Powers and Dr.
Apostolakis back, this might be a good time for you to
return to your curve on the acceptance criteria. I think
that would lend something to this, because I think what
you're saying is you will -- in the new process, you will
take this variability into account and develop not just the
one curve but a distribution and apply some of your
conservatism in your acceptance criteria; we're using the
two-sigma. So, if you could repeat that part of it, it
might be useful to these two.
DR. POWERS: Well, let me just interject that I
think I understand that argument, and I was surprised that
didn't come up immediately. I'm equally surprised that you
didn't appeal to the testing that you've done on some vessel
materials extracted from other plants and say that, yes, you
get a validation on at least this point. I mean, I can't
validate it for any specific plant, but I can take a vessel
that's been through this engineering sort of thing, abuse
and what not, and I can see that indeed, it's not obviously
flawed using this mean data.
DR. HACKETT: Okay.
DR. POWERS: I mean, I quickly get you trapped and
trained to produce -- prove a null hypothesis, and you can't
do that, but you can show the preponderance of the evidence
in support of your approach, and then, you can come to this
kind of an argument and say look: rather than putting the
conservatism on each of the elements that build up to this,
I'm going to put a conservatism on the final product.
DR. HACKETT: Right.
DR. POWERS: And I think that's where I certainly
become much more comfortable, because when you do it on the
individual elements of the thing, I don't know how
conservative you are at the end.
DR. HACKETT: Right.
DR. POWERS: Especially when you tell me that in
some cases, there are some nonconservatisms built into this
thing that creep into the process. They're sometimes
unquantifiable, and so now, have you built in enough margin
on it? Here, I can see did you build in enough margin.
DR. HACKETT: Right; that gets right to the heart
of it. I guess what I can say in a more global sense going
directly to the question that we've been able to show over
time is that what we're talking about in this process from
the materials perspective is this indexing to get to a
fracture toughness curve with sharpy specimens and RTNDT to
get to fracture toughness, your question being that if you
went there directly, and you're testing these specimens, you
know, maybe even large fracture toughness specimens from
actual vessel materials, the answers that we see over and
over again is by and large, those are better numbers than
what you'd get with the index, and the indexing process
generally is going to yield a conservative estimate of what
you'd get for the real toughness.
DR. POWERS: Sure.
DR. HACKETT: And that the real toughness of the
materials is generally greater.
DR. POWERS: I had occasion just recently to see
-- to witness a discussion of the materials issues
associated with the steel that was used to fabricate the
Titanic and the issue of fracture toughness there. There
had been some studies done some time ago, I think, arguing
that the fracture toughness of the steel in the Titanic was
low in salt brine temperatures, and these investigators said
that those tests probably weren't fair because they utilized
fairly high strain rates in looking at the toughness. I
think they used traditional sharpy tests with a hammer and
things like that.
And they went through and did studies at lower
strain rates under the argument that those particular carbon
steels were very strain rate sensitive. Are these steels
strain rate sensitive?
DR. HACKETT: They are; they absolutely are, and
that's why you've got consideration for both K-1-C and K-1-A
where, you know, the A is the arrests where the dynamic
value that you're getting at so that when you do the
probabilistic fracture mechanics, it is run in two loops
where you're going to look at the initiation event, which
would typically result from the thermal shock. Then, you're
going to see if that -- then, you run another loop to see if
that crack arrests. As you could imagine, that's a less
precise technology than just dealing with initiation, but
that is a factor, a major factor.
Should we revisit the --
DR. CUNNINGHAM: Do you think we should revisit
this or --
DR. KRESS: Yes, I think you might want to say a
few words about that still.
DR. CUNNINGHAM: The figure that's up here is a
figure from a 1982 Commission paper showing the information
they had at the time on frequency of challenges -- the pink
curve is a composite of the frequency of challenges; the PTS
challenges, the thermohydraulic response of the reactor
coolant system to that; and the probabilistic fracture
mechanics understanding at the time. It leads you to
generate a curve of frequency versus RDNDT for the surface,
reactor vessel surface.
What we're doing today in the program that's kind
of behind all of what we're talking about is a reformulation
or a revisitation of that curve; new PRA information; new
information on the initiators that could cause PTS events;
the frequency of those initiators; the state of technology
understanding of the thermohydraulic responses and the
information that Ed has been talking about on material
responses is all being lumped together, if you will, in a
model that will give us a new one of those curves.
I didn't mention it the first time through, but
part of this also is a more explicit analysis and
consideration of the uncertainties in that curve.
DR. KRESS: So you get a distribution rather than
just one value.
DR. APOSTOLAKIS: The distribution will be
vertical, right?
DR. KRESS: No, it will be both.
DR. APOSTOLAKIS: On the frequency.
DR. KRESS: It'll be on both.
DR. APOSTOLAKIS: No, the RTNDT is the independent
variable. It will be just a frequency. You know, what is
the frequency of exceeding a certain value of RTNDT or being
-- it can't be on both.
DR. WALLIS: Are the thermohydraulic
investigations are such that they lend themselves to
probabilistic output?
DR. APOSTOLAKIS: What does that mean, Graham?
DR. WALLIS: Well, I mean, are they just going to
give you a heat transfer coefficient, or are they going to
give you a distribution of heat transfer coefficients with
probabilities?
DR. CUNNINGHAM: My understanding is that it will
be in the form of some sort of a distribution, if you will,
of not just a point estimate.
DR. WALLIS: So they're going to give you
information in the right form to go into --
DR. CUNNINGHAM: Yes.
DR. WALLIS: -- a probabilistic type analysis.
DR. CUNNINGHAM: Yes.
DR. WALLIS: Thank you.
DR. CUNNINGHAM: That is correct. We talked about
-- Nathan Siu talked about that at one of the subcommittee
meetings. He'll probably come back and talk about it at the
next subcommittee meeting as well, the mechanism by which
we're doing the uncertainty analysis for this.
DR. APOSTOLAKIS: But make sure that you're
presenting it correctly. The horizontal axis is the
independent variable, and then, you have the uncertainty.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: It's very similar to what we do
with the consequences, the risk curves.
DR. KRESS: But you might also mention the
acceptance criteria and the two-sigma part.
DR. CUNNINGHAM: Okay.
DR. APOSTOLAKIS: And then, you have the family of
curves, and you select two-sigma up and down?
DR. POWERS: Of course, I'm going to say, well,
Tom, why two? Why not one and a half?
DR. KRESS: That's a really good question.
DR. APOSTOLAKIS: Why not a percentile? I mean,
we do that all the time.
DR. KRESS: Both those are really good questions,
and I think part of this whole meeting is how do you arrive
at the right acceptance criteria for this?
DR. APOSTOLAKIS: When you deal with frequencies
of events, I mean, we rarely work with standard deviation
and take in two sigmas and so forth. We usually go with a
95th percentile or some appropriate percentile. I mean, the
standard deviation makes much more sense when you deal with
normal distributions and so on. When you're dealing with
highly skewed distributions, it's not --
DR. KRESS: This is likely to be normal, this
particular one.
DR. APOSTOLAKIS: Normal? It will be normal? The
frequency? I would be surprised. I would expect it to
be --
DR. POWERS: I bet it looks an awful lot like a
normal by the time it's done.
DR. KRESS: Yes; this particular one will.
DR. APOSTOLAKIS: It will have a long tail.
DR. POWERS: But it will be interesting to see,
and I agree with you that the standard deviation is a useful
term when it looks almost normal.
DR. APOSTOLAKIS: Yes.
DR. POWERS: Or log normal, and if it's at all
distorted from that, then, the standard deviation is pretty
difficult to interpret.
DR. APOSTOLAKIS: Yes.
DR. KRESS: A more general term would be a
confidence level, so you could always use that.
DR. APOSTOLAKIS: Sure; well, the percentile is a
confidence level.
DR. KRESS: Yes, yes.
DR. POWERS: But it still begs the question: you
take 95 percent --
DR. KRESS: Yes; it's exactly the same question.
DR. POWERS: Why not 80?
DR. APOSTOLAKIS: Well, let us face that when the
time comes. Let us make sure we have a solid uncertainty
analysis first.
DR. POWERS: Sure.
DR. KRESS: Then, we'll figure out --
DR. APOSTOLAKIS: They have a lot of work to do
until that time, but they seem to be on the right track.
This is the way to do it.
DR. KRESS: Oh, yes.
DR. APOSTOLAKIS: This is the way to do it.
DR. CUNNINGHAM: The point we were discussing
earlier is not so much what should be done today but what
was done in 1982 --
DR. APOSTOLAKIS: Yes.
DR. CUNNINGHAM: -- or in the early 1980s, and the
5x10-6 value that underlies the rule is associated with a
mean RDNDT of 210 degrees, mean value.
DR. APOSTOLAKIS: So this is the frequency, now?
The curve gives me the frequency, which is 5x10-6 --
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: -- of the RTNDT being that value
or less?
DR. CUNNINGHAM: Yes, that's probably right, yes.
DR. APOSTOLAKIS: That value or less.
DR. CUNNINGHAM: That's right; now, so, the RDNDT
value that they choose -- they came up with was 210 degrees.
DR. APOSTOLAKIS: Yes.
DR. CUNNINGHAM: They, at the time, associated --
did some fairly simple analysis of the uncertainty in the
materials properties, okay? And they estimated that one
standard deviation in RDNDT was about 30 degrees, so they
said, well, to be highly confident that this screening
criterion will be conservative and that sort of thing, let's
add two sigma to the 210 degrees.
DR. APOSTOLAKIS: Right.
DR. CUNNINGHAM: And that's how they arrived at
270 degrees.
The story of how they arrived at 270 is actually
much more complex than that, but from the context of the
probabilistic element, that's how it seemed to come into
play at least from what I can understand.
DR. APOSTOLAKIS: Sure.
DR. CUNNINGHAM: Seventeen or 18 years later.
DR. APOSTOLAKIS: Okay; although, of course, the
uncertainties here are primarily model-driven, I would
venture to say.
DR. CUNNINGHAM: I agree.
DR. APOSTOLAKIS: So my idea at the time really
dealt with the parameter.
DR. HACKETT: Yes, yes.
DR. APOSTOLAKIS: You guys are going to deal with
model uncertainty.
DR. HACKETT: Absolutely, and I think we did
mention that explicitly the last time we met with the joint
subcommittee, but we are considering -- in every case where
we're looking at uncertainties, we're looking at both the
epistemic, the state of knowledge and uncertainty, and the
aleatory uncertainty that the randomness uncertainty on the
materials perspective is pretty obvious. Dr. Powers is
talking about transition fracture. It's endemic to that.
For these materials, they are strain rate sensitive. You
see a huge variability in material response that's just a
randomness type situation when they're in transition.
But then, there's also -- we're using RTNDT to
model how you get the fracture toughness, so we have to
address the model uncertainty associated with that
parameter, and that's just one example.
Another example is the top of the slide; fluence
calculations previously had been done on taking the end of
life maximum fluence, applying that everywhere throughout
the belt line and then assuming there's basically an
uncertainty on the order of plus or minus 20 percent is what
has historically been done or what Mark referred to back to
the 82-465 paper.
So we're trying to get a lot better than that this
time. We are looking at, for the plants that are going into
the basis for the generic re-evaluation here, we're looking
at plant-specific fluence maps and considering the variation
of the fluence as a neuphaly around the vessel. There's
also a new draft regulatory guide, which will be something
else we'll be coming and talking to the committee about
later this year, which has got a more defined methodology
for getting there, let's say, for fluence, so we are dealing
with fluence uncertainty in a more rigorous way this time,
too, and are accounting for the fact that, you know, the
belt line doesn't see the maximum fluence everywhere.
The last piece that we cover here is the
improvements in the fracture mechanics methodologies. The
code is called favor. It's largely an evolutionary
treatment from a previous code that was called visa, which
was developed basically here. Ron Gamble is here
representing the industry. Ron, I think, deserves a lot of
the credit for initiation of that and Jack Strosnider, who
is currently in NRR. That code has evolved into what is now
favor. And then, you see these elements here that are now
addressed that weren't addressed previously in a lot of
these analyses.
Just to touch on a few residual stress
distributions, the basis for what went into 82-465, that
type of thing wasn't considered; doesn't tend to be a big
factor, but it is now factored in. The mathematical
solutions for semi-elliptical surface flaws have been
refined over time. Those are now included in there as are
solutions for subsurface flaws, to the effect -- to the
extent that they might participate, you know, in
contributing to the failure frequency. So there has been a
significant amount of fracture mechanics technology
improvements built into the code that will make things more
accurate than they have been before, too.
DR. POWERS: Do you have to assume orientation
from the flaws?
DR. HACKETT: You do assume orientation, and
typically, you assume the worst orientation; in this case,
as an axial-oriented flaw in the weld, whether the weld is
-- this has been a debate also, but if you have an axial
weld, you assume there is an axial flaw in that weld, and
that's typically the worst case.
DR. POWERS: And are you going to do worst case
assumptions here in this new analysis?
DR. HACKETT: There will be both addressed from
the standpoint that, you know, the circumferential situation
is usually much less limiting, but the idea would be that if
you have a circumferential limited situation, you would
postulate a circumferential flaw, so that not being worst
case, because you could also say, well, I'll take a
circumferential flaw, and I'll still -- I'll take a
circumferential weld, and I'll still put an axial flaw in
it.
So we're not doing that. I mean, these are the
kinds of things that have been done before that we won't be
doing now. This analysis specifically will still be driven,
I believe, by an axial flaw in an axial weld situation, you
know, which could exist. And then, when you look at the
question back to this idea of the large opening in the
vessel, what's typically going to happen is you're going to
assume that that crack could open up until it hits the
boundaries. If you're talking about a vessel that was
fabricated from plates, you would basically open that entire
weld seam until it hits the free surfaces where it's joined
to the, you know, the upper and lower circumferential
plates, so that we're not -- we're trying deliberately not
to make the worst case assumptions in this particular
analysis.
DR. POWERS: You're going to think I spend an
inordinate amount of time looking at sunken ships. I did
have a chance to look at a pressure vessel from the
Skipjack, which was a submarine that sunk under high
pressure, in high pressure water, and the pressure vessel
failed not at the welds but at the bulk plate. So you're
also looking at the bulk plate.
DR. HACKETT: We are also doing that. I guess
that's a point I didn't emphasize on the flaw distribution
piece. The advanced NDE examinations in addition to
capturing the weld, and for obvious reasons, they were
focused on the welds, but we also scanned at least a plate
thickness or two to either side of the weld to pick up flaw
distributions that would be in the heat-affected zone
immediately adjacent to the weld or in the base plate
itself.
The base plates themselves, assuming, you know,
especially nuclear grade or Navy grade QA, are generally
much less defected than you would ever expect of the weld.
The welds are -- the welds or the heat-affected zone are the
areas you would focus on as a metallurgist, but you raise a
good point. That's not to say you couldn't end up with a
flaw in the plate, and that could be a problem, so that will
be addressed also.
DR. CUNNINGHAM: Ed was talking about one key
piece of the information base that's changed since the
1980s. Another key piece we wanted to talk about today was
in terms of Commission guidance in the area of user risk
analysis. Basically, I'm going to touch on four that are
particularly relevant: the safety goal policy statement,
the rules that were established for station blackout and
ATWS; the backfit rule and Reg Guide 1.174.
We discussed the safety goal policy statement any
number of times before this committee and with this
committee. I think in the context of revisitation of the
PTS rule, I think the key piece is the middle bullet right
there, that basically, in 1990, the Commission established a
subsidiary core damage frequency goal of 10-4, and that
starts us down a path, if you will, along with other things
towards thinking about allocation of reliability goals or
acceptable core damage frequencies.
Station blackout and ATWS rules actually were done
in the mideighties.
DR. APOSTOLAKIS: This is an interesting point,
Mark, which really brings up the dangers of allocating risk
goals. I believe that this 10-4 is indeed a goal for core
damage frequency under the assumption that the containment
will have a condition of failure probability of about 0.1.
Is that correct? In other words, it's not an absolute goal.
If a core damage event could lead simultaneously to the
failure of containment with probability 1, I don't think
that was the intent of this, intent of this.
DR. BONACA: That is correct, because the
relationship on individual risk which was 10-6 is tied to
that performance of containment.
DR. APOSTOLAKIS: Yes; in other words, this one is
to be taken together with the LERF criteria.
DR. CUNNINGHAM: You know, historically, the
original safety goal had goals for individual fatality risk
and latent fatality risk and proposed in effect a large
release goal of 10-6. That was in the mid 1980s.
DR. APOSTOLAKIS: Right.
DR. CUNNINGHAM: In 1990, they approved this one
in addition, and I guess at that point, they still had in
mind, I think, that there would be margin between core
damage and containment.
DR. APOSTOLAKIS: Exactly.
DR. CUNNINGHAM: Main containment failure.
DR. APOSTOLAKIS: Exactly.
DR. CUNNINGHAM: I think that's right.
DR. APOSTOLAKIS: Exactly.
DR. CUNNINGHAM: That's fair.
DR. APOSTOLAKIS: So in our case here, where that
margin may not exist, that goal probably is not applicable.
See, this is the problem with surrogate criteria, that you
are deriving them under certain assumptions.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: And if those assumptions are not
valid in a particular situation, then, you have to go back
to the original QHOs or whatever the case may be.
DR. CUNNINGHAM: Just to be clear, this 10-4 was
for the complete set of initiators.
DR. APOSTOLAKIS: Given the margin of the
containment.
DR. CUNNINGHAM: Yes, given all these other
things.
DR. APOSTOLAKIS: Yes.
DR. CUNNINGHAM: That is correct; the blackout and
the ATWS rules took another tack and said for these
particular initiators of core damage, we're going to have
goals, if you will, in rough terms, of having those be on
the order of 10-5 per year per reactor.
DR. APOSTOLAKIS: Yes; so, they reduced it there.
DR. CUNNINGHAM: That did it, too, and built into
that was their own analysis of containment performance and
that sort of thing associated with those.
DR. KRESS: The other point, George, is this was a
safety goal which was never intended to be a replacement for
adequate protection.
DR. CUNNINGHAM: Correct.
DR. APOSTOLAKIS: True.
DR. KRESS: Now we're talking about an adequate
protection rule in the PTS.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: Well, that's another issue here
that was confusing me a little bit, because we have
numerical guidance when it comes to goals.
DR. KRESS: Yes; so, how do you mix with adequate
protection with the safety goals?
DR. APOSTOLAKIS: Well, of course, if you had --
if the Commission follows our latest recommendation on these
things, then, it would be a piece of cake.
DR. KRESS: Yes; it would be duck soup.
DR. POWERS: And hell freezing over.
[Laughter.]
DR. KRESS: Yes.
DR. CUNNINGHAM: Dr. Kress' comments were a good
lead-in to my next slide, which talked about the fact that
the backfit rule and the reg analysis guidelines were
established in the eighties as well, and at some point,
there became a clear differentiation between -- on the
subject of when you can use cost-benefit analyses. PTS rule
was an adequate protection rule. For those rules today,
cost is not a consideration in the establishment of the
rule, costs and benefits. Blackout rule and the ATWS rules
were cost-beneficial safety enhancements where they were --
where costs and benefits were considered. So it
differentiates those.
The backfit rule today also now has an analysis
method for cost-beneficial safety enhancements that uses the
safety goals and looks at the potential benefit in terms of
a first screen of saying how much core damage frequency
would be reduced, and it builds into some consideration of
containment performance as well, so there is a -- a piece of
the logic is put into place with the safety goals and the
backfit rule and the reg analysis guidelines for certain
types of backfits.
Reg Guide 1.174 came along obviously a few years
ago and from a somewhat different direction. We're looking
at potential changes to license amendments to perhaps allow
a reduction in burden and that sort of thing. So we --
obviously, as the committee well knows, we have -- in 1.174,
there was a set of principles established on how you would
make those changes, allow those changes. It brings into the
issues that we've -- one issue we've talked about,
defense-in-depth, kind of explicitly into the consideration;
safety margins and then how much acceptable -- what the
changes in risk would be, what changes in risk would be
allowed.
All of those have come into play since the rule
was established that helped guide us in terms of how we
would think about revisiting the probabilistic aspects of
the rule.
DR. APOSTOLAKIS: So the rule is still viewed as
an adequate protection rule?
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: So 1.174 doesn't help you very
much. It's used for background but --
DR. CUNNINGHAM: Well, let's talk about the
approaches.
DR. APOSTOLAKIS: Okay.
DR. CUNNINGHAM: Because given all of that, given
the rule as it is, given all this difference in guidance,
we've talked about, well, what are different ways that you
could think about revisiting the rule, and what are some of
the characteristics of those? So we came up with, on this
slide, anyway, four possible approaches for revisiting the
screening criteria. One is leave it the same and just
acknowledge that it was based on what it was based on and
kind of move forward. The second one was, you know, let's
-- you know, ATWS and blackout and PTS are all, in one
sense, similar types of challenges. Maybe we should let the
core damage frequency for those be similar, so you would, in
a sense, allow the acceptable frequency for PTS to float up
a little bit to be in line with the others. That's a
possibility.
DR. UHRIG: A factor of two?
DR. CUNNINGHAM: Something like that, yes.
Another possibility is that we would use the basic
innocence. The principles and the approach that are in
1.174 -- that are documented in 1.174 are how you would
address changes to rules. They're focused on license
amendments, and then, you would -- your potential for
applying those same rules to a rule change, so that would
be, then, that you would look at the change in this would be
driven by consideration of the defense-in-depth safety
margins and small changes in CDF and LERF.
The third and the fourth approaches here are
variance of the same general approach. The third approach
would say that you're going to explicitly keep apart the CDF
and LERF. The fourth would say we don't have enough
information, in effect, to make that differentiation, and
therefore, we're going to treat them as the same.
DR. APOSTOLAKIS: The fundamental question -- I
understand, you know, looking at the principles of 1.174.
You're right that, you know, we spend a lot of time thinking
about those and so on, but the fourth -- it's not really a
principle -- well, it is. The principle that says that the
changes should be small.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: The question is whether that
applies here, because now, you're not talking about goals.
You're talking about adequate protection.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: Okay? And the fundamental
premise of the principle perhaps should be revisited,
whether -- when it's an issue of adequate protection, you're
allowed to increase the CDF for your convenience. So, I
mean, yes, sure, as long as you think about it, then, there
may be something -- I don't know what defense-in-depth you
have here, by the way. You don't seem to have much. You
have margins but not defense-in-depth.
DR. CUNNINGHAM: Well, that's one of the key
issues when you revisit this of how do you bring
defense-in-depth into this.
DR. APOSTOLAKIS: Well, you can't. When you talk
about the vessel, I mean, it's -- although you might have
safety margins for formal defense-in-depth.
DR. POWERS: I think -- yes; I think the vessel in
itself constitutes an element of the defense-in-depth
strategy.
DR. APOSTOLAKIS: Right; but if that element
fails, you see that a lot of your other defense-in-depth
measures follow suit.
DR. POWERS: The flaw is in order to be an element
of defense-in-depth, you've got to have -- you can't have
failure of the vessel ipso facto lead to failure of the
containment, the next element of defense-in-depth.
DR. APOSTOLAKIS: Well, that's what I was saying
here.
DR. POWERS: And that's what the fourth one says.
It's not obvious to me that's the case, but that's what the
fourth one says. And I don't know why they would say that.
DR. APOSTOLAKIS: Because the uncertainties are
very large in it.
DR. KRESS: And the way you invoke
defense-in-depth principles there is you look at the
uncertainty in developing the vessel failure, and you
recognize that in -- there is also a potential for it to
lead to containment failure, so it's a thing that impacts
both CDF and LERF. Therefore, you -- in my principles that
I outlined at one time, you make your acceptance criteria at
a very high confidence level on that particular one.
DR. APOSTOLAKIS: And lower.
DR. KRESS: And lower.
DR. APOSTOLAKIS: Yes, sure.
DR. KRESS: So you start with an overall
acceptance criteria; in my mind, that would have to be an
acceptance criteria that would be equivalent to something
called adequate protection, not to safety goals. And you're
going to allocate a portion of that to this particular set
of sequences, but since this particular set of sequences
impacts both containment and CDF --
DR. APOSTOLAKIS: Right.
DR. KRESS: -- you make the allocation of that
part of it much lower and yet a higher confidence level.
Now, those are qualitative principles.
DR. APOSTOLAKIS: Yes.
DR. KRESS: How you convert those into quantities,
I don't know -- I'm not quite sure yet.
DR. APOSTOLAKIS: Right.
DR. KRESS: But that would be the way I would
approach this sort of thing.
DR. APOSTOLAKIS: And I think that's a reasonable
way to approach it, and I think a lot of that is in the new
-- not policy; the new work that is coming out of Sandia
under your sponsorship.
DR. CUNNINGHAM: This is the framework for --
DR. APOSTOLAKIS: Yes.
DR. CUNNINGHAM: -- how we change Part 50?
DR. APOSTOLAKIS: And if you remember, they had
the core damage frequency and containment, but then, there
was an allowance for situations like earthquakes, for
example, where there is such strong coupling there that you
really have to look at the sequence separately, because, you
know, defense-in-depth measures have been compromised.
DR. KRESS: Yes.
DR. APOSTOLAKIS: And I think this is a mostly
similar --
DR. KRESS: This is the case.
DR. APOSTOLAKIS: -- sequence like that.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: But there are fundamental
differences here, which I'm sure Mark now appreciates or
even when you walked into the room probably appreciated that
we are not talking about goals anymore; we're talking about
adequate protection, so, you know, whatever principles you
invoke, I mean, 1.174, we will really use the goals, right?
DR. CUNNINGHAM: Yes, yes, that's right, exactly.
DR. APOSTOLAKIS: And the second one is that the
surrogate goals may not be applicable anymore, and the third
point, Tom, I think, again, when we use the words
defense-in-depth, we're using them too loosely sometimes.
DR. KRESS: Sometimes.
DR. APOSTOLAKIS: We're talking about
defense-in-depth in the context of the big picture,
preventing radioactivity from being released, or
defense-in-depth regarding the failure of the vessel, right?
And I think when it comes to the failure of the vessel, you
really don't have many options for defense-in-depth. Now,
for the bigger picture, you might still have something that
you may be able to do, right? So defense-in-depth really
can be interpreted only in a certain context which would
make the ATHENA people very happy, right, because context is
everything, right, Mark?
DR. CUNNINGHAM: Yes, sir.
DR. APOSTOLAKIS: No, you were not paying
attention so you --
DR. CUNNINGHAM: You're getting into a lot of the
issues that we're facing, and I'm just -- we recognize that
a lot of things, this is kind of uncharted waters for us,
and we're taking 1.174, which is for license amendments,
safety goal information, and we're trying to -- see, that's
probably our best set of guidance in terms of general
thinking on how to use risk information and then extending
that, if you will, into a different environment.
DR. APOSTOLAKIS: Are we going to have a chance to
look at the paper you're going to present to the Commission,
the final version?
DR. CUNNINGHAM: Before we send it to the
Commission?
DR. APOSTOLAKIS: When do you send it?
DR. CUNNINGHAM: It's due to the EDO in about two
weeks.
DR. APOSTOLAKIS: I guess we will not have a
chance, but can I make a suggestion?
DR. CUNNINGHAM: Okay.
DR. APOSTOLAKIS: That you actually elaborate on
these points we just discussed, and it seems to me that as
you know, certain stakeholders like the Center for Strategic
and International Studies have recommended to the Commission
that adequate protection be quantified, okay? And there has
been extreme reluctance on the part of just about everyone
to do that.
I think that reluctance and that recommendation
were really made at a high philosophical level, and as is
usual, you know, with philosophies and philosophical
approaches, engineers tend to be very impatient with those.
I think it's time that we start pointing out that there are
practical issues that we would have an easier time handling
if we had such guidance, such goals, such numbers.
DR. KRESS: And this looks like an ideal case.
DR. APOSTOLAKIS: And this is really an ideal one,
that if you guys had a quantified statement of adequate
protection, then, your life would be much easier, but you
don't, and that quantified statement, again, I don't want to
upset Gary Holahan -- it can't be one number. It's a whole
process and blah, blah, blah, but still, there will be a
number here just as a goal that's going to be in the context
of 1.174, and unless we can see these practical means, you
will never get anything, anywhere with a safety goal and
adequate protection letters and so on.
DR. KRESS: I'll tell you what will happen is that
the numbers in 1.174 will become surrogates for adequate
protection.
DR. APOSTOLAKIS: Exactly, exactly, and the same
risk Sandia has when they developed the other work on --
what do we call that?
DR. CUNNINGHAM: Framework for --
DR. APOSTOLAKIS: The framework, yes; see, they
have nothing else to go by, so they have to use some
numbers, and what is it they use? The goal numbers. And
before you know it, those will become adequate protection,
which is really not checking out the regulations, right? So
that's a second example.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: We have two examples.
DR. CUNNINGHAM: We see what we're doing here as a
test bed for at least a couple of things. One is that we're
going to test out the framework that you've been talking
about and see how that works in the context of this rule.
DR. APOSTOLAKIS: Right.
DR. CUNNINGHAM: Because if it works, we've got to
find our evidence on this rule, and then, there could be a
bunch of other rules that have to go through that same
process. So this is a test of that system. But you're
right; it's also a test of how do you deal with an adequate
protection rule in the context of a lot of other things, a
lot of other policies that have been focused somewhat
differently.
DR. APOSTOLAKIS: So there will be a discussion in
the Commission paper on this subject.
DR. CUNNINGHAM: I think the intent of the way we
see the Commission paper today is to lay out the issues.
DR. APOSTOLAKIS: This is a big issue.
DR. CUNNINGHAM: Yes, oh, yes.
DR. APOSTOLAKIS: Yes.
DR. CUNNINGHAM: And you're, of course, skipping
ahead, but this is one of the key issues, and we have to
tell the Commission that we're not intending to give them
the answer today but tell them this is an issue that we're
going to have to work out in the -- you know, X amount of
time. I'm not sure how soon; and discuss it with the
committee and go back and just tell them this is an issue in
this PTS --
DR. APOSTOLAKIS: Yes.
DR. CUNNINGHAM: -- revision.
DR. APOSTOLAKIS: I think they should be fully
aware of this fact, adequate protection and goals are two
different things.
DR. CUNNINGHAM: And as you say, this is a place
where all of that -- the theory comes together in a
practical application to a rule change where those of us on
the staff have to decide where do we want to go with this
rule change.
DR. APOSTOLAKIS: Now, if you look at these for --
I know you have discussed each one, but maybe we can do that
as a group. Is it fair to say that there are really two
sets here: the top two and the bottom two, the top two
really dealing with the absolute value of the core damage
frequency due to PTS and the bottom two, where you look at
it as a change.
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: And the question is which one is
philosophically right? Which one do you think -- do you
have any preference at this point?
DR. CUNNINGHAM: My personal preference, I think
of this in terms of the change.
DR. APOSTOLAKIS: So one of the bottom two.
DR. CUNNINGHAM: One of the bottom two is --
DR. KRESS: My personal preference is the top two.
DR. APOSTOLAKIS: Okay; Dr. Bonaca?
DR. BONACA: I pointed out really before that
there is a fundamental confidence in some of these; I mean,
they're offering us options, but if I look at option two and
option four, they lead you in different directions.
DR. APOSTOLAKIS: Yes.
DR. BONACA: One of these is a direction of the
basic uncertainty that you have on the containment
performance. You have to restrict your CDF criteria to
10-6, and that, to me, seems to reflect the whole
uncertainty we have on the phenomenon associated with PTS
issues. The option two you're proposing out of the blue
just for consistency with other rules, which makes no sense
-- it's 10-5, which is not a different reaction. And so, I
don't understand why you make differences in the
understanding of the PTS in theory essentially -- are the
basis of this initiative will lead you towards option two.
I just don't understand it.
And I'm only saying this for the purpose of saying
do you really want to present number four as an option to
the Commission? I don't think --
DR. CUNNINGHAM: Again, to back up, one of the
things we've changed in the context of the paper or will be
changing in the context of the paper is not to lay ideas out
as options for the Commission to consider. These, based on
what's happened over the last 10 or 15 years, here are
possible ways that you can -- or pieces of guidance that you
could use to adapt it and to consider this rule change. So,
I'm not going to argue that the second one is a very strong
argument; it's just historically, in other circumstances,
the Commission has said that sequences -- core damage
frequencies on the order of 10-5 are acceptable.
There are a lot of down sides to this, and
containment performance is one of them. And the
uncertainties and our understanding of the uncertainties on
these accidents is different. And all of that is arguments
not to take that approach.
DR. APOSTOLAKIS: In other words, if you don't
present these as distinct options, you may say, well, here
are some ideas, and one can pick, you know, something from
four and combine it with two. In other words, I can't have
in my thinking the fact that, you know, there were
frequencies discussed for ATWS and station blackout, but I
also worry about the fact that CDF and LERF here are really
the same thing, right?
DR. CUNNINGHAM: Yes.
DR. APOSTOLAKIS: I mean, I can -- so if you
present it that way, then I think Dr. Bonaca's concern is
mitigated.
DR. BONACA: Yes, it is; you know, again, and
Professor Seale made a comment that I thought was
interesting. They're different animals, too. In ATWS and
station blackout, you really probably have differences
between the CDF and LERF. For this, the PTS, I really need
to be more and more convinced that it is very high. Even if
you went to option C, it would be very hard to demonstrate,
to demonstrate, you know, within a certainty a difference
between CDF and LERF. So that makes a big distinction
between ATWS and station blackouts.
DR. APOSTOLAKIS: And the staff says, you know, we
recognize it and we'll look at it. We don't have to be
making a choice now.
DR. KRESS: The point of bullet two to me is that
you are allocating an overall criteria to a set of
sequences. It's an allocation process. And I'm saying --
when I say that's an appropriate thing to think about, how
do you allocate, and you need to consider that, things like
what you're talking about: what's the uncertainty in these?
What's the uncertainty in the other? Does it impact both
CDF and LERF and how much? Those are things you ought to
think about.
I think this is one of the big issues here is how
do you allocate your risk acceptance among the sets of
sequences? That's one of the big issues.
DR. CUNNINGHAM: Yes, and that's, in the context
of the blackout rules and things, it's just a straight
allocation on core damage frequency is overly simplistic --
DR. KRESS: Yes.
DR. CUNNINGHAM: -- because of the differences in
the uncertainties, including the differences in our
understanding of what containment performance would be.
It's in differences in the initiator frequencies. The
blackout starts with an event, a series of failures. You
can think of a PTS thing as it happens, and you've failed
the vessel, and you've melted the core. So the frequency of
the challenge is different. For all those reasons, it's an
overly simplistic approach to take, I think, and again,
we're not arguing to go there, but historically, we have
done things like that in the eighties. We have done some
sort of an allocation of core damage frequency, in effect.
It's happened piecemeal and things like that, but it's
happened. So it's something to think about.
DR. KRESS: What you need is an overarching
principle on how to deal with it.
DR. WALLIS: Another idea -- say you're looking at
the screening criteria. Rather than just looking at
options, I think you ought to explain why the changes to the
core, for what reason or on what basis. Because if
something is wrong with what we have now --
DR. KRESS: Yes.
DR. WALLIS: -- then, we would have a solution to
that. That should be clear before you launch into picking
one of these options.
DR. APOSTOLAKIS: So it seems to me you can jump
to 20 or 21.
DR. CUNNINGHAM: Yes, sure.
DR. BONACA: But certainly, if you could
understand what will be released, CDF and LERF can be
equated --
DR. CUNNINGHAM: Yes.
DR. BONACA: -- for this. It will provide -- shed
light on which approach to take.
DR. CUNNINGHAM: Yes; we're taking you to slide
20, where we think we have today where we want to go with
this paper. We can basically tell the Commission that we
think we need to take the principles -- several ways to
think about it. One is to take the principles that are laid
out in 1.174 and try to use them as a basis to develop the
process we would use and make the decision on the PTS rule
change.
DR. APOSTOLAKIS: Or, if possible, a new set of
principles.
DR. CUNNINGHAM: Or a new set of principles;
that's right.
DR. APOSTOLAKIS: Because this is an adequate
protection.
DR. CUNNINGHAM: That's right. So, one way to
think about it is an extension of 1.174. Another way to
think about it is this is using the framework that's in
option three and applying and testing that framework to see
how it works. It has the common element of its background
is safety goal oriented, and this is an adequate protection
rule, and we're not quite sure how that all falls out. We
think we want to go in that general direction, but I think
we need to -- we're going to plan to tell the Commission
about the two issues that we'll lay out.
One is this issue of containment, and I think we
need to provide a better description of the issues there and
try to address them at least to some degree in the staff.
They're very complex issues. At this point, I guess I'm not
convinced that handing it over and saying just equate the
two is okay. I think we need to do some more analysis
ourselves and think about this some more.
DR. POWERS: Can I get a better idea of why you
think there is a potential for early release of PTS
accident? I think when you say that, you're saying that
rupture of the vessel leading to a factor of corruption of
the containment.
DR. HACKETT: We'll be looking for the slide. One
of the things we talked about with the combined subcommittee
is the fact that if you assume -- we're making the
assumption that it's a large flaw on the side of the vessel
and then, you know, the huge dynamic impact that goes along
with that. At least this is the way I would construct it
just in an engineering sense that that will physically move
the vessel some amount and again, a very plant-specific, I
guess, response depending on how the vessel is supported and
so on, but then, obviously, everything associated with the
vessel is moving also.
I would be just thinking from an engineering or
materials perspective that a very high probability of
ripping loose some containment penetration as a result of
that sequence.
DR. POWERS: It seems to me I'm getting out of my
league here a little bit. It seems to me that we have other
accidents that result in vessel failures at high pressure in
severe accident space, and there has not been an assumption
that vessel failure in those ipso facto leads to a
containment failure; quite the opposite. My recollection is
some analyses done for Surrey and a few of those things,
quite the opposite conclusion was reached. They both felt
containment, just because you fail a pressurized vessel in a
good deal more hazardous configuration than this would be.
DR. CUNNINGHAM: To go back to an earlier slide
where we were talking about the general characteristics of a
PTS accident, we laid out a number of issues that are in the
context of containment performance. Ed was just talking
about some of the dynamic aspects of it. If you start it,
you're going to somehow move this vessel to some degree if
and when this large opening were to occur. That's going to
have some impact on penetrations and that sort of thing. So
you offer the potential for penetration failures. You have
the fact that something is going to happen to the fuel when
this blowdown, and the people are talking about the idea
that it's going to -- does it stay intact or not? What
happens to it?
You also have the issues of what's the blowdown
loads themselves. Running counter to that, then, is the
fact that, again, this is a water-generated accident, if you
will; if you've got a lot of water coming out of the reactor
coolant system; you've got a lot of water around the
containment; you have containment ESS.
DR. WALLIS: You're listing all these things. Are
these not candidates for analysis or --
DR. CUNNINGHAM: No, I think they're candidates
for analysis is where we want to --
DR. WALLIS: Because I get the feeling from all of
this, the feeling that some people feel that containment is
terribly at risk; therefore, we should assume that it fails.
That's a very superficial way to make a decision.
DR. CUNNINGHAM: That's why we have to do some
more analysis before we get to that decision.
DR. WALLIS: I don't think you should jump to one
decision before you know what the analysis shows you.
DR. CUNNINGHAM: Yes, and that's where we sit.
DR. HACKETT: I think that's a really good point,
and I think where we were coming from -- and Mark mentioned
it earlier -- is that if a large uncertainty would accrue to
address each one of these or potentially large at any rate,
it's not something -- certainly, it's not something that's
been addressed beforehand for PTS, and we'd be, as Mark
mentioned earlier, kind of carving out some new ground to be
delving into this.
DR. KRESS: Dana, this looked like a planned
sequence for a regression to me. So our regression -- it's
a lot of bloating in the containment.
DR. WALLIS: And now, we're talking about -- this
seems to be talking about CDF and LERF and risk-informing
it. We should be certain that when we talk about
risk-informing, you don't use it as a way of avoiding
tackling difficult clinical questions. I hope that's not
happening to this one.
DR. CUNNINGHAM: The third approach and the fourth
approach, the difference between them, in a sense, is, on
the third approach, you go in and do some more technical
analysis and see what it tells you. The fourth one says
right now, I'm going to deal with it as a policy matter and
say unacceptable, and I'm going to go right to the next one.
That's the differentiation we've got on the two options.
To come back to Dr. Powers' point, I think at the
subcommittee, we talked about -- my sense is we kind of
alluded to the direct containment heating accidents. I read
into what you said direct containment heating accidents.
It's hard for me to imagine that this is a more severe
challenge to a containment than a direct containment heating
accident.
DR. POWERS: I certainly don't come to that
conclusion, but I will accede that I'm unfamiliar with the
water rejection portion of this.
DR. CUNNINGHAM: So there are differences, and
there are things like that, but the DCH is a tremendous
challenge to these containments, and I'm not sure this is in
that league, if you will. But at any rate --
DR. APOSTOLAKIS: So we're going to see you again
in September, August, something like that?
DR. CUNNINGHAM: Yes; whenever the next
subcommittee meeting -- I think we need to go forward with
the paper telling them in general where we're going with
this.
DR. APOSTOLAKIS: Yes.
DR. CUNNINGHAM: I think we tell them that we have
to talk about the issues that have just been discussed here
and say in my mind, we have to provide more analysis,
technical analysis, technical work before we're at the point
where we're going to say that we're just going to make a
policy decision.
DR. POWERS: Dr. Kress raised a point that I think
deserves consideration as long as you're considering these
good technical issues. He said gee, if you had a rupture
and blew the vessel down, wouldn't you get air on the fuel?
Your LERF criteria right now are based on a source term that
doesn't consider air on the fuel. In another context, Dr.
Kress has suggested that it would be important to go back
and re-evaluate that LERF limit that you have in 1.174 when
your source term changes dramatically.
DR. CUNNINGHAM: Absolutely.
DR. POWERS: And you may have to consider that.
DR. CUNNINGHAM: I should have had that on the
slide as well. That's another technical issue embedded here
is the source term behavior in an air environment. If we
want to go forward with the paper and that issue and tell
them that we need to work that issue some more, and then,
the second issue is this -- the other one we're talking
about is we've got an adequate protection rule, and we're
trying, to some degree, to apply safety goal type guidance
to it, and how does that all work itself out? So we have to
work that issue as well.
At this point, this will be an information paper
to the Commission to say we've got to deal with these
issues, and this is what we're doing with them, and we're
going to go on with it.
DR. KRESS: Mark, you've got some really fun
things to think about.
DR. CUNNINGHAM: Yes; that's right.
DR. APOSTOLAKIS: Okay; anything else?
DR. CUNNINGHAM: I think we've covered everything,
and we'll come back at the next subcommittee and try to give
you an update on where we are on these issues.
DR. APOSTOLAKIS: Very good; thank you very much.
DR. POWERS: Are there any other presentations?
[No response.]
DR. POWERS: Then, I'm going to recess us until
12:45.
DR. SEALE: Mr. Chairman?
DR. POWERS: Yes.
DR. SEALE: Dr. Kress and I suffered -- and I do
mean suffered -- through three days of the CSUP meeting down
at the hotel on chairs of questionable orthopedic design --
[Laughter.]
DR. SEALE: And I think we should just all revel
in the comfort --
[Laughter.]
DR. SEALE: -- of our present ambience.
Would you agree, Kyle?
DR. KRESS: No, you're revealing your age. I'm
not that old.
[Laughter.]
DR. POWERS: If you're looking for some sympathy
on the committee, I assure you you have none.
[Whereupon, at 11:45 a.m., the meeting was
recessed, to reconvene at 12:45 p.m., this same day.]. A F T E R N O O N S E S S I O N
[12:45 p.m.]
DR. POWERS: Let's come back into session.
Members will note the fact that some thought that Dr. Shack
may have been kidnapped by rebels in Sierra Leone -- he is,
in fact, back and alive and well.
We'll turn now to proposed revision to the
Regulatory Guide 1.174, to see if a single word can be
adjusted in that sacrosanct document: an approach for using
probabilistic risk assessment and risk-informed decisions on
plant-specific changes in the licensing basis.
Professor Apostolakis, I think you're going to
take the committee through this.
DR. APOSTOLAKIS: We did discuss this issue at the
subcommittee meeting some time ago.
DR. POWERS: Would you explain to people when they
give us back the notes with draft stamped across them?
DR. APOSTOLAKIS: So who is doing the review
today? Gary?
MR. BARRETT: Bob Palla of the staff will make a
presentation.
DR. APOSTOLAKIS: No more accident management?
We know how to manage it.
DR. PALLA: Let's hope it's not in the
implementation stage.
[Pause.]
DR. PALLA: Shall we start?
DR. POWERS: Please.
DR. PALLA: I'll move right ahead.
My name is Bob Palla. I'm with the probabilistic
safety assessment branch, Office of Nuclear Reactor
Regulation. The purpose of my presentation today is to
discuss the proposed guidance on the use of the staff's --
guidance on staff's use of risk information in the review of
non risk-informed license amendment requests. This guidance
is intended to clarify responsibilities to consider risk in
regulatory decision making; the staff authority to question
risk implications of proposed changes and the staff's
authority to potentially reject the proposed changes based
on risk considerations.
The need for this guidance was identified as
policy issue four in SECY 98-300. The issue came to light
again in the review of electrosleeving for the Calloway
plant. It pointed out that this guidance would in fact be
needed and helpful; lay it on the table and make the
industry clear on how the staff would intend to use risk
information.
We did discuss this with the ACRS subcommittee in
September of last year, and at the time, we were working
from a draft version of SECY 99-246, essentially the same
document as what we forwarded up to the Commission about a
week or so after the committee and we had met. So we --
okay; let me just proceed here with the first slide.
This is in the way of background, and I'll walk
through this background quickly, and we can get to the heart
of things. The PRA policy statement encourages the use of
PRA in regulatory activities. However, licensees are not
required to consider or submit risk information. As a
result, the existing regulatory guidance that we've
developed is geared toward situations in which the licensee
voluntarily submits this risk information as part of a
risk-informed license amendment request.
Commission policy implies it's the staff's
responsibility to consider the change in risk as well as
compliance with the regulations when we review license
amendment requests. And these responsibilities are pretty
clear when the submittal is risk-informed, but when you have
a submittal that is not, then, how one takes risk into
account is not clear, and accordingly, in SECY 98-300, we
identified as a policy issue the need to develop such
guidance.
DR. APOSTOLAKIS: Now, I see here a statement that
is very interesting. The proposed license action satisfies
the existing design and licensing basis but introduces
significantly unanticipated risks. What happened to the
presumption of no undue risk to public health and safety if
the regulations are met?
DR. PALLA: Well, there still is the presumption
that -- I mean, it is a presumption that meeting the
regulations is -- provides adequate protection, no undue
risk. What we're talking about here is a situation that
defies that, really, calls into question this presumption.
DR. APOSTOLAKIS: Is it possible to meet the
regulations and still have undue risk?
DR. PALLA: Yes.
DR. POWERS: It seems to me that the language says
it's a presumption, okay? Here, he's talking about a
situation in which staff, in looking at it, says that
presumption may not be accurate in this case, because
there's something very unusual about this.
DR. APOSTOLAKIS: But it would also tell me that
the use of risk information should be mandatory if that's
the case.
DR. PALLA: Well, our thinking is that by and
large, this presumption is valid, and that the situations --
well, if one complies with the regulations, the situations
in which you actually do have undue risk are very rare, and
that's been our experience, that we did a search of various
license amendment requests, and it looks like very few of
those would come close to calling this presumption into
question. So our thinking is that again, it's better to not
ask for risk information in every submittal, but we think we
need to have some kind of ground rules for when we should
cross over and do ask for it.
DR. APOSTOLAKIS: But perhaps among technical
people, this is not a big issue, but I wonder if a member of
the public sees this, and I was expecting Dr. Wallis to
actually raise that issue. What kind of message does this
send, that if you can satisfy existing design and licensing
basis and yet introduce significant and unanticipated risks;
that doesn't sound too good.
DR. PALLA: We're not saying that's okay. We're
saying that's what we're worried about here. So we're
looking for those situations. We think they're rare, but
that's what we're specifically targeting here. And we're
stopping short of risking for risk information.
DR. POWERS: George, I know several instances in
which they have found that the regulations that existed at
the time did not cover the mark.
DR. APOSTOLAKIS: I know.
DR. POWERS: And then, they went to say -- where
they were reminded that prior applications had produced very
significant challenge to the plant and the evidence of the
regulations. Okay? So, I mean the fact that one might find
a situation out there for a particular plant in particular
circumstances where the rules simply weren't enough is not
within the bounds of credibility here.
On the other hand, equally true is that in 99
percent of the cases, existing regulations are enough. I
mean, I think that's what Bob said.
DR. PALLA: That's the concept.
DR. SEALE: Well, and the way in which it is
stated puts the onus on staff to come up with specifics as
to why it doesn't meet the adequate protection requirements
rather than requiring that kind of proof up front from the
licensee. That would be a considerable burden which they've
managed to avoid.
DR. APOSTOLAKIS: I think the whole thing comes
from the fact that they're constrained, I guess, by legal
considerations. It seems to me we have this new technology
that gives us new knowledge and all of that, and now, we are
questioning whether we should request this kind of
information, and this sentence there tells me that we
should. If there is a remote possibility, we should.
DR. WALLIS: As we get wiser, it's foolish to deny
that we know something more than we knew before.
DR. APOSTOLAKIS: Exactly; thank you very much,
Graham.
Let's go on.
DR. PALLA: Well, I just wanted to point a few
things out here. As I mentioned, we start the second paper
up. After meeting with the committee -- we did receive a
favorable staff requirements memorandum from the Commission.
The Commission basically said that the staff should develop
final guidance that clearly articulates what constitutes a
special circumstance in a clear and objective manner. It
indicated that we should engage the stakeholders and while
we -- you know, in that process and indicated support for
the use of the guidance that we had in SECY 99-246 on an
interim basis while we finalized the documents.
So we, as a result, issued a regulatory issues
summary 2000-07, which in essence summarizes the content of
SECY 99-246. It was transmitted to all utilities. It does
have one change from the guidance that I'll discuss with you
today, and basically, the flow chart has been changed to
reflect this; the Commission requested that in the interim,
if we run across a situation that we would consider meets
the special circumstance criteria that we would inform of
that. So the basic process that we included in the
regulatory issue summary is essentially the same one that
I'll be talking to you about today except it has that
additional step of notifying the Commission should we run
across any, and we have not so far since that.
Finally, we've issued the modified guidance
documents, and we forwarded them to ACRS and CRGR for
information. At the same time, we've put them in the
Federal Register. We plan to have a workshop next week to
discuss that with stakeholders.
I just want to briefly summarize the process that
was in SECY 99-246, and the reason I'm just going to repeat
what's in 246 is because that's essentially what is in the
modified documents, and in essence, we have created this
concept that there could be exceptions to the -- our
standard presumption of adequate protection, compliance
equals adequate protection. There could be some special
circumstances under which that presumption is not valid.
DR. APOSTOLAKIS: This raises a question that
bothers me a little bit. Also, I read the selected changes
in the text of Regulatory Guide 1.174. And I must say I'm a
little bit confused as to why an issue that seems to be
related to adequate protection appears in 1.174, which does
not deal with adequate protection. For example, in the
introduction, you are saying that the basis on which new
information reveals an unforeseen hazard for a substantially
greater potential for a known hazard to occur, in such
circumstances, the NRC has the statutory authority to
require licensee action beyond existing regulations to
demonstrate that the level of protection necessary to avoid
undue risk to public health and safety; that is adequate
protection.
Why is this in a document like 1.174?
DR. PALLA: Well, I would say it's in 1.174
because 1.174 gives guidance to the staff on how to use
risk. Now, we do, in this case, cross the risk information
with the concept of adequate protection. We don't equate
it, but we, in essence, in the appendix to the SRP that I'll
be focusing on, we use the Reg Guide 1.174 criterion
guidance in our review, and we then go further if that
criterion is not met to ultimately make a decision about the
acceptability of the amendment.
It seems like a logical place to put it. If there
was a better place, I guess we could put it there.
DR. APOSTOLAKIS: No, I mean, the five principles
of 1.174 include a very important one, which is principle
number four, that the changes in CDF and LERF should be
small.
DR. PALLA: Yes.
DR. APOSTOLAKIS: And I don't see how that
principle has anything to do with what we're talking about
here. If it's an issue of adequate protection, what does
small change mean, and would you allow any change?
DR. PALLA: One way to think of this is that this
is kind of a one-way test. The reg guide provides a basis
for saying that something would provide adequate protection,
but when that guidance is not met -- it's not the same, and
you can't categorically conclude that you have inadequate
protection, but it opens the door. We refer to it as a
trigger, that if you don't meet the 1.174 criteria, then, it
raises questions, and these questions have to be carefully
considered, and ultimately, one has to make a decision on
the acceptability of that.
Now, we refer in that, in the case that it's
unacceptable, we're saying that the underlying decision
would be that the protection is inadequate.
DR. APOSTOLAKIS: Now, again, that's not about
that particular principle that the changes should be small.
A licensee may submit a request that if it was submitted
under 1.174 would result in CDF saying that that's greater
than -- it would say -- that the licensee's core damage
frequency is down to 10-6. So it violates the principle
that the changes should be small, but it really doesn't
raise an issue about adequate protection.
So it's this blending of adequate protection
arguments with 1.174 that confuses me and I suspect is going
to confuse the licensees too. Is this a way of catching
those? In other words, a plant is not going the
risk-informed way, and he's asking me for something that's
too much? That may be fine; you don't want them to do that.
But don't invoke adequate protection to do that.
DR. PALLA: Well, the thinking here is that -- I
think you're concerned about just marginally not meeting the
criteria, the numerical guidance in the reg guide, and I
guess that it could appear to be a concern, but we were
trying to focus not only on numerics but on the other safety
principles collectively to make a decision that when one
looks at that integrated decision making process, those
safety principles are essentially the same kinds of
considerations that go into the judgment about adequate
protection.
So the numbers are a piece of it, but the other
considerations are other factors.
DR. APOSTOLAKIS: Yes; that brings me to my next
question, which is very much related. Let's say that we are
estimating the change, and you find that it's 4x10-5 for
CDF, which is about what the 1.174 allows, at which point,
you would say five or six or seven; when do you decide,
since you don't have any numerical guidance regarding
adequate protection, that now, you have a problem with
adequate protection?
MR. HOLAHAN: This is Gary Holahan.
George, I think we're going back a couple of years
to -- the questions you're raising are basic questions about
how Reg Guide 1.174 works in the regulatory process.
DR. APOSTOLAKIS: No, the question I'm raising is
whether 1.174 is relevant to what we're addressing here.
There's a clear statement.
DR. POWERS: I guess I'm confused on where we're
going. I'm still hung up on how you get into this process
before how you're going to react to what you find in the
process. Right now, I'm still hung -- hanging onto the you
come on, and you've declared this special circumstance, and
that's the part I don't understand. Now, once they've done
that, presumably, we'll go and find out what they really --
that's how they get into that.
DR. APOSTOLAKIS: I'm questioning handling the
special circumstances using principles from 1.174. I don't
think they apply, because the special circumstances,
according to the document, deal with undue risk. That is
adequate protection. So I don't see how you can use 1.174
to conclude that you have a special circumstance.
MR. HOLAHAN: George, remember, 1.174 is guidance
on how to do license amendments.
DR. APOSTOLAKIS: Yes.
MR. HOLAHAN: And the test to license amendments
is no undue risk. It is -- in fact, every license amendment
decision we make is effectively an adequate protection
decision.
DR. BONACA: It's interesting how you could go to
the same exercise thinking about with Calloway; they've
proposed this amendment, and they want 1.174. Think of it
that way. 1.174 never denied the amendment if they find
themselves in the region that they cannot justify this
change. And I believe that this allows the NRC to have the
symmetrical approach where they can implement the same
process and perform what Calloway should have done.
MR. HOLAHAN: With respect to Calloway, I think
the point is not necessarily that it would have been denied.
The point is that the rules of the game of how the amendment
was treated would have been much clearer. It would have
been much clearer if we had this guidance two years ago as
to what the expectations on the part of the licensee and on
the part of the staff were. It was still a difficult issue
with technical matters that will be argued over as to
whether they would have passed or failed the criteria. But
at least we would have had a clearer understanding of what
the expectations were.
MR. BARRETT: This is Rich Barrett with the staff.
I'd like to take a shot at that question. I think that the
concept of special circumstances is important here, because
I don't think that we want to say that a -- that you have --
that there are criteria in Reg Guide 1.174, and if you're
below those criteria, you get something approved, and iv
you're slightly above them, you cannot. What we're trying
to say here is that we only want to look at situations that
have -- I'll go so far as to say deterministic circumstances
which make the situation worth looking at from a
probabilistic or a risk perspective.
In the case of electrosleeves, it was the
introduction of an entirely new type of material into the
steam generator, the material having certain characteristics
which the staff, if we were doing this today, might call
that a special circumstance which would raise the question
of whether we should be looking at the criteria and the
guidelines of 1.174; and as Bob pointed out not only the
numerical guidelines but questions related to
defense-in-depth margins and the ability to monitor
performance and all of the other criteria in Reg Guide
1.174.
So what we're trying to say here is that there's
more than just a risk test, a numerical risk test. We're
basically saying that there is going to have to be a
threshold here that the staff is going to have to cross in
terms of something qualitative that raises this to the level
of a possible special circumstance.
DR. WALLIS: While we're on this general slide
here, wouldn't it be true that if there were an adequate
protection meter, then most of the problems would go away,
and for some reason, in the entire history of the agency,
having inadequate protection is somehow taboo?
DR. PALLA: Well, if you had an adequate
protection meter that sufficiently addressed all of the
facets of what goes into them through an adequate protection
procedure, that would be true. I think the challenge would
be to come up with such a meter, because it isn't just -- if
we had ideal PRAs, I don't think it would be numerical
still.
DR. WALLIS: Only the priesthood can decide. You
don't have the measure which the layman can understand which
says whether it's adequate protection or not.
DR. APOSTOLAKIS: Bob, nobody is arguing that it
should be a numerical criterion alone, but one could
envision the five principles of 1.174 with the fourth
principle replaced by other numbers as being applicable to
adequate protection questions. That would be a cleaner way
of doing it. Which brings me to my last point: you also
say here in the change that the staff -- if a licensee does
not choose to address this, the reviewers should not issue
the requested amendment until they have assessed the risk
implications sufficiently to determine that there is
reasonable assurance that the public health and safety will
be adequately protected.
Now how on earth are they going to do this without
numerical guidance? How does a reviewer decide that the
risk change is not great enough so that there is reasonable
assurance?
DR. PALLA: Okay; at that point, you're part way
through a process. You would have already had an amendment
request come in. It would be forwarded to the risk branch
for a look. The risk person would look at it; says, well,
even though it meets the requirements, there is some element
of what's being requested here that causes me concern for
some sequences that are significant to the risk profile for
this plant. So this is all, perhaps, qualitative judgments;
it would be.
If it wasn't a risk-informed amendment, it would
be coming in without that risk information, so the reviewer
would have to be making these judgments based on qualitative
considerations. And upon being convinced that there is a
concern here, one of the steps -- and there's a flow chart
in -- I think it's the next page in your package there --
that basically says management informed of potential risk
concerns. We've put in this process a step by which the
individual reviewer would not unilaterally make some
decision about an issue being a special circumstance. There
would be some iteration, some, you know, convincing done at
the staff -- you know, with their management to basically
reach some agreement that, yes, this does look like it's a
risk-significant consideration.
That could rise to the level that we might argue
that it's inadequate protection. And then, you'd go and
you'd ask for the risk information. Now, the point that I
think you were postulating was what if, after you asked for
it, you don't get it? And what we say in our guidance is if
the licensee chooses to not submit this, then, we basically
would need to convince ourselves that this issue that we got
ourselves all spun up about is, in fact, not an issue. So
we do put a burden on ourselves, number one, to convince
ourselves that it is a special circumstance, and then, once
we've done that, if we don't get enough information, I don't
think it's appropriate to turn around and say we got no
licensee response; the thing that we thought was a special
circumstance is okay.
You know, you can't do that. I don't think it
would be legitimate to do that. I think you then have to
establish a basis for concluding that it's acceptable, and
if you can't do that, then, I think you have a problem
approving the amendment request. And so that is kind of a
dilemma that's created if a licensee does not submit the
information that's necessary.
DR. POWERS: Let me ask a question, because I'm
not sure there's -- this is all this unusual. Suppose a
licensee proposes something; it comes to the appropriate
person on the staff. The appropriate person on the staff
says oh, gosh, these guys haven't looked at intermetallic
reactions. I'm pulling that out of the air.
DR. PALLA: Okay.
DR. POWERS: And so, he writes a request for
additional information; sends it off to the licensee and
says what about intermetallic reactions? The licensee comes
back and says never heard of such thing; don't have any
answers; don't know what you're talking about. I mean, the
guy has the same problem, right?
DR. PALLA: Yes.
DR. POWERS: And he has handled that as he's
handled it for years and years and years now. He either
convinces himself that his concern over intermetallic
reactions was misplaced, and there isn't a problem here, in
which case, he approves the amendment request, or he
persuades himself that, indeed, he was justified in worrying
about and he denies it, right?
DR. PALLA: Well, I don't think he denies it right
up -- I think he elevates it --
DR. POWERS: Oh.
DR. PALLA: -- and basically goes to the next step
to try to articulate what is it about this issue that --
DR. POWERS: Well, yes, he has to persuade
somebody, himself or maybe his boss or somebody else.
DR. PALLA: Right.
DR. POWERS: Yes.
DR. PALLA: And then, ultimately, if you can get
the information from the licensee, it makes the job easier
to make the judgment.
DR. POWERS: But if he doesn't --
DR. PALLA: If you don't, well, then, the staff
has to be -- I mean, our -- the purpose of our review is to
conclude that there is adequate protection. If you can't
get that information that's needed to do that, then, I don't
think you could reach the appropriate finding.
DR. POWERS: And I think that's been going on for
a long time, hasn't it?
MR. HOLAHAN: It goes on all the time, but in
making that judgment, the staff needs to articulate what is
it about this issue that makes it unacceptable, and in most
cases, and, in fact, the easiest path to saying no, this is
not acceptable is to say that this phenomenon, this issue,
this lack of information, this, you know, negative
information shows me that there's a piece of the regulation
that isn't being met, okay?
So the regulation may say something about the
extremely low likelihood of the vessel failing or something
like that.
DR. POWERS: And here, we've already posed that it
meets all regulations.
MR. HOLAHAN: Yes; that's right.
DR. POWERS: But there's still something there.
MR. HOLAHAN: That's right; that's right. And so,
this sort of issue, in which you're struggling with, you
know, how much do I know and is it good enough, say, some of
those are issues about am I meeting the regulation, and some
of those are, well, I think I meet the regulation, but I
still have residual concerns usually because the regulation
wasn't constructed originally to deal with these particular
types of cases or materials or issues or whatever.
DR. POWERS: I mean, you must be very
hypothetical, because I think I can take -- find something
in the general design criteria that's applicable to
everything in an absolutely sense.
MR. HOLAHAN: But in some cases, you may find that
the general design criteria or the regulation is specific on
a point, but under the particular circumstances, it gives
you the wrong answer.
DR. POWERS: I see; okay.
DR. APOSTOLAKIS: Speaking of hypothetical
situations, let's say that you are doing your risk
assessment, okay, either with the help of the licensee or
not, and you find that the change in CDF is 3x10-5, but
their CDF is already 10-6. Now, you have a problem. I
don't think you can really make a case of adequate
protection. On the other hand, the guy has violated the
fourth principle of 1.174. So does he get away with it just
because he did not submit risk information, or what do you
do?
DR. PALLA: Well, that's one of five, and you've
got to say, well, what about the other four? Does he have
any, you know, margins, defense-in-depth? Is this -- what
exactly is the situation in these other areas? And if he
fell down in numerous other areas, then, you might think of
elevating that.
DR. KRESS: I think George has a really good point
here, and the point is that in this process of dealing with
requested changes and whether they should be approved or
not, we're mixing in the concept of adequate protection when
we really don't know what adequate protection is, and we're
trying to equate it somehow with a change in CDF or a change
in something, and the two are just not related to each
other, and that confuses everybody.
What you ought to have is a rule that says we will
not allow a change in CDF or LERF more than this amount and
not try to confuse it with adequate protection. That is the
rule. That's the criterion, and whether they have a CDF of
this much or this much, we're not going to allow any change
any bigger than this, and it's an absolute, and you don't
confuse it with adequate protection; you don't even try to
throw it in that way. And I think that's the problem.
DR. APOSTOLAKIS: To tell you my personal view is
that the staff really -- and, of course, they're here; they
can correct me -- the staff really does not want to approve
changes that violate 1.174.
DR. KRESS: And that ought to be the rule. That
ought to be the criteria.
DR. APOSTOLAKIS: And if it was up to them, they
would do that.
DR. KRESS: Yes.
DR. APOSTOLAKIS: The problem is that legally,
they cannot publish it that way. They have to invoke
adequate protection.
DR. PALLA: Are you talking about just the
numerical criteria?
DR. APOSTOLAKIS: All of it; five, all five
principles.
DR. PALLA: It's very hard to judge those other
principles, though, but the numerical one is probably the
easiest one. These other factors are kind of diffuse so --
DR. APOSTOLAKIS: Right.
DR. PALLA: So I think that judgment is just very
clouded so --
DR. APOSTOLAKIS: It is very clouded, but at least
the only principle there that's not as much clouded is the
fourth principle, and there is a possibility that a proposed
change violates that principle without really disturbing the
others too much, and I -- if it's an issue of adequate
protection, I don't see how you can deny it. And yet,
Regulatory Guide 1.174 is violated.
So, in other words, by choosing not to submit risk
information, I have avoided one of the constraints there.
DR. PALLA: In fact, what the process we're
suggesting --
DR. APOSTOLAKIS: It's not your fault, by the way.
I'm not blaming you guys. I know your predicament.
MR. HOLAHAN: It seems to me that what we're
suggesting is that when there are circumstances that make a
particular license amendment look like it deserves
additional attention, it is, in fact, pulled in and given
the same treatment as though it were submitted under a
risk-informed application.
DR. APOSTOLAKIS: So it's going now down to 1.174.
MR. HOLAHAN: Yes, yes, and --
DR. APOSTOLAKIS: And I would agree with Dr.
Kress, then. Drop the words adequate protection. I mean,
that is a very noble thing to do, to apply 1.174.
DR. KRESS: I think it's a perfectly reasonable
process.
MR. HOLAHAN: I find that I can get through the
entire day without mentioning adequate protection by just
thinking about whether decisions are the right safety
decisions or not.
DR. APOSTOLAKIS: But would the Office of General
Counsel concur? That's really the question.
DR. PALLA: Well, I don't think they would,
because I think the basis for rejection has to be adequate
protection, and I think that --
MR. HOLAHAN: And the truth is the basis on which
all license amendments are approved, 1,500 every single
year, is adequate protection.
DR. APOSTOLAKIS: But, then, you don't invoke
1.174 in those decisions, Gary.
DR. PALLA: Well, you could.
MR. HOLAHAN: In fact, we do.
DR. PALLA: If you meet it, you do.
DR. POWERS: George, how do we get out of your
dilemma?
DR. PALLA: Go to the next slide.
[Laughter.]
DR. APOSTOLAKIS: No, there are two ways. One is
to drop references to adequate protection, and the other is
to drop the opposition to the three-region approach that
this committee has recommended.
MR. HOLAHAN: Well, okay, I think that's an
important point, which is the subject we're talking about
today and the relationship between risk-informed insights
and adequate protection is no different now, and it's not
being changed with respect to what 1.174 established with
this committee's enthusiastic endorsement just a few years
ago.
DR. APOSTOLAKIS: I agree with the latter part.
See, my point, Gary, is this: that you have the
CSIS; you have this committee coming to the Commission and
saying we need some numerical guidance regarding adequate
protection. The Commission looks at this and says why on
earth are they telling us this? This is philosophy. Why do
we need this? We don't need this.
Until the Commission sees practical problems that
arise because of the lack of that numerical guidance, they
will never change their mind. And if you show this to them,
they're not going to see it right away, and it seems to me
that this is your problem here, that you don't have any
numerical guidance regarding what is adequate protection.
MR. HOLAHAN: Clearly, what we are doing here is
we are defining a process that works in the context of the
existing regulations, existing case law, you know, the
existing guidance that we're using, okay? So, you know, by
definition, it is dealing with difficult cases within the
current regulatory context, okay?
And we need to do that because it's the only
regulatory process we have right now. And I think I'd like
to separate these two thoughts. One is I think the
committee has expressed a number of times that it would like
to see a different regulatory process with a different way
of addressing adequate protection and using risk information
and, you know, that seems to me that's an ongoing thought;
it's very interesting.
But in the meantime, we have to make the existing
regulatory process work. We need it to be as, you know,
clear and efficient for all the stakeholders as possible.
And so, you know, we're not trying to avoid those other
issues, but we need to deal, you know, with the existing
process at that point in time.
DR. APOSTOLAKIS: One last point, and then, that's
it. I don't see how the existing regulatory process guides
you to decide that the risk implications have been
sufficiently assessed to determine whether there is
reasonable assurance that the public health and safety will
be adequately protected. I just don't see the current
regulations doing that.
MR. HOLAHAN: Okay.
DR. APOSTOLAKIS: Okay; let's go on.
MR. HOLAHAN: I think I need to say that not only
do I see it, but I see it happening 1,500 times a year.
DR. APOSTOLAKIS: With numerical risk estimates.
MR. HOLAHAN: Well, maybe a few dozen times a year
with risk estimates.
DR. POWERS: I'm anxious to go through this slide.
MR. HOLAHAN: Well this, I think, is basically
what we've already talked about. Let me just explain
something up front about this, though, because license
amendments, when they come in, are processed in division of
projects, and the project manager would generally be looking
at these submittals and trying to determine what branches
need to have a role in this review.
We did put out some guidance. It's in office
letter 803. It was revised last December to include I'll
say the beginnings of some further guidance on how risk, you
know, considerations should be brought into that screening
process. But in special circumstances is one of the pieces
that's flagged in the office letter as being a reason to go
and consult with the PRA branch. We need to come back to
that office letter once we've fleshed this animal out some
more and updated and, you know, fill it out, but at this
point, there is a mechanism in place that an incoming
amendment would be screened; would be forwarded to the PRA
branch at least for consultation.
This isn't to say that it's automatically, you
know, assumed that it's going to be an in-depth PRA review,
but it will be forwarded to the branch to look at these
exact kinds of considerations, because this could be a
non-risk-informed amendment that comes to us, and we would
look at it to say is there something unique about this
request that we need to look at it for some reason and
possibly go into risk implications.
DR. POWERS: It's this unique situation criteria
on the very first box that causes me pause, because I can
argue that every single one of 1,500 licensing amendments a
year is unique. I'd have no trouble in defending that.
DR. KRESS: I think you what? Go in and make a
qualitative judgment that that change is likely to have some
influence on risk.
DR. PALLA: Well, it would generally be something
that's extreme.
DR. KRESS: Yes.
DR. PALLA: I mean, if you're looking at, let's
say, power up rates, and you come in with a couple percent,
5 percent, no big deal. If someone comes in for 20 percent,
you'd say hey, that maybe something that, you know, it's
unique in the sense that no one else has done anything close
to that before so --
DR. POWERS: I know that if you were doing it that
it would have that reasonable approach, but I also know that
as time goes on, different people do it, and they may have a
different threshold than you have in mind as you set this
chart up. And they will say ah, this one is unique.
DR. PALLA: Well, unique should really be thought
of as you believe it's potentially special circumstances.
That's what really is intended there. There's something
about the request that undermines or does not -- it includes
considerations that may not have been part of the
regulations. There's a hint that perhaps if this came in as
a risk-informed request that this would be a problem for
some reason or another. So you've got to carry a lot of
knowledge into that first box there. You have to have some
educated, you know, understanding about what the concern is,
why the existing regulations don't seem to cover it
adequately. It's not just unique, because there are a lot
of things that are unique.
DR. POWERS: I mean, everything is going to be
unique. I'll find something; this bolt on the lower
left-hand corner of this fixture is different than any other
bolt I've ever seen.
DR. SHACK: Unique is definitely the wrong word.
Either everything is unique, or nothing is unique.
DR. POWERS: That's right.
DR. PALLA: It's got to fit in the box, too.
DR. POWERS: What I'm trying to do or what I'm
concerned about is it seems to me that going through this
process ought to be a very rare circumstance, and it ought
not be used as an excuse for demanding that things be
risk-informed, all submittals being risk-informed or
threaten that all submittals need to be risk-informed.
That's where I'm concerned.
DR. PALLA: Well that's -- when you're in that
first box, you should be looking at what is it about the
regulations that -- for whatever reason I meet them, they
don't give me the intended level of safety that I would have
expected. That all goes on in the first box, you know; it's
more than unique; it's unique, plus you look at, you know,
you're looking at this situation and determining essentially
that you believe it's a special circumstance, that if you
approve this, there would be some significant increase in
risk and possibly rising to the level that you would say
it's inadequate and unacceptable.
So that's all that that box is intended to do, and
unique is not the best word, obviously.
DR. WALLIS: It's a pretty important box, because
if the answer is no, then, off it goes to the right and
disappears.
DR. PALLA: Yes; and this is where the burden
falls squarely on the staff's shoulders, because they're
making this judgment in the absence of the risk information.
They're trying to read into -- I've got a slide a couple
slides back here. There's really only a couple sentences in
the proposed modification to the guidance that we have that
tries to articulate a little bit more about what that is,
but it's an important part of the step.
And then, you go to the next -- you have to
structure all your arguments as to why this thing is a
safety concern up here. And then, once you've convinced
yourself, you take it, and you get the management buy-in on
this.
DR. WALLIS: I guess what I'm concerned about is
the reluctance to do this, because it's supposed to be a
special case. And then, here go these 1,500 applications
by. Suddenly, someone discovers at the end of the year,
gee, whiz, the net effect of all these things we've approved
is that we've violated 1.174. We've compromised the
principles without knowing it. We've let everything go to
another side, and then, gee, whiz, look.
DR. PALLA: Well, it's unclear you'd ever know
that, because these would be non-risk-informed. You
wouldn't have that risk information. You wouldn't know that
you would have exceeded it unless you know to ask. You're
kind of -- this is just a screening process. You try to
determine -- we're setting a threshold here for when we're
going to ask the risk questions.
DR. WALLIS: Well, there's no penalty for letting
it go to the box.
DR. PALLA: To the no, unacceptable box?
DR. WALLIS: At the end, the one there --
DR. PALLA: Oh.
DR. WALLIS: The one there on the right.
DR. POWERS: You have a really peculiar -- when
the answer is no, the application is yes.
[Laughter.]
DR. POWERS: It's a peculiar fault tree.
DR. PALLA: Well, yes, it is.
MR. HOLAHAN: In my mind, what's important to
recognize in this -- I really hate coming to the committee
and being such a pragmatist; it's much easier when I'm back
in my office. But the point is we're not inventing, you
know, these circumstances. I mean, license amendments come
in. They've been coming in for years. And the question is,
you know, as a pragmatic approach, should we try harder to
catch these unusual cases? If we don't do this, then they
automatically go --
DR. WALLIS: That's not the only question. The
question is when you've done this for a year, you look back
at it, and have you done it right? And you may need to
review the whole process of your 1,500 in a year and say
when we did all that, did we in some way violate the RG
1.174 safety principles and allowed them to be compromised.
MR. HOLAHAN: Well, let me say something about
that because we had a contract, and the contractor looked
through about six months worth of license amendment
requests. We had 700. And they looked at, you know, these
were risk-savvy people. They screened them carefully and
tried to determine, you know, what types of amendments
really warranted a quick look, a detailed look and basically
found that, you know, something kind of on the order of
maybe a tenth of them needed to have some -- a high level
look at risk and then a small fraction of -- they really
didn't find anything that they would have -- we consulted
and asked them about special circumstances; did it look like
any of these 700 would have risen to the level that they
would have, you know, invoked this special circumstances
thinking, and basically, the answer was no.
There is a lot of them that you would want to look
at the risk implications, but there weren't any in that
sample that would rise to the level that you'd reject this
thing because of risk considerations, and there were things
like that we didn't have any Calloways in this group but --
DR. WALLIS: That's reassuring; the whole purpose
of this is to make sure that you don't, in processing all of
these applications, some way allow you to violate the safety
principles.
DR. PALLA: But that was a sample. Now, that's
not to say we're going to always do that, but based on that
sample, we concluded that the instances that you'd have to
go drilling deeply for risk information are rare.
DR. WALLIS: You'd have to make sure that you're
doing whatever process like this you set up to make sure
you're doing it right. You have to keep on reviewing over
the year or in some way the feedback to see if you're
catching things right at box one, because that's the lead
box.
MR. HOLAHAN: It's a good suggestion. We need to
find an effective and efficient way of testing that, kind of
a reality check on the whole process, look at it
periodically after the fact, see how it's working.
DR. PALLA: Well, let me just move through this
thing a little bit more. There's a management buy-in step
here that's a checks and balances kind of thing to make sure
that the decision is not unilateral on the part of the
reviewer. Once the decision of a special circumstance, some
kind of a consensus is formed, decisions may -- let's try to
get the risk information we need to make a determination on
what does this amendment mean to us in terms of risk.
DR. KRESS: When you say management, is there some
sort of a -- going to be some sort of an upper-level review
group established that you take these things to or is it --
I don't know exactly what you mean by management.
DR. PALLA: The truthful answer at this point is
that I don't know exactly what that could involve, but I was
thinking at a minimum, it would be a branch level
involvement, branch chief, but it could possibly be --
DR. KRESS: Could possibly be a review from higher
management.
DR. PALLA: It could be; I don't know if Farouk
would be an appropriate thing. If they're rare enough, I
don't think it should be an impediment. I mean, it may be
high visibility.
MR. HOLAHAN: My guess is that there will be no
lack of management attention. My recollection, for example,
of a case that got enough attention, like the Calloway
circumstances; I believe it had every level of management
including the Commissioners.
DR. PALLA: That was an exceptional exception.
If we move forward; we ask for the information --
MR. HOLAHAN: And, in fact, if you recall, just --
let me interrupt -- remember: the Commission has asked to
be informed about these cases, okay, and that pretty much
assures that about five levels of management are going to
have to understand what it is in order to get it up there.
DR. PALLA: That's in the interim, the Commission
has asked to be informed. We've removed that step in this.
The expectation is we'd have a reasonable level of
management, and these things would be pretty rare, and you'd
get enough attention on it.
If we ask for the risk information, we get the
risk information, and it indicates that the reg guide safety
principles, numerical criteria, whatever, in whatever
combination is not met, it's not an automatic rejection.
It's basically a trigger has been met; you go to the next
step; look more carefully at exactly what made this thing --
why do we think this is special circumstances? How is it
that the existing requirements are not providing us what we
need? Are there other extenuating circumstances? Are there
some compensatory measures? Is it temporary?
We've reached the threshold at this point, and
then, we basically have to -- I'd say this thing gets
elevated up through the organization when it's clear that
there's a lot of fuzzy judgments need to be made about
whether this thing is adequate or not, and that's basically
the last box, to say do we or don't we think that the --
when one looks at all these considerations, is it adequate
protection or not?
DR. WALLIS: These fuzzy judgments is not a
reassuring expression.
DR. KRESS: I would be perfectly happy with this
process if the words adequate protection didn't appear in it
anywhere.
DR. APOSTOLAKIS: Drop adequate protection and --
DR. KRESS: They would be very nice for the
process.
DR. APOSTOLAKIS: As we said earlier, you've come
to this box -- you don't need to put it in this box, but if
you come to this box, and you see that 1.174 criteria are
violated, you say sorry; you have to go back and redo this.
MR. HOLAHAN: You understand whether it says so on
the chart or not, the basis for accepting and rejecting
license amendments is called adequate protection. That's in
the rules.
DR. APOSTOLAKIS: Everything is adequate
protection? We call everything license protection; you're
at the mercy of the reviewer. I thought we were trying to
get away from that.
DR. PALLA: Well, in this case, it's more than the
reviewer. You've ruled this thing all the way up the chain
so --
DR. APOSTOLAKIS: You don't have to call it
adequate protection; go through cost-benefit calculations.
MR. HOLAHAN: Not for licensing matters.
DR. APOSTOLAKIS: But, I mean, the argument is if
I say so, it's adequate protection. I mean, there's
something wrong with it.
MR. BARRETT: Well, these judgments are made in
the context of a lot of regulatory guidance.
DR. APOSTOLAKIS: Because you have nothing else to
go by, Rich.
MR. BARRETT: No, as we said, in 99 percent of the
cases, we have regulatory guidance; we have existing design
basis; we have, in some cases, prescriptive rules. This is
the exception. But the hurdle you have to cross is still
the same. It's a finding of reasonable assurance of
adequate protection. And that's one of the ground rules
that we have to work with in developing this guidance.
DR. APOSTOLAKIS: So what you are saying is that
in that box, question raised, what you're saying is that if
you violate 1.174, you don't approve.
DR. PALLA: No, I did not say that. I said that
you could potentially disapprove it at that point, but you'd
have to go to the next step, to relook at everything that --
every extenuating circumstance that applies. But this was a
case like in Calloway steam generators. We've deliberated
this at the office director level whether this is acceptable
or not. So you've got a lot of considerations go into that;
the uncertainties in these estimates; it's not just the
numerics. It's whether you've got margins, you know, what
the levels of risk are.
DR. APOSTOLAKIS: My question is this: if a
licensee voluntarily has submitted risk information, you
would be using 1.174, and granted, there are uncertainties
in everything, everything, everything. You would have
rejected it on the basis of 1.174, including the
uncertainties. Is it possible that the same request would
be approved if it goes this way? It should be, because
you're talking about the different --
DR. PALLA: I think the process is the same. I
think if it was risk-informed, I think you would jump into
this process without going through the first few steps.
MR. HOLAHAN: Right.
DR. PALLA: You just jump in there and --
DR. BONACA: It seems to me ultimately, if you
have an application, you make a judgment. And here, you're
trying to break the barriers with risk information as one
more thing to make a judgment. And I don't see -- I
understand about the issue -- what I don't see is how it's
divergent with what the staff is doing on every application.
It's simply some additional information. But this itself is
not -- it has the ability to render a judgment on an
application.
MR. BARRETT: You know, a lot of times, when we
get a risk-informed licensing action, it's submitted that
way because a licensee has made a judgment that it will not
be approved any other way. So Reg Guide 1.174 gives you an
opportunity, it gives you a mechanism for making a finding
of reasonable assurance of adequate protection and therefore
approving a license amendment. So we're already in adequate
protection space with 1.174. What we're saying here is that
this is a double-edged sword.
DR. PALLA: Can I proceed?
DR. APOSTOLAKIS: If you approve that, you would
approve it under this criteria; but you would have rejected
it under 1.174.
DR. PALLA: It's conceivable, but you should be
using the same guidance, though. It's possible if you send
it to two different reviewers that you'd come up with
different answers.
DR. APOSTOLAKIS: But aren't we depending on the
intelligence of the licensee to avoid risk information in
most cases, in other words, avoid 1.174?
MR. HOLAHAN: No, I think not; in fact, what this
does is it puts the licensee on notice that if this is an
unusual or controversial case, if you don't provide the risk
information, we're going to deal with it anyway, and if you
want this reviewed in an efficient manner, you ought to do
your homework up front to bring it to us with the complete
story.
DR. APOSTOLAKIS: I have no problem with that.
MR. HOLAHAN: I'd like to go back to another
point, which this issue of adequate protection and all that;
very often, what we find out is the issue is not whether you
have adequate protection or not. The issue is what is the
role of a regulatory guide? All regulatory guides are
definitions of at least one way of meeting the regulations.
They tell you what's safe enough. None of them tell you
what is unsafe. None of them draw that line to say this
would be unacceptable, okay? And we keep coming back to
that among all the regulatory guides, this is the only one
for which we set that expectation. Not only should it tell
you that something is acceptable, but this should have such
a clear regulatory role that it can also define what is
unacceptable.
DR. APOSTOLAKIS: This particular guide.
MR. HOLAHAN: This particular guide; we keep
coming back to having that expectation on this guide. And I
would say this guide is, in a legal sense, no different from
the other guides. It simply says if you do things this way,
the staff will accept them, okay? When you don't do things
this way, they're not automatically unacceptable, but it
becomes very difficult and in this sort of undefined
territory. Now, on what basis am I going to accept this?
And I think that's the nature of the regulatory process we
have. We don't have a clear line between acceptable and not
acceptable in the guidance documents.
DR. PALLA: That's right.
MR. HOLAHAN: And that's not a funny
characteristic of 1.174. That's a characteristic of all the
regulatory guides.
DR. APOSTOLAKIS: But there is a unique aspect of
1.174 that the defined numbers for core damage frequency
have global metrics, and I think we should be very sensitive
not to interpret those numbers as representing adequate
protection. And because they are the only numbers that are
out there, people are stuck in whatever they are doing to
recognize this is not adequate protection, but I'll use the
numbers anyway.
MR. HOLAHAN: And just like all the other guidance
documents, if there's one on an engineering matter, and it
says we should expect a factor of two in some engineering
margin; if the licensee says, well, I don't have a factor of
two in this case; I have a factor of 1.8, then, on what
basis do you say, well, I can't automatically say it's
acceptable; it's not the way we normally do things? Is it
automatically unacceptable? Or does it go to a more
judgmental situation where there is management attention and
other considerations, and we argue about is this a good
safety decision?
And the name we put on that safety decision after
it's done is called adequate protection. But putting the
name on it doesn't make the decision any harder or any
easier. It's just a tag we put on it after we decide.
DR. POWERS: And speaking of tags, I understand
from this flow chart special circumstance is a label that
you put on things and not a checklist of things; that that
checklist really exists on the first box, and it's an
amendment checklist; it's not a rule checklist. You come
out, and you say there's something special about this that
raises risk considerations.
DR. PALLA: Right.
DR. POWERS: Proceed.
DR. PALLA: Actually, all of what we have
discussed is essentially what we've embodied in the
regulatory guidance, and let me just explain how we decided
to structure this guidance. We had deliberated on whether
this belonged in the reg guide; whether this belonged in the
standard review plan. Initially, we were thinking there
should be information, appendices, new appendices to both.
We decided, well, let's put this in a self-standing
appendix, but it really belongs in the standard review plan.
It's not really -- it's really guidance to the staff more
than it is guidance to the industry, and so, we've put this
as a self-standing appendix to SRP Chapter 19.
We've made some minor modifications to the text of
the SRP and also to the text of Reg Guide 1.174, and those
minor modifications essentially indicate that there's a
potential for such special circumstances to exist, and it
makes reference to the new appendix in the SRP.
I indicated on the office letter 803, it made some
revisions in December of last year; that was the first step
towards incorporating some risk information in this document
to help managers screen amendment requests to determine
which ones should be looked at by the PRA branch. We want
to come back to that, and after we've gone through these
deliberations on special circumstances and this process, and
we will be updating that in the same time frame as we'll be
finalizing the other guidance documents.
Within the new appendix to the SRP, we took the
path of least resistance. We basically -- because we laid
these principles out in SECY 99-246; we had quite a bit of
deliberation; a lot of input from OGC on the process and the
league of principles involved. What we've done in the SRP
is tried to consolidate the basic process and rationale into
a concise summary description of how the staff should use
risk; you know, wanting to ask questions about risk; what to
do with the responses; what to do if you don't have
responses; how do you ultimately make a decision; trying to
embody all of that.
So we didn't try to get creative and do anything
new above and beyond the original SECY, except that we did
try to further articulate what are the special
circumstances, and we threw some examples in there that we
intended to clarify, and I expect to have some comments
about those.
DR. POWERS: Let me say that the first dash or the
second bullet really helps me a lot. These situations not
identified or addressed in the development regulations are
important enough to warrant a new regulation encountered on
a widespread --
DR. PALLA: These are -- in the attempt to try to
be more clear and objective, really, these are the two key
statements that we've added to what was in the SECY, so,
yes, if we didn't see it before, and if we saw enough of
them, we would think about moving towards a new regulation.
So that kind of establishes a threshold, if you will, of how
severe does this have to be.
DR. POWERS: Yes; that's a very useful thing.
DR. PALLA: And in the other one, the reviewer has
knowledge that the risk impact is not reflected by the
licensing basis accident, but you have reason to believe
that risk would be an impact and that you would have an
issue if it came in as a risk-informed application; this
kind of speaks to the Calloway issue and, you know, an
electrosleeving type of thing that if you'll look just
simply within design basis accident space, you know,
everything looks great. But then, you know, once you've got
steam generator tube ruptures or, you know, anything induced
by a high pressure core melt, this is a whole different
matter.
It's outside of the design basis and sends a flag
up, and if it, in fact, had come in as a risk-informed
request, we would have an issue there. So that's the second
piece to try to be more clear and objective.
We threw some bullets -- examples of situations in
the form of bullets. The bullets are on the middle of page
three, not of slide three but on page three of the guidance
document that you all have, and I don't know that I need to
go through them unless you have particular questions about
them.
DR. POWERS: I think it would be good if you would
spend some time thinking about your flow chart and see if
you can't get some of these ideas from this slide showing up
in that flow chart as questions that you ask.
DR. PALLA: You're saying in the future --
DR. POWERS: Yes.
DR. PALLA: -- try to embody those concepts within
the process chart?
DR. POWERS: Right now, you have this unique
circumstance or unique situation there which is really just
too nebulous --
DR. PALLA: Yes.
DR. POWERS: -- to really grasp, whereas, you have
some very specific criteria.
DR. PALLA: But those two concepts really are what
can be put in that box space there.
DR. POWERS: You need to get that kind of flavor
across in your flow chart, because that helped me a
tremendous amount.
DR. PALLA: Okay.
MR. HOLAHAN: Although perhaps references in those
boxes to the sections that talk more specifically about the
examples and --
DR. POWERS: That might be better than using words
that are too mushy to grab a hold of in some way because
those --
MR. HOLAHAN: Okay.
DR. POWERS: -- the first one clearly identifies
the kind of threshold you're looking for, and the second one
gives me something quantitative to grab a hold of, some
knowledge that I need to bring to bear on this subject that
allows me to say yes, this is something to worry about.
MR. HOLAHAN: Okay.
DR. APOSTOLAKIS: I notice that we have copies of
something from NEI. How much time will you need, Biff? Is
Biff here?
DR. BRADLEY: Fifteen minutes.
DR. APOSTOLAKIS: Fifteen minutes? So you should
wrap it up soon.
DR. PALLA: I'm almost at the end here.
DR. APOSTOLAKIS: Okay.
DR. PALLA: So why don't I just kind of throw a
couple up here?
DR. APOSTOLAKIS: Okay.
DR. PALLA: As I mentioned before, the
modifications to the case of the SRP and the reg guide are
kind of superficial. They just refer back to the appendix.
They make reference to the idea that even though amendment
requests meet the regulations, there could be a special
situation in which risk is increased, and the staff might be
requesting information, risk information in making the
decision. It makes reference to the appendix, which we
decided to place only in the SRP.
And finally, on the issue of the office letter, as
I mentioned, we've put some additional guidance in December
in the update, and it has some screening questions for
things other than special circumstances, but it does include
special circumstances as one of several. It does not
describe it because at the time, we had basically excerpted
information from SECY 99-246. It wasn't approved by the
Commission yet, and we didn't really have any further
guidance. So we do need to come back to that when we're
done here and bring the office letter up to snuff.
And this, the only other thing I want to say is
give a schedule for where we're going from here. We issued
the regulatory issue summary; it basically informed the
industry that this is the interim guidance that will be
used, and it will stay interim until we finalize this
process. We forwarded the draft mods to CRGR. We'll be
having a public meeting on this next week. We do intend to
take the feedback from the committee from the stakeholder
meeting; NEI, if they have any constructive criticism, roll
that into a revision of this document.
We'll be back here to discuss it, unless you don't
want us to come back.
DR. APOSTOLAKIS: You'll be back here in August?
DR. PALLA: Yes; is that too soon?
DR. APOSTOLAKIS: Are the members going to be here
in August?
DR. PALLA: Yes.
MR. MARKLEY: We do not have an early August
meeting. We have a late August, first of September meeting.
DR. PALLA: It's the end of the month, very end.
DR. POWERS: I think we need the conclusion of
NEI's presentation, and I hope you'll stay here so we can
discuss a little bit.
DR. PALLA: Okay.
DR. POWERS: This quite frankly smacks of a
management and a process issue with marginal technical
content that I'd like to discuss whether there really is any
input you can get from ACRS rather than grammatically and
lexicography.
DR. PALLA: And the end bullet there, September is
our date to try to wrap this thing up and have final
guidance to the Commission. That's it.
DR. APOSTOLAKIS: Okay.
[Pause.]
DR. BRADLEY: I'm Biff Bradley of the NEI staff.
We have with us today Alan Passwater from Union Electric who
is the licensing manager for Calloway. He's also on our
risk-informed regulation working group and has probably had
more direct involvement and participation in how this issue
plays out in application than most of us, if not all, and
some of you on the staff were probably involved as well.
And Al was going to make a few introductory remarks, and
then, I'll briefly go over the industry's positions on the
proposal.
MR. PASSWATER: Good afternoon. I want to thank
you for allowing us some time to talk about this issue. I'm
not here to relive the Calloway experience; once was enough,
and I think Gary and others would agree with that. But what
we did want to comment on today was what's being proposed in
the Federal Register notice that was issued recently, and a
lot of the things I know that we're going to talk about,
you've already hit on, so I don't think we'll dwell on those
issues.
But -- and I appreciate having the opportunity to
give the industry input. I'm on the NEI risk-informed
regulation working group, and we've formed a small subgroup
of that to address this issue, and since I do represent
Amron, UE and Calloway, I won the honor of taking the lead
on this issue as far as the non-risk-informed submittals to
the staff that the staff feels ought to address risk issues.
I think the industry in general, as you know,
acknowledges that the progress made by the staff is great in
utilizing risk insights where they're necessary and where
they provide additional information, and certainly, those
things are now available to the NRC, to the staff, to the
public, and it's not possible to stand up and say we just
need to ignore those where they exist.
I think that most of the concerns or the comments
that we have are more related to the process, and that's a
lot of what you've already discussed today. I think the
major one I want to hit on just briefly here is the fact --
it was in the first box that's in the flow diagram, and it's
a little vague as to exactly who is going to make this
judgment about whether or not unique circumstances exist and
then, from my viewpoint, just reviewing what was published,
it's a little hard to tell exactly what unique circumstances
are.
It kind of leaves the feeling that -- send it in,
and we'll know it when we see it, and from a practical
viewpoint, as a licensing manager, that whole thing comes at
the end of the process, the licensing process, and so, if I
have submitted an amendment that I have no idea needs to be
risk-informed --
DR. APOSTOLAKIS: Now, you're confusing me even
more. I thought I heard Gary Holahan say that 1,500
decisions are made that way, and now, you want more specific
guidance? I thought you were a veteran of these judgments.
MR. PASSWATER: No, I'm saying based on what was
published in the Federal Register --
DR. APOSTOLAKIS: Yes, but you know we don't
always go with the letter on these.
MR. PASSWATER: Okay; I recognize it's early in
the process.
DR. APOSTOLAKIS: Gary invoked the existing
process as working very well. We are doing 1,500 of these a
year; we know what it is; we recognize it when we see it.
And now, you complain that you don't like that.
MR. PASSWATER: I don't disagree with the way the
process has worked to this point, so what I'm commenting on
is what the suggested changes are to the process.
DR. APOSTOLAKIS: You're afraid of the
extrapolation.
MR. PASSWATER: Yes; okay.
DR. APOSTOLAKIS: On the one hand, it works; on
the other hand, it doesn't work. That's my personal
problem, not yours.
MR. PASSWATER: Okay.
DR. APOSTOLAKIS: I'm sorry.
MR. PASSWATER: Well, I think the way it's left
right now -- and I certainly agree that these ought to be
the unique situation, quote, ought to be a very rare
occurrence, and I think that's the intent. I'm not sure
that it's captured --
DR. POWERS: Well, you saw the second two -- I
mean, the two dashed bullets --
MR. PASSWATER: Yes.
DR. POWERS: -- under Bob's previous graph. Those
struck me as clarifying things so that it looked like it
would be rare. I mean, it would have to be something big
enough that if it were pervasive, it could be a new rule. I
mean, that's a nontrivial thing. It's a Brown's Ferry fire
or something like that.
And the second one is that they had to have
something that thought that this really got into a
risk-dominant accident sequence like rupture of a steam
generator tube, induced rupture of steam generator tubes;
you know, you had to have some risk information of some
sort, intuition or what not, that made you think that you
excited a risk-dominant sequence. That's helped me a lot
more than unique, because everything is unique.
MR. PASSWATER: Right; that is helpful, to give
specifics. And in the notice there were some examples
listed, too, which are somewhat helpful, but one of the
comments we've had earlier is that maybe they didn't go
quite far enough, and I'm kind of getting into what Biff is
going to talk about here in a minute, but that really kind
of wraps up what I see as, in general, our comments are and
the fact that they're more process-related than
substance-related as far as what's intended, but again,
having lived with, for many years now, how the SRPs are used
in evaluating license amendment applications, this one is
very unique to what I've seen in SRPs in the past, and it's
very broad. There's nothing it doesn't apply to. It
applies to any possibility I send in, with this judgment up
front that there either is or is not some kind of unique
circumstance.
So that's the front end of the process that I
wanted to comment on, and I will now turn it over to Biff to
talk about some of the specific comments.
DR. POWERS: I guess the take-home lesson is that
this unique situation clause at the first of the box
introduces some uncertainty in your mind on whether you've
done an adequate job in preparing your submission. That's
really the bottom line that you're coming out with.
MR. PASSWATER: That's on the front end of the
process. It's for something that to me is very clearly --
should not have to address risk issues; meets the design
basis, and that is adequate. But then, there's another
judgment added on to the end of that process at the start of
the NRC staff process that may change that.
DR. SHACK: But you agree that -- you know, I
mean, it's hard to enumerate special circumstances, since
they are expected to be unusual. I mean, it would seem to
me that the two general criteria that they've outlined that
are basically those bullets go a long way towards at least
telling you that, you know, I'm not sure I could look for an
enumeration of special circumstances. That's unreasonable.
MR. PASSWATER: No, and again, I'm stealing Biff's
thunder a little bit here, but I think it's more -- it's
definitional more than trying to enumerate all of them. I
agree; we couldn't. That's impossible to do.
DR. SEALE: It does give you room to hunt, though.
MR. PASSWATER: Yes.
DR. SIEBER: I presume the penalty to a licensee,
whether the declaration of a special circumstance is
legitimate or not, is additional time to process the license
amendment and the cost of it; is that correct?
MR. PASSWATER: Yes, and from a practical
viewpoint, to address the specific issues, the probabilistic
questions and the risk issues for someone which I think is
-- and I think this is pretty well representative of the
industry, has not had a complete review done of their PRA,
you know, basically, would require going back and submitting
the PRA, because it's very hard for anyone, including the
staff, to isolate on a specific issue requesting the
amendment, the risk associated with that without knowing
what's the foundation for that in the entire PRA.
DR. SIEBER: So that means a lot of --
MR. PASSWATER: Additional time.
DR. SIEBER: -- transmittals --
MR. PASSWATER: Right.
DR. SIEBER: -- RAIs --
MR. PASSWATER: Time cost and, you know, really
may affect the judgment as to whether or not the original
license amendment is cost-beneficial.
DR. SIEBER: Or the PSA is of good enough quality.
MR. PASSWATER: That would come into play, too.
DR. SIEBER: Thank you.
DR. BRADLEY: Thanks, Al.
I will be brief, because practically every point
I'm going to make has been brought up by the committee
already.
DR. POWERS: And I thought our reviews were
flawed.
DR. BRADLEY: As a matter of fact, you guys are
tougher than we are on the staff.
DR. APOSTOLAKIS: You are such a nice guy, Biff.
[Laughter.]
MR. PASSWATER: It's the new haircut he's got, the
personality.
DR. BRADLEY: As has been stated repeatedly --
DR. APOSTOLAKIS: Very diplomatic.
[Laughter.]
DR. APOSTOLAKIS: Keep going.
DR. BRADLEY: We concur; I wanted to start off
with a positive word, concur is one of them in my bullet
there that in rare instances, as discussed by the staff,
there are situations where you are going to need to address
risk issues. If we look at the history of this, they've
certainly been rare previous to this proposal for this
policy, so I think if we can capture what we've been doing
informally without the benefit of this and move forward on a
consistent basis, then, we will have succeeded in --
DR. WALLIS: We've seen this rare several times.
It bothers me. It's not a criterion. It's the way it ought
to be, and this shouldn't happen very often.
DR. BRADLEY: Right.
DR. WALLIS: But it's not a criterion that it must
not happen very often. If you guys start submitting things
which are on the borderline of what's acceptable, it's going
to happen more often. It's not something which must always
be rare. It ought to be expected.
DR. BRADLEY: Well, one assumes that the body of
regulations today is close to providing adequate protection.
I would presume this would continue to be rare.
DR. APOSTOLAKIS: This brings it back to a
question I've had for a long time: why is the industry so
sensitive to doing risk assessments and using risk guidance?
Is because the infrastructure is not there for a lot of the
licensees? They don't have the -- I mean, they've done now
the IPEs, and you see this sensitivity any time -- there
must be a reason why we should submit this. Oh, but then,
you're actually picking up the regulations if you ask us to
do this. Why? I'm trying to understand that.
MR. PASSWATER: I'll give you my opinion. I think
it's the unknown quantity. I think it's because it's new
and because, you know, I don't know how many times in the
last two years I've heard the metaphor about the two-edged
sword. And we know what one edge is; we don't know what the
other edge is, and I think there is a lot of uncertainty.
There have been some very good submittals made to the staff
and some very good reviews done by the staff on
risk-informed information, and some of those have been very
isolated on specific cases, and those are easy to get a
grasp on and easy to understand what goes along with what
the side effects are of those kind of submittals.
On the ones that are fuzzy, vague, as far as what
the requirements are, I think it's just the unknown quantity
of what's in the next round of questions associated with it,
and in some cases, I think it is also what each licensee has
done as far as the completeness of the PRA and how that
relates to the standards that are being developed as far as
PRA quality, et cetera, that go along with this.
DR. APOSTOLAKIS: So what you're saying is that
the industry is not convinced that its interactions with the
staff will be facilitated by the submission of risk
information.
DR. BRADLEY: I think we're in an evolution.
DR. APOSTOLAKIS: Yes, I understand that.
DR. BRADLEY: We're evolving, and there's a long
history of meeting the deterministic licensing basis and the
huge infrastructure, and it goes back into how that's done
and down to a very small level of detail, and I think the
general concern on the part of the industry is that you end
up having to do both, and as we grow more comfortable with
the use of risk arguments, and we work both sides of the
sword, I think those things will -- those -- that anxiety
will diminish. I know you raise that question a lot,
George.
DR. APOSTOLAKIS: I'm still trying to understand
it. I got a reasonable answer, I think.
DR. WALLIS: I don't think you should call it a
sword. I think you might think of it as a wrench, one end
of which is open-ended and one which is a loop, and they
both have their function in turning the nut.
[Laughter.]
DR. BRADLEY: New metaphor.
We had some, I think, some practical suggestions.
As Alan indicated, we basically view that it is a process
issue and making sure that the process that's laid out is
very clear; it's scrutable, repeatable, all those verbs we
always use and is stable. One of the issues -- and this is
another one of these words you guys always ask a lot of
questions about, but we want the understanding that evolves
here to hold the test of time, and we have had some
instances where regulations and policies do evolve
interpretations over time, and that's -- we don't want that
to happen here.
One element we think would be useful would be to
continue the interim policy of Commission notification.
Given that these are rare instances, given that in my view,
the Commission would want to be aware of situations where
the existing body of regulations is not providing adequate
protection, we don't see that there would be any harm or
burden to continue that policy going forward. So that's our
first firm recommendation.
DR. BONACA: At what stage? I mean, at the moment
of requesting information?
DR. BRADLEY: No; that's a very good process,
where in the process. I think it wouldn't be there, and I
don't think it would necessarily be at the absolute tail end
either, where you reject the request, but somewhere, you
know, at an appropriate stage in the process.
I acknowledge the efforts of the staff to try to
flesh out the definition and the examples and provide more
information on what is a significant and unanticipated risk.
However, we believe there is further work to be done to make
that definition clear. Some of the examples, I guess,
actually in my mind raised new issues to me. I think the
harder ones are not the things that are quantifiable. It's
the things that aren't quantifiable. I mean, you know, in
the extreme, one could suggest that changing your
organization chart in the tech specs is doing this, and
those are the things that are harder to deal with.
DR. POWERS: Yes; I think you raise a good point
there. I can look at something; a hypothetical submission
in which there is no risk information associated, but I
happen to know that at this particular utility, they've
changed their management structure and what not, and it
interfered with safety culture. And that's something that's
not in the PRA, but in most American PRAs, we don't really
deal with safety culture. However, I happen to know that in
Great Britain, for example, well, they did deal with safety
culture on that.
And so, I have some knowledge, then, that safety
culture might be important for risk. That seems like a
really tough issue to deal with, yet it seems like a -- I
hadn't triggered the first one, but I had triggered the
second one.
MR. HOLAHAN: I have to say I think that would be
unlikely. However, I can imagine, say, removing 90 percent
of the boxes on the chart might get your attention. It
would have to be something rather extreme.
DR. POWERS: Yes; you know, the organizational
structure is on the --
MR. HOLAHAN: I don't think that would trigger our
attention.
DR. BRADLEY: I'm not suggesting that that would
normally trigger it. It's just an example that we need for
a clearer definition, as clear as we can get.
Another point: I know NRC notes that the burden
is on them to identify situations. Just -- there is also a
burden, though; it's clear that there's a burden involved in
responding to these types of issues when they come up and
that the only way that burden is clearly going to be on the
licensee, particularly if it involves having to expand or
modify or enhance their PRA in order to provide enough
resolution of the risk impact to counter the arguments the
staff may make that may, in fact, be based on, say, a SPAR
model or a model of, you know, a similar facility where they
are having to infer maybe a similar impact.
And there's really no simple answer to this. I
recognize that the staff can't be expected to meet some, you
know, PRA quality standard or something before they can do
this, but there is a practical issue there that there is a
potential burden that would go back to the industry as well
as -- even though the staff maintains the burden of raising
this.
Finally, this is something I think -- per your
discussion this morning is exactly the same point that in
reading 1.174, there is a considerable text in there
discussing the fact that it was designed with a margin
beyond adequate protection and that the very small, the
small change guidelines could have been made more generous
and still preserved adequate protection, and I guess -- and
I don't know if this is one of the sections of the reg guide
that will be modified as a part of this, but I think this
received an adequate discussion this morning.
One, it's not clear in the SRP whether it's the
very small change or the small change guideline that comes
into play here, and the way the reg guide works is if you
get up between this very small and the small change, then,
you're generally requested to demonstrate complete baseline
CDF including, you know, shutdown and other things that most
plants don't have to quantify. So it's really not clear to
me exactly how 1.174 will be invoked here, and even if it
were that the guidelines of 1.174 are the right guidelines.
They may be the right guidelines for the trigger, and I
think the SRP discusses, you know, here's how we're going to
do the trigger. There's less definition in the SRP of once
we've triggered this, here's how we make the decision on
whether you're meeting adequate protection.
And I'd be the first to acknowledge that that's
not a trivial or simple thing to --
DR. APOSTOLAKIS: The staff says that someone will
determine that.
DR. BRADLEY: Right.
DR. APOSTOLAKIS: On page three, they say that
somebody will determine that. It seems that I'm the only
one puzzled by that.
DR. BONACA: We also heard that it's going to be
one of the inputs to the decision which is typical of the
decisions on amendments.
DR. APOSTOLAKIS: With risk information. That's
not typical at all.
DR. BONACA: No, no. There is information coming;
and that will be an issue, and it can be of any type of
engineering insight that will lead you to grant or not grant
an amendment.
Now, here, there is no information. A new type is
coming out from an application of risk information. It's
available to us as a community. And certainly, we want to
have, you know, have it right -- the situation at Calloway,
where you -- it's a legitimate question that's being
addressed, and risk information is helpful to make a
judgment. And I think in the context where you are not
using PRA information in and of itself to get the call, I
just consider that --
DR. APOSTOLAKIS: Either way, the first subbullet
there, I mean, where it says small change versus significant
risk information; that leads you to a problem. Either you
accept changes that violate 1.174, in which case, that's not
really desirable, or you are leaving it up to people to
decide, you know, what's a significant risk impact. It's a
fuzzy notion.
MR. MARKLEY: Almost all SERs have concluded that
-- almost all SERs that you'll pick up will have a statement
at the end and a conclusion that there was no undue risk to
public health and safety. That's the bottom line of all of
them whether they're deterministic or risk-informed.
DR. BONACA: They must do them in the most
expedient way to bring about information to bear on the
decision process using the guidelines in 1.174. It may not
be the most perfect, but I want to say that the fact that
you just -- make a good decision on a particular amendment,
then --
DR. APOSTOLAKIS: If I submitted the request using
risk information, would they grant me this in 1.174? Yes;
you violate the small change, but what the hell?
DR. POWERS: George, that's always been my
understanding.
DR. APOSTOLAKIS: That they would do that?
DR. POWERS: That if I sent in a risk-informed
application that had something larger than a small change in
CDF or LERF that that didn't automatically get kicked out of
the system. It simply elicited what was graciously
called increased management attention. But there was
nothing in that that said I absolutely could not have any
delta-CDF over some --
DR. APOSTOLAKIS: The increased management
attention is where you have sheds of -- shades of gray --
[Laughter.]
DR. APOSTOLAKIS: Grades of shave, too.
[Laughter.]
DR. APOSTOLAKIS: Approaching it really from below
and maybe a little bit above, but when I look at the figure,
and I see black, don't tell me that is increased management
attention. That's a clear message: don't even come. So I
think we have a little bit of a conflict here, but at least
Biff recommends it.
DR. BRADLEY: I think those were the major points
we wanted to make. I was interested in seeing the staff's
material today, and we were talking about those two
supplements and the one that said situations that would
warrant a new regulation if encountered on a widespread
basis; I thought that was interesting, because that would
imply that you would have to meet a backfit test, even
though for a CLB change request, you wouldn't. But it seems
to me that there's an implicit message in that bullet that
there is a backfit type of issue here.
DR. POWERS: I don't think so; if it's an adequate
protection issue, they don't have to meet a backfit
requirement.
DR. BRADLEY: Well, if it was going to warrant a
new regulation then -- okay, you're saying you would have to
exercise that provision of the backfit rule that says you
weight it as an adequate protection.
DR. POWERS: Weigh it as an adequate protection
issue.
DR. BRADLEY: Okay.
DR. POWERS: But, I mean, your point is still
valid. That gives you an idea of the magnitude of the thing
that you're talking about.
DR. BRADLEY: Yes; and in the second sub-bullet
that would warrant denial if it were evaluated as a
risk-informed application, I guess I think at the point, we
were just talking about where the guidelines of 1.174 don't
necessarily tell you what's a denial. They do talk about
enhanced management review, but I think that is -- this
issue of adequate protection versus 1.174 is basically the
heart of what we're having to deal with here.
Thank you.
DR. APOSTOLAKIS: I must say I'm very happy
hearing Biff speak that way, and thank you very much.
DR. BRADLEY: Thank you.
DR. POWERS: I still have this question: it's a
very process or management oriented issue. And I'm
struggling to say what is the technical content of that? I
think it's very interesting. I'm glad that you brought it
to us. It's always nice to see a modification of 1.174, but
I'm questioning whether there's anything we can contribute,
anything, to this modification.
MR. HOLAHAN: I think there is. I think the
committee made a valuable contribution to the way 1.174
looks, you know, in its original form, and that goes well
beyond just the numerical values or the charts on core
damage frequency and large early release, and if you think
about it, that's only a small part of 1.174. You can call
the rest of it all, you know, process if you like, but it
seems to me that process is the regulatory decision making
process, and I think that is a, you know, valuable and
legitimate role that this committee can address.
Now, I don't call it a policy matter that in the
current regulatory framework, regulatory decisions are made
by process. A policy decision is we're going to have
risk-informed regulation, and it looks more or less like
this. I think the committee views are valuable. I don't
agree with everything I hear at these meetings, but the fact
that they're argued out, I think, is valuable. You know, I
think that we want to follow up with it, and I think that
we'll be delighted to come back to the committee and report
on what we got in public comment and workshops and that sort
of thing.
DR. POWERS: Well, I think we would be intrigued
to hear it.
At that point, I will recess until 20 of the hour.
[Recess.]
DR. POWERS: Let's go back into session, and we're
going to examine a topic that would surprise me if it
elicited a strong comment: the proposed regulatory guide
and standard review plan section associated with NRC code
reviews.
Professor Wallis, is this an area that you think
you can lead us through with clarity and precision?
DR. WALLIS: If you're asking me to do so, I will
be pleased to do so. I wouldn't comment on my own.
DR. POWERS: All right.
DR. WALLIS: The staff has been working on a
review of analytical computer codes used in safety analysis,
and I don't need to tell this committee, but I'd like to
tell the record that this is a very important area; that we
don't have full-scale experiments on nuclear reactors, and
so, all of the predictions on what happens in accidents that
go into making the decisions are really based on computer
models, so they are, for obvious reasons, important.
To start, the industry and the public have faith
in what comes out of these codes. The letter that sent us
these documents did not make a distinction between the old
codes, those of the appendix K type, which are more
conservative than the newer codes, which are best estimate
or realistic, and maybe the staff will tell us a bit about
that. I think that the new documents are more directed
towards assessment activities. And that makes a difference,
because how you estimate these depends on what you want to
do with it, and as you get closer and closer, you may find
-- the staff will find the requirements on these codes make
it tougher, and there needs to be less uncertainty involved
in the process. This needs to come true somehow.
I don't want to make a speech, but I do need to
point out that over 10 years ago, the NRC spent 13 man-years
on the CSAU, the code-scale applicability and uncertainty
applicability and methodology process. This work has been
around for some time, but it's very rarely applied in its
full detail to codes, so I think we need to hear from the
staff about how what they propose is related to that study.
We have codes in the process of review, and others
are coming, so this is very topical for us, and it seems
very appropriate that the staff should be producing a
standard review plan which spells out clearly what
requirements these codes must meet and also a reg guide
which provides guidance about how applicants can go about
meeting those requirements. This isn't just the process for
-- these documents are well-prepared. The whole process is
going far more efficiently with far fewer doubts in the
minds of everybody about what's required; we don't need the
long process of RAIs and replies and re-replies and
requestioning and around and around about what is it you
really want, and you didn't say and so on, because this is
all going to be clear.
The staff has made three presentations to the
subcommittee, and even the last time, we still had questions
about parts of these documents, so we have suggestions for
improvements. And the question was should they -- was it
timely for them to come before the full committee; we went
around that a few times and eventually decided that it would
be worthwhile for the whole committee to be informed about
what's going on to decide what to do, and so, the staff are
here, and I would ask George Staudenmeier to get us going.
DR. POWERS: Just an item of clarification. The
titles sometimes have the word thermohydraulics in the code
reviews and they sometimes didn't. Is it understood that
this is referring to thermohydraulics codes, or is there
some generic applicability of this review plan?
MR. STAUDENMEIER: I think there is generic
applicability, and even within our branch, we deal with fuel
modeling codes. I think we obviously think it's applicable
to that, meaning the essence it's more and more directed at
our branch and most of all, our branch deals with these
thermohydraulics codes, so I guess I didn't intentionally
mean to limit it to thermohydraulics codes, although that's
mostly what we review.
DR. POWERS: Okay.
DR. WALLIS: It used to be clear, but there's a
letter that came from Gary Holahan talking about codes.
MR. STAUDENMEIER: We'll have to, I guess,
redefine the scope on that.
MR. LAUBEN: Excuse me; Norm Lauben. I think in
the standard review plan section and in the regulatory
guide, we talk about the concept of evaluation models being
all the computational engines and things that are required
to analyze those transients and accidents required by 50.34
and standard review plan chapter 15. So it isn't -- most of
the time, it's a thermohydraulic code engine that is the
basic part of this, but there are indeed other things that
are necessary to do an entire analysis to show compliance
with those things.
DR. POWERS: I guess my thinking on this is it's
going to be sufficiently challenging to come up with a good
standard review plan for the thermohydraulics codes, but
trying to add a generic character to it might pose even
additional challenges.
DR. WALLIS: I think, though, Mr. Chairman, if you
read the SRP that a great deal of it would apply to almost
any kind of code.
DR. POWERS: I'm sure some of it will. Maybe we
should be clear about what will be applicable.
MR. STAUDENMEIER: Okay; this is going to be
pretty much an abbreviated version of a presentation I gave
to the thermohydraulics subcommittee recently, and I think
it must have been well-received by everyone based on all of
the I love you email I got afterwards.
[Laughter.]
DR. SEALE: It's nice to know they're thinking of
you.
DR. WALLIS: Too bad we didn't get any in return.
[Laughter.]
MR. STAUDENMEIER: So, this whole effort was
started as a result of Maine Yankee ISAT and an NRR review
of the Maine Yankee problems and some points that were
brought up by the ACRS in other reviews that pointed to
generic problems in our code review process and also in the
documentation that's submitted by the industry for having
codes reviewed, so we are trying to put corrective actions
into place to try to correct these perceived generic
deficiencies in the process, and we want to deal with
adequacy of code documentation; of code assessment and
inconsistencies in the staff review process.
And the next slide is, I think, what a lot of
people perceive the development process to be for a lot of
the codes we get in here to review, and I'm sure we could
point to examples of many of these in a lot of the things we
review.
DR. POWERS: I particular appreciate the
management direct response.
[Laughter.]
MR. STAUDENMEIER: I'm sure you can construct a
diagram of the staff review process.
[Laughter.]
DR. APOSTOLAKIS: This can help with the flow
chart.
[Laughter.]
MR. STAUDENMEIER: I said you don't think it's
consistent with the findings?
[Laughter.]
DR. WALLIS: Let's move on, shall we, before
George asks a question?
[Laughter.]
MR. STAUDENMEIER: Okay; to address these
perceived problems, we're developing a standard review plan
to document the code review process and give guidance to the
staff for code reviews and model reviews and develop a reg
guide and a standard format and content guide to provide
guidance from the industry on what we think they need to
submit when they submit an evaluation model.
These are the topics we've tried to cover in the
standard review plan, leading off with the documentation,
which really is a definition and description of the whole
evaluation model, and as part of this documentation, we
think there needs to be something that defines the accident
scenario, defines what physical processes and physical
components need to be modeled to evaluate the accident.
There needs to be a section on the equations that you're
using for your mathematical model in order to evaluate the
accident; assessment of your mathematical model to show that
it's actually good at making predictions for what you're
trying to model.
DR. APOSTOLAKIS: Why isn't there a statement here
that your primary review is also an assessment of the
uncertainties in the predictions? Why should their
assessments be good?
MR. STAUDENMEIER: I think that's actually
discussed in the standard review plan. There is --
DR. APOSTOLAKIS: It's not important?
MR. STAUDENMEIER: No, I believe it is important
in assessment, as part of the assessment process and
determining what the uncertainties are and also in
quantifying your overall uncertainty of the calculation, and
I think that has to be less than what your safety margin is
in the calculation.
DR. WALLIS: Code uncertainty is one of your
headings along with these headings in the SRP, and you have
here some headings which are not headings in the SRP, so how
do you propose to bring that into line? The code
uncertainty isn't just the same as code assessment and code
theory in the actual process. But I think some other things
like analysis doesn't come out as a heading. So it's there
somewhere, but it's not organized the same way. It would
help if -- I think you need to be clear on what should be
headings, because eventually, they're going to check this.
MR. STAUDENMEIER: Okay; quality assurance is just
a process that needs to be in place to -- as a check process
to confirm that the development of the code and maintaining
of the code is done under a process where the quality is
actually controlled and that there is review of what is
being done in order to ensure a better product. Sometimes,
we do confirmatory analysis in order to do an independent
check on the code. There's a section covering that, since
that's -- it is often part of our review process.
We think we need to make some statements about
revisions to previously approved models, because a lot of
what we get in isn't a fresh code from scratch; it's a base
code with maybe some new model added to it, some new
modeling capability or some error corrections that have been
made in the code, and in the future, we're planning on
adding appendices for specific accident classes and types
that deal with modeling requirements and specific code
assessment for developing a code for certain type 10 and
certain classes of accidents.
Organization is a standard SRP organization; if
you want to look through the SRP, SRP sections generally
follow this type of organization. It covers areas of review
which describes the scope of review for whatever the section
is on. Acceptance criteria for the whatever you're doing;
the review procedures; the evaluation findings and
references that go along with the chapter.
We've gotten some informal comments from the
subcommittee that we're going to address before we release
it for public comment. We also have to go to the CRGR for
review before we put it out for public comment. Then, we
plan on putting both the SRP and the reg guide that Norm
Lauben is going to talk about that's a companion to this out
for public comment. The proposed public comment period is
going to be 75 days. It could be made longer. That's the
minimum you can have; if there's any desire to have it out
for longer than that, that could be changed.
And hopefully, our original schedule is to try to
release final versions of the standard review plan and reg
guide by the end of the calendar year. That will determine
how difficult the comments we get are to resolve. And then,
in the future, develop these appendices for specific
transient or accident classes that would cover specific
modeling requirements or assessment requirements for
specific accidents.
DR. WALLIS: I have to ask a question on the best
estimate versus conservative and so on. Are you addressing
best estimate codes here or --
MR. STAUDENMEIER: I think any code in general,
because, see, appendix K even has parts of it that are
really best estimate types of models, and if you read the
requirements for blowdown and heat transfer, I think it
specifically says -- it describes what you would probably
consider a best estimate model except for the grandfathered
case of allowing a Dougal-Rosenauer correlation, but I think
everybody -- I don't think there's a vendor left that has
the Dougal-Rosenauer correlation.
DR. WALLIS: It seems to me the real message is
it's got to be an adequate estimate, and for appendix K,
certain types of estimates are adequate, and best is really
the wrong thing; adequate for whatever purpose the submittal
is made. If it's made for some preparation of something
that gets you awfully close to a margin, then you're going
to examine it much more carefully. It's got to be a better
estimate. I wonder if that comes through or not. It's not
-- you're using some requirements for all codes. It depends
on the use and how carefully you scrutinize a particular
application.
MR. STAUDENMEIER: I guess in the case of appendix
K, there's no requirement for an uncertainty analysis; it's
perceived that some of the prescribed requirements in there
are good enough to keep conservatism in the overall model.
Many models are best estimate that are in appendix K,
though. Blowdown and -- models are pretty close to best
estimate.
DR. APOSTOLAKIS: The draft regulatory guide says,
in section 2.2, that there are three elements to this CSA
methodology, the third one being sensitivity and uncertainty
analysis that have been recast here that address sensitivity
and uncertainty which is under evaluation more than adequacy
assessment.
But then, when I go there, I try finding what
sensitivity and uncertainty analyses are expected.
DR. WALLIS: Are you asking a question about the
reg guide?
DR. APOSTOLAKIS: Yes.
DR. WALLIS: We're going to hear about that later.
Maybe it's best to raise the question that I think Norm --
the SRP is a separate document. I think the view of the
subcommittee was that the SRP and I guess our interpretation
is -- the purpose is to lay out very clearly what the staff
is going to look for in terms of documentation and code and
what they're looking for in performance, et cetera. The
guide is then going to make it more explicit about how one
might go about meeting those criteria. SRPs are quite
sparse; it's a high-level document.
MR. ELTAWILA: Can I make a -- this is Farouk
Eltawila. The SRP is guidance to the staff on the conduct
of the review. It does not institute requirements on
applicant or licensee performance and thus is a distinction
between requirement and guidance.
DR. WALLIS: But what are the requirements, then?
DR. APOSTOLAKIS: That's the problem.
DR. WALLIS: Are the requirements so vague that
it's all up to someone's guess or requirement?
MR. LAUBEN: Excuse me; Norm Lauben again.
Requirements -- are they going to address it?
MR. ELTAWILA: Yes.
MR. VERMEIL: The requirements, Dr. Wallis, are
specifically in 10 CFR 50.46 and appendix K and the
associated 50.44 and the associated parts of 10 CFR Part 50.
The guidance for how to implement those requirements is what
is provided in a regulatory guide such is the one that
research is developing and Norm will be talking about.
It is just that: it's guidance. It's not "a
requirement." The requirement is only in the rule.
DR. WALLIS: I understand the guidance part, but
the SRP essentially lays out a plan for the staff. The
staff is going to look for certain things.
MR. VERMEIL: That is correct; you are --
DR. WALLIS: So maybe I'm in error not to go back
to some earlier source, but I'm assuming that the SRP is
taken seriously, and if it says that --
MR. VERMEIL: Very.
DR. WALLIS: -- for the field equations should be
based on the natural laws of physics and not violate some --
that's actually a meaningful statement. That's not just
something that can be -- but it isn't in 10 CFR.
MR. VERMEIL: I'm not trying to say that that's
not a meaningful statement. It very much is. What I'm
trying to say, though, is there may be other ways than a
particular prescriptive manner of satisfying what the SRP
says the reviewer ought to be looking for.
DR. WALLIS: I understand that, but the fact that
the SRP asks for, say, a scaling analysis doesn't mean to
say that you can then not do it.
MR. VERMEIL: No, no, that's absolutely correct.
What it says is if there is to be one, then, that is what
the reviewer looks for. That's absolutely correct.
DR. APOSTOLAKIS: Actually, the SRP is just -- is
more specific regarding acceptance. So the reg is -- by
that specific advising.
DR. WALLIS: I see the SRP as something that a
professor might give to the students and say look, when you
submit your thesis, we're going to look for these kinds of
things.
MR. VERMEIL: That's exactly right.
DR. APOSTOLAKIS: I think you should also issue
another statement to the students as to what is acceptable.
MR. STAUDENMEIER: Okay; if you have any other
questions --
DR. APOSTOLAKIS: It just occurred to me: I don't
recall an instance where there weren't ACRS comments were
not used immediately with CRGR review. Are we sister
organizations? The staff comes here; immediately, CRGR
review follows ACRS comments.
MR. STAUDENMEIER: We're required to go through
both.
MR. VERMEIL: Yes; the process, Dr. Apostolakis,
requires -- I'll use that word -- that the staff go through
ACRS and CRGR before issuing documents of this kind for
public comment.
MR. STAUDENMEIER: And they often come with
conflicting requirements.
MR. VERMEIL: Yes, they do.
MR. STAUDENMEIER: Conflicting comments that you
can't satisfy both.
VOICE: That's to help you.
[Laughter.]
DR. WALLIS: Let me ask you about the CSAU. CSAU
was a major effort, 13 man-years of study, which was then
going to sort of answer the question how is the agency going
to handle this best estimate code type of thing coming
along. I can see how that might not play such a prominent
role in your stuff, because you're sort of saying these are
the requirements, and CSAU more addresses how to actually
satisfy them, but it's a great edifice in the past. There's
nothing since then, so you can't ignore it. And it may be
well of us to say what is it that you do that incorporates
it or replaces it or something, because it was the big
effort that even got people like me to sign off on it.
So, it's an authoritative document put out at that
time; how does it relate to this?
MR. LAUBEN: Let me just say that every activity
that we engage in usually has a specific purpose for which
it was done, and the CSAU was done in order to demonstrate a
method of how you evaluate code uncertainties in order to do
best estimate calculations. In order to get to that spot,
there are certain principles that you had to adhere to. But
most of those principles were not necessarily things that --
for which a lot of time and effort was spent on in CSAU.
However, when it comes to code development and
assessment, there are more things up front that you need to
spend time on, and we needed to do a little bit more
original work such as were reflected in a number of your
comments that would allow you to do this. Therefore, even
though those things are mentioned in principle, a very small
amount of the CSAU effort was spent on a lot of the up front
principles. So what we need to do in this document and what
we had to do, for instance, in the code development and
assessment work that we've done for AP600 was to spend a
little bit more time on the process up front.
Now, what -- specifically let me say that CSAU
assumed certain things to be true before the process was
even begun. It assumed that the code was complete, which
means that it wasn't necessarily going to be an extensive
review of the structure of the code. It assumed that all of
the assessment was complete. It assumed that the code was
frozen, and it assumed that all of the documentation was
complete and correct which -- now, many of these assumptions
may not be true; however, CSAU assumed that before they did
this, it had to be true.
So that's one thing that had to be emphasized a
little bit more in this kind of work. When we got to the
AP600 assessment, it assumed once again that much of the up
front work was done, and nobody was going to question the
code particularly, but rather, they were going to now
evaluate the code's ability to do AP600 analysis. So once
again, there was a lack of being able -- a lack of stating,
I think, as clearly as needed to be the principles that were
required up front.
So I'm just saying every time is a slightly
different purpose. SASM is the same way. SASM was trying
to look at experimental work. It didn't have much to do
with code -- with code development. So even though the
structure of these things may look similar in all three
instances, in all three instances, they had a slightly
different purpose for which they were done, and it's
important, I think, that we need to look at this is this is
something about a code development and assessment practices
that are required for a variety -- a variety of chapter 15
events, not just loca, for instance.
So, to a large degree, we have to make sure that
we get the basic principles, if you will, the high level
principles stated correctly and applicably for all of the
things we're trying to do.
DR. WALLIS: You've helped me a great deal,
because I think you have clarified for me one of my problems
with your document is that you say that CSAU, a lot comes
before CSAU. There's a lot of stuff before CSAU. CSAU
addresses how do you do scaling and uncertainty and
applicability.
MR. LAUBEN: Right.
DR. WALLIS: Yet, the reg guide does it the other
way around. It brings in CSAU at the beginning. I mean, we
should be talking about generalities like getting the
equations right and the principles. But later on, it
abandons it just when it's talking about uncertainty and
scaling.
MR. LAUBEN: I think sometimes, when fresh eyes
look at something, and you're used to being in the middle of
something, it helps to restructure your own thoughts as to
how you really want to order these things. I don't think it
necessarily means that there's an enormous amount of work
that needs to be done, but rather, you need to find all of
the principles and structure them in the right way that it's
a scrutable way to go about doing it.
DR. WALLIS: Are there questions for Joe? He's
really on the spot first?
[No response.]
DR. WALLIS: All right; shall we move on to Ron
Lauben, please? We move to the subject of the reg guide
that goes with this SRP.
MR. LAUBEN: Okay; I guess I won't put the first
slide up. It just says who I am and what I'm going to talk
about, and the purpose has already been discussed: to
present the background and content of draft guide 10-96, a
reg guide for transient and accident analysis methods used
to analyze events required in 10 CFR 34 and defined in SRP
chapter 15. When I talk about SRP chapter 15, I mean those
parts of chapter 15 that already exist to discuss about --
that describe specific transients in accidents.
Joe's subchapter of chapter 15 is going to discuss
the code requirements that would go to all those transients
and accidents that are specifically spelled out in chapter
15, the rest of chapter 15, chapter 15, once again, being
subservient to 10 CFR 34, 50.34, which says -- and appendix
A, which states that you need to do these transient and
accident analyses.
DR. WALLIS: Now, one approach here would be to
say here's the SRP; here are the listed things that it asks
for --
MR. LAUBEN: Yes.
DR. WALLIS: -- and then, here is some guidance
about how you meet them. So your structure would follow his
structure.
MR. LAUBEN: Well, I think that's a good point. I
think that is kind of the way it is usually done, although I
must say that in some cases, there may be -- in fact, this
has generally been the case over the years -- there are SRP
chapters for which there are no reg guides at all, and
that's usually the case, but, in fact, loca and rod ejection
are the only two exceptions to the 31 events in chapter 15.
DR. WALLIS: But the reg guides exist probably
because there have been difficulties --
MR. LAUBEN: That is correct.
DR. WALLIS: -- in interpreting, and therefore,
guidance is necessary.
MR. LAUBEN: Reg guides in the past, those two
events, the rod ejection and the loca, have been two of the
more difficult ones, so that it required some more guidance
on those two.
DR. WALLIS: Then it should be based a bit on
history.
MR. LAUBEN: Okay.
DR. WALLIS: If it turns out that there's a
requirement, that, say, the code should have equations based
on fundamental principles and approximation should be
explained and so on, then, maybe the guide should address --
should say something helpful to avoid mistakes being made in
the future of the same type.
MR. LAUBEN: I think the more I'm listening, the
more I realize that this has to be a fairly high level
document, and it has to be scrutable, and I think that to
worry about a great number of details might make it less
understandable to the people that need to understand it.
Okay; basically, what I planned to talk about
briefly was background and need for the reg guide; why do we
need it; the contents of the draft guide; your comments that
we got before the last meeting and status and summary. And
the thing that really started this was the Maine Yankee
independent assessment team that looked at -- actually
looked at non-loca transient and accident analysis methods
at Maine Yankee. It started with loca, but that was a
separate team; I think Joe was on that team that looked at
the Maine Yankee loca methods, but our charter with the ISAT
was to look at non-loca, and what we found out was hey,
there were some things about -- no surprise -- how they were
using the RETRAN code that caused us to have certain
questions.
And we decided then that what was really needed
was something to ensure sufficiency and consistency in the
level of documentation and validation. Now, this is not a
whole process by any means, but it says these were two
things that we identified at Maine Yankee; have a documented
process in place to identify and rank key phenomena for
relevant events, which is then used in code development and
assessment process. Once again, this is not necessarily
starting at the top, but these are simply the things that
were identified there.
And then, to implement this, the NRR Maine Yankee
task group recommended that a standard review plan section
be developed and a regulatory guide. Now, I put this last
bullet in because it's just what I said, is that the items
identified by ISAT by themselves do not constitute all the
parts needed for a description of transient and accident
analysis methods. Additional items, such as determination
of basic code capabilities, are being added to the draft
guide. So the draft guide, once again, wasn't -- excuse me;
the lessons learned from Maine Yankee were the lessons
learned from Maine Yankee. They weren't necessarily a
comprehensive set that would be needed to write a draft
guide.
DR. WALLIS: When you work on a problem, and
you're devising a solution, the reg guide is a solution to a
problem here, you need to set up the needs, and I wonder how
you determined what kinds of needs the reader of this guide
has.
MR. LAUBEN: Okay; the -- okay; that's -- I think
someone -- I think it was Dana --
DR. WALLIS: I can see with the SRP; I read it
through, and I can see obviously, the staff is addressing
its needs in reviewing.
MR. LAUBEN: Yes.
DR. WALLIS: But I have more trouble with the reg
guide. I mean, I'm trying to think: now, I'm an applicant
trying to write some stuff that I'm going to submit to the
staff looking for guidance, and I need -- it's not so clear
that the way it's structured helps me, so I wondered how you
found out what's the need out there that you're meeting.
MR. LAUBEN: I suppose the need that -- the one
that I identified most closely was the fact that when we
went up and looked at their message at Maine Yankee, we said
ah, this immediately stands out. You don't have assessment
that applies to the cases that you're doing, so you need
some guidance on assessment. And like I say, this sort of
ends up -- you're injecting yourself into the middle of the
process. Nowhere did we say -- well, in fact, I shouldn't
say nowhere did we -- we did not say in the report; however,
part of the review team did look at some of the fundamental
equations and had questions that, in some of the very, you
know, very simple codes that they used that their
fundamental equations and models probably did not model the
things that they needed to model to.
We didn't take that away as a serious enough
lesson from that review, but I think what I'm saying is the
way I came to it was what were the lessons learned from
Maine Yankee? And that isn't necessarily comprehensive
enough.
DR. WALLIS: See, if I can just speak for myself,
when I look at these codes, I think the first thing I'm
looking for is do the people running this stuff understand
what they're doing? Do they have a sort of professional
competence which shows that they understand the equations
they're using and the problems they're using them for that
they're appropriate and so on? That's the first sort of
impression I look for, and that message has got to get
across to whoever writes the documentation, besides QA and
all that, has to establish that you really believe this
person is doing a good job. That message has got to get
across first.
MR. LAUBEN: What I hope is that when I get to my
new road map, it will help a little.
DR. WALLIS: Okay; thank you.
MR. LAUBEN: Okay?
I think most of the things on this slide were
talked about by Joe a little bit that, you know, in fact,
you mentioned them, Dr. Wallis, that we came to the ACRS
three times and had three drafts; tried to factor in the
committee comments from the 11/17/99 meeting, and we're now
going to try to address some of your comments from this last
meeting as well.
And just so that people have an idea of what the
draft guide 10-96 has in it, we proposed, and once again,
these are proposals based on the Maine Yankee and I think
also the Siemens reviews, that we address analysis methods
for all chapter 15 events on a generic basis, stressing
verification, validation, documentation and quality
assurance; that we describe application of the evaluation
model concept, which includes all computer codes and
analysis methods and everything else not included in the
computer programs to show compliance with the analysis
required by 10 CFR 50-34; in other words, it may not just be
a single computer code that you need to show compliance with
those things. It may be a series of computer codes, other
analytical features, a lot of input and so forth.
Now, I thought that -- we thought that it would
probably be a good idea to stress some of the key principles
here of evaluation model development and assessment, and the
reason to do this is to show that this -- well, that these
are the things that are going to be featured in the guide
first of all. And one of the first things is determining
requirements of the evaluation model and the importance of
the key systems, components, processes and phenomena and to
note that a process like the hierarchical system
decomposition should be to assure -- should be used to
assure that all levels of the evaluation model development,
including are the equations the right equations, are
properly considered, and I think that's --
DR. WALLIS: What do you mean by hierarchical --
MR. LAUBEN: Well, this was the thing that if you
look in the SASM report decomposes all of the things that
are needed in the code including what phases, what, you
know, what is it that you really need to do with the code
that you're going to use, including appropriate phasing
equations and so forth.
DR. WALLIS: Does this say something about what
comes first? And I, again, in reading your reg guide, I
read this whole thing --
MR. LAUBEN: Yes.
DR. WALLIS: -- and I find that top down is sort
of long away, deep down at the end of it.
MR. LAUBEN: Right; it is.
DR. WALLIS: You don't put it first; if you don't
have --
MR. LAUBEN: Right.
DR. WALLIS: -- good equations; if you don't show
that you understand them, then forget the rest of it.
MR. LAUBEN: Yes; I think --
DR. WALLIS: That's the first thing you do. Then,
you go on to something else.
MR. LAUBEN: I almost made a copy of what it was.
DR. WALLIS: Yes.
MR. LAUBEN: But it's in there very early in the
SASM document, and it says --
DR. WALLIS: See, I didn't get from your reg
guide --
MR. LAUBEN: No.
DR. WALLIS: -- this -- it goes right into other
stuff.
MR. LAUBEN: Right.
DR. WALLIS: Before saying --
MR. LAUBEN: It does.
DR. WALLIS: -- you've got to establish --
MR. LAUBEN: It does.
DR. WALLIS: -- a base for this thing first.
MR. LAUBEN: Exactly; I hope when I get to the
road map, we'll say a little bit more about that.
DR. WALLIS: Hierarchical implies that there's
some sort of priority about --
MR. LAUBEN: Yes.
DR. WALLIS: -- going about this.
MR. LAUBEN: And when you decompose things
hierarchically, you have to make sure you've got the right
equations and the right way to do them and the right
numerics and so forth at first. So that's in there.
Second is once you know what you need, then, you
develop an evaluation model that meets those requirements.
Secondly, you develop an assessment base appropriate to the
requirements; and now, you're down to part of the CSAU part,
but you need to do those first things first and then assess
the adequacy once you've done your experiments and your
assessment base, and that's the new part. So here, we're
introducing the concept of uncertainties here.
DR. APOSTOLAKIS: So this is not in the draft
guide? I'm having difficulty finding it.
MR. LAUBEN: Not very clearly, and we are --
DR. APOSTOLAKIS: It's not at all.
MR. LAUBEN: Excuse me?
DR. APOSTOLAKIS: I can't find it at all.
MR. LAUBEN: Okay; I think that if you -- here's
the part that bothered me about uncertainties when I was
writing the draft guide, and that is how you consider them
for each transient or accident may be a little bit
different, and that is one of the main reasons why we were
going to write appendices, and I think we address
uncertainties in that appendix on loca, but what -- I think
there may be some more general ways that uncertainties
should be addressed, and I hope -- I'll show those to you in
the next slide, but I agree with you: until you read the
appendix, you probably don't see very much if anything about
uncertainties in the draft guide.
DR. APOSTOLAKIS: Now, what's the distinction
between -- SA stands for sensitivity analysis?
MR. LAUBEN: No, no, SA is scaling and
applicability, scaling and applicability. In other words,
when you're developing the experiments that you need to use,
you need to consider scaling, and you need to consider the
applicability of those experiments. So that's the SA part.
DR. WALLIS: This is in answer to the question
that I had, that you said at the beginning of your document
there are some principles, but I couldn't find the
principles.
MR. LAUBEN: Right.
DR. WALLIS: You're saying you're now going to say
what they are.
MR. LAUBEN: Here they are, right.
DR. WALLIS: And the rest of the document is going
to reflect those principles.
MR. LAUBEN: Well, in fact, I hope that the stress
and the revised document is going to be on the principles
and, you know, structurally how you --
DR. WALLIS: Well, it's a teaching tool, isn't it?
MR. LAUBEN: Well, yes, in a way.
DR. APOSTOLAKIS: But let me come back to this
uncertainty issue. On section 2.2 to 2.3.4, adequacy
decision, there is a statement that the process continues
until the assessments indicate that all important phenomena
are acceptably predicted. I don't know what that means when
you do an uncertainty analysis.
MR. LAUBEN: Okay.
DR. APOSTOLAKIS: The uncertainties are small
enough, you mean?
MR. LAUBEN: I'll tell you what: I think the
trouble with this is that people a lot wiser than me have
been struggling with this for years, and I don't -- other
than the discussion that we have in here on words like
excellent, adequate and so forth, which were indeed in
themselves a struggle for people to come up with, I don't
know -- and if somebody else wants to raise their hand and
talk about this difficult subject, I'll let them do it.
DR. APOSTOLAKIS: To make my point clear --
MR. LAUBEN: Yes, okay.
DR. APOSTOLAKIS: -- I don't think that this
statement makes sense if you have already asked them to
quantify the uncertainty. You have to be a little bit more
specific.
MR. DEMARTS: Can I interject?
DR. APOSTOLAKIS: Yes, go ahead.
MR. DEMARTS: This is Mario Demarts of research.
Basically, you have a set of tests, experimental tests,
which have some uncertainty. You can't repeat the test
twice and see if you get the same thing. So you have an
uncertainty on the database that you're going to use to
check your code against. On the other hand, you have your
code, and the code is essentially a mandatory condition
which can be ranged a little bit.
And so, at the end of that, depending on what
you're doing there, in the code, you have a certain range of
answers. In other words, in the CSAU methodology,
basically, you are looking at the key parameters that
controls your answer. You range them over the range that
you think this parameter has, and you see how they influence
the surface of responses. Therefore, you get an error back,
if you wish, on the answer of the code. You have also an
error back of the experiments.
Now, within that set, if you can say that you are
consistent, then, you say this is okay.
DR. APOSTOLAKIS: Yes; and my comment to that is
that this is the view of uncertainty that a deterministic
guide would have. If I were to quantify the uncertainty in
a particular code, I would develop a distribution using the
results of experiments, and the key element there, because
I've tried to do it, and you will have a hell of a problem
deciding or distinguishing between the sources of
uncertainty; in other words, is this deviating from a code
prediction because of some systematic issue? I have not
modeled it very well? Or they used another boundary
condition of which I'm not aware? Or is it just the
standard, you know, random variability that you expect when
you do measurements like that?
But in any case, I would develop a distribution,
and then, I would make a judgment as to the adequacy of the
prediction based on the distribution, because it may be very
broad but still acceptable depending on what I'm trying to
do. So I would not try to do sensitivity studies and change
key parameters and look at the range of answers from the
code and say, well, gee, if this is what I get here, and
this is what I have from the results, they look reasonably
similar. That's my view.
MR. DEMARTS: I can address that; in transients,
there are basically -- you have to look at transients in
that basis. In other words, a transient is very
deterministic in one phase, when it reaches a situation
where there is a bifurcation possibility. At that point, it
bifurcates one way or another, so if you wish to consider
that as a standard of uncertainty, basically, it will be all
over the place, okay?
So, in a portion of the transient, where, say,
nothing really extraordinary happens, what I just talked
about is suitable. In a portion of the transient where,
say, something happens if you reach that level; let's say a
level drop to a certain level of activation in an ADS system
stats, something of that kind, you've got to be very careful
in the definition that I just provided if you trigger that
or not. That is the kind of things that you have to look at
very, very carefully on this.
DR. APOSTOLAKIS: I suspect that I will have to
understand the issue much better. I can express a view
anyway. It seems to me that this is clearly an uncertainty,
but Dr. Wallis talked to us some time ago about various
assumptions, some of which are completely wrong in some of
the codes, but it seems to me that you have to do some
assumptions in approximate models, and these introduce
systematic deviations. And I'm talking about the modeling
uncertainty, in other words, not the physical thing when is
ATWS -- I mean, ATWS, auxiliary feeding water system is
going to come in and all that.
This is of a different kind, and in fact, this
raises a question: why isn't there a clear discussion of
these various kinds of uncertainties in this?
DR. WALLIS: I think the point that needs to be
raised here is the difference between the words acceptable
and adequate. An acceptable code, I think you have to spell
out, has to do enough so that the uncertainties can in some
way be estimated, in both modeling and other kinds of
uncertainties. A requirement in the code is that the
uncertainties should be assessable, measurable and expressed
in some form.
Whether this is adequate or not for some transient
is a different decision altogether.
DR. APOSTOLAKIS: Yes, yes.
DR. WALLIS: You cannot say, as you seem to say in
the reg guide, that it's adequate if it's acceptable.
They're different ideas altogether. It's -- a code is not
acceptable unless it can assess and tell you how uncertain
the predictions are.
Now, given that it makes predictions with a
certain uncertainty, it still may not be adequate, or it may
be adequate depending on what you want to do with this.
You've got to make that distinction.
DR. APOSTOLAKIS: But I also think, coming back to
my earlier comment, that a good discussion of the various
kinds of uncertainties that you are encountering with this
would be very helpful, and it shouldn't be hard to do. I
mean, there is a group working on pressurized thermal shock
that did that in their context, and they did a very good
job. Maybe you can talk to those guys.
MR. ELTAWILA: That's the same group that's
working on the thermohydraulic uncertainty. That's
Professor Magares.
DR. WALLIS: George, if you look at CSAU, there's
a whole box in the CSAU diagram that tells the different
kinds of uncertainties, right up front. The U in CSAU is
uncertainty. There's a lot in there already about
uncertainty.
DR. APOSTOLAKIS: I don't see it here.
DR. WALLIS: No, okay.
DR. APOSTOLAKIS: I'm sure there are documents
somewhere there.
DR. WALLIS: Ten years ago, but here, investigated
very thoroughly, and then, in light of great developments
and understanding of uncertainty since then, I'm sure it can
be improved.
DR. APOSTOLAKIS: And I would like to see a better
blend of the CSAU which really comes from the
thermohydraulics community and what the PRA types have done
in that form, and I think the PTS is a good example of that
blending, identifying the parameters; which ones are
epistemic, blah, blah, blah, that will help a lot. But
right now, I just don't see anything.
Anyway.
MR. LAUBEN: Like I said; to the degree that
differences in how it's going to be handled in different
accidents, we thought that this might be transient-specific.
DR. APOSTOLAKIS: Yes, it is.
MR. LAUBEN: Okay; and let me just continue on.
These first four principles that we've discussed have to do
with the process. The last two have to do with just good
engineering practices; in other words, when you do this,
establish an appropriate quality assurance protocol during
evaluation of the development and assessment process and
secondly provide comprehensive, accurate, up-to-date
documentation. And I don't -- I hesitated to say much more
than that, because every step of the process has to be
informed by these two principles.
DR. WALLIS: You talked about peer review in the
beginning of your document.
MR. LAUBEN: Yes.
DR. WALLIS: And then, you don't sort of reinforce
it later on about when it should be used. And by quality
assurance, you mean using peer review?
MR. LAUBEN: Yes; I do mean, and I think I said it
in here --
DR. WALLIS: It is included.
MR. LAUBEN: But it's going to be put in a
separate section; the quality assurance thing is going to be
a separate section by itself, and it's not going to be
looked upon like an appendix to this.
DR. WALLIS: The last thing you guys need is to go
through all this stuff and come to the ACRS, and the ACRS is
your first peer review.
MR. LAUBEN: Yes, okay.
DR. POWERS: Let me ask some questions about the
quality assurance here.
MR. LAUBEN: Yes.
DR. POWERS: I would assume that you meant NQA II.
MR. LAUBEN: Well, I mean, by it abiding by the
applicable parts of appendix A, and when it comes to code
development and design --
DR. POWERS: Are there certain code design
criteria?
MR. LAUBEN: Appendix B; I'm sorry, appendix B.
There are certain things that are talked about in there that
have to do with design reviews, and I discuss them in a
paragraph or two in the reg guide, and if there is something
additionally that should be emphasized, that would be --
DR. POWERS: I think I'd -- in NQA II, that people
repeatedly try to look at -- which is essentially appendix
B, the code quality assurance, and it's just never -- it
just really wasn't --
MR. LAUBEN: That's right; I think that's probably
true, and that's why we had to add the words about peer
review. Because I think that's -- it's more than just a
management check; it's a peer review.
DR. POWERS: I think the problem should be
configuration control and things like that.
MR. LAUBEN: Yes.
DR. POWERS: There are other standards to appeal
to.
MR. LAUBEN: Yes.
DR. POWERS: And I think they work better. I
mean, I almost hesitate to bring it up. There are certain
fairly onerous undertakings, but quite frankly, the quality
assurance on these codes is a very onerous undertaking. I
think I would appeal to something besides appendix B as
guidance on this.
MR. LAUBEN: Well, I do have to -- okay, that's
fine, except I just want to say I do have to tie most of
these things to the regulations.
DR. POWERS: I understand why; I understand.
MR. LAUBEN: All right.
DR. POWERS: I mean, what they tried to do in NQA
II was to take what's in NQA I, which is essentially
appendix B, and say okay, here's how you apply these things
to codes, because people were having so much difficulty
taking the words out of what was basically a hardware
quality assurance document and applying it to codes.
DR. SEALE: That's always a problem.
DR. POWERS: And let me ask you about the last
one: provide comprehensive --
MR. LAUBEN: Yes.
DR. POWERS: -- and accurate up-to-date
documentation. It's the comprehensive.
MR. LAUBEN: Well, I think that if you don't have
comprehensive documentation, there's no hope for any peer
review group to know what the heck is going on. So I think
that's what -- also, up-to-date. If it's -- accurate is
obvious. Up-to-date means -- and we could see that when the
original CSAU was done, the people complained that the
documentation for the code was not up-to-date, and
therefore, how can the peer reviewers, the CSAU people, do
this if they do not have comprehensive or up-to-date
documentation?
DR. POWERS: Yes, if it's not up-to-date; the only
reason you're worried about it not being up-to-date is it's
not accurate.
MR. LAUBEN: Okay; the next slide is to show what
we consider to be a reasonable perception of a road map. It
identifies four overall parts, and these sections are not
too unfamiliar.
DR. WALLIS: This is new, now, isn't it?
MR. LAUBEN: This is new.
DR. WALLIS: This is also new.
MR. LAUBEN: This is a new road map, and it's
meant to be higher level. If you will, the diagram that you
saw in the previous reg guide spends most of its time here
and a little bit of its time in one and not much of its time
in --
DR. WALLIS: So the purpose of this is to guide
the person who's sitting down --
MR. LAUBEN: Yes.
DR. WALLIS: -- to write documentation to submit
to the NRC.
MR. LAUBEN: Yes.
DR. WALLIS: To think about it first of all.
MR. LAUBEN: And to think about it in this way as
he's thinking about how he's going to do his development
work. That's correct; not just how he's going to document
it but how he's going to go through it to do the work that
he has to do to produce the product at the end. And a lot
of this was obviously informed by CSAU, and it was obviously
also informed by some of the process that we went through in
AP600 as well.
DR. WALLIS: This is helpful, because it gives a
structure to thinking; it also gives us a check list in a
way.
MR. LAUBEN: Yes; now, let me say what is -- you
can see some of the -- well, let me just go through, and let
me start by Roman numeral I, item one: identifies transient
power plant and figures of merit. I added figures of merit
because I think just because you know what plant you're
going to analyze and just because you know what transients
you're going to analyze, if you're going to do it for
different purposes or different -- you had a different
figure of merit --
DR. WALLIS: It might be uncertainty, for
instance.
MR. LAUBEN: It might be uncertainty; it might be,
if you're doing this for the PRA purpose of knowing success
criteria rather than a peak cladding temperature. So you
need to get your fingers of merit clear in the first place.
Secondly, I think number two: identify systems
component processes -- phases and fields that must be
modeled; in other words, let's not -- you know, everything
is not a correlation. You need to know -- you need to get
the basic engine right to begin with. So you need to know
what things you need to model here in your evaluation model.
And then, as a third thing, you can identify and rank
phenomena and processes.
DR. WALLIS: So, for example, two might say for
components, we're not going to model just straight pipes;
we're going to model all this and --
MR. LAUBEN: Oh, yes.
DR. WALLIS: And that will teach us -- and
therefore --
MR. LAUBEN: Pressurizers.
DR. WALLIS: And it doesn't make any sense for our
model in two to only have a straight pipe in it.
MR. LAUBEN: That's exactly right, right.
And number two is, then, knowing what your
requirements need to be, Roman numeral two. Then, you
select and or modify the system codes and other
calculational instruments and procedures that go to make up
this whole process. And the conditions that are required
for these things and their relationships; so if you need to
know how the whole thing gets put together, and that has to
be done.
And then, number two is determine the pedigree
applicability of the numerics governing equations for the
codes and calculational instruments. Then, while you're
doing this or after you start, perhaps, you go over to Roman
numeral III and ask yourself do I have an adequate
assessment base? And if -- from the assessment base that
you had, you may select certain experiments. If you don't,
then, you may need to perform experiments, and they may
indeed be, depending on the transient and these things that
you've established in number one, they may be separate
effects or integral effects tests, and then, you need to
make sure that between the database that you have to perform
standard analysis of the entire experimental database to see
how this applies.
And then, number III is, okay, appropriate
available plant transient data. This is especially true of
non-loca transients, where you may actually have some plant
transient data that you can use to assess the probit. Now,
you've got your base; you've got the evaluation model.
Then, you can go through the assessment part that we -- that
was the subject of most of the chart in the previous reg
guide.
DR. KRESS: Where in those boxes will I find the
uncertainty calculation?
MR. LAUBEN: Okay; I think what I said is right
here. Let me see; let's see. No, no, I had -- I did have
it there. Now, where is it? It says -- no, I had
uncertainty.
MR. ELTAWILA: It jumps out from your view graphs.
MR. LAUBEN: No, I had it in there. I don't see
where I use the --
DR. WALLIS: Well, you can do that when it comes
up. In several places --
MR. LAUBEN: Yes, but mostly --
DR. WALLIS: In the assessment phase as well as in
the model.
If you were to take CSAU figure one --
MR. LAUBEN: Oh, I know. I have it on this
previous paragraph, but I don't have it here, and it should
indeed be in this box here, right.
DR. WALLIS: It may be elsewhere as well.
MR. LAUBEN: Yes.
DR. WALLIS: If you took the CSAU figure one,
their road map.
MR. LAUBEN: Yes.
DR. WALLIS: And then sort of said where would it
map on here? Is it only in box four or --
MR. LAUBEN: Well, I'll tell you what: if you
look at CSAU, most of what's -- CSAU did not -- CSAU had a
Roman numeral I, II and III, and Roman numeral I, II and III
here, most of those things, would fit in Roman numeral I of
the original CSAU.
DR. WALLIS: This is in the CSAU.
MR. LAUBEN: Yes; but I mean also is developing --
documenting the code, which, you see there, was already
done. So that's --
DR. WALLIS: There are maps all over this thing.
MR. LAUBEN: What I'm saying is CSAU Roman numeral
I encompasses I, II and III here. And then, if you will,
and I've forgotten uncertainty, II and III of CSAU would be
in box four.
DR. WALLIS: I think my suggestion would be since
CSAU is such a monument, 13 man-years of work; they really
put a lot of thought into this; you ought to then tell the
reader how this relates to that.
MR. LAUBEN: Okay.
DR. WALLIS: You know, and the reader should be
able to start with this and then refer to CSAU for pieces of
it or what?
MR. LAUBEN: As a matter of fact, my druthers is
going to be that when I write this, I am going to
specifically reference specific parts of CSAU rather than
include them in the reg guide, in fact, and say this -- to
learn more about this, go see CSAU section da da da da da
da, and that would be the preferable way to do it. That
doesn't clutter up the high level structure that we want to
maintain in this document.
DR. WALLIS: I think that's great.
MR. LAUBEN: Okay.
DR. WALLIS: You've got to make the linkage.
You've got to put one in the perspective of the other.
MR. LAUBEN: Yes, yes, okay.
DR. WALLIS: I'm going through a redesign process
now.
MR. LAUBEN: Well, I'm hoping it's not too
significant.
DR. WALLIS: Well, I think it's a very
significant --
MR. LAUBEN: Very extensive is what I mean. I'm
saying I hope it's not going to be real --
MR. ELTAWILA: This is Farouk Eltawila again.
Really, I know that your enthusiasm about CSAU and things
like that, but I really think it could inhibit creative
ideas from coming in if we try to be so specific to mandate
CSAU, and that's not the intent of the reg guide. It's to
provide high level guidance about what does the staff
expect; that's all what's important. If we force CSAU on
them, I think it will be counterproductive for what the
staff is trying to accomplish.
DR. POWERS: We've encountered enough examples
where CSAU has proved difficult to understand to have it
apply that -- that's what the group says there that I think
many of the concepts behind CSAU are very valuable; some of
the purpose may have been peculiar to the issue they were
addressing at the time.
DR. WALLIS: I agree with everything that's been
said. This was the monumental study, so all of the
technical community bought into it. This was going to
change the way the codes were done in the future. I
remember all this from the time.
MR. ELTAWILA: Since then, there have been a lot
of developments all over the world.
DR. WALLIS: I respect that. I'm not saying it's
the way to do it, but you cannot -- you've got to reference
it in some way.
MR. LAUBEN: I will reference it more
specifically, and that will be fairly easy to do, I think.
DR. WALLIS: If you think some parts of it have a
really good message, say so.
MR. LAUBEN: Because I think in some places, I
literally plagiarized CSAU, so that's going to be easy to do
that.
DR. WALLIS: I'm amazed that you're doing this reg
guide presumably virtually yourself, and yet, this other one
took 13 man-years, so you have --
MR. LAUBEN: Well, of course, I didn't do -- well,
then, you're exactly right. I really didn't do it by
myself, did I?
Okay; I just wanted to say that once you get
through the assessment process in the existing reg guide,
there is an adequacy question, and in truth, the adequacy
questions need to be asked all the time. They aren't really
asked just here, but you have to ask it here in the end.
Once you've gone through this, are there any significant
inadequacies?
DR. WALLIS: It's like at the beginning: is it
adequate for the particular --
MR. LAUBEN: You're really asking where we stuck;
is it adequate? And then, if it's not, then, you go back
and you fix what's not adequate. But and, in fact, you may
find out that the plant isn't adequate, and you may even
have to go back and redo the plant. We found that out, for
instance, in AP600C. You know, you didn't have enough
capacity in your ADS4, so you had to actually go back and
change the plant.
DR. WALLIS: Adequate is the key question. You've
helped me a great deal again here. Through all of this, you
ask it; is each one of these adequate?
MR. LAUBEN: Exactly.
DR. WALLIS: You should help the reader with
what's the measure of adequacy going to be.
MR. LAUBEN: And I'm just saying I couldn't put
this question everywhere.
DR. WALLIS: See, that's the thing.
MR. LAUBEN: Okay.
DR. WALLIS: It could probably be about the
original reg guide draft; it's adequate versus acceptable.
MR. LAUBEN: I can always use help on that,
because I think that's one of the hardest things, in my
mind, at least, to define is adequacy and sufficiency and so
forth.
DR. WALLIS: The writer is going to be asking him
or herself all the time; is what I'm doing here adequate?
And you've got to have some way of measuring that or
estimating it.
MR. ELTAWILA: Norm, I think you're understating
the work that we have done for AP600. We defined what the
measure of adequacy, and we are going to use the same
standard that we --
DR. WALLIS: You're going to use AP600?
MR. LAUBEN: No, no, no, no, no.
MR. ELTAWILA: The principle; we're talking about
the principle.
MR. LAUBEN: There are some documents we're going
to try to make nonproprietary that we did, and I think we --
there was a lot of effort, as Farouk says, that was spent on
trying to find these things, and I understand some would say
that -- you and Dr. Apostolakis said --
DR. WALLIS: I think that I'd like to go back to
the purpose of all this. The purpose of all this is to come
up with a reg guide which really helps the future, so that
it's less work and better work and everything in the future.
And we're all working towards that end.
DR. KRESS: The adequacy determination in AP600
had almost -- wholly related to the predictions versus the
data.
MR. LAUBEN: Yes.
DR. KRESS: I think that's a risky way to go.
MR. LAUBEN: It is, and that's why we've included
these first few steps; I agree.
DR. KRESS: I think your adequacy cannot be that
completely. I think you've got to think of some other --
MR. LAUBEN: You've got to do steps one, two and
three, and you count there adequacy as well; that is
correct. I think we're acknowledging that.
DR. WALLIS: We all know the there are adequate
ways to fit data which have nothing to do with adequacy of
the --
MR. LAUBEN: But a lot of times, your adequacy
answer isn't good, better, best. Sometimes, your adequacy
answer is yes or no, you know, are these the right
equations? Yes or no?
DR. WALLIS: Or can it even model --
MR. LAUBEN: That's a yes or no answer.
DR. WALLIS: Right.
MR. LAUBEN: Okay; I just -- I don't know if this
is -- I counted your comments, Graham, and there were 31,
and I decided that 12 of these had to do with structure of
the guide, improper balance, lack of clear road map and
focus beginning with the overall code structure based on
fundamental equations, and I think -- I hope that what we've
shown here is that we are going -- we believe that we can
address these.
DR. WALLIS: These 31 are culled from about 100
that I had.
MR. LAUBEN: Okay; and there are 16 that I judged
to be easily accommodated, like numbering questions and
things like that, and there are three that I don't think
we're going to -- so it really looks like 28 to 3 here,
right?
[Laughter.]
DR. WALLIS: That's not too bad. Is that
adequate?
MR. LAUBEN: I don't know.
DR. WALLIS: Ninety-fifth percentile.
DR. POWERS: You get about a 90 on this. That's
pretty good. That's a good B+, isn't it?
MR. LAUBEN: Here's the three that I think we
believe that probably won't result in a change: add more
specificity about peer review, when it should be done. I
don't think that the degree of specificity would be
consistent with office letter 3A on reg guide development.
They specifically tell us not to be too specific about some
of these things, not necessarily this but --
DR. WALLIS: What you already do is make the
general statement that peer review should be used at various
stages in the process.
MR. LAUBEN: Sure.
DR. WALLIS: You're not explicit about where.
MR. LAUBEN: I may talk about some of the general
headings.
DR. WALLIS: What is the message somehow.
MR. LAUBEN: That's right; okay.
Number two, effect of changing regulations; I
don't know what to say about that. Regulations change
slowly.
DR. WALLIS: The message there was what I've said
here several times today.
MR. LAUBEN: Okay.
DR. WALLIS: That you cannot talk about adequacy
without saying what you're going to use the code for.
MR. LAUBEN: Well, okay.
DR. WALLIS: And as the regulations get
risk-informed, you're going to use the code in different
ways.
MR. LAUBEN: Well, they will. I agree with that.
I was also thinking about examples that I've had to
encounter in the last few months, and what I have
encountered is that the same questions we --
MR. ELTAWILA: Norm, if you look at box one --
MR. LAUBEN: Yes.
MR. ELTAWILA: -- one, identify transient power
plants and figure of merit.
MR. LAUBEN: Right.
MR. ELTAWILA: Figure of merit is the one that's
going to answer your question.
MR. LAUBEN: Good point. That's really what --
DR. WALLIS: Maybe you ought to put not just
figure of merit but requirements of the regulations or
something up at that level in the box.
MR. LAUBEN: Well, that implies that but okay,
sure, okay.
And then, number three, I think that there is a
very specific purpose for implementation. I think we meet
that.
DR. WALLIS: You meant something other than what I
thought you meant.
MR. LAUBEN: I think that's right, right.
Okay; so, that is it. The status and summary is
that DG10-96 on accident and transient analysis as it is
addresses the findings of Maine Yankee panel and other
review groups; timely inclusion of current ACRS comments is
the next step in the process of eventually releasing DG10-96
and SRP subchapter 15-01 for public comment. After
incorporation of ACRS comments, DG10-96 and the subchapter
and the regulatory analysis will be sent to OGC and CRGR for
review and after appropriate OGC and CRGR consent, the
documents will be released for public comment.
DR. WALLIS: Now, let's see; where are we here?
You had some comments from me; there were some comments from
the consultants and other members of the committee we can
share with you. I've shared some with NRR. The question I
have is whether it would help -- and this is a question we
had at the subcommittee meeting -- for you to have a letter
from the ACRS. I mean, you seem to be doing a good job of
modifying the document in light of comments that you
receive. Do you want a letter? Is it useful to have a
letter at this stage from the full committee, or should we
wait until you've --
MR. ELTAWILA: I think we don't want a letter at
this time. I think we would like to modify the document.
We'll send it to you concurrently with the -- we'll send it
to CRGR and OGC, and after we incorporate the public
comments, we'll come back to the ACRS, and then, at that
time, we'd like a letter.
DR. WALLIS: So you want a letter now?
MR. ELTAWILA: No, we don't want a letter now.
DR. WALLIS: You don't want a letter now?
MR. ELTAWILA: We don't want a letter.
DR. WALLIS: Because if you did get a letter, it
would have to be based on the document that we've seen, not
the document that we have a vision of today.
MR. ELTAWILA: Yes.
DR. WALLIS: So what does the committee think
about this?
DR. APOSTOLAKIS: So you are going to circulate it
for public comment with the --
DR. WALLIS: You were persuaded to bring it to the
committee. The subcommittee's feeling was that the reg
guide was not really ready for public comment. And what
we've seen today is a completely new vision of parts of the
reg guide so what --
DR. APOSTOLAKIS: But are they going for public
comment using the existing document that we have seen?
MR. LAUBEN: No.
DR. WALLIS: So you're taking our advice about not
going with the existing document.
MR. LAUBEN: That is correct.
DR. APOSTOLAKIS: But we will not have a chance to
look at the new document before it goes out.
MR. ELTAWILA: It's a different organization of
the document and adding -- just beefing it in certain areas
and reducing the level of the content of the document in
other areas. But I think the general flavor is as we
indicated here in the outline, in that diagram that Norm put
on there.
DR. WALLIS: So you will go out with a new
document which we will not see; is that right?
DR. APOSTOLAKIS: That's what they're saying.
DR. WALLIS: So the risk of not getting our input
at that stage is that if we have significant input later,
it's more difficult to make changes.
MR. LAUBEN: No, no, it's no more difficult to
make changes based on your input and the public input than
it is before we even send it out, Dr. Wallis. The process
is the same. We would treat any additional input on the
version that you see that goes out for public comment.
MR. VERMEIL: Yes, but doesn't it have to go back
out for public comment then if it's changed?
DR. WALLIS: Yes, if it's changed significantly,
it would have to go out again, wouldn't it?
MR. LAUBEN: We would make a judgment as to what
significant means, how much of a change.
DR. WALLIS: It isn't minimal; we know that.
MR. LAUBEN: No, it's above minimal.
[Laughter.]
DR. APOSTOLAKIS: As long as it doesn't touch on
adequate protection.
[Laughter.]
MR. LAUBEN: Fortunately, we're not dealing with
that here. This is not a regulation.
DR. WALLIS: but would it be useful for you, for
some members of the committee, at least, to see the next
version before you send it out or --
MR. ELTAWILA: We will send it to you at the same
time we send it to --
DR. WALLIS: Will the subcommittee see it?
MR. ELTAWILA: We really believe we don't need
another subcommittee meeting. If you insist about having a
subcommittee meeting, that's your choice.
DR. WALLIS: We are trying to be helpful, really.
MR. ELTAWILA: Yes.
DR. WALLIS: Trying to find the process which will
produce the best document is what we're trying to do.
MR. VERMEIL: It will get to you and CRGR
simultaneously, and there will be time between then and
actually publishing it for us to consider any additional
thoughts you might have, because we also have to consider
any thoughts that CRGR might have that would change the
document.
DR. WALLIS: Well, next time you bring it up here,
you will get a letter.
MR. VERMEIL: Yes.
DR. WALLIS: And if there are still some things
that we have trouble with, that will appear in the letter.
MR. VERMEIL: Yes.
DR. WALLIS: If we really like it, that will
appear in the letter.
MR. VERMEIL: Yes.
DR. SEALE: You hope.
MR. VERMEIL: I hope.
DR. APOSTOLAKIS: So the agreement is that this
letter will be written after the period of public comment?
MR. VERMEIL: Yes; that's generally how it works.
DR. SEALE: And if we want to make any comments at
the same time it goes out to help them, we can do that.
MR. VERMEIL: Okay.
DR. WALLIS: This is what the staff is proposing.
is that what the committee will agree to?
DR. POWERS: The committee will have to take a
vote on this. This is strictly advisory at this point.
Take it into discussion? My own view, right now,
it would not be unusual for the staff to go out for public
comments and come back to us with the resultant comments on
a document of this nature.
DR. KRESS: Yes, we've done it both ways.
DR. POWERS: Well, I think this one is
particularly susceptible to this because I think we're not
bending metal when we have to make changes here.
DR. SEALE: And they're on a roll. I mean, they
know where they want to go. It's not as if they're running
up against a stone wall.
DR. WALLIS: We don't want to slow that down.
DR. SEALE: Yes; yes.
DR. WALLIS: So we do get, then, the old question
about how much of our role is sort of participating with the
staff and how much of it is formal review. We may need to
sort that out a bit.
DR. POWERS: Professor Wallis, if you would like
to make a motion at this point, we can entertain this issue.
DR. WALLIS: The issue being?
DR. POWERS: The question of whether we should --
DR. WALLIS: Write a letter?
DR. POWERS: -- do this document before it goes
out to public comment or not.
[Pause.]
DR. WALLIS: I felt that the committee should make
a decision, so perhaps we should have a motion, then.
DR. POWERS: Yes.
DR. WALLIS: And so we simply have to make a
decision. I will make a motion that we should review it
before it goes out for public comment with the understanding
that I don't -- this isn't necessarily the view I favor, but
to get us going, I will make a motion that we review it
before it goes out for public comment.
DR. POWERS: Okay; I have a motion to review the
document before it goes out to public comment.
DR. SEALE: I'll second it.
DR. POWERS: All right; I have a second.
Is there any discussion of this motion?
DR. APOSTOLAKIS: What is the time frame for that?
When will we do this?
DR. WALLIS: We'll have to do it at the next
meeting of this committee.
DR. APOSTOLAKIS: You will have to be in the June
meeting?
DR. WALLIS: How fast can you move? When are you
actually intending to have the new document ready for it to
go out?
MR. ELTAWILA: We'll try to get it in the next
month or so.
DR. WALLIS: Okay; so, it would be possible for us
to review it.
MR. ELTAWILA: Yes.
DR. APOSTOLAKIS: The latest in July.
DR. WALLIS: Now, the reason that I think you
ought to consider this is that I don't want to get into a
situation where our ability to be helpful to the staff
ceases, and then, this results in something not being as
good as it might be down the road. I feel that we have made
considerable input already; I hope it's been helpful. But
it seems to have been a big change, and I'm a bit nervous
about saying now, the staff knows enough to go ahead without
some --
MR. ELTAWILA: Am I allowed to say anything during
this motion? I don't want to interfere in the process, but
what makes sense is to go to the CRGR, because you know that
the CRGR is going to look at this document very thoroughly.
So we might be able to satisfy you, and the staff has all
the intention to do the right things, but it might, in the
eyes of CRGR, constitute a backfit, and we want to be
careful about that, not to go all the way and to come here
satisfying you and to go to the CRGR and undo everything
that you're doing.
So I think it will be appropriate to send it to
CRGR; we'll send you a copy at the same time, and if CRGR
blesses it, it will go out for public comment, and during
that time, if you have any comments on it, we will be happy
to receive them and try to accommodate them as much as we
can.
DR. WALLIS: Well, maybe we could change the
motion a bit, that we would like to see it before it goes
out for public comment. We don't necessarily need to have
any input; we'd just like to decide at the time --
MR. VERMEIL: You would get it anyway, Dr. Wallis.
That's a given.
DR. WALLIS: -- to let you know whether at that
time, we feel it would be useful to see you again or whether
we should just let you go out with it.
MR. VERMEIL: You'll get it.
DR. WALLIS: We'd simply ask that we get given a
document -- can I modify my motion? We see the document,
and we decide whether or not we want to review it?
MR. VERMEIL: Absolutely, because you will get it.
DR. POWERS: I really honestly need to turn to my
parliamentarian on this. There is a mechanism for modifying
a motion. It's cleaner if it isn't. What do I do?
DR. WALLIS: You can substitute your own motion.
DR. SEALE: You have to ask me whether or not I'll
agree.
DR. WALLIS: That's all you have to do.
DR. APOSTOLAKIS: If the seconder agrees, then,
you can.
DR. SEALE: I will.
DR. POWERS: Okay; so, we have a modified motion
now, which is that we see the -- we receive a copy of the
document --
DR. WALLIS: And then, we decide whether or not it
would be useful for us to meet with the staff before it goes
out for public comment or not.
DR. POWERS: You're just asking to defer.
DR. WALLIS: Otherwise, we're buying a pig in a
poke, it seems to me.
DR. POWERS: You're not buying a pig in a poke.
DR. WALLIS: It's a little bit more risky than I'd
sort of like to be.
DR. POWERS: You're not getting a pig in the poke;
you get a chance to look at this again no matter what.
DR. WALLIS: But, you know, I'm not interested in
getting a look. I'm interested in the quality of the final
document being as good as possible. It's the only thing I'm
really interested in. And if we can intervene at some time
before it goes out for public comment in a way which will
increase the probability that it will be a really good final
document, I think we ought to have that opportunity.
DR. POWERS: It is cleaner.
I think you're just advocating a position of you
want to see this before it goes out for public comment. I
mean, that --
DR. WALLIS: We can see it, but it doesn't mean
we're going to have to bring the staff in here, yes; so
maybe that's good enough, then, if that's the understanding,
that see it doesn't mean a presentation like this,
understand.
DR. POWERS: The effect on the staff is the same,
is that they can't set a date for releasing it for public
comment until they get clearance from us. That's the
effect.
DR. WALLIS: Now, is that the effect? Is that the
effect?
DR. POWERS: Sure it is; sure that is.
I'd like to retain --
DR. SHACK: They can take the gamble that we'll
approve it and just go ahead and then worry about the
process.
DR. WALLIS: We may just write a --
DR. POWERS: That's what we've got a motion on the
floor right now to decide to do, right? Right as we speak
now.
DR. SHACK: Have we accepted the modified motion?
There was a second that changed it.
DR. POWERS: I don't think the -- the modification
is a reversal.
DR. SHACK: It's a reversal, yes.
DR. POWERS: I want to stick with the motion that
we had.
DR. WALLIS: The only reason I'm making the
motion --
DR. SHACK: All I want to know is which motion are
we discussing?
DR. POWERS: That's right.
DR. APOSTOLAKIS: That's important.
[Laughter.]
DR. POWERS: I will try to clearly articulate the
motion that we have. I have not accepted the change. The
motion is that we see it before it goes out for public
comment. Now, is there any further discussion on this
motion?
DR. APOSTOLAKIS: What does see mean?
DR. POWERS: Okay; it means you get the document
in your hands; that we would make a collective decision on
whether we wanted to bring the staff to the meeting --
DR. APOSTOLAKIS: Oh, okay.
DR. POWERS: -- or not and review it and make a
collective decision on whether we wanted to write a letter
to the Executive Director of Operations on this document or
not.
DR. APOSTOLAKIS: Okay.
DR. POWERS: Okay? Any further discussions of
this motion?
[No response.]
DR. POWERS: All those in favor of the motion to
see this before it goes out for public comment, please so
indicate.
[Show of hands.]
DR. POWERS: All those opposed and that would
prefer to have it go out for public comment and then come
back to the committee please so indicate.
[Show of hands.]
DR. POWERS: The committee votes to allow this
document to go out for public comment without being brought
to this. I certainly would appreciate if you could provide
us a copy of it when it goes out at the time.
MR. ELTAWILA: We will definitely do that.
MR. VERMEIL: We were going to do that anyway.
DR. SEALE: That's what I wanted.
DR. POWERS: Any further matters on this issue to
discuss?
[No response.]
MR. VERMEIL: So we'll do a Larkinsgram on this.
DR. POWERS: Yes, you'll get a Larkinsgram from
this that says something to the effect of the committee has
no objections to this going out to public comment and that
we'd like to see it after it comes back from public comment.
MR. ELTAWILA: Thank you very much.
DR. APOSTOLAKIS: Can we write additional comments
to the Larkinsgram?
[Laughter.]
DR. POWERS: At this point, I think I can dispense
with the transcript.
[Whereupon, the meeting was recessed, to reconvene
at 8:30 a.m., Friday, May 12, 2000.]
Page Last Reviewed/Updated Tuesday, July 12, 2016