488th Advisory Committee on Reactor Safeguards - December 5, 2001 - Afternoon Session
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
488th Meeting - Afternoon Session
Docket Number: (not applicable)
Location: Rockville, Maryland
Date: Wednesday, December 5, 2001
Work Order No.: NRC-132 Pages 1-62
NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers
1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005
(202) 234-4433. UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
+ + + + +
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
+ + + + +
DECEMBER 5, 2001
+ + + + +
+ + + + +
The Advisory Committee met at the Nuclear
Regulatory Commission, Two White Flint North, Room
T2B3, 11545 Rockville Pike, at 4:00 p.m., Dr. George
E. Apostolakis, Chairman, presiding.
GEORGE E. APOSTOLAKIS Chairman
MARIO V. BONACA Vice Chairman
F. PETER FORD Member
THOMAS S. KRESS Member-at-Large
DANA A. POWERS Member
STEPHEN L. ROSEN Member
WILLIAM J. SHACK Member
JOHN D. SIEBER Member
GRAHAM B. WALLIS Member
ACRS STAFF PRESENT:
MICHAEL T. MARKLEY
Risk-Informed 10 CFR Part 50 Pilot Program 4
DR. APOSTOLAKIS: We're back in session.
Before we start, I want to say something -- a few
words about something else. I think all the members
know that we have very good relationship with the ACRS
staff, but today it was made clear to me that some of
us are loved more than others. The staff found out
that it's Dr. Kress' birthday today, so they bought a
card and they had everybody sign, wishing him happy
birthday. So this is for you, Tom.
DR. KRESS: Yes. You know, it's tough
when you turn 55.
DR. APOSTOLAKIS: I know.
DR. POWERS: Dr. Kress, I'd like to point
out that we were going to get you a cake with candles,
but the fire protection group here said that that was
too much of a load on the building.
MR. ROSEN: And at 55, what's the
uncertainty band on that?
DR. APOSTOLAKIS: A factor of two.
DR. KRESS: I certainly appreciate this.
DR. SHACK: We called a little late, but
we still don't know too many 130-year-olds.
DR. APOSTOLAKIS: It's not normal the
DR. KRESS: I certainly appreciate this
from the staff and thank them very much. It's a very
nice card with -- I will take that home and frame it
and keep it.
DR. APOSTOLAKIS: Okay.
DR. KRESS: Thank you very much.
DR. APOSTOLAKIS: Now, back to business.
The next topic is Risk-Informed 10 CFR Part 50 Pilot
Program, Option 2. We had a Subcommittee meeting
yesterday when -- yesterday afternoon. We discussed
with the staff primarily the language of the rule and
then the treatment requirements and so on. We did not
get into the technical details of the categorization
process and other issues, but we will have another
Subcommittee meeting, it was agreed, maybe as early as
February where we'll go into details in these matters.
The staff agreed to present today to us
obviously a shorter version of their presentation
yesterday, but we also heard from NEI, Mr. Peitrangelo
and Mr. Heymer, yesterday, who unfortunately have
other commitments today and they cannot be with us.
But the staff promised to have one viewgraph where
they would identify some differences of opinion
without expressing any views since the other side is
And, Mr. Reed, you have the floor.
MR. REED: Thank you, Dr. Apostolakis. As
already mentioned, I'll be trying to go through a much
briefer version than what was delivered yesterday, for
the sake of time.
And I'll focus on the highlights from
yesterday's meeting, basically focusing in on the two
pieces of the draft rule language where I think most
of the discussion centered -- on the categorization
area, that's Paragraph C, and also in the RISC-3 area.
And then we'll also be discussing some of the early
comments or major issues, if you will, as requested by
Before we do that, though, real quickly,
a little bit of background here to get the Committee
same page. As you'll recall, SECY-99-256 provided the
rulemaking plan for Option 2 back in October of '99.
It attached an advanced notice for proposed rulemaking
that was published in March of 2000, which we got
something between 100 and 200 comments. And SECY-00-
194, which published in September of 2000, we provided
preliminary views on those comments and also some
additional thoughts on the regulatory approach.
South Texas, as the Committee's very
aware, was a technical effort that went on for I
believe it was about two years and was just issued in
early August. It was the majority of the tech effort,
if you will, from the staff for about the last year,
and it was a proof of content for Option 2. There's
been numerous stakeholder interactions.
We cite three public workshops of note.
Most recently there was one in November that I'll
speak to a little bit here in a second. We've had two
Commission briefings in September of 2000 following
issuance of SECY-00-194, and then also in conjunction
with the STP exemption approval, we also briefed the
Commission in July of this year. Just recently, last
week, I guess it was -- the 29th, I believe it was, in
the Federal Register we published the draft rule
language. And that, of course, was the major
discussion yesterday at the Subcommittee.
I will mention here a little bit the
request of the Committee. In our workshop that we
held back in November, when we discussed the draft
rule concepts, we discussed three alternative
approaches for addressing RISC-3 treatment. Those
approaches were, if you will, what I'll call almost a
pure commercial approach, Alternative 1, which would
basically have a simple statement in the rule that
says basically maintain the design basis functions for
this RISC-3 treatment, basically; Alternative 2, which
is very much like what you see in the draft rule
language, which has what I'll call minimum rule
attributes in the rule, more of a programmatic
approach but very -- a minimal amount of detail in the
rule, hopefully; and then Alternative 3, which would
be a much more detailed rule, which would, in my view,
very simply put, would be like putting the South Texas
FSAR in the rule. So there would be a lot of detail.
That was discussed during the workshop.
We got good stakeholder feedback, and we drafted the
rule that you see and what was published last week,
and it's pretty close to Alternative 2, although the
staff likes to think it's between Alternative 1 and
Alternative 2, I think you heard yesterday. NEI
believes it's about 1.95. So there's one little piece
A little more background to remind the
Committee of just the general structure and approach
here for Option 2 and how we're going from the old
safety-related, non-safety-related world into the
safety-significant, low safety-significant world.
This is the infamous four-box diagram. It shows the
RISC-1, RISC-2, RISC-3 and RISC-4 boxes. This is the
bins that the expert panel would put the SSEs into.
If you're safety-related and you're
safety-significant, as determined by this Risk-
informed categorization process, you're in Box 1,
RISC-1. If you're non-safety-related, and that
includes important safety equipment, okay, and you're
safety-significant, you're in Box 2. And if you're
safety-related, low safety-significant, you go to Box
3. And, of course, non-safety-related and low safety-
significant is Box 4.
Additionally, I show a little bit of the
requirements that would be in each box. Basically,
Boxes 1 and 2 all requirements continue to apply for
the safety-significant boxes, okay? So if they have
anything on them, and of course Box 1 has a lot of
special treatment requirements; those continue. Box
2, if there are any special treatment requirements,
those continue, and that will be a function of when
the plant was licensed and what happened during the
process, but it can have some special treatments
requirements there. Those would continue.
DR. WALLIS: Usually, when you have axes
like this, the most important things go up on the
vertical and to the right on the horizontal, and this
is not the case. The scale of deterministic increases
to the left.
MR. REED: Yes. We didn't --
DR. WALLIS: Why don't you switch them
MR. REED: I think you might have had that
comment before, actually, didn't you?
DR. WALLIS: I think I may have, but you
haven't done anything about it, have you?
MR. REED: No. I haven't addressed that
DR. APOSTOLAKIS: Well, it wasn't a
MR. REED: But, in general, that's the way
it works. The requirements stay on for Boxes 1 and 2.
A lot of effort was focused on Box 3 during the South
Texas project, and an all out effort's been focused on
that for Option 2, and it's a major area of discussion
with all the stakeholders.
And what we're trying to do here, just to
remind the Committee again, Option 2 is really only
risk informing what are called the special treatment
requirements, or those assurance requirements that are
in place throughout Part 50, 54, 52, Part 100, Part
21. They're there basically just to try to provide a
high level of assurance that in fact these SSEs
perform design basis functions. We're not changing it
to unknown requirements, okay, or to design basis
events or anything like that. That's Option 3. Just
these assurance requirements.
So one of the ground rules is for Option
2 that we have to maintain design basis, and that's
how we get stuck with this Box 3 and trying to do
something to provide a sufficient level of regulatory
assurance that we can say we're maintaining design
basis functions without in fact imposing all the
special treatment requirements. So that's kind of the
quandary we're in there.
DR. APOSTOLAKIS: Now, I think the line,
"ensure categorization assumptions," does not belong
there, because that's something you have to do anyway
for all four boxes. And it's not really a special
requirement. I mean, presumably, everything we do has
sound assumptions behind it, right?
MR. REED: Yes.
DR. APOSTOLAKIS: So I suggest that you
take that out.
MR. REED: Yes. Could have taken it out.
I mean it was a matter of judgment what I put in those
boxes. I could have -- yes.
DR. APOSTOLAKIS: I mean the other stuff
is different. Current requirements continue, I mean
that makes sense.
MR. REED: Yes. That's right.
DR. APOSTOLAKIS: So you're telling the
industry that they don't have to ensure categorization
assumptions for RISC-3? You're not saying that,
MR. REED: What I'm saying is that your
categorization assumptions have to be valid, okay?
DR. APOSTOLAKIS: Of course, but that's
MR. REED: And you have to maintain them,
DR. APOSTOLAKIS: That's understood.
Everything we produce here has good assumptions.
DR. SHACK: No, but I think it means more.
It means that the treatment is chosen to assure that
MR. REED: Yes.
DR. APOSTOLAKIS: Use a different verb.
DR. SHACK: It's not as though we're
making good assumptions or bad assumptions, that
you're also essentially looking for treatment
MR. REED: Well, it's actually --
DR. APOSTOLAKIS: Oh, you mean the
requirements that would make sure that the
categorization assumptions remain valid. You need a
better way to say that.
MR. REED: Yes. I think "treatment" is a
bad word, actually, to use, and I think that set off
an industry. What we're really saying is if you're
assuming something in the categorization process, you
ought to have a basis for that assumption, okay? And
take a look at what you're assuming. If you're
assuming a widget's going to perform --
DR. APOSTOLAKIS: So it's closer to what
MR. REED: It's exactly what you're
saying. So it's probably an unfortunate term to use
As I mentioned, just to try to hit the
highlights from the Subcommittee yesterday and not go
through the entire draft rule language, and I know
this Committee's very interested in the categorization
part, so we'll focus on the categorization piece
first, which is Paragraph C. I revised these bullets
since yesterday. These are now in my language, and I
do not make any attempt to take the actual draft rule
language and put it in here. Now, if you like, I have
the actual draft rule slides. If you want to discuss
it, we could put that up.
DR. APOSTOLAKIS: We have that too.
MR. REED: But, basically, what Paragraph
C is, first of all, it requires the use of an approved
categorization process. Appendix T, which in earlier
versions of this existed, does not exist in the draft
rule now. That was a stakeholder comment that if you
put a very detailed Appendix T in place, we're going
to get into the same sort of boxes and exemption
spaces that we have with things like Appendix R.
That's a very onerous type of situation to deal with,
and that can tie up a lot of resources, in fact, when
you get into that kind of framework. So we're
commenting them and basically using an approved
categorization process, which then requires a
submittal and a review by the staff.
You must have an expert panel or an
Integrated Decisionmaking Process to determine the
safety-significance of the function. In fact, this
entire thing really revolves around an expert panel
and giving them the sufficient information to make a
risk-informed judgment about the safety significance
of SSCs. So you must have that expert panel.
And that panel must use plant-specific
PRAs with internal events at full power, at a minimum,
okay? And in addition to that, we prefer you use PRA
as much as possible, but we're only requiring that at
full power internal events PRA. And if you're not --
you don't have everything in a PRA, then you must
evaluate the SSC function for whatever other tools you
have available -- safety significance in other modes,
in external events, shutdown, whatever, fires.
Whatever tools you have available, use that
information and give that to the expert panel.
DR. APOSTOLAKIS: Now, I got the
impression from yesterday's discussions, plus from
reading the NEI Table 4, there is more just having an
IDP. Both the industry and you would expect some
structure in the deliberations. Maybe that's
something you want to put here.
MR. REED: I think that's a good point.
I think it's also an observation that we've had to
date from watching the pilots, that when you get to
watching these expert panels and especially when they
get outside the quantitative piece, where you know --
they're very good at running through the PRA, the RAW,
the Fussell-Vesely, using the quantitative piece, but
when they get outside of that and they're in the
qualitative portion and discussion, I think that came
up yesterday, it's been an observation we've had, it
probably needs more structure. I think NEI
understands that too.
DR. APOSTOLAKIS: I can tell you, I mean
when we reviewed the South Texas application, the fact
that there was a lot of structure in the expert panel
deliberations was something that was a positive --
MR. REED: Yes. I think it makes it a
more efficient panel.
DR. APOSTOLAKIS: So you agree with that.
Let's see if you agree with the next. Must use a
plant-specific PRA with internal events at full power
and uncertainty analysis.
MR. REED: The key words were "and
DR. APOSTOLAKIS: That's right.
MR. REED: I'll have to look to Glenn or
Mike or Mark or whoever.
DR. APOSTOLAKIS: I'm just telling you
that there's going to be a lot of discussion of this
when we discuss --
MR. REED: Yes. I understand.
DR. APOSTOLAKIS: -- categorization
process. And there is evidence that if you use point
estimates which are ill-defined, you may not get the
categorization that would be the best. And as I said
yesterday, one of those papers was written by your
colleagues here. And if you use mean values, then
it's okay. You get a pretty good categorization. But
point values are -- we don't know what they are, and
the IPEs confirm that. I mean if you look at the
IPEs, some of the numbers are way out the mainstream,
so to speak. So it seems to me that that would be a
requirement at the end.
And, again, it's not difficult to do
anymore. This is not 1980 anymore. I mean there are
codes, it's done routinely with our distributions for
inputs all over the place. So I think that will be a
useful thing to put. Otherwise you're going to pay
the price or the licensee or the petitioner will pay
the price later. When you guys will start asking
questions -- why did you have this point value here
and there, what is the basis, and so on.
One of the things that most people don't
seem to appreciate is that an uncertainty analysis is
easier to defend than point estimates. It's much
easier to defend. So, anyway, that is going to come
up by at least one member of this --
MR. ROSEN: It also is very important to
the expert panel to know the bounds of uncertainty on
what it's using, because if you have the component
that comes out of the PRA at close to one of the
thresholds, Fussell-Vesely or RAW, say it's 2.1, well,
that's just more important than two, and you can't put
it in the low safety-significant category. But what
if it comes out 1.9? Now, you're tempted to put it in
the low safety-significance category, but you cannot
fail to remember in two years you're going to do an
update of the PRA, based on the actual performance of
the plant, and it's going to shift the numbers a
little bit, and you don't want to go there. You do
not want to be a plant that has a bunch of components
that move from category to category based on the
Bayesian update, because that could wreak havoc, and
you have to go back then and look at everything you've
done since the last update.
So the expert panel wants to know -- when
it gets a RAW value, it wants to know the uncertainty
on that number. It will want, typically, to place
things in a higher category. If the number's very
well known, that's one thing, but, typically, they're
going to move -- the expert panel will move things to
a higher category.
DR. APOSTOLAKIS: I think a lot of it
depends -- I mean there were at least two papers that
I'm aware of where people did the categorization, not
the same, necessarily, as you have, using point
estimates that they found someplace and then
distributions and the mean values of distributions.
And a general conclusion, although, you know, there
may be exceptions to that, is that if the point
estimate you're using is to the right of the mean,
it's higher than the median, most likely you're going
to get conservative categorizations. If it's to the
left, you will get non-conservatives in the sense that
an SSC that belongs, say, to Category 3 may end up in
4. And I think these are useful insights that will be
helpful to you.
MR. CHEOK: This is Mike Cheok.
DR. FORD: Can I ask a physical question.
Why do you have the --
PARTICIPANT: We've got a guy.
DR. FORD: Oh.
MR. CHEOK: This is Mike Cheok from the
staff. I guess we've all come prepared to talk about
uncertainties in February, in the February
Subcommittee meeting. But the one comment I have
today, though, is that the uncertainties that we are
talking about are the parameter uncertainties. I
believe that in categorization the big uncertainty
that would affect the results would be the stated
knowledge or the incompleteness uncertainties. And I
think we're trying to address that through the IDPs
and in a structured IDP process. I think we can
handle the date uncertainties in the parameters itself
but not the epistemic type uncertainties.
DR. APOSTOLAKIS: Very true. By the way,
the date uncertain is not epistemic. You're right,
you're right. But if we do that in the analysis part,
you know, take care of the parameter uncertainties,
then the burden on the panel will be less. And I
think that's what Mr. Rosen's comment was also about.
It will be less. So the more you do to help the
panel, the better off you will be.
MR. REED: Sure.
DR. APOSTOLAKIS: Dr. Ford?
DR. FORD: I have a physical-based
question: Why do you have the qualifier at full
DR. APOSTOLAKIS: Because that's the only
one they have.
DR. FORD: That's the only one.
DR. APOSTOLAKIS: That's the only one they
DR. FORD: Only one what?
DR. APOSTOLAKIS: PRA.
DR. FORD: Oh.
DR. APOSTOLAKIS: It's recognition of
DR. POWERS: Not for want of our trying.
DR. APOSTOLAKIS: Yes.
DR. FORD: The reason why I asked the
question, not being a PRA expert, is that there are
other degradation mechanisms that occur not at full
MR. REED: Yes. This process, though, is
going to be assessing the significance at all modes
and for all events. It's just that PRA at full power
is what -- we're going to require you have that.
DR. APOSTOLAKIS: You know, as the staff
stated yesterday -- I think they're right -- for the
modes for which you don't have PRA, then they're
taking you to a conservative categorization.
MR. REED: Exactly.
DR. APOSTOLAKIS: In other words, you're
paying the price.
MR. REED: Which would tend to make things
more put up in the higher boxes.
DR. APOSTOLAKIS: Unless you use point
MR. REED: Right.
MR. ROSEN: In terms of structure, we
talked about the questions also that the South Texas
MR. REED: Right.
MR. ROSEN: It provided a lot of
structure, and we'd encourage you to think about if
not having that, having something equivalent.
MR. REED: Yes, yes. I'm sure you have a
pretty good list of comments that you want to provide.
I know those are two are coming for February, and I'm
sure there's many more.
DR. APOSTOLAKIS: Sure.
MR. KELLY: This is Glenn Kelly from the
staff. Regarding the -- if you're using a non-pure
technique to analyze things at the Plant, those that
were for external events where, for example, they used
the five methodology of a seismic margins method,
these are things that are going to take you because
those methods inherently have some conservatisms in
it, theoretically. They should be giving you
However, for things like shutdown and mode
changes, where we do not have particular methods that
can be relied upon at this point to judge whether or
not something's important, where it's going to be more
engineering judgment at this point, you're not
necessarily going to end up with the same kind of
assured conservatism that you would for the external
DR. APOSTOLAKIS: But even for the
external events, when I read the NEI document, I had
a lot of questions, because I'm not sure it's fair to
say these are conservatives analyses; these are
screening analyses. So how one determines the --
MR. KELLY: The Fussell-Vesely it's not
clear, and that's correct, because it's not clear what
the Fussell-Vesely is going to tell you. And with the
five methodology --
DR. APOSTOLAKIS: Right.
MR. KELLY: -- you can't really do that.
What it will do is will potentially tell you whether
components in an area are important or not, but it
won't -- that's not the same as giving you a Fussell-
DR. APOSTOLAKIS: Well, the basic problem
that I see is that you don't have a CDF from those
MR. KELLY: That's correct.
DR. APOSTOLAKIS: And these measures
depend on the CDF. So these are the -- the sooner we
meet, I think the better off we'll all be.
DR. KRESS: Yes, George, that brings to
mind a question that's bothering me for some time. If
you look at importance measures, RAW and Fussell-
Vesely, both of them have the absolute value -- let's
say with respect to CDF --
DR. APOSTOLAKIS: Yes.
DR. KRESS: -- both of them have the
absolute value of CDF in them. That means you're
treating a plant with a low CDF differently than
you're treating a plant with a high CDF. You could be
treating them differently. And it seems to me like
that's not the right thing to do and that there might
ought to be some sort of a virtual CDF absolute that
you don't want to get close to, and you define an
importance measure with respect to a component as to
how much would contribute to getting up to that
absolute value that you're willing to accept as a kind
of importance measure for this kind of classification.
MR. ROSEN: I don't think you're right
about that, Tom. The risk achievement worth is a
ratio. So it has -- for a plant with a low CDF and a
plant with a high CDF, you're going to get a ratio.
That's what RAW is --
DR. APOSTOLAKIS: Well, that's what he's
DR. KRESS: That's what I'm objecting to.
DR. APOSTOLAKIS: He's objecting to it.
He says that --
DR. KRESS: That's exactly what I'm
DR. APOSTOLAKIS: -- you rank them, and
then you treat the Class 1 the same as in the other
MR. ROSEN: No. I'm not making myself
clear. Risk achievement worth is --
DR. APOSTOLAKIS: Both of them are ratios.
MR. ROSEN: Whether or not you're not at
South Texas, a very low overall CDF, or some much
earlier plant, they're both ratios --
DR. APOSTOLAKIS: Exactly.
MR. ROSEN: -- within themselves. So it
tends to wash out.
DR. KRESS: No. That's why I'm objecting
to it; it doesn't.
MR. ROSEN: Well, that's one I don't
DR. APOSTOLAKIS: I think maybe another
way of stating is the thresholds of 0.005 for Fussell-
Vesely and two for RAW could be different depending on
the absolute value of the CDF.
DR. KRESS: Absolutely.
MR. ROSEN: That's the part I don't agree
with. We need to work on that.
MR. CHEOK: I think the example here is
that if the baseline CDF is ten to the minus four, a
plant with a RAW value of two would get a delta CDF
increase of ten to the minus four. Whereas a baseline
plant CDF of ten to the minus six would only get an
increase of ten to the minus six. So the point there
DR. APOSTOLAKIS: That's the essence of
MR. CHEOK: And the response to that, Dr.
Kress, is that we do have a requirement in the rule
language that says that your increase in risk, i.e.,
CDF and LERF, should be small and conform to that
specified in Reg Guide 1.174.
DR. APOSTOLAKIS: Right.
DR. KRESS: Which means you have to end up
with different thresholds for what you use to
MR. CHEOK: That's correct. In essence,
those thresholds could be different for the different
plants, for different baseline CDFs.
DR. KRESS: Somehow I don't see that
concept in here.
DR. SHACK: But it's in the NEI document,
but they sort of admit that when you go through and
you don't -- you know, if in your first cut you don't
make it, you have to go back and adjust it.
DR. KRESS: It's in the NEI document,
which we haven't reviewed.
DR. APOSTOLAKIS: No, but the other side
DR. SHACK: He's right, it is capped in an
absolute sense by the 1.174 criteria.
DR. APOSTOLAKIS: But there's another side
to it. What if a licensee uses 0.005 and two, and
everything he finds is fine, the delta CDF is low and
so on? But if you change the thresholds and remove
some of the components from RISC-1 down to RISC-3, you
still get a delta CDF that's low. You don't allow
that because you have fixed the thresholds. So there
should be some flexibility there, shouldn't there?
DR. SHACK: Only if you worry about that
DR. KRESS: Well, if I'm the licensee, I
DR. APOSTOLAKIS: Well, I mean, gee, if I
were a licensee --
DR. SHACK: It's the licensee's problem --
DR. APOSTOLAKIS: No, because if you --
DR. SHACK: -- from a regulatory problem.
DR. APOSTOLAKIS: Well, the regulator must
say, "Here are some suggestions from the thresholds,
but we're open to listening to other comments." If
you put them there as an absolute -- but I think this
comes back --
DR. SHACK: They're not going to be in the
DR. APOSTOLAKIS: What?
DR. SHACK: They're not going to be in the
DR. APOSTOLAKIS: They're not going to be
in the rule. It's going to be in the guidelines.
MR. CHEOK: As a matter of fact, the top
event prevention essentially does what you would
suggest, George, in that you could pick and choose
what you want to put in RISC-1 and include that in
your success paths.
DR. APOSTOLAKIS: I hear that licensees
using some other methods for importance measures, and
they're getting many more components going down to
one. So you really want to not to ossify the methods.
But it comes down to the old argument -- just go and
find a gypsy, if she tells you what to do, the delta
CDF is all right.
DR. SHACK: It's acceptable. It may not
DR. APOSTOLAKIS: Yes, because of the
economy of the country. And I think the point that
Tom raised -- you're happy, Tim, right; we're arguing
It comes back to -- guys, there's one
DR. POWERS: If he wants to argue among
himself, he doesn't want any distractions.
DR. APOSTOLAKIS: I think if you follow
what Tom is saying with the absolute CDF, then you're
closer to the ROP. If you follow the current CDF,
you're closer to my interpretation of ROP, which will
come down. Now, as you're trying to maintain the
current levels or, as he's saying, as long as you're
below the regulatory goals, it's okay. So this point
will keep coming back, I think. Now, Tim, back to
MR. REED: Okay. Additionally, then the
rule requires the IDP to consider --
DR. APOSTOLAKIS: Cindi is smiling over
there. Why is that?
DR. KRESS: They probably had this
discussion among themselves at one time.
MR. REED: As I mentioned, it requires you
have an IDP, and then the IDP must consider all the
information I'll just briefly mention -- the PRA
results, the non-PRA information, defense in depth and
safety margins. So this is clearly a risk-informed
reg guide 1.74, if you will, type approach that we're
instituting here. If something's low, if an SSC's
low, then there has to be a justification for it to be
low safety significant.
DR. APOSTOLAKIS: Now, wasn't that an
issue that was raised yesterday, I believe, by my
colleague to the left here, supported by others, that
all this is focusing too much on CDF and LERF, and
that some components --
MR. REED: True.
DR. APOSTOLAKIS: -- SSCs are there to
prevent perhaps a higher consequence -- I mean, yes,
a higher consequence -- I mean low consequence, high
frequency and that there may be a way of having those
DR. BONACA: Yes. I mean the
classification has the consideration of frequency
consequence, and here it's just simply CDF. And
anything below core damage doesn't seem to be
MR. REED: At least my interpretation of
that was -- that's another way of addressing defense
in depth. If you look at the NEI document, they have
that table there, and it's a little bit confusing, but
it talks about the frequency of the event and then the
number of redundant diverse systems you have. For
very frequent events below consequences, we would want
to have redundancy and diversity there, in fact, to
make sure it's low. But there's other ways of doing
that. I think what you're suggesting is another way
to potentially do that.
DR. APOSTOLAKIS: Well, and also the
argument has been made in the past that the licensees
will take care of those for other reasons.
MR. REED: If they affect power operation,
DR. BONACA: My main comment yesterday was
that you're still struggling with justifying --
MR. REED: Yes.
DR. BONACA: -- defending functionality
for RISC-3 because you still believe, probably because
of that curve, that they need to be protected, those
functions. And you're not convincing me yet that
you're doing that. Conversely, if you take RISC-3
from the box and divide it then based on the risk
consequence, you could practically divide those
between those that you preserve and you can maintain
under Appendix B, and the majority will go under RISC-
MR. REED: Yes. I mean there's clearly a
set of SSCs in Box 3 which have no nexus to safety.
DR. BONACA: Absolutely.
MR. REED: And what have been termed
ornaments, if you will. And you were suggesting for
those we really don't need anything at all.
DR. BONACA: I would like to just insert
right here the reason why it's important, yesterday we
discussed what does functionality mean, and we used a
good example of 89.10, motor-operated valves. And we
concluded that for those in RISC-3 now, those MOVs
would not be stroke tested, right? They won't be
anymore under 89.10.
MR. KELLY: No, I believe -- this is Glenn
Kelly from the staff. The indication was that there
would be no requirements under RISC-3 that they
receive special treatment.
DR. BONACA: Right.
MR. KELLY: But they would be required to
retain their functionality.
DR. BONACA: And what does that mean?
MR. KELLY: And, certainly, to one extent,
I expect that inspector might ask a licensee if it
hadn't stroke tested its valve in five years why in
the world they think it's functional?
DR. BONACA: Because yesterday, Steve, you
commented that they would not be anymore under 89.10.
So, therefore, they would not be stroke tested.
MR. KELLY: That's correct, they would not
be under 89.10.
MR. ROSEN: They would not be under 89.10,
but that just means that they could be -- their
frequency would be longer. That doesn't mean they'll
never be tested.
DR. BONACA: But the reason why --
MR. ROSEN: They might be dynamically
tested now, and they might be statically tested.
DR. BONACA: Well, one of the reasons why
it did not work --
MR. ROSEN: It doesn't mean if you put
something in RISC-3 you're never going to test it.
DR. BONACA: I understand that, but let me
just say one of the reasons why they did not work
under design conditions was because the stems were
underdesigned. One of the reasons was because if you
don't stroke stress in that condition, the grease
hardens with time, and they don't work.
MR. ROSEN: Right.
DR. BONACA: So here we're leaving -- you
know, here I'm still left, as a member, very uneasy
about what this functionality means. We determined
through 89.10 that in order to demonstrate
functionality you have to stroke test them under
design conditions -- accident conditions, okay, in
addition to improving the stems. Because now we're
saying they low safety significance so we put them
there, and then we'll determine what you have to do.
And it seems to me that the very requirements we
implemented to assure performance are being removed.
That's what makes me uneasy. I mean I just don't know
at the end of the day what this demonstration of
functionality will mean.
MR. ROSEN: Well, it starts with the
answer to the question it does it matter whether it
works or not? And the answer is it's very low
DR. BONACA: Well, no, you see, because if
you use the curve I was discussing there, a certain
percent of those in RISC-3 will be still safety
significant in my book, as they were in the FSAR. The
rest would be not.
MR. ROSEN: They will be in RISC-3, if
they're safety significant.
DR. BONACA: Because you're using only
CDF, okay? And you're assuming that meeting Part 100
or not exceeding Part 100 is irrelevant. So you're
making certain assumptions that the guy on the other
side may not agree with you; in fact, they probably
DR. APOSTOLAKIS: Anyway, the comment is
that they should look at these things and possibly add
something else in addition to CDF and LERF, some
consideration. Because even defense in depth, I mean
it was said earlier that it will take care of these
things for defense in depth. Well, defense in depth
in the abstract doesn't mean anything. You have to
have something to defend, right? And, typically, when
we talk about defense in depth, we have in mind the
release or radioactivity over dame of the core. If
you say, "No, now I'm interested in something else,"
then defense in depth will have a different result.
MR. KELLY: This is Glenn Kelly --
DR. APOSTOLAKIS: I mean the concept is
the same, but --
MR. KELLY: This is Glenn Kelly from the
staff. In the pilots, this issue came up about the
defense in depth. And defense in depth, as I
understand how it's supposed to applied here is
defense in depth for both the deterministic -- each
component is supposed to have its functions defined
for -- its deterministic safety functions as well as
any functions that are attributes that are given to it
or taken credit for it under the PRA evaluation. And
then you look at to what extent, if you remove the
treatment for this equipment, whether or not you're
affecting defense in depth in particular for your
deterministic evaluation, because, in part, the PRA
order is already counting in some ways for the defense
So they are supposed to be looking at,
which is one of the reasons why something like standby
gas treatment system would probably be retained under
a defense-in-depth argument, because it doesn't affect
core damage frequency, it doesn't affect LERF. But it
is a defense-in-depth thing for your Part 100. So I
would expect that that would end up being a RISC-1
rather than a RISC-3.
MR. ROSEN: That's why the expert panel is
there. If you didn't do that, you wouldn't need an
expert panel for the components that are for model,
the model components. You'd just go "click" and you'd
get the answer, but that's not how you do it. You get
the answer and then you subject it to a review by the
DR. APOSTOLAKIS: But this is part of what
we asked for structure of the --
MR. REED: Exactly.
DR. SHACK: But even in that case, when we
looked at the questions that the panel looked at in
South Texas, it didn't address Mario's concerns,
because the questions were all aimed at preventing CDF
DR. APOSTOLAKIS: But Mr. Kelly now
MR. ROSEN: But the questions aren't the
only -- the questions that are asked are answered for
the expert panel, but they're not the only questions
the expert panel asks. It asks questions about
shutdown, it asks questions about --
DR. SHACK: I think what Mario is asking
is do you make them more explicit? Does it sort of
somehow get dragged in? Maybe, but if the criterion
are RAW and Fussell-Vesely and the questions are the
questions, then it looks like all the explicit
criteria are CDF-oriented.
DR. APOSTOLAKIS: So let's wrap it up by
saying that the deliberations of the panel, I think,
we need a lot of guidance there. We agree that's part
MR. REED: Certainly more structure.
DR. APOSTOLAKIS: We shouldn't rely on the
kindness of the panel too much.
DR. SHACK: Even if they're strange.
DR. APOSTOLAKIS: Even if they're strange.
Thank you, Will. You're the only one who appreciated
MR. REED: Additionally, the draft rule
requires -- and this is what I refer to as the bottom
line -- that the potential increase in CDF and LERF
will be small. And this is -- I think it's already
been mentioned to the Committee -- this is sort of, in
effect, maintaining the current risk profile of the
plant, this approach here, unlike some more absolute
value. So this is currently the way we're going.
You'll have to monitor the performance
condition of SSCs. It can affect the categorization
results. And if you find integrated situation or you
get -- whatever information you get back, you need to
take action and maintain the validity of that SSC,
SSCs categorization, and then to maintain the
categorization through time. So it's update the PRA
and categorization as you either change the
configuration of the plant or you obtain operational
data. So this is make it valid and keep it valid,
DR. APOSTOLAKIS: I don't understand this,
"must monitor the performance of condition of those
SSCs that can affect the categorization results."
What does that mean? How would the results be
affected by the performance?
MR. REED: I think that's an effort -- and
I might need a little help on this -- but that's an
effort to note that -- there's thousands and thousands
of assumptions, as you guys are well aware. Some are
very important, some are not important. What you
monitor, you want to monitor those that really affect
the results, the categorization results, and that's
what you want to focus your energy on and maintain.
There are assumptions in there that really don't make
any difference at all, whether there's RISC-3 or
DR. APOSTOLAKIS: But if I have a
component that has been categorized RISC-1 or even
RISC-3 and I find that its performance is below par,
how would that change the categorization? I mean I
would probably be looking somewhere else why the
performance is poor.
MR. REED: First of all, you'd make sure
that you're still in keeping with the assumptions you
made for its reliability, availability and capability.
That would be the first thing you'd do. And,
hopefully, the assumptions you're making for that
component, the data you're collecting, is in keeping
with that. If not, then you'd have to alter those
DR. APOSTOLAKIS: But would that be
subsumed then by the last bullet?
MR. REED: Exactly. Yes, they really do
DR. APOSTOLAKIS: Yes. That's why I don't
MR. REED: Yes. A lot of these actually
do overlap. I mean I break them out for bullets, but
DR. APOSTOLAKIS: As I update my PRA, if
I have lots of --
MR. REED: These three really all kind of
MR. ROSEN: It will affect the shifting
the categorization of a component. But you don't want
to do that.
MR. REED: Yes. That's actually --
MR. ROSEN: You guard against that in your
MR. REED: That's actually a very good
MR. ROSEN: Either way it creates havoc.
MR. REED: Yes.
MR. ROSEN: Operationally -- handling the
MR. REED: Yes. That's an excellent
DR. APOSTOLAKIS: I would just delete it
and make it part of the last bullet.
MR. ROSEN: And I know that because some
of ours shifted, and it's not fun to have to deal with
DR. APOSTOLAKIS: Which way?
MR. REED: If they go up, that's the
MR. ROSEN: Once they move up, that's a
MR. REED: Yes. Once you take something
out of the special treatment requirements and release
DR. APOSTOLAKIS: But why did they move,
Steve? Because of performance or --
MR. ROSEN: I don't remember now.
DR. APOSTOLAKIS: Okay.
MR. ROSEN: There were very few cases, but
there were a couple that were particularly nasty.
They had to be fixed, and we had to go back and look
at what had we done on these components since we
categorized them, et cetera, et cetera.
DR. APOSTOLAKIS: Okay.
MR. REED: Okay. The other area that we
had, I think, the most discussion on yesterday was the
RISC-3 treatment, and I've already mentioned the RISC-
1 and RISC-2 treatment, in passing, that those
requirements are basically you're maintaining all the
requirements on those SSCs. RISC-3, the focus,
though, is to have sufficient regulatory requirements
to maintain the design basis functions, basically.
So what we're doing in this portion of the
draft rule is, first, taking off the special treatment
requirements, and then we're replacing them with the
requirements you see in Paragraph 50.69(d)(2). And,
basically, that says you must have processes to
control design, procurement, installation, maintenance
inspection tests and surveillance, corrective action,
oversight and configuration. So it's not simply a
matter of saying, okay, you're going to maintain your
design basis function; it's also a matter of having
these processes in place, okay?
And if you look at the draft rule, you'll
see that under each one of these headings, we have at
least one or, in some cases, two sentences that
provide a minimal set of attributes for what we're
thinking that you need to do for each of these
processes. But the bottom line is, is that you need
to apply the pertinent programmatic requirements to
provide reasonable confidence of the capability of
RISC-3 SSCs to perform the safety-related functions
under the design basis conditions.
Now, this is -- I think the Committee had
quite a bit of discussion yesterday -- how did we
determine this was a sufficient level of regulatory
confidence and so forth? And I think this is clearly
an area where we've had a lot of discussion in South
Texas. We've had discussion on Option 2; I think
we'll continue to have that kind of discussion. And
I'll point out it's an area that I think is a major
issue area here in a couple of slides.
MR. ROSEN: The NEI wanted to discuss
those, and they characterize this as something in
between Appendix B and no Appendix B.
MR. REED: Yes, they did.
MR. SCARBROUGH: Right. This is Tom
Scarbrough with the staff. It's sort of -- if you
look at it, it's sort of on the order of what might be
imposed for station blackout equipment and ATWS
equipment. It's somewhere -- it's not the no controls
whatsoever so that you can down to what's commercial
practice kick and count, receipt inspection or tool
It's not all the way down to that level.
It's probably somewhere in between. It's something
that you might expect for under station blackout
equipment or ATWS. But in terms of the actual
details, we haven't gotten down to that level. We're
trying to say that it's not down to the complete low
level, but it's somewhere in the middle.
MR. ROSEN: Yes. I think we felt, and I
think the industry continues to feel, that if you end
up with a third program, in other words, no program,
the full program and something in between, then a lot
of the benefit of this tends to go away.
MR. SCARBROUGH: Right. Well, that's why
I think it's important to realize that what we're
talking about is probably on the order of what's
already been implemented for the station blackout and
ATWS equipment. It's equipment that's a lot of
commercial grade, but it has some controls over its
procurement, and that's on the order of what we're
talking about. There's a lot more flexibility.
I think there was a question that
yesterday came up in terms of what are some of the
differences, and we were talking about that today in
terms of some of the major differences are the
flexibility. You have these very high-level
attributes that we're talking about. There's a lot
less prescription in the QA criteria. You don't have
the 16 criteria; you have eight processes which are
very general. You don't have the very specific
control of measuring test equipment. You still will
deal with that under the general functionality
requirement, but you don't have the very specific type
of QA criteria, which you have to have very specific
procedures and documentation and recordkeeping.
There's a lot more flexibility in how you do that.
And we talked about procurement yesterday.
It's a lot more flexible in terms of getting it from
vendors. For certain procedures -- for certain
activities, such as some types of maintenance where
it's not technically difficult to do, you can do a lot
more maintenance without detailed procedures, more
general procedures, like skill a craft type of
procedures, that would be allowed.
In a specific area, such as motor-operated
valves, which we were talking about, in this case, the
way the rule is currently drafted, this draft, 55(a)
is still applied, and so they still would have ISI and
IST processes. And 55(a), in the most recent revision
of the regulations, applies an overlaying design basis
capability verification long term for motor-operated
valves. So that would be accommodated even for these
low risks. But how they would accommodate that is
through the risk-informed programs that are being
developed by ASME, the code cases and such. They
allow less margin and longer frequencies for testing
for these low risk ones. So that's how they sort of
take care of that.
DR. APOSTOLAKIS: Okay. Understand.
MR. SCARBROUGH: And so those are some
areas where that would -- they'd be able to apply this
flexibility and get a lot of gain. But they're still
covered but just less assurance than we have right now
for Appendix B.
MR. REED: Okay. Getting on to trying to
identify some of the areas where there's perhaps some
differences or concerns between staff and
stakeholders, and I'm referring to them as early
comments here. I just broke them into what I'll call
areas of concern with the draft rule language and then
some of the areas of concern with the implementation
guidance. And this is, by no means, an exhaustive
list, but it just gives you some of the more big
ticket items that at least I heard yesterday and I've
heard in recent weeks.
Certainly, I think you heard that there's
a view that the language that we have in the draft
rule in this RISC-3 treatment area that it may be too
detailed. In fact, perhaps some of that detail ought
to be moved to the Quality Assurance Program, as was
suggested yesterday. That's one area that we need to
There was also a suggestion that we should
probably use what we have available to us in terms of
regulatory vehicles today. We've spent a lot of time
in the last couple years developing design basis 50.2
guidance. We should probably use that. We have a
Commitment Management Guidance document out there, NEI
99-04, that the staff's endorsed. And 50.59's
recently been revised, and there was a suggestion that
that ought to be utilized to the maximum extent in the
framework you heard yesterday.
There was also concern that the license
amendment process that we're suggesting -- right now,
if you look in the draft rule language, it requires,
prior to approval, a submittal. It also requires that
to be a 50.90 license amendment. And so there's some
concerns of is that the appropriate way to go about
this? Can we do that without that process? That's
perhaps some burdens associated with the license
And I think, in general, there were a lot
of comments that would go to this bullet saying that,
basically, we just need more dialogue to understand
what this draft rule language means. I mean I look at
some of it, frankly, and sometimes I'm having a hard
time figuring what it means myself. So, certainly, if
we would have the statement of consideration -- if
this was a proposed rule, we'd have the statement of
consideration to be explaining that, but we don't have
that in this format. We're trying to work with
stakeholders early. We don't have that available to
us, so it's understandable that stakeholders are
looking at the words and don't know what they mean and
there's some concerns there, so we just need more
Which leads me to the next piece, which to
get more comfort in the draft rule, we probably need
to have a little bit more comfort in how you're
implementing it. And that leads to the issues in
implementation guidance. Staff has really reviewed
the categorization part of the implementation -- NEI
00-04. We've provided two rounds of comments, and
we're about to provide a third round. So we've done
a pretty good amount of review there on that part of
On the treatment end of it, unfortunately,
really they didn't have much treatment in the very,
very early versions, and now, in the most recent
version, if you look at that, it really allows you to
pretty much -- it's not really guidance; it's really
the description of typical processes. It's not
aligned to the way the draft rule is today. I think
it would have to be written more like guidance and be
aligned to our draft rule for us to really provide
constructive comments. So it remains to be seen how
we'll address the treatment portion.
In the categorization area, like I said,
we're having another round of comments. I'd just
simply point out, there's a lot more than simply the
long-term containment integrity issue, but that's an
issue that is a big one, I think. But there's many
others that have been identified and also identified
during our observation of the pilot activities. And
we'll roll all those up and get those to NEI in the
Also, I became very well aware yesterday
that this Subcommittee has some significant concerns
on the robustness of the categorization process, and
so we want to meet with the Committee as soon as we
can -- February, hopefully; if not, March -- to get a
list of all those concerns. And, frankly, I would
think at that point in time that it would be a good
idea to have NEI also present and discuss their
document. They are in fact the author of a lot of
that, and they probably would be -- it would be good
to have them here to defend their document, as a
DR. APOSTOLAKIS: Now, Mike, do we have to
wait until February?
MR. MARKLEY: It's up to you, whenever you
and the staff and NEI are ready to sit down and talk.
DR. APOSTOLAKIS: It has to be in
Subcommittee. We should not transmit any comments
MR. MARKLEY: They would be individual
member comments and have to be captioned that way, but
we could do that.
MR. REED: I mean I really see the staff
kind of in the mode of listening to the Committee's
comments. And we'd like to get those as soon as we
can to try to factor those into what we're going to do
with the categorization guidance.
MS. CARPENTER: This is Cindi Carpenter
from the staff. We're available any time that you are
to discuss that.
DR. APOSTOLAKIS: But NEI has to be
MS. CARPENTER: Right. Exactly.
MR. REED: Yes. That's another -- got to
work them into it too. We can work with Mike.
DR. APOSTOLAKIS: I would rather do it
sooner than later, because the more we wait, the more
resistance there will be to the comments, because the
MR. MARKLEY: But, George, it would seem
to me it also would be good if the time when you have
it, the staff has refined their comments more fully on
the NEI document as well, so that you're looking at
something they've got, whether it's a response back to
them, and you're dealing with the resolution of those,
as opposed to necessarily just throwing these out to
get folded in.
MR. REED: We were thinking that way too,
but I'm not sure how much we'll have refined by early
February. Hopefully, we will, but it's most important
for us to understand these concerns. So I think that
I'd rather have that happen, get the concerns as soon
as possible. And if we can have our comments refined,
we can deliver those too at the same time. Hopefully
we can get those to NEI first as part of the process.
DR. APOSTOLAKIS: Well, maybe we can
submit -- first of all, during the Subcommittee
meetings, the comments that members make are
individual; they're not Committee positions anyway.
Maybe send something to them in the form of questions
from individual members? Because I really would hate
to wait until February.
MR. KELLY: This is Glenn Kelly from the
staff. And you can correct me, Mike Cheok, if I'm
wrong, but it's my understanding that we're pretty up
to date in categorization with NEI. And we've
certainly had a lot of discussions with them there.
And they reasonably well understand where we are, and
I think we understand where they are. We're not
necessarily in the same place with treatment, so we
can probably come and discuss categorization right
away, and we'd have to wait on treatment. So even if
you could split those two up --
DR. APOSTOLAKIS: Oh, yes.
MR. KELLY: -- that's fair.
MR. MARKLEY: George, maybe the best way
to put this thing together is that if we can refine a
list of questions or something that would be the
context of what's discussed at the Subcommittee, I
think that may be the better way of doing it.
DR. APOSTOLAKIS: Sure. I think that's a
good point. So I can invite the members to submit
questions if they wish.
MR. MARKLEY: It really just becomes the
agenda for the meeting at that point.
DR. APOSTOLAKIS: Yes. And, again, things
that we've covered here, like put more structure to
the deliberations of the panel, we don't have to put
those down. That's part of the transcript now. But
there are some detailed technical questions that I
would really hate to wait until February to give them
to you. So Mr. Markley and the Committee will work
together to come up with the best way to do it.
MR. REED: Okay. Actually, I've already
said this, but I'll say it again. On the treatment
portion of the NEI guidance, as I mentioned, it's
really not written in terms of guidance, and obviously
could not have been aligned to the draft rule since
the draft rule just recently came out. So we need to
work with NEI and figure out the most sufficient way
to get that tuned up to fit with the draft rule and be
written as guidance.
And I simply point out, and I think the
Committee's well aware of it, but in the RISC-3 area,
I think the key to success of Option 2 is
understanding the details of implementation of RISC-3;
in fact, hopefully getting agreement on that
implementation. That sorts out whether in fact this
is cost beneficial or not for industry to follow it,
and it will be the key factor determining success,
ultimately, of Option 2.
MR. ROSEN: What do you define as success
of Option 2?
MR. REED: Actually, I define it as
everybody understanding -- personally, I'd like to
have both sides completely understand what's going on,
okay? We need to have regulatory assurance, okay?
That's a given. We need to maintain design basis.
And when we get to the point where we've defined the
minimum level of requirements for RISC-3 that we have
to have, okay, as low as we can, and we understand and
we can transmit to basically deliver the information
in industry, when we expect the means to implement
that, then it becomes a function of what does that
mean in terms of cost and benefit?
MR. ROSEN: Well, that's what I think is
the definition of success, is that staff is able to
define a set of requirements that does in fact
maintain design basis functionality.
MR. REED: A minimum, yes.
MR. ROSEN: And licensees see that as not
so high a barrier to entry that a fairly large
percentage of them choose to adopt Option 2. If you
set a criteria that preserves design basis
functionality and everybody agrees but the licensees
say, "Well, it costs so much, and I wouldn't bother to
change anything," then I would say that's failure.
That's a lot of work for nothing. There's no change.
MR. REED: Yes. I mean I have to start
with the prerequisite that I maintain design basis
functions, and there's a minimal set of requirements
to do that. I hope that it works out. That's cost
beneficial. If it's not cost beneficial, then
obviously it's not -- it doesn't behoove anybody to
follow this approach. This is a voluntary initiative,
and staff shouldn't be taking its resources to put
together a voluntary initiative that nobody will
follow. So we're both interested, both industry and
the staff, of what the costs and benefits are. And
the sooner we can get there, I think the better off we
And then, quickly, just one slide, George,
on where we're going from here. We're continuing our
efforts to review the NEI implementation guidance, and
in that regard, developing guidance for review of the
submittal, which would look at the categorization
process and the PRA quality issues. We're continuing
to observe pilots. We've observed two pilot plants to
date, two pilot expert panels and two more are coming.
DR. APOSTOLAKIS: Yes. Tim, I had a
question on that. Is the purpose of the pilots to
observe the IDPs or to put the whole process under
scrutiny? Because the reports that you got back from
your colleagues, which I must admit I read very
quickly, but the reports that you got back from the
two pilots were primarily dealing with the
deliberations of the panel. Will there be any effort
to see if this is how they did the categorization,
they're calculating Fussell-Vesely correctly? In
other words, the whole works, not just what questions
the panel asks.
MR. REED: Well, clearly, the expert
panel, in my mind, is the culmination of the process.
They have all the information delivered to them. For
example, the pilot I was involved with, observing Quad
Cities, Quad Cities, basically, the BWR Owners Group
gave them about a 300-plus page document that did the
categorization, okay? And we got an opportunity to
take a look at that. And then they additionally
provided data sheets and what have you to help the
panel do its job.
So, in a sense, you get to -- when you
observe the IDP, you get to see the RAW information in
addition to watching the panel perform. And so it's
a good opportunity to interface with the pilots in an
efficient manner with minimal interruption and still
get an awful lot of feedback.
DR. APOSTOLAKIS: But are you going to
question what's in the 350-page input to the panel?
MR. KELLY: I can speak about the two that
we already had. Generally, we didn't really get the
information early enough ourselves to be able to do
any kind of independent test. What we did do is, as
Tim mentioned, is to observe the third part of the
categorization process. We did not look at how they
came up with defining the functions for each
component. We did not look at how they did the
Fussell-Vesely or RAW calculations for each component.
Those, given that you have a PRA, are relatively
simple calculations to set up. We were more concerned
about that they had things such as had defined all the
important functions for the components, had considered
potential problems associated with failure of the
DR. APOSTOLAKIS: The statements in the
NEI report, though, regarding common cause failures I
must say I don't --
MR. KELLY: You have some issues. That's
one of our issues.
DR. APOSTOLAKIS: I'd like to see a
natural implementation of this and say, "This is what
we did." For RAW it says -- I believe it says that
you don't need a common cause failure.
MR. KELLY: Well, this is one of the --
DR. APOSTOLAKIS: It exists for a year,
and I'm trying to understand what that means.
MR. KELLY: Well, one of the problems
associated with common cause failure is that I don't
think in either one of the pilots that they understood
what they were doing regarding common cause failure.
DR. APOSTOLAKIS: Other than that, did you
like what they did?
MR. KELLY: Well, I mean this is one of
our comments. We recognized that they didn't really
understand that. And there were a number of things
that they didn't understand, but this is a learning
process for them also. As we had mentioned at the
Subcommittee, generally they had about six or seven
experts about various parts of the plant, one of which
was generally a PRA expert, and he was generally the
only person that had any clue about really what RAW
and Fussell-Vesely or any of the PRA stuff meant. And
that was a real limitation in how the IDPs were run.
Hopefully, in the future, that part will be -- they'll
be better educated in where the risks are, as defined
by the PRAs and things like that. We did identify
that in our comments that came out of the pilots. I
think Mike has a comment.
MR. MARKLEY: I guess, for the record, I
guess it's not that fair to say that they did not
understand common cause failures. I guess the
understanding was different from what the staff would
understand it to be. They, basically, did not account
for the common cause failures in the calculations of
importance measures. The staff position was that we
have to account for it somehow, either through the
random failure probability, such as the beta factors.
We are still in the middle of discussions with NEI on
DR. APOSTOLAKIS: I expect that. I didn't
name it that, by the way.
Don't look at me that way. Are you done,
MR. REED: Just about. Additionally,
we'll continue issuing revisions of the draft rule
language onto the web as they become available. Of
course, we're going to meet with ACRS, as I point out
here, in the February/March time frame, whatever we
work out with Mike Markley. We have to start the
regulatory analysis and of course develop the proposed
rule package, which we've started on now.
DR. APOSTOLAKIS: Good.
MR. REED: So those are the tasks that we
have to perform. Thanks.
DR. APOSTOLAKIS: Any comments, questions
from the members? Well, thank you very much. And I
must say I'm really happy to see that you're so
willing to ask questions yourselves and learn from
that what the Committee has to offer. That's very
good. That's why I don't want to wait until February
when perhaps you will start being more defensive.
So this is early in the process, and I'm
really very happy to see that you have the right
attitude. So thank you very much, appreciate it. You
responded to the Subcommittee's request very well,
even though you only had a day. So we'll meet again.
And we'll discuss with Mr. Markley what the
appropriate way would be to --
MR. ROSEN: Get comments in.
DR. APOSTOLAKIS: -- give you comments or
maybe have another Subcommittee soon or somehow
communicate with you.
DR. POWERS: If you're looking for input
on that, I really grow nervous when members start
offering comments outside of the normal processes of
MR. REED: Maybe we'll have to have a
meeting and round them up at that point.
DR. APOSTOLAKIS: Sure. That's part of
the consideration of whether we send comments. Okay.
Thank you very much. Now we have to work on our
reports. I think we start at 5:30.
(Whereupon, at 5:09 p.m., the ACRS
Advisory Committee Meeting was concluded.)
Page Last Reviewed/Updated Monday, August 15, 2016