478th Advisory Committee on Reactor Safeguards (ACRS) - December 7, 2000
128
1 UNITED STATES OF AMERICA
2 NUCLEAR REGULATORY COMMISSION
3 ***
4 478TH ADVISORY COMMITTEE ON
5 REACTOR SAFEGUARDS (ACRS)
6
7
8 U.S. Nuclear Regulatory Commission
9 11545 Rockville Pike
10 Room T-2B3
11 Rockville, Maryland
12
13 Thursday, December 7, 2000
14
15 The above-entitled meeting commenced at 8:32 a.m.,
16 pursuant to notice, the HONORABLE DR. DANA A. POWERS,
17 chairman, presiding.
18
19
20
21
22
23
24
25
. 129
1 ACRS COMITTEE:
2 DR. DAN A. POWERS, Chairman
3 DR. GEORGE APOSTOLAKIS, Vice Chairman
4 DR. THOMAS S. KRESS, ACRS Member
5 MR. JOHN D. SIEBER, ACRS Member
6 DR. GRAHAM B. WALLIS, ACRS Member
7 DR. ROBERT L. SEALE, ACRS Member
8 DR. WILLIAM J. SHACK, ACRS Member
9 DR. ROBERT E. UHRIG, ACRS Member
10 DR. MARIO V. BONACA, ACRS Member
11 DR. LEITSCH, ACRS Member
12 MR. PAUL A. BOEHNERT, ACRS Staff
13 NILS J. DIAZ, NRC Commissioner
14 DR. JOHN T. LARKINS, ACRS Executive Director
15 MR. SAM DURAISWAMY, Federal Official
16 APPEARANCES:
17 R. BARRETT, NRR
18 T. J. KIM, NRR
19 M. RUBIN, NRR
20 R. CARUSO, NRR
21 E. CARPENTER, NRR
22 F. ELTAWILA, RES
23
24
25
. 130
1 P R O C E E D I N G S
2 [8:32 a.m.]
3 CHAIRMAN POWERS: The meeting will now come to
4 order. This is the second day of the 478th meeting of the
5 Advisory Committee on Reactor Safeguards. During today's
6 meeting the committee will consider the following: South
7 Texas Project exemption request; control room habitability;
8 proposed final Regulatory Guide D.G. 1053, Calibration and
9 Dosimetry Methods for Determining Pressure Vessel Fluence.
10 Our highlight today, however, is that Commissioner
11 Diaz has agreed to come over and discuss with us some of the
12 issues concerning the Safety Research Program. I think we
13 understand that he is going to give us a brief discussion
14 and then throw it open to us to ask him questions, so it
15 should be a most interesting discussion.
16 The meeting is being conducted in accordance with
17 the provisions of the Federal Advisory Committee Act. Mr.
18 Sam Duraiswamy is the Designated Federal Official for the
19 initial portion of the meeting.
20 We have received no written comments from members
21 of the public regarding today's session.
22 We have received a request from Mr. Peter Lagus of
23 Lagus Applied Technology, Incorporated for time to make oral
24 statements regarding control room habitability.
25 A transcript of portions of the meeting is being
. 131
1 kept and it is requested that speakers use one of the
2 microphones, identify themselves and speak with sufficient
3 clarity and volume so that they can be readily heard.
4 I want to remind members that today during the
5 lunch time break we have an opportunity to meet with some of
6 the candidates for membership that are being considered by
7 Mr. Rogers' screening panel. They have been divided up into
8 groups one and two, to meet in sequence with these possible
9 candidates, and there will be some possible candidates
10 tomorrow as well.
11 DR. LARKINS: Do we have that?
12 CHAIRMAN POWERS: We should have that.
13 DR. LARKINS: If anybody needs a copy of that
14 thing, let me know.
15 [Discussion off the record.]
16 DR. WALLIS: There is an overlap in timing too. I
17 guess we see three over lunch hour. The timing overlaps on
18 the handout.
19 DR. LARKINS: There is an overlap. There are
20 three different sets of interviews. One is the screening
21 panel and then there's two separate groups for the ACRS
22 members.
23 DR. WALLIS: That's right. We are seeing Mr.
24 Ballinger and Mr. Kumar at the same time.
25 CHAIRMAN POWERS: No.
. 132
1 DR. LARKINS: No.
2 DR. WALLIS: It doesn't say Friday, it says
3 Thursday.
4 DR. LARKINS: They made a mistake on it.
5 DR. WALLIS: Okay, so there is a mistake, thank
6 you.
7 CHAIRMAN POWERS: I understand we are also getting
8 a photograph taken some time today?
9 DR. LARKINS: Yes, there is supposed to be a
10 photograph. I thought it was tomorrow.
11 CHAIRMAN POWERS: Is it tomorrow?
12 DR. LARKINS: Yes, right before the Christmas
13 shindig.
14 CHAIRMAN POWERS: Okay, so tomorrow neat and
15 pretty for a photograph of the committee.
16 This morning we are going to look at letters
17 concerning the South Texas Project exemption, control room
18 habitability, Reg Guide 1053.
19 I think we have decided that we don't need to
20 write anything on the core power uprates.
21 Welcome -- we are ready for you at any time.
22 COMMISSIONER DIAZ: I am concerned about that.
23 [Laughter.]
24 [Pause.]
25 CHAIRMAN POWERS: Commissioner, as you know, we
. 133
1 are in the throes of preparing a report to the Commission on
2 NRC Research, and so it is indeed a great pleasure to have a
3 well-known researcher such as yourself appear to us and give
4 us some insight and what I have called a pep talk on
5 preparing this research, so I will turn over the floor to
6 you.
7 COMMISSIONER DIAZ: Thank you so very much.
8 As usual, it is a pleasure to be here among those
9 whom I consider my colleagues and sometimes I do miss the
10 ability to sit down and talk shop. I was so gratified a
11 couple of days ago. I received a report from Oak Ridge
12 authored by R.V. Perez on how to use multiple detectors and
13 cross correlations and the amazing thing -- I was able to
14 read it.
15 [Laughter.]
16 COMMISSIONER DIAZ: The second thing, I understood
17 half of it.
18 [Laughter.]
19 COMMISSIONER DIAZ: Which to me was a very good
20 thing. I think that brings back very good memories.
21 I know you have a strong interest in me
22 immediately turning to some of your research areas, but I
23 really have a strong interest in your opinions on a couple
24 of other issues that you might have formed, so since I am
25 speaking first I take my issues first.
. 134
1 [Laughter.]
2 CHAIRMAN POWERS: Fair. Fair.
3 COMMISSIONER DIAZ: I think there is a suspicion
4 and probably there is mounting evidence that we are
5 embarking on regulatory reform, I think. It might be even
6 proven guilty right now in a court of law, and I kind of
7 like that.
8 You might remember, but you probably don't, that
9 in my very first speech as an NRC Commissioner I quoted
10 Cardinal Newman with something that is very, you know, to me
11 very revealing about how humans should do things, and he
12 said in the higher world it is otherwise but here below to
13 live is to change, and to be perfect is to change often.
14 We are not striving for perfection but we really
15 are giving it a try. At the same time, all of these years,
16 and it is amazing that in the little more than four years
17 that I have been here, and I think it's over three years the
18 first time I talked to you, I have been stressing two
19 things, that in the nuclear regulation to me are
20 indispensable because of the tie-ins with the social
21 concerns, the political perceptions that do play a role not
22 in what we do things but how the rest of the country works
23 and the awareness of the public on issues and that is that
24 we need to change with predictability and fairness.
25 Those are really two things that we always need to
. 135
1 have. Therefore, although we do change and we are changing
2 often we try to bring elements of predictability and
3 elements of fairness into it.
4 If I would have written a speech for today, which
5 I didn't, I would have called it something like "Forks in
6 the Road" or "Predictability and Fairness" or a combination
7 of both, like "Forks in the Road and Predictability and
8 Fairness." However, I did not write a speech, which you are
9 very lucky that I did not.
10 I want to bring two issues before I go back to
11 some of the ones that concern you, and the first issue is
12 one that has been really bothering me for some time, because
13 as you know, we have a regulatory fabric that has evolved
14 through time, and it has really areas in which there is not
15 that much consistency, areas that lack definition, areas
16 that have overlap, areas in which a rule exists but the rule
17 is essentially forgotten and the only thing that is
18 remembered is how it is applied via a series of hierarchies.
19 I am concerned about the hierarchy of regulations
20 and how in a time of change how can we conserve a hierarchy,
21 how can we look at a rule and know that 10 steps removed
22 that rule is what is taking place.
23 I think that as we grew rapidly and rules and Reg
24 Guides proliferated there is always a connection, but it is
25 connection as tight as we need it to be to be predictable
. 136
1 and to be fair is a connection really always there in a way
2 that you can follow it down.
3 Is it compatible all the way through? Is it
4 consistent with each other and with the rest of the
5 regulations, and this is an area that I believe ACRS has
6 unique expertise to look at some of their main regulations
7 and see whether the fabric as it was made was made
8 consistent.
9 I at times have been concerned and I expressed
10 this in public meetings that, you know, the hierarchy exists
11 and I will go through at least one example, that sometimes
12 it is only the lowest level of the regulation that really is
13 updated, maintained, made perfect, and used, and the rest of
14 them remains in the background, and that might be adequate
15 for a time in which there was not the level of knowledge of
16 today, there was not the technical know-how, there was no,
17 you know, information technology, but it might be becoming,
18 okay, more of a liability than an asset.
19 It is important to realize that we sometimes
20 establish hierarchies but interpret them at the lowest
21 level, and I am not sure that that is what it was intended.
22 I will give you an example. The Commission has
23 been dealing with how do we deal with radioactive materials.
24 Now that sounds very common. However, the main -- the
25 main -- law that governs what we do, when you look at it,
. 137
1 the Atomic Energy Act, doesn't have the term "radioactive
2 materials." Isn't that amazing? Byproduct materials, mill
3 tailings, source materials, special activities, but it
4 doesn't deal with radioactive materials.
5 We were in this argument of how to do and then I
6 think it's important to realize the Commission is going to
7 have to deal with the term "radioactive materials."
8 How many years after the Atomic Energy Act is it
9 important that we define what we deal with? I believe so.
10 If it is not in the Act, should we still try to ignore it as
11 we try to go and make judgments on clearance of materials or
12 what is from the public health and safety viewpoint
13 important? I don't think so. I think we need to look at
14 the hierarchy and what is missing we need to provide better
15 definition in every place that that is missing.
16 If it is not in the Act and we need it, we can
17 deal with it. If it is not in 10 CFR and it complements,
18 adds to it, we should deal with it. If there is too much
19 with it in there, we should deal with it, but we should not
20 ignore it.
21 Let me jump to an example that is very current --
22 steam generator tube failure, something that is very much
23 right now in everybody's plan. We just got an action plan
24 from the Staff. Let's look at the hierarchy of regulations.
25 Of course, there's 10 CFR, and then there's Part
. 138
1 50, and then you have in there 50.34 that calls for a series
2 of requirements, calls for Appendix A and B, very, very,
3 very clearly, and calls for then the tech specs, and then
4 there is an association with 50.55(a) codes. There is Part
5 100, and then comes Reg Guide 1.121, and then there is a
6 relationship to the ASME Code, Section 3 and 11, although I
7 understood yesterday, and I don't know this to be so --
8 there is the story of the chicken and the egg in here. I
9 thought the ASME Code had come first --
10 SPEAKER: No.
11 COMMISSIONER DIAZ: -- and then the Reg Guide, but
12 no. I just found that it is the other way around, and so
13 there is an issue in here, in the hierarchy. The hierarchy,
14 you know, first it says 50.55(a) you have the Code and then
15 you have that and then it seems like it was the other way
16 and there are conditions in there that are used and are
17 applied and I think that sometimes we need to take a moment
18 and ponder.
19 For example, as you know, I believe, continue to
20 believe and I think I will never stop believing that
21 Appendix A is central to our safety criterion. It is not
22 only general design basis but it goes into how we should
23 erect and construct tests. It goes into, in many cases,
24 operation. What's the problem with Appendix A?
25 Well, it was never completed because people keep
. 139
1 adding lower level regulations to give it specificity. The
2 problem with Appendix A is that it lacks criteria that would
3 actually make it more useful, but it is still there, and
4 then we use it whenever it comes to mind and then we ignore
5 it whenever we want to, okay, and try and go around it
6 rather than through it.
7 Regarding the steam generators, let me read you
8 General Design Criteria 14. By the way, when was the last
9 time that you guys scanned Appendix A of Part 50?
10 CHAIRMAN POWERS: Well, we have just been working
11 on steam generators so yesterday.
12 COMMISSIONER DIAZ: People come to work for me,
13 one of the first things that I do in the first week is I get
14 Appendix A and say read it, take Appendix B, read it,
15 because it sprawls in every direction and we think we can
16 fix it here, fix it there.
17 But let me, you know -- General Design Criteria 14
18 is very interesting. Reactor Coolant Pressure boundary
19 shall be designed, fabricated, erected and tested, so as to
20 have an extremely low probability -- I can see George
21 Apostolakis smiling -- extremely low probability of abnormal
22 leakage, extremely low probability of rapidly propagating
23 failure, and extremely low probability of gross rupture.
24 Now that seems to me, you know, that is a criteria
25 that is telling us something, but we didn't define what
. 140
1 extremely low probability is. We didn't define what
2 abnormal leakage is. We didn't define what a rapidly
3 propagating failure is.
4 I was making the comment yesterday that the most
5 beautiful thing about a failure, okay, when you have a
6 stress and you rupture a tube is that in the event itself
7 there are all the elements to avoid for it to keep growing.
8 Once you rupture, you know, pressure tends to equalized.
9 There are all kinds of phenomena, so you get a little
10 rupture here and this idea that is it going to keep going
11 and going and going like the little rabbit is not true.
12 In itself the failures have the means of
13 self-termination.
14 So we now look at steam generators and we focus on
15 the endpoint, and we say what do we need to do to make these
16 things better? My question is should at one time or another
17 we look at what is the foundation of that and how do we
18 establish a better -- better rules that all of these things
19 can be founded on.
20 This is not a small thing, but I think we need to
21 think outside of the box. We need to come and look at what
22 is it that this Agency and this country needs in the long
23 term, and this foundation and this hierarchy is very
24 important because the more separated is a rule from its
25 beginning and foundation, the more it tends to be divergent.
. 141
1 There is no doubt about it, because interpretations are made
2 and is it the fact that implementation then becomes such a
3 strong feedback that it changes the original intent of the
4 rule and is that what we really want? Maybe so. Maybe so.
5 Is that what we want?
6 But it should be a decision that is made with full
7 contemplation and analysis of the fact.
8 Having said that, I'm going to go into a very,
9 very similar, but not the same subject, one that I think is
10 very much in the -- right now, and that it ties into what
11 kind of a hierarchy are we going to have?
12 How do we put together, you know, risk-informed
13 regulation and deterministic regulation. And I believe that
14 as we inch forward in the construct of risk-informed
15 regulation, we will eventually reach a difficult place in
16 the road, a fork in the road, which I hope is clearly
17 marked, a road that might lead some people to mostly
18 deterministic regulations, and a road that I believe other
19 groups of licensees and stakeholders will go into mostly
20 risk-informed regulation.
21 There might be some in between, but as time will
22 pass, they will find uncomfortable to have, you know, one or
23 the other. They will find that eventually they either get
24 trained in risk-informed regulation, or they reject it.
25 So I believe that time will see that there will be
. 142
1 two main groups, not exclusively; one going one way; another
2 one going the other way.
3 This point of decisionmaking, this fork in the
4 road, should be clear of obstacles to either one of them,
5 and to have well-defined processes.
6 I believe that ACRS has role in determining, you
7 know, when we get to this fork, how should it be clear of
8 obstacles, and what well-defined processes should exist.
9 Today, there are two recurrent policy issues that
10 are in front of us and need resolution before we even reach
11 that fork: One is selectivity, and the other is the
12 application of backfit analysis to voluntary rules.
13 Selectivity, I think it is, you know, as the Staff
14 proposes, and I believe I go with the Staff that we should
15 not have selectivity of parts of a rule. We either take it
16 or don't take it.
17 I believe that we still are saying it's voluntary,
18 so you might be able to select one rule or the other at the
19 present time, but not two to the end, all right? That
20 certainly would be complicated.
21 Let me try to focus on the issue of backfit and
22 bring it back to predictability. You know, NRC voluntary
23 rules or voluntary -- rules are not subjected to the backfit
24 analysis, yet voluntary really applies only to the adoption
25 of the rule. Compliance is mandatory once you adopt it.
. 143
1 Thus, it would be possible, using this criteria,
2 to assemble throughout years and time, a compendium of
3 risk-informed regulations embodying many of the relevant
4 requirements of Part 50 that have little or no backfit
5 analysis, because adoption of each component will be
6 voluntary.
7 So you can end up with -- each one of them came
8 in, they were voluntary, somehow they essentially start
9 being put in a compendium to say these are all the voluntary
10 lists and it may be this thick, okay?
11 And all of them, you know, each one of them, when
12 voluntary, did not have -- and now you have a total set of
13 rules without the discipline of the backfit, with changes to
14 this compendium of these rules, be exempted from backfit,
15 because they are voluntary.
16 Now, you add selectivity to it, and hierarchy, and
17 you start getting my gist. The reality is that the NRC is
18 in an unchartered course at the time, with the complexity
19 due to the potential duality of regulations, could grow to
20 the point that it might be difficult to manage.
21 And some people say, well, every reactor is
22 different, every reactor, you know is -- and we deal with
23 that. But do we want to make regulations in a manner that
24 at the front end, okay, they are plant-specific?
25 Do we want to have a standard that then you really
. 144
1 go down to the plant level, you might have the plant
2 differences? And I believe that we all agree that
3 proliferation, whether it is of nuclear weapons, or of
4 nuclear regulations, or interpretation of those regulations,
5 might not be the best way we want to go.
6 The fact is that if you look at it -- if you look
7 at the new reactor oversight programs, revised reactor
8 oversight programs, you would -- you see that in a certain
9 way, it is a deliberate attempt to maintain, you know, let's
10 call it a more centralized control of what would have been
11 individual opinions, okay?
12 It actually creates a level of checks and balances
13 so that no longer is the Resident Inspector or the Senior
14 Resident Inspector alone in making judgments. He is now
15 supported by the Regions, by headquarters.
16 There is no a tie-in. There is a deliberate
17 attempt to make these things closer together, close-knit,
18 not have each one doing his own things.
19 If we end up with a set of rules that have not had
20 the discipline and then they go down and each one takes
21 their own different approach from the tech specs or
22 interpretations of Reg Guides, we might find ourselves with
23 a very difficult case in hand.
24 I strongly believe that predictability needs to
25 anchor risk-informed rulemaking. There has to be fairness
. 145
1 and equity for every risk-informed rule, as for the sum
2 total.
3 When you went with this, five years from now,
4 there has to be some common fabric that says this has
5 complied with some criteria that is acceptable, that is
6 makes -- a criteria that includes always the fact that it
7 should be inside the envelope of adequate protection.
8 It should be consistent with the other rules, but
9 it also should have some disciplined approach at what it is.
10 And in essence, we need to look at it. And I
11 don't know what to do with it, but I can assure you that I'm
12 extremely interested in how is that we're going to be able
13 to put these things together.
14 And I believe that the Commission has been facing
15 this issue, isolated, again, isolated here. We just saw
16 50.54; we just see this, we just see that.
17 And again we're patching things. We should not be
18 patching. We should be able to look at the mantle and say
19 this is what we need to do.
20 This is the effective and efficient way of doing
21 this regulatory process. If we're going to have a highway
22 that is called risk-informed regulation, it should be clear
23 when we enter into it, what are we going to do, what are the
24 licensees expected to do, and it should not be, well,
25 there's a little voluntary here, but a now there is a little
. 146
1 less in here.
2 Are we going to start adding minor requirements
3 that were not in the original rule because now we have a
4 voluntary rule, and because it's risk-informed it has a
5 little more leeway? I say that's contrary to risk-informed
6 regulation.
7 That's contrary to the way we started it. That's
8 contrary to the principles.
9 And there are times in which principles need to
10 start at the very beginning and not avoid to putting them at
11 the back end.
12 I sincerely believe that we need to look at this
13 in a special manner. I do not advocate throwing the backfit
14 in Section 51 on -- indiscriminately at risk-informed
15 regulation or at voluntary rules. But neither can I support
16 throwing the book away.
17 I believe there has to be some way in which
18 discipline be brought into this, so when this compendium
19 comes out, it is tied in, it is defendable. There are no
20 obstacles, because we might come up with the best set of
21 risk-informed regulations in the world, and have no takers.
22 CHAIRMAN POWERS: That's very likely.
23 COMMISSIONER DIAZ: Have no takers because it was
24 not a discipline, it was not an approach that people can say
25 that this is predictable, I can trust the process, I can see
. 147
1 it when it starts to what it ends.
2 And that, my dear fellow colleagues, is an issue
3 that we will be facing more and more in the days, weeks, and
4 years ahead. And it's a large issue, and it's an issue on
5 which I believe you should weigh in.
6 And that finishes my remarks. I was very good.
7 [Laughter.]
8 COMMISSIONER DIAZ: Twenty-five minutes, that
9 might be a record. Can we have that in the minutes.
10 CHAIRMAN POWERS: And give you a gold star for
11 this? I give you a gold star immediately for discussing the
12 idea that we call coherency in the regulations. And I give
13 you another gold star for what have called bringing up the
14 clean sheet approach toward risk-informed regulation.
15 Because as we have said before when we met with
16 you, you've got a tangle in the Appendix A, that you can
17 change anything you want to in the body of 10 CFR Part 50,
18 and nothing for the license changes on iota because it gets
19 nailed by the Appendix A.
20 COMMISSIONER DIAZ: The design basis can always
21 come and get you.
22 CHAIRMAN POWERS: It is a set of regulations that
23 has been built up on time like layers of sediment, but every
24 once and awhile, a gopher has gone in and mixed things
25 together.
. 148
1 I think it has always been my belief that as we go
2 through Options II and III, that really is a chance for the
3 Staff to see what kinds of problems you have trying to do a
4 patchwork, you know, change this regulation and put a little
5 risk information, and then go over and change this one.
6 You can't do it, because you start getting this
7 interleaving and pretty soon you say, no, I've got to stop
8 and go back with a clean sheet.
9 That clean sheet approach becomes even more
10 important if the Department of Energy is right and they are
11 going to bring forth a fourth generation of reactors that
12 are not based around water technology.
13 Then you have to have something that is rather
14 independent or reactor type.
15 COMMISSIONER DIAZ: Or even if they are a
16 Generation IV water technology.
17 CHAIRMAN POWERS: Even if they are water
18 technology, it's going to be such a different water
19 technology it doesn't match well with the regulations.
20 I think you have brought a new idea into this that
21 I also find very attractive. I think what you were saying
22 to us is, gee, I can use the backfit rule as the criterion
23 by which I define these new clean-sheet regulations.
24 And I hadn't heard that before, but I am attracted
25 to it, because that has all the elements that the Committee
. 149
1 has been anxious to see in defining a broader set of risk
2 metrics.
3 And so it might be very strong. I'm interested in
4 what Dr. Kress sees in this.
5 DR. KRESS: I think that's a very attractive idea,
6 because the backfit rule has built into it, that
7 defense-in-depth elements, through the fact of that backfit
8 -- the safety goal test of the backfit.
9 It has the cost/benefit, which I think is
10 mandatory in there, and then it has the acceptance criteria.
11 You have to start with a set of principles like that if
12 you're going to have a risk-informed regulation.
13 And I think that's as good a set as any to start
14 with.
15 CHAIRMAN POWERS: One of the questions that always
16 -- you know, you always come up with, is where to start.
17 And it seems to me that it's fine, what the Staff is doing
18 right now.
19 They went through 50.54 which was an easy one.
20 You really don't need risk information to do 50.54. You
21 need technical insight.
22 And they're now embarked on 50.46, which is a
23 tough one. And I see this as kind of a learning exercise
24 for them, and certainly we observe this learning going on.
25 But now what's the next step in this process? It
. 150
1 seems to me that I would have had somebody looking at a
2 clean-sheet approach, but even if you don't do that, maybe
3 the thing to do is to start with Appendix A.
4 Because the problem with Appendix A right now, I
5 think, is that it would give us the same kind of problems
6 that we had with 50.59, in that the language that was
7 written in is the language of a pervious generation in which
8 risk was measured with a decade-marked ruler, whereas now we
9 measure risk with units.
10 Two times ten to the minus third is different to
11 us than one times to the minus third, whereas 20 years ago,
12 ten to the third was not different than ten to the two.
13 And so I'm afraid that the lawyers are liable to
14 force us into changing the Appendix A because of its
15 absolutist language that's incompatible with a much refined
16 capability to calculate risk.
17 COMMISSIONER DIAZ: You're absolutely right. I
18 believe that there is some early work and some interchanges
19 that need to take place. And I think that -- well, I really
20 should not go any further.
21 [Laughter.]
22 DR. SEALE: I'll make a comment there. I remember
23 I think it was before we left the Phillips Building, I asked
24 Paul Boehnert and a couple of other people around here, who
25 was the genius that wrote Appendix A?
. 151
1 Now, you confirmed -- if you will, you voted with
2 me a few minutes ago when you pointed out that Appendix A
3 was truly a remarkable document.
4 And while we may want to update the nomenclature
5 in which some of those statements are cast, I really feel
6 very strongly that we need to endorse the idea that it does
7 have genius associated with it, for all of the reasons that
8 you mentioned here, in particular, this fourth generation
9 question and things like that.
10 So, as we do violence, perhaps, to some of the
11 words, I think we need somehow to also recognize the
12 fundamental genius that went into that set of criteria.
13 COMMISSIONER DIAZ: Let me make a statement: In
14 the last four years, I have visited many places, and
15 interacted with many outstanding world nuclear safety
16 experts.
17 And to a few of those, maybe ten, I asked the
18 question, are you familiar with our regulations? Yes.
19 And I said, which is the most important nuclear
20 safety regulation for nuclear power plants in the book? And
21 the answer always came back, Appendix A.
22 DR. SEALE: Yes.
23 COMMISSIONER DIAZ: We consider it the foundation
24 of nuclear reactor safety. And like I said, it's
25 incomplete; it was never finished; it didn't have the
. 152
1 specifics. We need to -- but it is for Generation IV or for
2 a risk-informed road, it remains a fundamental issue.
3 MR. SEALE: Maybe you ought to identify those guys
4 and give them a medal?
5 COMMISSIONER DIAZ: Maybe I should. I know you
6 had some questions on research. I'm going to leave you to
7 ponder the small issues, and I hope that by tomorrow, you
8 have them all resolved.
9 [Laughter.]
10 CHAIRMAN POWERS: We'll get right on that.
11 COMMISSIONER DIAZ: There is much we can talk
12 about this. And I will at another time. We don't have do
13 this, and I'll be happy to sit in a meeting and argue with
14 you. I think I'll find that to be fun, yes.
15 You did send me a few questions, all of them
16 penetrating questions. I don't know that I want to answer
17 all of them, but I'll give it a crack, because I'm early
18 today. I have another 15 or 20 minutes.
19 The first question you asked me is to design a
20 useful research program, it is necessary to have some view
21 of when it is that the NRC must conduct an independent
22 assessment, rather than just reviewing licensee submittals
23 and proposals.
24 Could I provide views on when the NRC should do
25 independent assessments? This is a loaded question.
. 153
1 I believe that that goes to the issue that I
2 talked about. How do we prioritize? How do we make a
3 judgment this is crucial or this is ordinary? We need to
4 have the capability to perform independent assessments on
5 any major generic issue that affects safety. I mean, you
6 know, we want one, it is steam generators. Do we need to
7 make independent assessments and do we need to have
8 expertise oin the issue of steam generator tube failure?
9 I have asked the staff fundamentally the question,
10 are we going to use a de facto zero tube failure criteria?
11 Of course, the answer is no, we are not going to use zero.
12 Are we are going to use a de facto zero release criteria?
13 We can't do that. Where is it? Well, it is going to be as
14 tight as it should, okay, so we can, you know, make sure it
15 satisfies things. And then you start looking and wondering,
16 you know, when you get to the consideration of the tube
17 boiundary and you say, oh, it has to withstand three times
18 the operating pressure, and how long is that? Where did it
19 come from, and waht materials did it apply?
20 You see, that is where independent assessments are
21 needed. We need to have the capability to do that. And we
22 need to have the capability to do that, okay, at times I
23 would say internally. We need to have the technical
24 capability to do some of those things. And I, you know,
25 applaud Chairman Meserve's initaitive in trying to further
. 154
1 develoip the personnel of the agency, because I believe that
2 we are now at a point in which demands on technical
3 expertise are higher, because people have gone now from the
4 very first level. The level is -- anything happens, don't
5 worry, NRC, I fix it. Okay. Doin't worry, I fix it. I am
6 guilty, I will fix it.
7 Right now that is no longer there, people who
8 actually consider what is in the best interests of
9 maintaining this plant operating, including coinsideration
10 of safety, because consolidation merges the regulations.
11 All of the things are coming into play. So there is a new
12 set of variables, and that new set of variables creates, you
13 know, the fact that we need to control more and more our
14 technical judgments. And that brings into case the issue of
15 independent assessments.
16 Again, if this is an issue that we want to spend
17 time, I would suggest that we sit oin a roiundtable and I
18 will be happy -- but you get my gist.
19 The next question says, let's justify some
20 research programs, at least in part, as a mechanism from
21 preserving core competencies in particular fields. Could I
22 provide views on what are the most important? You probably
23 think, but I do believe that the most important -- the most
24 important competency that the NRC should have across the
25 board is what is radioactivity from a public health and
. 155
1 safety viewpoint. How do we measure it? How do we infer
2 what it is? Because this agency, its main objective is
3 radioloigical protection. Okay.
4 And you know pumps and valves are great, but the
5 bottom line is that is what we need to know. That is the
6 girdle that has to support everything. Whenever we do
7 something is has to end in soime terms, okay, of how do we
8 protect radioloigically public health and safety. And I
9 believe this question of how low is low enough, and where it
10 is, and how does it place in this nation's set of rules and
11 regulations is a major issue. And we need to be competent
12 about it. We need to know how to measure. We need to know
13 where to place it. Okay.
14 Maybe risk is the wrong word, oaky. Where to
15 place it. Where do we place it? How do we actually provide
16 a structure in which radiological protection is always the
17 end point? And we are capable to assess it, measure it,
18 communicate it in a proper manner. Okay.
19 You want my second tier, I believe that most of
20 hte problems that we have got to look at power plants are
21 really going down to the issue of thermal-hydraulics. Do
22 you have cooling? Do you have a heat sink? Do you have
23 water? You know, it comes down to that issue.
24 Tied into that becomes the integrity of those
25 components and the foiur materials becomes, you know, an
. 156
1 issue that is significantly important. And I can go down
2 and do, okay. I'll tell you that criticality and
3 criticality manipulations are very low in my risk of
4 important things, because we know how to do it. We know
5 there are no problems with it. We know that is an area that
6 is really, you know, fundamentally well taken care of for
7 this present generation of reactors at least.
8 In between there you will find issue that come up
9 that are related to either one of these things. I think it
10 is important that we establish a hierarcy in the core
11 assessment. And you notice that I get one issue and I
12 hammer on it. That is because I am simple and, you know,
13 can't think multi-dimensional.
14 All right. Research has opportunity to join
15 international comparative research programs, you know that.
16 Often these programs are not perfectly aligned with agency
17 needs. Could I provide some on how we preserve an
18 international, but how far should we go? I think there has
19 to be really an alignment, because everything costs money.
20 Nothing is free.
21 You know, participating in a program,
22 international program on wehat are the mating habits of the
23 ugly duckling, it, you know, could be very interesting. And
24 I am not saying that we do any such things or our programs
25 have some alignment. But there might be alingment that is,
. 157
1 you know, better, that better utilizes our funding, because,
2 you know, people say, well, but, you know, we only take one
3 trip and the amount of money. It is the amount of time that
4 the staff puts into it that really becomes the issue, it is
5 the amount. And it is the diversion of our, you know,
6 capabilities into areas. And I think it is very important
7 that we have the best set that we can get.
8 You know, nothing is free. Nothing is free.
9 Everything costs. And we are at a time when we need to
10 prioritize where we invest our resources. And I don't know
11 that I know any program that we should not be in. But I
12 think that is an issue that Research and you should look at.
13 Because I am really, you know, I see them as points and
14 phrases. I really lack the insight to go, you know, into
15 it. I can get a six months leave of absence, go and do
16 that, that might be worth -- but I think it is an important
17 issue to be raised, and it is an issue that needs to be --
18 CHAIRMAN POWERS: I mean it is just, we have a
19 reality that no country really has enough money to pursue
20 all of the things that one would like to pursue. So,
21 cooperative research is very important. And even in that,
22 you have to -- we don't have enough money to participate in
23 everything.
24 COMMISSIONER DIAZ: That's right. That's right.
25 There has to be a prioritization. There has to be a
. 158
1 hierarhy. And what are the things? You know, I mean
2 somebody says right now there is an international program
3 with steam generators, probably we should jump it.
4 CHAIRMAN POWERS: There is one by the way.
5 [Laughter.]
6 COMMISSIONER DIAZ: I know there is. And there
7 are ones inside in here that are also very good.
8 The next one, the Commission is moving toward some
9 risk-informed regulatory processes, you know, where it it
10 saying there is the fork in the road in here. And we don't
11 have enough sufficient risk information, some of it is all.
12 Three years ago I went into the lion's den, okay,
13 which was a meeting of the power plant CEOs. And I told
14 them flat-out that any efforts that they could put in in
15 getting better improved state of the art risk-informed
16 processes and, you know, starting with better PRAs, will pay
17 for itself handsomely over time. And I think that many
18 people thought, well, I don't want to spend the $2 million.
19 And now they are getting bits and pieces, okay, and they are
20 finding that -- how do I deal with that insight? Hoiw do I
21 deal with it?
22 So I believe it is fundamental that we have, you
23 know, the best set that we can possibly have. You know,
24 oine appraisal of the steam generator, one, in a
25 risk-informed sense, could pay over and over for the best
. 159
1 set of risk-informed regulation. And I believe my fellow
2 Commissioners, you know, think that -- you know, and we
3 haven't voted on this, but I sense that we all believe we
4 should get better information. That, you know, we need to
5 find ways of defining this quality of PRAs. And I think as
6 we progress on this, we should make it clear that it will
7 make life simpler for everybody if we would have a better
8 set. Okay.
9 Again, people will opt for the voluntary options,
10 and I think that's fine. And I think that is needed, we
11 need to be respectful of the different, you know, licenses'
12 interests and/or actions, but if we are going to be making
13 licensing decisions, we need to have better information.
14 And I think there is movement in that area, and I think it
15 is going to get better before it gets worse.
16 There is a question on best estimate analysis for
17 thermal-hydraulic assessments. If the codes work in the
18 sense that they match limited sets of applicable data, is it
19 important that codes have firm technical foundations? I
20 think I already answered that question at a public meeting,
21 and I have been asked again. I am definitely convinced that
22 we should have solid technical foundation for everything we
23 do, because that is the only thing we cointrol. It is the
24 technical, you know, competency and the technical foundation
25 of all we do.
. 160
1 Having said that, I think there are priorities
2 which we need to place. We don't want to go into third,
3 fourth, fifth decimal place effects no matter how
4 interesting they are scientifically. We should be able to
5 scope them out and say, this is the third significant
6 figure, we know how to do it. Okay.
7 And the last question is a very interesting
8 question. I am very, very happy that somebody thought about
9 this because it deals with human performance. Human
10 performance is likely, I hope so, to remain a key aspect of
11 plant safety. They are not going to have robots there very
12 soon. And important to evaluate in the assessment of risk
13 posed by the plants.
14 Much of the human performance technical community
15 is persuaded that the safety culture of the plant has an
16 important effect on human performance. Should there be an
17 effort to include measures of safety culture in
18 probabilistic risk assessment? Right now, no, I don't think
19 so. I don't think we have the know-how to do that. I don't
20 think that, you know, safety culture to me becomes a little
21 mushy at times, okay. Somebody was talking to me yesterday,
22 we are going to talk about safety culture to this country
23 that, you know, -- we should not be talking to it about
24 safety culture, because we should be talking about simple
25 civil rights, before we talk about safety culture.
. 161
1 But the issue is, what is the role of human
2 factors? And I think, you know, I will go back to first
3 principles. What we do in PRA or, you know, we do many
4 things, we try to make, you know, estimates. Most of the
5 times we do in probabilistic space, we need to achieve, you
6 know, find a distribution. And, you know, for mechanical
7 failures, I guess a lot of people use the way of
8 distribution. Some people look at power plants, you log in.
9 But in reality, the main issue with human
10 performance issue is how can be form a distribution that is
11 appropriate, okay. To infer their importance in operations?
12 And, you know, there are some human failure distributions,
13 there are some, but the problem is that they really do not
14 come close to defining those interactions in a nuclear plant
15 setting. And the reason is we have very special
16 requirements.
17 Does mechanical failures of componets in nuclear
18 power plants follow the same, you know, distribution of
19 failures as in a fossi plant? No. Why? Well, you know
20 better than I do. It poses a different set of requirements.
21 I mean Appendix B in itself, which became, you know, a more
22 central rule than A, although it really was done just, you
23 know, and it is done to get anything that is required by
24 Appendix A, had really done very good work in preventing the
25 same type of failures that we find everywhere.
. 162
1 Is it important that we eventually achieve soime
2 measure of how human performance impacts on the use of
3 risk-informed regulation as we get closer and closer to
4 maybe taking that fork in the road? The answer is yes.
5 As long as we understand perfectly well that
6 safety has two important dimensions, and you're absolutely
7 right and I agree with you that's it's the norms and
8 attitudes of people, though I'm not sure that we should get
9 into that.
10 But the organizational part is a very important
11 part. As an Agency, we do inspect things. You have said
12 many times that a corrective action program is the
13 cornerstone of --
14 COMMISSIONER DIAZ: Absolutely.
15 DR. APOSTOLAKIS: So why not try to understand
16 those things?
17 And I realize that in many quarters, the moment
18 the Office of Research starts something, there is immediate
19 suspicion that the regulations are following in six months.
20 Another thing that I'm sure you're aware of is
21 that on issues like this, there are no other agencies in
22 this country that really care. It's not as if the NRC
23 Office of Research doesn't do it, the NSF will do, because
24 the NSF will say, no, no, this is specific to reactors, go
25 back to the NRC.
. 163
1 DOE is interested almost exclusively in Generation
2 IV reactors now. They don't care about lightwater reactors
3 or the NRC's job.
4 So there is this additional consideration. And
5 finally, I don't know whether it was an oversight or
6 deliberate or not, but you skipped the question on whether
7 the risk assessment clause should be subject to public PRA
8 review.
9 COMMISSIONER DIAZ: It was a deliberate oversight.
10 [Laughter.]
11 COMMISSIONER DIAZ: You know, you might have
12 noticed that I seldom do things deliberately, but that one,
13 I did.
14 [Laughter.]
15 DR. APOSTOLAKIS: So back to this organizational
16 thing.
17 COMMISSIONER DIAZ: Let me try to separate it.
18 Let me first take the issue of the NRC becoming intrusive
19 with the management of nuclear power plants. I'm opposed to
20 getting involved with the management of nuclear power
21 plants.
22 We don't manage nuclear power plants; we regulate
23 them. And it's to me the licensee's prerogative to deal
24 with the management, okay? That's one issue.
25 Now, the issue of how do they, even if there are
. 164
1 studies that we might make, how do they establish serious
2 processes to get better know-how into how the human factors
3 allow them to operate the plant safer and maybe more
4 economically?
5 I think that is -- there is an obvious need to
6 have that set of information. But is it us that should do
7 it? I really doubt that we have the resources and the
8 inclination to do that.
9 I do believe that we need to continue with our
10 human performance programs, but they apply to management
11 structures. I really have serious problems with that.
12 And I see the need for the know-how to be there.
13 I completely agree with you that they need to know how it
14 percolates into the different parts of their risk
15 assessment, plant operations, and it's probably overdue by
16 now.
17 Now that we're at this stage that people get
18 additional knowledge, but for the NRC to get involved in the
19 -- and the potential pitfalls in getting involved in
20 management, I'm opposed to it.
21 And I have already overstayed my welcome by five
22 minutes. But it's your fault.
23 [Laughter.]
24 CHAIRMAN POWERS: You never overstay your welcome,
25 Commissioner.
. 165
1 COMMISSIONER DIAZ: Gentlemen, thank you so very
2 much. It's always a pleasure to be here, and I look forward
3 to doing it again.
4 CHAIRMAN POWERS: Thank you, sir; we very much
5 appreciate it.
6 I'll take a break for 15 minutes.
7 [Recess.]
8 CHAIRMAN POWERS: Let's come back into session.
9 The topic for the rest of this morning is the South Texas
10 Project Exemption Request. Jack, I think you're going to
11 lead us through this?
12 MR. SIEBER: Yes. If you look in your binders at
13 Tab 9, you will find about 280 pages of introductory
14 material. And on your tables, you probably have received
15 the response to the request for information that was sent in
16 by the South Texas Project for the exemption request that
17 they have earlier made.
18 We have actually been considering and the Staff
19 has been considering, this prototype approach for Option II
20 for the South Texas Project for several years now.
21 And it would appear as though we're getting down
22 to rug-cutting time, so to speak, and, in fact, the
23 scheduled would indicate that sometime in April or at least
24 early Spring, this process should come to some kind of
25 closure.
. 166
1 Today we're going to talk about exemption
2 requests, which follows from the classification of
3 structures, systems, and components into various categories
4 indicating their safety significance and their risk
5 significance.
6 There were seven exemption requests for basically
7 non-risk and non-safety-significant components and
8 structures and systems. And as an example, the exemption
9 requests seek exemption from qualification requirements,
10 maintenance rule, Appendix A, GDC-2 and 4, QA requirements
11 in Appendix B, inspection and testing, which is GDC-18, ASME
12 Section 11, and 10 CFR 50.59, and use of commercial grade
13 components to replace safety-grade components as part of the
14 repair and replacement program.
15 So, this is actually a pretty sweeping set of
16 exemptions to implement this program, and I'm sure that
17 South Texas believes that they need these exemption requests
18 in order to make it worthwhile to adopt a risk-informed
19 classification system.
20 So with that introduction, what I'd like to do is
21 introduce the folks from the South Texas Project, and you
22 can introduce each other, and have you begin your
23 presentations.
24 MR. McBURNETT: Thank you and thank you for the
25 opportunity today to come and discuss this with you. I'm
. 167
1 Mark McBurnett and I'm the Director for Quality and
2 Licensing at the South Texas Project.
3 On my left is Glenn Schnizel, and he's the Project
4 Manager for our QA project, which includes this Option II
5 effort. And on his left is Rick Grantom, the Manager of our
6 organization that's responsible for the risk assessment
7 activities at South Texas.
8 I'll start off with -- we have a draft Safety
9 Evaluation Report in hand, and I think it represents a
10 significant piece of work on the part of the staff. I think
11 it does a very good job of bringing us in and getting to the
12 real issues that stand between us and completing and getting
13 the exemptions granted.
14 It's not been an easy task. Certainly the rules
15 are complex and intertwined and have deep roots. We still
16 have a lot of work to do, as the draft Safety Evaluation
17 Report talks about, but there are -- we have ongoing
18 discussions with the Staff.
19 One of the things you will see as we go through
20 the slide package today is that the slides were prepared
21 about two or three days ago, and submitted and then, in
22 turn, we met with the Staff yesterday. So a number of the
23 issues, we have either resolved or made some progress on
24 yesterday.
25 Glenn will do most of the presentation today, and
. 168
1 as he talks through that, he'll hit a couple of them, and
2 he'll say that we've made significant progress on them, and
3 these others, we're focusing some attention.
4 Our effort today is going to go through and try
5 and discuss just some of the key issues that we see between
6 us and getting the exemptions granted. Clearly, we're
7 focused on achieving success here, and, of course, success
8 means our goals have to be satisfied, the NRC's goals have
9 to be satisfied, and there are some compromises there as to
10 what makes that all work, but it's got to work for all of us
11 together.
12 I'm saying that we're focusing on a process that
13 categorizes components in the station as
14 non-risk-significant, low-risk-significant, medium
15 risk-significant, or high-risk-significant, independent of
16 whether or not they are safety-related.
17 And from there, we're going into to saying that
18 those are in that low- and non-risk-significant category,
19 should be able to be treated with our normal commercial
20 practices in lieu of the special treatment requirements in
21 the regulation, particularly the special treatment
22 requirements for seismic, equipment qualification, Class I-A
23 and the ASME components.
24 With that short introduction, I'm going to let
25 Glenn go ahead and get into our presentation.
. 169
1 MR. SCHNIZEL: Thank you, Mark. And again, on
2 behalf of the South Texas Project. We welcome the
3 opportunity to come in front of the Committee and present
4 the information on our draft Safety Evaluation Report, and a
5 little bit of what we're doing with risk insights.
6 The South Texas Project, as has been stated, has
7 been involved with risk insights for a number of years. One
8 of the first major areas that was process-oriented involving
9 risk insights was our graded quality assurance approach.
10 We received that safety evaluation report in
11 November of 1997. The intent of the graded quality
12 assurance approach was to grade the controls that we have
13 over some of our safety-related components.
14 As we got into the implementation activities, it
15 was soon recognized that a lot of the regulations wove
16 themselves very tightly into those controls over those
17 components.
18 And what we expected to gain from our graded
19 quality assurance program wasn't necessarily going to be
20 achievable. What we did see is that there are four primary
21 areas that were causing us a challenge.
22 And those were in the area of ASME, seismic,
23 equipment qualifications, and Class I-E. The main gain that
24 we expected to receive from our graded quality assurance
25 approach was going to be with procurement of replacement
. 170
1 components.
2 And those were areas that were challenging that
3 ability to procure replacement commercial components. Once
4 we recognized that, we started into some considerable
5 dialogue with the Staff.
6 The culminated in us submitting our exemption
7 request. That was to exclude certain components from the
8 scope of the special treatment requirements required by
9 regulation. That was submitted in July of last year, 1999,
10 and that approach did represent the prototypical Option II
11 pilot as called in SECY 98-300 for risk-informing Part 100.
12 DR. APOSTOLAKIS: Can you tell us -- I mean, I
13 understand the Class I-E, seismic and so on, but by ASME,
14 what do you mean?
15 MR. SCHNIZEL: What we were looking at there is
16 replacement of ASME components with non-ASME components. We
17 were looking at if we had qualified ASME components that
18 required replacement, they failed, they were degraded, that
19 we could replace those with commercial, non-ASME components.
20 That was one of our premises with graded quality
21 assurance.
22 MR. SIEBER: I think another aspect of that is
23 that I read your submittal, was exemption from the
24 requirements of Section 11.
25 MR. SCHNIZEL: That is correct.
. 171
1 DR. APOSTOLAKIS: Okay.
2 MR. SCHNIZEL: Now, as far as 98-300, just to
3 refresh your memories, one of the main areas under 98-300
4 that we used as a foundational statement to build from, was
5 the following: Under Option II, SSA, structures, systems,
6 and components of low safety significance would move from
7 special treatment to normal commercial treatment, but would
8 remain in the plant and be expected to perform their design
9 function but without additional margin, assurance, or
10 documentation.
11 So that was the basic foundational statement that
12 we began with when we started this Option II approach.
13 Once our exemption request was sent to the Staff,
14 we did receive some requests for additional information in
15 January of this year. Those were responded to, and we had a
16 lot of interaction with the Staff, both face-to-face
17 meetings and phone calls, to try to work through some of
18 those areas.
19 We did submit a revised exemption in August of
20 this year, and as reported, just within the past three
21 weeks, received a draft Safety Evaluation Report. That
22 draft Safety Evaluation Report has preliminarily been
23 reviewed and we're currently actively engaged in discussions
24 with the Staff to work through some of the issues.
25 Some of the key targets that are coming up for the
. 172
1 short term: South Texas is targeted to respond to the draft
2 Safety Evaluation Report on January 15th of next year, with
3 the target of April 14 for grant of the final SER.
4 MR. SIEBER: Let me ask a question: Could you
5 briefly explain what commercial treatment is, particularly
6 with regard to maintenance?
7 MR. SCHNIZEL: From a commercial treatment
8 standpoint, it's essentially how we maintain what we call
9 our balance of plant. We use what we consider good business
10 practices of documenting deficiencies, providing appropriate
11 instructions to our maintenance craftsmen in order to
12 perform maintenance.
13 And then we do post-maintenance tests as required
14 in order to ensure that those components have been
15 satisfactorily restored to their desired function.
16 We do have preventative maintenance activities
17 that we do on those components. So we do use our corrective
18 action program readily on our balance of plant side. So
19 that's part of what we use for our commercial practices.
20 MR. SIEBER: But that's pretty much under the
21 total purview of individual licensees. As an example, some
22 people keep their plant in tip-top shape; other people look
23 at the budget and say I'll do what I have to do until the
24 money runs out.
25 And so there's basically some external constraints
. 173
1 on how one could possibly interpret commercial treatment,
2 right?
3 MR. SCHNIZEL: Well, you do have to remember that
4 we are operating as a business also, and our business is to
5 produce electricity. And if the balance of plant side,
6 using commercial practices, is unreliable, it doesn't serve
7 the basic business premise.
8 So our underlying goal is to ensure that the
9 balance of plant is as reliable as it possibly can be, to
10 ensure that we can produce that power.
11 MR. SIEBER: Thank you.
12 MR. SCHNIZEL: Continuing on, under SECY 98-300,
13 again, what we would see from Option II is that we would
14 adjust the scope of structures, systems, and components to
15 which this special treatment would apply.
16 Option II does not change regulations. Changing
17 regulations would be considered under the Option III
18 approach. Under Option II, low safety-safety-significant
19 components would remain in the plant, there's no plan of
20 going out and summarily removing low-safety and
21 non-risk-significant components, just because of their
22 safety significance.
23 We would move from a special treatment to normal
24 commercial treatment. These low safety components would be
25 expected to perform their design functions, but without that
. 174
1 additional margin, assurance, or documentation that's
2 associated with the high-safety-significant components.
3 And one key point is that safety-related
4 components would remain safety-related. Option II is not
5 looking at reclassifying the components from safety-related
6 to non-safety-related.
7 As STP began into the Option II approach, there
8 were some foundational beliefs that we --
9 DR. SHACK: What does that mean? I mean, if
10 you're no longer going to give them safety -- you know,
11 special treatment, is it just a name that's left?
12 MR. SCHNIZEL: We're beginning to operate at
13 essentially a higher plateau where now you're dealing with
14 safety significance and non-safety significance, as opposed
15 to safety-related and non-safety-related, and making all of
16 your decisions deterministically. And now you're applying
17 that risk-informed insight into how you make your decisions.
18 MR. McBURNETT: Another part of the answer to that
19 question is what the definition of safety-related does, it's
20 what makes the tie between all the general design criteria,
21 the design basis and analysis of the design basis of the
22 plant, which we're not changing.
23 Safety-related leads to that, and we're trying to
24 -- so we don't want to change that, so that's why we're not
25 going after declassifying things. We started down that path
. 175
1 and found out that that's not where we want to go to yet.
2 That's Option III.
3 MR. SCHNIZEL: Some of those foundational
4 principles or beliefs of South Texas began under the Option
5 II approach, was that the emphasis would be placed on proper
6 categorization of the components. If we had a robust
7 categorization process, that would properly bucket these
8 components into the right risk box, Risk 1, 2, 3, or 4.
9 And then the treatment that you would apply
10 against those would ensure that those components are going
11 to properly perform. So really the categorization is the
12 cornerstone or the key to this entire proces.
13 DR. APOSTOLAKIS: Well, on that front, though, I
14 looked at your response dated August 31st to the NRC Staff's
15 requests for additional information. And in Attachment 4,
16 your respond to Question 20(a), which asks about how you
17 handle common cause failures.
18 And basically you're saying that you're using a
19 commercially-available computer code, and then you proceed
20 to describe how the two importance measures of Fussel Vesely
21 and risk achievement worth are calculated, and I must say
22 this is all new to me.
23 And it looks completely arbitrary. The equations
24 are given without any basis. Statements are made that are
25 wrong.
. 176
1 I suspect you are not aware of it, because you use
2 the code in good faith. And I really don't know where that
3 leaves us. I mean, you are obviously pursuing a robust
4 categorization, but I must say that my confidence in how
5 robust it is, based on reading pages 83, 84, and 85, is
6 shaken.
7 MR. GRANTOM: Let me add some clarification on
8 that. What you're seeing in that question is how you
9 correlate the contribution of a component, Component X, to
10 the total common cause contribution.
11 So, we're calculating common cause correctly; the
12 computer code we use is calculating all the proper
13 combinations, including the global common cause, the
14 doublets, all those types of things are being calculated
15 properly.
16 And our code has been software-QA'd; it's been
17 backed up by the vendor, so the common cause is being
18 calculated correctly.
19 Now the question comes up as to how do I determine
20 Component X's contribution? Originally, we counted the
21 entire component -- every time it showed up, we counted the
22 entire component's contribution again. So therefore in
23 doing that, we were double-counting, triple-counting the
24 components.
25 Both us and the Staff agree that we were extremely
. 177
1 conservative in applying the common cause contribution to a
2 component in a very conservative manner. Now, what we
3 determined when we realized we were very conservative in
4 that regard is, if you have a three-component common cause
5 group, then how do you determine, out of all of the common
6 cause combinations, what a particular component's
7 contribution is to that or how you do the bookkeeping to
8 extract all the common cause terms and relate it back to a
9 particular Component X.
10 We elected to do that by dividing the total common
11 cause contribution by the number of components, so that a
12 third of it went to Component X for all the common cause
13 terms and likewise for the entire common cause group.
14 One gets the impression in looking at that that we
15 are adding risk achievement worth -- things of that nature,
16 but what we are really trying to do is we are trying to show
17 that common cause contribution in a system in common cause
18 group and we are trying to correlate the total common cause
19 contribution to a component.
20 Now we have had conversations with the Staff about
21 this and the Staff has given us their impression of how it
22 should be calculated, which we agree. Our computer code
23 does exactly the way that the Staff has indicated.
24 The problem is how do you correlation all of those
25 common cause doublets, triplets back to a specific component
. 178
1 and that is the way we have elected to do it, right?
2 DR. APOSTOLAKIS: Yes, and what I am saying is
3 that there doesn't seem to be any basis for this.
4 For example, it says that RAW is calculated as 1
5 minus the sum of RAW (i) minus 1 and I don't know why that
6 is.
7 I mean the definition of RAW is that you calculate
8 the risk metric assuming the component is down. It is a
9 very simple definition. There is no reason to take
10 one-third of the common cause failure probability and all
11 that. You just put the component down and calculate a new
12 risk.
13 The Fussell-Vesely it's the same thing and I mean
14 I don't see there is any theoretical foundation to this.
15 I understand that there was a problem before,
16 being overly conservative, but that does not justify doing
17 what is here. I mean somehow we have got to make sure that
18 when calculations are made they have some theoretical
19 foundation.
20 I am not talking about abstract theories here. I
21 mean there is a clear definition. The RAW is the Ratio of
22 the, say, CDF, assuming that component is down everywhere,
23 wherever it appears. It is down, divided by the nominal
24 risk. Why is that so difficult to do?
25 MR. GRANTOM: That is what we are doing. I mean
. 179
1 the calculation of the computer code is calculated. When
2 you are looking at common cause, what is the RAW of a common
3 cause term? Well, the RAW of a common cause global term is
4 basically of system-level importance. It fails all three
5 pumps in a system, well, it fails the entire system, so when
6 we are looking at global common causes you are really
7 looking at system-level importances as opposed to component
8 level.
9 When you are looking at an independent failure
10 versus a common cause failure of two other components, a
11 single and a doublet, and all combinations of that, you are
12 still looking at failure of the entire system and so you are
13 trying to determine what is the common cause component of a
14 particular component in trying to include that additional
15 worth and not forget the common cause but somehow to include
16 it in the contribution of the component.
17 This work, as far as I am aware, has not been done
18 before, and we are the first to step into this to try to
19 correlate those.
20 Now if you look at Reg Guide 1.174, Reg Guide
21 1.174 would tell you you don't include common cause at a
22 component level to determine component level importance.
23 You just do independent failures and you use sensitivity
24 studies on common cause to determine system level
25 importances. That is all they would have you do.
. 180
1 Now we have tried to take this a step further and
2 pull common cause into the importance of a component and
3 include both of those contributions.
4 There is the debate of whether we were too
5 conservative and in the next set of equations whether we're
6 nonconservative. One of the recommendations we have on the
7 table with the Staff is well, maybe we should go back and do
8 it the way Reg Guide 1.174 says to do it and just use common
9 cause as the sensitivity studies and not include it in the
10 importance calculation.
11 DR. APOSTOLAKIS: Well, I think we are getting
12 into too much mathematical detail, which may not be
13 appropriate for a meeting of this type, but I really have, I
14 remain to be convinced that what is being done on these
15 three pages makes sense.
16 DR. SCHINZEL: We will continue to work with the
17 Staff to seek resolution that we both agree is a proper
18 approach and try to ensure that we can communicate that to
19 show the basis of what we're using.
20 DR. APOSTOLAKIS: I raised the concern, not the
21 Staff, so by continuing to work with the Staff I don't know
22 how much that helps me.
23 I need help, Mr. Chairman. Right now I cannot
24 support a letter that says this makes sense. In fact, I
25 would be very negative.
. 181
1 I believe what is in these three pages just
2 doesn't make sense, even though the intent that Mr. Grantom
3 mentioned is well understood and I agree that there is a
4 need to do what he just said. I don't think that what is
5 being done responds to that.
6 CHAIRMAN POWERS: Okay.
7 DR. APOSTOLAKIS: And I don't think this is the
8 appropriate forum to get into details why did you subtract
9 one or do this and that, so maybe we should go on and --
10 CHAIRMAN POWERS: Yes, let's go ahead.
11 DR. SCHINZEL: Okay. I will continue.
12 Again some of the basic premises or beliefs of
13 South Texas had for Option 2 going into this, repeating
14 again emphasis placed on the proper categorization -- we
15 consider that to be the most important of these.
16 Again we believe that current commercial practices
17 that are in effect on our balance-of-plant side, these are
18 sufficient for the low safety and non-risk significant
19 safety-related components.
20 We do have safety-significant components on our
21 balance of plant side. These commercial practices have been
22 effective in ensuring that these components are going to be
23 reliable, and we feel that these same practices will
24 certainly be sufficient for our least important
25 safety-related components.
. 182
1 The details of the commercial practices, as far as
2 getting into the details and having the details there and
3 ensuring that the details are understood, to ensure that
4 safety is ensured, we don't feel it is necessary for again
5 the least important components.
6 As far as placing additional controls over these
7 least important components for the commercial practices,
8 again we don't feel that those are warranted, that the
9 commercial practices as are suffice.
10 For the high and the medium components, these will
11 continue to be governed by existing regulations and for the
12 non-safety related high and medium components that are
13 identified they would be evaluated for additional enhanced
14 treatment. We would evaluate the treatment that is
15 currently being applied against those, determine whether the
16 functions are being properly ensured, and if there are cases
17 where we see that we need some additional treatment we will
18 apply that.
19 Again we started in with the Option 2 approach for
20 primarily procurement benefits, and so again the focus from
21 a procurement standpoint required that we gain some relief
22 in the areas of ASME, EQ, seismic and 1(E).
23 DR. BONACA: Could you go back again? Maybe I
24 missed something on the fifth bullet.
25 What do you mean that they will be evaluated for
. 183
1 enhanced treatment?
2 DR. SCHINZEL: If we identify a safety significant
3 non-safety related component, currently those have balance
4 of plant commercial controls applied against those.
5 DR. BONACA: Correct.
6 DR. SCHINZEL: In the process of performing our
7 graded quality assurance categorization evaluations, we
8 identify critical attributes for those components. We will
9 look specifically at the critical attributes on those
10 components and determine do we feel that we are doing all
11 the right activities that's going to assure that that
12 function can be performed when called upon, when demanded.
13 What we may do is increase some of the controls as
14 far as performing preventive maintenance activities more
15 frequently. It may entail additional quality oversight over
16 the repair activities, additional supervisory involvement in
17 the oversight of the work activities in the field, and
18 additional documentation -- those types of activities.
19 DR. BONACA: But you are not proposing to step
20 them up to full Appendix B?
21 DR. SCHINZEL: No, we are not proposing either
22 reclassifying those as safety-related or stepping those up
23 into a full Appendix B program.
24 DR. BONACA: Okay. How are you going to track? I
25 mean since you may have quite a diverse set of requirements,
. 184
1 I mean identified on an attribute basis, how are you going
2 to track them? Are you going to have a family of components
3 that have, you know --
4 DR. SCHINZEL: For the components themselves the
5 components are included in our risk significance basis
6 documents that identify the detail and the bases for why
7 these components are classified the way they are.
8 When you get into the applications, some of that
9 will be on specific preventive maintenance activities that
10 we essentially handle on case by case, component by
11 component activity.
12 In other areas, it may entail a change to one of
13 our process procedures.
14 DR. BONACA: Okay.
15 DR. SCHINZEL: if we have a maintenance procedure
16 that we identify in that administrative guideline that for
17 anything that is safety significant, balance of plant, these
18 are the controls that we are going to place on it, our
19 planners, our work supervisors will follow those, so our
20 procedures are going to be our primary implementation for
21 this.
22 DR. BONACA: Okay, thank you.
23 DR. SHACK: But you would see no instance, for
24 example, when you would have to apply an environmental
25 qualification test to a component?
. 185
1 DR. SCHINZEL: Currently the answer there is no.
2 Again, looking at the reliability of the balance
3 of plant components, a lot of these components are operated
4 in humid, hot conditions. They have been designed to
5 operate properly and be reliable in those conditions, and we
6 would continue to operate as we do today. If there's areas
7 where we can provide additional assurance, now recognizing
8 the importance of these components, if there are some soft
9 parts from an environmental standpoint we may place some
10 additional quality oversight to ensure that we are putting
11 in the right gasket, that the bolts are properly torqued on
12 this one particular component.
13 DR. BONACA: But for example I mean if you found
14 one of these components is located into an area which is
15 subject to high energy line break, would you requalify the
16 component to the requirements of high energy line break for
17 that room?
18 DR. SCHINZEL: Part of the evaluation that we
19 would do, again recognizing that the importance of these
20 components could involve fairly involved engineering
21 evaluations, recognizing again the critical attribute and
22 the importance of that function.
23 As far as qualifying it for high energy line
24 break, that would be dependent upon what that engineering
25 evaluation results in.
. 186
1 DR. BONACA: We want them to function.
2 DR. SCHINZEL: Absolutely. We are wanting to
3 ensure that that component is going to function in the arena
4 that it is going to be demanded.
5 DR. BONACA: Okay.
6 DR. SCHINZEL: As far as the exemption request
7 itself, again the exemption is seeking to exclude the low
8 safety and the non-risk significant component from the scope
9 of special treatment.
10 We have included a listing of the regulations that
11 we are seeking exemption from and we have given a brief
12 summarization off to the side of generally what those
13 regulations entail.
14 DR. APOSTOLAKIS: Do you make any distinction
15 between low safety significant and non-risk significant? I
16 mean there is --
17 DR. SCHINZEL: We have a distinction as a
18 classification when we go through the initial categorization
19 of these components.
20 DR. APOSTOLAKIS: I understand that.
21 DR. SCHINZEL: When you get into the actual
22 application, we are essentially heading down closer to be in
23 line with the four box approach --
24 DR. APOSTOLAKIS: I see.
25 DR. SCHINZEL: -- that the Staff is promoting.
. 187
1 DR. APOSTOLAKIS: So both go to Risk 4?
2 DR. SCHINZEL: They would be Risk 3 or Risk 4.
3 Risk 3 would be your safety-related low and non
4 risk-significant.
5 DR. APOSTOLAKIS: Thank you. In retrospect, you
6 have used four categories, right? -- and as you are aware,
7 the Option 2 just proposes two, low and high.
8 Would you still do four if you were starting
9 today? Does it make any difference?
10 DR. SCHINZEL: I think the answer is yes. I think
11 right now the four box approach, it may be an interim step,
12 but it allows for that distinction between safety-related
13 and non-safety related, high safety significant and low
14 safety significant, allows you to transition through the
15 proper controls over each of those areas, and we do need to
16 make some determinations on long-term, how all those high
17 safety significant components are going to be treated.
18 DR. APOSTOLAKIS: I was not referring to the four
19 boxes of Option 2.
20 Option 2 really considers two classes with respect
21 to risk, high and low. You have considered four. Would you
22 still do four or as it turns out doing two would be good
23 enough?
24 MR. GRANTOM: I think we would still do four. I
25 think there is an important aspect of having a medium box.
. 188
1 Things that are high generally are high all the time under
2 all configurations, but you do have this group of components
3 that kind of reflect in a sense a meeting.
4 You needed a buffer between high and low in a
5 sense.
6 I do not expect components to go from low to high.
7 I expect components to migrate possibly from medium to low
8 or medium to high at times, but that buffer of having medium
9 in there is an important buffer to keep components from mass
10 migrations of components going from low to high and having
11 all sorts of implementation issues of when you are trying to
12 backfit what you have done or what relief you have had now
13 that something has all of a sudden popped into the high
14 categorization due to an update, the PRA or update of the
15 performance.
16 I think the medium category offers an important
17 buffer in there and stability.
18 DR. APOSTOLAKIS: I am trying to understand how
19 that would affect what the Staff is planning to do with
20 Option 2.
21 You remember Risk 1, 2, 3, 4. Some of your medium
22 risk significant SSCs would be sometimes in one box and
23 sometimes in another. Some of them will be in one box and
24 some of them in the other?
25 MR. GRANTOM: In general the Box 1, Risk 1, would
. 189
1 be medium and high safety significant components and Box 3
2 would be low and non-risk significant.
3 DR. SCHINZEL: I think one thing --
4 DR. APOSTOLAKIS: But if that is the case, Rick,
5 why bother? That is what I am saying.
6 DR. SCHINZEL: I think one thing that it does for
7 us, by having the medium category, like Rick said it allows
8 a buffer.
9 It allows us to recognize the components that are
10 potentially marginal. They are essentially low in that
11 safety significance box, and as we do some of our
12 categorizations if some of those components fall let's say
13 high in the Risk 3 box or low in the Risk 1 box, there's
14 recognition that those are potentials for migration back and
15 forth.
16 DR. APOSTOLAKIS: But as far as what you are
17 supposed to do to them, there is no difference.
18 In the future you may say this was very close and
19 maybe I will do something about it.
20 DR. SCHINZEL: That's correct. Currently there is
21 no difference.
22 DR. APOSTOLAKIS: Okay.
23 DR. SCHINZEL: It provides insight for us
24 primarily, I believe.
25 DR. APOSTOLAKIS: It is primarily for your own
. 190
1 benefit. Maybe in the future you may want to do
2 something --
3 DR. SCHINZEL: Right. We recognize those that are
4 in between low and high and also we recognize those
5 components that are non-risk significant. They absolutely
6 play no role.
7 DR. APOSTOLAKIS: That was a good qualification,
8 thank you.
9 DR. SCHINZEL: Okay.
10 DR. APOSTOLAKIS: I am not always very negative.
11 [Laughter.]
12 DR. SCHINZEL: As far as the draft Safety
13 Evaluation Report, we did receive that on November 15 and
14 again these are some of our initial insights into the draft
15 Safety Evaluation Report.
16 We have had discussions with the Staff as recently
17 as yesterday to gain some clarifications into these areas,
18 but as far as the draft Safety Evaluation Report, as we
19 first read it, there appear to be several areas that the
20 Staff was agreeable in granting exemptions. Those were
21 given on the board.
22 Some of those areas that had some restrictions
23 that we felt were there we have worked through with the
24 Staff and we have resolved some of those areas and currently
25 we would say that there are no restrictions that apply.
. 191
1 One of those is the maintenance rule area.
2 DR. BONACA: Let me ask you a question again now.
3 Going back to those high safety significance, no
4 safety-related components, if you go to the previous slide,
5 would any of this requirement apply to those?
6 DR. SCHINZEL: Well, for the non-safety related
7 they would not. Now generally we are applying the
8 maintenance rule against those components already.
9 DR. BONACA: Okay, so if they have them in the
10 maintenance rule --
11 DR. SCHINZEL: But again, this exemption is to
12 exclude low safety and non-risk significant components from
13 the scope of these regulations, so if they are high and
14 medium this exemption really doesn't apply to that.
15 DR. BONACA: I understand that. I am just asking
16 what you are planning to do.
17 There is an issue of reciprocity. Either these
18 programmatic requirements are important for safety
19 significant components, typically they are applied to
20 safety-related components which were believed to be safety
21 significant or they are not, and I am trying to understand,
22 you know, what you are trying to do for those safety
23 significant, high safety significant components which, you
24 know, again -- before you said they are not going to be in
25 Appendix B. You are going to evaluate them. But some of
. 192
1 these requirements which I believe are very important are
2 not going to be applied to those components and I just
3 wanted to understand better what your plans were.
4 MR. McBURNETT: The difficulty really comes in
5 when -- this is a small category of items we find, but
6 non-safety related but has high safety significance, to be
7 able to go back and make a backfit these requirements into
8 place since those components weren't purchased and designed
9 from the beginning, they are already in the plant, it is
10 difficult to go back and make them meet the rules precisely.
11 However, what the intent would be would be to look
12 at what makes sense out of this to give the component the
13 necessary reliability and the issue that we have identified
14 that makes it high safety significance, and draw out the
15 things that make -- that are important to make it work in
16 that circumstance.
17 DR. BONACA: I appreciate the fact that he put it
18 in the maintenance rule, which is an important step. What
19 about, for example, 50.59? I mean, the NRC will have no
20 involvement in changes that you make to those components,
21 right, because they won't fall under 50.59.
22 MR. McBURNETT: They will still fall under 50.59.
23 Everything that's in the FSAR will.
24 DR. APOSTOLAKIS: Didn't you request an exemption
25 from that?
. 193
1 MR. SCHNIZEL: We are requesting an exemption to
2 that, again for the low safety and the non-risk-significant.
3 For the balance of plant components, at South Texas we have
4 a single design change program. And so that design change
5 program applies to both safety-related and
6 non-safety-related.
7 And so the 50.59 that we use in that process
8 applies to both.
9 DR. APOSTOLAKIS: By the way, are you also doing
10 -- applying the risk importance measures to LERF, or is it
11 only CDF? Your categorization uses both?
12 MR. SCHNIZEL: It uses both.
13 DR. APOSTOLAKIS: I have seen your criteria where
14 if RAW is greater than 100 and so on. If RAW is greater
15 than 100, either for CDF or LERF, is that the statement
16 that's missing there?
17 MR. GRANTOM: We do -- what you see in the picture
18 is the CDF. What we found is that from a LERF point of
19 view, there are very few components that directly contribute
20 to that. Containment supplemental purge valves are
21 primarily those.
22 And so we pretty much have already included those,
23 and categorized those into the high safety significance so
24 we don't have to focus so much on the LERF issues, and
25 that's why you don't see the emphasis on all the pictures or
. 194
1 all the charts with the corresponding LERF.
2 DR. APOSTOLAKIS: But you did do, since you are
3 really pushing the envelope here, you did do both?
4 MR. GRANTOM: Yes, we did both.
5 DR. APOSTOLAKIS: But in the future, for somebody
6 else, perhaps doing a good job on the CDF and maybe
7 supplementing it with some insights from your work on LERF
8 would be good enough, you think?
9 MR. GRANTOM: I would think so.
10 DR. APOSTOLAKIS: So that's the reason why you
11 don't emphasize that you did both, but you did do both?
12 MR. GRANTOM: We did do both.
13 DR. APOSTOLAKIS: Okay.
14 MR. GRANTOM: And that's how we identified the
15 components that were important and have just included those
16 into the high category.
17 DR. APOSTOLAKIS: Good.
18 MR. SCHNIZEL: Okay, moving on, in addition to the
19 areas where the Staff identified that the exemptions
20 appeared to be appropriate, there were several areas where
21 the Staff indicated that either it was indeterminate at this
22 time, or it appeared that based on the information provided,
23 that an exemption could not be granted. And those areas are
24 shown at -- primarily focus with inservice inspection,
25 repair and replacement of ASME components with non-ASME;
. 195
1 equipment qualification; seismic, I-A.
2 All of these areas are areas where we haven't
3 closed the door on them, and I don't believe the Staff has
4 closed the door on any of these yet. We're going to
5 continue to work with the Staff on these areas to seek
6 resolution, and to seek a successful outcome.
7 And we intend to have extensive involvement with
8 the Staff over the next four or five weeks to try to work
9 through successful outcomes on each of these.
10 DR. APOSTOLAKIS: Now, the word, cannot, is
11 supported by some technical argument someplace?
12 MR. SCHNIZEL: The Staff generally has a basis
13 that's provided in the Draft Safety Evaluation Report that
14 identifies why it cannot be supported. We have some
15 differences of opinion as far as the soundness of some of
16 those bases, and that's what we're working through right now
17 with the Staff.
18 DR. APOSTOLAKIS: A related question: In the
19 previous slides, 6, for example, you really show what one
20 might fairly call massive change here. You know, there are
21 all these 10 CFR 50-dot-something's that you're requesting
22 to be exempted from.
23 And the categorization of the SSCs is done on the
24 basis of single SSCs, right? You find the importance of --
25 MR. GRANTOM: At the component level.
. 196
1 DR. APOSTOLAKIS: At the component level. Is
2 anybody concerned that here we're affecting whole classes of
3 SSCs and something that may not be important on an
4 individual basis, if I have a thousand of those, maybe the
5 class becomes important, the set, the whole set?
6 Has that issue come up and have you resolved it
7 satisfactorily?
8 MR. GRANTOM: We've done some sensitivity studies
9 that were very insightful. You have to keep in mind that
10 the risk-cranking analysis that supports the categorization
11 of the PRA is not just rank the average CDF calculation. We
12 do a series of 21 different sensitivity studies, all
13 including common cause, external events, the full PRA
14 quantifications.
15 And what we do with all of these studies here is
16 that -- what was your question, Doctor?
17 DR. APOSTOLAKIS: You know, let's simplify it. In
18 fact, now that you mention sensitivities, I remember you did
19 a lot of good ones. I may have, say, three components.
20 Each one, on the basis of Fussel Vesely and LERF, ranks low.
21 But taken together as a group of three, perhaps
22 they become important.
23 DR. SHACK: That's the A-4 question.
24 MR. GRANTOM: Right. We have common cause, so
25 that's a grouping. One of the other sensitivity studies
. 197
1 that we did as a result of one of the Staff's questions is,
2 we changed the failure rates of all of the low safety
3 significant components that were in the PRA by a order of
4 magnitude, by a factor of ten.
5 DR. APOSTOLAKIS: I remember that.
6 MR. GRANTOM: And that sensitivity study produced
7 a change in core damage frequency on the order of about two
8 percent.
9 DR. SHACK: That's all the components at once?
10 MR. GRANTOM: All of them at once, all of them
11 together, an order of magnitude in the failure rate, which
12 is a pretty bounding calculation that --
13 DR. APOSTOLAKIS: But did you also change the
14 common cause failure parameters?
15 MR. GRANTOM: I don't know; I'd have to check on
16 that.
17 DR. APOSTOLAKIS: The failure rates usually, you
18 know, we have such redundance in the plants that they appear
19 in powers of something --
20 MR. GRANTOM: But I don't believe we actually
21 changed our beta, gamma, factors by a factor of ten also.
22 DR. APOSTOLAKIS: That would be an interesting
23 sensitivity study.
24 MR. GRANTOM: It could be.
25 DR. APOSTOLAKIS: You are pioneering here, Mr.
. 198
1 Grantom, so you're getting all these questions.
2 MR. GRANTOM: And when we get a grad student to
3 tell us how to correlate those to the component, it will
4 help us also, too.
5 So what we're trying to do here is, we're trying
6 to bound it by using those failure rates at a factor of ten
7 to get an indication of what would happen.
8 And you have other controls in here. The
9 maintenance rule, even at the component level, if a
10 component fails, a higher function, we track that failure
11 down to the component level also.
12 DR. APOSTOLAKIS: But was that issue raised by the
13 Staff and you debated it with them, the issue of classes of
14 components?
15 MR. SCHNIZEL: Yes, it was, and that was the
16 reason that we did that sensitivity study, failing all the
17 LSS components, and looking at a ten times factor increase
18 in failure rates.
19 DR. APOSTOLAKIS: You know, this is an amazing
20 result, though. I mean, you take the failure rates and
21 multiply them by ten, and we see no impact. You really
22 wonder what dominates the PRA.
23 It can't be lambda, so anyway, that's --
24 MR. SCHNIZEL: That's really the premise of Option
25 II, of putting the focus on those high safety significant
. 199
1 components and reducing the focus on the least important
2 components. And that just shows you --
3 DR. APOSTOLAKIS: You increased by ten, the ones
4 that you have found already to be of low significance?
5 MR. SCHNIZEL: Yes.
6 DR. APOSTOLAKIS: That makes me feel better.
7 DR. SHACK: Before you go on, let me come back to
8 this ASME, non-ASME thing. As I understand it, your
9 proposal with the ASME components, you're going to use
10 non-code-stamped components, is what you want to do, but
11 you're going to do the same analysis I would do for an ASME
12 component.
13 Are you going to require, for example, the same
14 kind of code minimum yield strengths? Would you put
15 sufficient restrictions on this commercial component to have
16 what I would believe to be the basic requirements for an
17 ASME component?
18 MR. SCHNIZEL: We're currently working with the
19 Staff on trying to reach a final resolution there. We
20 believe that we've offered a bounding approach that would
21 ensure that we properly factor in allowable stresses, using
22 B-31.1 as the bounding code for some of the replacement
23 components as opposed to ASME.
24 And we haven't worked through an agreeable
25 position yet, and that's one of our main challenges over the
. 200
1 next several weeks.
2 DR. SHACK: Okay, but you don't feel you've come
3 to a head-to-head roadblock?
4 MR. SCHNIZEL: Not necessarily. We've had some
5 initial discussions. We need to continue with those, and we
6 haven't come to a roadblock where we have to agree to
7 disagree. We stated an opinion and the Staff has a
8 different opinion.
9 Continuing on with page 9, as far as some of the
10 additional insights -- and I'll try to go through these
11 fairly quickly, again, we have had some discussions with the
12 Staff, and some of the positions that are stated in here
13 have been altered slightly based on those.
14 When our initial look at the draft Safety
15 Evaluation Report did give us the impression that additional
16 treatment can be imposed or is being asked to be imposed on
17 LSS and NRS components, and in the draft Safety Evaluation
18 Report, it was making statements concerning that the STP
19 commercial practices that we currently have on the balance
20 of plant side, aren't sufficient in all cases for the
21 safety-related, low, and non-risk-significant components.
22 Some of the areas that are identified there have
23 been resolved. We have worked through resolution on the
24 National Consensus Standards and some of the testing
25 activities. We still have some areas to work through with
. 201
1 engineering evaluations or engineering analyses for
2 procurement activities for replacement components.
3 But again, part of our premise, stepping into this
4 is that a single program, a single commercial program was
5 going to be adequate for both the balance of plant side and
6 the least important safety-related components.
7 And that appears to be challenged right now,
8 whether we can actually get to that point. It does look
9 like either we'll have to have two separate programs or we
10 may have to imply some additional burden on the program that
11 we're currently using on our balance of plant side.
12 So, again, we'll continue to work actively with
13 the Staff on trying to reach resolution there.
14 In the categorization areas, one area that was
15 pointed out in the draft Safety Evaluation Report is that we
16 factor in LERF and that factors in the early release
17 consequences, but we may not fully factor in some of the
18 latent effects associated with the release.
19 And so that's currently being reviewed and
20 assessed by our staff. This may require revision to our
21 categorization process, add some additional deterministic
22 insights into that consideration. And we've categorized a
23 number of components so far and it may require some
24 reevaluation of those components to ensure proper
25 categorization.
. 202
1 In the draft Safety Evaluation Report, control
2 over some of the treatment aspects is an issue for us. The
3 draft Safety Evaluation Report is requiring some fairly
4 extensive detail to be included in the FSAR, and some very
5 stringent controls to be placed over the details that are
6 added.
7 With the detail that's included in this FSAR,
8 essentially we would have to come back to the Staff, seek
9 prior NRC approval for any type of change to what's included
10 in the FSAR, and that would even include beneficial changes,
11 minor editorial type changes.
12 And we feel, from a pilot approach, there are
13 going to be lessons learned and feedback that we're going to
14 garner. That we need to have the flexibility to be able to
15 incorporate some of those lessons learned into our process.
16 And there are, recognizing again, the
17 categorization and the categorization being the most
18 important facet of this Option II approach, knowing that the
19 categorization is robust, we feel that there are some other
20 areas where consideration should be given for adjusting
21 scope and specifically in some of the testing areas.
22 One area would be motor-operated valves where
23 there's never a testing program, again, recognizing low
24 safety significance, whether we need to have the same
25 robustness of testing that's currently required in those
. 203
1 areas.
2 DR. APOSTOLAKIS: Are you saying -- the way I
3 understand what you said is that some of the requirements
4 that the Staff is asking you to impose on what has been
5 categorized as a low safety-significant component, the
6 motivation for that, for those requirements might be the
7 lack of confidence in the robustness of the categorization
8 process?
9 And you're coming back and saying, no, our process
10 is robust, so that's the issue there?
11 MR. SCHNIZEL: I believe one of the real premises
12 of the issue comes down to the Option II approach. From our
13 standpoint, we feel that if we can properly categorize the
14 components, that we can apply commercial treatment against
15 those components and those components are going to properly
16 perform.
17 The Staff believes that there is a degree of
18 functionality that has to be assured, and that's part of the
19 requirement that has been placed on the Staff. It is to
20 ensure that that functionality is going to be retained.
21 Now, it's difficult to transition directly from
22 saying commercial is going to be okay, if you don't have
23 something that you can have a control or some type of
24 restriction to ensure that that functionality is going to be
25 there.
. 204
1 So that's part of the crux to the problem that I
2 believe we're running into right now.
3 DR. APOSTOLAKIS: So it's not really lack of
4 confidence in the robustness of categorization? I mean,
5 they just want to see the functionality, some assurance that
6 there is functionality.
7 MR. SCHNIZEL: I believe that's the primary
8 motivator there. Now, generally there is a belief that the
9 categorization is robust. You know, there are some areas
10 that we are continuing to refine with the categorization,
11 but I think there is overall confidence in the
12 categorization.
13 DR. LEITCH: A question about the categorization:
14 I think that is the process that's described on pages 16,
15 17, and 18 of the Safety Evaluation Report. It lists some
16 critical questions and then scores for each critical
17 question. And then there's a table at the top of page 18
18 that talks about the score versus the risk category.
19 And it seems to indicate that scores between zero
20 and 20 fall into non-risk-significant category. I guess I
21 can see how those scores would rate safety significance into
22 low, medium, and high.
23 But is it entirely logical to say that because a
24 score falls below 20, for example, it gets to be
25 non-risk-significant? In other words, I can see this
. 205
1 process being used to grade low, medium, and high, but how
2 does this numerical value then put it into the
3 non-risk-significant region?
4 MR. SCHNIZEL: We have those five critical
5 questions, and as we started in with the graded quality
6 assurance categorization process, we used to answer those
7 questions just with a yes or a no.
8 And as we started through that process, we
9 identified that that really wasn't giving us all the insight
10 that we needed, and so what you're seeing there is the
11 grading and the weighting that goes along with those.
12 Now, what we do recognize is that there are
13 opportunities where a single question may be important in
14 one area only, and it could still come in with a score less
15 than 20.
16 DR. LEITCH: That's right, that's my question,
17 exactly.
18 MR. SCHNIZEL: And what we have identified in
19 there, I believe, if you continue reading around that table,
20 it identifies that there are some areas or considerations
21 that if one single question has a degree of importance
22 associated with it, and it scores a grade of 12 or higher,
23 that as a minimum, that component is going to be categorized
24 as low.
25 Just because it comes in under 20, doesn't mean
. 206
1 that it's automatically going to be non-risk-significant.
2 DR. LEITCH: Okay.
3 MR. SCHNIZEL: If the a single question is
4 answered with -- if we have a score of three or 12 or more
5 as a final grade, then it's classified as a minimum low.
6 DR. LEITCH: Okay, thanks. That exactly answers
7 my question.
8 DR. SCHINZEL: On page 12, continuing, the
9 reliance on deterministic criteria, this kind of goes to
10 some of the discussion that we have already had.
11 It does appear that a portion of the response in
12 the draft Safety Evaluation Report is relying on
13 deterministic bases for response, and again we have
14 identified here some of the foundational premises for a
15 risk-informed approach.
16 We have already talked about the 10 times increase
17 in failure rates, and we have also done some studies where
18 we have evaluated failure rates for safety related and
19 non-safety-related components, industry-wide, looking at
20 over 70 billion component hours of information and at least
21 13 billion component hours of those were on non safety
22 related components.
23 What we essentially found is that the failure
24 rates between components, safety related and non safety
25 related are essentially the same, so that draws some
. 207
1 question as to what are we really getting with some of the
2 additional burden in controls that we are placing on these
3 components.
4 On the next page we had some initial concerns with
5 the particular area of ASME on whether certain key areas
6 were being viewed as design requirements as opposed to
7 special treatment requirements, and we are still working
8 with the Staff on these areas.
9 We don't have a final resolution there. We do
10 know that there's a potential that if these are going to be
11 looked on as design requirements, essentially Option 3
12 areas, that ASME may not be viable for Option 2.
13 There is a draft code case that is being worked on
14 by the ASME code committee. We are also evaluating that for
15 inclusion and matching that against our exemption request.
16 On high safety and medium safety significant
17 components, again in the draft Safety Evaluation Report
18 there's a requirement concerning documented engineering
19 evaluations for beyond design basis considerations.
20 We are discussing this actively with the Staff,
21 again haven't reached a consensus on this.
22 It does pose some challenge, mostly burden
23 associated with getting these engineering evaluations done
24 and we are working with the Staff to determine the necessity
25 for ensuring that that does occur.
. 208
1 As far as conclusions, I am going to turn it back
2 over to Mark for concluding statements.
3 MR. McBURNETT: One of the first things I would
4 like to assure you, you know, you are goal here is to have a
5 robust categorization process and we will assure that we get
6 any question related to categorization resolved to
7 everybody's satisfaction.
8 With that, we will get to an answer on the common
9 cause issues that gets everyone comfortable.
10 There are a couple of key points I would like to
11 make at the conclusion of this and we have kind of hit
12 around them as we have gone through the discussion, but one
13 of them has been this discussion of the assurance of
14 functionality.
15 Through the process, the concept that we have
16 going into Option 2 effort, is we have a very robust
17 categorization process that we can depend on to tell us
18 which things are not risk significant or low risk
19 significance and in doing so that those things that fall
20 into those categories we can rely on commercial practices as
21 an acceptable means to assure they provide the desired level
22 of function.
23 Where our energy is focused as far as ensuring
24 that this is done right is the categorization process and
25 where we believe that the regulatory interest ought to be
. 209
1 focused is on the categorization process.
2 We spent quite a bit of time and are spending
3 quite a bit of time going through, talking about what is
4 adequate treatment for ensuring functionality on these, what
5 are now low or non risk significant items, so we are still
6 spending a lot of our energy focused, a lot of our
7 regulatory energy focused on the least significant items.
8 We look at that and say we think we have our focus a little
9 bit off there.
10 A lot of that is really driven by the way the SECY
11 paper, 98-300, is written and the direction it gives in
12 going and saying we have got to assure functionality. The
13 Staff has to make a finding that the process we use for
14 assurance of functionality is adequate, so that is driving
15 us to doing a lot of effort in that area, and again it seems
16 to be continuing to cause to apply a lot of resource,
17 regulatory resource, and the station will continue to apply
18 resource, but from the regulatory standpoint on the items
19 that are of a less risk -- less significant. That is one
20 key feedback out of this.
21 The second point I wanted to make was that overall
22 for South Texas to be successful at this, what we need out
23 of it, we have gone into this -- this is an expensive
24 proposition as far as going through and building the
25 categorization process and all the infrastructure that
. 210
1 manages that and maintains it, and it is a life of the plant
2 expense. It is something that has feedback loops and you
3 have got to continue to monitor and it is something that
4 doesn't go away once things are initially categorized.
5 We have taken on this, but the expectation back
6 out of that is that we will basically through savings in the
7 areas of seismic, EQ, Class 1(E) and ASME, will be able to
8 return a benefit to the station from a monetary standpoint
9 as well as improve the overall safety performance of the
10 station.
11 DR. APOSTOLAKIS: In a past presentation in the
12 context of GQA your colleagues told us that the expected
13 benefits were on the order of $2 million a year. Now that
14 this is expanded, can you give us a number?
15 MR. McBURNETT: Really to achieve that number, as
16 we see it, we have to get into these items.
17 DR. APOSTOLAKIS: Oh, okay, so including seismic,
18 EQ and so on -- I see.
19 MR. McBURNETT: To achieve that benefit that we
20 advertised several years back we headed into graded QA and
21 found that, well, that doesn't get us far enough to get that
22 benefit that we want to get out of this effort, so we see
23 there is that level of return.
24 We still believe there is that level of return in
25 it but in order to get there we have to get some relief on
. 211
1 these particular ones, and we are still having quite a bit
2 of dialogue with the Staff on those issues, so it is --
3 those are the hardest issues out of the whole stack is in
4 those four categories so we continue to work on those.
5 DR. SHACK: Do you want to venture a guess on what
6 you could get out of the draft SER as it stands?
7 MR. McBURNETT: No. Hadn't really thought it that
8 much through to give you a legitimate answer.
9 That kind of concludes our presentation unless
10 there is any other questions that we can answer -- we're
11 running just a couple minutes over, if you will forgive us.
12 MR. LEITCH: On Viewgraph Number 12, industry data
13 demonstrates that failure rates of safety related and
14 non-safety related -- that is, I guess the industry we are
15 talking about there is the nuclear industry data? You are
16 not including like fossil plant data?
17 DR. SCHINZEL: No. We went into the INPO
18 databases, the EPIX database and the old NPRDS database and
19 we looked at the inputs that were provided through that
20 database and again understanding that most of those inputs
21 are safety-related type inputs, but there were also 13
22 billion component hours' worth of non-safety related
23 information there, so that's what we were drawing that
24 conclusion on.
25 It is nuclear-related experience.
. 212
1 MR. LEITCH: But in your concluding slide when you
2 talk about commercial controls, is that comparable to what
3 you would expect in non-safety related nuclear? When you
4 say "commercial" it says to me that that might be how you
5 would build a fossil plant or a supermarket.
6 DR. SCHINZEL: I think there is a significant
7 difference in commercial controls if you go to a paper mill,
8 if you go to an apartment complex than nuclear commercial
9 controls.
10 Again we would be applying our nuclear commercial
11 controls that we have been using successfully on our balance
12 of plant side and those commercial controls have ensured
13 that our availability and reliability on the secondary side
14 has been in excess of 90 percent. The plant is staying on
15 line. It is in good material condition. It's being
16 maintained extremely well. It is reliable for us.
17 DR. BONACA: But many of the exemptions you are
18 seeking impose -- are from requirements that assure
19 performance under accident conditions. I mean you can't
20 only refer to this comparison.
21 For example, you qualify equipment for
22 environmental qualification that the equipment will never
23 see, hopefully, and that will provide you a comparison if
24 you had it of the performance of safety-related equipment
25 qualified versus not qualified.
. 213
1 DR. SCHINZEL: What we do expect is if we have a
2 qualified, safety-related component today that is low safety
3 significant and it fails, and now we need to replace it, and
4 this exemption has been granted, we would take those same
5 design parameters that we had originally with that component
6 and we would go to a commercial vendor and we would say we
7 have a component here that needs to operate up to a 300
8 degree environment. Give us one that has the soft parts, the
9 fasteners that can operate in that type of an environment,
10 and we believe that we can get those.
11 DR. BONACA: Sure.
12 DR. SCHINZEL: Now we wouldn't have a certificate
13 of compliance. We wouldn't have all of the quality
14 paperwork that goes along with that and we wouldn't need the
15 controls of retaining that long term, but what we would have
16 is appropriate assurance that that component when called
17 upon in those design conditions, we would have reasonable
18 assurance that it will operate.
19 DR. BONACA: And you believe that the conclusion
20 that you draw in the second bullet will still stand?
21 DR. SCHINZEL: Yes, we do.
22 MR. SIEBER: Any other questions?
23 [No response.]
24 MR. SIEBER: If not, thank you very much,
25 gentlemen.
. 214
1 DR. SCHINZEL: Thank you.
2 MR. SIEBER: I would like to call on the Staff to
3 make their presentation.
4 [Recess.]
5 CHAIRMAN POWERS: We need to follow pretty closely
6 to schedule.
7 MR. SIEBER: All right. Perhaps while you
8 continue to work that we can use the handouts that were
9 provided and we can get started.
10 MR. NAKOSKI: Actually I am not going to be able
11 to run this version of PowerPoint slides.
12 [Pause.]
13 MR. NAKOSKI: Good morning, and thank you for the
14 opportunity to brief you on the status of the Staff's review
15 of the South Texas exemption process.
16 The first I would just like to go over kind of
17 what I would like to talk to you about today, kind of a
18 brief timeline of where we started and where we are, a brief
19 discussion of the Staff review process, where we are with
20 the categorization process, the treatment process, change
21 control and a preliminary assessment of the exemption.
22 Let me turn to Slide 3. As South Texas indicated,
23 we received their application in July of 1999. We have had
24 a number of meetings with South Texas. We issued our
25 request for additional information in January of this year.
. 215
1 We had several meetings after that to discuss
2 draft responses to facilitate resolution of the REIs.
3 We received their revised submittal at the end of
4 August and we issued the draft Safety Evaluation on November
5 15th.
6 That brings us to today, to discuss this with you.
7 Going forward, we are looking to resolve the open items by
8 the middle of February of next year and have the final
9 Safety Evaluation prepared and provided to the Commission by
10 the middle of March with a target of briefing the Commission
11 at the end of March and issuing the final Safety Evaluation
12 in mid-April of next year.
13 DR. APOSTOLAKIS: So there is no other
14 contemplated ACRS review?
15 MR. NAKOSKI: That is correct.
16 If you could turn to Slide 4 on the Staff review
17 process, we tried to apply the processes that we have had in
18 place for reviewing individual risk-informed regulatory
19 decisions under what would be considered Option 1 as
20 described in SECY 98-300.
21 Some of the key principles are that it meets
22 current regulations, unless it is subject to an exemption
23 and that is kind of where we are at with South Texas. It's
24 subject to an exemption.
25 We need to maintain some sort of defense-in-depth.
. 216
1 We need to have the categorization, the treatment in
2 combination. We need to have some maintenance of
3 defense-in-depth concept. We need to maintain safety
4 margins consistent with the risk significance of the
5 components.
6 Changes to risk, overall risk, are small, and the
7 Commission has safety goal policy statements that we tried
8 to embed in our review and really we are looking at that's
9 being addressed by the licensee's categorization process.
10 Then finally we really need to look at what the
11 impact of moving forward with the change -- what is the
12 impact? Is it monitored? How is it monitored? Are there
13 performance measures that are being used.
14 Through the process we were challenged to develop
15 some criteria against which we would measure a submittal and
16 in July of this year we sent South Texas some draft review
17 guidelines that have -- they really became kind of a
18 foundation upon which we built to go forward and assess the
19 submittal.
20 The guidelines became the attributes that we would
21 expect to see programs and processes that the licensee would
22 implement to categorize and treat low risk components.
23 That brings us to where are we in the
24 categorization process, and this is described in Section 3
25 of the draft Safety Evaluation.
. 217
1 We generally found that it was acceptable to
2 define the risk significance for the exemptions that South
3 Texas was seeking subject to resolution of the open items,
4 but it is also important to recognize that categorization is
5 tied to the confidence of functionality provided by the
6 treatment process, because there are some assumptions and
7 some performance levels that are assumed in the
8 categorization process, in the PRA that is based on some
9 level of treatment.
10 We really need to have assurance or have
11 confidence going forward that the performance assumptions
12 remain valid or they are appropriately addressed in updates
13 or changes to the categorization or PRA.
14 We did identify a number of open items on the
15 categorization process.
16 The first one is open item 3.1 on page 12 of the
17 DSE and we had questions with the equations on the
18 importance measures on common cause failure similar to the
19 questions that were raised here -- is the change that they
20 proposed the right way to do it? Actually, we came to the
21 conclusion that it was non-conservative. It under-estimated
22 the common cause contribution.
23 DR. APOSTOLAKIS: Why are you focusing so much on
24 common cause failure? I mean it seems to me that the
25 failure rate of a particular component may appear in two or
. 218
1 three different, four terms in the CDF expression, one of
2 which is a common cause failure. I mean there may be
3 others.
4 For example, while you are doing maintenance on
5 one train, then there is a probability the other one might
6 fail, so the failure rates and all that will appear in that
7 term, which is not really common cause failure related.
8 MR. NAKOSKI: I'll give a quick answer, and then I
9 will ask --
10 MR. RICHARDS: John, why don't we see if there is
11 somebody here from the Staff that can speak to that. Yes,
12 we do.
13 MR. LEE: I am Sam Lee. In assessing the
14 importance measure of a component, the licensee for a graded
15 QA approach had proposed a method which also captures the
16 importance of the common cause in which that particular
17 component is part of, so they have used that approach for a
18 graded QA and for the exemption request they have modified
19 it a little bit to what they perceived as an over-estimate
20 for the graded QA approach, so they in assessing the
21 contribution from the common cause factors, they
22 fractionalized it as was indicated earlier --
23 DR. APOSTOLAKIS: I understand.
24 MR. LEE: -- and that is what we were evaluating
25 and whether that fractionalization of the contribution from
. 219
1 common cause is indeed the right way to go.
2 It appears that the first method is an
3 overestimate and it appears that the revised proposal is
4 less I should say overly conservative, so we wanted to make
5 sure that that was done the right way, and we actually took
6 a very simplified example to demonstrate that it may not be
7 as conservative as -- excuse me, it may not be as less
8 conservative as it was --
9 DR. APOSTOLAKIS: Yes, but that addresses the
10 issue of how do you take the common cause failure
11 contribution and assign it to one component.
12 MR. NAKOSKI: I think the question you asked was
13 why are we focusing on common cause.
14 DR. APOSTOLAKIS: Yes.
15 MR. NAKOSKI: Because that is I think the area
16 where we identified a concern in the way that they were
17 calculating it and treating it.
18 DR. APOSTOLAKIS: And what I am saying is there
19 may be another concern, that a failure rate for a particular
20 component may appear in more than two terms in the CDF
21 expression, one being the random failure, the other being
22 the common cause and there may be other things.
23 Why isn't that an issue?
24 MR. NAKOSKI: And again I will have to defer to --
25 DR. BARRETT: If I could give kind of a
. 220
1 qualitative answer -- my name is Richard Barrett. I am with
2 the NRR Staff.
3 I think one of the concerns we have with regard to
4 reductions in treatment is that if you are reducing the
5 treatment requirements for general classes of equipment, and
6 you do it in a way that doesn't preserve functionality, you
7 could be undermining the functionality of redundant
8 equipment so I think that is qualitatively why we are
9 concerned about common cause.
10 Steve Dinsmore of our Staff has perhaps some other
11 insights into it from a quantitative perspective.
12 MR. DINSMORE: Hi. I am Steve Dinsmore from NRR.
13 I guess our feeling was that when you do the RAW
14 on the individual component you pick up pretty much
15 everything except the contribution from the common cause.
16 If you have got a cut set where you have a
17 maintenance outage plus this component, you fail that
18 component, I mean that is all going to be included in the
19 RAW from the individual component.
20 DR. SEALE: Speak up a little louder.
21 MR. DINSMORE: What we worried about was again the
22 common cause because it has such a potential impact, and
23 when we did graded QA we were kind of looking at -- we said
24 if this equipment is subject to common cause, we should look
25 at the importance of the function, which could be failed.
. 221
1 That was pretty conservative. South Texas has changed that.
2 What we are kind of trying to do now is to figure
3 out how to determine the importance of the failure of this
4 component in all or from all reasons.
5 DR. APOSTOLAKIS: I guess what I am saying is that
6 it is not always the whole component, the unavailability
7 that appears in other places. There are parameters like
8 failure rates that may appear in several terms and you can't
9 set a failure rate equal to one or anything. I mean there
10 is a difference between dealing with component
11 unavailabilities and dealing with properties of the
12 components.
13 The other thing is I keep hearing, you know, this
14 may not be conservative. I think the first question you
15 should be asking is is this correct? Whether it gives high
16 or low values is irrelevant if it is incorrect.
17 MR. DINSMORE: Well, we are aware that both of
18 them are not correct, in that sense. The question that we
19 were asking is are they usable -- if we understand what they
20 produce and we can use them.
21 We are trying to develop along with Dr. Abramson
22 in Research a correct -- insofar as the equation is easily
23 defined or what the equation's producing is easily
24 defined -- so that is kind of what the discussions going on
25 now between us and South Texas are.
. 222
1 MR. NAKOSKI: Still on Slide 5, another question,
2 an open item we identified was a criteria for use of
3 Fussell-Vesely importance measure to categorize SSCs as HSS,
4 and it turns out it was essentially a typo in their
5 submittal that got translated into one of their procedures
6 that they are correcting.
7 MR. NAROSKI: Since we, like South Texas, are
8 relying heavily on our categorization process, we expect
9 that people who would be implementing this, that would be
10 exercising their judgment, have strong background at both
11 the plant and the PRA and deterministic backgrounds.
12 As a result, we feel the need to have some fairly
13 stringent qualification attributes for the expert people and
14 the working group members as a condition in the FSAR.
15 DR. APOSTOLAKIS: I guess that confuses me.
16 Doesn't this go against performance-based regulatory
17 philosophies? I mean, have you -- I understand that the
18 licensee has categorized, what, 40,000 components?
19 MR. NAROSKI: Forty or 50,000, right.
20 DR. APOSTOLAKIS: Do you have any evidence that
21 some of them were mis-categorized? Have your experts looked
22 at some of those and said, gee, the expert panel was wrong;
23 we better look at their qualifications?
24 MR. NAROSKI: I'm not sure that we're questioning
25 the qualifications of the people that have done it at South
. 223
1 Texas, to date. What we need is, going forward, some
2 confidence that the people that will be doing it ten years
3 from now or 15 years from now, have the skills -- similar
4 skills and backgrounds to the people that are doing it
5 today.
6 Going forward with these exemptions, we're
7 approving a process. They won't have completed the review
8 of all of these thousands and tens of thousands of
9 components at their plant at the point that the exemptions,
10 should we grant them, are granted.
11 We need to have assurance that that process that
12 forms the basis for our granting the exemptions remains
13 valid. One of the ways that we feel we need to do that is
14 in assuring, providing ourselves confidence that there are
15 controls on who can make up the expert panel, who can make
16 up the working group so that we have confidence that these
17 people have the right backgrounds, the right skills to come
18 to decisions on the categorization.
19 They are a fairly subjective process in the
20 deterministic portion of the categorization. There are some
21 fairly subjective judgments that go into making those
22 decisions that have a direct impact on classifying
23 components as NRS, LSS, MSS, or HSS.
24 DR. APOSTOLAKIS: But do you have any evidence
25 that those judgments were not the right ones, or do you have
. 224
1 any suspicion that the expert panel and the working group
2 right now does not meet the qualifications that you would
3 like them to have?
4 MR. NAROSKI: Today we have no evidence that
5 that's the case.
6 DR. APOSTOLAKIS: And yet it's an open item.
7 MR. NAROSKI: Because we haven't worked and
8 finalized the position that's acceptable both to the Staff
9 and to South Texas. That's why it's an open item.
10 They came in and described their process, the
11 controls and experience for their expert panels and working
12 group panel members.
13 We did not feel that it was sufficient to give us
14 the confidence that going forward, that our basis for the
15 exemptions would remain valid.
16 MR. BARRETT: If I could add a word, John, again,
17 we always have in the back of our minds, Option II and what
18 we have written on Option II, including our take on Appendix
19 T, which we gave the Commission in SECY 99-256, I believe.
20 And one of the things we focused on in this area
21 was some significant differences between the qualification
22 standards that were proposed for this exemption, and the
23 qualification standards that we proposed to the Commission
24 for the general Option II.
25 So, you know, as we continue to talk with South
. 225
1 Texas, whatever we settle on, it presumably has some
2 implications for the larger question.
3 DR. APOSTOLAKIS: Why, for example, couldn't you
4 take 100 components randomly from the 40,000 and have your
5 own panel go through them and gain confidence that what they
6 are doing is okay?
7 MR. NAROSKI: We have taken several of their risk
8 significance bases documents that contain the basis for
9 their determination of the categorization of these systems
10 and components.
11 DR. BONACA: I understand you're worrying about
12 the future.
13 MR. NAROSKI: Yes, going forward.
14 DR. BONACA: So if you --
15 DR. APOSTOLAKIS: South Texas is future?
16 MR. NAROSKI: That's correct, South Texas is
17 future. We are approving a process. If they would have
18 completed the review and provided a list of all the
19 components, this probably wouldn't be an issue.
20 This is a process that they are planning to
21 implement after the exemptions are granted, should we grant
22 them, through their remaining life of the plant.
23 We need to have a basis, we need to have
24 confidence that our basis remains valid. We can't grant an
25 exemption without having confidence that the basis will
. 226
1 remain valid, going forward.
2 DR. APOSTOLAKIS: But you will also have a
3 monitoring program, you will also be free to check. Anyway,
4 okay, I understand.
5 MR. NAROSKI: Okay. Another area that we
6 identified as an open item is how to treat those components
7 that mitigate the consequences of accidents, primarily by
8 maintaining containment integrity.
9 CDF doesn't necessarily deal with containment
10 failures. LERF deals with certain aspects of containment
11 failures, but there are other failures out there that could
12 have as large or equivalent impacts on public health and
13 safety as large early release from a containment or early
14 containment failure or just outside the definition of a
15 large early release, or a late large release. There are
16 scenarios that aren't necessarily captured by CDF and LERF.
17 We see that as a limitation and we're working with
18 South Texas to kind of come to a position as far as how to
19 address that.
20 DR. APOSTOLAKIS: Did we discuss this when we
21 developed Regulatory Guide 1.174; is it there? I don't
22 remember.
23 MR. BARRETT: I believe that there was a lot of
24 discussion when Reg Guide 1.174 was developed, as to whether
25 or not there should be other criteria such as late
. 227
1 containment failure or some risk type of measures.
2 And I think there was a conscious decision made
3 that CDF and LERF were the best measures to use for this, in
4 general. Our concern here is that there's a whole class of
5 equipment in the plant for which those criteria are simply
6 irrelevant.
7 And we're asking South Texas to propose a
8 criterion that will allow us to have a rational basis for
9 saying -- for instance, a lot of the equipment in
10 containment, which category it falls into.
11 DR. APOSTOLAKIS: I guess my question is, are we
12 going beyond 1.174 here?
13 MR. BARRETT: Yes.
14 DR. APOSTOLAKIS: We are?
15 MR. BARRETT: Well, we're going beyond the literal
16 reading of 1.174, yes. We're not trying to say that this
17 equipment is necessarily of high safety significance; we're
18 saying that there should be a criterion, a rational
19 criterion by which they can be judged, and we're asking the
20 licensee to propose something that has an equivalency to CDF
21 and LERF.
22 DR. APOSTOLAKIS: Is there any discussion in 1.174
23 that one is allowed to do this?
24 MR. BARRETT: I don't believe so, no.
25 DR. APOSTOLAKIS: Didn't the Commission tell us
. 228
1 this morning that the regulatory system must be predictable?
2 We issue Regulatory Guide 1.174 and now we go beyond it?
3 This is not a good day for me; is it?
4 [Laughter.]
5 DR. KRESS: He also says that if it's the right
6 thing to do, we should do it, whether the regulations allow
7 it or not.
8 DR. APOSTOLAKIS: It all depends on what doing it
9 means.
10 DR. KRESS: Determining whether or not there is
11 some significance to these other components having to do
12 with the release of fission products and doses and things
13 like that that aren't captured by LERF.
14 DR. BONACA: Because the metrics don't capture
15 everything.
16 DR. APOSTOLAKIS: So how do I know then in the
17 future, if I choose to go to the risk-informed way and I
18 pick up on 1.174, that there may not be issues that will be
19 raised?
20 DR. BONACA: I think we have raised the issue of
21 whether or not the metrics of 1.174 are --
22 DR. APOSTOLAKIS: But it's not in the books.
23 DR. BONACA: I understand that.
24 DR. APOSTOLAKIS: All right. The thing that
25 surprises me is that everybody seems to think it's so
. 229
1 natural to do this.
2 DR. BONACA: I think it's natural because if I
3 only look at Fessel Vesely, I mean, I will be left with that
4 concern.
5 DR. APOSTOLAKIS: The Regulatory Guide doesn't say
6 that.
7 DR. BONACA: I know, and I understand that.
8 DR. LEITCH: I'm a little confused when I look at
9 page 16 of the draft SER. There's a list there of critical
10 questions.
11 The first question is: Is the function used to
12 mitigate accidents or transients -- are we saying here that
13 waiting is not appropriate? I'm just not -- I don't
14 understand how that categorization process doesn't address
15 the second bullet.
16 MR. NAROSKI: Is the function specifically called
17 out in the emergency operating procedures or emergency
18 response procedures?
19 DR. LEITCH: I'm sorry, the first question, does
20 the function used to mitigate accident or transients -- I'm
21 wondering how that does not address your second bullet on
22 your slide.
23 MR. NAROSKI: And I have Mr. Dinsmore back there
24 who is prepared to answer that.
25 MR. DINSMORE: This is Steve Dinsmore from the
. 230
1 Staff again. Our understanding is that when they -- that
2 question may only apply to CDF and LERF, to sequences which
3 could cause a core damage frequency, or which could cause
4 core damage or could cause a large early release.
5 DR. APOSTOLAKIS: This is under deterministic
6 risk, so it's not really CDF or LERF here.
7 MR. NAROSKI: What we understand is that when they
8 review that and answer that question, they're looking at
9 does it mitigate core damage frequency, or does it mitigate
10 large early release frequency? It's not necessarily does it
11 mitigate accidents or transients.
12 MR. DINSMORE: Or accidental release through the
13 sewer system.
14 MR. NAROSKI: Period. They focus on core damage
15 and large early release. That's our understanding.
16 MR. DINSMORE: And since they're not complaining,
17 I guess that's the right understanding.
18 DR. APOSTOLAKIS: Is this the right understanding?
19 MR. NAROSKI: Rick? Come up to the mike if you
20 want to respond.
21 MR. GRANTOM: Rick Grantom. When I go through the
22 categorization process, their focus isn't on core damage
23 frequency, and they're looking -- you're talking about
24 accidents and transients, so you're trying to follow this
25 thread all the way through to lead to a large early release
. 231
1 frequency, too.
2 DR. APOSTOLAKIS: So the Staff is right?
3 MR. GRANTOM: Yes, on that.
4 DR. LEITCH: Thank you.
5 MR. SIEBER: I'd like to sort of speed up a little
6 bit. We have 15 minutes left, and about 15 slides, so let's
7 see if we can move a slide a minute.
8 MR. NAROSKI: The next open item is support for
9 the ASME Section 11 ISI exemption. Really it relates to
10 passive functions.
11 In the categorization, PRA doesn't really deal
12 directly with passive functions. The categorization process
13 similar to that doesn't really support the categorization of
14 passive components such as piping, which would get into the
15 basis for the ISI program.
16 Another open item on categorization is the use of
17 general notes. This is a process that was in development
18 actually as we were getting the submittal.
19 It's been finalized. They agreed that they will
20 submit that to us so that we have an understanding of what
21 the process is for the development, application, and control
22 of general notes.
23 The reason we're concerned with that is that this
24 potentially could categorize more components than any other
25 part of their process.
. 232
1 If you go to Slide 7, in the area of treatment,
2 for high safety significant, medium safety significant SSEs,
3 we generally found that it provides confidence of
4 functionality, however, there was an open item that deals
5 with the description of the attributes of the processes for
6 treating the beyond-design-basis functions of
7 non-safety-related and safety-related components.
8 And as you heard from South Texas, they would
9 still, these HSS and MSS components, would still remain
10 within the scope of the existing NRC regulations that are
11 applicable to those components.
12 For low safety significant and non-risk
13 significant components and functions, we again generally
14 found that it provides confidence of functionality, or there
15 were some open items that we identified.
16 Principally what we are looking for, the Staff is
17 looking for, is the attributes of the processes that they
18 plan to use to treat low risk components. We're not
19 necessarily interested in how they're going to do it. We
20 want to know what are the key points, key features of their
21 processes.
22 And in procurement, we identified four attributes
23 that weren't clearly specified in their submittal or
24 committed to in their proposed FSAR section. The first is,
25 is what they ordered or what they got what they ordered, you
. 233
1 know, some sort of check that that's the case; that they,
2 where appropriate, use vendor recommendations.
3 One of the key points that we're discussing is the
4 use of engineering evaluations to provide confidence that
5 the replacement components meet the design basis inputs, the
6 design features, and we're working with them to resolve
7 that.
8 Another is the use of National Consensus
9 Standards, and as you heard this morning, we pretty much
10 have come to closure on that, that aspect, with South Texas.
11 If you could go on to Slide 9?
12 In the area of installation, what we're looking
13 for there is pre-operational and pre-service testing.
14 Basically, after they've installed it, they need to do some
15 sort of testing, and they have described that they do, and
16 it shows that maintenance that was done correctly and that
17 the thing works as expected.
18 And then, again, there is the National Consensus
19 Standard question.
20 In the area of maintenance, we're looking for them
21 to include the attribute that they consider vendor
22 recommendations; that the corrective maintenance will
23 implement the results of the outcome of the corrective
24 action program.
25 That they basically do appropriate
. 234
1 post-maintenance testing; and, again, National Consensus
2 Standards.
3 If we could go to Slide 10, in the area of
4 inspection, test, and surveillance, there's a common theme
5 going through here: Vendor recommendations are considered;
6 National Consensus Standards.
7 When they do testing, though, one of the things
8 that we would -- we need to provide confidence of
9 functionality is that where they can, they take credit for
10 testing and compare it to conditions at design basis,
11 basically to show that it can function at design basis
12 conditions.
13 It doesn't necessarily mean that it has to be
14 tested at those conditions; that's pretty much impractical
15 in most cases, but that they have some sort of method to say
16 I've tested it under these conditions and here's why I have
17 confidence, some evaluation or analysis that is necessary
18 that shows that it would still function under design basis
19 conditions.
20 MR. SIEBER: Quick question on this: One of the
21 exemptions is relief from the requirements of ASME Section
22 11. Section 11, I think, has both ISI and IST.
23 MR. NAROSKI: Yes.
24 MR. SIEBER: So if you eliminate the IST, part of
25 it then -- the periodic tests that you would use to assure
. 235
1 functionality disappear.
2 MR. NAROSKI: That's correct for these low-risk
3 components.
4 MR. SIEBER: And on the other hand, your
5 requirement is that it still should be functional, so how do
6 you establish that, by some other kind of test?
7 MR. NAROSKI: Using existing tests. I mean, they
8 do system tests under tech specs, they do post-maintenance
9 testing; they do -- well, in the system tests, if they can
10 show that, for example, a valve has to stroke and the valve
11 strokes and operates correctly under normal plant
12 conditions, there are a couple of cases that valve could be
13 in its normal plant conditions could be equivalent to the
14 design basis conditions.
15 That could be given credit as this valve will
16 function under design basis conditions. You need to do an
17 evaluation that says that that's the case.
18 MR. SIEBER: What about a pump, for example, a
19 pump that's normally in service, its operating parameter
20 degrade as time goes on and clearances open up and so forth?
21 If you eliminate the testing, you have no way or basis to
22 establish rate of degradation and when it fails to achieve
23 its minimum design flow rates, pressures, et cetera.
24 MR. IMBRO: Perhaps I could add something here,
25 and one minor clarification to start: Section 11 doesn't
. 236
1 really address IST. IST is addressed through the O&M Code,
2 so it's really the O&M Code that we're talking about here.
3 That's a minor point.
4 But what we're I guess saying in the exemption is
5 that the Staff will grant them an exemption to O&M testing,
6 however, this is based on the fact that these components
7 have a very small contribution to CDF, but we would expect
8 and we would ask them to commit that they do have some
9 testing program that would assure or provide some assurance
10 that the components would be functional at the design
11 conditions, and it would also have the attribute that
12 includes service-induced aging, so that there would be some
13 type of trending that the program would have, which would
14 clue them to potential incipient failures.
15 So it's not that they won't do testing; the
16 testing will be outside of regulatory control, and they will
17 do it as much as possible through the normal programs that
18 they have available at this station.
19 MR. SIEBER: IST is one of those programs.
20 MR. IMBRO: Yes.
21 MR. SIEBER: Okay, go ahead.
22 MR. NAKOSKI: The final area we had -- where we
23 were looking for additional attributes, still on Slide 10,
24 is management and oversight.
25 Again we're looking at the vendor recommendations
. 237
1 and the use of national consensus standards and a commitment
2 to control surveillance equipment, measuring and test
3 equipment going forward.
4 We have identified two confirmatory items where we
5 saw that the licensee requested to do something that was
6 contrary to existing commitments that we couldn't fully
7 support.
8 Well, the first one is where we saw
9 inconsistencies in their program description. The first is
10 Class 1E components that are not fully qualified. If they
11 can't get a replacement on it, they plan to isolate it from
12 the 1E circuitry, but their process as described to us
13 didn't really describe what they would do for ensuring
14 functional capability of the replaced component.
15 They make an assumption on when a low risk SSC
16 exceeds its qualified life that it is assumed to be
17 functional. It is really not supported. They don't provide
18 a basis for why that assumption should be valid or
19 acceptable.
20 They talk about functional requirements that
21 envelope credible design basis conditions and the Staff's
22 position -- and as we understand their submittal they say,
23 well, it's a design basis condition. They will maintain the
24 inputs. They will maintain design basis for design basis
25 conditions, not necessarily credible design basis
. 238
1 conditions. The design basis is what it is. You need to
2 maintain functionality within that envelope and similar
3 "designed to function in installed environment" -- it is
4 again the design basis environment.
5 Go to Slide 12. The next confirmatory item deals
6 with that part of South Texas's request where they basically
7 sought our approval to wipe the slate clean on commitments
8 for LSS and NRS components.
9 We are really not in a position I think to do that
10 because we don't clearly understand the impact of that. It
11 is kind of an open-ended exemption from 50.59. There is
12 currently an NRC endorsed guidance out there on how to
13 control commitments, and we are seeking a commitment back
14 from South Texas or a confirmation from South Texas that
15 they will continue to follow that NEI guideline.
16 We can go to Slide 13. Earlier we were talking
17 about having confidence that the basis for our exemption
18 remains valid going forward, and that gets to the heart of
19 controlling changes to the processes that form the basis for
20 our granting any exemptions that we may grant.
21 Again, I just need to emphasize we are looking to
22 approve these exemptions based on accepting processes. I
23 don't think we have done that before so we are kind of
24 breaking new ground there.
25 We need to have confidence going forward that the
. 239
1 basis for our exemption remain valid and that changes don't
2 invalidate that.
3 DR. APOSTOLAKIS: Isn't it an important part of
4 your basis for the exemption the calculations by either
5 computer program that South Texas uses? Why aren't you
6 concerned about the accuracy and the correctness of the
7 results of that program?
8 Have you reviewed it?
9 MR. NAKOSKI: For the PRA for example?
10 DR. APOSTOLAKIS: Yes.
11 MR. NAKOSKI: I don't believe we reviewed -- and
12 maybe Rich can help me out there, but I don't believe we
13 reviewed the calculational programs --
14 DR. BARRETT: Well, Steve, go ahead. Steve
15 Dinsmore -- this is Rich Barrett.
16 We have a great deal of experience with South
17 Texas PRA, probably more than any other PRA in the American
18 nuclear industry.
19 We had a detailed review of that PRA performed by
20 the Office of Research in the early '90s, also, when we
21 reviewed their graded QA submittal.
22 Perhaps Steve could talk a little more to the
23 details because I was not here at that time but we did
24 perform some oversight of that PRA at that time so it is one
25 PRA that we have a great deal of experience with and a great
. 240
1 deal of knowledge with. Steve, would you like to add
2 something?
3 MR. DINSMORE: This is Steve Dinsmore.
4 Just that again when we did the GQA review we
5 essentially looked at the quality assurance which was used
6 to develop the calculational program, the RiskMan program.
7 We got in touch with I guess it was PLG then and
8 we talked to them on the telephone and got their QA records
9 on how they developed the program and we considered that
10 that was sufficient to assure that the program was doing
11 what we thought it was doing.
12 MR. NAKOSKI: So the short answer to your question
13 I guess is yes, we have reviewed your calculational methods.
14 DR. APOSTOLAKIS: No, you haven't. You asked them
15 how they did it and you just accepted what they told you.
16 MR. NAKOSKI: Steve, you are headed back to the
17 microphone?
18 DR. APOSTOLAKIS: Would you do the same to a
19 thermal hydraulic code, if a company develops a code like
20 that and submits the calculations?
21 Would you call them up and say what quality
22 control measures did you use and if you found them
23 satisfactory, and you say this is good enough?
24 MR. DINSMORE: Steve Dinsmore again. I think
25 there was also benchmarks run. I mean they had a whole
. 241
1 process and we didn't repeat the process, no, that's
2 correct.
3 MR. NAKOSKI: Gene, you want to address that?
4 [No response.]
5 DR. APOSTOLAKIS: I think I got the answer, yes.
6 MR. NAKOSKI: Okay. We tried to develop a set of
7 change controls that would give us the confidence going
8 forward that the basis will remain valid but give South
9 Texas flexibility to change their procedures.
10 That is why we are focusing in what we are asking
11 for them to commit to in the FSAR on the process attributes
12 that -- the processes that we are going to base our
13 conclusions on for the exemptions.
14 We feel that that would give them flexibility in
15 changing their programs, their procedures, without
16 necessarily requiring a large number or a significant need
17 to revise the description in the FSAR which would get into
18 the need to either come in for a new exemption or an
19 exemption amendment.
20 Part of what was driving this is during graded QA
21 we discovered that 50.59, having a description of the
22 process in the FSAR without some other change control, 50.59
23 by itself really wasn't sufficient.
24 We worked with South Texas on graded QA, came up
25 with a change control process that relied very heavily on
. 242
1 50.59. Subsequent to granting or approving the graded QA
2 changes were made to their process that were consistent with
3 the requirements of 50.59 that we didn't anticipate and that
4 is kind of what is driving us to feel or to need a more
5 stringent control on changes to assure that our basis
6 remains valid.
7 Really what that ends up with, the kernel, the
8 center of the control is really they can't change the FSAR
9 description without our prior approval.
10 Please go to Slide 14.
11 MR. SIEBER: Why don't I interrupt here just for a
12 second. I think that Slide 14 and 15 are just roll-ups of
13 how you are treating on a regulation by regulation basis of
14 the information that you just gave us, and I think we can
15 all read that.
16 If anybody has any questions, just glancing
17 through it, you can ask them now. If not, I think that in
18 the interest of keeping to our schedule that I want to thank
19 the Staff for their presentation. I also understand that
20 NEI would like to say a few words at this time.
21 If that is not the case, let me know.
22 MR. HEYMER: This is Adrian Heymer from NEI.
23 In the interest of time we don't have anything to
24 add that hasn't been said by South Texas, specifically Mark
25 McBurnett's concluding comments, so I think all I would say
. 243
1 is very similar to what he would say.
2 MR. SIEBER: Okay. Well, thank you very much and
3 my thanks and appreciation to all of the presenters.
4 Mr. Chairman --
5 CHAIRMAN POWERS: I guess I am still left with the
6 question of how we approach understanding the derivation and
7 bases for these remarkable equations on pages 83 and 84.
8 I mean they appear here without reference. Is
9 there some mechanism to ever understand where these
10 equations come from or are they forever --
11 MR. SIEBER: Yes, why don't we ask South Texas.
12 MR. GRANTOM: This is Rick Grantom.
13 These equations were developed as an attempt to
14 try to correlate all the common cause contributions to a
15 component. We intend to get together with Dr. Apostolakis.
16 We will bring the software vendor in -- PLG EQE also -- but
17 this is not a question about the software calculating common
18 cause correctly. That has been reviewed. That is part of
19 the software quality controls that the vendor has.
20 That is part of the installation and verification
21 that we use at South Texas when we install this code on our
22 own computers. That part is not in question.
23 The part in question is how do you extract from
24 all of the common cause terms the contribution that a
25 component has to all of those various terms.
. 244
1 Dr. Apostolakis' question on you could have
2 maintenance in one train and a failure in the other one,
3 those are explicitly handled in the PRA in what we call
4 alignments. There's maintenance alignments, testing
5 alignments, and normal alignments and under those alignments
6 the cut sets can be different. That is explicitly handled
7 in the software that quantifies the PRA.
8 This question goes strictly to common cause and
9 how we extract common cause to the components, and we made
10 an attempt to respond to the Staff's concern about that by
11 doing something that is essentially outside of the computer
12 code, RiskMan.
13 DR. KRESS: On that question of alignments, you
14 have estimates of how many types of alignments you are going
15 to have throughout the lifetime of your plant and the
16 duration of the time they will be in those alignments?
17 MR. GRANTOM: We have alignments based on our
18 operational procedures for what that system is permitted to
19 do, so it can be in a normal alignment which is set up for
20 normal operations.
21 There is an alignment for testing.
22 There can be an alignment for maintenance.
23 DR. KRESS: And then those are factored into your
24 CDF calculations?
25 MR. GRANTOM: Yes. They are. They are explicitly
. 245
1 modeled.
2 MR. SIEBER: Getting back to Dr. Powers's
3 question, I guess the answer is no.
4 CHAIRMAN POWERS: Okay. I take it the answer is
5 no.
6 On that cheery note, I will recess us for lunch.
7 We will be --
8 DR. SEALE: Dispersed.
9 CHAIRMAN POWERS: We will resume at ten minutes of
10 1:00. I will remind the members that they are obligated to
11 participate in some lunchtime interviews either in Group 1
12 or Group 2.
13 [Whereupon, at 11:50 a.m., the meeting was
14 recessed, to reconvene, at 12:50 p.m., this same day.]
15
16
17
18
19
20
21
22
23
24
25
. 246
1 A F T E R N O O N S E S S I O N
2 [12:50 p.m.]
3 CHAIRMAN POWERS: Let's come back into session.
4 The next topic of business is control room habitability.
5 Dr. Kress, I understand that you spent a lot of
6 time habiting control rooms, so maybe you can help us in
7 this area.
8 DR. KRESS: I habitually habited. Yeah, our
9 Subcommittee on Severe Accident Management recently had a
10 subcommittee meeting November 15th, at which we primary
11 reviewed an NEI document, 99-03, that has the title of
12 "Control Room Habitability Assessment Guidance."
13 This document was developed by NEI in response to
14 concerns that we have heard about before over control room
15 habitability that surfaced actually in the '80s, mid '80s or
16 so, in which audits showed that the in-leakage values for
17 the plants exceeded their design basis, sometimes by a
18 considerable margin. And the staff and the NEI have had I
19 think what I would call extensive interactions on this
20 document. And I think this is in the hope, on both their
21 parts, that perhaps the document might serve as the basis
22 for a Regulatory Guide.
23 The document looks to me like it does provide some
24 excellent guidance, but there are still some open items
25 between the staff and the NEI. And in my mind, there are
. 247
1 several of these, but in my mind the major open items that
2 you might want to think about while we are hearing
3 presentations are whether or not component testing is
4 sufficient to determine the total in-leakage at least for
5 the first baseline in-leakage testing, if there is to be
6 one.
7 And then the second open item that I think you
8 might want to focus on is whether or not to include some
9 sort of a limiting in-leakage value within the technical
10 specs. There are other items, but I think those are the
11 ones that are major.
12 So, with that, I guess I will turn the floor over
13 to -- who is starting? Is it NEI, Mr. Cozens of NEI, or the
14 staff?
15 MR. COZENS: I am bringing this panel to make it
16 quicker.
17 DR. KRESS: Okay.
18 MR. COZENS: I am Kurt Cozens with NEI, I am the
19 Senior Project Manager responsible for the control room
20 habitability issue. At our subgroup presentation, we had a
21 two-and-a-half presentation and were asked to considerably
22 shorten it, and we had a lot of details. And we have indeed
23 shortened the presentation, but because of the breadth of
24 technical issues that are concerned with the control room
25 habitability issue, I have invited three gentlemen to help
. 248
1 me make the presentation so we can give you the proper
2 information.
3 To my left here I have Ken Taplett from South
4 Texas. I have Steve Schultz from Duke and Bob Campbell from
5 TVA. They will each be making a short part of the
6 presentation, and we thought that this panel format would be
7 a little bit easier, rather than having people jump up and
8 down just for a couple of slides, and we thought that would
9 be a little bit smoother in the presentation.
10 So let me just continue. Just to kind of baseline
11 everybody what the purpose of why we wrote NEI 99-03, it
12 provides an approach for assuring adequate protection of
13 control room operators during abnormal events, and we
14 specifically have focused on the radiological and toxic gas
15 events.
16 It has always been our desire to obtain NRC
17 agreement with that approach. This is a guidance document.
18 We have the desire to provide it to industry so that they
19 might be able to follow it, that it would be a better set of
20 instructions, a learning document, guidance, and some
21 background piece so they can understand how we have arrived
22 at the recommendations that we have indeed arrived at.
23 You will hear more from the staff. There are some
24 differences that remain out, as Dr. Kress has mentioned.
25 And you will probably be hearing more about that later
. 249
1 today.
2 The last time the full committee had seen this
3 document was the August 1999 draft. It has significantly
4 been rewritten. The staff presently has the October 2000
5 draft, and this draft contains the various positions that
6 are listed on this slide, which I wanted to review briefly
7 with you to demonstrate some of the changes that we have
8 made in the document since you had seen it previously.
9 First of all, for the control room in-leakage,
10 basically focusing on the unfiltered in-leakage aspect of
11 that, and for the toxic gas, we are recommending that
12 licensees perform a baseline test of the in-leakage and a
13 baseline assessment of the toxic gas to determine if there
14 has been any changes on the toxic gas.
15 For the in-leakage, we are also recommending a
16 periodic assessment, which we will have some discussion on
17 what that actually constitutes. For the toxic gas, we are
18 also recommending that plants periodically assess if they
19 have adequately characterized their toxic gas risk following
20 the existing regulatory guidance and requirements that
21 presently exist within their licensing bases.
22 We have examined the issue of smoke infiltration.
23 The staff raised a concern about the fact that if we do
24 understand that these control rooms are not as leak-tight as
25 we had once assumed, that the issue of smoke needs to be
. 250
1 addressed. We have looked at this particular activity. We
2 are going to talk about it more, but we have come back with
3 a recommendation that licensees undertake a qualitative
4 approach on managing smoke, being that there is no clear
5 quantitative approach that we could define that could be
6 done, and we have more to say on that.
7 DR. KRESS: On your first two bullets up there,
8 you talk about periodic assessment of the baseline of the CR
9 in-leakage and a periodic evaluation of the toxic gas.
10 Could you give us a little better idea of what is meant by
11 that?
12 MR. COZENS: We have slides on both of those
13 topics.
14 DR. KRESS: You have slides later, okay.
15 MR. COZENS: Yes.
16 DR. KRESS: I will wait.
17 MR. COZENS: We are going to cover those in more
18 depth.
19 DR. KRESS: Okay. I will wait.
20 MR. COZENS: I just wanted to do an overview here.
21 DR. KRESS: Okay. Thank you.
22 MR. LEITCH: When you say control room, you are
23 talking about that structure, that facility that is
24 continuously manned, as contrasted from -- I mean a number
25 of these plants have a control structure with a control room
. 251
1 and an auxiliary electric equipment room or a cable
2 spreading room? I mean you are just talking about the
3 control room, per se?
4 MR. CAMPBELL: Yeah. This is Robert Campbell with
5 TVA. What we are talking about is the control room
6 envelope, the part of the plant that is to remain habitable.
7 Now, sometimes that includes some control panels, sometimes
8 it doesn't. But, generally speaking, it will include the
9 main control panels where the operators are. It will
10 include some facilities like toilets, locker rooms, things
11 like that, and maybe a mechanical equipment room or so. And
12 that is the area that we are talking about.
13 It could be housed in a structure called the
14 control building. In some plants it is the whole building.
15 But that is the area we are talking about. And I guess I
16 will just reiterate here is that no two plants in the
17 country are alike in what they consider as they control room
18 boundary.
19 DR. KRESS: Is the control room boundary spelled
20 out in your technical specs or in the licensing basis
21 somewhere? Where would I go to find, for a particular
22 plant, what its control room envelope is?
23 MR. TAPLETT: This is Ken Taplett, South Texas
24 Project. That would be, at least our plant, I think most
25 other plants, that would be defined in your updated final
. 252
1 safety analysis report.
2 DR. KRESS: In the final safety.
3 MR. TAPLETT: So it would be in that report, and
4 that would be part of your licensing basis.
5 DR. KRESS: Thank you.
6 MR. LEITCH: In that context, is one of the toxic
7 gases considered to be halon used in the fire suppression
8 system and portions of that building? Is halon considered a
9 toxic gas in this sense?
10 MR. COZENS: The scope of the toxic gas that are
11 covered in this document are those which are part of the
12 current licensing basis of a plant, which would have
13 included things that -- let's see, the two Reg. Guides are
14 1.78 and 1.95, is that correct? That are defined via those.
15 That is typically what most plants are characterizing as
16 toxic gases.
17 MR. LEITCH: Okay.
18 MR. COZENS: We have, as you might have inferred
19 from some of my comments, the 99-03 document establishes its
20 contents based upon the existing plant licensing basis. We
21 are here to assure that we are maintaining our licensing
22 basis.
23 The staff has identified the potential that some
24 plants, when they did their initial licensing, or subsequent
25 even, may not have properly evaluated their limiting
. 253
1 accident scenario. There was some guidance in the Standard
2 Review Plan which most people at that time pretty much
3 assumed that, with the LOCA, was the limiting accident, and
4 that is what many plants did.
5 With increased knowledge, there are some
6 situations where that may not be true, and they may not have
7 always analyzed it. That is yet to be determined. It
8 hasn't necessarily been evaluated. But we have defined
9 those accidents that are presently defined in the licensing
10 basis as those which a plant should consider and assure
11 themselves that they have indeed examined their limiting
12 accident.
13 The next bullet I have down here is what I
14 nominally call my housekeeping. Before you go out and do a
15 lot of this effort, you ought to make certain that your
16 paper for your procedures, your design, your operation, your
17 plant -- and the physical plant actually match up. Just do
18 a little bit of housekeeping to assure that you are matching
19 up with what you say you are. And we have advised plants
20 that they should indeed take that step early in the process
21 before they go on to look at other larger questions.
22 This particular document has realized that since
23 the early evaluations at plants, technology has moved on a
24 long ways and the initial methods of evaluating radiological
25 doses and other analysis methods such as the source term
. 254
1 have moved along and grown, and improved, and we are looking
2 at ways of using these to address issues that might be
3 identified.
4 One of the major points that we have brought into
5 this document is a realization that not only do you need to
6 examine your plant once, you need to be diligent in managing
7 your plant and establish a program to assure that as the
8 plant is used, and things are changed, modifications, new
9 equipment, whatever might come in, that you are indeed
10 maintaining the plant in accordance with the requirements of
11 the regulations and your licensing bases.
12 So we believe that a plant needs to put into place
13 a program that would maintain this plant, its maintenance,
14 its operation, and we have some more details on that which I
15 will let the slides go into that rather than take it at this
16 point.
17 But our overall goal has been to look at the
18 entire program of control room habitability and ask
19 ourselves, what is reasonable to give adequate assurance
20 that the control room operator is indeed receiving the
21 protection that he needs during an abnormal condition? And
22 we believe what we have put forth in the document actually
23 achieves those goals.
24 And so with that kind of as an overview of the
25 document, where it stands presently, it is about 200 pages,
. 255
1 so you may have not read it all, I am going to turn to Ken,
2 Ken Taplett, and let him talk about some baseline test
3 results. Ken.
4 MR. TAPLETT: Good afternoon. Again, I am Ken
5 Taplett from South Texas Project.
6 We believe that baseline testing does need to be
7 conducted to determine your in-leakage so that you do
8 provide a basis for your in-leakage assumptions. The basic
9 baseline test attributes, as you can seen in the slide
10 there, is the tests, whatever test method you use, it has to
11 be comprehensive, in other words, the value you get has to
12 have tested everything on that envelope boundary. It should
13 be done under accident configuration lineups, that is an
14 important aspect. And it should be in accordance with the
15 recognized test standards.
16 Now, acceptable baseline test methods that we
17 explored, or that we looked at, are the integrated tracer
18 gas test method has been the method used to date by
19 licensees, and that is certainly a recognized baseline test
20 method. But the task force looked at, are there other
21 options that licensees might be able to use other than the
22 tracer gas test? And one of these that we have developed is
23 the component test method, which I will have some slides
24 that will go into some more detail on what that test
25 consists of.
. 256
1 And the document does discuss, it leaves the
2 option open also for licensees to develop any other
3 alternate test method, as long as they have made the
4 baseline test attributes that are in the document. Go to
5 the next slide.
6 Okay. As I said, the tracer gas test method, I
7 think you are all aware, has been used in the industry. Its
8 standard is ASTME 741, which I am sure you also know. We
9 believe this test method is based on sound principles, and
10 it has improved control room integrity for those plants that
11 have used it. It is a proven test method used in other
12 industries. Again, I am sure most of you know that. So it
13 is a good method.
14 However, it has been expensive for licensees to
15 perform this test, and it has been beyond the capability of
16 the plant staff to perform the test. All licensees to date
17 have had to use contractor support.
18 However, this is a valid method for control room
19 designs. And we recommend it for a number of control room
20 designs, particularly the non-pressurized control rooms.
21 Now, there are various factors that affect the
22 accuracy, and these are just some of them listed here. You
23 have to make sure that your tracer gas test is uniform
24 throughout the volume. For some of the test methods you
25 have to know, you have to have an accurate measure of your
. 257
1 control room volume, and environmental effects can affect
2 this test as well as pressurization flows, particularly if
3 your plant is pressurized.
4 DR. KRESS: For the non-pressurized, you recommend
5 it for non-pressurized which means for pressurized ones, you
6 are not recommending it.
7 MR. TAPLETT: Pressurized control rooms that meet
8 certain aspects, and my next slide will talk about what
9 other things you want to consider besides just being a
10 pressurized control room.
11 DR. KRESS: Yeah. Well, my question is, could it
12 be used for pressurized, you are just not recommending it?
13 MR. TAPLETT: Oh, yes, it could be used. In fact,
14 there have been a number of pressurized control rooms that
15 have used it.
16 MR. COZENS: This is Kurt Cozens. I might
17 emphasize that what we are saying by this slide is that
18 where it is the recommended method for a non-pressurized
19 control room, for a pressurized control room, we consider
20 that there are options that might be chosen, particularly if
21 you meet certain conditions. And so that is the caveat that
22 we are trying to express in this slide.
23 DR. KRESS: Okay.
24 MR. TAPLETT: The task force looked at, you know,
25 could we develop a different test method that might be less
. 258
1 costly, a straightforward method that plant staffs -- within
2 the capability of plant staffs to use. And the idea behind
3 this is if you can do a plant assessment, a thorough plant
4 assessment of your plant, and you can demonstrate that your
5 control room design is rigorous such that you have few
6 vulnerabilities to in-leakage, and then if you can measure
7 those components that are vulnerable to in-leakage and
8 quantify that, then we believe that you should be confident
9 that you have demonstrated your total in-leakage.
10 DR. KRESS: Basically, about how many components
11 are we talking about?
12 MR. TAPLETT: It is plant-specific. I will tell
13 you we have done assessments on five plants. For one
14 example, one plant it will be three different -- I guess you
15 start adding up dampers, I recall something like five or six
16 dampers and then a duct, so that adds up to about five
17 components.
18 DR. KRESS: So it not a lot of components?
19 MR. TAPLETT: It is not a lot of components. In
20 fact, if you were going to test a lot of components, we
21 would recommend, just from an economic standpoint, it is
22 probably wiser to go with tracer gas, use tracer gas.
23 DR. KRESS: At what point do you feel like there
24 is enough components that you can't be certain you have
25 identified all of them that might be leakage sources?
. 259
1 MR. TAPLETT: I don't think the number of
2 vulnerable components is an issue. I think you can identify
3 them all.
4 DR. KRESS: You can, you think you can identify
5 them all.
6 MR. TAPLETT: I think the issue is if you are
7 testing a large number of components, each of those tests
8 costs a lot of money.
9 DR. KRESS: So it is an economic decision then.
10 MR. TAPLETT: Yeah, maybe just easier -- make more
11 sense to a tracer gas test.
12 MR. LEITCH: You don't consider cable penetrations
13 components? I mean there must be hundreds.
14 MR. TAPLETT: Let me discuss how we handle that
15 when I describe the test.
16 MR. LEITCH: Okay.
17 MR. TAPLETT: Because, yes, we are considering
18 cable penetrations.
19 MR. LEITCH: And what about doors into the control
20 room?
21 MR. TAPLETT: The same thing. I will show how we
22 consider those.
23 Okay. The procedure is fairly straightforward,
24 but it has to be performed rigorously and it has to be
25 comprehensive. The procedure, the first thing you need to
. 260
1 do, and this test method, again, as we said, is only
2 applicable to positive pressure control rooms, or
3 pressurized control rooms. You have to demonstrate that
4 your control room is at a positive pressure with respect to
5 all adjacent areas, and that has to be a thorough
6 measurement.
7 Now, if you have got a control room with several
8 spaces, you need to be very thorough in each space, make
9 sure that any -- and when we talk about cable penetrations,
10 we talk door penetrations. We have to make sure that we
11 have taken a pressure -- we have measured pressure in a
12 sufficient number of places to ensure that each door and
13 each cable is represented by a pressure reading. If we can
14 do that, and it is positive with respect to the outside
15 space, then even if the penetration leaks, it would be
16 out-leakage. So we wouldn't actually measure the specific
17 leakage through all those penetrations, we would just prove
18 that it would be out-leakage by this positive pressure
19 measurement.
20 Now, the second step though is not every place
21 within your control room envelope is positive with respect
22 to adjacent areas. And principally this is within your
23 ventilation systems themselves. You may have dampers that
24 realign when you go into the emergency mode. You may have a
25 duct from another -- a ventilation duct from another
. 261
1 non-control room ventilation system that may remain
2 operational during the accident condition, it may be a
3 higher pressure. Well, any of these conditions, you have to
4 then, since you don't have the favorable differential
5 pressure condition, you have to measure these components,
6 and these are the components we are talking about.
7 Plant staffs can look at their design drawings and
8 can verify their design with field walkdown, and can
9 identify these components.
10 Now, just for instance, again, the key is the
11 control room has to be positive with respect to all adjacent
12 areas. And as we discussed, we think this is a method that
13 is only going to be useful for those plants that have a
14 small number of vulnerabilities, have a rigorous design,
15 have good boundary construction, good ventilation system
16 construction that would not be susceptible to leakage.
17 DR. KRESS: Now, when you measure the leakage of
18 one of the components that you feel needs to be tested, how
19 do you make that measurement?
20 MR. TAPLETT: How do you make that? Well, some of
21 the standards we use, there is ANS-510, which is a direct
22 pressurization method, or you can use a pressure to K
23 method.
24 DR. KRESS: Build a little chamber around the
25 thing?
. 262
1 MR. TAPLETT: But, basically, you pressurize one
2 side of it and then you measure.
3 DR. KRESS: Both sides.
4 MR. TAPLETT: Measure if there is any leakage on
5 the other side. And you have got to set the right -- you
6 have got to have the same differential pressure condition
7 you would have during accident conditions.
8 DR. KRESS: Well, how do you measure whether there
9 is leakage on the other side, use smoke?
10 MR. TAPLETT: You set up another boundary and you
11 have got -- either you can measure, you pressurize it up to
12 a certain pressure and then hold it there.
13 DR. KRESS: And then watch it decay?
14 MR. TAPLETT: And then if it decays, then there is
15 a calculation to determine your flow.
16 DR. KRESS: Okay. Watch the decay in pressure.
17 MR. TAPLETT: Or you could actually -- you can
18 actually have an instrument, how much flow you need to keep
19 to keep the pressure constant, how much flow is required.
20 So there is different, and those are, you know.
21 DR. KRESS: Okay.
22 MR. TAPLETT: We think this is an acceptable
23 method for baseline tests, remembering it is only applicable
24 to certain plants, because it is comprehensive in that you
25 are measuring everything. You are measuring everything
. 263
1 across your control room boundary, either by a positive
2 pressure test, or where you don't have those favorable
3 conditions, you would use a component test.
4 Plants, and it is within, you know, the capability
5 of staff to determine where those vulnerabilities are. We
6 will do the positive pressure tests and we will do the
7 component tests under the same flow, or airflow and pressure
8 conditions that you would have during the accident
9 condition. And there are standards available for these
10 component tests. Licensees have done duct testing, have
11 done damper testing, so there are standards available.
12 DR. KRESS: Well, let me ask you one other
13 question I have. Has anybody at a specific plant, for
14 example, done a component testing and a tracer gas testing
15 and compared the two results?
16 MR. TAPLETT: That has not been done yet. That is
17 the next step we are looking at and getting the plant to do
18 that so we can get some correlation.
19 DR. KRESS: Sort of a verification.
20 MR. TAPLETT: Right.
21 DR. KRESS: Okay.
22 CHAIRMAN POWERS: Tom, would one particular
23 facility be enough?
24 DR. KRESS: Well, that is always a question. That
25 is one sample point.
. 264
1 CHAIRMAN POWERS: Better than none.
2 DR. KRESS: It is better than none, of course.
3 But, yeah, you know, I don't know how you decide how much is
4 enough on something like that.
5 DR. SHACK: Again, how are you confident you have
6 identified all the potential in-leakage pathways?
7 MR. TAPLETT: Well, we believe, you know, you can
8 look at your design drawings. I mean plant staffs can look
9 at their design drawings and they can go out and verify it
10 during field walkdown. And also based on industry
11 experience in doing this kind of stuff, we have come up with
12 a number, you know, we have identified where all these
13 possible vulnerabilities would be. We believe you should be
14 able to identify everything. It is, you know, it is really
15 design drawing review, understanding your system, and plant
16 engineers and plant staff know that, or can do that.
17 DR. SHACK: I guess they were designed to be
18 leak-tight initially, or at least reduced infiltration so
19 that there is -- I mean there is construction details so
20 that there really are a limited number of penetrations?
21 MR. COZENS: This is Kurt Cozens. Let me help you
22 here just a little bit. The first aspect is to be assured
23 that the control room has positive pressure to all adjacent
24 areas. That eliminates all standard things like floors,
25 walls, ceilings, penetrations through the walls, door seals,
. 265
1 things of this nature, that frequently are kind of difficult
2 to quantify individually.
3 So by a more bulk test you are able to demonstrate
4 with many measurements of delta P that that volume is
5 actually a positive pressure and leakage is outward. That
6 eliminates from consideration many typical sources that one
7 might think of. And we believe that, you know, we have seen
8 some in the past, you know, this number of measurements, and
9 really looking for this probably has not been done.
10 But then you go to the next phase. Okay. If all
11 my leakage is outward, is there any potential for other
12 in-leakages? So you have to say, okay, do I have a set of
13 ducting that is more positive, is higher pressure than the
14 control room itself, that should be leaking, that might, or
15 you have got some ducting?
16 Am I capturing this right, Ken? And so that is
17 the concept that first you do the bulk and then you look for
18 the exceptions. And those exceptions, by walkdowns and
19 draw-in examination, we believe you can identify all of
20 them, because there should not be that many. They are
21 physical features.
22 MR. LEITCH: But then if I understand it
23 correctly, then a tracer test would not necessarily equal --
24 in other words, the sum of all the leakages in a component
25 test would not equal a tracer test, because the component
. 266
1 that is the tracer would also measure those positive things
2 where there was positive pressure, it was all excluded.
3 MR. CAMPBELL: The tracer gas test is going to
4 measure all the air that is coming into the boundary.
5 MR. LEITCH: Right.
6 MR. CAMPBELL: And so it is really not going to be
7 calculating what is leaving, it is going to be calculating
8 what is coming in. So, again, the tracer gas test would
9 essentially be -- and one way of looking at it would be
10 essentially doing what you are doing in the positive
11 pressure test. You are eliminating everything that is
12 leakage out and you are only measuring stuff that is leaking
13 in.
14 I mean that is not an exact picture, but that is
15 kind of like what we are talking about here. We are
16 eliminating everything that is leaking out of the system. A
17 tracer gas test, or an integrated one will measure
18 everything that is coming into the boundary. And so by
19 doing a component test, we are eliminating everything that
20 is leaking out and then we find the things that can be
21 leaking in and we measure those things that could be leaking
22 in. And then we add all that up together and that gives us
23 a number.
24 DR. KRESS: One of the attractive things you said
25 about component testing is that could be under the complete
. 267
1 control of the licensee without having to call in
2 consultants and a contractor to do it. So they could have
3 -- they have their own flexibility of how often to do it and
4 how to carry it out, and how complete it is and things like
5 that.
6 MR. TAPLETT: This is Ken Taplett again. Yeah,
7 these tests have been performed, these are the types of
8 tests that have been performed by licensees. You know, they
9 have done -- people have experience doing duct testing,
10 damper testing, this type of thing, so it is not a new test
11 methodology. It has never been done in a comprehensive
12 standpoint to determine total in-leakage, but these type of
13 tests have been performed.
14 Again, a strength of this test is it recommends
15 for positive pressure control rooms with few in-leakage
16 pathways. What plant test results have indicated is for
17 plants that have large pressurization airflow, this is the
18 makeup airflow to give you your positive pressure. This can
19 leak to nick uncertainties in the unfiltered in-leakage
20 value that the tracer gas results give you.
21 And we have just shown here, looking at plant test
22 results for a neutral control room, which we certainly
23 recommend tracer gas as a good method, the uncertainties are
24 less than 10 percent. But for pressurized control rooms
25 they have ranged from 30 to 60 percent.
. 268
1 DR. KRESS: How were these uncertainties ranges
2 determined? I mean where did you get these numbers?
3 MR. CAMPBELL: This is Robert Campbell with TVA.
4 What we have here, and I think Pete Lagus can -- he is going
5 to talk later on this afternoon, and he can certainly
6 elaborate on it. But what we have seen is that errors are
7 associated because there are actually a couple of numbers
8 calculated. And so what we do is we will calculate the
9 total volume of air coming into the control room and we
10 subtract off of that the pressurizing airflow. And the
11 errors associated with both those numbers are evaluated and
12 applied to the difference between those two numbers.
13 So what you end up with is even though you may run
14 a very accurate test, because you are getting down to the
15 very fine points of it, that error becomes large and
16 associated with this smaller number. So that is why when
17 you do a control room pressurized test, or a test on a
18 pressurized control room, you will see these larger ranges
19 of uncertainties, I will call them.
20 DR. KRESS: It is a classic difference between two
21 large numbers to get a small one?
22 MR. CAMPBELL: Yes. And I am sure Pete can refine
23 that later on.
24 MR. TAPLETT: Okay. So with that in mind on the
25 test results we have seen, we think the benefit of the
. 269
1 component test method -- we have said this already, but the
2 component test method is a method that is -- these methods
3 have been routinely performed by the plant staff in that
4 these types of tests have been performed, not in a
5 comprehensive standpoint that we are suggesting here, but
6 they have been performed. And the measurement of
7 uncertainty of an individual component test is less than 5
8 percent. So you start --
9 DR. KRESS: That is on a specific component?
10 MR. TAPLETT: A specific component.
11 DR. KRESS: Component, okay. So if you had 10
12 components, you would have a 50 percent error maybe?
13 MR. TAPLETT: I don't think it would add up that
14 way.
15 DR. KRESS: It is a plus or minus.
16 MR. TAPLETT: I think you would still end up
17 finding yourself somewhere within plus or minus 5 percent.
18 CHAIRMAN POWERS: When you say plus or minus 5
19 percent, is that the precision of the measurement?
20 MR. TAPLETT: Yeah, the precision of the
21 instruments. Most of the instrument precision is down in
22 the range of plus or minus 2 percent, and so it is
23 probably --
24 CHAIRMAN POWERS: What I am trying to figure out
25 is how did you know that the measurement was accurate to any
. 270
1 percentage at all. You must have had something else that
2 measured it something. Was it a calibrated component that
3 had a known leakage? I mean how did you know what the
4 accuracy was?
5 MR. TAPLETT: Well, what we base this on is the
6 precision of the instrument.
7 CHAIRMAN POWERS: Okay. So it is really a
8 precision and not an accuracy.
9 DR. KRESS: It is the precision of the instrument
10 as applied to the calculational methodology, I think.
11 MR. TAPLETT: It is a direct measurement of the
12 component, but the numbers that come from the precision of
13 the instrument.
14 So, in conclusion, basically, some licensees do
15 have control room designs that they should be able to take
16 an option, or should be able to use this test method as an
17 option, and should have confidence it is going to give them
18 a reasonable -- or is going to give them a total in-leakage
19 number that should give them a reasonable assurance that
20 they have defined what their in-leakage is and can use this
21 in the accident analysis.
22 I'll pass to Bob Campbell, who is going to talk
23 about the control room habitability program.
24 MR. CAMPBELL: This is Bob Campbell, TVA. One of
25 the critical aspects of 99-03, at least in my opinion, is
. 271
1 that once we go through this process of establishing your
2 control room and doing your baseline tests, how do you keep
3 it there. And what we are saying is that each utility or
4 licensee should have a program established after they have
5 gone through this process.
6 Some of the elements of the program one is a
7 periodic assessment of their control room habitability.
8 Part of that is to look at the system material condition.
9 Are seals degrading? Are gaskets getting brittle? Are
10 there any holes being developed in your system due to
11 construction or maintenance?
12 If you look at in-leakage challenges, some of the
13 factors are, again, construction. Are you putting in new
14 conduits? Are you making some modifications to the plant
15 where you are poking holes in?
16 The next thing is toxic gas challenges. New
17 chemicals being brought onsite, increasing the quantity of
18 chemicals stored onsite.
19 DR. KRESS: When you say periodic assessment of
20 those things, how do you determine the frequency of the
21 assessment? Do you have some guidance on that?
22 MR. CAMPBELL: Right now the document is pretty
23 loose in that area. And when we talked at the subcommittee
24 meeting, I believe we have made a commitment to go back and
25 make that more robust.
. 272
1 I will let Kurt answer that.
2 MR. COZENS: We have realized that there are
3 factors that should be considered when any plant looks at
4 its plant-specific situation, things considering such as the
5 margin between what you have measured and what you can
6 operate at. There are the differences of what is going on
7 in your plant. And we acknowledge that these are
8 plant-individual features that should be considered when
9 making a decision. We do not believe that one schedule for
10 all plants to follow is probably the appropriate answer.
11 But what we have not done to date is provide more
12 robust guidance on how to consider those factors as they
13 might apply to an individual plant. We don't have a
14 complete answer on that yet, but we are going back to
15 consider that further and we are working on that presently.
16 DR. KRESS: When you arrive at some more robust
17 guidance on that, it is likely to end up being
18 plant-specific?
19 MR. COZENS: It will be plant-specific.
20 DR. KRESS: Would it be a subject for performance
21 measurements like Appendix J for containment leakage? You
22 know, you could have, depending on the performance of the
23 individual plant, you could maybe extend the periodic tests
24 and something like that.
25 MR. COZENS: Possibly. It has been discussed, but
. 273
1 I honestly don't know what the answer is going to be at this
2 point in time. That has been definitely discussed.
3 DR. KRESS: How would this show up as a regulatory
4 commitment if it were, you know, if NRC says, yeah, that is
5 the way to go, we like that? Would it be a volunteer
6 program or would it be --
7 MR. COZENS: We have basically proposed that
8 licensees make a commitment to staff to follow the program
9 that we just discussed. It would be an Appendix B type
10 program.
11 DR. KRESS: Okay.
12 MR. CAMPBELL: Continuing on, another aspect of
13 this program would be configuration control. Again, they
14 are what we call a barrier control program, which is
15 basically breaching. For example, maintenance has to be
16 done on a door, or a damper or a duct, or conduit, so there
17 would be controls put in place so that the hole that is
18 needed is limited in size, that there are controls put in
19 place to assure that that hole is sealed back to as-required
20 condition, that there is a PMT that is put in there to
21 verify that that particular aspect has not been degraded.
22 Procedure controls identifying control room, that
23 control items are taken care of within plant procedures,
24 operating procedures, maintenance procedures, that there is
25 a consideration of this.
. 274
1 Design change control, that in going through the
2 design change process, plants will have a requirement to
3 assess the impact on the control room habitability within
4 that design change and make sure that there are aspects
5 within that design change that will not invalidate any
6 control room requirements. And analysis change control
7 similar to the design change control process, any analysis
8 changes are looked at within -- for effects on control room.
9 And then the last aspect of this will be training.
10 Make sure that the plant staff is trained in these areas,
11 that operators are trained, engineering staff, maintenance
12 staff.
13 Factors that affect the assessment or test
14 frequency, we are not ruling out tests, but we prefer to go
15 through an assessment process. And, again, we are saying
16 that we are going to make that robust. But some of the
17 things that go into that are the number of in-leakage
18 sources. If a plant has, say, 2,000 feet of ducting that is
19 outside the control room, obviously, it is going to do a lot
20 more work than a plant that doesn't have anything inside the
21 boundary. So that is what we mean under there, that is an
22 example.
23 Differential pressure margins, that the plant has
24 a requirement to maintain 1/8th of a inch water pressure to
25 all adjacent spaces. And if it is maintaining 6/10ths of an
. 275
1 inch of water to adjacent spaces, we know we probably are
2 not leaking through the walls or the doors, or the floors or
3 the ceilings. The margin of the measured baseline value to
4 the allowed value that is in the design analysis. If I have
5 an analysis that says I am allowed 500 CFM, and I go out
6 there and measure 100 CFM, again, that is a factor is in the
7 utility's favor.
8 The factor that goes against the utility, if they
9 measure 200 and they are allowed 200. They may end up
10 having to do a test on a more frequent basis, like every
11 refueling outage.
12 DR. KRESS: Give me a better idea of what an
13 assessment is, as contrasted to a test.
14 MR. CAMPBELL: The assessment, as I see it, would
15 be the 99-03 process, whatever frequency the plant would
16 choose, they would go back through the 99-03 process. They
17 would look at their boundary, they would do walkdowns. They
18 would look at degradations to the system. And then they
19 would say, is there something there that tells me that I no
20 longer have a reasonable assurance that my control room is
21 adequate? And then, what do I need to get that assurance
22 back? And that is what I see, I see 99-03 as the
23 assessment.
24 DR. KRESS: Sometimes it may be another component
25 test.
. 276
1 MR. CAMPBELL: It may be a another component test.
2 It may end up being an integrated tracer gas test. We don't
3 know because, again, each plant is different, and what we
4 have come to find out, this is very plant-specific, at least
5 that is my view.
6 And then testing when appropriate. I am not
7 ruling out tests, and I believe that when plants go through
8 this process, they will be doing tests, and there will be
9 periodic tests when they go through this. But that is yet
10 to be seen since we haven't done it yet, but that is my
11 expectations.
12 MR. LEITCH: Would part of that be consideration
13 of new sources of toxic gas?
14 MR. CAMPBELL: Yes.
15 MR. LEITCH: From, you mentioned in-plant, but I
16 am thinking about outside of plant, like transportation
17 accidents, nearby chemical facilities and so forth. Is that
18 things that --
19 MR. CAMPBELL: Yes, that is correct. We have -- I
20 believe the document says right now that a plant, as a
21 minimum, or we recommend as a minimum, that if you are in an
22 urban area, you reassess that every five years, or if there
23 is any new construction, new facilities being built in the
24 area, if there is something that has changed, the plant is
25 obligated to go back and do it. But they should do it no
. 277
1 less than five years if they are in an urban area, no less
2 than 10 years if they are in a rural area.
3 Control room some infiltration. What the 99-03
4 document is looking at under this item is a qualitative
5 assessment of how the plant will respond to smoke and assure
6 that the plant can be maintained safe. What you see here is
7 a flow diagram that gives us a success path for dealing with
8 smoke.
9 But prior to entering this flow diagram, there are
10 certain things that 99-03 recommends. One is that remote
11 shutdown panels, or what we call backup control room, are
12 located outside the control room envelope itself. Verify
13 that one event, one fire or smoke event will not take out
14 both the backup control room and the main control room.
15 Verify that there is paths available for the operators to go
16 from the main control room to the backup control room in the
17 event that they have to evacuate. And it goes on, several
18 things like that, scubas, training, that type of stuff.
19 So once we have established all those things, then
20 we get into -- how do we handle if smoke does infiltrate?
21 We say, is the ventilation system the source of the smoke
22 coming in? If it is, then you shut it off, and then
23 everybody is happy. And we go down through the flowchart.
24 Essentially, this is a path to assure that the control room
25 operators can stay within the control room, or, if not, that
. 278
1 they have a path to get to the backup control room and be
2 able to man the station in a safe condition from the backup
3 control room.
4 DR. KRESS: Now, the HVAC system and the
5 ventilation system is there to keep the room at an
6 appropriate temperature for both the operator and the
7 electronics, I presume. Is that a problem if you do this
8 step?
9 MR. CAMPBELL: That would be a problem if the air
10 conditioning system was the source of the smoke.
11 DR. KRESS: Yeah.
12 MR. CAMPBELL: If it is a pressurizing system, a
13 lot of control rooms have ventilation air, you know, just
14 for breathing and stuff. If it is the ventilation system,
15 no, that wouldn't be.
16 DR. KRESS: That wouldn't be a problem there.
17 MR. CAMPBELL: But if it is the air conditioning
18 system that is bringing in the smoke, for example, if I have
19 a duct that is going through an area of fire, then, yes,
20 that could be a source and you wouldn't be able to isolate
21 it. But what I am seeing both in practical experience and
22 in analysis space is that that is not an immediate problem,
23 it takes times.
24 DR. KRESS: It takes a while for things, yeah.
25 MR. CAMPBELL: It takes a while. So it would give
. 279
1 the operators time to safely control the shutdown should
2 they need to leave the control room.
3 My personal opinion is that probably, we probably
4 get the fire under control before you would ever reach any
5 temperature limits.
6 MR. LEITCH: If you substituted toxic gas
7 infiltration for smoke infiltration, it would seem to me
8 this logic diagram would be quite similar.
9 MR. CAMPBELL: It would be, but right now the way
10 the plants are designed, in accordance with the Reg. Guides
11 1.78 and 1.95 that Kurt mentioned, is that we look at toxic
12 gas sources and we calculate what the concentration of the
13 toxic materials are in the control room, and we take
14 appropriate steps to protect them from that.
15 This would be as if the toxic material got into
16 the control room, and we are not allowing toxic materials in
17 the control room.
18 MR. SCHULTZ: I am Steve Schultz from Duke Energy
19 and I am here to describe today the activities of the
20 analysis subgroup of the 99-03 effort. If you go back about
21 12 months at the initiation of this effort, it was
22 determined that there were several areas that were
23 worthwhile examining, and both the staff and industry
24 determined that, given, as Kurt had mentioned earlier, what
25 was happening with regard to alternative source term
. 280
1 developments, that we ought to take a look at this with
2 respect to how it would impact control room habitability
3 analysis.
4 And as a result, we formed an industry subgroup
5 that included eight to 10 members who met on about a once a
6 month frequency with the staff for one or two days with the
7 staff, one or two days on our own, to examine different
8 approaches that could be taken in the analysis efforts for
9 determining control room dose.
10 DR. KRESS: When you do this, do you take into
11 your thinking that for those plants that have filtered
12 in-leakage, the effect that the aerosols may have on your
13 filter? Is that part of the thinking when you talk about an
14 alternative source term?
15 MR. SCHULTZ: It is part of the thinking that goes
16 into the overall evaluation, but it is not a portion of the
17 particular dose evaluation that we were discussing.
18 The point I am trying to make here is that we had
19 a couple of paths that had been identified. The staff was
20 moving toward development of Reg. Guide 1.183, which
21 prescribes methodologies for utilizing the alternative
22 source term in calculations. We wanted to examine elements
23 of that to see what we would -- how we might modify
24 calculations that used the TID source term as a basis.
25 With regard to that, there were two efforts there,
. 281
1 one associated with calculating control room dose under new
2 measurements of unfiltered in-leakage, and second was
3 calculational, that we were first examining a calculational
4 methodology that might disposition results from unfiltered
5 in-leakage evaluations. That is looking at whether we
6 should develop operability evaluation methodology.
7 We determined, after looking at that for a few
8 months, that a better approach would be to focus on
9 licensing method development and assume that if a plant
10 became -- entered a condition where their unfiltered
11 in-leakage value was greater than what had been assumed in
12 previous analyses, that potassium iodide or self-contained
13 breathing apparatus could be used as compensating measures
14 on an interim basis. And that is a section that you will
15 find in the 99-03 document.
16 We also examined, besides the dose evaluation
17 calculations, quite a bit of the technology and dispersion
18 modeling for control room habitability calculations. And I
19 will get into that in more detail in a few minutes.
20 But we had principally agreed that an effective
21 approach could be using the alternative source term with a 5
22 rem TEDE limit. That with a TID source term, a 50 rem
23 thyroid limit could be utilized in place of the standard
24 6.4.2 30 rem guideline.
25 On the next slide, I show the highlights of some
. 282
1 of the changes that have been developed in this work. Some
2 have been agreed to as part of the document by the NRC
3 staff. Some are still under review by the staff. With
4 respect to the fuel handling accident, we take from the
5 alternative -- from the Reg. Guide 1.183 the revisions to
6 the isotopic gap fractions and the decontamination factor of
7 200 per 23 feet of water that was derived as part of the
8 work for that document.
9 Additionally, for the steam generator tube
10 rupture, a spiking factor of 335 is adopted versus the
11 spiking factor of 500, which has been used in the past.
12 For steam generator tube rupture and main
13 steamline break, again, the spike duration of less than
14 eight hours has been examined. We proposed to the staff
15 that for a steam generator tube rupture, a spike duration of
16 four hours could be justified, for main steamline break a
17 spike duration on the order of three hours could be
18 justified. That position paper is still under review at
19 this point in time.
20 Other elements associated with the design basis
21 LOCA, which we take from the revised Regulatory Guide, is
22 the elimination of the ECCS passive failure for events where
23 previously certain plants needed to assume a 50 GPM leak for
24 30 minutes, at 24 hours after the LOCA, and for BWRs, the
25 Reg. Guide 1.183 now endorses a reduced leak rate for times
. 283
1 greater than 24 hours at 50 percent of the initial rate, and
2 that has been adopted here.
3 We also have under review by the Reactor Systems
4 Group at the staff, a position on enhanced mixing,
5 containment mixing rates. In Reg. Guide 1.183, the position
6 is that an exchange rate of two per hour can be justified,
7 but if additional mixing can be justified, credit can be
8 taken. So we have proposed a methodology by which licensees
9 could perform that evaluation.
10 CHAIRMAN POWERS: I noticed in the notes from your
11 subcommittee meeting that you had asked for more information
12 on this spiking factor of 235. Did you get additional
13 information?
14 DR. KRESS: I have not received anything yet, no.
15 CHAIRMAN POWERS: Do we have any idea where this
16 number came from?
17 DR. KRESS: No, we don't.
18 CHAIRMAN POWERS: This is a number that appeared
19 out of space?
20 MR. SCHULTZ: No. Let me speak to that. The
21 original document that we referenced -- well, there are two
22 references to it. Of course, it is quoted in the Reg. Guide
23 1.183, but we have looked at what we believe to be the
24 source for that, which is the data set that was developed at
25 INEL by Adams and Atwood. And additional work has been
. 284
1 done, or the data set now exists at the Pacific Northwest
2 Laboratory.
3 CHAIRMAN POWERS: Now, if I recall the Adams and
4 Atwood paper, they don't quote spiking factors. He didn't
5 even believe in spiking factors, he didn't even like them.
6 It was another way of --
7 MR. SCHULTZ: He was proposing an alternative
8 method of addressing it. And he suggested that his
9 alternative method would result in substantially lower
10 results than if one had assumed a spiking factor of 500 or
11 335.
12 If we look at -- what that was is a set of data
13 that looked at the pre-trip activity level, coolant activity
14 level, dose equivalent iodine, and then looked at the
15 post-trip value, and assembled about 150 data points as a
16 function of the pre-trip value.
17 What we have done is examined that data set and
18 made assumptions about how one might apply that generically,
19 looking at letdown flow, looking at iodine decay, and
20 looking at the variability that one might assume could be
21 from beginning of the event to the peak of the spike. Then
22 we took a 95 percent, looked at the data set of all those
23 peaking factors and taken a 95 percent value of that data
24 set.
25 CHAIRMAN POWERS: If I recall the data set
. 285
1 properly, and if one casts the numbers in terms of spiking
2 factor, which Adams and Atwood don't, there are spiking
3 factors, one of them is as high a 11,000. So how does that
4 compare to what you call the 95 percent here?
5 MR. SCHULTZ: They did quote a value or a number
6 that they did not call spiking factor. But this is a
7 portrayal of their data, and I have labeled it, in the
8 handout I provide a draft because it is a portrayal and
9 these data points are what I would say very close to what
10 they represent in the paper, but they are our interpretation
11 of the data set. We have worked with the data set from
12 them.
13 The values that you are talking about are the few
14 values that are at very low initial iodine concentrations,
15 10 to the minus 3 and the last initial concentrations. And
16 in the accident evaluations, we are assuming that we are at
17 the tech spec value of the initial accident as of the tech
18 spec value of concentration, which is generally 1 microcurie
19 per gram, off the scale out this way.
20 CHAIRMAN POWERS: I mean these are actual physical
21 measurements. I mean it is not out of the hypothesis. And
22 it says, again, where it goes up, as the concentration goes
23 down. So why wouldn't you take the number down here where
24 you really are? Why would you go up and take the tech spec
25 value?
. 286
1 MR. SCHULTZ: Could you repeat the question?
2 CHAIRMAN POWERS: Well, you have said, oh, gee, my
3 tech spec value for my liquid concentration is 1 microcurie
4 per gram. So let me take the spiking factor up there.
5 Well, the spiking factor is lower up there. Okay.
6 Your plant really operates probably less than a
7 tenth, probably less than a hundredth of a microcurie per
8 gram. Why wouldn't you take something that is more
9 realistic down there? A spiking factor associated with
10 those kinds of environments?
11 MR. SCHULTZ: Well, you are assuming in the event
12 that you are at a 1 microcurie per gram concentration in the
13 coolant.
14 CHAIRMAN POWERS: Why would I do that? I am
15 never, ever in this plant at 1 microcurie per --
16 MR. SCHULTZ: Well, one could change the analysis
17 technique and assume a lower initial value. But that is not
18 what is done in the analysis methodology, 1 microcurie per
19 gram is assumed.
20 CHAIRMAN POWERS: Yeah, and what I am trying to
21 understand is why. I mean the reality is here you have got
22 a measurement that says spiking factor goes up as the
23 concentration goes down apparently. I usually operate at a
24 lower value. I am never -- I have met a plant up around 1
25 microcurie per gram in the last 20 years, I bet. I am
. 287
1 usually down here where these low numbers are. Why wouldn't
2 I take a realistic number?
3 Why would I do something imaginative that seems
4 non-conservative? It doesn't even seem -- it is neither
5 realistic nor conservative.
6 MR. SCHULTZ: Well, I understand the dilemma, but
7 as we apply the analysis now, the assumption that industry
8 has used and the staff has agreed with is that you use the
9 tech spec value or you assume the tech spec value of the
10 dose equivalent iodine coolant concentration.
11 DR. KRESS: Is that because of the fact that you
12 have that tech spec value, that there might be some point in
13 your operation you might approach that tech spec value and
14 then have the accident, or is that the assumption?
15 MR. SCHULTZ: Yes. The assumption in the
16 licensing analysis is that since the plant could be in that
17 condition, one should assume that value in the design basis
18 analysis.
19 DR. KRESS: It is a design basis concept.
20 CHAIRMAN POWERS: Well, if we assume that, I still
21 don't see how we get 335.
22 DR. KRESS: Well, I am not quite sure how we do
23 either.
24 MR. SCHULTZ: What's that?
25 CHAIRMAN POWERS: I still don't understand where
. 288
1 335 comes from.
2 DR. KRESS: I can't get it off of that plot.
3 MR. SCHULTZ: Well, you can't. You can't see it
4 from the data, but if you take all of the data, all of the
5 factors that are calculated, and take 95 percent of them,
6 they will be bounded by 335. This is a plot that starts the
7 data set at 5 times 10 to the minus 3, so it leaves off
8 those data points that are very high. But in any case, the
9 whole data set was used in getting to the 335 value.
10 DR. KRESS: Now, is that the data set that was --
11 that the authors recommended they multiply them by 3 because
12 of the time delays in getting the data points?
13 MR. SCHULTZ: Yes, it was.
14 DR. KRESS: Did you multiply it by 3?
15 MR. SCHULTZ: And that is why, if you look -- I
16 should have explained that earlier for those familiar with.
17 That is why the value is here. I did not multiply these
18 values by 3, so the high value here is 3,000 versus 10,000,
19 that Dr. Powers spoke to before.
20 DR. KRESS: I see.
21 CHAIRMAN POWERS: Well, why didn't you follow the
22 recommendation?
23 MR. SCHULTZ: Well, we did two things. We looked
24 at it from that perspective and we also expanded our
25 analysis to look at -- or develop a larger spreadsheet, I
. 289
1 guess I would say, to look at a spectrum of other potential
2 release rates -- excuse me, potential durations of the
3 spike, and essentially did a sensitivity study as a function
4 of initial, again, initial concentration in the coolant.
5 And in performing that evaluation, you can show
6 that for, again, the design basis licensing evaluations that
7 are performed, you are well bounding the data with this
8 combination of 1 microcurie per gram and a spiking factor of
9 335.
10 CHAIRMAN POWERS: George, you thought you were
11 having a hard time earlier today, I am really at sea here.
12 DR. KRESS: These parameters you are talking
13 about, they would be possibly used if the assist and
14 measured in-leakage were such that if you used the old
15 values, that you might exceed the dose limits, so that this
16 is a way to fine tune the calculation and still live with an
17 in-leakage that you may have measured?
18 MR. SCHULTZ: I think, if you go back
19 historically, the simple answer is yes. That there were two
20 areas that staff and industry wanted to look at. One was
21 the testing issue and it was clear from testing that has
22 been done, that unfiltered in-leakage values are higher than
23 have been assumed in the analysis.
24 Industry had countered that by saying, yes, but
25 the analysis is very conservative. And so the approach that
. 290
1 was taken was let's then improve the analysis and get to a
2 point where we have more realistic evaluations in concert
3 with the testing results.
4 Are there any other questions on that first slide?
5 DR. KRESS: I was wondering where your 200
6 decontamination factor came from, and was it strictly for
7 iodine as a gas or iodine as an aerosol?
8 MR. SCHULTZ: I believe the answer to the question
9 is yes. I am going to look to the audience for -- yes,
10 John.
11 MR. DUFFY: John Duffy, PSEG Nuclear, I am a
12 member of the Control Room Habitability Task Force. I
13 believe that 200 is an overall decontamination factor for
14 all iodine species.
15 Steve Levine, who is not in attendance, could
16 probably speak better to this issue, but this is the value
17 from the alternate source term Reg. Guide. And the
18 assumption is that all the particulates are scrubbed out and
19 there are two different decontamination factors. None of
20 the organic iodine is assumed to be scrubbed and the
21 partition factor for element iodine is on the order of 500.
22 So the 200 is an overall factor.
23 DR. KRESS: I thought you were applying this to
24 fuel handling accidents.
25 MR. SCHULTZ: That's correct.
. 291
1 MR. DUFFY: Yes. That's correct, that is the
2 context.
3 DR. KRESS: Where were your sources coming from,
4 the gap?
5 MR. DUFFY: Yes.
6 DR. KRESS: And I thought those values in the
7 alternative source term were for fission products that come
8 out of a melting core, transferred through the primary
9 system and in containment. There seems to be a mismatch
10 there. I am not sure the two are compatible.
11 MR. SCHULTZ: This is a separate evaluation as
12 part of the Regulatory Guide 1.183, but it entered into the
13 design basis evaluation for application of alternative
14 source terms. So it was additional information that was
15 added as part of that Regulatory Guide. Some of the other
16 elements fit into the same kind of category, they are not
17 directly related to the severe accident source term itself,
18 but they are improvements to analysis based on the enhanced
19 understanding.
20 DR. KRESS: Based on the insights.
21 MR. SCHULTZ: And insights. The other major topic
22 that I mentioned earlier is in the meteorology and the
23 dispersion models associated with the control room. And
24 what the subcommittee has been working with staff on is the
25 use of ARCON96 as a methodology for performing the chi over
. 292
1 Q calculations. It is an advanced modeling technique. It
2 is a critical component for the overall analysis because the
3 evaluation of chi over Q can be very sensitive for certain
4 sites.
5 However, an overall application of ARCON96, we
6 couldn't come to a conclusion with the staff as to how the
7 stack releases or high energy steamline valve releases could
8 be handled using the code methodology as it currently
9 exists. And so a program has been initiated by the NRC
10 staff to enhance ARCON96 to allow better modeling of both
11 those phenomena.
12 And this is what we can call a joint effort
13 between NRC and NEI. NRC is providing the contract for the
14 enhancement of ARCON96. We have been providing ourselves
15 technical inputs and also been looking for benchmarking
16 information that can be used with the code.
17 We have found very little information that is
18 available for this purpose and are looking for more. But we
19 have found some and are sharing that information with the
20 contractor.
21 DR. KRESS: With respect to the meteorology and
22 dispersion modeling, I seem to recall there was a question
23 about whether or not to consider sister plants that are
24 adjacent or nearby. Did that ever get discussed and ironed
25 out?
. 293
1 MR. COZENS: The way the 99-03 is prepared, it
2 directs licensees to consider all accidents that are in
3 their current licensing basis. If that is not under the
4 current licensing basis, we have not recommended to them
5 that they do that.
6 We have taken the position that that is not a
7 generic issue, that if it needs to be worked out, it should
8 be worked out on a plant-specific basis, because there are
9 uniquenesses to each.
10 MR. SCHULTZ: Are there any other questions on the
11 analysis approach?
12 [No response.]
13 MR. SCHULTZ: Thank you.
14 MR. COZENS: This is Kurt Cozens again. I would
15 like to discuss the technical specification issue. The
16 staff had recommended that we consider managing this program
17 that we have mentioned previously via a technical
18 specification. Having considered that, we looked at what
19 are the options that might exist, and these are listed on
20 the slide.
21 First of all, the status quo, make no changes to
22 what we have proposed already. The second position we
23 considered was a commitment to a control room program. That
24 is the present position which is in the document.
25 We have considered three different options that
. 294
1 involved the tech spec itself. One would be an
2 administrative program committing to -- a tech spec
3 administrative program committing to a control room
4 habitability program, but the program itself would not be
5 part of the tech spec.
6 The next one would be a moving control room
7 habitability HVAC tech spec requirement that would be placed
8 into a TRM technical requirements manual and adoption of a
9 periodic in-leakage assessment requirement through a CRH
10 program. This is an item that we have given a great deal of
11 thought to but have not adopted at this point in time.
12 The last item, which is I believe the approach the
13 staff has proposed to date, is the adoption of an in-leakage
14 surveillance in the current tech spec. This is an approach
15 that we have looked at, and we have looked at the criteria
16 and haven't felt to date that, when we read the
17 requirements, nor have looked at the examination of the
18 issue, that we have gone through about 25 percent of the
19 plants to date. This is not a shutdown issue. We don't
20 believe that this type of requirement necessarily should be
21 in the tech spec. We don't necessarily believe that it is.
22 DR. KRESS: What is the down side of putting it in
23 there?
24 MR. COZENS: It is our belief that the proposed
25 AOTs that the staff has proposed, which would be a 14-day
. 295
1 AOT, were likely will shut down not only one plant with the
2 control room, but with plants that have multiple resulting
3 shutdowns unnecessarily and that are not affecting the level
4 of safety. We are not disagreeing that we need to address
5 the issues at hand, that the plants need to have a program,
6 but that they do indeed -- you know, that is not the
7 appropriate regulatory methodology for that.
8 But after we had been through this, we had
9 committed to the staff to reassess this. Committing to --
10 our observation is committing to reassess control room
11 habitability, in-leakage periodically is the right thing to
12 do and it is recommended by NEI 99-03. I think that is one
13 thing we are talking technically here. We believe that is
14 the right step to take.
15 We believe that the current tech spec surveillance
16 adequately addresses the operability of the control room
17 habitability HVAC systems. We also believe the technical
18 specifications should focus on parameters and indications
19 observable and controllable by the operator. We do not
20 believe that this is one of those situations, that the
21 in-leakage is one of those features.
22 We believe that it is sufficient to maintain the
23 control room habitability in accordance with an Appendix B
24 criteria. But realizing that our particular task force is
25 more a technical group that has been wrestling with the many
. 296
1 aspects of this particular issue, and that there is another
2 group that deals regularly with tech spec issues, the
3 standardized tech spec, which would be the Tech Specs Task
4 Force, we have asked them to take on this issue of
5 addressing this and determining in their mind what do they
6 believe the appropriate steps are.
7 They have agreed to take this on. We do not have
8 a schedule on that, but at this point in time, that is about
9 all I can report on that. But we believe they are a better
10 body to deal with it, being that they are dealing with all
11 other tech spec modifications that are coming through the
12 system for the standardized tech spec.
13 DR. KRESS: Are there more questions of NEI before
14 we go to the next part of the agenda?
15 MR. COZENS: I have one more slide.
16 DR. KRESS: You have one more slide. I'm sorry.
17 Yeah, it is on the back.
18 MR. COZENS: I apologize for this running a little
19 longer than we had planned.
20 DR. KRESS: That is mostly our fault.
21 MR. COZENS: I would like to discuss our future
22 actions. But before I do, I want to recognize not only the
23 task force as a whole, but the industry as a whole for their
24 considerable contributions to defining what we presently
25 have in 99-03. It is we believe a substantial document that
. 297
1 has indeed captured many of the issues that the staff has
2 identified as a concern, and that we are trying to build a
3 consensus as we can to what is reasonable and acceptable for
4 industry to do. We believe that through their concerted
5 efforts we have defined a reasonable approach that gives
6 reasonable assurance that we can protect the operator.
7 As Dr. Kress had mentioned in his opening remarks,
8 there are several issues that are on the table that have not
9 been fully resolved. The first one is test options, control
10 room habitability versus tracer gas testing. Will control
11 room habitability be an acceptable option?
12 To support this, we have gone out to survey the
13 industry on more specifics about their -- we have gone out
14 to survey the plants that have tested with using the tracer
15 gas, to obtain more specific information. And we have
16 indeed used that partially in this presentation, as well as
17 some other features that we are still considering, trying to
18 graph a little bit more what the particulars of this fairly
19 complicated activity involve.
20 We are examining the tech spec issue, we are
21 looking at the need, what an appropriate AOT would be, or if
22 there should be one, and the test frequency of that. And as
23 I had mentioned earlier, we asked the Tech Spec Task Force
24 to assist us on that. And I believe the NRC has some
25 interaction with that group.
. 298
1 We have taken a consideration of the staff. The
2 staff has asked us to modify the document to effectively
3 make a common licensing basis for all licensees, whether not
4 considering what is in their present licensing, we are
5 examining the backfit considerations of that at this point.
6 We don't know where that will go at this point, we are just
7 examining it.
8 As Steve had reported in his discussion on the
9 evaluation methods, we are looking at ways to eliminate the
10 unnecessary and excessive conservatisms, and working with
11 staff to address all aspects of that, both conservatisms and
12 non-conservatisms. And we are hoping that those discussions
13 will continue, because we believe that those have been very
14 beneficial for both staff and industry.
15 At one point in time the staff had written the
16 industry, NEI specifically, and indicated that they would
17 not be providing us detailed comments on NEI 99-03. We have
18 gone back to the staff, asked them to reconsider that. They
19 have agreed to give us some feedback on this to highlight
20 the main issues, the basis for their concern, what is the
21 wrong the document, and why they think it is wrong. And
22 those are due to us, we understand, by December 31st.
23 With this document, we are planning to continue
24 our process to issue it to the industry and expect to have
25 it out by the end of January. It is going to be an
. 299
1 aggressive schedule, but we are really targeting that,
2 pushing hard to make that happen.
3 When the draft Regulatory Guide, that I believe
4 the staff will be talking about, is issued, we will be
5 developing comments on that using some of the insights we
6 have developed during the development of 99-03 and other
7 information we have gathered through various surveys and
8 studies of the issue.
9 And depending on how things work out, we believe
10 there is a benefit to have an industry-wide workshop where
11 we would have an opportunity to share, educate, converse and
12 help licensees understand this issue better and what they
13 need and should do.
14 DR. KRESS: Is your goal still to get 99-03
15 endorsed in the Reg. Guide?
16 MR. COZENS: I believe the staff has decided that
17 is not the most efficient path for them.
18 DR. KRESS: Does that detract from the benefit of
19 99-03?
20 MR. COZENS: It is a detractive option, but
21 content is important. I just want to emphasize that, you
22 know, the comments we receive from staff, we are going to be
23 studying seriously to see what adjustments are appropriate
24 in our document. We are going to be looking at that. That
25 will be the first formal set of comments we will have
. 300
1 received in writing, on I should emphasize the October
2 version of that document.
3 So, with that, that completes our comments, and I
4 apologize for the duration of our presentation, it did run
5 over about 20 or 30 percent of what we had intended. And so
6 if there is no more comments, questions?
7 DR. KRESS: Seeing none, we still have some --
8 CHAIRMAN POWERS: We have got lots to do.
9 DR. KRESS: We have got lots to do.
10 CHAIRMAN POWERS: We have got lots to do.
11 DR. KRESS: Well, we have got presentations by the
12 staff.
13 CHAIRMAN POWERS: Unless it is pressing.
14 MR. BARRETT: Mr. Chair. I was going to say, Mr.
15 Chairman, I am Richard Barrett with the NRC staff. We could
16 considerably compress our comments.
17 DR. KRESS: That would be helpful if you could.
18 MR. BARRETT: I thought it would be.
19 DR. SEALE: You may be suitably rewarded, too.
20 DR. KRESS: So why don't we turn it over to you
21 now and let's hear the staff's presentation.
22 MR. BARRETT: Actually, I think we could do the
23 presentation from the table here. And there are really only
24 a few points that I think need to be made.
25 First of all, I would like to echo what Mr. Cozens
. 301
1 said. I think that what you see in 99-03 represents a
2 significant amount of work on the part of the industry,
3 working with our staff in what I would call definitely a
4 good faith effort to address every aspect of control room
5 habitability. And if any of you are familiar with where we
6 were in August of 1999, you will recognize immediately what
7 an enormous improvement this current version of 99-03 is
8 over the version that we had at that time.
9 I think the second thing I would like to say is
10 that the presentation you just saw was an excellent outline
11 of the current status of this issue. It characterizes the
12 industry's current technical status. I think it also fairly
13 characterizes the industry's positions on the key issues,
14 and characterizes fairly the staff's current positions on
15 these issues. And as Mr. Kress said earlier, the key
16 issues, as he said, were related to sufficiency. What
17 constitutes a sufficient test for the initial test and, of
18 course, for the periodic reassessment of testing?
19 Secondly, the question of what is the proper way
20 in which the NRC should control this program, and Mr. Cozens
21 outlined all of the options and our current position.
22 And then thirdly, one that I am not sure was fully
23 characterized, and that is the question of what is the role
24 of a licensee's current licensing basis. And I think what
25 is important to us there is that, in the current licensing
. 302
1 basis for many, if not all of the plants, there are enormous
2 conservatisms, conservatisms on top of conservatisms. And
3 there are also non-conservatisms. And in some cases there
4 are licensees who only are required to look at a subset of
5 the accidents.
6 And the position that the staff is taking is that,
7 as we properly bring ourselves forward into this decade,
8 into this millennium, with new information about source
9 terms, and we can properly remove the conservatisms, or at
10 least reduce the conservatisms to make everything more
11 realistic in terms of the way we do meteorology and
12 transport, and the way we do source terms, we believe that
13 it is not proper to retain the non-conservatisms, that we
14 should bring all of this information up to date so that we
15 have something that is more realistic in every aspect.
16 So, our position is not to retain those
17 non-conservatisms that are in the current licensing basis
18 for some plants, and to request that licensees commit to
19 analyzing a full spectrum of accidents.
20 DR. KRESS: Do you have a list of what those
21 non-conservatisms might be?
22 MR. BARRETT: Jack Hayes of our staff could tell
23 you more about that.
24 MR. HAYES: Some of those issues might be, for
25 example, the assumption of suppression pool scrubbing for a
. 303
1 MARK I and MARK II containment. Another one might be the
2 fact that some plants don't include iodine spiking in their
3 evaluations of a main steamline break or steam generator
4 tube rupture, you know. So that is just a couple of the
5 examples.
6 DR. KRESS: Thank you.
7 MR. BARRETT: So if I could just wrap up, I could
8 say that it is a credit to everybody who has been involved
9 in this that we have reduced this thing to a very few
10 issues, and we are working to further reduce these issues.
11 Right now the staff is in the process of drafting some
12 regulatory guidance that -- what we are going to do is we
13 are going to take these issues and we are going to examine
14 them in-house so that the entire management chain of NRR has
15 an opportunity to look at these issues, and the options, and
16 ask ourselves -- what is the right thing to do given what we
17 know about this issue, what we know about the risk
18 significance of this issue, and what we know about how this
19 issue compares with how we have dealt with other issues in
20 the recent past?
21 And then in the spring of this year, we will put
22 these positions out for public comment, and that will be an
23 opportunity for it to be aired by everyone.
24 DR. KRESS: You mentioned what you know about the
25 risk implications of these issues. Has someone done a risk
. 304
1 evaluation of this, of the issues?
2 MR. BARRETT: No. No, we have not. We have not
3 done a risk evaluation of the issues, of this issue. There
4 has never really been a motivation to do so. But we have
5 looked at it qualitatively. We understand the risk
6 associated with the radiological part of this, which has to
7 do with accident management, of course, the toxic gas, and,
8 of course, the smoke, which would be attendant to a fire,
9 which also gives -- which we know the risk significance of
10 fires and how this might complicate response to that. But
11 it is only qualitatively that we have looked at it.
12 DR. KRESS: Thank you.
13 MR. BARRETT: Finally, there has been some
14 discussion today, and I think there is some vagueness about
15 the role of 99-03. We had earlier talked about having 99-03
16 be an industry guidance that would be endorsed by an NRC
17 Reg. Guide. I think that that was probably never a
18 realistic option.
19 99-03 is a very complex technical document, it is
20 not a regulatory document. I think that probably where we
21 will end up is that 99-03 will be a technical document that
22 will be heavily referenced in regulatory guidance, because
23 it contains a wealth of information about how to test, how
24 to analyze, how to maintain and operate your plant.
25 So we plan to continue working, we are going to
. 305
1 give the industry our comments on the October 13th version.
2 To the extent that it requires further clarification, we
3 will continue to work with them to make that document, as
4 they say, as they have said, make it the best guidance
5 document they can possibly make it. And so we are hoping --
6 our vision is that we will end up with regulatory guidance
7 that will deal with the issues at the level of detail that
8 you saw in the NEI presentation today, coupled with a 99-03
9 that will give an operating utility the kind of guidance
10 they need to implement that guidance.
11 DR. KRESS: Is your Reg. Guide -- I don't want to
12 put you on the spot, but is it likely to endorse parts of
13 the 99-03? Like, for example, the control room habitability
14 program and the smoke control program.
15 MR. BARRETT: I think that it will probably
16 endorse the concept of a qualitative, for instance, or not
17 -- let me not pre-judge what our management will say, but,
18 for instance, the commitment on the part of the industry to
19 have a qualitative way that you saw of dealing with smoke,
20 it will endorse that, but it probably will not endorse the
21 details of how you go about doing it.
22 DR. KRESS: Okay.
23 MR. BARRETT: I think that is probably really all
24 we wanted to say today, you know, above and beyond what the
25 industry has already said.
. 306
1 DR. KRESS: Okay. Thank you very much for that
2 very precise and concise wrap-up, I think it was helpful.
3 Did you have a question of the staff before we
4 move on to the next --
5 MR. SIEBER: Yeah, I do. And it can probably be
6 answered yes or no. In your last slide, Jack, you talk
7 about spiking factors and duration, and so did NEI. NEI
8 proposes a spiking factor of 335. Does the staff agree with
9 that number?
10 MR. HAYES: The staff were the original ones who
11 generated that value.
12 MR. SIEBER: 335. Is that consistent with Adams
13 and Atwood?
14 MR. HAYES: Well, Adams and Atwood really didn't
15 put forth a spiking factor. What they had --
16 MR. SIEBER: They do have a database, though.
17 MR. HAYES: They do have a database, and that is
18 the database which we utilized. It was I think a paper by
19 Adams and Saderson, who did put forth the spiking factor.
20 We utilized Dr. Adams' data and he provided that to us, and
21 we did an evaluation to come up with the factor of 335.
22 MR. SIEBER: Okay. The Adams and Statler paper
23 was two years before, it was in the '80s, I think.
24 MR. HAYES: Right, that's correct.
25 MR. SIEBER: Do you agree that spike duration can
. 307
1 be as low as three hours?
2 MR. HAYES: We are presently evaluating that. We
3 have not drawn a conclusion on that.
4 MR. SIEBER: Oh, okay. So in your last slide, you
5 weren't actually going to tell us what the number.
6 MR. HAYES: Right. What the last three slides
7 were intended to do was to represent the considerations that
8 the staff must take under review in the development of these
9 Regulatory Guides. It was not intended to indicate that
10 there was positions one way or the other.
11 MR. SIEBER: Okay. One final question. If
12 control room habitability requirements are not in tech
13 specs, there still are regulatory requirements in Appendix A
14 and Part 100 that apply, right?
15 MR. HAYES: By Appendix A, you are talking about
16 the GDC-19?
17 MR. SIEBER: Yes, sir.
18 MR. HAYES: Yeah, GDC-19 still applies and so does
19 Part 100.
20 MR. SIEBER: Okay. So there are requirements out
21 there that licensees must satisfy whether it is in their
22 tech specs or not?
23 MR. HAYES: Consistent with whatever their
24 commitment is as far as those particular regulations are in
25 their licensing basis.
. 308
1 MR. SIEBER: Okay. That's all I have, Dr. Kress.
2 DR. KRESS: Thank you.
3 At this time we have a presentation by Mr. Pete
4 Lagus of Lagus Applied Technologies. I don't know if I
5 pronounced that right or not.
6 DR. LAGUS: Perfectly.
7 DR. KRESS: Okay.
8 DR. LAGUS: And I promise -- I am a fairly rapid
9 speaker anyway, and I promise, due to the lateness of this,
10 I will try to speak even more rapidly.
11 DR. KRESS: Well, it would help, but you have the
12 full allotted time.
13 DR. LAGUS: Okay. Well, the motivation for this
14 is I didn't attend the last ACRS NRC NEI meeting wherein you
15 discussed many of the issues regarding tracer gas testing in
16 the control rooms, but the following two weeks I certainly
17 heard about it, because I got an awful lot of telephone
18 calls asking me to clarify the whole question of how one
19 defines the uncertainty of these measurements, what the
20 uncertainties are, what we might do to reduce them, and et
21 cetera, et cetera, et cetera.
22 And there were enough questions that seemed to
23 have some common themes, that when I learned that this
24 meeting was occurring, I asked Secretary Brenner if he could
25 give me a couple of minutes to address some of these issues.
. 309
1 And I would like to thank the committee for giving me this
2 time.
3 Basically, what I am going to tell you about here
4 is that we have done -- I am only going to talk about tracer
5 gas measurements that I personally have done. So I have
6 done tracer gas measurements of air leakage in 16 nuclear
7 power plant control rooms that encompass 22 reactors.
8 Now, in practice, when we do an air in-leakage
9 measurement, we get a single value of in-leakage for each
10 control room emergency ventilation system, we call a CREVs,
11 operating mode. And for pressurization, as we understand
12 that the pressurization systems are somewhat different from
13 the recirculation systems, for pressurization CREVs, each
14 in-leakage value is obtained from between 10 and 20
15 individual data points, while for recirculation CREVs, the
16 volume normalized in-leakage rate, because what we get
17 directly actually is an air change rate, which is a volume
18 normalized air leakage rate -- in other words, it is the air
19 leakage rate in CFM divided by the volume of the control
20 room envelope -- is determined from at least 100 data
21 points.
22 That, incidentally, is one reason why the
23 statistical errors on the control room, the recirculation
24 control room measurements tend to be much tighter than the
25 ones in the pressurization measurement.
. 310
1 The problem is that out of over 60 individual data
2 sets that we have, only four of these data sets have been
3 repeated. So if we would like to talk about the
4 repeatability of these measurements, it is very difficult, a
5 priori, to talk about repeatability when out of 60, you only
6 have four repeats. And what we then require is some method
7 of data analysis that allows us to statistically address the
8 uncertainties in our measurements.
9 Fortunately, for us, there does exist an ANSI/ASME
10 standard, PTC 19.1, and it is probably in -- the ones that I
11 am familiar with are the 1985 and 1990 versions. I believe
12 there is a 1998 version which, as I understand, hasn't
13 changed tremendously. But what this basically does is it
14 combines both bias or systematic uncertainties. We are
15 going to get rid of the word "error", all right. What we
16 are talking about is uncertainties here, and the bias or the
17 systematic errors of the measurement equipment, with the
18 random uncertainties of the actual measurement data. So you
19 are really combining two different kinds of information.
20 Now, I am trained as an experimental physicist,
21 and we tend to get very jaded about the kinds of
22 measurements and measurement errors that we do. But,
23 unfortunately, we also get a little sloppy sometimes, and
24 that seems to follow over into our engineering brethren, and
25 we tend to get sometimes errors that are, quote,
. 311
1 "engineering judgment." And engineering judgment,
2 unfortunately, because we are all human, can sometimes be
3 colored.
4 The nice thing about this particular standard is
5 that it does combine these uncertainties in a statistically
6 meaningful way. It provides 95 percent confidence limits.
7 It substitutes a calculational format for our, quote,
8 "engineering judgment." And especially it seems, in the
9 ventilation areas, there tends to be an awful lot of
10 engineering judgment substituted for rational error
11 analysis.
12 And the interesting thing about it as I began to
13 look into this a number of years ago is that formalism is
14 used extensively, for instance, in the steam turbine
15 industry. One of the examples, that they actually work
16 through in excruciating detail in PTC 19.1, has to do with
17 the performance of a steam turbine.
18 Well, I am not going to give you a course in
19 statistics in 10 minutes, but the basic idea behind the ASME
20 measurement uncertainty is a root sum square in which one
21 combines bias errors, these are the systematic uncertainties
22 because of uncertainties in our measurement equipment, with
23 the random uncertainties that are generated in the act of
24 measuring data. All right. And it is a fairly self-evident
25 definition. But when all of the numbers that you have heard
. 312
1 me talk about for uncertainty, this is what I mean. If I
2 tell you the number is 126 plus or minus 36, what I mean is
3 the upper 95 percent confidence limit is 136 plus 36, the
4 lower confidence limit is 136 minus 36, at the 95 percent
5 confidence level.
6 Now, the systematic uncertainties, or what we call
7 bias, when I was an undergraduate, they were called biases,
8 and them somewhere, I am not sure if the political
9 correctness crowd took over or what, but now I understand
10 that they are called systematic uncertainty, so we will go
11 with the new jargon. But typical of the systematic
12 uncertainties are, in a case of testing of pressurization
13 CREVs, the concentration of the injection gas.
14 The flow meter that one uses to inject the tracer
15 gas into the system. The analyzer calibration, yes, one has
16 to calibrate an analyze, one needs calibration gases. One
17 needs an analyzer calibration, one needs an analyzer
18 response. In the case of a recirculation system, the volume
19 of the control room envelope. These are systematic
20 uncertainties.
21 Random uncertainties basically result in
22 variability of the measured data set. Random uncertainties
23 aren't uncertainties that we get in our measurement system,
24 they actually come out of our data. And the things that
25 will affect these things are the plant configuration and the
. 313
1 operation, HVAC configuration of the operation, external
2 influences, wind temperature, what-have-you, like that, gas
3 sampling. In other words, how the actual sampling is done.
4 If poor sampling technique is used, one will find that the
5 random uncertainties, the uncertainties in the measured data
6 set are going to be, perforce, larger.
7 Incomplete mixing. One of the fundamental
8 assumptions behind everything that one does in E7.41, which
9 is the controlling standard here, is that we have a
10 well-mixed volume, because we are really tricking this whole
11 thing into thinking that what the chemical engineers call a
12 well-stirred reactor. So we have to make it think it is a
13 well-stirred reactor so the mathematics works for us.
14 Another reason that one might get some
15 uncertainties in the measured data set is that the tracer
16 concentration itself is not in steadystate, which is
17 required when one is testing a pressurization CREV. So
18 these are kind of examples of what the uncertainties are.
19 In summary, for uncertainties in 60 individual
20 data sets for both pressurization and recirculation CREVs,
21 the random uncertainties are larger than the systematic
22 uncertainties for pressurization CREVs, across the board.
23 And if you think about it, I think Dr. Kress pointed out
24 that what we are measuring here is the small difference
25 between two large numbers, and that, in fact, is why our
. 314
1 random uncertainties get larger in this particular case,
2 because we attempt and, in the best of all possible worlds,
3 we will measure the makeup flow rate and, if you will, the
4 total in-leakage rate into the control room envelope with a
5 tracer test simultaneously. So we are actually measuring a
6 difference between, let's say, 2200 CFM and 2000 CFM, and we
7 all understand what happens when we do difference
8 measurements.
9 The Urss for the pressurization CREVs averaged
10 approximately 40 percent, plus or minus 40 percent. Those
11 are 90 percent confidence limits. Okay. I would like them
12 to be better, but the reality of life is this is what the
13 statistics, this is what the physics and this is what the
14 mathematics gives you.
15 However, interestingly enough is that -- and I
16 don't have the numbers at the tip of the fingers right now,
17 but for the vast majority of the plants that we tested at
18 least initially, even if one took the lower confidence
19 limit, in other words, took the mean and subtracted off the
20 40 percent or whatever it was, the resulting value was still
21 above the original habitability analysis in-leakage. So,
22 yes, it is a problem, but for most of the plants that we
23 have done so far, and I am not saying that this will be true
24 in the future, but for most of the plants that we have done
25 so far, that hasn't really been the problem.
. 315
1 The random uncertainties, as you might expect, for
2 the recirculation CREVs, are smaller than the systematic
3 uncertainties, and one of the reasons being is that we sort
4 of do our world famous statistics trick, we throw lots of
5 data points at it, and that really helps us out in terms of
6 our statistics. And the uncertainty, the 95 percent
7 confidence limits for the recirc average approximately, plus
8 or minus 5 percent, assuming a 2 percent uncertainty in the
9 control room envelope volume, and that goes through
10 linearly. If you want 5 percent in your control room
11 volume, then you are going to boost up the 95 percent limits
12 that we presently see to around 8 percent. I believe in one
13 of the slides this morning people were saying less than 10
14 percent, and that is certainly a realistic and defensible
15 number.
16 Now, repeatedly, which is always the question.
17 You know, obviously, one way to get rid of these huge
18 numbers, or these huge error bars, is to do again what we
19 like to do in statistics and that is make lots of
20 measurements, and let's average them. The problem is that
21 is time-consuming, that is expensive, and last time I
22 checked, nuclear power plants were not in the business of
23 doing scientific research. So it is very difficult to get
24 yourself into a situation where you can repeat these
25 measurements.
. 316
1 However, it turns out, just as a friend of mine
2 would say, through a juxtaposition of fortuitous
3 circumstances, we actually have four repeat CREVs data sets,
4 and sort of the codicils in here are that there were no
5 modifications to either the control room envelope or the
6 CREVs between the time that the initial measurement was made
7 and the retest was made. In other words, they didn't go,
8 oh, well, we got that number, now we went in and we did all
9 this sealing and we fixed all this stuff, no. Basically,
10 the system was tested, probably after it had been sealed or
11 fixed, and then some time period later was tested again.
12 The tests were all done in the same ventilation
13 lineup and physical plant lineup as far as we were
14 physically able. I mean you have realize that some of these
15 plants are enormous. One control room envelope, or two
16 control room envelopes particularly we have recently tested
17 are essentially six-story, poured concrete buildings inside
18 larger poured concrete buildings. So one tries to do as
19 best one can in getting the physical configuration back in
20 an identical position, particularly in terms of doors and
21 things like this.
22 It turns out that of these sets, two
23 pressurization CREVs were in different plants, and then we
24 had two different recirculation CREVs in the same plant.
25 This was a little bit of an artificial test because in this
. 317
1 particular plant, instead of doing a recirculation test in a
2 mode in which the plant might operate under a normal
3 emergency condition for a variety of reasons that we don't
4 have time or interest to go into right now, the plant was
5 operated at an artificially high induced pressure across the
6 entire control room building, if you were, of approximately
7 2/10ths of an inch of water.
8 For the recirc CREVs, this is the kind of thing
9 that we saw. The legend on here is D and D are obviously
10 the same CREVs. These were taken two years apart. D prime
11 was a different CREVs in the same plant, and D prime taken
12 two years apart.
13 As we can see, three out of the four are well
14 within the 95 percent error bars. This one, we are
15 considerably over the 95 percent error bar. We are probably
16 at 99.999 percent to get that one. We are like 5 standard
17 deviations up. I do not know why. These systems were
18 completely contained inside the control room envelope. The
19 CREVs themselves were adjacent to each other inside one huge
20 mechanical equipment room.
21 Our best, and this is a guess, and I will tell you
22 that freely and fully, our best guess is that -- and this
23 was particularly a six-story, poured concrete control
24 building, and we probably didn't have, or potentially we
25 didn't have one door completely closed, because, remember,
. 318
1 we were inducing an artificial 2/10ths of an inch of water
2 pressure across this building. It turns out that if you
3 have a 7 foot high crack an eighth of an inch wide, you can
4 let 50 CFM through that, like that. That would give you
5 this kind of number.
6 So I don't know why, but it shows you the
7 unreproducibility that we get under real test conditions in
8 real plants. It is not bad.
9 We have also two sets of in-leakage values, repeat
10 in-leakage pressurization CREVs. And I guess, I am sorry, I
11 am going a little bit fast here. What I should point out is
12 what I am showing you is the diagonal hatching is actually
13 the mean, the cross-hatching is the upper 95 and the dots
14 are the lower 95 on both of these plots.
15 Here we had Plant H, well, CREVs H triple prime in
16 Plant H -- unfudged, guys. These are two measurements taken
17 a year-and-a-half apart. All of the input data between
18 these two things are completely different. We were
19 flabbergasted when we came out with the same number.
20 Because when we did it, I was sure we had make a mistake.
21 We actually went back and cold-calculated again while we
22 were at the plant.
23 The second one here was another CREVs in a
24 different plant. These tests were taken identical
25 situations, three weeks apart. And here we can see that the
. 319
1 means differed by about less than 40 or about 15 to 20
2 percent, whereas, the upper and lower 95s were a little bit
3 closer to 40 percent in this case.
4 So in the case of the sets that we have been able
5 to look at our retest data, it appears that in most of the
6 cases, our means are considerably closer to each other than
7 the upper and lower 95 percent confidence limits would lead
8 you to believe, but it is the best we can do under the
9 circumstances that we have got, because I doubt that anybody
10 seriously wants to do one of these tests 10 times in a row
11 in a plant.
12 So, in sort of summary, repeatability is better
13 than the 95 percent Urss limit for the two pressurization
14 CREVs, and it is within the limit for three of the four
15 recirculation CREVs. And, as I say, I believe I am
16 reasonably confident that a door was not closed, but
17 something I can't prove, and I will never be able to prove.
18 One issue that we seem to keep coming back, and I
19 certainly get an inordinate number of calls about this is,
20 how do we define an acceptable in-leakage value? You know,
21 you say tomato, I say tomato, and off we all go.
22 Now, interestingly enough, OSHA faces the same
23 dilemma in a different context. As you may or may not all
24 be aware, there are a list of some 400 substances that are
25 regulated, whose concentrations are regulated in the
. 320
1 industrial environment, and they must be -- the
2 concentrations of these substances must be less than
3 something called the PEL or the permissible exposure limit
4 for the particular substance based on an eight hour a day,
5 40 hour a week exposure.
6 Now, an OSHA inspector will come into a plant for
7 some reason. He will do a sampling of a xylene atmosphere
8 for a potential violation. We have exactly the same problem
9 that we have with our control room. Now, he is going to
10 make one measurement. Okay. OSHA inspectors have the right
11 to fine based on what they learn here.
12 Well, the inspectors use, and there is a whole
13 methodology that has been worked out inside the OSHA
14 formalism, use a one-sided confidence limit test on the
15 airborne concentration values. And the measured
16 concentration value is compared to the PEL and the
17 confidence limit, which is based on a set of standards,
18 sampling and analysis protocols that OSHA has gone to the
19 trouble of defining standard errors for. And so, very
20 crudely speaking, a value that lies below the upper
21 confidence limit is not considered to be a violation.
22 So if, for instance, the PEL for xylene, I will
23 tell you, is 100 parts per million, and for the particularly
24 sampling and analysis error that is normally used is about
25 .15. So, if you came into my plant and measured 105 parts
. 321
1 per million, even though you have exceeded, on that one
2 measurement, you have exceeded the PEL, my plant is probably
3 not in violation, because if we take into account the mean
4 and the upper confidence limit, okay, I am not in violation.
5 And this has been, this is an accepted formalism, it is an
6 accepted regulatory formalism and, as far as I understand,
7 it is also an accepted legal formalism. The fines are
8 levied on the basis of this and apparently paid.
9 So what kind of occurred to me in closing is that,
10 for in-leakage, NRC could propose to accept a similar kind
11 of confidence limit value concept. The NRC, obviously,
12 would want the upper confidence limit. And the industry
13 would say, well, no, no, no, the lower confidence limit is
14 where we want to be. And I think there needs to be some
15 negotiation here. But my point is there is already an
16 intellectual and a regulatory framework that encompasses
17 this. It might be something to think about.
18 And in this control room habitability context, if
19 there were interest in doing this, what one needs to do is
20 to establish standard errors of both the analysis method and
21 the sampling method. This can be done, but it obviously
22 hasn't been done at this point. You need to precisely
23 define the test performance, the instrumentation,
24 calibration, sampling and analysis protocols. This can be
25 done, it hasn't been done.
. 322
1 And if one has these kind of definitions, one can
2 control these confidence limit excursions, because,
3 willy-nilly, you can not just say, oh, geez, Lagus, that is
4 a great idea, we are going to do it tomorrow, because then I
5 have a real impetus to be just as sloppy as I can. If NRC
6 wants the top value and the industry says, no, we get the
7 bottom value, the bigger my error bar is, the further down I
8 drive this number. That is probably not an intellectually
9 valid way to operate.
10 But there should be a middle ground here, and I
11 merely bring it up because in a different industrial
12 context, this problem is faced probably every day in the
13 working environment, and this is one solution that has been
14 proposed.
15 Questions? Thank you.
16 DR. KRESS: Well, thank you.
17 MR. DUFFY: John Duffy, PSEG Nuclear. I have one
18 question about the test methodology, whether or not, when
19 you perform a test like this, you can categorize whether the
20 in-leakage that you are measuring is filtered versus
21 unfiltered. For example, in recirculation system, my area
22 of greatest vulnerability would probably be the floor to the
23 cable spreading room. In that case I would expect that the
24 in-leakage would be unfiltered. However, for a pressurized
25 system, more than likely the greatest area of vulnerability
. 323
1 is in the duct work or the fan shaft itself. And depending
2 upon the fan filter configuration, that can either be
3 filtered in-leakage or unfiltered in-leakage.
4 I am wondering how your test methodology deals
5 with that issue.
6 DR. LAGUS: Okay. That's a good question. And
7 the answer is you deal with that -- unfortunately, you have
8 105 control room ventilation systems and they are all
9 different. But what we will do is, especially in the
10 pressurization systems, in addition to just doing the
11 integrated leak test, because I can't get out of a plant if
12 I tell somebody what a leak is, without telling them where
13 the leaks are, we will then make additional measurements
14 during this time in, for instance, runs of ductwork, through
15 rooms and things like this, and we are then able, again,
16 using these conservation of mass concepts, to fractionate
17 out how much in-leakage is occurring in a particular run of
18 duct or through a particular damper, through a particular
19 air handling unit housing, or through a particular
20 ventilation system all by itself.
21 So, yes, one can do that, and, in fact, if one of
22 these tests is properly run, you always get more than just
23 the gross in-leakage number, simply because of the fact the
24 gross in-leakage number by itself doesn't tell you anything
25 except that you may or may not have a problem.
. 324
1 DR. KRESS: Okay.
2 DR. LAGUS: Thank you.
3 DR. KRESS: Thank you very much. You can have the
4 podium back.
5 CHAIRMAN POWERS: Okay. Thanks, Tom. We will
6 recess for 15 minutes.
7 [Recess.]
8 CHAIRMAN POWERS: Let's come back into session.
9 Our next discussion concerns a proposed final Regulatory
10 Guide, DG-1053, Calibration and Dosimetry Methods for
11 Determining Pressure Vessel Neutron Fluence, which ought to
12 be pertinent to any number of activities that we're hearing
13 about.
14 And to lead us through this particular subject, we
15 have the esteemed Dr. Shack.
16 DR. SHACK: Well, we had a meeting of our
17 Materials and Metallurgy Subcommittee on November 16th to
18 hear a full briefing on this. We had a consultant, Dr.
19 Sulfinides from the University of Missouri, Rolla, who
20 attended the meeting as a consultant. His report is
21 included in your notebook to discuss this.
22 As Dana noted, the calculation of the vessel
23 fluids is important for a lot of reasons. It's fundamental
24 to the calculation of the vessel fracture toughness, which
25 is important to us for our pressure/temperature limits,
. 325
1 pressurized thermal shock and other things.
2 The Staff, in reviewing this over the years, had
3 found that the licensees had used the wide varieties of
4 methods and assumptions in doing these calculations. This
5 sort of necessitated an extensive and laborious review to
6 assess the validity of this, and they thought it would be
7 helpful to provide a Regulatory Guide that would outline a
8 method, or at least methods that would provide acceptable
9 results for the NRC.
10 They've done the Reg Guide, and then they've done
11 a rather interesting thing, I think, in providing a NUREG
12 with a benchmark problem that the licensee can calculate and
13 use his methods and test himself and see how well he can do
14 it.
15 The methods are intended to be best estimate
16 methods. They provide guidance on how to calculate the
17 uncertainties in the analysis, and I think Dr. Chokshi will
18 be short of introducing the subject for us.
19 DR. SEALE: I might say this was -- I heard a
20 presentation on this particular thing among the very first
21 presentations that I heard when I came on the ACRS eight
22 years ago. I thought you might be interested in a
23 benchmark.
24 DR. SHACK: They work slowly but thoroughly.
25 DR. SEALE: Yes, I got that.
. 326
1 DR. CHOKSHI: Good afternoon, and as Dr. Shack
2 mentioned, I'm going to repeat some of the things he already
3 said and just briefly describe why we are here and give you
4 an oral status.
5 I'm accompanied by Bill Jones of the Staff, who is
6 the Research Project Manager, and he give you the more
7 detailed overview of the Guide's content, and then Dr. John
8 Carew of Brookhaven National Laboratory will go into the
9 technical details, positions, and some of the very more
10 interesting aspects of the Guide.
11 John has been with this developmental guide since
12 the beginning, and probably we'll see why it has taken this
13 many years. So let me just start with briefly going over
14 the purpose of the presentation, which is to obviously at
15 this point, at this stage, where the Guide is ready for the
16 final publication.
17 It has been a long time in the making, and it has
18 gone through a number of public comment exercises, and
19 involvement of interested groups.
20 What we would like to do today is to give you a
21 summary of the key technical positions, and also describe
22 briefly, selected public comments which have been of more
23 significance.
24 And obviously, I think the purpose of, and our
25 purpose for being here is to secure approval, and if
. 327
1 everything goes well, the next will go to CRGR, and then we
2 hope to publish the Guide by the early Spring.
3 Before I give you the status of the Guide, I think
4 I'll just retrace what this Guide does and why it's needed.
5 As Dr. Shack mentioned, there are a variety of methods out
6 there, and I think there's a need for providing standardized
7 methods.
8 And NRR in 1992 requested Research to come up with
9 a Guide which will outline some acceptable methods that can
10 be used by the NRC Staff.
11 The Regulatory Guide basically provides acceptable
12 methods for meeting the requirements of the PTS Rule, 10 CFR
13 50.61, Appendix G, Fracture Toughness Requirements and also
14 the surveillance requirements of Appendix H.
15 Again, the Guide is one acceptable method, and it
16 provides the methods for the fluence calculations.
17 So Research received the user need letter from NRR
18 in 1992, and the initial effort was started in 1993, and
19 work has been done at Brookhaven National Laboratory.
20 After the publication or after we were ready to
21 release the first draft, the Staff briefed ACRS and CRGR,
22 and I think I understand that ACRS at that time said come
23 back when you ready with your final draft. So we are here
24 now.
25 A number of things have happened as to why it has
. 328
1 taken so long. The scope --
2 DR. APOSTOLAKIS: Maybe you should define the term
3 of ACRS membership. If a project takes longer than the
4 tenure of the people here.
5 [Laughter.]
6 DR. CHOKSHI: Well, I think that as you will hear
7 from John, there are good reasons why. The thing has
8 evolved.
9 DR. APOSTOLAKIS: I have no doubt.
10 DR. CHOKSHI: I should expect that. But I think
11 the scope of the Guide has undergone some substantial
12 revision. We went to the public comment, and Bill and John
13 will touch on that.
14 Our last public meeting was in September of 1999,
15 and after that, we received additional comments, and in last
16 November, we disposed of all the comments, prepared the
17 final version and briefed the Subcommittee on the 16th.
18 The Guide has been concurred in with NRR and OGC,
19 and again I think that if this Committee also sees fit that
20 it can move forward, then we hope to publish it by 2001,
21 early 2001.
22 DR. SHACK: We don't want to rush into these
23 things.
24 [Laughter.]
25 DR. CHOKSHI: Not at this stage.
. 329
1 Anyway, what I think we will do is, as I
2 mentioned, Bill will give you an overview, and John is going
3 to describe much more, some of the technical details, and
4 finally, Bill will come back with one more viewgraph
5 summarizing and reiterating the schedule.
6 So, if there are no questions, I will ask Bill
7 Jones to come up here.
8 MR. JONES: As Dr. Chokshi said, I'm the Office of
9 Nuclear Regulatory Research Project Manager for this
10 Regulatory Guide, and have been so for about a year and a
11 half. I am a bridge between what Dr. Chokshi has told and
12 what Dr. Carew will tell you.
13 I'm going to go pretty fast with just an overview,
14 talking about the objectives and the contents of the
15 Regulatory Guide, and emphasize some proportions in the
16 Regulatory Guide, some proportions of the material.
17 The bottom line is that we believe this Guide
18 provides acceptable practices for fluence estimation and
19 calculation of the use of the dosimetry data.
20 Our aim is achieve greater accuracy and
21 completeness and reliability, and to describe acceptable
22 methods for doing fluence estimation. And that's really the
23 bottom line or the summation of what's on the Objectives
24 slide.
25 CHAIRMAN POWERS: When I look at things like
. 330
1 measurements of metal ductility or other measurements of
2 vessel embrittlement, I see data that are always
3 extraordinarily scattered, with lots of scatter.
4 That leads me to believe that I don't have to be
5 very accurate in the estimates of my fluences.
6 MR. JONES: The rules, the regulations that the
7 fluence estimates go into have an implicit assumption of
8 about 20 percent uncertainty. And John will be talking
9 about that 20 percent, but that's in the rules that the
10 fluence estimate calculation fits into.
11 CHAIRMAN POWERS: So only a plus or minus 20
12 percent is plenty good enough?
13 MR. JONES: Well, that's the implicit assumptions
14 in the rule, in the regulations that the draft Guide -- that
15 the proposed Guide supports. --
16 DR. SEALE: But the devil is in the details.
17 MR. JONES: That's true.
18 DR. SEALE: The thing you measure is a consequence
19 and an energy which is, god, how many orders of magnitude
20 lower than the energy where all the problems are done. And
21 so you're really inferring an energy distribution, which is
22 important for the damage mechanisms at depth.
23 And so it's not even the spatial distribution of
24 what you measure that's important; it's the energies that
25 happen to be in the processes of attenuation as you come
. 331
1 down to that energy, going through the medium.
2 To be 20 percent in the process of attenuating by
3 six orders of magnitude with a change in energy of a
4 million, that's a hell of a calculation.
5 CHAIRMAN POWERS: Well, it may be a tough
6 calculation, but I don't have to do it very accurately.
7 DR. SEALE: If you want it to be 20 percent, you
8 better.
9 MR. LOIS: This is Lambros Lois. Let me extend
10 what Dr. Seale is talking about. Way back in the late 70s,
11 early 80s, when we started to -- when we realized the need
12 for an accurate calculation of the fluence and assembled a
13 number -- about ten, actually -- laboratories and
14 universities to do the same calculation in a round-robin
15 fashion, we found a 300 difference between top and bottom.
16 And we realized that nobody really knew how to do
17 it. And then we've established a research effort to get
18 some benchmarks and proceed from there to stabilize the
19 calculations.
20 So this is the culmination of a very long effort,
21 both in dosimetry as well as in calculation methodologies.
22 CHAIRMAN POWERS: I'll just interject that
23 usually, it seems to me, I see plots of metal ductility
24 where the temperature as a function of fluence, one of which
25 -- one of those axes is a logarithm, and I see datapoints
. 332
1 typically with error bars that I can't cover with my thumb.
2 So being off by a factor of three doesn't strike
3 me as wildly different than the uncertainty in the
4 measurements of the mechanical properties.
5 MR. LOIS: Let me state that the plant or a
6 pressure vessel can go up or down. I mean, the reactor can
7 be shut down for a one-degree difference. It is to our
8 interest as well as to the industry's interest to increase
9 -- decrease the uncertainty in the calculated value of the
10 RTPTS.
11 MR. JONES: The Regulatory Guide itself, about a
12 third of the guide deals with the fluence calculations, and
13 about a fifth of the Guide deals with dosimetry
14 measurements, and the rest of the Guide is what we would
15 like -- what we want reported and implementation.
16 So the overview I'm trying to provide here is that
17 calculation is about a third, so it's the major part of the
18 Guide, and dosimetry is about a fifth, and then you talk
19 about what we want reported and implementation.
20 And the last slide in the small package is my
21 summary slide, and I'm going to be talking about that at the
22 end, so if there are no questions, I'm trying to create a
23 bridge for Dr. John Carew who has been involved in this
24 since it started.
25 If there are no questions, I'll turn it over to
. 333
1 John.
2 DR. SHACK: It will make a difference to the
3 utility of course whether its vessel will last for 60 years
4 or 20 years.
5 CHAIRMAN POWERS: Do you think so?
6 [Laughter.]
7 DR. SEALE: You have a way with words, Shack.
8 CHAIRMAN POWERS: I mean I just don't --
9 DR. SHACK: Fluence is pretty much directly
10 related to the time you have operated the vessel.
11 DR. WALLIS: It's the easiest part to calculate.
12 [Laughter.]
13 MR. CAREW: I am John Carew and I will be
14 discussing the Regulatory Guide for the calculation and
15 determination of the neutron pressure vessel fluence.
16 Let me start out by just noting, making some
17 observations, the reactor pressure vessel fluence is
18 required for the determination of vessel embrittlement and
19 lifetime.
20 The vessel fluence is used to determine the
21 adjusted to reference temperature for the nil ductility
22 transition, RTNDT, and is input to the PTS rule, Part 50.61
23 for determination of the reference temperature for
24 pressurized thermal shock RTPTS.
25 The difficulty in determining the pressure vessel
. 334
1 fluence stems from a combination of several factors. The
2 first and probably the most important is the fact that the
3 neutron fluence undergoes several decades of attenuation
4 between the core and the pressure vessel.
5 If you look at the pressure vessel in the next
6 slide, you see the neutron source, the neutrons that are
7 generated in the core region and propagate through the
8 downcomer, the thermal shield, the barrel of thermal shield
9 out to the pressure vessel.
10 During this attenuation the neutron flux as seen
11 in the next slide attenuate by three decades, out to the
12 pressure vessel. Due to a spectrum change here you see a
13 slight increase in the barrel in the pressure vessel, but
14 this is part of the basic problem.
15 Because of this strong exponential fluence
16 attenuation the calculation is very sensitive to the
17 material and geometry representation of the core and the
18 vessel internals, the space energy neutron source, and the
19 transport numerical transportational scheme. As a result,
20 to determine the fluence a detailed, multigroup,
21 multidimensional analysis is required for an accurate
22 fluence calculation.
23 DR. WALLIS: Why don't you measure it at the
24 vessel?
25 MR. CAREW: Why don't I measure at the vessel?
. 335
1 The reason is at the vessel you know very well how many
2 neutrons you have, but because of the attenuation through
3 all those internals you lose three decades of neutrons and
4 the amount you are left with is very uncertain, so even
5 though you can measure at the vessel, the trick is to
6 measure at the core --
7 DR. WALLIS: Why not measure it at the vessel?
8 MR. CAREW: Well, why don't you measure at the
9 vessel?
10 DR. WALLIS: If that is what you are interested
11 in.
12 MR. CAREW: Well, they do. They measure at the
13 vessel and we use those measurements but to calculate those
14 measurements -- well, we do measure at the vessel and we
15 measure in the capsule, we measure in the surveillance
16 capsules, but those measurements are not perfect, and you
17 need a calculation as well.
18 Let me finish this discussion here --
19 MR. LOIS: There is one more regulatory
20 requirements, Dr. Wallis, and this is that the 10 CFR 50.61
21 requires to predict the status of the vessel at the end of
22 life or times in between, so you can measure it, what it is
23 now, but you still have to have a reliable calculation to
24 push it forwards. Thank you.
25 MR. CAREW: So because these difficult
. 336
1 calculations are required, in practice there is a wide range
2 of fluence methods and approximations and cross sections are
3 used to simplify the calculation.
4 This is further complicated by the fact that it is
5 difficult to determine how accurate and what the errors are
6 that are introduced by these approximations because of the
7 limited number and uncertainty of the capsule fluence
8 benchmark data. Nevertheless, an accurate fluence
9 calculation is required because certain vessels have a
10 limited end of life margin to the reference temperature of
11 PTS.
12 The Regulatory Guide responds to this problem or
13 this concern by providing accurate fluence methods which
14 will ensure a fluence determination to within 20 percent as
15 required by the PTS rule.
16 In the following I will discuss first -- I will
17 review the Regulatory Guide. I will discuss the industry
18 comments and their resolution on the pressure vessel fluence
19 benchmark problem.
20 The Regulatory Guide was developed for the
21 Materials Engineering Branch by B&L and with input from NIST
22 and Oak Ridge.
23 The purpose of the Guide was to document
24 calculation and measurement methods for determining the
25 vessel fluence that are acceptable to the NRC.
. 337
1 The scope of the Guide was to determine fluence
2 for the PTS rule in Reg Guide 199 at Appendix G.
3 The Reg Guides includes a detailed description of
4 fluence calculation and measurement methods, procedures for
5 qualification of those methods and those calculations,
6 reporting requirements, and also it requires the calculation
7 of the benchmark problems in the NUREG Report 6115 for
8 fluence qualification.
9 The primary calculational tasks in determining the
10 pressure vessel fluence are, first, to determine the
11 geometrical and material compensation data, determination of
12 the core neutron source, and to propagate that source those
13 neutrons from the core out to the vessel and into the
14 cavity.
15 The Guide allows two types of calculations to
16 determine the fluence. The first calculation is a discrete
17 ordinates calculation, and the calculation -- the first step
18 in the calculation is to determine the amount or the number
19 and energy of the neutrons being produced in the core. The
20 number and the location of these neutrons in the core is
21 determined first by the assembly and fuel bundle power and
22 pin exposures, and the energy dependence of these neutrons
23 is determined by the isotopic fission spectrum.
24 The next step is to determine the material -- use
25 the component material compositions and the microscopic
. 338
1 cross-sections to determine the cross-sections that are used
2 to propagate these neutrons from the core to the vessel and
3 determine the attenuation.
4 DR. WALLIS: This R-Theta -- this is a cylindrical
5 coordinates, is that right?
6 MR. CAREW: Right.
7 DR. WALLIS: The z as well?
8 MR. CAREW: I'll get that in a second. You'll see
9 that right here.
10 DR. WALLIS: Oh, it is there.
11 MR. CAREW: It is in there, yes.
12 The next step is to actually propagate these
13 neutrons' source from the core out to the vessel, and this
14 is done by discrete ordinate transport code in which the
15 whole problem is discretized. The location of the neutrons
16 is made discrete. The energy of the neutrons is made
17 discrete. Their direction and space is made discrete. So
18 it is a discrete ordinate transport code calculation, and it
19 is done in basically two steps.
20 It's done with a radial planar calculation, and
21 R-Theta geometry, and then an axial calculation, an R-Z
22 calculation. These are combined in a synthesis procedure
23 and a three-dimensional fluence is determined.
24 The next step, the next type of calculation that's
25 allowed by the Regulatory Guide, and I might just interject
. 339
1 at this point and say in these Round 2 set of questions or
2 comments in 1996 one of the things that were introduced, the
3 industry wanted us to expand the scope of the Guide to
4 include this second type of calculation. At that time, this
5 type of calculation had never really been submitted to the
6 NRC or wasn't accepted at the time. There was no basis.
7 This was to do a Monte Carlo transport
8 calculation, so the Guide -- this was a major increase in
9 the scope of the Guide. We had to include this in the Guide
10 and develop the methodology included in the Guide and then
11 included in the benchmark NUREG report.
12 But in any case the Guide now incorporates the
13 Monte Carlo technique and this technique is basically very
14 similar, almost identical to the discrete ordinates
15 methodology except for one crucial step, and that is the
16 transport of the neutrons from the core out to the vessel
17 inner wall.
18 In this calculation, as you might expect, that
19 transport was done statistically where each particle neutron
20 is tracked from the core out to the inner wall of the vessel
21 and its fate on the inner vessel wall is recorded and of
22 course this problem is a very difficult problem. It's a
23 difficult problem in Monte Carlo because of the fact that
24 the neutron flux is attenuated by three decades and going to
25 the inner wall and consequently you need a thousand neutrons
. 340
1 to make one hit on the vessel inner wall and therefore you
2 have all the statistical problems that result in the large
3 variances that you get, but that is one of the difficulties
4 in this type of calculation, but it has the advantage of
5 course that all the surfaces and the geometry and the
6 energies can be represented exactly.
7 CHAIRMAN POWERS: What size of sample do you have
8 to use to get reliable --
9 MR. CAREW: Well, these calculations really
10 take -- at B&L it would take about a month to do, so you
11 have to run 400 million histories, something of that order,
12 and as you know the variance goes down by one of a score of
13 events so you don't increase that inexpensively.
14 Now the Regulatory Guide, Appendix B at 10 CFR 50,
15 requires that the calculational methods by qualified, and
16 the Reg Guide has a procedure here for qualifying and
17 ensuring the accuracy of the calculations. The procedure
18 includes two steps.
19 The first step is doing an analytic sensitivity
20 analysis where the calculational input is all perturbed,
21 sensitivities are recorded, and using estimates of what the
22 uncertainties in the input is we multiply these together,
23 combine them statistically, and make an estimate of what the
24 uncertainty in the calculation is.
25 In addition to this analytic approach, we use
. 341
1 benchmark comparisons, and there's two types of comparisons
2 that are made. Basically we compare to operating capsule
3 measurements at a plant when they exist and also we use
4 comparisons to benchmark experiments that have been done
5 sponsored by the NRC, the pool critical assembly benchmark
6 experiment at Oak Ridge, and other pool site facility PCPSF
7 experiment.
8 In addition, there is a test comparison or
9 calculation of the benchmark problem, which also is used to
10 assess the accuracy of the calculations, so this
11 qualification procedure results in an estimate of what the
12 uncertainty is. It should be less than 20 percent to
13 satisfy the PRS criteria.
14 The Guide also allows the situation where the
15 uncertainty is greater than 20 percent. In practice the
16 fluence is calculated and using the qualification procedure
17 the uncertainty determined and if the uncertainty is less
18 than 20 percent, equal to or less than 20 percent, the
19 fluence the best estimate fluence is a calculated value and
20 when appropriate a bias is applied.
21 If the uncertainty is between 20 and 30 percent,
22 the fluence is increased by an amount that ensures that the
23 possibility or probability of a under-prediction of the
24 fluence is not increased, and then if the uncertainty is
25 greater than 30 percent that Guide is not applicable.
. 342
1 Let's see. In summary, the Guide provides a best
2 estimate prediction of the fluence rather than a bounding
3 approach. The accuracy is less than 20 percent. It covers
4 the range from the top of the fission spectrum from 15 MEV
5 down to about .1 MEV and the qualification is done by both
6 benchmarking comparisons and an analytic uncertainty
7 analysis.
8 The applicability of the Guide is to input for
9 Appendix G and Reg Guide 1.99. The Guide includes the
10 application to both PWR and BWR core vessels and geometries
11 and fuel designs.
12 Again I would like to note here in defense of our
13 schedule and when we had the meeting in 1996 one of the
14 comments was that we would expand the scope to include BWRs,
15 and that was done and now the methodology applies to both
16 BWRs and PWRs.
17 In addition, the Guide applies to vessel fluence
18 reduction schemes, core designs like a low leakage or ultra
19 low leakage core designs, partial length shield assembly
20 cores, and life extension calculations.
21 Let me get a status of where we are. We had the
22 initial NRC reviews in '93. We came before this committee.
23 The NRC -- it was released for comment. We had a formal
24 review meeting with industry. We had a set of round one
25 comments.
. 343
1 In September of '96 we met again with industry to
2 review the update and the resolution to their comments. At
3 this point the comments were evaluated and incorporated and
4 at the request of this meeting, in response to that meeting
5 the Monte Carlo and BWR benchmark problem was expanded --
6 the Monte Carlo was incorporated and the scope was expanded
7 to cover BWRs as well.
8 September of '99 we met again with the industry
9 and presented the new products, the Monte Carlo application
10 and the BWR as well as all the other updates.
11 At that meeting industry requested that we provide
12 to them a resolution of comments document, an RCD document,
13 which provided -- addressed each one of their earlier
14 comments and provided the basis and resolution for those
15 comments.
16 DR. WALLIS: You were doing these calculation
17 methods and exercising them and getting results all this
18 time. What was industry doing? Were they also using,
19 developing calculation procedures?
20 MR. CAREW: Well, industry was developing this,
21 but industry was also, I might just say, at this time that
22 during this period there were several submittals that
23 actually referenced the methods that were developed here and
24 actually applied those methods.
25 DR. WALLIS: And used those methods.
. 344
1 MR. CAREW: And used those, and in addition, we
2 have now received just in the last few months a BWR
3 submittal which has not only used the methods that we have
4 developed here in the Reg Guide, and they have referenced it
5 extensively, but they have also taken the benchmark problem
6 that we have developed and released for the first time in
7 September '99, they have submitted a topical report which
8 has taken that benchmark problem --
9 DR. WALLIS: Essentially the methods were
10 developed by a contractor for the U.S. Government, so they
11 were not developed independently by industry?
12 MR. CAREW: No, I wouldn't say that. B&L did
13 not --
14 DR. WALLIS: That is the impression I got.
15 MR. CAREW: We did not develop all these methods.
16 DR. WALLIS: You did not?
17 MR. CAREW: These methods are a compilation of a
18 lot of history dating back for the last 30 years.
19 DR. WALLIS: So you coordinated them in some way.
20 MR. CAREW: Right. We did develop some of these,
21 but I wouldn't say --
22 MR. LOIS: The Government paid for the development
23 of the standards, the benchmarking.
24 MR. CAREW: In any case, we provided this -- we
25 provided the resolution of comments document to the industry
. 345
1 in March, and to the stakeholders for their review.
2 And we received industry's comments in May, and in
3 August of this year, we have completed the evaluation of
4 those comments and incorporated the changes into the Guide
5 as appropriate. We expect to or hope to release the Guide
6 in early 2001.
7 Okay, now, let me discuss at this point, the
8 comments that we received from industry in May. First of
9 all, we had comments from the Nuclear Energy Institute, Penn
10 State University, Bechtel, and from the smaller
11 organization, Don't Waste Michigan, DWM.
12 MR. JONES: Excuse me, John. I need to interject
13 --
14 DR. APOSTOLAKIS: Don't waste what?
15 VOICES: Michigan.
16 MR. JONES: I need to interject at this point that
17 the Hopkins comments from Bechtel were informally received,
18 and, therefore, you didn't get them in your package. That's
19 why you don't have them in your package.
20 MR. CAREW: This is a group, a small group
21 concerned about one specific reactor. But in any case, the
22 comments concerned editing, organization, methods, and
23 qualifications.
24 There were comments that both requested tightening
25 requirements and relaxing requirements. The resolution
. 346
1 included a response to all comments, and a decision basis
2 for each comment.
3 The response was based on a team consensus and the
4 resulting changes from those comments have all been
5 incorporated into the Guide.
6 DR. WALLIS: Who was it that looked at whether or
7 not the methods were accurate and viable and correct?
8 MR. CAREW: Well, I would say that we first
9 published this with most of the methods, I'd say 95 percent
10 of the methods, back in 1993, and since then we've
11 incorporate -- we had many comments, and to be honest about
12 it, there were very few that had an impact in changing the
13 methods.
14 The methods we had in 1993 are basically the ones
15 we have today, except for the Monte Carlo applications. So
16 industry saw those comments. They were widely distributed,
17 and actually most of them are being used by industry now.
18 And so industry commented on them, and we had an
19 internal team put together that commented on them. And you
20 will see in one of my last slides where we have some pretty
21 significant -- we have comparisons that compare the various
22 methods that I think will indicate that they are basically
23 correct.
24 CHAIRMAN POWERS: That they work?
25 MR. CAREW: Yes.
. 347
1 DR. WALLIS: If the ACRS would review this, it
2 wouldn't find any mistakes?
3 MR. CAREW: I don't think so.
4 CHAIRMAN POWERS: One of the things that's
5 remarkable about this slide is, did you ask people in Europe
6 or Japan or places like that to comment on this?
7 MR. CAREW: People in Europe have seen it. AEA,
8 in particular, has looked at the Monte Carlo part of it.
9 We've gotten comments. We've actually sent this to AEA, I
10 personally sent this to AEA and asked them for comments.
11 AEA has looked at it. I'm trying to think -- and it was
12 presented at international meetings.
13 It was presented at an ASTM meeting in Europe back
14 in 1993.
15 MR. LOIS: In 1996.
16 MR. CAREW: In 1996, so it's been --
17 DR. WALLIS: So it's out in the open?
18 MR. CAREW: It's definitely out in the open.
19 DR. SEALE: These methods are being used in the
20 criticality calculation community and a lot of other places,
21 all over the world, actually.
22 MR. CAREW: Okay. I might just, in terms of a
23 summary, I might just indicate that the types of comments we
24 had were as follows:
25 About 15 percent concerned -- these were the
. 348
1 comments we received from industry in May -- 15 percent were
2 concerning updates of the references; 25 percent concerned
3 the wording we selected; 26 percent requested clarification
4 of the wording; nine percent requested additional guidance;
5 and 11 were typographical; and 14 percent were technical
6 comments.
7 DR. WALLIS: If you were to cast that picture in
8 terms of how much work was required to resolve them, I would
9 suspect the wording ones would probably be resolved.
10 MR. CAREW: Some of them.
11 DR. WALLIS: Some of the technical ones might take
12 longer.
13 MR. CAREW: I can tell you they were a lot more
14 difficult. There were ones where we had to go get or find
15 references that were like 30-40 years old. There were
16 wording changes. Sometimes the wording would iterate many
17 times amongst ourselves, et cetera, and we're still doing
18 that.
19 You'll see in one of the comments -- clarification
20 was fairly straightforward. The additional guidance, even
21 that has been very sticky, as you'll see in one of the
22 comments we're still dealing with.
23 Typographical were easy, and the technical were
24 fairly -- you know, weren't that bad either. But as I said
25 before, the comments that really hung us up were the
. 349
1 requests to extend it to BWR application and the inclusion
2 of Monte Carlo.
3 Let's see, okay, the approach we took is as
4 follows in resolving the comments: First, every comment,
5 all comments, every single comment was thoroughly evaluated
6 for possible inclusion.
7 The requirements for inclusion of a comment was,
8 first, it had to be consistent with the scope and purpose of
9 the Guide. It had to be technical valid, a correct comment.
10 It had to be a significant part of the fluence
11 determination.
12 It had to be consistent with accepted
13 methodologies. If someone proposed something and it had not
14 been reviewed by NRR, or accepted, wasn't commonly -- it
15 would not be included, although for the Monte Carlo, we
16 allowed -- there, we deviated a little bit, and even though
17 that method had not been accepted at the time, we included
18 it.
19 The level of detail should not be overly
20 prescriptive, and allow the sufficient freedom to apply
21 engineering judgment to address plant-specific unique cases
22 which we can't foresee.
23 The response was based on a team consensus, and
24 that's it.
25 Let me see, okay, now, at the back of the handout,
. 350
1 I've included all of the comments that we have for the
2 Subcommittee. Of those comments, there were two that were
3 selected to be of significant concern that should be
4 considered for this Committee, and I've included these, too.
5 And the first comment is the following: Is that
6 the Regulatory Guide does not -- the methods are not
7 applicable to BWRs because of the large, extensive,
8 three-dimensional spatial dependence of the voids and the
9 fuel compositions in BWRs.
10 And our response is, first, that even in a PWR, as
11 you go and move around the core, either radially or axially,
12 because of the initial enrichment and the burnup, the
13 inventory of the fissile isotopes varies substantially, even
14 in a PWR.
15 And the Regulatory Guide has a method for
16 determining, using exposure in the core, the fuel design,
17 the burnup, to determine what the fissile isotopics are, and
18 in determining what fraction of the neutrons are borne and
19 what fraction of fissions occur, the U-235 versus the
20 Plutonium-239, and given that fraction, what the
21 corresponding yield of neutrons is, and what the energy
22 distribution is.
23 So the guide has that methodology built into it,
24 and the fact that in a BWR you have this additional
25 complexity in the core because of a large amount of voids,
. 351
1 you have a harder spectrum, more Plutonium-239, that is
2 really well within the capability and the limits of the
3 methods that are in the Guide.
4 So this, the fuel composition, is really not a
5 significant concern, and it's well within the methods that
6 are in the Guide.
7 CHAIRMAN POWERS: If I had a MOX core?
8 MR. CAREW: No problem.
9 CHAIRMAN POWERS: No problem?
10 MR. CAREW: No problem.
11 Okay, now, with respect to the void distribution,
12 indeed, there is an added complexity in BWRs concerning the
13 void distribution and the attenuation of the neutrons out to
14 the vessel. However, this is addressed and identified in
15 the Reg Guide as a specific item to be concerned about.
16 In addition, the application of the methods of the
17 Regulatory Guide have been applied in one of the benchmark
18 problems in NUREG 6115, which I'll discuss in a few minutes,
19 and you'll see that the methods are fully applicable to the
20 BWRs.
21 Okay, the next comment was kind of a sticky
22 comment, and I'll -- let's see, the comment is the
23 following: The Regulatory Guide should include guidance on
24 the selection of the weighting factors used to determine the
25 fluence calculational uncertainty.
. 352
1 In other words, the PTS rule requires that the
2 calculational uncertainty be 20 percent or less. Now, the
3 Guide has a qualification procedure in which that 20 percent
4 is determined by a weighted sum of the uncertainty -- the
5 uncertainty in the calculation is a weighted sum of the
6 uncertainty determined with the analytic uncertainty
7 analysis, the uncertainty determined by the comparisons to
8 operating reactor measurements, and the uncertainty based on
9 comparisons to the simulator measurements.
10 And the question -- and the commenter is asking
11 the question --
12 DR. APOSTOLAKIS: What, exactly, do you mean by
13 uncertainty? What's sigma?
14 MR. CAREW: Okay, let me finish the sentence. And
15 the commenter was asking the question, how do you determine
16 these weights? Now, what is sigma?
17 Sigma is the one sigma standard deviation in the
18 fluence prediction, determined by either this uncertainty
19 analysis, analytic uncertainty analysis, or based on the
20 comparisons to measurements.
21 DR. APOSTOLAKIS: And you are taking a linear
22 combination of standard deviations?
23 MR. CAREW: Right. I'm saying that I have three
24 --
25 DR. APOSTOLAKIS: Isn't that very unusual?
. 353
1 CHAIRMAN POWERS: Yes, it's very unusual.
2 MR. CAREW: That's very unusual.
3 [Laughter.]
4 MR. CAREW: No, no, this is -- what we have is, we
5 have three estimates of what the uncertainty is, and these
6 are the values, and what we're taking is a weighted mean,
7 what's called a weighted mean of those three estimators.
8 DR. APOSTOLAKIS: But wouldn't it be more
9 appropriate to take the variances?
10 DR. WALLIS: W is partial differentials.
11 MR. CAREW: No, forget that these -- forget what
12 these are, if you want. Make believe these are velocities
13 or something, and the question is, do you want to find out
14 what -- if you have three estimates of what the velocity is,
15 how do you determine what your best estimate is?
16 And the answer would be, one answer would be,let's
17 take a weighted mean and weighted average of these three
18 values, and the question then would be, what is the -- what
19 do you weight that with?
20 DR. WALLIS: There are three independent
21 assessments.
22 DR. SHACK: You're not adding variances; it's not
23 a contribution for this and a contribution for that.
24 MR. CAREW: Right.
25 DR. WALLIS: He's asking three people.
. 354
1 DR. SHACK: It's an expert judgment.
2 MR. CAREW: Right. I have three values, three
3 people to tell me what the velocity of the car is, and I'm
4 saying, okay, how do I know what the right velocity is? I'm
5 going to take a weighted mean, a weighted average, and now
6 the question is, how do I determine the weight? How do I
7 assign those weights?
8 CHAIRMAN POWERS: I mean, there are bigger steps
9 than this, it seems to me. You make a real assumption when
10 you come in here and say I'm going to take an arithmetic
11 average, even if it's weighted.
12 MR. CAREW: Excuse me?
13 CHAIRMAN POWERS: If I'm going to take -- you're
14 making some sort of an assumption when you say I'm going to
15 take an arithmetic average, even when you weigh it. I mean,
16 you could tend something else.
17 MR. CAREW: I agree.
18 CHAIRMAN POWERS: Why this?
19 MR. CAREW: Okay, let me finish. Okay, this is
20 the -- this is what is called the weighted mean, and the
21 question is, how do I determine the weights; that was the
22 question.
23 CHAIRMAN POWERS: I assume the sigma-zero at the
24 bottom is really a sigma-S.
25 MR. CAREW: You've got it. That's a typo, very
. 355
1 good.
2 Let's see now, all right, when we wrote the guide
3 -- well, let me just say this: In practice, these weighting
4 factors depend upon the application-specific details which
5 can vary widely.
6 For example, if you're doing the weight, the
7 weight you would assign to the analytic uncertainty analysis
8 will depend on the accuracy and completeness of the input
9 uncertainties.
10 If, for example, some plant -- you don't have
11 vessel diameter measurements of it -- well, if in some
12 plant, you do have vessel diameter measurements
13 uncertainties measurements available, then for that plant,
14 you may have 24 radially and axially -- asmuthially and
15 varying measurements, which would tell you, give you a
16 pretty good idea of what that vessel diameter is, which is
17 one of the dominant uncertainties in the vessel prediction.
18 In addition, if you had a very good estimate of
19 what the number of neutrons being produced in those
20 peripheral bundles of the core, again, you'd have a pretty
21 good estimate of what the analytic -- what the uncertainty
22 was.
23 So in that case, the input, the values you're
24 going to input in that analytic uncertainty determination
25 are very good, and you would expect that your analytic
. 356
1 uncertainty analysis would provide you with a pretty good
2 estimate of what the uncertainty really was, and it would be
3 a reliable estimate, and you probably in that case would
4 want to give that or assign that a high weight.
5 DR. WALLIS: I would think you would want to do
6 something like use the reciprocals, because the ones you're
7 most uncertain about are the ones you want to throw out.
8 MR. CAREW: Yes, and I'm coming to that, and
9 that's exactly right.
10 Okay, now --
11 DR. WALLIS: But you start with the reciprocals.
12 MR. CAREW: Let me see, now, on the other hand, if
13 you were in a situation -- another factor would be if you
14 were trying to assign the weight to the uncertainties, based
15 on those comparisons to measurement, and if the measurement
16 data was very good, you'd assign them a high weight.
17 If, on the other hand, you had a plant where the
18 measurement data was very bad, you assign them a low weight.
19 And then in addition, another factor you might
20 want to consider is that if you had one of those estimates
21 where the quality assurance was very poor, and had a very
22 complete story and one estimate had very poor quality
23 assurance, you might want to assign that estimate a low
24 weight.
25 So, anyway, the point here is that in practice,
. 357
1 the weight you would assign to those different estimators
2 would depend upon the situation you're in.
3 DR. WALLIS: How about the difference in the
4 estimates themselves? You can get different estimates.
5 MR. CAREW: But that's a cheat. You can't look at
6 the estimates. You have to base the weights on how good you
7 think those values are, and how good -- what the uncertainty
8 in those estimates is, not on what the values are.
9 Because of the fact that the weights depend on the
10 very specific application and these other factors --
11 DR. WALLIS: You get the answer you want by
12 varying the weights.
13 MR. CAREW: Right.
14 DR. WALLIS: That's not --
15 MR. CAREW: That's cheating.
16 DR. WALLIS: -- a very good approach.
17 MR. CAREW: Right, okay, but because of the fact
18 that these weights depend on the specific situation, in the
19 Regulatory Guide we intentionally did not specify those
20 weights. In order to allow the analyst sufficient
21 flexibility to accommodate all cases and let him use his
22 engineering judgment, giving him -- having confidence that
23 he can use good engineering judgment and assign reasonable
24 weights to those --
25 DR. WALLIS: That's a bad word though. That's
. 358
1 almost equivalent to "fudgery" --
2 MR. CAREW: Well, if you want to fudge, this guy
3 does not prevent you from fudging, but in any case, let me
4 continue.
5 This is what our intention was. We wanted to
6 leave it up to the analyst, let him have the freedom to use
7 his best engineering judgment and make good assignments of
8 those weights.
9 Well, after getting this comment we decided to put
10 additional guidance in the Guide. Basically we have
11 actually included an example of how to assign weights in
12 this situation.
13 DR. APOSTOLAKIS: And what is it? These are --
14 the ultimate result is a standard deviation, right?
15 MR. CAREW: Excuse me?
16 DR. APOSTOLAKIS: The ultimate result of the
17 weighting is a standard deviation?
18 MR. CAREW: Right. That is what we are trying to
19 estimate.
20 DR. APOSTOLAKIS: What is the point estimate? I
21 mean you are giving me the standard deviation, about what?
22 MR. CAREW: Oh, the fluence prediction.
23 DR. WALLIS: The predictions, the three
24 predictions.
25 MR. CAREW: Well, it gets complicated.
. 359
1 DR. WALLIS: Why?
2 MR. CAREW: The question is this. I am trying to
3 predict the fluence.
4 DR. APOSTOLAKIS: Right.
5 MR. CAREW: And the PTS rule requires that I
6 predict that fluence to within 20 percent, so I have my
7 methodology and I want to assure that when I use this
8 methodology I can predict the fluence to within 20 percent,
9 so what I want to do is I want to make an estimate of what
10 my fluence prediction uncertainty is, what the sigma for
11 that value is, so the question is what is my sigma?
12 DR. APOSTOLAKIS: What is your prediction? You
13 have got three predictions. What does the prediction come
14 from?
15 MR. CAREW: Hold on -- let me go back.
16 This represents the sigma for my fluence
17 prediction. That is one prediction. Now --
18 DR. APOSTOLAKIS: And the prediction is?
19 MR. CAREW: I am trying -- now a second prediction
20 would be predicting what the uncertainty is, the sigma is,
21 for my fluence prediction.
22 DR. WALLIS: The prediction is just the analytical
23 one, isn't it?
24 MR. CAREW: Now -- but I have three estimates of
25 what these uncertainties are.
. 360
1 DR. WALLIS: Right.
2 MR. CAREW: Now let me go through the example.
3 Maybe that will clarify it.
4 So we have added to the Guide an example. We said
5 okay, and this is what the example is, basically. It says
6 the following.
7 In practice I want to determine what the sigma is,
8 what the uncertainty in my fluence prediction is. Let's
9 suppose I have three estimates. I have an estimate based on
10 the analytic uncertainty and because I have good data I feel
11 that that analytic uncertainty prediction, sigma A, is good
12 to 15 percent. Okay?
13 Now suppose I also have -- for various reasons I
14 have an estimate of what that sigma is based on comparisons
15 to operating measurements.
16 DR. WALLIS: Of what? Of capsule measurements of
17 fluence?
18 MR. CAREW: Yes, right.
19 DR. WALLIS: First of all, I would like to know if
20 my prediction agrees or it doesn't agree with the
21 measurement, not just what the uncertainty is.
22 MR. CAREW: Yes, I start out by saying in this
23 exercise I am determining what the -- I want to determine
24 what the accuracy of my prediction is based on the operating
25 reactor measurements, so I have all these capsule dosimeter
. 361
1 measurements and I have a calculation and I predict them,
2 and I get --
3 DR. WALLIS: If your prediction is off from the
4 measurement by 50 percent then all these uncertainties
5 become rather irrelevant. Don't they?
6 MR. CAREW: No. No, no, no -- that means --
7 DR. WALLIS: Oh, you are separating bias --
8 MR. CAREW: Bias are out right -- uses a sigma --
9 so in that case that you just mentioned, right, suppose I
10 make all my comparisons to the capsule measurements and I
11 have 10 measurements. I get the deviations. I get the
12 average sigma, and suppose the sigma is 50 percent, okay? --
13 so that probably, so now the question is those operating
14 reactor measurements standard sigma was 50 percent. That
15 sigma O was 50 percent, and now the question -- let's
16 suppose in this case that the accuracy of that estimate of
17 50 percent was 20 percent, okay?
18 So the sigma on my sigma is 20 percent. In other
19 words, my fluence prediction is good to 50 percent. That's
20 both my estimate and now I know I have a 20 percent
21 confidence in that value.
22 Finally, I make comparisons to the simulator, and
23 the estimator of the simulator is 25 percent.
24 I think it is best to change the discussion here
25 to velocities, so suppose I have three velocities now.
. 362
1 Forget the sigmas.
2 CHAIRMAN POWERS: No. That is a vector. Graham
3 will have a heart attack -- two speeds.
4 MR. CAREW: All right, two speeds.
5 But suppose I basically have three speeds. I have
6 asked three different people what the speed is and they have
7 given me three different numbers, and I suppose I trust the
8 first guy and I think his number is good to 15 percent, the
9 second person to 20 percent, and the third fellow to 25
10 percent.
11 Now what you do in that situation, one way to
12 solve that problem is to say I have three estimators, three
13 different estimators, and I want to determine what the best
14 value is, so I will use a weighted mean in which the weight
15 is determined by one over the sigma squared of each
16 estimator, and that is what we are doing here. One over
17 sigma squared --
18 DR. WALLIS: You are ending up with one over
19 sigma, because you are multiplying it by sigma.
20 MR. CAREW: One over the variance or one over
21 sigma squared -- no, but this is a different sigma.
22 DR. WALLIS: Oh.
23 MR. CAREW: That is why you got velocities. It is
24 velocity over the sigma squared.
25 Okay, so in that case if you have three different
. 363
1 velocity speeds, and one is good to 15 percent, and one to
2 20 percent, and one to 25 percent, you get a weighted mean,
3 a weighted average. You weight each speed by one over 15
4 percent squared, one over 20 percent squared, one over 25
5 percent squared, and you have to renormalize them so all the
6 weights are one.
7 DR. WALLIS: This is a very pretty pattern. Does
8 it have a logic to it?
9 MR. CAREW: Yes. This is --
10 DR. WALLIS: Does it have a mathematical proof to
11 it?
12 MR. CAREW: Yes. It's what you get -- it's
13 maximum likelihood that --
14 DR. WALLIS: So you are using a principle, you are
15 not just creating a picture?
16 DR. APOSTOLAKIS: This is now different. You said
17 that I have three estimates of the speed to within 15
18 percent, 20 percent and so on, so what is my best estimate
19 of the speed? So I will take the first estimate of the
20 speed times the square of the inverse of this, so now I have
21 a best estimate of the speed.
22 That is not what you said earlier. You said you
23 get an estimate of a standard deviation.
24 MR. CAREW: I changed the language to speed
25 because there were several sigmas involved there.
. 364
1 I was trying to make an estimate of what the
2 variance was -- the standard deviation was.
3 DR. APOSTOLAKIS: So the Ws then on the slide,
4 whatever --
5 MR. CAREW: Was one of a sigma squared.
6 DR. APOSTOLAKIS: But that sigma squared is not
7 the same as a sigma that appears in the --
8 MR. CAREW: Exactly. Right. That is why I
9 changed the language to speed because I didn't want to
10 confuse it, so there's two sigmas. There's a sigma -- in
11 other words I do an analytic uncertainty analysis, I do
12 sensitivities and I come up -- I do, I propagate through the
13 problem and I come up with an estimate of what the sigma is.
14 Say it is 15 percent -- no, say it is 50 percent. My
15 analytic uncertainty estimate says my fluence prediction is
16 good to 15 percent but the sigma on that 50 percent is, say,
17 15 percent, so I weight that sigma with one over 15 percent
18 squared.
19 DR. APOSTOLAKIS: But still what is your best
20 estimate of the actual fluence, not the uncertainty, the
21 fluence itself? That is the analytical --
22 MR. CAREW: That is a different question.
23 DR. APOSTOLAKIS: Yes, I know.
24 MR. CAREW: Okay. Let me finish. The best
25 estimate of the fluence prediction is I simply carry out all
. 365
1 the steps in the Regulatory Guide and it will give me the
2 best estimate fluence prediction.
3 DR. APOSTOLAKIS: The analytical calculation, what
4 you call analytical calculation?
5 MR. CAREW: Right. The discrete ordinates
6 transport calculation.
7 DR. APOSTOLAKIS: Right.
8 DR. WALLIS: That gives you an estimate. There is
9 no criterion for best being used.
10 MR. CAREW: It is a set of rules.
11 DR. APOSTOLAKIS: Maybe I am wrong. Let's go back
12 to that other slide with the sigma and the weighted sum.
13 So I do the calculations --
14 MR. CAREW: Right.
15 DR. APOSTOLAKIS: No, forget about sigma -- and I
16 get a value for the fluence, then I do this --
17 MR. CAREW: Wait. Let me finish --
18 DR. APOSTOLAKIS: I don't?
19 MR. CAREW: You're right. You do the -- you get
20 the value of the fluence and then I have the best estimate
21 of fluences.
22 DR. APOSTOLAKIS: Right.
23 MR. CAREW: Then I have to make sure that I
24 conform to PTS and get the uncertainty in the value.
25 How do I get that?
. 366
1 DR. APOSTOLAKIS: How do I get that?
2 MR. CAREW: Now there's three steps you have to
3 carry out. You first take your code and do some
4 perturbation calculations, and determine analytically what
5 your -- your error propagation calculation and you determine
6 what the uncertainty is.
7 DR. WALLIS: Suppose you have got a perfect
8 experiment and there was no sigma for the experiment. Then
9 it would be weighted to be 1 and the others would be zero by
10 your method, and so the only uncertainty in your calculation
11 has to do with the uncertainty in the experiment. Doesn't
12 make sense -- which is none.
13 MR. CAREW: It's true --
14 DR. WALLIS: You have a perfect experiment and you
15 automatically have a perfect calculation -- doesn't make
16 sense.
17 MR. CAREW: No, you are saying -- consider the
18 case where you have one estimator --
19 DR. WALLIS: The experiment is perfect.
20 MR. CAREW: One estimator where the estimate is
21 perfect and the variance is zero. You divide by the
22 variance. The weight is infinite. That is the way it
23 should be.
24 DR. WALLIS: Yes, but it doesn't say anything
25 about the quality of your calculation, the fact that you've
. 367
1 got a perfect experiment.
2 MR. CAREW: No, no, no. That wasn't a perfect --
3 the sigma was not for the sigma on the experiment. The
4 sigma --
5 DR. WALLIS: The calculation, the number that you
6 give and say this is the fluence that you have calculated.
7 I am saying that the fact that you got a perfect
8 experiment is no measure of the quality of your experiment.
9 MR. CAREW: Let's see -- you are saying that
10 suppose that --
11 DR. WALLIS: The experiment is perfect.
12 MR. CAREW: In this context what that means is the
13 following.
14 DR. WALLIS: -- zero is one and the others are
15 zero.
16 MR. CAREW: Right. You are saying that let's
17 suppose you have a situation where your analytic uncertainty
18 analysis --
19 DR. WALLIS: 15.
20 MR. CAREW: -- tells you the uncertainty in your
21 fluence prediction is 50 percent, but suppose that 50
22 percent is truth. There is no error in that. That is what
23 you are saying. That would mean that the sigma in the
24 analytic uncertainty would be zero.
25 DR. WALLIS: No, I am saying the sigma in your
. 368
1 experiment, the measurements, the second one, is zero.
2 MR. CAREW: That's fine.
3 DR. WALLIS: Then I think --
4 MR. CAREW: No, this is not an uncertainty in the
5 sigma in the experiment. This is a sigma constructed by
6 comparing a set of measurements and a set of calculations.
7 The measurements may have zero error in them but
8 the calculation will differ from that.
9 DR. WALLIS: This is a combined -- it's very
10 different from the uncertainty sigma --
11 DR. SHACK: The sigma in the experiment is the W
12 sigma. The sigma that he is using here is the difference
13 between the experiment and the prediction of the experiment.
14 DR. SEALE: Yes.
15 DR. SHACK: So in your case if the W would be
16 infinite and if the experiment was perfect he would use that
17 estimate, the difference between the experiment, because the
18 experiment is perfect, and the prediction as his sigma.
19 DR. APOSTOLAKIS: There is a fundamental
20 assumption here that the analytical prediction of the
21 fluence itself remains the best estimate no matter what I do
22 with the sigmas.
23 DR. SEALE: Well, and that is one of the
24 uncertainties.
25 [Laughter.]
. 369
1 DR. SEALE: Well, of course it is. You have
2 100,000 cases or 10 million cases in the Monte Carlo, I get
3 a difference in the statistical uncertainties.
4 DR. APOSTOLAKIS: I know, but he brings other
5 elements into this.
6 DR. SEALE: I appreciate that.
7 DR. APOSTOLAKIS: So there may be a systematic
8 bias.
9 DR. SHACK: But he's got a bias -- if you go back
10 to Figure 4 he has the bias, he takes his analytic
11 prediction and if he knows he has a bias he fixes it.
12 MR. CAREW: Right. This is getting more
13 complicated than it really is. It's a very simple problem.
14 We have got three people on the street and I want
15 to know what speed the car was going. I ask Mr. A, asked B,
16 and asked C. That is the problem and they give me three
17 different numbers and I say I want to find out what the
18 speed of the car is, and all I do is I say I am going to
19 figure out, I am going to make a linear combination and
20 somehow weight those. I can rely on the maximum likelihood
21 method to tell me how to weight those values.
22 That is a rigorous method and I go and what it
23 says is that you ask the first person what is your speed and
24 what is your uncertainty of that speed. He says it is 50
25 miles an hour plus or minus two miles an hour, so his sigma
. 370
1 is one over two squared.
2 I go to the next person and say what is your
3 estimate of the speed? He says 40 miles an hour plus or
4 minus one mile an hour, so I said 40 and his weight is one
5 over one squared. I have to normalize them at the end.
6 I go to the third person. He says 75 miles an
7 hour plus or minus 25 miles and hour and I say fine, 75, and
8 I weight him one over 25 squared, renormalize all the
9 weights, and that is the answer.
10 It is a simple problem. The only question here
11 is -- I am not talking speeds. It gets a little bit more
12 complicated --
13 DR. APOSTOLAKIS: I understand that.
14 MR. CAREW: -- because I have a fluence
15 prediction. The speed is now the uncertainty in that
16 fluence prediction.
17 DR. WALLIS: Your analytic uncertainty by itself
18 is one thing. When you start comparing the experiments your
19 analytic uncertainty is also bound up in your comparison
20 with experiments, isn't it?
21 MR. CAREW: Say it again? No. These are three
22 independent estimates, completely independent.
23 DR. WALLIS: No, but doesn't the analytical
24 uncertainty come into the way in which you make comparisons
25 with experiments? Doesn't it have to?
. 371
1 MR. CAREW: No, it is completely independent.
2 The fluence prediction -- I am using the same
3 fluence prediction but when I calculated the uncertainties
4 all I am using is the -- I am taking as the independent
5 estimates of what the vessel diameter is, which never enters
6 into the other one.
7 MR. SIEBER: You are actually getting the fluence
8 prediction from the series of capsule measurements, right,
9 as opposed to external dosimetry --
10 DR. SEALE: Yes.
11 MR. SIEBER: Maybe I could ask a more fundamental
12 question just to help me understand what it is you are
13 dealing with.
14 If I am a plant manager and the plant is 25 years
15 old and I started out on day one, and you construct a core.
16 You have a bunch of capsules in the reactor. You will pull
17 one of those as your tech specs tell you to do from time to
18 time, make a measurement, and estimate the RT PTS.
19 Now sooner or later you run out of capsules, okay,
20 and so now you move to another method which is external
21 dosimetry, but you've got to integrate, in fact redo this
22 kind of a calculation all the way from the beginning of core
23 life knowing what the geometry was for each reload that you
24 had, okay? That is not real easy.
25 MR. CAREW: That's right.
. 372
1 MR. SIEBER: You have to do a steady state core
2 model for every cycle and then when you are done with that
3 you integrate all these measurements and you do a
4 calculation like this for every cycle because the geometry
5 changes from cycle to cycle.
6 In fact, it changes through the cycle as burnup
7 proceeds, so the ultimate answer is a combination of
8 different methods, okay? This is only one of them. Am I
9 correct or not correct?
10 MR. CAREW: Well, your fluence prediction that we
11 are talking about here, the basic best estimate fluence
12 prediction has all those things you just mentioned built
13 into it.
14 For a reactor you go and you do all those things
15 and you ask all the question and you get the temperatures
16 and the defuel and et cetera, and you add them all up, each
17 cycle, you add them up and you get your end-of-life fluence
18 prediction and that is the prediction.
19 MR. SIEBER: Okay.
20 MR. CAREW: And now the question here we are
21 trying to do is we are trying to say, okay, we had the
22 prediction, now how do we know how good that prediction is?
23 As I said, there's three ways.
24 MR. SIEBER: So these three weighting factors, do
25 they apply to different aspects of this integrated process
. 373
1 or is that for a single point in time for a single cycle?
2 MR. CAREW: Well, what you would do is you have
3 your prediction methodology and you would test it by
4 taking -- you would have maybe five capsules you took out
5 during the lifetime of the reactor --
6 MR. SIEBER: That would be five years or
7 something.
8 MR. CAREW: -- and you would predict each one of
9 those capsules and after that prediction you would have five
10 predictions and five measurements and you would compare
11 them.
12 From those predictions you would get the standard
13 deviation between the measurements and calculations, and
14 that would be sigma sub 0 in that slide, and that is your
15 standard deviation or your error in your prediction based on
16 those capsule measurements.
17 MR. SIEBER: Now the funny thing is when you pull
18 the first one you can extrapolate by first principles --
19 MR. CAREW: Sure.
20 MR. SIEBER: -- what the rest are going to look
21 like and then as you pull each one they don't fall on that
22 line.
23 MR. CAREW: Right.
24 MR. SIEBER: That has been my experience.
25 DR. WALLIS: I was going to suggest we move on.
. 374
1 MR. CAREW: Agreed.
2 DR. WALLIS: I think we have established the
3 principle that the ACRS is interested in rational ways of
4 estimating uncertainty and defensible methods of analysis.
5 MR. SIEBER: You have clarified my point.
6 MR. CAREW: Okay.
7 DR. APOSTOLAKIS: By the way, I don't know if I
8 like your analogy with speed. I wouldn't ask each expert to
9 tell me how confident he or she is. I would evaluate that
10 myself because experts tend to be overconfident and all that
11 so I really don't know if you are giving me an example to
12 say for an example or you are really serious about it.
13 I would have to ask Dr. Shack. Is this worth
14 really pursuing to death? How important is it? I am
15 utterly confused.
16 MR. SIEBER: What you want to do is use
17 engineering judgment, which is what the Guide always does.
18 CHAIRMAN POWERS: Give equal weighting to all the
19 experts, right. We should be applauding this guy.
20 DR. WALLIS: We are applauding you. I think it is
21 wonderful that you are not just waving your hands, you are
22 actually trying to do something analytical. And it is
23 probably right.
24 DR. SHACK: We should move on.
25 DR. WALLIS: That is probably right, we should
. 375
1 move on.
2 MR. CAREW: Yeah, that is probably the case.
3 Okay. The Regulatory Guide requires, as part of
4 the qualification procedure, that the pressure vessel
5 fluence benchmarks be calculated. And now what are the
6 pressure vessel fluence benchmarks? Well, these are a set
7 of benchmarks whose purpose is to ensure an accurate fluence
8 prediction and quantify the uncertainty, to standardize
9 fluence methods and to streamline the licensing process.
10 Now, how does this work? In the first place, the
11 fluence methods are included in the Regulatory Guide. The
12 NUREG report includes the definition of the problem and the
13 solutions. The licensee would calculate the benchmark
14 problems and compare his results to the referenced
15 solutions. Based on the agreement between the referenced
16 solutions and his predictions, the method would be accepted
17 in part, based on those comparisons.
18 Now, I can just say that I can tell you that we
19 released the benchmark problem report in a draft form in
20 September '99 and we have already had a fuel vendor apply
21 that problem in exactly this way. He has submitted a
22 topical report to Lambros, where he has compared taking the
23 benchmark problem and compared his solutions to the
24 reference solutions. So, okay, that is the purpose.
25 The report or the problems include typical types.
. 376
1 We have three types of PWRs because of the low leakage
2 fluence reduction schemes. We have a standard core, a low
3 leakage core and a partial link shield assembly core, and we
4 also have one BWR core, because they really don't use any
5 novel fuel designs for fluence reduction.
6 The problems included detailed descriptions of the
7 materials, geometry and pin-wise source distribution. I
8 mention the pin-wise source distribution because in the
9 core, on the periphery of the core, there is a gradient, a
10 steep gradient in the pin-wise distribution and that, for a
11 best estimate prediction, must accommodate that.
12 Typical operating reactor geometry materials are
13 used and the fluence, or the problem involves a complete
14 execution of all steps required for the RT PTS predictions.
15 Let's see, the PWR standard core uses typical
16 core-to-cavity dimensions and materials, complete input.
17 DR. WALLIS: What do you do about the neutron
18 source, don't you really have to solve the problem of the
19 flex distribution in the core and the energy distribution
20 and stuff first?
21 MR. CAREW: Right.
22 DR. WALLIS: You have to do the whole thing first,
23 you have to look at the details.
24 MR. CAREW: Right. Exactly. But we designed this
25 problem, so we have short-circuited that, and all we do is
. 377
1 we provide them with a XY -- with a pin-wise power
2 distribution, and an exposure distribution.
3 MR. WALLIS: Provide them.
4 MR. CAREW: We give that to them, it is
5 well-defined. And then they take that.
6 DR. KRESS: That is the benchmark.
7 DR. WALLIS: That is the benchmark.
8 MR. CAREW: Right.
9 DR. WALLIS: But in practice.
10 MR. CAREW: Well, in practice, yeah, they have to
11 go -- as he said, right.
12 DR. WALLIS: But how could their -- their overall
13 criticality and all that stuff has to be in order to get it
14 right.
15 MR. CAREW: Right. Okay. Let me see. The PWR
16 problem includes a standard wall capsule and a cavity
17 capsule. There is a map of the PWR problem, the core is in
18 red. We have the downcomer, a barrel thermal shield and
19 pressure vessel. We have got a capsule on the inner wall,
20 the back of the thermal shield, and the further vessel, and
21 a capsule in the cavity. That is the PWR problem.
22 We also have a BWR problem.
23 DR. WALLIS: Do all these things have thermal
24 shields?
25 MR. CAREW: Not all of them. Not all of them. In
. 378
1 the problem, they do.
2 Here is the BWR problem. Notice here that we have
3 the jet pumps and risers included. You have the reactor
4 pressure vessel. And I just might mention in passing, a
5 point, that if you recall, in a BWR, you have got a fuel
6 channel where we have voided moderator on inside, between
7 the bundles we have solid water. And we are going to do a
8 calculation both with the solid water and void fraction
9 region separated, and them homogenized with the Monte Carlo.
10 You will see in a few minutes.
11 I might add -- well, okay.
12 DR. WALLIS: So you have got two flow regimes in
13 there?
14 MR. CAREW: We don't -- yes.
15 MR. WALLIS: Original license is trying to
16 separate.
17 MR. CAREW: Exactly. Exactly. Here is the BWR
18 problem, here we have seven axial regions to account for the
19 distributed void fraction, reflected downcomer, pressure
20 vessel, core plate, upper separators. This is the BWR
21 problem.
22 I just might mention in passing that these
23 problems are a complete specification. So we include all
24 the dimensions, the vessel radius, inner radius, the
25 temperatures. The inlet temperature is important because we
. 379
1 are propagating the core through the -- neutrons through the
2 downcomer, and the temperature of the water, dense water is
3 more attenuating than non-dense water. Materials are
4 specified. So it is a complete specification.
5 DR. WALLIS: You ought to say barrel thickness
6 3.81 plus or minus all those things.
7 MR. CAREW: Yeah, that is a good point. That is
8 an excellent point. One advantage of this, that this
9 benchmark problem has over other calculations, is that when
10 you calculate an experiment, you know that you don't know
11 exactly what the vessel diameter is. You don't know what
12 the temperature is. You don't know what the compositions
13 are. You don't know what the source is. This is completely
14 divided down into six, seven significant figures, so it is
15 an excellent and efficient way --
16 DR. WALLIS: So these are exact.
17 MR. CAREW: Exact numbers.
18 DR. WALLIS: In your work example. But in
19 reality, you are going to have to look at all those
20 uncertainties.
21 MR. CAREW: That's right, and that is the
22 uncertainty question. Okay.
23 MR. SIEBER: Do you account in a PWR for the boron
24 and all these water --
25 MR. CAREW: As it turns out, the boron is a
. 380
1 thermal absorber, so it is not important for fast flux.
2 MR. SIEBER: Okay.
3 MR. CAREW: We specify the compositions here at
4 the core region. We have all the materials, the fissile
5 isotopes, the atom densities. The number densities are
6 specified as part of the problem.
7 The problem solution is tabulated.
8 DR. WALLIS: Do you have to worry about bouncing
9 back from the concrete shield of these neutrons?
10 MR. CAREW: You do have to worry about that,
11 especially if you are in the cavity.
12 DR. WALLIS: Yeah, but nobody worried about the
13 wall.
14 MR. CAREW: Especially when the cavity --
15 DR. WALLIS: It only affects the fluence, does it?
16 MR. CAREW: When the cavity is thin, it does,
17 yeah.
18 Let's see. We have tabulated solutions, the
19 fluence solutions presented for the flux above 1 MeV, .1 MeV
20 and DPA, displacements per atom. We have got accelerated
21 wall capsules and internal cavity locations, and we also
22 include sensitivity calculations.
23 The methods that are used to carry out these
24 calculations and determine the reference solutions are those
25 that are used in the Reg. Guide. We have S8 P3 transport
. 381
1 calculations and R theta RZ synthesis, the latest ENDF-VI
2 cross-sections for the transport, sufficient spectrum and
3 for the dosimeter cross-sections.
4 I will just go to the next. All right. The next
5 slide is for the following, is for the MCNP calculations.
6 Okay. This was the result of some of the earlier comments
7 to include Monte Carlo. We have done these calculations
8 with the discrete ordinate method, DORT method, and we have
9 also done the exact same calculations with the Monte Carlo,
10 MCNP Monte Carlo code.
11 We have done the Monte Carlo calculations for a
12 standard core, a partial links PWR standard core, a PWR
13 partial link shield assembly core, and a BWR core. The
14 advantage of the Monte Carlo calculation is you can
15 represent every surface in the problem exactly. You are
16 given the coefficients for the surfaces, so all the geometry
17 is exact.
18 We have a pin-wise description of the peripheral
19 powers and we have the ENDF-VI cross-sections, the same
20 cross-sections that are used in the DORT calculations. So
21 the DORT and the Monte Carlo have the exact same nuclear
22 data set. And I can just tell you that the Monte Carlo code
23 calculations use a variance reduction technique called
24 splitting. The flux, in these comparisons I am going to
25 show, the flux above 1 MeV is provided. And I can tell you
. 382
1 that the DORT and the MCNP codes differ to within about 5
2 percent. And this is consistent with the MCNP statistics,
3 which are about 5 percent.
4 And due to the DORT geometry, in the DORT
5 calculation you represent the problem in R theta geometry,
6 polar coordinates, when, in fact, it is Cartesian, so there
7 is an error in doing that. There is DORT numerics, the
8 iterations, convergence, whatever, selection of the mesh,
9 numerics, and the fact that you have a R theta and RZ
10 synthesis, you have got a combined, that also induces an
11 error.
12 And you are going to see that these calculations
13 agree to a very, very good agreement. And I think this
14 answers -- I think you had an earlier question, how do I
15 know these methods are correct? I can just say that this a
16 very interesting exercise in that --
17 DR. WALLIS: Well, you are saying it works, yeah,
18 but I was going to say if we examine the theory, we don't
19 want to find something there.
20 MR. CAREW: Well, here, let me finish. So here we
21 have the discrete ordinate method, which discrete ties is
22 the whole problem, spatial mesh, the location of the
23 neutrons, their energy and their direction. It has all
24 kinds of approximations of doing that. And then we have an
25 entirely different code. All kinds of different -- a whole
. 383
1 different set of approximations to track these things
2 statistically and ensure that these variances are minimized.
3 We take these two codes and we go out and we do --
4 the Monte Carlo calculations took a month, 400 million
5 histories, and we find that they agree. And this basically
6 confirms that the methods in the regulatory --
7 DR. WALLIS: It gets better and better as you go
8 along.
9 MR. CAREW: If you have got the money, we will
10 keep running the calculations. But anyway, we come back and
11 we compare the results and they agree. And this basically
12 confirms that what is in the Regulatory Guide, these two
13 different techniques agree and the methodology is correct.
14 Now, the first case I will show you is the PWR
15 core, the flux above 1 MeV, and the results are in very good
16 -- this is a function of -- this is a neutron -- flux above
17 1 MeV as a function of angle, azimuth, and the results, as I
18 can tell you, are very good.
19 The next calculation is a calculation of a partial
20 link shield assembly core in which -- this is a core where
21 there is a weld opposite in this lower region here, and the
22 core design, in the core design, the fuel in the bottom
23 3-1/2 feet of this core had been replaced by stainless steel
24 rods. And so this is an extreme case with an extreme axial
25 flux distribution, and it would be an extreme test for the
. 384
1 synthesis method. And you will see in the next curve that
2 the synthesis method agrees very well, especially about here
3 where the weld is where it is of most concern.
4 On the next slide I have a comparison of the BWR
5 calculations. And, again, the flux above 1 MeV as a
6 function of azimuth, and, again, the results are very, very
7 good for using two entirely different methods, codes, et
8 cetera.
9 MR. SIEBER: You get downcomer and jet pump areas
10 there.
11 MR. CAREW: Yeah.
12 MR. SIEBER: That's all right. I know what --
13 MR. CAREW: It's okay. It is an interesting
14 question.
15 MR. SIEBER: I know what they look like.
16 MR. CAREW: This has all that in there. You can
17 see that, if you will notice, you can see here that what it
18 controls -- it is kind of interesting. What controls the
19 vessel fluence is the following, it is this distance. As
20 you might expect, the distance between the core boundary,
21 the fuel, the source and the sink, the location of interest.
22 And here you can see there is a minimum distance here, so we
23 would expect the peak here, another peak here, and another
24 peak here. So there is three peaks, and that is exactly
25 what you will see.
. 385
1 There is a peak at 45 degrees, there is a peak
2 here, and a peak here. And these are just really tracking
3 the core periphery geometry. And, yes, this calculation has
4 the risers, the jet pumps, all that detail in there.
5 MR. SIEBER: But they really don't mean much as
6 far as the shape of the flux, as opposed to the shape of the
7 core itself, core geometry.
8 MR. CAREW: You mean the --
9 MR. SIEBER: The modeling of the downcomer, jet
10 pumps, risers.
11 MR. CAREW: Well, I can say this, that the
12 presence of the jet pump, if we take the jet pumps out, the
13 flux goes up by about 15 percent, so it has an affect.
14 MR. SIEBER: Okay.
15 MR. CAREW: So, you know, all those things do.
16 But this was a calculation I was alluding to before. Here
17 is a calculation where in one case we have taken the
18 contents of the fuel channel and the bypass region between
19 the channels, between the fuel elements, and homogenized it.
20 And the other case, we have a case where we have had the
21 discrete geometry explicitly, where the voids are inside the
22 channel, the channel box, and then we have solid water on
23 the outside. And we were concerned that this might be some
24 kind of a streaming effect down the channels or
25 what-have-you.
. 386
1 It turns out that this case here is the
2 homogenized case. The dotted curve is the explicit water
3 case. And we ran this for 400 million histories, and we
4 couldn't see any difference. So it was kind of a null
5 result, but maybe we will just do it once.
6 That basically finishes what I was going to say.
7 If there are any questions, I would be glad to answer them.
8 CHAIRMAN POWERS: Well, I will just say that I
9 think you have done a fantastic job here of getting a
10 standard. It looks like you have killed this problem. Most
11 researchers usually complete their presentations by saying
12 and now what work needs to be done. Can you give me a hint
13 on what work needs to be done?
14 MR. CAREW: You took the words right out of my
15 mouth. No, but, you know, you are right, we have answered a
16 lot of the questions, I have to admit. But there are some
17 interesting aspects.
18 The Monte Carlo calculation is interesting, but is
19 very expensive and is not talking -- well, anyway. It is
20 interesting that no one really -- there are lot of people
21 that do these calculations and no one has really shifted
22 over to Monte Carlo because it is so expensive and long
23 running time. But they run, the other calculation runs in
24 an hour or so. And this certainly was nice to able to give
25 us a lot of assurance, but these other facts that we left
. 387
1 out of the other modeling, et cetera, the synthesis
2 procedure, they are correct.
3 Thank you.
4 DR. SHACK: There is nothing like a better
5 calculation to show you how good your approximation is.
6 CHAIRMAN POWERS: Okay. That completes these
7 session. Oh, we have got --
8 MR. JONES: I just want to show you, go through
9 the last slide. In the little package, it would be the
10 summary. And I think the level of effort that John has
11 made, I mean that is really representative of the work that
12 has gone into this. I have only been involved for the last
13 year or so.
14 I think the benchmark calculations provide an
15 added dimension that lets people really verify their
16 methods. We have had a lot of substantial public
17 participation and we have had a lot of comments. And I
18 think, factoring that in, the guide is ready, and that is
19 really what that slides says. And if you have questions, we
20 will answer them.
21 DR. WALLIS: Well, the calculations are very nice
22 and the methods are very impressive. How does it compare
23 with some sort of measurements?
24 MR. JONES: How does it compare?
25 DR. WALLIS: How do any of these things tie in
. 388
1 with the measurement?
2 MR. LOIS: This is Lambros Lois. We see this
3 every day, Dr. Wallis. For a benchmark code, they compare
4 very, very well. Of course, once in a while you see some
5 outcomers that are out of extreme -- I should say,
6 measurements, but most of the time we find why those
7 deviations exist. But by and large, the comparison is
8 pretty good. Pretty good in this case --
9 DR. WALLIS: What is your measure of very good?
10 MR. LOIS: Less than 20 percent, one sigma. That
11 is my definition of this case.
12 MR. JONES: If there are no other questions?
13 MR. SIEBER: Just one other question. It has been
14 my experience that perhaps some utilities do their own
15 calculations, but a lot of fuel vendors do that, you know.
16 I couldn't afford to keep a staff to do stuff like this and
17 so I would hire it out. Did you get a lot of comments from
18 the fuel vendors?
19 MR. JONES: In your package, there are a group of
20 NEI comments. There are fuel vendor comments in there. And
21 fuel vendor representatives have been active in the public
22 meetings and have contributed comments, yes.
23 And I believe at least two vendors have submitted
24 methods nearing what is in the guide, that I am aware of.
25 MR. LOIS: Can I add something to that? Two of
. 389
1 the vendors, one of them has already been approved. There
2 was one utility who has submitted their own methodology, and
3 we have right now in-house reviewing another vendor.
4 MR. SIEBER: Thank you.
5 MR. JONES: I want to thank you for your time and
6 we appreciate your comments.
7 DR. SHACK: And I have a one page maximum
8 likelihood analysis of the best estimate, if you have three
9 estimates. And it is, of course, a weighted average of the
10 three.
11 MR. SIEBER: So how fast were they going?
12 DR. SHACK: Sigma squared, normalized sigma. So
13 there is a principle.
14 CHAIRMAN POWERS: Thank you. I think at this
15 point we can dispense with the transcription.
16 [Whereupon, at 4:29 p.m., the meeting was
17 recessed, to reconvene, at 8:30 a.m., Friday, December 8,
18 2000.]
19
20
21
22
23
24
25
Page Last Reviewed/Updated Tuesday, July 12, 2016