ML20195G936
| ML20195G936 | |
| Person / Time | |
|---|---|
| Issue date: | 06/21/1988 |
| From: | Advisory Committee on Reactor Safeguards |
| To: | |
| References | |
| ACRS-T-1677, NUDOCS 8806280176 | |
| Download: ML20195G936 (115) | |
Text
R06S7= M 77 F
ORlSWAL
~
UNITED STATES i
o NUCLEAR REGULATORY COMMISSION
=ma..
mmmmes===mmaammmme===
m================
=============,
1 ADVISORY CO?'JIITTEE ON REACTOR SAFEGUARDS
)
In the !!atter of:
)
)
SUBCOMMITTEE ON THER!!AL HvDRAULIC
)
)
l PHENOMENA
)
)
l (Open Session)
)
)
e l
O i
is
.W,,
T%g->; RS9?.hu Q
a Ag b.. a u cr.
o 90 do1Retilove iom [-\\(",i'a 0'.'.,
,, ice -
DATE:
June 21, 1983 p
PAGES:
1 through 102 I
LOCATION:
Washington, D.C.
m m m m m m m m m m m mm mm mm m mm m m m mmmmme===
HERITAGE REPORTING CORPORATION O
- omwn, 1224 L Street, N.W., Sake ese WanWeston, D.C. 2000$
(282) 628 4088 S006280176 86:0621 PDR ACRS T-1677 D C D
1 0
1 UNITED STATES NUCLEAR REGULATORY COMMISSION 2
ADVISORY COMMMITTEE ON REACTOR SAFEGUARDS 1
3
)
In the Matter of:
)
4
)
)
5 SUBCOMMITTEE ON THERMAL HYDRAULIC
)
OPEN SESSION PHENOMENA
)
6
)
7
- Tuesday, June 21, 1988 8
Room 1046 9
1717 H Street, N.W.
Washington, D.C.
20555 10 The above-entitled matter came on for hearing, 11 pursuant to notice, at 8:30 a.m.
12 BEFORE:
HR. DAVID A. WARD 13 Research Manager on Special Assignment p\\_/
E.
I. Du Pont de Nemours & Company 14 Savannah River Laboratory Aiken, South Carolina 15 ACRS MEMBERS PRFSENT:
16 DR. WILLIAM KERR 17 Professor of Nuclear Engineering l
and Director of the Office of Energy Research 18 Univesity of Michigan Ann Arbor, Michigan f
19 MR. CHARLES J.
WYLIE 20 Retired Chief Engineer Electrical Division 21 Duke Power Company Charlotte, North Carolina 22 CONSULTANTS:
23 l
I. Catton 24 V.
Schrock l
C. Ticn 25 O
HERITAGC REPORTING CORPORATION -- (202)628-4888 J
c:.
ll 3
2
[-
1 ACRS COGNIZANT STAFF MEMBER:
2-Paul Boehnert 3
NRC STAFF PRESENTERS:
4 B. Jones 5
.i 6
7 l
8 9
10
-11 5
12 O
13 14 l
15 l
16 i
17 i
18 l
l 19 i
l 20 21 l-22 e
i 23 1
l 24 25 i
i O
HERITAGE REPORTING CORPORATION -- (202)628-4888
3 1
P R O C E E D 1 H G 1.-
)
2 CHAIRMAN WARD:
The meeting will now come to order.
3
'This is a-meeting of the Advisory Committee on Reactor 4
Safeguards, Subcommittee on Thermal Hydraulic Phenomena.
I am 5
Dave Ward,.the Subcommitteo Chairman.
6 The other ACRS members in attendance are Mr.-Kerr 7
and Mr. Wylie.
We also have ACRS consultants Mr. Catton, Mr.
8 Tien, and we expect Mr. Schrock.
9 The purpose of this meeting is to continue our 10 review-of the Westinghouse revised ECCS evaluation model for 11 the two loop upper plenum injection plants.
Most of the 12 meeting will be closed to protect information deemed to be 13 proprietary by Westinghouse.
[}
14 Paul'Boehnert on my right is the cognizant ACRS 15 staff member for the meeting.
Rules for participatiori in the 16 meeting have been announced as part of a notice previously 17 published in the Federal Register on June 8th, and again on 18 June 17th, 1988.
A transcript of the meeting is being kept.
19 It is requested that each speaker identify himself or herself 20 and speak with sufficient clarity and volume so that he or she 21 can be readily heard.
22 And for the proprietary portion of the meeting, we 23 will keep a separate transcript.
24 Let me make a few comments to review the situation.
25 We are picking up wnere we left off from the May 27th meeting.
HERITAGE REPORTING CORPORATION -- (202)628-4888
4 1
The original question at that meeting was what I will call the
()
2 inapplicability of existing codes to model the upper plenum 3
injection configuration and process for ECCS.
4 After a rather considerable period of time, and I 5
guess rather uncharacteristic patience on the part of NRR, the 6
licensee and Westing..vuse as their agent have chosen to 7
correct this deficiency, and at the same time, move to a best 8
estimate analytical approach.
9 This would be and I think is laudable.
However, it 10 is not yet opportune to move to the best estimate approach as 11 the new rule, but instead they have taken the approach we have 12 characterized as the SECY 83-472 approach, but in fact, they rS 13 have it seems to me, have stepped back half pace from that to V
14 something more like an intermediate position between the old 15 Appendix K evaluation model and 83-472.
16 Coupled with this has been what seems to have been a 17 marginally adequate review by the staff or at minimum, 18 inadequate documentation or documentation requirements placed 19 by the staff on the licensee.
20 Some of the concerns from the May 27th Subcommittee 21 meeting involve this what I will call a quasi-best estimate 22 approach, in as much as they did the original question of 23 accurate modeling of the UPI system, and for example, 24 questions such as is the licensee taking advantage of some
\\_)
25 features of the best eatimate approach without providing HERITAGE REPORTING CORPORATION -- (202)628-4888
5 1
sufficient rigor in identifying and quantifying appropriate
(
2 margins for uncertainties, which I see as a, as a burden 3
. requirement of the best estimate approach.
4 The second question might be is the regulatory staff 5
giving sufficient attention to and planning for sufficient 6
resources to provide what seems to be a more difficult perhaps 7
in terms of more judgment being required, but a more difficult 8
review process for the best estimate approach via 83-472 than 9
has been necessary in the past using evaluation models?
10 Now we have cried to define the agenda for today 11 that will answer the questions and clarify the issues and the 12 confusion about the issues that was-raised at the May 27th 13 meeting.
We have put some of the responsibility back on the
[}
14 licensee and Westinghouse to clear up what they thought the l
l 15 Subcommittee was getting wrong in, at the May 27th meeting, j
16 and we have also asked both the staff and Westinghouse to 17 restate their perceptions of the overall strategy at the 18 beginning of their presentation as a help toward getting us i
19 directed down the right road in this review today.
l 20 We have received some additional documentation and 21 addendum to the WCAP document and another, plus another thick l
22 document which provides Westinghouse answers to the staff 23 questions about Volume 1 and 2 of the WCAP document.
I guess 24 I have been puzzled as to why this what seems to me to be very 25 important, quite thorough docunentation, wasn't made available HERITAGE REPORTING CORPORATION -- (202)628-4888
.__~
6 L
-1 to us prior to the last meeting.
(f 2
Now this material was frequently referred to during 3
the previous metting.
We heard comments such as the 4
Subcommittee is asking the same questions the staff did and 5
the answers are in the WCAP addendum documents, so I am going 6
to be interested'in whether.the consultants think that this 7
document was, at least these two documenta really were 8
properly characterized at the May 27th meeting as having the 9
questions and the answers that we were looking for in our 10 previous ~ meeting.
11 In addition, Dr. Schrock was to receive further 12 documentation on decay heat modeling, and you did receive 13 that?
Good.
And Dr. Catton was to receive further
{}
14 documentation on the pipe break modeling, and did you get 15 that?
16 HR. CATTON:
I got it, t
17 CHAIRMAN WARD:
We would like to hear your comments 18 on these at the appropriate points in the meeting.
19 Okay.
Do any of the other members have comments 20 they would like to make at this time before we start?
- l I
21 DR. KERR:
I do not.
22 MR. CATTON:
I could comment on documentation right 23 now.
24 CHAIRMAN WARD:
Okay.
Do you think it would, well, l
O j
25 I will leave that up to you.
I HERITAGE REPORTING CORPORATION (202)628 4888
7 1
MR. CATTON:
Hight set the stage.
The question I (m
(_)
2 had was the same as the one that Virgil had, and that is that 3
you have a different kind of break, how do you calculate the 4
flow?
They sent me two documents, one I think if I understood 5
all those things tells me how they detect how far the breaks 6
will go.
I think that was adequate.
7 The second part of the question dealt with flow rate 8
through the break, and how did tney arrive at the, this 9
business of stopping the change in the size at point--I forget 10 what it was.
I believe it was
.5.
Lidn't answer that.
11 Didn't address it in what I had.
At least I don't believe it 12 did, but I don't think the answer was adequate.
("
13 CHAIRMAN WARD:
Virgil, do you have anything you V) 14 want to say at this point.
15 MR. SCHROCK:
Decay heat part?
You have an item for 16 it?
I will wait until that time.
17 CHAIRMAN WARD:
Fine.
John, do you have anything 18 you would like to say at this time?
Okay.
Well then, let's 19 go to the first item on the agenda, and that's Mr. Jones of 20 NRR, and Bob, will you tell us--I mean we are on, we are open 21 now, and will you tell us at any point in your presentation we 22 need to go into closed session?
23 HR. JONES:
Okay.
I guess as far as the, when it is 24 time to close, Westinghouse could assiet me.
I have a feeling 25 that from what I have here, I do not believe we will have to HERITAGE REPORTING CORPORATION -- (202)628-4888
8 1
go into a closed session except in response to some of the, (m) 2 some questions, and maybe the best way to handle that will be 3
if we come up to our proprietary information, if that can be 4
identified at that point, maybe we just put it off until the 5
end of my presentation, and go back and try to address it.
6 CHAIRMAN WARD:
All right.
Fine.
7 MR. JONES:
I have reordered or tried to interpret 8
the agenda somewhat so I, what I would like to do is first go 9
into a discussion as to what kind of a review standard we used 10 in looking at the Westinghouse model, and then I will follow 11 up with a discussion on the future EH reviews which is to 12 address your questions about having adequate resources, et 13 cetera.
{}
14 (Slide) 15 MR. JONES:
As you are well aware, Westinghouse har 16 taken a SECY 83-472 approach to evaluating or coming up with a 17 new UPI model, so I think it is appropriate to start a 18 discussion as to what are the SECY 83-472 requirements?
What 19 are in the document?
20 Basically they are listed here on the slide.
First 21 pff, the analyses ought to be performed with a quote, best 22 estimate code.
And I will, I will go into how we address 23 these standards in a few minutes.
24 HR. CATTON:
How do you decide that you have a best (3w]
25 estimate code?
HERITAGE REPORTING CORPORATION -- (202)628-4888
9 1
MR. JONES:
I will get into what we did in this j
)
2 process specifically.for the Westinghouse model.
3 MR. CATTON:
I will tell you why I asked the 4
question.
In all of the paper that I have in hand, I' don't i
5 see a single comparison of calculated heat transfer 6
coefficient measured, no scatter plots.
7 MR. JONES:
And we did not do that.
8 MR. CATTON:
I don't know how anybody, anybody'can 9
evaluate a code whose basis is heat transfer without looking-10 at the heat transfer code.
Frankly, it amazes me.
11 MR. JONES:
We have relied on judgment, knowledge, 13 you know, experience of consultants in using various codes 13 that are'out there, and their familiarity with the
[}
14 correlation.
We did not, and we did not, and I will be honest 15 with you, we do not intend to ever sit down and go line by 16 lining through codes or through each and every correlation 17 that is comparison to data.
We rely on the judgments of our 18 consultants as to whether they believe these correlations are i
19 reasonable or reasonable best estimate.
20 MR. CATTON:
You are not paying a whole lot of 21 attention to the effort that is underway or has been for the 22 past year.
23 MR. JONES:
CSAU method.
24 MR. CATTON:
But you don't look at the heat transfer 25 coefficients.
You don't know what is in the code.
You cannot HERITAGE REPORTING CORPORATION -- (202)628-4888
10 1
make a statement on the adequacy of the code until you see
()
2 them.
Unlees ivu are some--
3 MR. JONES:
We have what the correlations are.
4 Okay.
They are documented.
5 MR. CATTON:
There is a big difference between what 6
you think the correlations are, and what the code does with 7
them when it generates a number to be used internally, and if 8
you don't look at them, you have no way of knowing because 9
there are lines upon lines of logic that incorporate those 10 correlations.
Their actual use may create a different number 11 than you think it does.
Unless you look at them, you don't 12 know where you are at, unless you are an oracle.
13 MR. JONES:
We also rely--that is part of the reason
{}
14 for the code comparisons, to get a handle as to how good the 15 code is.
Let me go into that.
16 MR. CATTON:
I found these problems in TRAC.
That 17 is why I am wondering what you used to compare it with.
18 MR. JONES:
Let me go into specifically what we have 19 done.
Okay.
20 SECY 83-472 then goes on to discuss the need to 21 calculate a realistic and then a conservative PCT, whether 22 conservative PCT is supposed to be representative of the, of 23 at least the 95th percentile PCT for the large break LOCA.
24 Now part of the reason that standard came about was 25 we were in the process of looking at the GE model at the time HERITAGE REPORTING CORPORATION -- (202)628-4888
11 1
where they were using a realistic case and doing a compilation
()
2 of the uncertainties through response surfaces.
Thus the 3
realistic had a lot of importance in the overall process.
As 4
you are aware, the realistic does not have a large bearing on 5
where or how Westinghouse did their calculation, but that is 6
what is listed in the 83-472 requirements.
7 The conservative PCT is to account for various 8
uncertainties, and this is not obvicusly an exhaustive list 9
but these are the ones that were specifically called out in 10 83-472, and these are, include items like code can predict 11 experiment, what operating parameters you put into your input 12 model to reflect the plant, and in cases such as nuclear 13 parameters which are not used in the majority of the V("%
14 experiments, how you are going to treat the nuclear part of 15 the problem.
16 The actual method of dealing with these 17 uncertainties were not specified in 83-472.
Rather all it 18 says is the applicant shall develop one, and then the staff i
19 would obviously have to review it.
20 Ultimately what becomes the licensing standard under 1
21 83-472 is a new evaluation model which is developed by using 22 your best estimate code, augmented by putting in the models l
23 which are absolutely required by Appendix K such as Baker l
24 Just, and calculate a new PCT in that fashion, and to assure r
25 that the new evaluation model is, has sufficient conservatism, HERITAGE REPORTING CORPORATION -- (202)628-4888
12 1
you must demonstrate that your Appendix K calculation is t
(
2 greater than this conservative PCT but still less than the 3
criteria 50.46, 2200 degrees.
4 MR. SCHROCK:
As you presented it, sounds as though 5
you could dispense with all of the top items, and do the 6
Appendix K calculation, and that's -hat.
7 It seems to me you ought to be focusing on what is 8
the difference in this best estimate process from the old 9
rule, and what that amounted to as I recall is that it frees 10 the applicant of doing myriad calculations using evaluation 11 model out there in some Never Never Land calculating nonsense, 12 and allows him to identify what the critical problem is and
(}
13 then to demonstrate that if you do apply the old Appendix K 14 requirements, that in fact that predicts a worse result than 15 the new methodology.
16 I think you are wrong in stating that this document, 17 SECY 83-472, gave you know, guidance on how to apply the 18 conservatism to the best estimate result.
I think it gives 19 you some freedom to make engineering judgments.
It 20 acknowledges the fact that you cannot do it on strictly 21 statistical bases, but it doe;n't say that you can discard all 22 the knowledge you have of statistics, all of the information 23 that you have, and go directly to some half-baked kind of 24 argument that this is more conservative than you are going to 25 get from a more rigorous calculation.
You have not done the HERITAGE REPORTING CORPORATION -- (202)628-4888
13 1
rigorous calculation.
You have no basis for a comparison.
()
2 HR. JONES:
Let me, first off, let me just clarify 3
the Appendix K calc.
Appendix K calculation required under 4
83-472 is required by the regulations.
Okay.
That is not 5
where the primary focus of the review effort is.
6 The primary focus of the review effert is on this 7
conservative PCT because in essence that is your licensing 8
standard.
The other is your regulatory requirement.
9 HR. SCHROCK:
I think you misunderstood me.
My 10 comment about the focus related to why the Appendix K item is 11 there.
This thing was written in such a way as to allow the 12 use of best estimate methodology, still conforming to the old
(}
13 law.
Okay.
14 HR. JONES:
Yes.
i 15 HR. SCHROCK:
You imagine that you are conforming to 16 the old law by doing one best estimate calculation, one 17 evaluation model calculation, excuse me, at the conclusion of 18 the whole process, instead of doing many, many best estimate 19 calculations throughout the process.
20 Isn't that all that is saved by the application of 21 SECY 83-4727 What other advantage is there in it?
22 HR. JONES:
You stil.1 have to do a myriad of 23 calculations with your best estimate model such as breaks, 24 and--
25 HR. SCHROCK:
That makes good technical sense.
HERITAGE REPORTING CORPORATION -- (202)628-4888
14 l
1 MR. JONES:
You are ctill required under Appendix K I
()
2 rules to do a break spectrum analysis to set your, what your 3
worse case Appendix K break.
That is still required, and in 4
. fact, that's what Westinghouse had to do in that avaluation.
t 5
So it did not totally alleviate the requirements under
_6 Appendix K.
We were trying to find a way to incorporate all l
7 the research since 1973 into the licensing process without 8
changing the rule.
A realistic change was being processed at 9
the time or was being picked up again, and as you are aware, 10 we are hopefully coming to the end of that process, but this 11 was to give relief at that point in time which was, you know, 12 this document, it says in '83.
So we were trying to work 13 within the framework of the Appendix K while still allowing
(}
14 you to go best estimate.
15 The concern that we had at that time was gee, if I 16 just go back and put in what I think are best estimate models i
17 and then just throw the Appendix K required features on top of 18 it, how do we assure that that is sufficiently conservative?
19 And we didn't have a very good answer for that.
It wasn't j
20 obviour that just doing that would be sufficient.
i 21 Thus came the process of doing the best estimata 22 model, doing the calculation, trying to come up with an 23 estimate of what the 95th percentile PCT was, and we said i
24 that's good enough, and if your Appendix K model is larger 25 than that, then you demonstrated that your Appendix K model i
(
HERITAGE REPORTING CORPORATION -- (202)628-4888
15 1
will keep sufficient conservatism to cover the uncertainties
()
2 and the best estimate model and its application No the plant.
3 Once you have gotten this done, actually you do 4
Appendix K calculations for the plant.
You ton't redo the 5
best estimate model necessarily.
6 MR. SCHROCK:
I don't think vsu heard a word I said.
7 MR. JONES:
This is the reoceas.
This is what the 8
83-472 process is and that is how it was applied to GE.
It is 9
not different than how it was applied to GE.
That was the 10 intent of the process.
Maybe it was wasn't communicated that 11 way in the past, but that is the intent and how the approach 12 is taken.
13 MR. CATTON:
I dor.'t believe that was the intent
(}
14 some of us who have been involved in the process from the 15 beginning would understand.
16 CHAIRMAN WARD:
Virgil, would you restate?
I think 17 you made a good, a good statement of the issue.
18 MR. SCHROCK:
Okay.
The point is that the, the 19 advantage of the SECY 83-472 is that it permits the applicant 20 to identify the governing situation through the use of best 21 estimate methodology, which is calculating as best we can what 22 actually happens in the plant as a response for a postulated 23 initiating event.
24 Then at the conclusion of that, you comply with the 25 existing law by doing one calculation by the evaluation model, HERITAGE REPORTING CORPORATION -- (202)628-4888
7___________________
16 1
that demonstrates that the new methodology is in fact bounded
()
2 by the old law, and therefore, you met the spirit of the old 3
law, but you have gotten to that point through calculations 4
that make some sense technically, not through calculations 5
that produce absolute dissimilarity between response of the 6
system, and what you are calculating.
Okay.
7 And so that's the advantage of it.
Now why is it, 8
why isn't it cost saving?
Maybe it produces a result that is 9
better in the end and gives more freedom in the operation of 10 the plant.
Maybe it allows power level changes.
Maybe it 11 does a number of things, but just in the process itself, it 12 ought to be a more satisfactory way of getting to the end 13 result and in fact it ought to even in my estimation, be a
(}
14 more economical one.
15 Now what we hear is that well, the cost of doing 16 these best estimate calculations and applying through any kind 17 of mathematical sense at all, the knowledge that we have of 18 the uncertainty in our best estimate calculation to arrive at 19 a 95 percent probability PCT is prohibitive.
I don't believe 20 that.
Okay.
And that's what has not been demcnstrated here 21 at all, that that is prohibitive.
It has not been attempted 22 here.
23 It has, the a priori premise is that that is 24 prohibitive.
Therefore, we will not do it.
We will find a 25 way of producing some new sort of arguments that lead us to an HERITAGE REPORTING CORPORATION -- (202)628-4888
17 1
upper bound on our PCT calculation through something which is
()
2 not a best estimate calculation.
It is not an evaluation 3
model calculation.
It is ill defined.
4 MR. JONES:
Let me--the second half of your question 5
hopefully I am addressing through this presentation, but with 6
respect to the first part, I do have to say the Appendix K 7
calculation is not a single calculation necessarily.
In fact 8
in the case of GE, it requires spectrum of breaks to still be 9
analyzed under Appendix K.
10 MR. CATTON:
But by best estimate.
11 MR. JONES:
Best estimate and evaluation model; it 12 did require that, and your technical specifications for the 13 plant must still be demonstrated by the Appendix K calculation
(}
14 and in compliance with the regulatory, with 50.46.
15 What the best estimate approach does is buy you, 16 generally decrease your peak clad temperature and allow you to 17 widen or increase your technical specification values for such 18 things as peaking factor, but that is how it is done.
That is 19 how it was applied in GE, you know.
It was not a single 20 calculation.
And that is documented in that SER back in the 21 83--
22 CHAIRMAN WARD:
Would you go back to the first slide 23 again, please?
I know you are trying to get beyond that.
24 HR. JONES:
I am not trying to push you.
25 CHAIRMAN WARD:
I think the problem is, is that back HERITAGE REPORTING CORPORATION -- (202)628-4888
18 1
when this83-472 method was developed, and I guess there could
()
2 be a number of perceptions of it, but at least the perception 3
of some of us were that that first bullet was kind of the 4
centerpiece of that, that that was, that that was the whole, 5
the benefit in both practical terms and safety terms as far as 6
giving you a better understanding of what was going on in the 7
plant, was to concentrate the analysis on the first bullet.
8 HR. JONES:
It still is.
9 CHAIRMAN WARD:
Okay.
And then, too, as a part of 10 that was to provide, make an estimate of the 95 percentile 11 level.
Then going back to the EH was sort of tokenism to 12 comply with the law.
We have--that is sort of exaggerating e 13 little bit, but in this process, we seemed to have, well, if
(}
14 not reversed it, changed that where the impression I have is 15 that the best estimate analysis that has been done for this 16 UPI plant has been kind of tokenism, and the emphasis is on, 17 well, really on the other two.
18 HR. JONES:
I guess, well, as I get into the 1
19 presentation, hopefully you will get the flavor of what it is.
20 but the review effort is indeed concentrated on the best 21 estimate process, the best estimate code, and then how you do 22 your uncertainty analysis.
That is where the review effort 23 primarily concentrates.
24 CHAIRMAN WARD:
You are saying that, and that's--
O l
25 HR. JONES:
I am just trying to clarify that l
l l
l HERITAGE REPORTING CORPORATION -- (202)628-4888
19 1
Appendix K is not that much of a tokenism.
It is not, not an
()
2 insignificant effort, and that's all I am trying to clarity.
3 That's not an insignificant effort, It is not just one 4
calculation, once and go home.
It is still, because of all 5
the legal requirements that are wrapped up in there, does l
6 require some significant efforts on behalf of the applicant.
7 CHAIRMAN WARD:
Okay.
8 HR. CATTON:
Before you leave, I just want to make 9
sure I understood your answer to my question.
When I asked 10 you about the heat transfer coefficients, you said that you 11 used judgment.
Is that rigl.t?
12 MR. JONES:
It's combination of judgment and the 13 license, you know, the contractor's familiarity with the
[
[}
14 variance in thermal hydraulic codes and correlations that are 4
15 out there, coupled with the analyses that were done of the 16 plant data that provides it.
17 No, we did not go into individual models and do a
18 individual code, model comparisons to data, but in the area of 19 the heat transfer, within this approach, for the Westinghouse 20 application, you have the G-2 FLECHT SEASET data, THTF data, j
21 so you have done a lot of calculations in the area of heat 22 transfer.
23 No, you did not take the correlation separately and 24 say how well does it fit the data, but you are looking at the f
()
25 effect of the heat transfer models as put into the code, i
HERITAGE REPORTING CORL0tATION -- (202)628-4888
20 1
interrelated with all the other models, et cetera, and how it
()
2 is reflected in the, its ability to predict these data sets.
3 HR. SC'10CK:
I can only conclude that Westinghouse 4
is receiving inadequate review.
5 HR. JONES:
And I will say this is not a, unlike 6
what we did f or GE.
It is not.
We didn't do thih with GE, 7
either.
8 HR. SCHROCK:
We will show you some ways in which it 9
is unlike what you did with GE.
10 HR. CATTON:
I find there is a significant 11 difference between what is being done now and what was done 12 for GE.
13 HR. JONES:
There is, certainly is significant
(}
14 difference in how the uncertainties are taken.
15 HR. CATTON:
Unless the dance team that get up in 16 front of this group wasn't saying it like it was.
17 HR. JONES:
Certainly the way the uncertainties werm 18 treated had a significant difference.
19 HR. CATTON:
Significant difference in what details 20 were looked at, too.
21 HR. JONES:
Okay.
22 HR. TIEN:
Could I ask one question?
I am new.
I 23 missed the last meeting, so I didn't get all the fun part, 24 but--
}
25 HR. JONES:
Depends on which side of the table you HERITAc r. REPORTING CORPORATION -- (202)628-4888
7
- i 21 1
.were on!
.r~\\
. \\ '/
_1 2
MR. TIEN:
The first question I want to ask, I just
\\
'i
-3' want to make sure that you still maintain officially as a 4
regulatory person, there is no'need, in fact it is not 5-necessary to check about the heat transfer coefficient 6
correlation at all, and you rely completely based on whatever 7
your consultant would say that is correct.
Is that your 8
official position?
9 MR. JONES:
We use consultants to help make those 10 judgments.
11 MR. CATTON:
Your consultants didn't look at them, 12 either.
And in this particular case, your consultant is a 13 young man who doesn't have the experience I don't believe to
(}
14 even come to such conclusions.
15 MR. JONES:
Well--
16 MR. TIEN:
I think that is very important.
If you 17 say this is not' exactly your position, you know, you feel it 18 is necessary to check on that, I feel much more comfortable, 19 but you if you are maintaining there is no need to do that at 20 all, and all the information regardless what you got is 21 sufficient, then I must say I am worried.
22 MR. JONES:
We have worked with the consultants in 23 all these meetings with Westinghouse and this was not a 24 non-trivial review effort by no means.
P.ut you know, I think 25 I clearly stated we did not sit down and look at the heat HERITAGE REPORTING CORPORATION -- (202)628-4888
22 l
1-transfer correlations in the level of detail I think Professor
()
2 Catton is getting across.
3 MR. CATTON:
That is just a symptom.
You have got a 4-whole lot of fluid that is contained in something, and it 5
contacts the outside world through heat transfer coefficients, 6
friction factor relationships.
If you don't have those things 7
right, you don't know where the hell you are.
If you 8
incorporate, you have an unbelievable amount of logic in a 9
code that is a couple hundred thousand lines long.
There is 10 nobody iTo can tell you whether or not all that logic is in 11 there right and there is only one way to ferret that out.
You 12 have to have the code generate some of this stuff and compare 13 with data to see if what you intended to have in there is
[}
14 indeed in there.
15 As far as I can tell, this was not done.
Now I l
16 think Westinghouse--
l 17 CHAIRMAN WARD:
In answer to that point, Bob seems 18 to have said a couple of times, and I would like to get an 19 answer, that the staff hasn't done it, but your consultants 20 have or have not?
Could we clarify that point?
21 MR. JONES:
Rupe, do you want to try to--
22 MR. BYERS:
Rupert Byers from Sandia--during our 23 review, which by the way I was not connected with from the 24 beginning, we started by looking at the documentation 25 COBRA / TRAC which was a parent code of WCOBRA/ TRAC.
We asked HERITAGE REPOPTING CORPORATION -- (202)628-4888
23 1
Westinghouse to provide us with a-list of all changes that
)
2 were made from COBRA / TRAC which was developed in the Pacific, 3
isn't that right, Pacific International laboratories, all the 4
changes that went into that code to turned into the W code 5
TRAC that we had, so we did not in fact look at lines of 6
coding, but we did look at all the documentation that we could 7
get a hold of, and all the documentation from Westinghouse 8
telling us what new models were going into the code.
And 9
because of the fact that the COBRA / TRAC has been qualitatively 10 assessed not, no accuracy quantifications were done, but 11 because of the fact that a lot of special effects calculations 12 were done just to address things like the effective heat
_{ }
13 transfer coefficients, I think it would be, if you look at the 14 results of those special effects calculation, for example, in
~
15 which the heat transfer process is dominating ones, you don't 16 believe that you can infer some kind of judgment on whether 17 the code is doing a good job or not, you might as well quit.
j 18 Do you really, Dr. Catton, do you think it is 19 necessary to look at every line of code in a review process?
20 HR. CATTON:
I don't believe that I ever said you 21 have to look at every line of code, but one thing you have to 22 do I believe is to look at what the code calculates and you 23 have to look at it in sufficient detail to come to a l
I 24 cenclusion now, or judgment.
I will give you an example.
25 The early version of COBRA TF used the Rosenau, also l
l HERITAGE REPORTING CORPORATION -- (202)628-4888
24 1
had complicated split of where the heat was going either to
()
2 liquid or to vapor.
We have found that this is a disaster in 3
TRAC.
Now you just accepted that.
You accepted COBRA.
The 4
only question you asked was what do the changes do?
5 Well, there was another group that was much smaller, 6
probably spent a lot less money, that took a look at that, but 7
looked at where you're tying your split thermal hydraulics to 8
the surroundings, and they found that the code came wanting, 9
and there was a significant penalty that may come out of that 10 Rosenau that it was, how it was treated.
I don't remember the 11 numbers exactly, but a hundred to 150 degrees F was an adder 12 that they felt would have to be put in because of it.
You l
13 didn't find it because you didn't look at it.
You just
{)
14 accepted what you were given and went from there.
I don't 15 think that that is good review.
It is inadequate.
And that's 16 not a microscopic look, by the way, although that might help.
17 MR. BYERS:
Well, I guess the part of this L ole 18 thing, it is connected with the way this is being approached.
19 All the so-called assessment comparisons from it, from the 20 COBRA / TRAC results, are indeed producing the uncertainty 21 number, and what the source of those is from the way the 22 uncertainty number is used, what the source of that 23 uncertainty is, is in the end I don't think really important.
24 If you have a set of calculations and comparisons with the f
25 experiment, and I that I say this is wrong by such and euch an l
HERITAGE REPORTING CORPORATION -- (202)628-4888
25 1
amount, this should show up in the bias.
(I 2
HR. CATTON:
But see you certainly do have an 3
uncertainty number when you are handed your exercise, but in 4
my view, you don't have any confidence, or if you do, you are 5
diluted.
I don't know how you can get that from what, at 6
least what I have seen.
7 MR. JONES:
Let me just try to make one comment.
I 8
think what we are talking about is a difference in philosophy, 9
and today the--
10 MR. CATTON:
I don't think so, Bob.
There is no 11 philosophy here.
At least there shouldn't be.
There is a 12 technical question put on the table.
How do you address it?
13 MR. JONES:
I think--I somewhat disagree with you.
{}
14 I think it is somewhat philosophical in how do you get to--
15 MR. CATTON:
You introduce philosophy into this sort 16 of thinking and you are in deep trouble.
l 17 HR. JONES:
Let me--the approach was back in the GE l
18 days, you know, in talking to the reviewer that did that work, j
19 and talking to Brian Sheron back then, and going through chis
('
20 process, the approach was that we would get a handle on the 21 adequacy, the overall code schedule and adequacy of the models 22 not by an individual look at models necessarily, but by taking 23 the code, applying it to the data, and that is how we would 24 come up with the code uncertainty.
O l
25 The CSAU methodology is a different approach.
It is l
HERITAGE REPORTING CORPORATION -- (202)628-4888
26 1
a different way of attacking the problem, but this is the way
()
2 that we attacked the problem in the GE, in the GE approach, 3
and it is the same way we are attacking it here in these 4
process such as the heat transfer model that you are talking 5
about.
We~believe that they are adequately addressed by the 6
code comparisons that have been done to the FLECHT, SEASET, 7
G-2 data, et cetera, to come up with what the bias and the 8
standard deviation is for the code.
9 And it is not CSAU methodology.
I will agree with 10 that, but I do think it is'a difference in how you wish to 11 attack the problem.
We think this is a reasonable way of 12 going about it.
And I understand the CSAU methodology is l
{}
13 another way of doing it.
It is different.
14 MR. CATTON:
I don't mean to tell you you should use 15 the CSAU methodology, but there are certain things I think you 16 really ought to look at.
One is heat transfer.
And another l
17 is the friction factor relationships and interfacial drag.
I i
18 don't see any of those brought out anywhere, and I believe I 19 will just stop.
I think a review without some investigation 20 of these things is inadequate and unfair to Westinghouse.
21 CHAIRMAN WARD:
Larry?
22 MR. HOCHREITER:
Couple of comments--first of all, 23 the reviewer that started on the program is Dr. Buxon, and he 24 was with the program through the completion of Volume 1, and
.O 25 Dr. Buxon has ample experience in digging in, ferreting out l
HERITAGE REPORTING CORPORATION -- (202)628-4888
27 1
model shortcomings and various codes, and I think is one of
()
2 the more apt consultants the NRC has in terms of reviewing 3
codes and models and applications of this sort,.so there was 4
not, this was not a breaking-in ground for a new reviewer.
We 5
had a very experienced reviewer who knew the details of the 6
codes, and knew how to ask the right questions.
7 In terms of the heat transfer, we will be speaking 8-to that in Part B, and somewhat this part C of our part of the 9
agenda, but we do believe that the comparisons we have done, 10 we have looked at the heat transfer ourselves, okay.
It does 11 not appear in the documentation that you do have, but we t
12 believe that you can characterize heat transfer request 13 correctly by looking at not only the clad temperatures, but
}
14 the vapor temperatures.
ity of the fluid, amount of mass 15 that is stored in bunt _
a we have made those comparisons, 16 will be showing some of waose to you today, and those are 17 documented in the information that you have.
18
-MR.
CATTON:
I am not questioning what you guys did.
19 I am questioning the review, but just a comment to what you 20 said, I think looking at temperatures is sort a bottom line.
21 There is a stage I think you have to go through where you try 22 to decide whether it is an overall or reasonable code, and 23 part of that is making sure that whatever the models are that 24 you put in are indeed effectively implemented.
25 I don't see that step, either, because that means HERITAGE REPORTING CORPORATION 202)626-4888
____.__ _._._..,___ _.,,.____.__ _-- (. _ __._,___, _ _. _ ___.__
28 1
you have to look at things.
When you do a FLECHT calculation, s_)
2 you really ought to compare those heat transfer coefficients 3
they got out of the FLECHT with the ones you are using because 4
there may be compensating errors somewhere, and the same thing 5
with CCTF.
You ran a whole bunch of CCTF studies, but I don't 6
see any of that other end of it.
7 Now for an EM evaluation, which we are all used to 8
doing, it doesn't matter, and that's the flavor that I get is 9
that really this is, has been an EM kind of elevation, that 10 this has not been a best estimate evaluation, as sucia, it is 11 inadequate.
12 HR. HOCHREITER:
Remember in the experiments you
(}
13 have just quoted what you measure.
The heat transfer, the l
14 total rod heat flux, the code is calculating high partition in i
l 15 heat flux, and in some experiments you will have the right 16 measurements that allow you to incur precisely how well the l
17 code is doing a job in partitioning the energy, and there are 18 very few of those experiments.
FLECHT is one of the 19 exceptions because of vapor temperatures.
20 HR. CATTON:
Still you ought to be--
21 HR. HOCHREITER:
CCTF did have vapor temperatures, 22 showed what the vapor superheat is and how well the clad 23 temperatures as well, so--and I did protect the, absolutely, 24 level of the heat flux correctly, then the partition of the 7-)
(/
25 heat flux and we will show you some of this stuff.
We have HERITAGE REPORTING CORPORATION -- (202)628-4888
29 1-looked at that. -And the staff has reviewed that information.
If 2
MR. CATTON:
They didn'tfdocument their review that 3
I can. find.
4 MR. HOCHREITER:
They have looked at it.
5 CHAIRMAN WARD:
Okay.
Thank you.
Go ahead.
- 6 MR. JONES:
Just to continue on with just kind of an 7
overview of what we did, to a large extent, this kind of 8
summarizes where we went, first after, and this kind of gets 9
me in, the first line kind of gets me into the overall 10 approach and'whether this needed to be a best estimate and how 11 well you need to define 95th percentile, but in essence, what 12 we were doing was looking or examining the Westinghouse 13 modeling methodology to provide reasonable assurance that the
}
14 criteria in 10 CFR 50.46 was met.
This did not mean th't we wera ensuring that you had 15 a
16 the 95th percentile or even that you were necessarily close to 17 it.
It was reasonable assurance that you meet the criteria.
18 Now how did we do his with respect to the approach or 83-472 19 guidelines?
20 DR. KERR:
Excuse me, Mr. Jones.
Are yo ocing to 21 tell us what reasonable assurance means in that con txt?
22 MR. JONES:
Yes, and that is essentlally v. it these 23 next three bullets are trying to address in the remaisder of 24 the slides.
.O 25 To give a feel for the review, we did, we be sically HERITAGE REPORTING CORPORATION -- (202)628-4188
30 1
have what we see are basically three significant issues that (n,)
2 we are trying to address during the review.
3 First off, is COBRA / TRAC a, a best estimate code 4
which can be applied for scaling the UPI response?
That is, 5
do we believe that it is a reasonable code and can do the job?
6 Second, given what the Westinghouse approach is, 7
does the superbounded PCT with, with its uncertainties added 8
to it, provide a peak clad temperature which is at least at 9
the 95th percentile, and have we examined the uncertainties 10 and assured they have been, you know, they have been 11 addressed?
Sometimes it is quantitative.
Sometimes it is 12 qualitative, but have. hey been addressed?
And then have they
('T 13 met the legal requirements of A7pendix K?
And I am going to
\\,_)
14 go into the details on how we did that in a few minutes.
15 There have been comments that this was a, or implied 16 to me, the impression I have is that the "ommittee thinks this 17 was somewhat of a cursory review.
This was not a cursory 18 review by no means.
It was a review that did have some change 19 in the guard during it.
This was about a two and a half year 20 process.
I picked up this review about a year and a half ago 21 or year ago when we reorganized.
Gene Hsii has been involved 22 throughout the entire process.
Consultant made a change over 23 about a year ago, but it has been a long, drawn-out process.
24 We have looked at many draft documents from Westinghouse.
25 They have been significantly modified.
l HERITAGE REPCRTING CORPORATION -- (202)628-4888
31 1
What you see is essentially the final product.
You
()
2 do not see all the changes that went up to that process, to 3
where we are now, but it has been a long effort, required at 4
least four and a half person years, and about 350 K 5
contructors, so it was not an insignificant effort by no 6
means.
7 MR. CATTON:
Just for the sake of comparison, what 8
is the CSAU effort going to cost?
9 MR. ZUBER:
Well, I think this is misleading.
My 10 answer, it is misleading as far as the cost.
I think the 11 entrance cost is high.
12 MR. CATTON:
I understand that.
Okay.
If you were 13 to repeat that process, what is it going to cost?
)
14 MR. ZUBER:
I would say no more, should not be more 15 than 400,00, 500,000.
Let me say we didn't do 150 16 calculations--20 runs.
So we have the flexibility of doing 17 it.
But it cost us more--not to mislead the audience--it cost 18 us more.
It cost us probably around a million and a half, but 19 the reason is because we had to produce two, three QA 20 documents which we didn't have.
We have to set up the 21 methodology.
I may draw a comparison, it is almost like going 22 to the movie, you pay your entrance cost to get into the 23 movie.
After that, it cost you to buy peanuts or whatever.
24 It is only a few pennies or a few dimes.
The entrance cost to s-]
I 25 enter into this activity is already paid.
To run these codes HERITAGE REPORTING CORPORATION -- (202)628-4888
32 1
I think and having them, technology applications still not be
'( )
2 very large.
The expenses not should not be very large.
3 CHAIRMAN WARD:
What is the implication there?
4 HR. CATTON:
The point is that it sounds to me like 5
any review that NRR undertakes of this tyr s is going to cost 6
them about a half a million dollars, or less.
7 MR. ZUBER:
Pardon me.
It should be even less.
Let 8
me say I don't know how project calculations goes on the 9
vender's end because we are using--which may have a lower 10 computer cost, but establishing the important phenomena, 11 arranging the parameters, the determining what is weak in the 12 code, once this is established, the thing should be very easy 13 to fall off.
The entrance costs at least from our side has
[}
14 been paid.
Doing this admittedly should not be very 15 expensive.
16 MR. CATTON:
But still there is dollars associated 17 with it, and one of the concerns I have had is the NRR was not 18 going to put forward the resources to do this kind of review, 19 and if they are not going to do it, then I don't think the a
20 rule should go through.
21 MR. JONES:
I will address that in the second 22 presentation.
23 MR. CATTON:
That is what I was trying to get to l
24 here.
l 25 MR. JONES:
That is coming later.
This, that was l
HERITAGE REPORTING CORPORATION -- (202)628-4888
33 1-just to give you an effort because the impression that was
(')
2' left at least from the last meeting was that we didn't do much 3
of a review, and it was a long, arduous process.
4 HR. ZUBER:
May I comment?
Maybe my answer not be 5
misunderstood--this was, as I say, the uncertainty for a TRAC 6
code.
Now if you ask what is the cost of reviewing it, and anL 7
applicant makes it even less than that, I did not really--what 8
was your question?
How much it does cost to determine the 9
uncertainty in TRAC or how much it would cost to review the 10 process?
11 MR. CATTON:
I was trying to get a measure of what 12 the cost, what the cost would be to take another code and go 13 through the exercise, like NRR has done for Westinghouse.
For
[}
14 example, you had CE submittal.
That didn't cost very much, 15 but that was for other reasons, but if you went through the 16 review process.
What would it cost?
17 MR. ZUBER:
For us to review it?
18 MR. CATTON:
Yes.
19 MR. ZUBER:
I think should be less because one could 20 just spot-check the important phenomena and do it through a 21 consultant or through RES.
I mean that should be less than 22 number like I quoted the number, I quoted starting with a 23 quote and proceeding and through the whole, do the whole 24 analysis and evaluation.
25 HR. JONES:
I am just going to say I am not sure I HERITAGE REPORTING CORPORATION -- (202)628-4888
34 1
agree with what his numbers would be, but I will try to
(,)
2 address that later.
3 MR. CATTON:
I am not sure--less than what?
'4 MR. JONES:
I do not believe any of these reviews S
would be very simple.
They won't be costly.
They will be--
6 MR. TIEN:
I think this offer impossible.
I would 7
like to hear whether it is possibility to have a, not a 8
full-blown review, maybe by another approach, but a, just a 9
few spot-checks of key elements, and how much that would cost, 10 or I would like to hear your comments later on.
11 MR. JONES:
It depends on how it is done.
Let me 12 try to get into that later, although it is not a line item.
13 Later you are going to have to ask it again.
{' )
14 How did we look at WCOBRA/ TRAC?
What was che 15 process we went through to try to determine whethcr this code 16 was adequate?
17 To some extent, Larry Hochreiter has added some 18 stuff that in part was before me, but there have been a lot of 19 meetings where we have discussed a lot of things also which 20 are not, which are probably not reflected on this slide, but 21 basically the process we went through was we took a global 22 look at the field equations, the closure relationships, the 23 models, looked at them to try to get a feel as to whether, you 24 know, there were, generally was the model going to appear to l
-s 25 be capable of simulating important phenomena?
We looked for HERITAGE REPORTING CORPORATION -- (202)628-4888
35 1
are there a lot of assumptions?
Are there neglecting
()
phenomena, et cetera?
That was the first thing we did.
2 3
We looked at the COBRA / TRAC developmental assessment 4
to determine, get a feel for the overall adequacy of the code 5
and some of its limitations.
6 We then looked at the specific modifications that 7
Westinghouse made to, to the COBRA / TRAC model for modifying 8
it, to simulate the UPI phenomena and address some of the 9
COBRA / TRAC development assessment problems, and then finally, 10 a fairly large effort was spent in going through the 11 WCOBRA/ TRAC comparisons to the data to get a feel for the 12 overall adequacy of the code in comparison to experimants.
As
/~
13 you can see, it is not a detailed review of the equations,
(-)'
14 like I stated before.
15 (Slide) 16 MR. JONES:
Bottom line, first off, we think that 17 the developmental assessment did provide us some qualitative 18 demonstration of the code adequacy.
Since the W COBRA / TRAC 19 model is based on the COBRA / TRAC development assessment, we 20 think that applies in a qualitative sense only because of the 21 changes made, but it does give us a good feel for where the 22 model is and we think it is generally adequate.
23 Our review of the specific models added by 24 Westinghouse, we went through those and we think that they are 25 capable of simulating major UPI phenomena such as the pooling, HERITAGE REPORTING CORPORATION -- (202)628-4888
36 1
CCFL, entrainment, et cetera.
We think that the equations in
()
2 the code contain the basic features that you are going to need 3
to do an analysis of the UPI model.
Specifically the three 4
field two-phase flow model allows'you to handle the heat 5
transfer in the the dropping of liquid down in the, from the 6
upper plenum.
7 In looking at the data comparisons, as a whole, we 8
think that overall it demonstrates that WCOBRA/ TRAC will 9
provide a reasonable agreement with the data.
10 We, you know, we used the PCT data primarily to come 11 up with the bias and code uncertainties, but we also looked at 12 the overall system response--pressure slows, pressure drop, 13 void fraction.
We think generally it was in good agreement
{}
14 with the experiments, and we attempted to look for 15 compensating errors, not necessarily along the detail that you 16 were implying, Dr. Catton, but we did look for things like if 17 the code overpredicted temperature with-level low, lower than 18 the experiment?
For example, in the CCFT data or the FLECHT 19 data; if we under predicted the heat transfer, was level 20 higher?
21 We are looking to see whether there was that level 22 of bias; was the code self-consistent between the predictions, 23 some of global system parameters and some of the results we 24 were coming out, and we did not find any apparent compensating 25 errors.
I know in the last meeting there was a lot of HERITAGE REPORTING CORPORATION -
(202)628 4388
37 1
. discussions of ones that were readily apparent where models
()
2 were tuned.
We did not see that from our review.
3 MR. CATTON:
We had lengthy discussion the last time 4
on blowdown peak, reflood peak, and so forth.
5 Are you going to address that question?
6 MR. JONES:
Yes.
That's coming.
7 MR. SCHROCK:
What did you learn in your review 8
about the noding of the--
9 MR. JONES:
Noding.
10 MR. SCHROCK:
Aren't you interested in that?
11 MR. JONES:
Well, first off, the noding used in the 12 experiments at least for the, philosophically, the basic 13 approach was the same between the experimental predictions and 14 the planned application, so we think we had the noding covered 15 to some extent there.
There were specific noding studies 16 done.
17 MR. SCHROCK:
I don't understand that comment.
18 Experiments are small-scale systems.
Plant is a big system.
19 The noding decisions are inevitably different, but wl.at do you 20 mean by that?
21 MR. JONES:
Philosophically, again use that word--
22 MR. TIEN:
There is no such thing in noding 23 experiment.
24 MR. JONES:
Basically the approach used was, had the 25 right words, but they used like the same heat transfer length HERITAGE REPORTING CORPORATION -- (202)628-4888
38 1
between the experiments for a node, between the experiment,
()
2 and then what they applied in the plant, so they tried to keep 3
that.
They didn't just use five axial nodes, look at four 4
inch foot rod and five axial rods, look at a 12 foot plant 5
rod.
They used kind of same noding type breakdowns.
That's 6
what I mean by they were consistent.
7 MR. TIEN:
What do you mean noding?
8 MR. JONES:
To look at things like multi-dimensional 9
flow in the downcomer, et cetera, in order to select some of 10 the models.
11 MR. TIEN:
Excuse me.
I couldn't understand.
What 12 do you mean by noding?
In the experiment when you do 13 experiment, you don't have--
[}
14 MR. JONES:
The simulation of, the simulation of the 15 experiment.
16 MR. TIEN:
Yes.
When you get the, compare with 17 experiments--
18 MR. JONES:
That's what I am talking about.
The 19 noding used or the general methodology used to come up with 20 the noding for the experiment is the same methodology used to 21 come up with the experiment, the noding for the plant.
There l
22 were questions with respect to, for example, how many channels 23' do I simulate the coring?
And that's one ot' our outstanding 24 questions with respect to this review.
25 A four channel model is used for the plant.
The HERITAGE REPORTING CORPORATION -- (202)628-4888
39 1
final plant predictions, the CCTF data was done with three (3
ts_)
2 channel model, and before we issue the SER, we need to have 3
the four channel calculation to see whether it has any 4
significant effect on the predictions, whether it will 5
significantly impact the uncertainty evaluations.
If it does, 6
we are going to have to hold, but that work has to be done 7
before we issue SER.
8 MR. TIEN:
I understand what you mean.
My question 9
really is are you sure you actually get sufficient detailed 10 experimental data which you can make adequate noding which you 11 know, averaged out like a particular segment and which 12 compares exactly with the numerical?
I doubt it.
It is good
(~)N 13 to say it, but I don't think you can ever get--
14 MR. JONES:
That is accounted for in the uncertainty 15 evaluation.
That is accounted for in the uncertainty 16 evaluation.
17 DR. KERR:
I am puzzled by the comparison of an 18 experiment node which apparently I assume means points in 19 which one makes measurements?
20 CHAIRMAN WARD:
No.
21 MR. JONES:
These are fluid volume nodes would be a 22 certain height, and they use the same height and they would do 23 the plant calc.
24 CHAIRMAN WARD:
A COBRA / TRAC calculation of the--
f-V) 25 DR. KERR:
I know, but when you talk about noding, HERITAGE REPORTING CORPORATION -- (202)628-4888
~
40 1
and experiment, I assume you have an experiment that is a
()
2 continuum.
It is a distributed system.
It is not a point 3-system.
When you do noding, you are representing a continual 4
system by point systems that are connected in a certain way, 5
so to say that you--I don't understand what it means to 6
compare.
7 CHAIRMAN WARD:
He is talking about his COBRA / TRAC 8
modeling of the experiment.
9 DR. KERR:
But he seems to be implying somehow the 10 experiment is noded.
11 HR. JONES:
No.
I am saying to simulate the 12 experiment, you have to, with the code, you have to make
{}
choices as to what level of detail to put in your simulation 13 14 model in terms of height, width, radial distributions, et 15 cetera, that those choices were made to do those experimental 16 predictions with WCOBRA/ TRAC.
17 What I am saying is the general approach used, for 18 example, in setting the node size height-wise is the same 19 approach that would then be used to do the plant.
Now 20 comparison to the data, they would take you, you would only 21 get one temperature prediction essentially out of the rod for 22 the code.
Then you would have to compare that to the data and 23 you have a myriad of data points and the scatter in the data 24 at given elevation and experiment was covered within the O
25 uncertainty evaluation itself, but the code was compared to HERITAGE REPORTING CORPORATION (202)628 4888
41 1
the average of the data.
()
2 Then you have the scatter of the data to consider as 3
part of the overall evaluction of the unccrtainty, but the 4
noding itself was consistent at least in the general approach 5
utilized to simulate experiment of the plant, so you do not 6
have to answer or attempt, attempt to not have to answ r why l
7 is the noding different, what impact does that make on the 8
uncertainty evaluation for the code, because noding can 9
influence the ability of the code to predict the experiment.
1 10 MR. TIEN:
I think there is still a point just Bill 11 mentioned.
That is in the experimental simulation, are you 12 sure you can actually get the correct noding for your 13 simulated experiment?
It is a distributed system or a finite
{}
14 element system?
Th3t's the question.
Your point is that's 15 taken care of by the uncertainty analysis.
16 MR. JONES:
Define correct.
This is a choice they l-17 made to simulate the experiment.
This is how they simulate it 18 with their modeling guidelines.
They are using those same l
19 models guidelines when they model the plant, and that 20 consistency is what we wanted.
f 21 MR. HOCHREITER:
One of the--unfortunately you 22 missed the last meeting.
One of the components of the 23 uncertainty that we put in when we considered the overall code 24 uncertainty is how one models the experiments, and in there we 25 took specific experiments and we did change the noding, and we l
l HERITAGE PEPORTING CORPORATION -- (202)628-4888
i 42 i
1 looked at how the calculated peak cladding temperature changed
()
2 relative to the lc*a, and in a number of experiments we did do 3
axial noding where we subdivided the experiment into a 4
different number of fluid volumes.
5 Basically'we doubled the nodes in many cases, and in 6
those cases we generally found very, very small differences 7
but that difference we did factor into the uncertainty, that 8
would then apply to the code's calculation of the PCT.
9 MR. CATTON:
I made a comment at the last meeting 10 that is in this area, and I will repeat it.
11 I looked at all of the information that was given 12 us.
I found that with the LOFT they used four core nodes, 13 underpredict the temperature, in some cases up to 150 degrees.
[}
14 I will comment on that underprediction again in a minute.
15 On CCTF, they used six to seven nodes, slight 16 underprediction and sort of increasing with elevation.
17 FLECHT which was they used 12 nodes, and they 18 overpredict.
Okay.
l 19 Now by the way, that was a no UPI CCTF that I looked l-20 at.
Somehow that information should lead me to an 21 uncertainty.
What did you do with that information?
What is 22 the uncertainty associated with it?
The answer to that 23 question will address everybody, past questions that have been 24 raised.
25 MR. HOCHREITER:
I can answer that question.
HERITAGE REPORTING CORPORATION -~ (202)628-4888
1 43 1
MR. CATTON:
I want him to answer it, Larry.
/'T
'(_j
.2 MR. JONES:
The uncertainty from those?
3 MR. CATTON:
You were given information in these-4 reports.
What did you do with it?
5 MR. JONES:
The uncertainties between the various 6
experiments were average and standard to come up with biases 7
and standard deviations determined.
8 MR. CATTON:
You see, you were answering their 9
questions about your view of nodalization.
That's different 10 than what is available to you.
In one case they had four core 11 nodes, that short core.
In another case, they had CCTF which 12 is four full-length six to seven nodes, and they had FLECHT, 13 which is 12 nodes, and you have a definite trend to
{}
14 underprediction.
You had to have addressed that.
15 HR. JONES:
Gene, do you have any comments on that 16 one?
17 MR. HSII:
No.
18 MR. CATTON:
Okay.
I think--
19 MR. HSII:
The noding study, to assist us in the 20 experiment data, Westinghouse has performed sensitivity study 21 on the noding, the effect of noding on the PWR.
They assess 22 the uncertainty, and with respect to the application to the 23 PWR, they tried to use the same kind of noding with respect to 24 the assessment work.
25 MR. CATTON:
I would conclude from that that they HERITAGE REPORTING CORPORATION -- (202)628-4888
44 1
damn well better use twelve nodes of the core, unless I could
()
2 see some very strong equivalent to the core.
I don't believe 3
you should use 12, but that certainly is the evidence that is 4
in your report.
5 HR. HOCHREITER:
I would disagree with you.
The 6
data that you have, there was only one log test that we 7
underpredicted.
8 MR. CATTON:
But you have to remember the common 9
wisdom for the LOFT L2 series is t'..ase reflood temperatures 10 are on the order of a hundred degrees low because of the 11 external thermocouples.
You really need to add a hundred 12 degrees to it, then make that statement, and you didn't do
{}
13 that.
14 MR. HOCHREITER:
We are going to address that, and 15 what I will show you later in the agenda is what happens to 16 those answers when you do that.
Okay.
17 Again, in terms of the noding, when we came up with 18 this factor that we put into the uncertainty, we went back and 19 we used those LOFT tests and we compared them to more recent 20 calculated LOFT test which had six nodes and we used that 21 effect into our modeling to the total code uncertainty.
22 MR. CATTON:
Did you do this splitting blowdown and 23 reflood?
If you didn't, then you haven't addressed the 24 question of I am raising.
25 HR. HOCHREITER:
He came up with that number--
HERITAGE REPORTING CORPORATION -- (202)628-4888
45 1
MR. CATTON:
You are not addressing the question
()
2 blowdown is a different case entirely.
3 MR. HOCHREITER:
I have got another way of doing 4
that that I am going to show you.
5 HR. CATTON:
It is going to be very difficult to 6-buy.
I 7
MR. SCHROCK:
I raised the. question of the noding 8
because I am concerned about the upper plenum calculation.
9 The whole point of our meeting is that UPI, and what have you 11 0 done to understand the effect of noding of the upper plenum?
11 HR. JONES:
My understanding is there were separate 12 noding studies done for that.
13 HR. HSII:
Yes, sensitivity study on the upper
[
14 plenum.
15 MR. JONES:
They took the worst configuration for 16 them all, and they put hot rods under, I forget whether it is 17 guide tube open hole support columns anymore, but they put the 18-hot rod in the worst location, et cetera.
Those specifically 19 asked--
20 MR. SCHROCK:
You tried last time to tell us this 21 and 7. couldn't understand it.
I don't understand it yet, and 22 I hope that during the course of the day you can come up with 23 some clarification on this question of what you know about the 24 effect of noding of the upper plenum on the result of 25 COBRA / TRAC predictions in the plant.
HERITAGE REPORTING CORPORATION -- (202)628-4888
4 6~
1 MR. JONES:
There is certainly'some difference in
()
2 how you model that, three, four channels, the interface to the 3
upper plenum, and it is not insignificant.
It is not 4
insignificant.
5 MR. SCHROCK:
These are too general comments.
6 MR. JONES:
That is why we will be, we have asked 7
them to go back and look at them.
8 MR. SCHROCK:
Intuitively I know that is not the 9
case.
It is not a insignificant thing.
That is why I am 10 raising the point.
11 MR. JONES:
We are asking that.
That is why we 12 tried to be consistent through the process of the CCTF to the 13 plant, and since they have changed their plant model we are
[}
14 now requesting, going to require them to go back and redo the 15 CCTF calculation using the same approach.
16 MR. HOCHREITER:
If I could interject something, we 17 did do upper plenum noding sensitivity studies.
We also did 18 sensitivity studies looking at which type of hardware in the 19 upper plenum would influence the temperature response in the 20 most unfavorable direction, so we did simulate the specific 21 types of hardwear in the upper plenum, tried and isolate in 22 which again type of hardware you should be using to represent 23 the hot assembly in the core, and that's one of the uniqua l
24 features of COBRA / TRAC.
You can't isolate (4 single hot 25 assembly in the core, so you can investigste the influence of W
HERITAGE REPORTING CORPORATION -- (202)628-4888
47 1
the structures in the upper plant on the flow that is
()
2 delivered to the hot assembly through the upper core plate and 3
the upper fuel nozzles, and we did do specific sensitivity 4
studies looking at how one would break up the upper plenum in 5
different nodes to model the UPI injection, and as Bob pointed 6
out. one of the open items that we are working to complete is 7
to try and use the same type of logic that we used to model 8
the PWR, to go back and model the CCTF.
9 CHAIPMAN WARD:
Did you conclude that there was 10 significant uncertainty associated with, with noding in the 11 upper plenum?
d 12 HR. HOCHREITER:
We had always felt when we started 13 the program that if we used a core noding in the upper plenum
/}
14 it would be conservative because it would force the water to 15 spread across the entire upper PWR and then selectively try to 16 flow down in the countercurrent flow method through the holes 17 in the upper core plate in the fuel houses.
j 18 HR. SCHROCK:
How do you know that is conservative?
i 19 HR. HOCHREITER:
Because there were CCTF tests that i
20 it showed that when you had injection of one side of the 21 vessel like in the plant, there was penetration, CCTF test 76 22 versus CCTF test 59.
It showed that there was selected 13 penetration of the water in the region where the injection t
24 was.
25 We took initially more conservativa approach.
When HERITAGE REPORTING CORPORATION -- (202)628-4888
1 48 1-the UPTF test came alor.
ch / ran A UPI type injection
()
2 scheme, in a UPTF ft.1 s:(1-ver et It was very, very clear 3
that the water juet
.. m
- u. thr ' m the injection port, hit 4
the immediate structut s,
r..en renetrated straight down 5
into the core.
6 In fact, the crude mass balance on the dats 7
indicated that all the water that came in penetrated.
And 8
that was a full-scale test, so we felt justified at that l
9 point.
We could go ahead and model that effect, and that's 10 why we went to the channel model where we segregated the outer 11 region of the upper plenum over the low power fuel assemblies.
12 There would be lower stem generation in the region of the 4
13 upper plenum between the core and the inside of the core
[}
14 barrel from the remainder of the upper plenum.
15 And when we did those calculations, we did see it, 16 we did see more penetration.
We show that in our reports, and 17 to the Nke staff, and the question that came back was well, 18 your model verification has been done for a cruder model, you i
I 19 should go back now using the modeling approach on the PWR and 1
3 20 go back, look at the CCTF test, and that was the staff's 21 comment on our model change, and we are in the process of i
1 22 doing that.
23 HR. SCHROCK:
What I requested is your statement 24 that a computational hard fact that forces the liquid to 25 spread over the entire upper plenum is inherently going to HERITAGE REPORTING CORPORATION -- (202)628-4888
_ - =
49 1
lead to a conservative end result in the calculation.
How do
()
2 you reach such a conclusion?
3 MR. HOCHREITER:
We calculated that in'this case and 4
again, it was seen in tests.
5 MR. SCHROCK:
What do you mean you calculated it?
6 HR. HOCHREITER:
The code, you know, we modeled it 7
tisat way, and then we modeled more of 'a separation of where 8
the water is injected and you see penetration in the second 9
case.
10 If you look at CCTF test 59, the water is basically 11 injected into, toward the center of the bundle, and it pools 12 more easily.
13 In CCTF test 76, the water is injected primarily on
[}
14 one side of the bundle, and it was more easy for the, it was 15 easist for the water to penetrate.
When you look at the UPTF 16 test, that to me is the clincher.
It doesn't pool.
17 MR. SCHROCK:
Where it penetrates is not where the 18 peak clad temperature is located, and so to argue that the 19 mere fact of penetration in peripheral regions of the reactor 20 is automatically beneficial and will lead to lower peak clad 21 temperatures, the whole process is very complicated.
22 MR. HOCHREITER:
You have to follow the process 23 through I agree.
T 24 MR.
CHROCK:
You come to conclusions that are based 25 on statements such as this that if you force the water to HERITAGE REPORTING CORPORATION -- (202)628-4888
50 1
spread over the upper plenum in your calculation, that that is
()
2 the inherently conservative thing to do.
I don't believe it.
3 I just don't believe it.
4 HR. HOCHREITER:
When we force water to spread 5
across the upper plenum, we think it is conservative because 6
you have less delivery of the water to the core.
7 HR. SCHROCK:
You deliver the water to the core not 8
when the temperature is highest but when it is, the 9
temperature is lowest.
10 HR. HOCHREITER:
Let me finish, sir.
What happens 11 in this case is that you don't fill the vessel as fast and 12 when you spread the water across, you don't get a lot of
/}
cooling from the water falling down through the core.
You get 13 14 more cooling as you fill up the vessel and actually fill the 35 vessel full of water, so what you want to do is fill the 16 vessel as fast as you can, and it doesn't matter where the 17 water comes from, as long as you are filling the vessel fast.
18 HR. SCHROCK:
My whole point is these are just 19 intuitive arguments.
I can make an intuitive argument maybe 20 very different from yours.
21 HR. HOCHREITER:
It is verified by test data.
22 HR. SCHROCK:
What test data?
23 HR. HOCHREITER:
CCTF.
24 HR. SCHROCK:
On a plant in which the scales are 25 quite different, so the radial distance that water that has
[
HERITAGE REPORTING CORPORATION -- (202)626-4888
51 1
penetrated is required to traverse is very different than it
(
).
2 is in CCTF experiments, and so you don't have any experiment 3
that really tests this aspect of the physics of the problem in 4
the PWR.
5 HR. HOCHREITER:
Again--
6 MR. SCHROCK:
That's the whole issue here is how 7
well can the code calculate what is happening in the real PWR?
8 MR. HOCHREITER:
Dr. Schrock, the UPTF tests which 9
are full scale, show that the water that was injected 10 penetrated down to the outside of the core, filled the vessel.
11 If you fill the vessel faster, the heat transfer improves.
12 MR. SCHROCK:
It may or it may not.
That's
(~N 13 something that has to be found.
If we get countercurrent flow u.)
14 in the hot channel at an earlier point, it may do a better job 15 with cooling the core than--
16 MR. HOCHREITER:
Not compared with filling the 17 vessel full of water.
18 MR. SCHROCK:
You don't reflood the whole core at 19 that rate.
20 MR. HOCHREITER:
The injection rate of those pumps 21 is large.
22 MR. SCHROCK:
Again, it is an intuitive argument.
23 What I am looking for is hov on the basis of real 24 interpretation of the calculations you can reach these.
gsb 25 MR. JONES:
They have--
HERITAGE REPORTING CORPORATION -- (202)628-4888
52 1
CHAIRMAN WARD:
Is it really fair to characterize (f
2 that as just an intuitive argument?
I mean they are 3.
calculating the refill rate.
4 MR. SCHROCK:
My point is he doesn't, he doesn't 5
really 1:now where this water that has penetrated in the 5
peripheral channel has gone in the core, and how quickly, how 7
quickly it is going to--it fills the vessel, sure, but how 8
does it fill the vessel?
9 MR. HOCHREITER:
Very fast.
10 DR. KERR:
Goes to the bottom of the vessel, 11 MR. HOCHREITER:
You have the cold leg accumulators, 12 sir, the cold leg accumulators which are filling the downcomer
/~T 13 and filling the lower plenum and the accumlators do run out
(./
14 for a period overlapping with the upper plenum injection so 15 when the upper plenum injection system is on, the accumulators 16 have already been on and they do continue to run for some time 17 period, generally have filled at least a portion if not all 18 the downcomer and the lower PWR with the accumulator flow, and so the water that is coming from the UPI system which is 19 l
20 falling down through the outside of assembly very quickly 21 starts to fill the remainder of the vessel and the core.
And 22 what our calculations indicate is that the flow basically is 23 coming down through the outside.
It crossflows over through 24 the fuel and you are basically filling the core up as a pool 25 because there is virtually no crossflow resistance in the HERITAGE REPORTING CORPORATION -- (202)628-4888
53 1
water, just filling, and so_the vessel will fill faster.
()
2
_You do get improved cooling. ~You go into more a 3
conventional reflood type behavior inside the vessel..The 4
whole' assembly acts in, and you fashion--and you don't get as 5'
much UPI penetration _directly-down into the assembly because 6
you are filling much more effectively through the colder-7 channels in the steam flows and the cooling is much better.
8 MR. TIEN:
Could I make two statements?
- First, 9
again I think it is very misleading to say those, this is a 10 nodal uncertainty.
Actually it is entirely due to the 11 physical--it is very different because near the boundary 12 peripheral bundle the updraft velocity is much lower.
You 13 have a lot of penetration from the beginning, so it has 14 nothing to do with_the noding uncertainty.
That we have to be 15 very clear.
That is totally different.
You mix that, you get 16 into trouble all the time.
17 And I think if you divided that into different 18 segment, peripheral and center, then you perform an analysis, 19 I am still nc,t sure what you say is correct because, because 20 you have a lot, you have some penetration.
You get a lot of 4
.1 steam generated.
You have the center portion of the CCFL 2
22 break down much later than normal.
You may still get high 23 peak period somewhere before you filled up the whole, you 24 know, vessel, so I think what you need is really a more
(
25 physical, you know, understanding and the, if you just say HERITAGE RE?ORTING CORPORATION -- (202)628-4888
54 1
this is noding and nodal and uncertainty, that's really wrong.
()
2 MR. HOCHREITER:
I didn't mean to confuse you with 3
that.
4 CHAIRMAN WARD:
You have to have enough nodes to 5
model these important phenomena.
But--
6 MR. JONES:
They ran--
7 HR. SCHROCK:
What I challenged is the statement 8
that if the calculation artificiallly drives the distribution 9
of liquid in the upper plenum to a more uniform radial 10 distribution than actually occurs, that this is an inherently 11 conservative situation.
I don't think you have a sound 12 technical argument to prove that.
13 It may well be that if it occurred in that fashion, 14 that you would cause a redistribution of the vapor flow 15 through the upper tie plate and that in fact you get 16 countercurrent flow into the hot channel at an earlier time, 17 these arguments that have been precented to make that point 18 have been intuitive, and not based on sound interpretation of 19 the experiments or interpretation of theoretical aspect or 20 anything else.
21 MR. JONES:
Westinghouse was using a three channel 22 model in their most, the majority of their calculations, and 23 they switched to the four channel model four or five months 26 ago based on the UPTF whatever.
That did provide some 25 verification if you wish for their intuitive argument.
HERITAGE REPORTING CORPORATION -- (202)628-4888
55 1
What we are still uncomfortable about is the rm
()
2 possibly an artificial thing, that we wanted them to go back 3
and look at CCTF for that very reason, to just make sure it 4
correlates with how they came up with their code 5
uncertainties, whatever, to make sure that the basic flow 6
anse in the core, et cetera, could be verified via the 7
experiments with that noding arrangement, so we have, we are 8
not buying totally just the intuitive argument.
That is why 9
there is a requirement that they close the loop so to speak 10 through the experiment.
Novak?
11 MR. ZUBER:
I would like to clarify.
We have some 12 data from SATF.
This is not a UPI test.
The mechanism of the
{}
13 flow is, has a bearing on this discussion.
What we found out 14 is that the guide tubes and guide tubes, the plate, the area 15 increases, and therefore there is a de-entrainment of the 16 liquid in that region and the liquid goes down along the 17 peripheral boundary and then fills the vessel from the bottom.
18 Now we don't model this in our calculation, but it 19 means we entrain more water in the hot plate which is a 20 conservative, it is a penalty, so that nodalization, what we 21 have found out really penalizes it, but we do have this factor 22 that the liquid entrained goes down the peripheral of the 33 vessel from the bottom, so this is what we are observing in 24 SATF.
If we nodalize it, pay the cost of more calculation 7S V
25 model more accurately, obtain better results, but for our HERITAGE REPORTING CORPORATION -- (202)628-4888
56-1 sake, we didn't have time or it wasn't cost effective, so we
()
2 didn't do'it, and we are paying a price for it.
3
.MR.
SCHROCK:
'I understand the water goes down the 4
peripheral channels.
It does that because it has subcooling.
5 It condenses enough steam and it overpowers whatever steam 6
flow there may be exiting through those-peripheral channels 7
and I have no doubt.that it goes down those channels, 8
What I am questioning is the a priori assumption 9
that it will be conservative if you do something in the 10 calculation that makes that water spread and not go down those 11 peripheral charnels, I don't know a basis on which we can 12 make that conclusion at this point:
()
13 MR. ZUBER:
Our calculation if we model the enlarged 14 area we have more--
15 CHAIRMAN WARD:
That was the assumption that is made 16 in the past.
17 Mk. SCHROCK:
That it is conservative?
18 CHAIRMAN WARD:
Yes.
19 MR. CATTON:
We are still doing that.
20 CHAIRMAN WARD:
With the four channel model we are 21 not doing that.
I mean to what extent are you dependent, is 22 your current analysis dependent on making that sort of 23 assumption?
s i
24 MR. HOCHREITER:
The four channel calculation now, I
)
i 25 mean we will calculate how the water spreads, because you HERITAGE REPORTING CORPOR?. TION -- (202)628-4888
57 1
inject into the outer region where the UPI injection is truly
(
2 located, and the water, some of the water will penetrate where 3
it can penetrate, so it will spread across.
4 You still get water out into the center region of 5
the upper plenum and it still flows across, but new because 6
you're primarily modeling really the low power fuel assemblies 7
on the outside of the core, the steam generation rate there is 8
lower, as Dr. Schrock pointed out, and so there is less 9
competition to hold that water up.
The water really does have 10 more, more freedom if you would, and it enn penetrate more 11 easily in those assemblies.
12 CHAIRMAN WARD:
You said that UPTF experiments shows 13 that almost all the water penetrates--
/}
14 MR. HOCHREITER:
Yes, that is correct.
15 CHAIRMAN WARD:
Your model isn't--
16 MR. HOCHREITER:
Our model would take the outside 17 region of the upper plenum and treat it as one channel.
If we 18 really wanted to get the water to penetrate, we should 19 subdivide that into sectors, and have one sector where the 20 injection is, and then as Dr. Schrock pointed out, that 21 subcooling would be better preserved and that water would 22 truly penetrate faster.
23 CHAIRMAN WARD:
So the fact that you are probably 24 not modeling the full extent of the penetration, now are you rs) i (J
l 25 claiming that's a conservative assumption?
HERITAGE REPORTING CORPORATION -- (202)628-4888
58 1-MR. HOCHREITER:
I don't think we have made the
)
2 claim that it is conservative.
If you look at tho UPTF tests 3
you would come to that conclusion.
I don't think we have made 4
the claim.
5 CHAIRMAN WARD:
Is there some way--is that d
6 contributing to your superbound number or something else?
7 MR. HOCHREITER:
We use the four channel model for 8
the superbounded, but we don't take any additional penalty on 9
it, if that's what you mean.
10 MR. CATTON:
You do argue that that leads you to 11 conservatism.
I can't recollect where it is.
12 MR. HOCHREITER:
The four channels.
[}
13 CHAIRMAN WARD:
The fact that you still have some 14 non-penetration with the four channel model--
i 15 MR. CATTON:
Two percent of the flow carrying into i
16 the stcam generator, you are arguing that there is more 17 pressure, as a result that leads you to conservatism?
18 MR. HOCHREITER:
I can't remember where that comes 19 out in the PWR calculation.
He is talking PWR.
20 CHAIRMAN WARD:
Yes.
i-l 21 MR. CATTON:
That argument made about the fact you 22 don't penetrate at all hydraulic, it is carried into the steam 23 generator--
24 MR. HOCHREITER:
I think we make a common--we are 25 talking about the CCTF tests.
HERITAGE REPORTING CORPORATION -- (202)628-4888
59 1
CHAIRMAN WARD:
Is that where your problem is, n(_)
2 Virgil?
3 HR. SCHROCK:
My problem through this whole thing is 4
that what I hear is kind of sweeping qualitative judgments 5
that I don't see technical basis for.
6 HR. HOCHREITER:
We backed that up by calculations.
7 We have backed that up with calculations.
8 HR. JONES:
We believe--
9 HR. HOCHREITER:
Now there is test data to support 10 the judgments that we have made.
i 11 HR. JONES:
That's the point I was going to make is 12 we wanted to assure that they selected a model for CCTF and we 1
1 j
[
wanted to assure the consistency to the plant calculation.
13 14 Now they have changed--we are still going to force them to go 15 back and make sure that is okay, but there is to some extent, 16 an inherent judgment that the uncertainties, et cetera, are 17 reasonable given the experimental comparison and they 18 reasonably apply to the PWR, but that, again that's part of 19 why we want to go back and close the loop on the four channel 20 model, because if they improve or, you know, depending on how 21 it changes their prediction, that might have some impact on 22 their uncertainty.
What does that mean as you carry it 23 through the process?
Does it help or hurt the 95th percentile 24 effect?
You will not issue the SER until it is done.
25 HR. CATTON:
This is the price you pay when you HERITAGE REPORTING CORPORATION -- (202)628-4888
r-60 1-argue that you are using a best estimate model when it is
()
2 really not.
I don't think, at least in my own mind, looking 3
at the CCTF UPI tests, and also the UPFF tests, there no doubt 4
in my mind what happens-when the water runs down and it, you 5
have a section of the core that is just like normal refloods.
6 It doesn't even really know that the downflow region is there.
7 Water runs down, runs into the pool, and you add to the 8
effects of reflood.
This happens in UPI.
It happens in the 9
UPTF, but here we are talking about a best estimate nodel and 10 it kluged up.
I don't belief that-you can really use it as 11 the basis for your argument.
12 CHAIRMAN WARD:
What did you say?
{}
13 MR. CATTON:
K-L-U-G-E-D-;
that you can't, you can't 14 really by itself use, use it.
It was already done.
You can't 15 use it as a basis for the argument.
I buy, by the way, the 16 use of the experimental data to come to a conclusion, but I 17 certainly wouldn't base it on your codes.
18 HR. TIEN:
Could I just give a scenario that shows 19 how complex a situation and also how sometimes difficult to 20 say conservative and not conservative?
21 Again, I'm not saying this is the case.
You have 22 some penetration from peripheral.
Depends on how large the 23 penetration, and there you have water downflow.
You have 24 steam going up.
You carry a lot of carryover droplets.
25 Now ordinary CCFL correlation does not have the HERITAGE REPORTING CORPORATION -- (202)628-4888
61 1
carryover liquid droplets which give you a lot of upper
()
2 momentum, which prevents your even vapor, upper plenum, 3
breakdown of the CCFL, and which delay your CCFL.
I don't 4
know.
That become very non-conservative.
In fact, very small 5
penetration may hurt you.
Large penetration will give you 6
very good situation.
7 Again, I don't see, I say this i s correct.
It needs 8
a lot of thinking, so that's just give scenario because CCFL 9
correlation never really put the liquid droplets, how much 10 entrainment you have, and that changed your upper momentum 11 drastically, change your CCFL correlation completely.
12 HR. YOUNG:
That was one of the reasons why we chose
/~%
13 COBRA / TRAC and I will try and explain it later in the meeting, b
14 COBRA just has dispersed drop.
15 HR. TIEN:
How do you characterize that?
It is very 16 difficult.
This has tremendous uncertainty.
I worked on that 17 myself.
18 HR. ZUBER:
If I may just amplify one moment on 19 this, we have data on this again from SATF.
You are right, 20 Mr. Tien.
You have a CCFL, the droplet entrainment, but this 21 is in the center part of the core.
That liquid is carried in 22 the upper plenum, is de-entrained.
CCTF is, I mean just is 23 not a UPI test, and the liquid saturated.
What happens when 24 the 1m center part of core liquid is entrained?
You have, 25 CCFL gets into the upper plenum, gets entrained in this HERITAGE REPORTING CORPORATION -- (202)628-4888
1 i
62 l
1 enlarged area, forced down in the peripheral channel.
You
()
2 obtain the correlation of the fluid which redistributes the 3
cooling, so this is what we observed in, just recently in the 4
CCFT tests.
It is difficult to model, but it is beneficial.
5 You obtain the circulation.
6 MR. TIEN:
I agree.
It is very possibly this will 7
be a very conservative case, but on the other hand, the 8
process is so complex, you really never know exactly unless 9
you have some good tests to really back you up.
That's what I 10 am trying to say.
I think that's what Virgil is indicating, 11 It is so complicatsd.
It is very hard to say for sure this is 12 conservative.
i 13 MR. CATTON:
It isn't the code that gives you a warm
(}
14 feeling.
It is the experiments.
MR. JONES:
That and the ability of the code to i
reflect the experiments.
I am not going to give up thut the 4
7 code is not doing a reasonable job, because if it does not, t
i 8
and we really believe that, we are done.
We shouldn't accept
.9 the model and go, i
20 HR. CATTON:
Nobody said it is not doing a J
l 21 reasonable job.
22 MR. JONES:
That is the conclusion we have come to.
l l
23 And I don't want to--
24 MR. TIEN:
I agree it is a reasonable job.
Whether 25 it is a good job or not--
HERITAGE REPORTING CORPORATION -- (202)628-4888
-63 1
MR. CATTON:
Or the basis for you concluding it is
()
2 reasonable.
3 MR.' JONES:
I would like to move on.
4 CHAIRMAN WARD:
Have you got one point here?
5 MR. HOCHREITER:
I am just curious to know what the 6
phrase kluging--would you--
7 MR. CATTON:
Would you like me to amplify?
8 MR. HOCHREITER:
I am not too sure, because I am 9
concerned about whether you are thinking the code, we have 10 taken specific models in_the code, and changed them to make 11 the code give a conservative answer, and when we model the 4
12 tests, any of the tests, we try to use the code in the best i
1 13 estimate fashion, the best models that we think belong in the
[}
14 code.
15 We do the PWR calculations, again we try to use a 16 code again in the best estimate fashion.
Now we will change 17 some models if we don't think there has been ample 18 verification of those models, and be conservative in those 19 models, and we went through whose the last time.
Okay, but 20 when we talk best estimate, we are tying to use a code in a 21 best estimate fashion.
i 22 Thermal hydraulic models, when we go and we do 23 conservative calculations, primarily the way we are putting i
l 24 the conservatism into the calculation is through the input i
I 25 and the state of the plant so that we are analyzing--we are t
HERITAGE REPORTING CORPORATION -- (202)628-4888
64 1
not going into this specifically screwdriving down heat
()
2 transfer correlations or anything else like that, so we are, 3
when we talk about using the code in a best estimate fashion, 4
we think we are, but the point or the reactor operating 5
characteristics that we are modeling we feel are conservative.
6 MR. CATTON:
I am just unsettled when you use the 7
word best estimate, and the way you divide up the core into 8
channels.
You see, if you say best estimate, that means you 9
try your best.
And you clearly don't. m.
10 Now you may give good argument as to why you don't 11 and you may argue that all these things you ought to do but 12 don't do are on the conservative side.
You may well do that.
13 But to me it is still the kluge and it is not best estimate.
{}
14 It won't be best estimate where you have sufficient number of 15 channels whenever you run that, if you are going to get 16 downflow, you get the downf*.ow, upflow in other regions.
I 17 don't see that.
18 MR. HOCHREITER:
I think that is where the four 19 channel came in.
20 HR. CATTON:
Four channels is insufficient if that 21 outer channel is a ring.
Thte doesn't detract from what Jones 22 has to do because we are still not into the real world of best 23
- estimate, You are still I think, what was it, SECY 83 24 something or other, 472, which I think allows you to not quite (q>
l 25 go to best estimate.
HERITAGE REPORTING CORPORATION -- (202)628-4888
i 65 1
HR. HOCHREITER:
Let me hypothecize a case.
Okay.
()
2 And we had some experience with this with some of the 3
calculations we did-What if we did do what you say, and we 4
did model the core up very well?
We used the best realistic 5
conditions that we could, and including things like break 6
size, power levels, peak linear heat?
The answer comes out 7
TSAT.
What do I do?
When the celebration ends, then what do 8
I do?
9 HR. CATTON:
I don't think the experimental data 10 shows that.
That is hard to imagine.
11 HR. HOCHREITEP.:
They have never run a full UPI type 12 blowdown transient.
13 HR. CATTON:
That's true.
{}
14 HR. HOCHREITER:
What do I do?
How do I run my 15 sensitivity studies from a base point of TSAT?
?.6 HR. CATTON:
That's a good question.
17 HR. HOCHREITER:
Question that doesn't have a good 18 answer.
19 HR. CATTON:
UPTF shows that is not going to happen 20 I think.
All this upper plenum or there is this hot leg 21 injection.
All it really does is enhance the reflood rate in 22 certain regions of the core.
23 HR. HOCHREITER:
If you have a smaller, more 24 realistic break, then there is a very good chance you could 3
w/
25 quench during blowdown.
HERITAGE REPORTING CORPORATION -- (202)628-4888
66 1
MR. CATTON:
Certainly.
p()
2 HR. HOCHREITER:
And then you come into reflood with 3
a quencl.ed core..
4 MR. CATTON:
Okay.
5 MR. HOCHREITER:
Your answer is TSAT.
What do I do?
6 Do I go in to the Commission and say my answer is TSAT?
7 MR. CATTON:
I don't know where not, if you are 8
dealing with best estimate.
9 MR. HOCHREITER:
I have a reason why not.
But I 10 mean when we ran the 1.0 break, okay, the answer at the end of 11 blowdown was basically TSAT.
12 MR. CATTON:
There was no reflood peak at all?
13 MR. HOCHREITER:
We didn't see any point in running (a~}
14 the transient out much further.
So what I am saying is, what 15 I am saying is if we used very realistic conditions, okay, 16 particularly the conditions more than the models, If we used 17 very realistic conditions, we would have gotten very easy 18 FCTs, when we did the sensitivity study, they probably would 19 have been meaningless.
That's why in our philosophy we took 20 more limiting plant conditions.
21 MR. CATTON:
Somehow I don't really understand.
You 22 are saying gee, I don't like the answer I am getting because 23 maybe I am going to have difficulty with the Commission so I 24 am going to screw it up and get something that is nice.
I-kJ 25 MR. HOCHREITER:
The conditions that we have put in HERITAGE REPORTING CORPORATION -- (202)628-4888
67 1
for the plant, Ivan, the conditions that we have put in are
()
2 conditions that the plant can reach.
3 MR. TIEN:
Still limits.
4 MR. CATTON:
To me there are two parts to this whole 5
business.
One is your best estimate code itself.
That's 6
thermal hydraulics.
And you want to know what the thermal 7
hydraulic models is going to do to you for some given set of 8
circumstances.
9 Now when you go to license that piece, there is a 10 whole other group of people that come in and say you have got 11 to--you have your linear power and you have got to do this.
12 You have a failed pump.
Those are separate issues.
They may
[}
drive you into this other regime, but I wouldn't be perplexed 13 14 at all.
If when you ran the plant as you think it should be, 15 it turns out it never heats up beyond saturation, that's okay.
16 You still have to test the thermal hydraulics because you 17 are--
18 MR. HOCHREITER:
I don't think I would like to come 19 in front of Committee saying a PCT that is TSAT.
20 CHAIRMAN WARD:
I don't understand that.
21 DR. KERR:
I am concerned about this because I have 22 an idea that people who are trying to teach operators, for 23 example, to deal with emergency situations, use these codes to 24 tell them what the physical circumstances are, and if what I 25 am hearing is that you do not believe that you should tell HERITAGE REPORTING CORPORATION -- (202)628-4888
68 1
people what you think the physical circumstances are, I think
()
2 we are putting people who are faced with accidents through 3
these plants are--maybe you don't believe the TSAT and if you 4
don't believe it, that's one thing, but if you really believe 5
that's what is occurring, you shouldn't tell anybody that, I 6
am concerned.
7 HR. HOCHREITER:
I didn't mean to give thac 8
impression because we did do what we called like the S0th 9
percentile type calculation.
Still used what we would still 10 consider a conservative break size based on our approach in 11 getting the break sizes, but that calculation showed about a 12 peak clad temperature of 1300 degrees, and that's probably--
13 DR. KERR:
Not so much peak clad temperature; I have
[}
14 even heard people talk about--take the codes, run the codes in 15 order to try to decide what to do, presumably best estimate 16 codes; and these best estimate codes aren't one's best method 17 of calculation, what is going on?
In the first place, I am a 18 little skeptical of their approach, but if anybody ever tried 19 to use it--
20 MR. SCHROCK:
Our problem seems to be the use of the 21 English language.
We are talking about the best estimate 22 calculation of the transient sequence following an identified 23 set of initial conditions.
24 What you are saying is that you can view this as a 25 conservative set of initial conditions versus some HERITAGE REPORTING CORPORATION -- (202)628-4888
69 1
non-conservative set of initial conditions.
The initial n()
2 conditions are the initial conditions, and they may have 3
uncertainty about them in the actual situation, but the fact 4
is that if you specify a given set of initial conditions, you 5
have a computational tool that will tell you how the system 6
responds to those conditions, and what you are trying to 7
assess is how well that, does that calculation do in 8
responding to those specified initial conditions?
9 I don't understand you.
You are confusing me with 10 this description of conservative boundary conditions.
It 11 makes no sense to me at all.
I think what, what you have to 12 do is to justify the fact that the code does calculate within 13 some degree of accuracy what actually happens for postulated
[}
14 conditions.
Okay.
And you have to postulate nany, many 15 different conditions and find out how well it does for a whole 16 range of postulated conditions.
It may do well on one set of 17 postulated conditions and poorly on another set of postulated 18 conditions.
Certainly you found out that when you tried to 19 apply the large break codes to small break problems, didn't 20 you, some ten years ago?
21 HR. HOCHREITER:
Well, we had different codes.
Let 22 me try to answer you.
We are going to come back to that.
In 23 terms of our presentation, on the conditions.
When you do do 24 the calculations, though, you are changing conditions, you 25 don't have an absolute basic.
So you are looking at relative HERITAGE REPORTING CORPORATION -- (202)C28-4888
70 1
changes and the influence of changing.one. condition versus
()
2 another.
3 MR. SCHROCK:
I don't follow you on your point 4
there.
5 MR. HOCHREITER:
You were implying the code does 6
well in one set'of conditions, does poorly in another..When 7
you are doing the plant calculation, you don't have an answer 8
to make that judgment.
All you have is the influence of the 9
change that you have made in PCT.
10 When you compare the code to data there, you have 11 the answer that you compare it to.
That's the only real truth 12 you have of the data, and so you validate the code over the 1
13 range of conditions you expect to see, the PWR.
(}
14 MR. SCHROCK:
It has been demonstrated through the i
l 15 course of history that a given code will have performed well 16 on past experiments and you do a new experiment, and low and 17 behold, you had two new codes to do a good job on past 18 experiments and now it doesn't do a good job on the new t
19 experiment t
20 MR. CATTON:
Compensating errors do not necessarily 21 scale.
22 CHAIRMAN WARD:
Okay.
Let#s try to move ahead for a 23 while here.
24 MR. JONES:
Okay.
Just to give an overview of the 25 uncertainty methodology that was employed by Westinghouse.
HERITAGE REPORTING CORPORATION -- (202)628-4888
71 1
basically we started out, and this was somewhat of a tough
().
negotiation period to get through this, but we started out 2-3 basically with the, defining what to the best of our ability 4
was, are all-the contributors to the uncertainty, and to try 5
to categorize them basically into these two groups, what I p
6 will call code development and verification, dealing with the 7
adequacy of the model itself, and the application phase which 8
is the application of the codes to che plant.
9 From there, each of these were--some were not 10 considered.
We said the effect was small.
We then went into 11 quantification of these uncertainties to quantify the code 12 development and verification of uncertainties, primarily based 13 on the comparisons of the WCOBRA/ TRAC code to the experimental
)
14 data, and consider such things as the data variability within 15 the test initial boundary conditions, the noding used for the 16 test calculations, et cetera.
Those were all part of coming 17 up with the code uncertainty.
18 There were some key models which were felt to lack 19 data, and somewhat important to the PCT calculations so that 20 there was some sensitivity studies done and bounding input 21 parameters were used.
22 Now with respect to the application phase, and this 23 is where we get into the argument over is that the best 24 estimate approach or not primarily, for applying this to the 25 plant, basically what was done is bounding input variables we HERITAGE REPORTING CORPORATION -- (202)628-4888
72 1
used, and so that they would come up with the so-called
(')
2 superbounded PCT when they bounded most of the models except 3
there were a couple, you know, that were talked about, some of 4
these so-called models where they were biased somewhat.
Most 5
of the thermal hydraulic heat transfer models were not 6
altered.
All that was altered was the input conditions for 7
the plant, things like peaking factors, power level, those 8
kind of items were set at bounded type conditions and used to 9
come up with the superbounded PCT.
10 This could, arguably you could have done this in a 11 couple of ways in coming up with the superbounded PCT or 12 thio--well, let me move--then this conservative PCT, I am 13 trying to use consistent nomenclature through this slide,
['}
14 which is this estimate of the 95th or 95th plus percentile, 15 was then what I call to superbounded PCT plus all the code 16 uncertainties along the combined--
17 MR. SCHROCK:
Initially, Mr. Jones, you said that 18 the superbound you believe has been demonstrated to be at 19 least at the 95 percent level.
How do you justify that 20 statement?
21 HR. JONES:
Let me go on.
I think that's an 22 approach on the next slide, but thet's the basic approach.
23 CHAIRMAN WARD:
Okay.
Wait a minute.
Now let's, 24 you know, look at these two categories, code development and 25 verification, and then the application.
Let's go back to the HERITAGE REPORTING CORPORATION -- (202)628-4888
73 1
issue of the fraction of UPI flow that penetrates down into
(')
2 the, into the vesselm 3
Which category is that in?
4 MR. JONES:
That would be to a large extent in the 1
5 comparisons to the experimental data, albeit limited set of 6
data.
That is the CCTF data primarily, and that is the one 7
that again that is the reason we are saying you have got to go 8
back and redo this.
Because of the change in the basic plant 9
model, you have now somewhat--you don't have the flow from the 10 experimental application of the code in the plant that we 11 desire in the noding standard to reflect that basic phenomena.
12 The code calculates, puts the water within a given 13 node and all the specific models figure out where it was 14 going, so it is kind of a noding type issue more than the 15 models themselves because size of nodes is going to have some 16 influence as to what the thermal hydraulic conditions are that 17 could be shoved into the various correlations and models.
18 HR. CATTON:
You have UPTF and we have CCTF and 19 there is a whole series of counterpart tests, and as I 20 mentioned before, coding, compensating errors do not e
21 necessarily scale.
It seems to me that somewhere it would 22 have been incumbent upon you to ask for counterpart 23 calculations to be done--SCTF, CCTF, UPTF--if you really want 24 to get at uncertainty.
Instead, I get the feeling that you 25 used discussion, rationalization, and judgment instead of just HERITAGE REPORTING CORPORATION -- (202)628-4888
1 1
74 1
grabbing hold of the data that is available and doing some
()
2 calculations.
3 HR. JONES:
UPTF is very recent.
i 4
HR. CATTON:
Very who?
I 5
HR. JONES:
Very recent, and I will be honest with 6
you.
The issue of the uncertainty, et cetera, that process in 7
our mind, was to a large extent, the code uncertainty itself 8
was finished almost a year ago, with some minor refinements.
9 So we have kind of made these decisions almost a year ago.
We 10 did not expect to be this long from there on.
11 HR. CATTON:
CCFT counterpart must have existed for 12 some time.
13 HR. JONES:
We went off, you know, the G-2 tests
{
14 which have the top down reflood to look at some of thoso.
15 HR. CATTON:
G-2 test is not relevant here.
That's 16 counterflow, Larry, end what you have in the UPI is water 17 coming down in one place and normal reflood process in 18 another.
Now you may well argue that the G-2 looks like your 19 model, but your model is not right when you spread the water 20 over the top.
21 HR. HOCHREITER:
There is still counterflow.
22 HR. CATTON:
But the heat transfer that is taking 23 place where the peak is going to occur is not the same.
You 24 have the water coming down in one part of the core, and you 25 have an almost normal reflood process going on in another part HERITAGE REPORTING CORPORATION -- (202)628-4888
e 9
75 f
1 ot the core, and that's not your G-2 test.
()
2 MR. HOCHREITER:
Some of the G-2 test results are 3
calculated.
4 MR. CATTON:
It is CCTF tests.
4 i'
5 MR. HOCHREITER:
Wers counterflow tests.
i/[
6 MR. CATTON:
In your report you don't separate them 7
out.
8 MR. HOCHREITER:
That is correct.
9 MR. CATTON:
So the way your G-2 tests are 10 presented, they really back up your models if indeed that was 11 the plan, but it is not.
12 HR. HOCHREITER:
Even with tho four channel 13 calculation, you will still get counterflow in the hot 14 assembly for some time period, and then you will go inte 1
15 alternately between co-current upflow and counterflow 16 analysis.
You have got to represent the whole core.
The 17 outside assemblies are going to be in counterflow.
18 MR. CATTON:
I hear you, but from my understanding 19 of the UPI and what the Germans have had to say, my discussion 20 with people like More11e in Japan, once that penetration takes 4
f 21 place, you can treat the problem of, you can almost separate 22 the problems, and you no longer have a G-2 type configuration l
23 so the heat transfer relations and arguments you used based on i
i 24 G-2 are not applicable.
FhECHT is more applicable to the l
l 25 reflood side, and you don't care where it is coming down as i
HERITAGE REPORTING CORPORATION -- (202)628-4888
76 P
1 long as you know the are.
-n r
(_)
2 MR. HOCHREITER:
Y~t have to remember we started 3
this program in 1985.
4 MR. CATTON:
I understand that, but we have NRR here 5
arguing the benefit, the positive aspects of the G-2 test and 6
I am just questioning it.
That's all.
It doesn't deserve 7
much more.
8 CHAIRMAN WARD:
Okay.
Let's go ahead.
p 9
MR. JONES:
Okay.
Do you have any more questions on 10 this, Dave?
You have the one that you started asking.
11 CHAIRMAli WARD:
Go ahead.
12 (Slide) 13 MR. JONES:
Now where do we cove down on the
(}
14 uncertainty methodology review process?
We think that the 15 pro:ess of identifying the contributors and sources of the 16 uncertainty was important to us and we think we have 17 identified them to the best of our ability in this position.
18 Westinghouse in their original methodology did not 19 come up tith separate PCT uncertainties for the blowdown or 20 the reflood phase of the event.
They have. however, nince the 21 last Subcommittee meeting, gone back and looked at that, and 22 basically yes, you come up with some somewhat different 23 numbers as to what these uncertainties are between the various 24 two phases of the event, but the overall impact on the PCT is 25 negligible.
HERITAGE REPORTING CORPORATION -- (20?)628-4088
f 77 1
Primarily why that comes about, because the reflood 2
tests, a large majority of the uncertainty was based on the 3
reflood tests anyway, and when you break out the blowdown 4
peak.s, it doesn't influence the reflood peak that much.
You 5
do get different numbers for blowdown, but it doesn't make the 6
blowdown exceed the reflood peak, and I think Westinghouse is 7
going to present that later.
Okay.
We have some numbers, but 8
that's the bottom line on it.
And we have looked at it.
9 MR. CATTON:
Your conclusions are 180 degrees out of 10 phase with the observations of TRAC PF1.
11 MR. JONES:
Different code, different method, 12 different approach.
(
'3 13 HR. CATTON:
Don't say that unless you know, unless v
14 you know.
From what I gather, the early COBRA-TF, its 15 internals were very similar to TRAC.
PF1 has the Rosenow 16 approximation.
It has the splitting of the heat flux, vapor 17 and liquid, and wehave all of these things in it.
They 18 changed some things, but I haver.' t seen the impact of the 19 changes anywhere ir. any of the papers I have had.
All I have 20 heard is words, and words from you saying based on judgment, l
21 it is okay.
22 HR. HOCHREITER:
I'm sorry--
23 MR. CATTON:
Can I finish?
What they are finding is 24 that the blowdown peak is about like Westinghouse.
Numbers 25 are about the same, but they find that as they go to the first HERITAGE REPORTING CORPORATION -- (202)628-4888
78 1
reflood peak and the second reflood peak, that uncertainty
{)
2 gets larger and larger and it gets larger and larger for a 3
number of reasons, and the number of those reasons one could 4
also rationalize from what Westinghouse has done, and the 5
number that is added to the calculated value goes from on the 6
order of 200 degrees up to I believe six or seven hundred 7
degrees, and you consider all the uncertainties, the three 8
peaks are the same order of magnitude.
9 Now if that's the case, this is for the sort of base 10 case that they have running, you start looking at the change 11 in areas from where the reflood peak rises, that means most of 12 your, attention should_be on the reflood peak, and uncertainty 13 is much larger.
14 Now I think within your own agency you ought to 15 rationalize this before you come to a conclusion.
Either that 16 or you ought to say that what Reeearch is doing is irrelevant 17 and doesn't matter.
18 HR. JONES:
We didn't make our judgments on the 19 research results.
We have made it on the information we had 20 in front of us from Westinghouse.
21 MR. CATTON:
I think the information you had in 22 front of you was inadequate.
23 MR. JONES:
Okay.
I mean I can only answer what we 24 have done.
We did not do the CSAU approach.
O 25 MR. CATTON:
I don't believe any rational person HERITAGE REPORTING CORPORATION -
(202)628-4888
79 1
could make the conclusion you did.
You cannot do that.
What
()
.2 you have to do is look at the data and look at the phenomena 3
and you have to conclude that you're more uncertain the 4
further you go out in a transient.
There are issues like 5
scaling.
Scaling really doesn't matter a whole lot for the 6
blowdown peak, but there are certain aspects of scaling that 7
play a very important role in the reflood peak, and nitrogen 8
plays a big role in the second reflood peak.
Here is another 9
example.
Westinghouse treats nitrogen as steam in their 10 modeling.
How could'you accept that?
11 MR. JONES:
Westinghouse also showed that by trying 12 to treat the nitrogen, they improved the heat transfer.
They 13 improved the process, dropped the temperature, so ti.at was one
(}
14 of the conservative models or one of the nodels they felt that 15 they didn't have adequate enough verification for and selected l
16' a bound so any impact the nitrogon model may have been through 17 the CSAU methodology and its influence on the peak disappears 18 because it is now conservative.
It has been treated in a i
19 conservative manner,.and that uncertainty in essence has been 20 already absorbed into the calculation.
So there are I
21 differences, l
22 MR. CATTON:
But that uncertainty is in the reflood 23 peak, not the blowdown peak.
I 24 MR. JONES:
Yes.
I said they have now separated 25 them.
l l
HERITAGE REPORTING CORPORATION -- (202)628-4888
80 1
DR. KERR:
How do you know that something is 5,)
2 conservative:when you don't know what the; uncertainty is?
3 MR. JONES:' Well, we did have calculations and we 4
looked at.the, what they were telling us and why.
We.believe 5
that what'happens is that the nitrogen comes in and depresses 6
.the water level in the downcomer which refloods the core and 7
gives you improved heat transfer.
That makes physical sense 8
to us.
9.
DR. KERR:
This may be, but if you don't know what 10 the uncertainty is in the assumption, how do you know that the 11 final result is conservative?
12-MR. JONES:
We look at it both with and without, and
{}
13 so we went with the without.
We believe that any modeling of 14 the nitrogen and its effects on the transient will improve the 15 prediction, improve the, will drop the temperature.
16 CHAIRMAN WARD:
You are saying you know the 17 direction of the effect but you can't quantify it?
18 MR. JONES:
By ignoring the effect so to speak, we 19 believe we bound any uncertainty.
That may have the benefits 20 that you would have gotten from that model.
You essentially 1
21 end up continuing the rise and instead of getting the drop in l
22 the heat transfer and starting to rise up again.
We believe 1
23 that's a conservative approach.
24 HR. HOCHREITER:
Let me try that to answer Dr.
25 Catton's comment.
HERITAGE REPORTING CORPORATION -- (202)628-4888
81 1
CHAIRMAN WARD:
Bill, are you satisfied with the I)-
2 response you got on that?
3 DR. KERR:
For the time being.
4 CHAIRMAN WARD:
Okay.
Go ahead.
5 MR. HOCf. aITER:
I think what--first of all, we 6
treat the uncertainties differently than CSAU, and we are 7
going to go through a discussion of that now.
What Mr. Jones 8
was talking about was just the uncertainty for the code, the 9
code calculation.
10 When we do the plant calculation, when we add in the 11 fact of the plant boundary initial conditions and accident 12 conditions, you are absolutely right.
The integral effect is
{}
13 penalizing reflood much more than penalizing blowdown because 14 there are fewer things in the plant initial and boundary 15 conditions that really affect the blowdown peak whereas nearly 16 everything you do affects the reflood peak because of the time 11 history of the events, and that's why in our superbounded 18 calculation the reflood peak just for the calculation itself 19 without the code uncertainties is 1814 degrees, or is nominal
'20 calculation.
It is like 1300 degrees.
There is a 500 degree 21 penalty, and that's coming from the assumptions that are made 22 on the input of plant state of the conditions.
23 Now numbers ne was referring to are just the code 24 uncertainties which we add on to that number that represent 25 how the code compares with experimental data.
HERITAGE REPORTING CORPORATION -- (202)628-4888
82
-l' MR '. CATTON:
I understand what you are'saying.
~ (/"'s,)
2-MR.-HOCHREITER:
What you implied was something 3
different.
4 MR. CATTON:
I feel you have some uncertainty.
5-associated with the blowdown peak.
You have, you have a 6
different-uncertainty associated with the reflood peak.
The 7'
uncertainty associated with the blowdown peak maybe is-on the 8
order of 200 degrees Farenheit give or take some.
I don't.
-9 remember the numbers.
The uncertainty associated with the 10-reflood peak may be on the order of five or six hundred 11-degrees.
12 HR. HOCHREITER:
Our numbers would agree to that, if 13 you add the plant conditions.
}
14-MR. CATTON:
Well, I mean this is without plant 15 conditions.
This is just a straight out fixed circumstance.
16 Now when you add your, do your calculation for your 17
,lant with all its uncertainties but which again are boundary 18
-conditions, you then have to go to the calculated value and 19 add that uncertainty.
And you add a different value to the 20 blowdown than you do to the reflood.
21 Now it seems to me that trying to treat them both 22 the same, in one sense you are penalizing yourself, and in
-23 another you don't know where the hell you are at because if it 24 is blowdown peak that is higher, I would rather add a smaller 25 number.
What are au adding?
Are you taking the average of HERITAGE REPORTING CORPORATION -- (202)628-488'
f.
83 1
the two?
P)
(_
2 CHAIRMAN WARD:
Haven't you, you have got a 3'
'different approach to this?
4 MR. HOCHREITER:
We have an approach.
We are going 5
to go through it today.
6 CHAIRMAN WARD:
Can we hear about this later?
7 MR. CATTON:
Sure.
8 CHAIRMAN WARD:
Why don't we hear about it later?
9 All right.
You know, I think it is important.
I want to hear 10 about it, but--
11 MR. SCHROCK:
Don't we have to get at the heart of 12 this problem, though, of arguing about the conservatism in the 13 calculated results in the frame of boundary conditions?
{}
14 That's not conservatism in the calculation.
That's changing 15 the-boundary conditions to produce a different transient, and 16 what we should be getting out at in this discussion is the 17 what is the uncertainty in the calculation of a given 18 transie:'t ?
Okay?
Meaning that you begin with a statement of 19 what the plant status is, and what the forcing function is 20 that causes this transient to occur; you break a pipe, and you 21 started with an initial condition, then what does the code 22 calculate?
23 What Ivan has said is just without argument it seems 24 to me.
It is true that there will be one uncertainty that is Os i
25 associated with the peak clad temperature that occurs during HERITAGE REPORTING. CORPORATION -- (202)628-4888
- ~-
84 1
the blowdown process and there will be a different one I) 2 associated with peak clad temperatures that occur in the 3
reflood phase, and it is inevitable that the second one is
~4
-going to be larger because it'is an accumulated error for a 5
longer period of time through a transient evolution.
6 MR. JONES:
Nobody is arguing that they are 7
different.
We have gone back and looked at them.
We thought 8
the approach that was taken was an acceptable approach which 9
was the lumping of them.
It turns out that the lumping of 10 them really did not alter the answers at all to a significant 11 degree.
And you know, I mean we are talking a few degrees I 12 believe difference between whether you lump them or not lump
/~%
13 them.
Yes, it does have i.'apact on the blowdown peak.
It does V
14 change where it would be, but not relative effects.
15 MR. CATTON:
There are factors cf two at least in 16 the magnitude of the delta T vhat has to be added as a result 17 of the uncertainty.
What are you using?
18 MR. JONES:
Dr. Catton, I am not justifying the CSAU l
l 19 methodology, and the answers that they are coming up with is
[
20 different than Westinghouse.
Let me give you the numbers.
l 21 HR. CATTON:
Just hold it a minute.
I am not 22 talking about the CSAU or any method.
The uncertainty is 23 larger on the reflood than it is on the blowdown.
l 24 MR. JONES:
That is not what we are coming up with.
(
25 MR. HOCHREITER:
Wait a minute.
HERITAGE REPORTING CORPORATION -- (202)628-4888
85 1
MR. CATTON:
You are not reviewing your vender ~very 2
well because he thinks he is.
3 MR. JONES:
The numbers I have got here--
4 MR. HOCHREITER:
You are talking two different 5
numbers.
That's the, what I was trying to point out before.
-6 MR. SCHROCK:
That is what we tried to point out a 7
month ago.
You have to get those things separately.
8 MR. HOCHREITER:
Those are two other numbers we talk 9
about.
Let's.try this slow.
We talk about the reflood peak, 10 only the reflood peak.
When.we put into our-superbounded 11
. calculation the changes in the plant boundary and operating 12 conditions--
(}
13 MR. CATTON:
Can we stay away from those and talk 14 about uncertainty in calculations?
.15 MR. HOCHREITER:
It is mixed in with~CSAU and that 16 is the basis you.are using for your statements.
17 MR. CATTON:
I certainly for one would like to see 18 the uncertainty in your calculational tool.
19 DR. KERR:
I would like to do--just give him a 20 chance to finish one statement at least.
-21 MR. HOCHREITER:
Thank you.
When we do the 22 superbounded calculations, we are adding the uncertainties in 23 plant operating and bounding conditions and those do affect 24 the blowdown peak more than they affect the reflood peak--they
- 5
- affect the reflood peak more than the blowdown peak.
And in 2
HERITAGE REPORTING CORPORATION -- (202)628-4888
86 1
fact that you look it our nominal calculation versus our
,.(_j 2
superbounded calculation, there is a difference of 500 3
degrees.
4 Now to that we add the code uncertainty, and we, as 5
you will see later in the presentations, we have separated the 6
data into a blowdown and a reflood uncertainty, and we will 7
add different numbers to the different peaks, okay, 8
representing different sets of data.
But when I think Dr.
9 Catton refers to numbers or degrees of five and six hundred 10 degrees, he has in there both plant and accident boundary 11 conditions in addition to the code uncertainty, and if you 12 take our numbers like the 500 degrees for the plant accident
/~
13 boundary conditions, and then you add to it the code
(_-}
14 uncertainty, and I really don't remember what our number is, 15 but it is around 200 degrees, you get the same sort of numbers 16 that Dr. Catton is talking about here, so I don't see that 17 there is an inconsistency except that we have separated them 18 by looking at the plant calculation where we included them as 19 part of plant calculation, and then adding on the code 20 uncertainty from experimental comparisons.
The end results 21 are the same, though, and we will get to that in our 22 presentations.
23 MR. JONES:
Okay.
24 MR. SCHROCK:
Do the last two bullets then remain 7sd 25 preserved in your argument?
The next to last one, I just HERITAGE REPORTING CORPORATION -- (202)628-4888
7.-
87 1
can't buy on the' face of it.
The last-one, I can't believe
)-
2 that you have a basis for it.
3 MR. JONES:
Again, it is a judgment to some extent, 4
that the superbounded case that uses bounding inputs for the 5
plant parameters does indeed represent an extremely 6
conservative case.
It is not a response-surface methodology.
7 that has been employed.
It is somewhat similar to the old 8
Appendix K approach in the sense of using bounding input.
We 9
think we have enough.
10 DR. KERR:
Does belief in the bullet mean you 11 believe it, or that it is generally believed?
12 MR. JONES:
That belief or judgment is ours.
I mean 13 that is I think where Mr. Schrock and I maybe have different
(}
14 beliefs.
15 MR. SCHROCK:
Absolutely; 180 degrees opposed, 16
. guaranteed.
17 MR. JONES:
You know, the--
18 DR. KERR:
You don't mean that you don't believe 19 that he believes that, do you?
20 MR. SCHROCK:
I think he believes it, but he has no, 21 no basis for believing that.
22 MR. JONES:
It is our engineering judgment for what 23 it is worth, you know, that by setting these things to 24 bounding conditions, things like power, the uncertainty in the 25 power level, the tech spec value for peaking factor, how you HERITAGE REPORTING CORPORATION -- (202)628-4888
88 1
model the rod for stored energy, all of those which are the,
()
2 probably the most important factors for coming up with the 3
increase in the PCT, drive the calculation to a highly 4
conservative method in Appendix K.
5 DR. KERR:
At the rick of some redundancy, I am 6
going to ask Mr. Schrock to repeat his statement about what a 7
code calculates.
8 CHAIRMAN WARD:
Bob, I guess I--are you claiming 9
that, that the uncertainty, I mean you are really wanting to 10 take conservatism in plant conditions to account for 11 uncertainties in the code?
12 HR. JONES:
No.
That's not what I am trying to say.
(~)/
The code uncertainty war, calculated by the comparisons to tne 13 14 data, and all the other manipulations we had to go through the 15 data variability, the test and boundary conditions, the 16 noding, et cetera.
That is the cade bias, the code 17 uncertainty.
That gets added to this superbounded case which 18 accounts for the plant parameter uncertainty so to speak.
19 CHAIRMAN WARD:
Two separate things.
20 HR. JONES:
To come up with this conservative PCT 21 which we are saying is above the 95 percent; we are saying you 22 exceed 95 percent primarily because we believe this, with 23 respect to the plant application phase, is in excess of 95 24 percent for the uncertainty associated with input parameters.
25 The code uncertainties we think is a reasonable estimate of HERITAGE REPORTING CORPORATION -- (202)628-4888
89 1
say the 95 percentile uncertainty.
Okay.
n t) 2 HR. SCHROCK:
The thought in this is that you are s
3 falling back to the same kind of reasoning that led to 4
Appendix K in the beginning, and that is that if you make what 5
you believe to be conservative assumptions at various stages 6
along the way in the calculation, that you can be assured that 7
the end result is very, very conservative.
8 Well, what you have, and in a transient type of 9
problem such as this then, the so-called evaluation model goes 10 off calculating something that is totally unrelated to what 11 the plant will really do, and that's not a good kind of 12 calculation to be making because it makes no sense in the 13 physical realm.
You are not describing in any realistic sense
{}
14 what will actually occur in the physical system.
15 What you are doing by these arguments is retree.cing
- 1. 6 back to the mentality of Appendix K, and that's in effect what 17 we are trying to get away from.
18 MR. JONES:
Let me--the issue of how you treat the 19 uncertainty in the input parameters we have all, we even say 20 in the reg guide that we are going out with the new rule that 21 you can use bounding numbers if you want to.
It is in the reg 22 guide, to reflect the plant, and that's what I am talking 23 about.
24 CHAIRMAN WARD:
To what extent are you using 25 bounding plant parameters to help you cover for your ignorance HERITAGE REPORTING CORPORATION -- (202)628-4888
90 1
or uncertainty in the code?
That's the issue here.
_( )>
.2 MR. JONES:
Generally we are not really relying on 3'
it.
Now there is a little bit of uncomfortableness say.with 4
the scaling argument.
Zero; that's a little shaky to us, 5
but--
6 CHAIRMAN WARD: ' Why don't you pump up the 7
uncertainty allowance or whatever for the, keep it clean, if 8
that's where your uncertainty is?
9 MR. JONES:
I still have to come up with the what do 10 I plug in?
11 CHAIRMAN WARD:
You don't know what number you are 12 plugging in when you just make the plant calculations a little 13 more conservative,'either.
You don't nave any basis for that
{
14 than just guesses at a hundred degrees for the other one.
15 HR. JONES:
There is some justification provided i
16 that says scale--there is apparently no scale bias.
17 Westinghouse made argument for that.
There is some 18 indications in the CSAU methodology that scale seems to be 19 negative.
20 CHAIRMAN WARD:
Why allow for it in the more R21 conservative plant calculations?
l 22 HR. JONES:
It is their choice.
Okay.
It was their 23 choice.
You have at this point in time, once you have got the 24 code uncertainty, you are basically down to okay, how do I O
25 treat the plant parameters?
And I have got two ways I can go HERITAGE REPORTING CORPORATION -- (202)628-4888
91 1
about it.
I can go out and vary these numbers and do some
/m.
(_)
2 sort of a response surface to come up with a number, or I can 3
on the other extreme, it is, the set bounding values come up 4
with a highly conservative number for doing the plant.
5 What Westinghouse has chosen is the latter.
6 MR. CATTON:
That doesn't reduce our uncertainty in 7
the calculations at all.
8 HR. JONES:
No.
It has no impact on the 9
uncertainty, of what we judge to be the code uncertainty.
10 HR. CATTON:
You really have to separate I think 11 the, your calculational tool for the thermal hydraulics from 12 the plant.
(~)
13 MR. JONES:
That is what we tried to do.
x_/
14 HR. CATTON:
You didn't.
You wind up addressing all 15 questions about the uncertainty in the calculational tools by 16 saying but we tre more conservative in the plant.
That's 17 begging the issue.
18 HR. JONES:
That is not what I said.
That is not 19 what I said.
What I have said is this is how we determined 20 the code uncertainty.
This is how we did it.
21 HR. CATTON:
It is probably us, but I don't 22 understand how you did it.
23 MR. JONES:
This is really how we did it.
We looked 24 at code uncertainty.
We tried to pick on that if you wish, as g3 (J
25 much as we could to come down with a number that we felt was HERITAGE REPORTING CORPORATION -- (202)628-4888
92 1
comfortable and had a reasonable basis.
Then we looked into D)
(,
2 how we are treating the plant uncertainties, what you are 3
reusing and why are you using it, how are you are putting it 4
in your model.
Why do you think it is 95 percent, et cetera?
5 I mean there were earlier approaches.
We were talking about 6
.how to statistically combine various calculations to come up 7
with the impact of the plant uncertainty, and this is where 8
they ultimately fell down.
9 MR. CATTON:
If I look in the SER, I should find 10 some kind of a statement then as to what you believe the 11 uncertainty of the calculational tool is?
12 MR. JONES:
Uncertainty of the calculational tool I 13 can tell you.
I can tell you that, and I think Westinghouse
}
14 will tell you that later.
I didn't do it as part of their 15 presentation in closed session.
16 CHAIRMAN WARD:
They don't quite trust that, so when 17 they go to define the uncertainty in the plant input 18 parameters, they Jack that up a little bit to make up for it.
19 Sounds like that.
20 MR. HOCHREITER:
Mr. Chairman, that was never the 21 intent.
These two things are treated separately.
We have to 22 take the code.
As Dr. Catton said, we compare the code to l
23 data.
We define the uncertainty for the code based on i
l' 24 comparisons of test data.
When it comes to the defining the
()
25 uncertainty ander plant conditions, that there are several HERITAGE REPORTING CORPORATION -- (202)628-4888
m 93 1
options.
You can use an nominal calculation and you can go
( f' 2
through a design of experiments type series of calculations, 3
like those done in CSAU and.all the plant input parameters, 4
develop a response surface, get a probability, PCT probability 5
distribution, and come up with a 95th percentile and its 6
uncertainty, and then you have to combine that uncertainty 7
separately with the skewed PCT.
THAT's what CSAU is doing.
8 What we did was we said let's envelope and bound the 9
plant in accident boundary conditions.
Let's examine 10 everything that can vary in the calculation for the plant, and 11 take it as bounding and then add to it separately the 12 uncertainty for the code that comes from the experiment.
13 There is no tradeoff in our method.
[}
14 CHAIRMAN WARD:
Okay.
Well, that's fine.
That's 15 good.
That sounds great, but that doesn't seem to be what Mr.
16 Jones is saying.
17 MR. JONES:
That is what I am trying to say.
18 MR. CATTON:
I think he clearly brought out--and 19 that is that you are not carrying your uncertainties in 20 blowdown calculational capability in the reflood.
21 MR. HOCHREITER:
For the plant?
22 MR. CATTON:
No; just in general you are not 23 carrying the uncertainties in your ability to calculate 24 blowdown into your calculation of reflood.
You are taking t.s 25 PCT separately, and I think basically that's the heart of the l
HERITAGE REPORTING CORPOhATION -- (202)628-4888
94 11
-matter.
r'%
(_)
2 MR. HOCHREITER:
In'the representation.
3 MR. CATTON:
I am'just telling you now the 4
observations.so you can address it.
l
'S MR. HOCHREITER:
That will be addressed in our 6
presentation later.
7 MR. JONES:
Dave, what I am'saying, I am going 8'
through--this process is essentially what Larry said.
We 9
looked at the code separately.
We looked at the plant 10 application separately, and what I am just calling 11 conservative PCT has just been consistent with the, with the 12 phrase I used from SECY 83-472, which is this conservative PCT
(}
is just an expression.to represent tho 95th percentile PCT or 13 14 thereabout.
15 What we are saying is the conservative PCT which is 16 the sum of this plant calculation, which is set at bounding 17
_ condition, plus the code uncertainties, we believe is in 18 excess of 95 percent.. An argument is based on this, we think 19 this is quote, at 95 percent, this is something greater than 20 that.
Therefore, we think the overall 95 percentile or 21 conservative PCT as defined in 83-472 must exceed 95 percent.
22 That's all we are saying here.
23 CHAIRMAN WARD:
For some reason, you are being a 24 little more conservative in the plant conditions, but Larry 25 says that's not the, to make a tradsoff.
That's just for some HERITAGE REPORTING CORPORATION -- (202)628-4888
95 1
unknown reason, unknown to us anyway.
You are being a little q_)
2 more conservative in that side of it.
3 MR. JONES:
That was their choice.
4 MR. HOCHREITER:
That is, we thought that from an 5
economic point of view in terms of the number of calculations 6
we would have to run, we thought that this an easier thing to 7
do than to try and develop it on the surface.
As you see by 8
the number of calculations we did run, it is not clear that 9
that was the case.
10 MR. JONES:
What I would like to do now is I don't 11 think there is much interest in the specific Appendix K calc 12 which is the next slide because I think that's, we have 13 discussed that at the last meeting anyway.
Let's move on to
{)
14 the second issue which is future EM review.
15 CHAIRMAN WARD:
Let me just look at this.
16 MR. JONES:
Unless you have got a question on it.
17 CHAIRMAN WARD:
Well, I guess the, maybe you will 18 get to this in the next--
19 MR. JONES:
There is not much in the next one.
I am 20 not sure what you were after, so multiple questions mostly on 21 the next one, but--
22 CHAIRMAN WARD:
I guess maybe I could just put it in 23 terms of a question.
The new rule is going to be approved and 24 available for you when?
25 MR. JONES:
Decemberish?
HERITAGE REPORTING CORPORATION -- (202)628-4888
96 1
MR. ZUBER:.The rule is now in front of the
~d V
2-Commission.
3 CHAIRMAN WARD:
Okay.
So-perhaps later this year.
4 That's not important.
5 When that's available, what sort of approaches to 6
qualifying a core or whatever you call it, for large break 7
LOCA and ECCS performance will you accept?
Are there' going to 8
be three approaches--the old traditional EH approach,83-472, 9
and the new rule--or is83-472 going to disappear or what?
10
-MR.
JONES:
Let's--yes.
Really after we revise the 11 rule, there will he three ems allowed in the sense that we 12 have evaluation models allowed, just evaluation models meaning 13 compliance with the rule.
There will be the use of the j"
14 current approach, Appendix K.
There will be the SECY 83-472 15 approach, and the best estimate calculation approach under the 16 revised 50.46.
17 Now the reason I say they are allowed is a sense 18 that we have approved some of these SECY 83-472 models, and if 19 people wish to continue to use them, they can use them.
There
'20 is nothing that says you can't.
We will not entertain future 21'83-472 models.
It doesn't make sense primarily because really 22 the standard on the 83-472--I am going to get my foot in my 23 mouth on this one, but essentially the standard 83-472 for the 24 best estimate calculation in 95th percentile is the same as 25 the revised 50.46.
. HERITAGE REPORTING CORPORATION -- (202)628-4888
97 1
As I say, I put my foot in my mouth.
Basically the
().
2 way Westinghouse did business in this calculation, to come up 3
with this 95th percentile calculation with the superbounded, 4
plus the code uncertainties, that would meet the intent of the 5
new rule.
6 We have got to check a few little things, but in 7
essence it meets the intent.
Now we would expect that the 8
83-472 models will, will basically become converted into 9
compliance with the revised EM.
There would be some paperwork 10 that would have to be done.
We would have to connect a few 11 items.
I do not belief that is going to be a large, large 12 efforts So titat effectively in the future we would expect to 13 get down to 2 EM, two LOCA calculatienal approaches, 1.ppendix
}
14 K,
or best estimate.
That's what we think we are going to 15-have.
16 CHAIRMAN WARD:
Okay.
Because 83-472 won't amount 17 to anything significantly different from the estimates?
18 HR. JONES:
I don't believe so.
I checked with the 19 guy that did the GE review of the SAFR model.
He didn't think 20 it would be a big deal.
And I think in the case of 21 Westinghouse's model, we certainly wouldn't do much 22 different--one minor exception dealing with it is possibly 23 with the blowdown peak and the separation.
Now that we are 24 going to force, if that becomes a licensing standard and how 25 they might apply that in the future, we may need to put some HERITAGE REPORTING CORPORATION -- (202)628-4888
98 1
restrictions on. -That's what I see in here as maybe I) 2 additional restrictions or something, depending on how we 3
bless.these models.
4 CHAIRMAN WARD:' It could be as the CSAU method is 5
developed more fully, you know, is completed, that I mean 6
that's going to be a tool for the staff to judge the 7
acceptability of an application.
And it may be that, that 8
some different judgments will be made two years from now using 9
that as a tool that you made over the last year reviewing this 10 application.
Is that a possibility, or is that an 11 understandable question?
12 MR. JONES:
I understand what you are saying.
What 13 I can't do, I can't tell you exactly how we will use the CSAU
[
14 standard or approach on a given applicatien.
15 What we will be looking at primarily, which is one 16 of the reasons, one of the things that Wescinghouse is going 17 to address next, are-you applying the basic features of the 18 CSAU methodology as part of your, the applicant's way of 19 coming up with their code uncertainty?
Are they applying it 20 either literally or basically trying to get at the same type 21 of issues, and maybe biased to a conservative manner such as 22 the plant application uncertainties.
Hay get applied still 23 conservatively just on a cost basis, but because I don't know 24 how we would use the CSAU method ourselves with somebody 25 else's code--I mean we couldn't really do that very HERITAGE REPORTING CORPORATION -- (202)628-4888
.99 1
conveniently, I don't believe.
L
)
2 The other concern that I have heard is the staff 3
willing_to or will the staff put the resources necessary to do 4
this job, the review of best estimate codes?
And you know, I 5
guess I addressed that a couple of months ago at one of the 6
committees.
I guess it was this one.
7 CHAIRMAN WARD:
Would you repeat--let me ask the 8
question.
Is it going to be a bigger technical job for the 9
staff to review best estimate applications than Appendix K 10 applications?
11 MR. JONES:
Definitely; not even a doubt about it.
l 12 especially since most of the Appendix K models are there; it 13 will definitely be bigger.
It is more complicated, and it 14 will be a much larger affort.
15 We also expect we are going to have to use a 16 contrector.
It is not going to be done totally in house.
17 There is no doubt about that.
I 18 The one thing I do not want to mislead anybody on is 19 we are not going to go out and jump up and down to start 20 reviewing these things.
As these applications come in, we are 21 going to put them through our priority system as far as, you 22 know, how fast you are going to get review, if at all.
May be 23 put off for a few years.
24 CHAIRMAN WARD:
That will tak? care of it.
25 MR. JONES:
Really we are jumping at safety l
HERITAGE REPORTING CORPORATION -- (202)628-4888 i
100 1
significant issues such as mid-loop and you know, at this
,,(,)
2 point in time, Station Blackout, et cetera, we are going to 3
have much higher priority for responding to plant problems, 4
inspections, et cetera, than we--what we view thic primarily 5
being, as leading to, while I agree that there are safety 6
benefits associated with it, the fact of the matter is how it 7
is going to be reflected primarily we think it is just goir.g 8
to be improved flexibility for the plant's operation, 9
flexibility, more margin, maybe more economics, but the staff 10 resources are such that it is, we really don't have a lot of 11 people to put on that.
But when we do it, we are not going to 12 go out and just say well, it is low p.;ority or not.
That's
~ '
11 important, yes, that's fine, we expect to do as, do a complete 14 review although some people may disagree with this last 15 one--was a complete review at least from a rescurce standard 16 point.
We will apply the resources necessary.
It will be 17 stretched out scheduling.
RES has discussed possibly helping 18 us out on that, and we very well may take them up on their 19 offer.
We will use them as a contractor, kind of in-house 20 contractor to us, but that's, that's where it is at.
21 HR. SCHROCK:
It seems to me your first statement is 22 a very troublesome one--review effort expected to be similar 23 to that under SECY 83-472.
This exercise on the UPI in my 24 mind, has been a disaster from the standpoint of applying the f~k-25 intended regulatory process under SECY 83-472.
I don't think HERITAGE REPORTING CORPORATION -- (202)628-4888 t
r 101 1
you have done that.
Now you are telling us that the new rule
(_)
2 is going to be interpreted in the same fuzzy way that you have 3
interpreted SECY 83-472, and that really bothers me.
4 I would rather that this whole exercise had never 5
occurred and that you didn't have that crutch to lean on as 6
you go into the use of the new rule.
7 MR. JONES:
I think we have been faced with this 8
kind of an approach anyway just by the cost of doing some of 9
these calculations.
I think we would have probably started 10 seeing some cutting the corners from the standpoint of cost.
11 And we would have had to make the judgment.
12 MR. SCHROCK:
In effect what you have told us is the 13 staff has the authority to do it the way the staff concludes
)
14 it has to do it given resources, capabilities, et cetera, et 15 cetera, and quite independent of the background of what has 16 occurred to improve the basis of licensing of reactors.
There 17 is no improvement at all if you do your job in that way.
18 MR. JONES:
In a sense, what you are getting, in its 19 simplest terms, is you are getting a thermal hydraulic model 20 which will better reflect the overall system response.
When 21 you start to deal with a specific plant application, and 22 coming up with some bottom line r sker if you wish for what 23 the 95th percentile PCT is, it in that point that you may 24 still end up with some conservative bounding applications.
25 I mean I, my understanding is that by switching to a HERITAGE REPORTING CORPORATION -- (202)628-4888
102 1
bounding calculation versus something like a nominal
()
2 calculation, the basic overall system response, system 3
hydraulics, don't significantly change.
All that changes is 4
you are starting with higher fuel rod temperature so you go up 5
a little higher in your initial blowdown peak, and because you 6
have got more power, it takes a little longer to reflood the 7
core in the hot channel in the long-term.
You change what the 8
PCT response looked like but the overall system response in 9
how it comes down, comes across, fills up, does not 10 significantly change by changing the plant parameters.
11 CHAIRMAN WARD:
Okay.
Thank you.
Let's take a 12 break for fifteen minutes.
13 (A brief recess was taken.
Please refer to page 105
[}
14 of the closed session transcript for continuation of the 15 proceedings.)
16 17 18 19 20 21 22 23
(~)
25 HERITAGE REPORTING CORPORATION -- (202)628-4888
1 CERTIFICATE
()
3 This is to certify that the attached proceedings before the 4
United States Nuclear Regulatory Commission in the matter of:
s 5
Name:
ACRS--Subcommittee on Thermal Hydraulic. Phenomena 6
7 Docket Number:
8 Place:
Washington, D.C.
9 Date:
June 21, 1988 10 were held as herein appears, and that this is the original l
11 transcript thereof for the file of the United States Nuclear 12 Regulatory Commission taken stenographically by me and, thereafter reduced to typewriting by me or under the direction 13 14 of the court reporting company, and that the transcript is a
(}
trueandaccuratsrecordoftheforegoiggprocawdings.
15 16
/S/
b_ h
'7 4
17 (Signature typed):
Catherine S, Boyd 18 Official Reporter 19 Heritage Reporting Corporation 20 21 22 23 24 a
25
()
Heritage Reporting Corporation (202) 628-4888
O-
.O
,i l
l REVIEW STANDARD FOR BEST ESTIMATE j
LOCA ANALYSES AS APPLIED TO THE l
WESTINGHOUSE TWO-LOOP PLANT ECCS EVALUATION MODEL 1
1 l
i j
l j
PRESENTED TO PRESENTED BY l
ACRS SUBCOMMITTEE
R.
C. JONES ON T/H PHENOMENA NRR/ DEST /SRXB
)
J U N E 21, 1988 492-3232 WUP11
=..u.w1 O
O
,C SECY 83-472 REQUIREMENTS l
t
- l l
t PERFORM ANALYSES WITH BEST ESTIMATE CODE
+
CALCULATE "REALISTIC" (~ 50%) AND "CONSERVATIVE"
+
(~95%) PCTs
+ "CO NSERVATIVE" PCT ACCOUNTS FOR UNCERTAINTIES IN:
- CODE PREDICTIONS TO EXPERIMENTS l
PLANT OPERATING PARAMETERS l
NUCLEAR PARAMETERS NOT CONSIDERED l
UNCERTAINTY METHOD MUST BE DEVELOPED BY APPLICANT l
+
t CALCULATE PCT WITH BE CODE AUGMENTED WITH REQUIRED FEATURES OF APPENDIX K (NEW EM)
APPENDIX K MUST BE GREATER THAN "CONSERVATIVE" PCT
+
BUT LESS THAN 22OOF WUPl2
O W
i t
OVERVIEW 'OF NRR REVIEW PROCESS l
I
{
- WESTINGHOUSE MODEL AND METHODOLOGY WAS REVIEWED AS A WHOLE TO DETERMINE WHETHER REASONABLE j
ASSURANCE THAT 10 CFR 50.46 WAS M ET l
l SIGNIFICANT ISSUES PERTINENT TO SECY 83-472 l
+
IS COBRA / TRAC A BEST ESTIMATE CODE SUITABLE FOR CALCULATING UPI PLANT RESPONSE?
l
- DOES THE SUPERBOUNDED PCT PROVIDE A PCT WHICH
{
IS AT LEAST AT THE 95 PERCENTILE AND HAVF UNCERTAINTIES BEEN ACCOUNTED FOR?
i
- ARE APPENDIX K REQUIRED FEATURES PRO PERLY l
lMPLEMENTED IN THE EVALUATION MODEL?
REVIEW PERFORMED WITH THE ASSISTANCE OF SANDIA
+
AND WAS A SIGNIFICANT EFFORT (2.5 YRS /4 PSY/8350K)
WUPl3
i_
9
,O l
l-i REVIEW PR CESS FOR WCOBRA/ TRAC l
f EXAMIN ED FIELD EQUATIONS, CLOSURE RELATIONS AND MODELS TO DETERMINE GENERAL CAPABILITY TO MODEL IMPORTANT PHENOMENA lt l
EXAMINED COBRA / TRAC DEVELOPMENTAL ASSESSMENT TO j
+
DETERMINE OVERALL ADEQUACY OF CODE AND ITS l
ABILITY TO MODEL UPI PHENOMENA l
1.L I
REVIEWED WCOBRA/ TRAC MODIFICATIONS TO ENSURE l
l THAT UPI PHENOMENA WERE PROPERLY MODELED l
REVIEWED WCOBR y' TRAC COMPARISONS TO DATA TO DETERMINE OVERALL ADEQUACY OF CODE WUPl4
..p...
t h
,U t
WCOBRA/ TRAC REVIEW F NDINGS l
EXTENSIVE COBRA / TRAC' DEVELOPMENTAL ASSESSMENT QL'ALITATIVELY
+
DEMONSTRATES CODE ADEQUACY
- WESTINGHOUSE IMPLEMENTATION OF UPI SPECIFIC MODELS ENABLES WCOBRA/ TRAC TO SIMULATE MAJOR PHENOMENA
+ WCOBRA/ TRAC FIELD AND CLOSURE EQUATIONS CONTAIN BASIC FEATURES a
NEEDED FOR BE ANALYSIS OF UPI
- THREE-FIELD TWO-PHASE FLOW MODEL ALLOWS TREATMENT OF l
SIMULTANEOUS LIOUID FILM AND DROPLET FIELDS f
- CONSTITUTIVE MODELS CAN PREDICT MAJOR UPI PHENOMENA l
l
- WESTINGHOUSE WCOBRA/ TRAC DATA COMPARISONS
- REASONABLE AGREEMENT OBTAINED WITH PCT DATA
- OVERALL SYSTEM RESPONSE, e.g. PRESSURE. FLOWS, IN GENERAL AGREEMENT WITH EXPERIMENTS i
- NO APPARANT COMPENSATING ERRORS OBSERVED IN THE l
COMPARISONS
" WCOBRA/ TRAC IS SUITABLE FOR BEST ESTIMATE UPI ANALYSIS WUPIS
9
,9 l
U NCERTAINTY M t_ i HODOLOGY LIST CONTRIBUTORS OF UNCERTAINTY AND CATEGORIZE l
- CODE DEVELOPMENT AND VERIFICATION l
- APPLICATION PHASE 1
l' i
QUANTIFY CODE UNCERTAINTIES
+
COMPARISONS TO EXPERIMENTAL DATA
- DATA VARIABILITY, TEST INITIAL AND BOUNDARY CONDITIONS
- NODING
+ IDENTIFY KEY MODELS LACKING DATA AND IMPORTANT TO PCT CALCU LATION
- PERFORM SENSITIVITY STUDIES i
- USE BOUNDING MODEL USE BOUNDING INPUT PARAMETERS FOR APPLICATION PHASE
+
s (SUPERBOUNDED PCT) l CALCULATE "CONSERVATIVE" PCT AS SUPERBOUNDED PCT PLUS
+
CODE UNCERTAINTY ADDERS WUPl6
..3.-,,...
., h UNCERTAINTY METHODOLOGY REVIEW FINDINGS l
t
- IMPORTANT CONTRIBUTORS AND SOURCES OF UNCERTAINTY IDENTIFIED
- WESTINGHOUSE METHODOLOGY DOES NOT CALCULATE SEPARATE PCT l
UNCERTAINTIES FOR BLOWDOWN & REFLOOD PEAKS
- SEPARATE PCT UNCERTAINTIES EXAMINED, DID NOT MODIFY
!-ti r
OVERALL RESULT
+ CODE UNCERTAINTIES OTHER THAN THOSE RELATED TO PWR APPLICATION ADDRESSED
- PWR CALCULATION USED SIMILAR NODING TO THAT FOR CODE ASSESSMENT
- EXAMINED FOR COMPENSATING ERRORS, NONE APPARANT i
- NO CLEAR SCALING BIAS OBSERVED t
t USE OF BOUNDING INPUTS FOR APPLICATION PHASE JUDGED j
+
CONSERVATIVE
- ESTIMATES INDICATE THAT SU PERBOUNDED CASE MAY HAVE SEVERAL HUNDRED DEGREES CONSERVATISM t
"CONSERVATIVE" PCT BELIEVED TO EXCEED THE 95 PERCENTILE
=
REQUIRED BY SECY 83-472 MCPl7
.)
~_
i l-t I
APPENDIX K MODEL CALCULATION l
+ METHODS FOR COMPLIANCE WITH APPENDIX K REQUIREMENTS PROVIDED REQUIRED APPENDIX K MODELS (E.G. MOODY) ADDED l
+
- APPROPRIATE INPUTS USED IN CALCULATION I
+ REQUIRED SENSITIVITY STUDIES PERFORMED TO IDENTIFY i
I'l WORST CONDITIONS l
+ NON-COMPUANCE ITEMS:
i
- A CODE MODIFICATION WHICH DOES NOT ALLOW RETURN TO NUCLEATE BOILING WILL BE REQUIRED I
- EXEMPTIONS REQUIRED FOR TWO APPENDIX K REQUIREMENTS WUPIB
Y
.a i :
l l l 1 i
O YSB2 B E X3 NR2 D OS3 EJ /
T
.T2 N
S 9 E C E 4 S
D E R/
R R
P R
S N
WE I
VE R
9 M
E E
R U
k A E N T
T E T
U M
I F
MO8 O MN8 T OE 9 D CH1 E B P TU 1
NS H2 E
/
SST E ER N
R CN U P AOJ
9
.O l.
i FUTURE EVALUATION MODELS
- At-iER 10 CFP, 50.46 REVISED, THREE ems ALLOWED CONTINUED USE OF CURRENT RULE (APPENDIX K
h APPROACH)
- SECY 83-472 APPROACH (8EST ESTIMATE +
APPENDIX K CALCULATION)
REVISED 50.46 (BEST ESTIMATE CALCULATION) r SECY 83-472 BEST ESTIMATE CALCULATION (~95%)
IS j
ESSENTIALLY THE SAME AS THE REVISED RULE.
l MINIMAL REVIEW REQUIRED TO CONVERT FROM l
SECY APPROACH TO NEW RULE l
WILL ELIMINkTE THE REQUIRED APPENDIX K.
f I
CALCULATION l
4 LOCA2 1
Q i
s f
FUTURE BE' LOCA AN ALYSIS R EVIEWS j
REVIEW EFFORT EXPECTED TO BE SIMILAR TO THAT UNDER l
l CONTRACTOR ASSISTANCE WILL BE NEEDED i
I STAFF RESOURCES ARE LIMITED AND ARE APPLIED USING A l
PRIORITY SYSTEM
- SAFETY SIGNIFICANT ISSUES HAVE HIGHEST PRIORITY l
- IMPROVED FLEXIBILITY LOWEST PRIORITY
- BE LOCA CALCULATION REVIEWS WILL NOT HAVE HIGH PRIORITY STRETCHED buT REVIEW SCHEDULES POSSIBLY HAVE REVIEWED BY RES LOCA1
,