Thermal-Hydraulic Phenomena - January 18, 2002
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
Thermal-Hydraulic Phenomena Subcommittee
Docket Number: (not applicable)
Location: Rockville, Maryland
Date: Friday, January 18, 2002
Work Order No.: NRC-177 Pages 612-628
NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers
1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005
(202) 234-4433UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
+ + + + +
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
THERMAL HYDRAULIC PHENOMENA SUBCOMMITTEE
+ + + + +
JANUARY 18, 2001
+ + + + +
+ + + + +
The subcommittee met at the Nuclear
Regulatory Commission, Two White Flint North, T2B3,
11545 Rockville Pike, at 8:30 a.m., Graham Wallis,
GRAHAM B. WALLIS, Chairman
THOMAS S. KRESS, Member
VIRGIL SCHROCK, Consultant
Ralph Landry . . . . . . . . . . . . . . . . . . 615
Adjourn. . . . . . . . . . . . . . . . . . . . . 628
MR. LANDRY: I think I'll sit down for
this next couple of minutes. Hopefully I won't say
anything stupid to disclose proprietary material in
the next few minutes.
MR. BOEHNERT: We'll shoot you if you do.
MR. LANDRY: What I'd like to do --
MR. BOEHNERT: He's going to go back to
his original slides.
MR. LANDRY: Going back to my original
slides. What I'm going to do is skip up to page six
and then to page eight because we've just had the
presentation on the review that is being performed on
the statistical methodology.
Some of the comments that the subcommittee
made with regard to S-RELAP 5 for the small break LOCA
and I've lumped some of the comments together in just
identifying three items. One thing the committee said
was that the staff should insist on complete
documentation before issuing a final SER. When we
look at the material that has been submitted for the
realistic large break LOCA on S-RELAP 5, our
impression at this point is that there is a great deal
of material and, in fact, far more material than we
typically see on any submittal of a code.
Whether it is the absolute perfect
complete set, I'm not sure we could say right now but
I would say that this is a very large set of material.
It includes a full set of manuals on the code itself
and on the subsidiary codes. It includes other
documentation which is contained on the three CD roms
which we have received.
One of the points that was brought up
yesterday by Professor Schrock was looking at the
decay heat model and how that model has been
substantiated. The vendor has produced licensing
calculations or calculation notebooks -- excuse me --
along the way which they do on every calculation they
do. Normally, we do not ask for submittal of
calculation notebooks. They are available for audit
at any time. We will discuss with them further though
whether part of the justification for the decay heat
model which they are using should be supplied as
additional material, perhaps in an RAI or some
submittal form. But we will discuss that with them
further. We do not anticipate asking for the
calculation notebook, but perhaps we should put on the
record an extract from that notebook which justifies
what they have done.
Another comment that the committee made
was that the staff needs to consider how broad-based
the assessment of realistic LOCA should be, not only
to ensure adequacy but also to the measure of
uncertainty. We've been making comments throughout
this presentation. Doctor Orechwa just gave you his
views and the approach he's taking to doing the
review, and we will continue to discuss this, look at
developing his views further, and we would look
forward to feedback from the subcommittee on your
views on what he has said and the approach he has
Another comment that the committee made
was that there should be an independent evaluation of
code runs when S-RELAP 5 is submitted as realistic
LOCA. We are doing that, as I indicated yesterday.
We have at least one person now --
CHAIRMAN WALLIS: That means running the
MR. LANDRY: Running the code ourselves on
input models but also running parts of the code,
looking at models and sub-routines within the code,
varying parameters within those models and
correlations to determine the effect on the result.
We have a person who is putting in time working with
the code, worked from the source code, modifying the
models, modify parameters, recompile the code and run
the code. That would, of course, be taken to be
understood that that is a modification we have made to
the code and it is not the frozen code, that the
results that we get are looking at an effect of change
within the code and would not be taken that this is
UMAR01 or whatever the latest frozen version of the
code is. So that would have to be that understanding,
that these are modifications we made to try to
understand what the code does.
As part of that effort, we will be looking
at some of those models and some of the equations
within those models and spot checking. Are they
programmed to represent the model equation or the
correlation that is in the documentation? So we will
be doing independent evaluation with regard to some
spot checking also.
So I hope that those comments respond to
some of the five points that you had in your letter.
I lumped a couple of those together in these
responses, but we are taking those points that you
made quite seriously and we're trying to respond to
them and we're trying to factor those into our review
on the realistic large break LOCA.
Some of our conclusions. We do feel that
the S-RELAP 5 submittal is probably the most complete
we've seen to date.
CHAIRMAN WALLIS: Of any code?
MR. LANDRY: Of any code. At least in the
last several years, this is the most complete of any
code we've had submitted. The documentation is
massive. All the pertinent documents for S-RELAP 5
have been submitted and for models that are carried in
as outside models into the code.
The effort of the review is focusing on a
number of areas that have not been previously reviewed
in the code. There are some things that we will be
going back to look at, but I've outlined some of the
specifics that we are looking at in this review.
Again, due to our staff resources, we can't review
everything in the code. We have to pick out what has
not been reviewed and items that we feel are very
important and focus very heavily on those. The review
will include independent evaluation of the models and
the code and we do, as has been said a number of times
the last few days, have a very aggressive schedule for
this review. We are doing our best to adhere to that
schedule. We're pressing the staff and the consultant
to adhere to that schedule to get information and
questions very quickly because we want to proceed with
MR. KRESS: Does your exercise of the code
include the ability to do the non-parametric
MR. LANDRY: We have not planned to do the
statistical analysis with the code. We are, again
because of staff limitations, there's only a certain
amount we can do and my feeling is that it would be
more beneficial and better use of the resources if I
focus them on areas of the code that should be
evaluated internally and independently.
MR. KRESS: You will evaluate the ranges--
MR. LANDRY: That is part of the questions
that we talked about yesterday and part of other
questions we'll be asking to ensure that the
correlations and models are being used within the
range of validity and applicability. And Yuri brought
up some of that this morning, too, in his discussion,
concern that you're using data properly and
If there are no other questions, those are
the conclusions and where the staff review is on the
code. I've taken a number of notes the past two days,
a number of items that we're going to look at. One of
the points that we're going to discuss internally is
what can we do in the way of looking for compensating
errors or effects.
One of the examples that always comes to
mind on that is talking about the LOFT test L-22.
When we ran the L-22 test, as one that was managing
the project then, I was sitting there in the control
room when they ran the test and we had a beautiful
plot that was done with RELAP 406, I believe it was,
up on the screen and here comes this plot and here
goes the temperature screaming up, almost overlaid the
CHAIRMAN WALLIS: -- heat up or something.
MR. LANDRY: The blowdown part overlaid it
beautifully and the curve drops down and goes on up to
reflood and here comes the temperature screaming down.
We thought, holy cow, what did we do wrong? And then
the temperature goes back up. So afterwards we spent,
after we got some rest because we worked about three
days without sleep before we did that test, after we
got a little rest, we started looking at the data
trying to figure out what in the world happened and
found out that we had indeed had a core quench. Right
away at that point, the two people that had the codes,
INEL with RELAP and Lionel Los Alamos with TRAC,
started trying to figure out how they're going to
predict this. Lionel came back with a calculation
that overlaid the quench perfectly. They used a new
correlation. We sat back and looked at that and
started looking at the correlation and looking at it.
We said, you got the temperature right but you don't
have a single other parameter right. Every single
hydraulic parameter is wrong. So you got the right
temperature for the wrong reason.
Prime example of a compensating error and
a big concern that we do have is we have to make sure
to find some way to determine are the values we're
seeing right for the right reason or are there
We want to look at break flow split a
little bit on these double-ended guillotines. How is
the break flow being split between the two ends of the
pipe? The comment that Professor Schrock was bringing
up about the split break is very valid. There are no
data for a split break. There are data for T-flows
and, indeed, the split break starts as a T. But it
transitions from a T to flow from two directions into
what should be a T, and there are no data that I'm
aware of to do that. So we have to give some more
thought to that.
Critical flow calculation and model. One
of the things that we specifically tasked our
contractor to look at is the critical flow model, the
break flow models. So we'll be looking at that
CHAIRMAN WALLIS: I think we'd like an
explanation of how you go from two fluid to
homogeneous or whatever and how things are fitted
MR. LANDRY: That's one of the things that
I think we'll be asking Doctor Ward to look at
further. The regulatory implications of biases.
That's part of a discussion that we'll be having with
Doctor Orechwa and the views on the uncertainty
analysis, I believe.
We'll be looking further at things like
UPTF and some of the other tests and the applicability
or how you deal with some of the oscillations that
have been shown and discussed in this meeting. So
those are a number of the notes that I have for
further input we'll be looking for in the transcript
of the meeting and we'll also be waiting to hear from
the subcommittee and from the subcommittee's
consultant any additional comments, especially with
regards to things that you think as part of the
independent evaluation should be looked at within
particular models. We have a person dedicated to
doing this work and I would like to have the specifics
to keep her busy so that we look at the right
questions instead of come up in August and say, you
didn't look at the right questions. I'd like to use
the person correctly and look at some of those right
questions plus the other work she's doing.
CHAIRMAN WALLIS: We can not promise that
there won't be questions in August. My experience is
that no matter how much you go over this, there's
always the chance that something will come up.
MR. LANDRY: I realize that.
CHAIRMAN WALLIS: But we don't want to
surprise you in August.
MR. LANDRY: We'd like to give the best
shot we can.
CHAIRMAN WALLIS: We simply don't have
time. I've had that experience of reviewing
something. You think you've done it and then you find
out that it triggers something and you dig further
into something else and you find something you didn't
know you were going to find. Can't be sure.
MR. LANDRY: One final comment. A comment
with regards to statistics and liars and statisticians
is often times attributed to Mark Twain or Samuel
Clemens. Actually, it originates with Disraeli and
Mark Twain was quoting Disraeli.
CHAIRMAN WALLIS: I would think it's
probably been used before that. Almost as soon as
they were invented, someone probably saw the
MR. LANDRY: Julius Caesar may have talked
about the statistical evaluation of whether he should
be stabbed or not.
CHAIRMAN WALLIS: Well, they probably had
59 conspirators to make sure.
MR. NUTT: Then you have a 95 percent
probability he's dead.
CHAIRMAN WALLIS: Well, I think that we
don't have any ACRS letter on this and we're not to
that point. What will happen is that I anticipate
that I will make a very short presentation to the full
committee saying we've had this meeting and we've
heard various things and this is where we are in the
Do any of my colleagues wish to say
anything at this time before I wind up the meeting?
I'm looking forward to your comments by email so that
I can prepare for the full committee meeting.
MR. BOEHNERT: I think the only thing we
have to sort out is this business about what we want
to transmit or what we have to transmit to the staff
and Framatome and how to do it.
CHAIRMAN WALLIS: I'd like to know what we
transmitted last time.
MR. BOEHNERT: Yes, we got to dig that up.
CHAIRMAN WALLIS: Tom, do you have any
last minute remarks for us?
MR. KRESS: No. I think -- well, you
covered my items that I think he ought to think about
already that I had on my list. I might want to say
overall I was favorably impressed with the approach of
using PIRT and the uncertainty and sensitivity to
determine the 95/95 values for your acceptance
criteria. Looks like a good approach to me. There
may be some problems with some of the details but I
was impressed with it.
CHAIRMAN WALLIS: Virgil.
MR. SCHROCK: I think I was impressed in
many ways and disappointed in a few others. I think
I've commented on the places where I thought there was
a problem. I'll try to make those more pointed and
clear in my written comments. I do think there's a
problem with this so-called best estimate approach
being picked apart this way and to have continuum of
things ranging from Appendix K method to best estimate
method. Not very much clarity about what it is that
finally guarantees the thing is satisfactory.
I guess the decay power is the key one
that we talked about. Ralph's about not requiring
these notebooks, I guess you need to think about
whether the requirements that you have are going to
serve adequately in this environment. Seems to me
that the key point there was that if you're going to
claim that what you have is a conservative approach to
the thing, you have to demonstrate its conservatism.
I don't see in that case any other way of doing it
than doing a large number of best estimate
calculations which are exercising the full range of
technical information in the ANS standard that's now
CHAIRMAN WALLIS: Well, I was hoping to
stop at 11 but it's gone.
MR. BOEHNERT: Just barely.
CHAIRMAN WALLIS: And the reason is that
we started a minute late. I'd like to thank
Framatome. I think a lot of useful information was
transmitted in this meeting and it's given us a good
start in our review of your submittal or the role that
we play in the review process. I'd also like to thank
you for your general appearance of openness to discuss
questions, respond frankly. That's been very helpful.
We don't have any comment yet on the documentation I
think because we haven't really had time to study the
details. It's a lot of stuff, but we look forward to
So thank you very much. We'll see you
again some time in the future and hope this all works
out well for all of us. Thank you.
MR. HOLM: Thank you.
CHAIRMAN WALLIS: The meeting is
(The meeting was adjourned at 11:05 a.m.)