Thermal-Hydraulic Phenomena/Future Plant Designs - February 15, 2002
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
Combined Thermal-Hydraulic Phenomena/
Future Plant Design Subcommittee Meeting
OPEN SESSION
Docket Number: (not applicable)
Location: Rockville, Maryland
Date: Friday, February 15, 2002
Work Order No.: NRC-232 Pages 730-828
NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers
1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005
(202) 234-4433 UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
+ + + + +
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
(ACRS)
COMBINED THERMAL-HYDRAULIC PHENOMENA/
FUTURE PLANT DESIGN: SUBCOMMITTEE MEETING
+ + + + +
FRIDAY
FEBRUARY 15, 2002
+ + + + +
ROCKVILLE, MARYLAND
+ + + + +
The Subcommittee met at the Nuclear
Regulatory Commission, Two White Flint North, T2B3,
11545 Rockville Pike, at 8:30 a.m., Thomas S. Kress,
Chairman, presiding.
COMMITTEE MEMBERS:
THOMAS S. KRESS Chairman
DANA A. POWERS Member
VIRGIL SCHROCK Consultant
WILLIAM J. SHACK Member
JOHN D. SIEBER Member
GRAHAM B. WALLIS Member STAFF PRESENT:
Paul A. Boehnert
Medhat El-Zeftawy
ALSO PRESENT:
Walton L. Jensen
Jared Wermiel
Ed Throm
Mike Corletti
Ralph Landry
Edward Cummins
Bill Brown
Rick Austin
Jerry Wilson
Milorad Dzodzo
I-N-D-E-X
AGENDA ITEM PAGE
Reconvene, T. Kress 733
Staff Position on Codes/Testing (cont.)
Applicability of AP600 Codes
Reactor Codes, W. Jensen, NRR 734
Westinghouse Response, W. Representative 780
WGOTHIC Containment Code, E. Throm, NRR 784
Westinghouse Response, W Representative 805
Subcommittee Caucus (Open): 814
Comments on Meeting Presentations
Follow-on Actions
Decision to Bring Review to ACRS/
Instructions to Presenters
Adjournment 828
P R O C E E D I N G S
Time: 8:33 a.m.
CHAIRMAN KRESS: This is a continuation of
our Joint Subcommittees on Thermal-Hydraulic Phenomena
and Future Reactor Designs, dealing with the subject
of Phase 2 review of the AP1000 certification.
I guess we will ask if the members want to
have any preliminary thoughts before we start. If
not, Ill call on Jared Wermiel to introduce the set.
MR. WERMIEL: This is Jared Wermiel. I am
Chief of the Reactor Systems Branch.
This morning we are going to talk about
the review that was performed under the Phase 2 of
AP1000 of the those computer analysis codes for
accidents in transients and for containment analysis.
The first discussion will be with Walt
Jensen of my staff. In the Reactor Systems Branch we
had a team of my staff review NOTRUMP, LOFTRAN and
WCOBRA TRAC, those codes being used for analysis of
accidents in transients for AP1000, in order to
determine the applicability of these codes for AP1000
utilizing the information that we obtained during the
AP600 review and the new information that was provided
by Westinghouse as part of this Phase 2 review.
After Walt finishes, Ed Throm from the
Plant Systems Branch will talk about our review of
WGOTHIC, the code that is being used for analysis of
containment performance for AP1000.
So I'll let Walt go ahead.
MR. JENSEN: I am Walt Jensen, NRC staff,
and I am going to talk to you about our review of
LOFTRAN and NOTRUMP for AP1000.
(Slide change)
MR. JENSEN: First let me show you some --
the background slide. This details the differences
between AP1000 and AP600 for the passive safety
systems. We concentrated on the differences between
the two plants and how NOTRUMP and LOFTRAN would
handle the differences, because we just finished
reviewing both of these codes and approving them for
AP600.
So a lot of the review we didn't want to
repeat again, but as you can see, the makeup tank is
a little bit bigger than AP600, but they are the same
height but they are fatter. I believe that it was
Steve Bajorek in review of the scaling concluded that
the CMTs for AP1000 were still within the scale of the
CMT test data.
The accumulators are the same size. IRWST
is just about the same. They are going to pull a
little bit more water in, and the lines going to the
reactor system are larger. So they will have more
flow.
ADS-1, 2,3 are the same. ADS-4 is larger
with a lower line resistance and will remove steam and
water a lot faster than they did for AP600 -- or they
would for AP600.
(Slide change)
MR. JENSEN: The passive RHR is slightly
larger, the same design. The lines going to the
passive RHR are larger, larger lines with less flow
resistance. So more flow will flow through heat
exchangers, and more heat will be removed.
I have some containment data, and Ed Throm
will talk about the containment when I'm done.
CHAIRMAN KRESS: On your other slide, the
first one --
(Slide change)
MR. JENSEN: This one? All right. Okay.
CHAIRMAN KRESS: Down at the bottom it
says the ADS-4 venting is designed to allow for stable
IRWST/sump injection during long term cooling. What
does that mean?
MR. JENSEN: I guess I can't elaborate on
it.
CHAIRMAN KRESS: Does that mean it has a
certain capacity to the pump and the sump, and you
want to match the drainage out this way, out this
thing to it?
MR. JENSEN: This would be for the --
There are check valves that open to the reactor
system, once the blowdown was completely finished, and
allow the sump water to drain into the reactor system,
and the ADS-4 then depressurizing the reactor system
completely. Then this is part of the long term
cooling process.
This is not modeled in NOTRUMP, but it is
modeled in WCOBRA TRAC which is picked up after the
NOTRUMP analysis is completed. The NOTRUMP, work
stops when the IRWST begins to inject.
CHAIRMAN KRESS: One reason I asked the
question is I thought that valve was designed so that
you can depressurize at the right rate to get the
injections from the other systems during the
depressurization, and that this might also be a
consideration.
MR. BROWN: I think it's more -- You know,
think of it in terms of -- Bill Brown from
Westinghouse -- that it probably -- When I hear this,
it makes it sound like this really is maintaining the
depressurization so that, as you would send the steam
out, that essentially you don't repressurize; because
if you didn't, then you would have this continual
cycling of having to get some injection. You would
have to wait for the depressurization to occur through
the valve again, if it was undersize, for example, and
then have to wait for that whole thing to clear out,
and then you get some more injection again, where by
the sizing of it and design, you can keep the pressure
-- once you depressurize it, keep it down rather than
repressurizing and going through a cyclic --
CHAIRMAN KRESS: That fixes the minimum
area then?
MR. BROWN: Pardon me?
CHAIRMAN KRESS: That fixes the minimum
area.
MR. BROWN: Yes, what you need to do that,
yes. Yes.
CHAIRMAN KRESS: Thank you.
(Slide change)
MR. JENSEN: LOFTRAN -- I'll talk about
LOFTRAN first. It's probably the easiest. This is an
old code that Westinghouse has been using for years to
calculate transients, Chapter 15 transients in the
reactor system, including steam generator tube rupture
and steam line break.
It is used with other codes to evaluate
the maximum reactor system pressure, fuel temperature
and DNBR that would be experienced for various Chapter
15 transients.
It was approved first after a lengthy
review in the mid-eighties and then again for AP600 in
1998. It models the entire reactor system, but it
doesn't handle two phase conditions very well except
in the pressurizer. Pretty much if two phase occurs,
it is assumed to be homogeneous. So one should avoid
conditions for which two phase would occur with a
LOFTRAN analysis.
(Slide change)
MR. JENSEN: For AP1000, we looked at some
of the accidents that might occur. ADS-1, 2 and 3
could be analyzed with the code in a manner similar to
a stuck open relief valve, but the analysis stops
before any two phase conditions occur, and it's for
DNBR only. The inadvertent ADS-4 opening would be
done by a LOCA code.
The PRHR is larger and has a higher heat
flow -- higher heat flux. The data -- The
correlations in the code have been benchmarked and fit
to the actual test data, and the data includes --
Excuse me.
CHAIRMAN KRESS: Did they fit that test
data by putting a penalty on the area?
MR. JENSEN: This is the -- They use a
Rohsenow correlation, I believe, and they fit the
exponents in the equation just so they would --
CHAIRMAN KRESS: Oh, they adjusted the
correlations.
MR. JENSEN: They adjusted the
correlations so it would fit the test data. The test
data then includes the conditions that would be
encountered in AP1000.
CMT draining -- Let's see. Oh, yes, the
split cold legs -- LOFTRAN doesn't directly model the
split cold legs for AP1000, but an external model is
used to calculate the flow rates for asymmetric cold
leg conditions such as a stopped reactor coolant pump,
a locked rotor or sheared shaft.
The flows are calculated outside the code
and input into LOFTRAN. The models were reviewed for
AP600, and we think the same models should be
appropriate for AP1000.
MEMBER WALLIS: Well, if LOFTRAN is no
good for two phase flow, why is it used for split cold
leg?
MR. JENSEN: It can -- As long as the flow
is single phase, it can be used for --
MEMBER WALLIS: But it's not, is it?
MR. JENSEN: Pump -- reactor coolant pump
trip, and they would evaluate the DNBR, and this would
be done before two phase conditions occur.
MEMBER WALLIS: Before two phase
conditions?
MR. CORLETTI: Walt, this is Mike Corletti
at Westinghouse. Could I just clarify?
I think the split cold leg that you are
talking about there is not a LOCA. I think you are
talking about the fact that we modeled two cold legs.
MR. JENSEN: Ah, yes. Then again, CMT
draining is not expected. So they don't worry about
the void --
MEMBER WALLIS: You mean it treats both
cold legs as the same?
MR. CORLETTI: Yes, Dr. Wallis.
MR. JENSEN: Treats them as the same.
MEMBER WALLIS: By split, you mean they
are different. That split means that there's a
different flow in each one. Is that what you mean?
MR. JENSEN: Yes, sir. That's what I
meant to say.
CMT draining is not expected, and no void
formation is expected to occur in the pressure balance
lines. Westinghouse does have a penalty factor here
to use in case CMT -- in case the pressure balance
lines do become voided, but they don't think this will
occur.
The steam generators are larger than
AP600, and we were concerned that steam formation
might occur in the reactor coolant loops in the steam
generator tubes perhaps, in the top of the U-bend or
in the CMT pressure balance line in a place that
hadn't been reviewed.
So Westinghouse did not submit a steam
line break and proposed to do that in Phase 3. So
this is open right now, though we have done a
preliminary steam line break which shows that no void
formation occurs in the system besides the upper head.
So this should be resolved fairly easily in Phase 3.
(Slide change)
MEMBER SCHROCK: Did you say 2 or 3?
MR. JENSEN: In Phase 3 Westinghouse will
submit the steam line break.
CHAIRMAN KRESS: Now just for my
edification, this code, LOFTRAN, looking at it, and
it's used for a certain portion of the accident
analyses, and you have these considerations that you
listed on the previous slide which were things you
thought might be things to be concerned about and to
look at.
Now you are jumping to a slide that says
LOFTRAN is acceptable. My question for my edification
is: What went on in between these concerns and
reaching this conclusion? Did you take the code
calculations that Westinghouse made and looked at them
and somehow made judgments about those concerns or did
you -- What is the review process that gets you to
this point?
MR. JENSEN: Okay, good question.
Westinghouse submitted their input manual and their
user guidelines to running the code for AP1000. We
reviewed those. The code has been benchmarked against
SPES data and CMT data and PRHR data.
The scaling of the data was reviewed and
found to be appropriate to use for AP1000. We did a
RELAP analysis, a steam line break analysis which
showed very little cooling.
RELAP entrained a lot of liquid in the
steam generator as we blew the steam generator down,
and carried a lot of liquid away, where LOFTRAN
conservatively assumes only steam that's relieved in
a steam line break. The water stays in the steam
generator and is available to remove more steam to
boil and remove more heat from the reactor.
So this is a conservatism, and then, of
course, it had just been reviewed for AP600 and we
rely on that review for things that are similar
between the two plants.
CHAIRMAN KRESS: I thought you said a few
moments ago that the code should not be used for a two
phase problem. But now you've just described a
problem with two phase in the steam generator.
MR. JENSEN; Yes. I should have added to
the components that the code has been designed to
model in two phase. I should have added the steam
generator secondary side, and the code does handle two
phase in steam generator secondary and in the
pressurizer, and everywhere else it assumes
homogeneous. So I need to thank you.
Well, that's all I have for the LOFTRAN:
CHAIRMAN KRESS: We didn't dwell on that
open issue. Would you? We would like to know.
MR. JENSEN; Yes, sure can. The open
issue -- Well, I mentioned the steam line break that
Westinghouse -- for phase 3.
CHAIRMAN KRESS: Now that's the open issue
you just talked about.
MR. JENSEN: But preliminary analysis
shows they don't get boiling except in the upper head
and the steam generator and the pressurizer, and they
calculate a return to power. But we will be looking
at that in detail in phase 3.
MEMBER POWERS: You indicated that this
particular code had been benchmarked against data from
the SPES facility, I believe.
MR. JENSEN: Yes.
MEMBER POWERS: If I developed a
substantial masochistic streak and wanted to look at
that comparison against the test data, where would I
go to find it?
MR. JENSEN: It is in the topical that was
submitted for AP600, the final --
MEMBER POWERS: Maybe you could just give
me that reference before you leave today.
MR. JENSEN: Okay, I sure will. You're
going to have to remind me so I don't forget.
MEMBER POWERS: I think I have most of
that someplace.
MR. BOEHNERT: Well, we've got it here,
too, I think, if you don't.
MR. JENSEN; I expect you do have it.
MEMBER POWERS: Well, I am reminded of a
presentation once when a fellow said I benchmarked my
code against some data. We asked him to see that, and
the data and the predictions were at right angles.
Whereas, he had indeed done what he had said, it's
just that the comparison was very poor.
MR. JENSEN: Well, you have already seen
this, because you reviewed it for AP600. It's not
something you have to review again, I don't think. Do
you know?
MR. LANDRY: Dana, Ralph Landry from the
staff. That is all in the submittal that was on the
LOFTRAN for AP600, and it is in the -- can't remember
the WCAP number that it had, but the LOFTRAN submittal
had all the RAIs and all the responses to the RAIs.
What Walt is referring to was in RAIs and
responses, but that is in the two binders on LOFTRAN
for AP600. All those comparisons are shown.
MR. CORLETTI: This is Mike Corletti,
Westinghouse. I believe it was in the LOFTRAN
validation report. I have two WCAPs here. I'm not
precise on which one it is for sure, but it's either
WCAP 14234 or WCAP 14307.
MR. LANDRY: I think it was 434, Mike.
MR. BOEHNERT: 234 or 434?
MR. LANDRY: 234.
MEMBER POWERS: Maybe if somebody just
scribbled this down, I would have some hope.
MR. JENSEN: Thank you, Mike. Anything
else on LOFTRAN?
CHAIRMAN KRESS: So, basically, your
conclusion is, its use for AP1000 is okay with the
possible exception of this one open issue that you're
going to look into further?
MR. JENSEN: Right. That's right.
CHAIRMAN KRESS: Let me understand that
open issue and be sure I understand it. LOFTRAN
predicts not much voiding, and you got a lot of water
left in there to keep the core cool.
MR. JENSEN: No, no. LOFTRAN is single
phase. What we are worried about with the things we
are looking at with LOFTRAN is DNBR before the rods go
in, and maximum pressure for the surges up in the
pressurizer. Maybe it's used to input the code to
calculate the seal temperature, but there's no core --
that would be calculated with LOFTRAN. That would be
with the LOCA code. It's transients, DNBR.
CHAIRMAN KRESS: Okay, but what is it you
are worried about with the open issue again?
MR. JENSEN: Yeah, okay. Well, if there
are voids in the system that are calculated in the
system, LOFTRAN doesn't separate the voids. It's all
homogeneous.
CHAIRMAN KRESS: And you did RELAP
calculations that said there might be voids there?
MR. JENSEN: No, I didn't. RELAP didn't
calculate any voids.
CHAIRMAN KRESS: Just worried that there
might be voids there?
MR. JENSEN: There were no voids in RELAP,
but we were worried that the code would not be capable
of analyzing the physics of what's going on in the
system if voids occurred, except in a limited number
of places.
CHAIRMAN KRESS: Okay, I think I
understand.
MEMBER SCHROCK: This evidently didn't
include a systematic review of the parameter range for
the applications of the correlations in the code, or
did it? Two are mentioned.
MR. JENSEN; Not systematic, no. No, we
didn't do that. We did look at the PRHR heat
exchanger correlations and assured ourselves that the
data they used was within range. The code was
reviewed in a lot more detail for AP600, and I suspect
this was done. I don't know.
MEMBER SCHROCK: Thank you.
(Slide change)
MR. JENSEN: NOTRUMP: Now NOTRUMP -- It's
a small break LOCA code, and it's used to calculate
the inventory in the core and whether the core becomes
uncovered or not. And if it does become uncovered,
Westinghouse's model, small break model, includes the
SB LOCTA code that's used to calculate peak cladding
temperature.
Westinghouse is hopeful that there will be
no core uncovery calculated by NOTRUMP, and so they
won't be in need for a core heatup calculation. But
again, this was -- The NOTRUMP code was -- It's an old
code, and it was approved by the staff for operating
plants after a long and rigorous review in 1985, and
then again for AP600.
It uses five conservation equations with
slip models to evaluate the steam and the water
velocities. It does not -- The present model doesn't
include momentum flux for area and density changes.
Westinghouse did a study --
MEMBER WALLIS: Those are the ones that
are usually controversial. So they just leave them
out?
MR. JENSEN: Well, yes. It also made the
code -- Apparently, it made the code run unstably.
MEMBER WALLIS; Easier to run. It makes
it easier to run. No, really. You can do that. You
can make the code unstable with momentum flux term.
MR. JENSEN: Yes. Now they did a
sensitivity study for AP600 and concluded that the
momentum flux had very low effect except for ADS-4,
because of the high velocity effect.
CHAIRMAN KRESS: And they did something to
compensate for not having those. Is that where they
artificially reduced the water level?
MR. JENSEN: Yes, sir, they did that, and
they also talked about adding a resistance, extra
resistance in the IRWST line.
CHAIRMAN KRESS: To account for the what
would have happened if you had momentum.
MR. JENSEN: Right, and they propose to do
a similar thing with the resistance for AP1000.
CHAIRMAN KRESS: But not do the penalty on
the head of the water level in the IRWST? To do the
resistance instead?
MR. JENSEN: That is true, to do the
resistance and not the water level penalty.
CHAIRMAN KRESS: And that resistance would
be determined by comparing with SPES data?
MR. JENSEN: This was determined for OSU,
I believe, in this case.
CHAIRMAN KRESS: OSU data? Okay, OSU
data.
MR. JENSEN: And then I think they added
some additional resistance for conservatism. Now
would that same thing be appropriate for AP1000?
Well, that's kind of tied up in the overall --
CHAIRMAN KRESS: That's the question,
isn't it?
MR. JENSEN: Yes, that's the question. We
have a number of issues involving ADS-4 that need to
be worked out.
(Slide change)
MR. JENSEN: Well, we made a RELAP model
for AP1000, and we tried to make it as close as we
could, to use the same assumptions that NOTRUMP was
using, so we could compare the two codes, one against
the other.
There's a single failure. One of the ADS-
1 lines had assumed ANS plus 20 percent, and the
containment back pressure is set to atmospheric, just
like in NOTRUMP.
The first thing we did, we got one of the
old AP600 runs, and yes, it does have a lot of hash in
there.
CHAIRMAN KRESS: Is that numerical or is
it real?
MR. JENSEN: Numerical, mostly on the back
part.
MEMBER WALLIS: We don't have any real.
MR. JENSEN: Some of it may be real, but
I think this is numerical, and particularly it's
caused by the steam properties at low pressure, and
they were doing -- The code had trouble converging.
This is an older version of the code, too.
It's 2.3.3 gamma, and they have recently released a
3.3 beta to the count numbers, and that's what we used
for AP1000, but this is an old run for the older
plant.
What you see is that --
CHAIRMAN KRESS: This is like inducing a
LOCA by opening up ADS-1?
MR. JENSEN: Yes, that's the way it is.
ADS-1 sticks open.
CHAIRMAN KRESS: Well, then does the
depressurization continue through ADS-2, 3, and 4?
MR. JENSEN: Yes, sir. As I said, this is
predicated on the draining of the CMT.
CHAIRMAN KRESS: Draining of the CMT.
Okay.
MR. JENSEN: And then, of course, the
accumulators inject. CMT injects, and things are
pretty stable until ADS-4 opens, and that causes the
void fraction in the core to soar. That's this big
soaring in the void fraction.
CHAIRMAN KRESS: I see.
MR. JENSEN: And then the peak is turned
around the IRWST injection, which continues to inject
and then drops the void fraction to a fairly low level
which, unfortunately, in this version of the code kind
of jumps around.
MEMBER SCHROCK: This code to code
variation might give you a warm feeling, but don't you
need some demonstration through data as well? Are
there some data for this type of accident?
MR. JENSEN: Well, yes, AP600. Of course,
there was the staff test and OSU and ROSA test, and
RELAP was benchmarked against numerous of those tests
back in the AP600 review.
MEMBER SCHROCK: Well, benchmarked could
mean a lot of different things, I suppose, in terms of
the data compared with the prediction. But have you
looked again at how well RELAP did for this particular
transient in order to satisfy yourself that using it
as a comparison now against NOTRUMP provides useful
insight?
MR. JENSEN: Yes, I think it did a pretty
good job on stuck open ADS-1. Well, maybe -- I don't
remember whether they did a 1 or what, but they did a
stuck open ADS, I'm pretty sure, out of the
pressurizer. I think it did a pretty good job.
It had some trouble on some of the OSU
beyond design basis runs, and it had problems with
ADS-4 entrainment. One of the reports, I believe,
thought that there was too much entrainment in ADS-4,
and another report thought there was too little
entrainment. But, yes, it has the problem.
So we are not going to review AP1000 based
on what RELAP says. It helps us to understand what's
going on, so we get an idea about the conditions in
the plant and helps us to know what questions to ask
to Westinghouse. But I wouldn't put a great deal of
confidence in what RELAP predicts, other than it did
do a pretty good job on a lot of these test results.
CHAIRMAN KRESS: Do you have a similar
curve for -- calculated by NOTRUMP, just to see what
they look like?
MR. JENSEN: No, not -- Well, what I
wanted to show you now was for the AP1000.
(Slide change)
MR. JENSEN: As you see, the hash has gone
away. I do have some comparisons with NOTRUMP I'm
going to show you in a minute.
MEMBER WALLIS: You are going to show
comparisons with the sort of map that Dr. deMarzo
showed us yesterday of the actual level of the liquid
in the core at the critical time?
MR. JENSEN: What I was looking at here
mostly was the void fraction. But, yes --
MEMBER WALLIS: Inventory versus time or
something like that. She showed us a plot of total
amount of liquid versus time, and depending on what
you assumed about entrainment, you got different
answers.
MR. JENSEN: Right. Okay. Well, of
course, the minimum liquid in the core, if you looked
at the collapsed core level, would occur when the void
fraction is the highest. That's right before the
IRWST injects.
CHAIRMAN KRESS: What exactly do you mean
by void fraction here? This is in the core?
MR. JENSEN: This is the top node of the
core.
CHAIRMAN KRESS: So the top node of the
core. Okay.
MR. JENSEN: Now RELAP has nine nodes, I
believe, in the core.
CHAIRMAN KRESS: But it's the top node you
are dealing with.
MR. JENSEN: Right.
MEMBER WALLIS: So that would mean here
that the top node void fraction is almost one in this.
MR. JENSEN: That is true.
MEMBER WALLIS: The question I would ask
is what's it in the node below that, and so on?
MR. JENSEN: Right. I didn't bring it.
It's lower in the node below it.
MEMBER WALLIS: It is.
MR. JENSEN: It finally gets down to -- IN
the bottom of the core, it gets down to a void
fraction of around 50 percent in this tail that comes
out.
MEMBER WALLIS: Well, the collapsed level
is going to be less than half of this that you've got
here, if the minimum void is 50 percent.
MR. JENSEN: The collapsed level is
probably around 30 percent. I did look at some of the
collapsed levels, and it's around 30 -- 33 percent, I
think. So it's pretty low, but the pressure is low,
and the steam takes up a lot of room.
So if you just looked at the quality, you
looked at the mass ratio, and it would be -- mass
ratio would be -- It would look a lot better. It
would be like 30 percent or 50 percent.
CHAIRMAN KRESS: But this is telling me
that with AP1000 for the same accident sequence for
AP600 that your top node of the core is pretty well
uncovered or pretty high void in it?
MR. JENSEN: It has a lot of voiding.
CHAIRMAN KRESS: It stays there quite a
while?
MR. JENSEN: Yes, sir. That's exactly
what I wanted to show you, that there's more voiding
than AP600, and the voids are higher, and they stay
there high a long time.
CHAIRMAN KRESS: For this particular
sequence?
MR. JENSEN: Right, and other small break
LOCAs look very much like this, because actually, once
the CMTs start to drain and ADS opens, they all look
pretty much the same. Then they are all really
controlled by the ADS, especially ADS-4, because it's
just so big compared to these little breaks. But they
all look pretty much the same.
MEMBER SCHROCK: You have containment
pressure as a parameter on the slide. Does
containment pressure play any role here?
MR. JENSEN: Yes, it would.
MEMBER SCHROCK: Isn't it critical flow
through the break throughout this time period?
MR. JENSEN: No, it goes down to
subcritical flow.
MEMBER SCHROCK; In 4000 seconds?
MR. JENSEN: Yes. It's about the time
that the IRWST begins to inject that the flow drops to
subcritical. If the pressure is higher in the
containment -- Westinghouse in fact did some
sensitivity studies, and they got substantially less
voiding when they used a higher containment pressure.
Of course, they used a constant
containment pressure, I believe, and if one was going
to take credit for containment pressure being higher,
one would have to calculate it. It would probably
still be pretty low until ADS-4 fired off. ADS-1, 2
and 3 go down into the IRWST and get quenched, and
only when ADS-4 goes off then is there a large flow
into the containment.
So it would be somewhat more work and
cause probably some iterations between codes to take
credit for the containment pressure.
CHAIRMAN KRESS: So what is it about the
AP1000 that makes this difference between it and 600?
Is it because your flow out to ADS-4 is greater, and
it can't be made up as fast by the passive systems
that are feeding water into it?
MR. JENSEN: That's part of it, and
probably a lot that -- mostly, it's the power. The
power is bigger. The power density is higher.
CHAIRMAN KRESS: So it's pouring out more
steam.
MR. JENSEN: More steam. More steam is
coming out. The accumulator is the same size. CMT is
just a little bit bigger, and so more reliance is
being placed on the ADS-4 to depressurize the plant
and get injection from the IRWST.
MR. CORLETTI: This is Mike Corletti from
Westinghouse, Walt. Could I just --
MR. JENSEN: Oh, please do.
MR. CORLETTI: The main focus of the phase
2 was really looking at the code applicability and the
performance of the code and how -- are the phenomena
similar between the two plants.
I think what we are going to find -- What
we did is a preliminary set of analyses with bounding
power shapes. I think what we probably find -- I
don't think Walt really focused on is it safe -- the
safety of the plant was not -- They weren't really
making judgments on that.
CHAIRMAN KRESS: That's not the issue
here. Just looking to the applicability.
MR. CORLETTI; Right. And I think what
we'll see for our safety analysis that we present in
the design control document you are going to see -----
With the AP1000 specific power shapes and some of the
other specifics that we've put in, you are going to
see voids probably not quite this high as presented
here. But that will be all reviewed as part of the
phase 3 design certification.
CHAIRMAN KRESS: Okay. That's a good
clarification.
(Slide change)
MR. JENSEN: Now here I do have a
comparison between NOTRUMP and RELAP, and this is for
a two inch cold leg break. This is actually the only
--
MR. BOEHNERT: Could you take that other
slide off, Walt? It's hard to see.
MR. JENSEN: Oh, I'm sorry. Sure. And
should I turn the projector off?
MR. BOEHNERT: Yes. Thank you.
MR. JENSEN: All right.
CHAIRMAN KRESS: Now this would be almost
like that ADS-1 case you just showed?
MR. JENSEN: They look very similar.
CHAIRMAN KRESS: Very similar
MR. JENSEN: This break size is the only
break that we really can compare to, this two-inch
break, with NOTRUMP. There were a number of analyses
done by Westinghouse about a year ago, and they were
with some old assumptions for AP1000. So only
recently have they redone this one. So we are
comparing the RELAP analysis.
They thought this would be the best to
compare with RELAP. But this is the only one we have
right now for comparison. So what we'll get for the
other breaks remains to be seen.
So as you look down this table, you see
there are some differences. The first thing you see
is the reactor trip. We tripped ours on a
overpressure delta T at 9.4 seconds, and Westinghouse
has tripped a lot later.
They wait until they get a lower pressure
signal in the pressurizer before they trip, and that
occurred a little bit later. I guess they don't take
credit for the overpressure delta T.
The safeguard signal came in about the
same time. This would be on low pressurizer pressure.
Reactor coolant pumps tripped on the safeguard signal
about the same time, but we are using a 15 second time
delay that was in AP600, because that then is just a
modification of the old AP600 that we used for that
plant. It's probably about the biggest RELAP deck
ever made.
So it looks like we got some different
delay times that we are using between us and
Westinghouse. So we're going to need to work out some
of these differences.
MEMBER SCHROCK: This table is AP1000?
MR. JENSEN: This is all AP1000. I'm
sorry, sir. Yes.
MEMBER SCHROCK: Do you have a
complementary table for AP600?
MR. JENSEN: No, sir, I don't. I didn't
want to spend much time on AP600, because we had
already done that one.
The CMTs begin to drain a little sooner in
this NOTRUMP analysis. Accumulator injects sooner,
and then look at ADS-1. It's really a lot faster than
RELAP.
ADS is based on the core make-up tank
volume. So apparently we are losing more core make-up
tank volume or CMT volume, losing it faster in NOTRUMP
than in RELAP.
Then we got the high void fraction, 90
percent. It even, in fact, goes up to 95 percent at
3660 seconds when the IRWST injects.
So everything -- We've compared a number
of the parameters, the ADS-1, 2, 3, and 4 flow, CMT
injection, and they all looked very similar except for
a shift, because NOTRUMP is doing everything faster.
So why is NOTRUMP boiling down faster than RELAP?
MEMBER SCHROCK: Was that supposed to be
best estimate? Isn't NOTRUMP the conservative code?
MR.JENSEN: That is true. That is true.
But part of it is the way the plant is described, and
it looked like to us it was the break flow model.
This is what RELAP calculates for the
break flow. This decompression difference occurred
before any of the ADS started to inject. The only
thing being lost when the system was compressed was
coming out of the break.
So what you see here looks pretty much a
conventional plant for a small break LOCA. There is
a subcooled blowdown.
CHAIRMAN KRESS: Now this is the NOTRUMP
here?
MR. JENSEN: Excuse me. This is RELAP.
CHAIRMAN KRESS: This is RELAP?
MR. JENSEN: This is RELAP. This is where
RELAP says the break flow is. I'm going to show you
the comparison in a minute, and I wished I had -- I
should have written RELAP on here.
So we have a subcooled blowdown where
RELAP is using the Henry Fauske correlation, and then
we have a saturated blowdown, which is this flat part,
and then a two phase blowdown where it comes down.
ADS is open here, and sludge, water and steam are
coming out of the break at this time.
So this is what RELAP gets. Now what does
NOTRUMP do?
(Slide change)
MR. JENSEN: I got a lot of things on
here. The solid line is AP600. It just happened to
be on this slide. I wish it wasn't. This is a
Westinghouse figure, but the dashed line is NOTRUMP
for AP1000. So that's what we want to look at, this
dashed line for AP1000. You see it?
They, too, got the subcooled blowdown.
They used the Zaloudek correlation, which gives a --
It is conservative, and it gives a higher flow than
RELAP does in the subcooled blowdown.
Then they have the flat saturated part.
But then they get the big spike of water coming out of
the break that RELAP doesn't calculate. All right.
Now what it looked like to us that caused this was the
downcomer description in the codes.
RELAP divides the downcomer into eight
radial nodes between the downcomer and the core
barrel. So the eight radial nodes of the CMT line are
in individual nodes. I had a figure that said
Westinghouse Proprietary. So I didn't bring it.
Each pole leg is a individual node. So
it's a highly segmented downcomer. So when the break
occurs in the cold leg, the flow reverses in that
section of the downcomer, but the CMT water is free to
continue the flow down to its segment into the lower
plenum and up into the core.
Now Westinghouse does not segment their
downcomer. So they have it in a single segment like
a pipe with -- I think they have maybe three axial
nodes.
So what happens in NOTRUMP when the CMTs
eject, the whole downcomer, being a one-dimensional
pipe, tends to flow the break and carries the cold CMT
water with it. When this cold subcooled water comes
to the break, the critical flow model that is their
Zaloudek correlation for a subcooled blowdown, just
blows out a lot of water out of the system.
So this is why they are losing water, we
believe, faster than we are. I suppose it's
conservative to do this, to lose water faster. This
is why we are different, I think.
MEMBER SCHROCK: This flow rate does not
include the ADS flows. It's the flow through the
break?
MR. JENSEN: This is the break flow.
MEMBER WALLIS: So we get a difference of
something like 200 pounds per second for about 200
seconds. That's 40,000 pounds of water. Now is that
important compared with the inventory of the system?
MR. JENSEN: I think it's a lot of the CMT
water.
MEMBER WALLIS: How much water is in the
CMTs?
CHAIRMAN KRESS: 2500.
MEMBER WALLIS: Twenty-five?
MR. CORLETTI: There's 5000 cubic feet of
core makeup tank.
MEMBER WALLIS: How many pounds is that?
MR. CORLETTI: Times 62.
MEMBER WALLIS: So it's hundreds of
thousands of pounds of water.
MR. JENSEN: So let me go back to my
comparison slide again.
(Slide change)
MR. JENSEN: So we got very similar
answers to NOTRUMP for this break. The void fractions
we got were slightly higher, but they were pretty high
in NOTRUMP, too.
We did some other breaks, too, and we did
a pressure balance line and, of course, we did ADS-1.
MEMBER WALLIS: I think what you are
saying is the details are very different, but the
overall picture, when you look at the whole thing, is
about the same. Now the timing of events is different
from RELAP and Westinghouse, and the amount of water
you get in and out at various times is quite
different, but when you look at the overall picture at
the end of things, result is about the same. Isn't
that what is happening?
MR. JENSEN: That is true, and the details
are similar except for the break flow. I wanted to
tell you, too, we did a double-ended DVI line break,
and we got a little bit of core uncovery with RELAP.
RELAP had a hot channel in it -- hot pin.
It didn't have a hot channel. It had a hot pin, and
it calculated a peak cladding temperature of about 400
degrees Fahrenheit. So it's not anything -- not a big
core uncovery, but it dipped down at the core a little
bit in RELAP, and we don't have an up to date DVI line
break from NOTRUMP. So we don't know what that is.
Then again, we're not going to approve the
plant or disapprove the plant based on RELAP, because
we are not sure we believe it.
MEMBER WALLIS: Well, I don't know. You
have to use some sort of judgment here. If RELAP is
predicting disaster, then you really want to know why
there is no disaster.
MR. JENSEN: That's true. It has been
benchmarked against the data, and we have some
confidence in it, but we think it may be a little weak
for the entrainment in the ADS-4 line and the hotleg.
That's the same trouble that we're having with
NOTRUMP, and we'll have to iron in phase 3.
MEMBER WALLIS; You didn't show any
comparisons between your predictions and, say,
something like the APEX facility or something that
would give some confidence that you were not too far
from reality with your predictions?
MR. JENSEN: Well, like with LOFTRAN, that
was all done in great detail for AP600, and I could
give you some references for that, too, if you would
like.
CHAIRMAN KRESS: Now I recall those
comparisons for AP600. They had these three
categories, good, better and best, or something like
that. It depends on where the calculations fit within
the uncertainty band on the data.
Best I remember, for most of the accident
tests run, they fell within the uncertainty bands of
the data, which gave me some confidence that they were
doing pretty good.
MR. JENSEN: It looked pretty good. It
did.
MEMBER WALLIS: But then if I look at what
you just showed us for the break, the AP1000/AP600
curves are very different. We don't have any modeling
of AP1000 by something like APEX. So we have to take
it on faith that --
MR. CORLETTI: Dr. Wallis, this is Mike
Corletti. We had hoped that our scaling was really
showing that APEX was scaled for AP1000. So that was
the basis for using the same validation.
MEMBER WALLIS: I don't quite know what
you mean by scaling here. They are certainly not
getting the same curves. Look at that curve you
showed us just now with the flow right out the break
versus time. It's certainly not the same scenario.
MR. JENSEN: This one?
MEMBER WALLIS: No, no, no, no, the one
with RELAP 5, AP600, AP1000, two-inch 49, whatever
that is.
MR. BOEHNERT: This one right here.
MR. JENSEN: The break flow.
MEMBER WALLIS: And we got three curves.
RELAP5 seems to be absolutely fascinated with 200 and
stays along there. Is that what the crosses mean?
MR. JENSEN: Yes, it does. It stays
there, because the -- Then the flow to the break is
saturated. It's saturated water in there. I mean, it
just kind of sits --
MEMBER WALLIS: If I look at these two
curves, I'd say, well, AP600 and AP1000 don't have the
same response, do they?
MR. JENSEN: No, they don't.
MEMBER WALLIS: Quite different. So
what's being scaled by something like APEX? Is the
APEX response like the 600 or like the 1000 or
something else entirely? I don't know.
MR. JENSEN: Of course, the difference in
the break is the same size for both AP600 and AP1000,
and with the break the same size --
MEMBER WALLIS: Ah. Not the same scaled
size. Is that what it is?
MR. JENSEN: No, the break is not scaled
in this. The break is a two-inch break.
MEMBER WALLIS: It's not a scaled break.
So you shouldn't compare based on the same size, but
we should perhaps compare some other break size which
is properly scaled between 600 and 1000? That's
probably the trouble.
MR. CORLETTI: Dr. Wallis, in our report
that we submitted that was precisely what we showed,
that it appeared that the larger break -- To get
equivalent performance, you needed a little bit larger
break for AP1000.
Really, what we saw was really a time
shift for breaks of the same size. So that seemed to
be consistent with -- That led us again to a judgment
that the plants really do operate from a scale basis
the same.
MR. CUMMINS: Dr. Wallis, Ed Cummins.
To me, those curves look alike, with the
AP1000's spike just delayed. If you just translate it
to the right, then they look very similar.
MEMBER WALLIS: I don't know quite what
you mean by similar.
MR. CUMMINS: On AP600 we also got a
spike. It was just earlier in time.
MEMBER WALLIS: And Everest looks like Mt.
Washington, if you take two foot out and put them
close together.
MR. CUMMINS: They look a lot closer to
RELAP.
MEMBER WALLIS: This is not very -- I
think the real thing is that, if you scaled the break,
you'll be able to show a much better comparison. That
would be more meaningful than maybe trying to wring
something out of this picture which is not very
meaningful.
CHAIRMAN KRESS: And when you scale the
break, the size that you scale it to is going to be
different, depending on which one of those periods you
are in, because one of them, you are going out
critical flow, and the other one going out subcooled,
and you will get a different scaling ratio for the
break size.
So you have to be careful with just saying
how you scale it. But I agree with you. If he scaled
it to the break size for the different periods on that
curve, why you will probably get similar.
MEMBER WALLIS: I think we concluded for
AP600 that you got all these bathtubs indirectly and,
depending on some rather small changes, it can make a
difference whether this bathtub goes in before that
one and all that. But at the end of the day, if you
look at whether or not you get uncovery that period,
it doesn't matter too much what you did before.
That's perhaps where the focus should be
in these studies, whether or not you get uncovery and
how sensitive it is to where all these different
things are happening before, which really don't really
make that much difference at the end state.
MR. BROWN: All right. Bill Brown. I
think what we'll find is that if you take all those
and get -- Once you get ADS, I think these will all
look the same. I think that you are going to see some
differences up front a little bit, but once you get
ADS --
MEMBER WALLIS: Once you get ADS-4.
MR. BROWN: Exactly. I think they
basically will look very, very similar and will
converge to that.
MEMBER WALLIS: Well, the idea of ADS --
MR. BROWN: Is to turn it all into -- a
little break into a bigger break and all look the
same.
MEMBER WALLIS: What happened before
doesn't matter.
MR. BROWN: Right. Right. All the
history is lost.
(Slide change)
MR. JENSEN: Now these are some of the
components we looked at in the AP1000 review. The
accumulators are the same size. ADS1, 2 and 3 are the
same size, and the CMT -- they are the same height and
fatter, and these were all compared to test data in
AP600. So --
CHAIRMAN KRESS: Is each CMT 23 percent
more volume or is that the total for the two of them?
MEMBER WALLIS: Probably both.
MR. JENSEN: II think each one would be 24
percent more volume. Now the PRHR heat exchanger is
22 percent larger, but because the inlet and exit
paths have a reduced resistance, it's designed to
remove 72 percent more heat.
Now NOTRUMP has some problems with high
heat flow in a PRHR heat exchanger. I told you how
for LOFTRAN that they had modified the exponents in
the Rohsenow correlation to fit the test data.
NOTRUMP uses the Tome correlation for boiling in the
IRWST, and Westinghouse didn't fit the X points in the
T correlation.
MEMBER WALLIS: Pretty nostalgic. This
has worked on in the late fifties.
MR. JENSEN: Probably so.
CHAIRMAN KRESS: It modified the exponent
or the coefficient?
MR. JENSEN: I don't remember what this --
Probably both of them, but they refit the curves. Now
so Tome is nonconservative in comparison to the PRHR
test data at high heat flows. So Westinghouse has
benchmarked against a 1.5 foot per second flow rate in
the tubes.
So then the question is what is the flow
rate in the PRHR heat exchanger tubes, and is it
greater or less than 1.5 feet per second? So this is
another RELAP calculation.
So anyway, the flow rate, we found, is
considerably higher than 1.5 feet per second. So what
did they do with NOTRUMP in this area where the code
is stated not to be completely out?
Well, PRHR heat exchanger heat flow is
just a medium in points, and it probably doesn't have
a great deal of significance for a small break LOCA,
and Westinghouse has even done a preliminary study
that they reduced the heat transfer area by 50
percent, and it made very little difference in the
course of the LOCA. But they are going to qualify
their assumptions during phase 3 for the PRHR heat
exchanger flow.
CHAIRMAN KRESS: Now this PRHR heat
exchanger is mostly there for the long term cooling?
MR. CORLETTI: No. A PRHR heat exchanger
is primarily there for transient. So it's there on a
LOCA event, but it really was not the sizing basis.
What we've seen in most of the LOCAs is, once you
depressurize, go two phase, it really does not become
a big factor in the transient behavior.
CHAIRMAN KRESS: This is one place where
you have a pump, and it's an active system?
MR. CORLETTI: No. This a passive RHR
heat exchanger. It sits in the refueling water
storage tank.
CHAIRMAN KRESS: It's that C-shaped thing?
MR. CORLETTI: Yes.
CHAIRMAN KRESS: It's the water flowing
to the inside of the tubes?
MR. CORLETTI: Right. It's by natural
circulation.
CHAIRMAN KRESS: By natural circulation.
MR. CORLETTI: Yes. And it's really a
sizing basis. It's for transients like loss of normal
feed, feedwater line break, those sort of events. But
it is modeled in NOTRUMP in the LOCA, and it doesn't
have much effect unless you have a very, very small
break.
CHAIRMAN KRESS: What's the heatsink on
that when you get outside the IRWST?
MR. CORLETTI: The refueling water storage
tank, and if it would -- for a long transient would
heat up and boil, steam would condense on the
containment shell, and we have it return back to the
refueling water storage tank. So it essentially can
stay as a heatsink --
MR. CUMMINS: Using containment then,
basically.
MEMBER WALLIS: I wonder if we could move
on to the bottom line?
MR. JENSEN: The bottom line? Sure, the
bottom line.
MEMBER WALLIS: This is a real bottom
line? This hasn't been manipulated by the accountants
in some way?
(Slide change)
MR. JENSEN: This is where we are right
now with phase 2. So we think NOTRUMP is pretty good
except for three areas, and maybe just two, and the
big one is the liquid entrainment from the core into
the upper plenum and out the hotleg and out ADS-4.
Westinghouse proposes in phase 3 to
benchmark NOTRUMP against the WCOBRA/TRAC.
MEMBER WALLIS: I don't see how that
works. I mean, you say one code against another. The
physics is wrong. How is that going to help you?
MR. JENSEN: Okay. Then they are going to
benchmark WCOBRA/TRAC against test data.
MEMBER WALLIS: Ah, they are going to do
that. That's going to be essential.
MR. JENSEN: Finding the right test data
may be where the difficulty lies.
MEMBER WALLIS: You're going to hold the
line there?
MR. CUMMINS: Yes.
MEMBER WALLIS: Not to say it remains an
issue, but supplemental verification will be performed
or something.
MR. WERMIEL: Ultimately, Dr. Wallis --
This is Jared Wermiel. Ultimately, we will need some
confirmation either through test data or some data
that we can all agree that it is valid that, indeed,
if they do do a sensitivity study on this phenomena,
that the sensitivity is telling us the right thing,
and ultimately it's an answer that we can rely on.
It will also depend in large measure on
some qualitative arguments about how significant this
entrainment -- You've heard argument yesterday. How
significant really is this issue? We are still not
sure, and that will be a source of continual
discussion during the phase 3 review.
MEMBER WALLIS: Yes. Thank you very much.
MR. JENSEN: Okay. then the second one is
the PRHR heat exchanger model, which is open, but we
don't think it's going to have much effect.
Then finally, we have looked at --
MEMBER WALLIS: I'm sorry. Proposes to
reduce the heat transfer area -- that's simply --
That's in the numerics. They are going to say, well,
if you don't like it, we'll just assume it's half as
big, and that's conservative. Is that -- It's not
actually physically reducing?
MR. JENSEN: Right. Hopefully.
MR. CUMMINS: That's right.
MR. JENSEN: And they will look at the
test data to see.
MEMBER WALLIS: Let's just cut it in half.
MR. JENSEN: Oh, I see what you mean.
Excuse me. Well, us analysts, we don't deal with
hardware very much. In fact, we sometimes forget
about that it's actually a plant that's going to look
like this. It's just a bunch of numbers.
All right. Then lastly, we've looked at
one break size, the two inch cold leg break, and there
was no core uncovery. But in the course of phase 3 we
are going to be looking at the entrainment out of ADS-
4, which will affect what is going on in the core, and
we'll be looking at a lot of other break sizes.
So it's very likely that core uncovery
will be calculated.
MEMBER WALLIS: So this statement, only a
limited number of breaks have been analyzed -- that
means by you or by Westinghouse?
MR. JENSEN: By Westinghouse.
MR. WERMIEL: By Westinghouse.
MEMBER WALLIS: Well, you said you are
going to be looking at it. Does that mean you are
going to be doing more analysis?
MR. JENSEN: We will do more analysis. We
have done -- looked at several break sizes already,
and we got a small amount of core uncovery for one.
Westinghouse is going to look at some
more, but I think the final runs are going to be --
have to be after the ADS entrainment issue is solved.
So will there be core uncovery? I don't know, but the
SB LOCTA code and the NOTRUMP code really haven't been
looked at for core uncovery in the passive plants.
There's a transition boiling model in
NOTRUMP that we didn't look at for AP600, because no
core uncovery was calculated, and the SB LOCTA code --
I'll look back to see when that was reviewed last. I
think we looked at it -- It was LOCTA4 back then, back
in about 1972, back when the original ECCS model came
through. So we will probably want to look at that
again if there is any core uncovery predicted.
Let's see. I have one more slide, if you
can bear with me.
(Slide change)
MR. JENSEN: This is what RELAP predicts
for the flow in the hotleg when the ADS-4 is
operating.
MEMBER WALLIS: Thirty meters a second?
MR. JENSEN: Thirty -- Yes, please, I left
off the -- The top curve should say steam flow, and
the bottom curve should say liquid flow.
MEMBER WALLIS: Ah. You liquid flow is
actually catching up with the steam.
MR. JENSEN: The liquid is the bottom one,
and it's just kind of bouncing around zero. It's
fairly low. So ADS-4 opens at about 2900 seconds, and
then the IRWST comes on at around 3500 seconds, which
is this big water slug coming out. But I guess what
I'm trying to show here is the steam velocity. It's
pretty high in this hotleg.
RELAP predicted the flow was annular. It
didn't think the flow -- based on its flow regime map,
didn't think it had stratified flow at all. It
thought it had annular flow. So it's carrying all
this then out the ADS. So this is --
MEMBER WALLIS: What's happening in the
vessel? Is the vessel level higher than the hotleg
here?
MR. JENSEN: The vessel two phase level is
up even with the hotleg, I think. It's carrying out
two phase.
MEMBER WALLIS: Even?
MR. JENSEN: It's up in there, up in that
region. Whatever is coming up is carrying it out.
MEMBER WALLIS: I'm just saying the flow
regime is going to depend very much on how the liquid
gets in there at the end. It comes in as droplets.
It's not going to be instant in the annular flow, for
example.
MR. JENSEN: Right. So this is the kind
of thing we can use RELAP for. Maybe we don't believe
the flow regime, but we can say, well, we know what --
it's telling us what the steam velocity is, and we
want to be sure that the data that we are using, when
we find some to benchmark WCOBRA/TRAC, is applicable
for velocities of this magnitude.
So we have a lot to do here.
MEMBER SCHROCK: These predictions were
done with RELAP?
MR. JENSEN: Yes. This is RELAP. That's
all I have to tell you about NOTRUMP. Entrainment is
going to be a big problem.
CHAIRMAN KRESS: Okay. With that, we are
at the point where we would like to hear what
Westinghouse has to say about that.
MR. CORLETTI: I guess we can just
summarize our plans going forward. We are doing our
plant calculations with NOTRUMP. We also are
preparing our topical report, which is our ADS --
using COBRA/TRAC, looking at the ADS-4 IRWST
transition phase.
We are validating that against the test
data at OSU that was already performed for AP600. We
are going to show that and also show plant
calculations with that as part of this topical report
we will be submitting with our application.
The purpose there we hope to show is that
with such a code that NOTRUMP is still performing a
conservative representation of the transient. That is
going to be part of our phase 3 review.
MR. CUMMINS: This is Ed Cummins. To
summarize our position, it's very similar to the
staff's. In LOFTRAN their comment was we were
sensitive to two phase events, if they are predicted.
Our current analysis doesn't predict two
phase events for the steam line break, but the staff
hasn't reviewed that, and it's intended to be reviewed
in phase 3.
In NOTRUMP, basically, we discussed
entrainment yesterday, and I think we wouldn't quite
agree with this is a big problem. We think in some
respects it is self-limiting, because H is cubed as
well as steam flow. But this is something that we
don't intend to resolve in phase 2. We intend to
resolve in phase 3.
We agree that, if NOTRUMP shows core
uncovery, that that invokes this LOCTA code and that
it was appropriate for the staff to review it.
Thirdly, we agree that the heat exchange
model in NOTRUMP is applicable for less than 1.5 feet
per second and, if it's greater, then it has to be
adjusted downward to be conservative, and we intend to
do that.
CHAIRMAN KRESS: Okay. At this point we
are going to turn to WGOTHIC, but I wonder if the
members would like a break first.
MEMBER WALLIS: I think, as we just have
one more topic, we might take a break now and then --
CHAIRMAN KRESS: That's what I thought we
might want to. So let's take a 15 minute break. Be
back at ten o'clock.
(Whereupon, the foregoing matter went off
the record at 9:47 a.m. and went back on the record at
10:03 a.m.)
CHAIRMAN KRESS: Okay, let's turn now to
the Gothic novel part of this thing. The word Gothic
invokes visions of Frankenstein. No comment about the
speaker, of course. Okay.
MR. THROM: Good morning. My name is
Edward Throm, and I am currently in the Plant Systems
Branch of NRR, and I'm in the new division called
Design Review Section.
I'm going to be going over the WGOTHIC
review for the AP100.
(Slide change)
MR. THROM: Since it always comes up, I
did look it up again, and GOTHIC stands for Generation
of Thermal-Hydraulic Information for Containment. So
it's a word that has a meaning, and WGOTHIC is an
extension of the numerical applications incorporating
GOTHIC4.0 code.
The extension is the inclusion of the
Kline model which is the Westinghouse model that
addresses the passive containment cooling system,
which is basically using condensation on the inside of
the containment and evaporation of the water flow on
the outside of the containment.
In the supplemental package I do have a
caricature of the system. Water is basically poured
onto a bucket on top of the containment. The bucket
is not really on the containment. It's elevated from
the containment, and then it is distributed through
two sets of weirs such that there is uniform flow of
the water down the shell of the containment.
Air through buoyancy driven forces comes
down the downcomer, up a riser section and out through
the chimney, and the water is evaporated and the heat
carried off through the chimney channel.
(Slide change)
MEMBER WALLIS: And the water evaporates
before it reaches the ground or the bottom?
MR. THROM: Yes. Part of the modeling
that is done in WGOTHIC is to make sure that the
amount of water that is being evaporated is only the
amount of water that can be evaporated.
What would normally happen is there is a
drain line at this elevation, and any excess water
would flow down that drain.
MEMBER SCHROCK: Is the aspect ratio
realistic in this picture?
MR. THROM: No. That is just a
caricature. I cut it out of another document.
MEMBER SCHROCK; It's much taller.
MR. THROM: Yes. There are -- I couldn't
find any real good pictures, but it was just a sense
to remind you again of what the passive containment
cooling features are.
WGOTHIC is described in Westinghouse
topical report, WCAP-14407, and the staff's safety
evaluation was presented in NUREG-1512. So those two
documents are the code.
Basically, the code approved in Part 52.
Approval of the methodology is done as part of the
design certification. So the approval of WGOTHIC for
the AP600 is part of that design certification.
It's also necessary to point out that in
developing and approving WGOTHIC we came up with what
we called an evaluation model, not necessarily at the
level of detail you see in an Appendix K model but,
nevertheless, we needed a term to identify what we
were talking about.
So the evaluation model conserved the
aspects of using lumped parameter networks for the
representation of the containment, issues concerning
circulation and stratification, and issues concerning
the use of the PCS flow and the mass and heat transfer
models that were developed to model and conserve the
passive containment cooling system.
(Slide change)
MR. THROM: As we are all aware, the large
scale test facility which Westinghouse used was not
well scaled for the blowndown portion of a transient.
So the problem we had was not having a good scaled
facility to assure ourselves that we really understood
circulation, stratification, and some of the other
issues that would be of importance if we were trying
to do a quality calculation against a good set of
data.
So during the development of the model, we
looked at the international code databases.
Specifically, if you want to read the information, you
can look at Section, WCAP-14407 or Section 21.65 of
the WCAP -- of the NUREG 1512.
In order to feel comfortable or justify
the use of the lumped parameter approach, we went out
into the international database, looking at the
Battelle Model Containment and the HDR.
The HDR -- there were a couple of
experiments that were done there where they sprayed
water on the outside of the containment shell. What
you see from the application of the GOTHIC code or
tools like GOTHIC to the international database is the
lumped parameter tends to homogenize the steam with
the noncondensables, basically the air in the
containment, and as a result, what's being condensed
near condensing surfaces usually has a lower steam
concentration that you might expect, and you generally
always overpredict pressure, and pressure is one of
the key markers for containment analysis.
When you apply regulatory requirements to
these calculations, you really come up with a very
conservative calculation for pressure.
This basically -- We confirm
Westinghouse's contention that the lumped parameter
model is appropriate for use.
(Slide change)
MR. THROM: The other issues that you
wound up with in not having a properly scaled facility
were circulation and stratification. In order to
address these concerns, Westinghouse presented, and we
ultimately accepted, a very conservative approach.
One approach is that after blowdown,
Westinghouse turns off the heat sinks below the
operating deck. So if they calculate steam going back
down below deck, they don't take credit for the
condensation.
In order to address the issues with
circulation and stratification, two things came up.
One is the developing of a condensing surface on a
floor. Westinghouse has elected not to take credit
for that.
The other big concern was potentially
getting an air blanket on the operating deck as part
of the stratification issue perhaps in the long term.
So Westinghouse also takes no credit for condensation
on the upper deck.
So that was the mechanism for addressing
the uncertainties and saying we really know where the
steam is, and we can take full credit for it. We've
elected to be conservative in those particular
aspects.
The heat and mass transfer correlations:
There was a lot of separate effects tests done by
Westinghouse. There is published data, and all of
that information went into developing the correlations
for mass and heat transfer used in the WGOTHIC code.
As we are all aware, there's a lot of
scatter in data. You can draw a line. You can try to
draw a 95 percent confidence line. What Westinghouse
did was they put bounding multipliers on the mass and
heat transfer correlations to basically bound the data
when doing the calculation.
They also used the conservative mass and
energy approaches that are prescribed in the standard
review plan where, during either the LOCA or the steam
line break, you maximize the energy release and the
mass release to its ultimate almost. You look at a
large double-ended break. You use Moody multipliers
for the low -- You make sure all the stored heat is
reduced quickly.
In the steam line break, similar
conservative mass and energy methodologies are
prescribed, and Westinghouse basically uses those in
their evaluation model.
So overall, the model that has been
developed and approved is a very conservative
calculational tool.
(Slide change)
MR. THROM: You start looking at some of
the differences between the AP600 and the AP1000, and
it was pointed out yesterday, this is just simply a 76
percent power uprate evaluation. Nonetheless, these
differences aren't as marked as they might be.
Even though the initial inventories for
both LOCA and main steam line break, the inventory and
the energy stored in the primary or secondary system
is larger. The pipings haven't changed. The coldleg
where the break is going to be for the LOCA is the
same size in the AP600 as the AP1000.
So your concern there is now the rate as
being slightly different, not significantly different.
Similarly, in the steam line break they have the flow
restricter in the line, which again, even though you
have the larger inventories and energies to release,
the rates are not appreciably larger than you would
expect for the AP600.
MEMBER WALLIS: Well, it's a race, because
the total amount of energy is bigger.
MR. THROM: Yes.
MEMBER WALLIS: So that, if the rate is
rapid, you are putting in an impulse of significantly
more energy into the containment.
MR. THROM: Right.
MEMBER WALLIS: Large LOCA, for instance.
MR. THROM: Yes. But you see that there
is the compensating factors within the containment are
basically the free volume is increased so that
compliance helps out a lot. The containment shell
itself is a little thicker, and they do put more water
on than they did previously with the AP600.
CHAIRMAN KRESS: For the question of the
mixing in the containment, the lumped parameter node,
have you given any thought to using a CFD code to look
at that?
MR. THROM: Not at the staff level, no.
We think that, in general, if you look at scaling of
the break jet, you are going to see that it kind of
supports the well mixed assumption. All the evidence
out there does support it.
Where you see some conservative in the
evaluations, I think, for some of these real high
energy jets, you would expect the steam to almost hit
the ceiling or the dome of the containment where it's
really cold due to this water, and get significant
condensation.
Whereas, in the lumped parameter model, as
I said earlier, you are going to be always mixing this
with the air. So there's going to be a tendency to
decrease the condensation rate and basically calculate
a somewhat higher pressure.
CHAIRMAN KRESS: Well, I was worried about
hydrogen stratification, which goes the other way, you
know. You don't want to assume well mixed there. You
want to know what you actually have.
MR. THROM: Right. And we don't believe
we have that particular issue for the design base
events.
CHAIRMAN KRESS: Because that's not enough
hydrogen to deal with.
MR. THROM: That's dealt differently.
There are other things that come into play when you
look at hydrogen, when you look at equipment
qualification, when you look at subcompartment loads.
GOTHIC is part of the tool, but there are
other things that come into play when you look at
those. So what we are really focusing on here right
now is the Chapter 6 evaluation for the peak pressure
response for the containment.
(Slide change)
MR. THROM: Westinghouse looked at these
changes. First of all, there's nothing that changed
in the PIRT. The rankings are basically the same.
Actually, they are the same. The process has not
changed. It's condensation on the inside of the
shell, evaporation of water on the outside of the
shell.
So we've not identified any new phenomena
that need to be included in the models. However, we
did feel quite concerned that the mass and heat
transfer models still be used within applicable
ranges, and the focus of the review was on that
particular aspect, to go back and look at the mass and
heat transfer correlations and look at the expected
response of the AP600 to assure that the mass and heat
transfer models and the tests that were run to develop
those models covered the range of the AP600.
MEMBER WALLIS: These are natural
convection models?
MR. THROM: Natural convection.
MEMBER WALLIS: High Rehle numbers beyond
most of the database.
MR. AUSTIN: This is Rick Austin from
Westinghouse. The Rehle numbers are very high for
AP1000 and AP600.
CHAIRMAN KRESS: 1010?
MEMBER WALLIS: Are they beyond the
database, the Rehle numbers?
MR. AUSTIN: In the chimney they are.
There's a table in WCAP-15613 that compares the heat
and mass transfer correlation -- the dimensionalist
groups for the heat and mass transfer correlations,
with a test data range, the AP600 range and the AP1000
range.
MR. THROM: And what he is talking about
is the chimney area, which is up here. And because
there is insufficient data to cover those ranges,
Westinghouse uses the Ichida correlation in that area,
and that's deemed to be a conservative correlation.
So that's kind of the one aspect where
within the data that's the place where it wasn't
adequately covered by any experiment. So Westinghouse
in the evaluation model -- they elected to use Ichida
correlation for that particular regime.
Basically, what you see when you look at
the dimensionalist numbers, the Grashof number,
Reynolds number that you use in the correlations, for
the most part, the AP1000 brings those numbers up into
the higher range of the data, but they are covered.
That's covered in the WCAP that
Westinghouse provided and verified by -- They have
done scoping calculations and backed out the numbers
and have shown that the peak numbers are within the
range, and they are not really pushing the upper
bounds of any of the ranges.
When they looked at the film flow, they
found out that the lower end for the Reynolds number
in the heat flux data that they have for film
evaporation was still being covered.
The one concern that we did have was with
the higher power, higher energy being reduced by a
LOCA, we were somewhat concerned with the peak shell
temperature before a credit would be taken for the
PCS, and it was probably from a statement that was
made in one of the topicals that said, well, if you
got really hot, you know, there might be a problem,
but we didn't find any design basis events that would
do that.
We challenged Westinghouse on that, and
they did an evaluation that showed that the peak shell
temperature prior to credit for the PCS water would be
about 180 degrees. So the film model that's being
used in WGOTHIC is applicable.
We were concerned that, if the shell
temperature had gotten up into the 212 degree range
where we might have to worry about a boiling front or
potential breakup of the film, that that would be a
concern. But calculations they have done and some
preliminary calculations we have done indicate that
that is not going to be a problem for the AP1000. A
lot of that is due to the thicker shell.
MEMBER WALLIS: This run down the shell is
-- Does it run as rivulets?
MR. THROM: It's a film.
MEMBER WALLIS: It's definitely a film?
MR. THROM: Yes.
MEMBER WALLIS: It's so much water that it
really wets everything?
MR. THROM: Yes. Yes. It doesn't
necessarily cover the full circumference of the shell.
They have -- The water distribution tests were run by
Westinghouse. These were basically full scale,
partial height tests where they were done cold, but
for different flow rates they came up with the
coverage that you would have on the shell, and that's
part of the model.
You would expect, actually, somewhat
larger coverage in a heated environment, but we don't
take any credit for that, because we don't have any
data to really do that. So we look at, for example,
early on about 90 percent of the shell being covered.
Later on, as the water flow rates
decrease, it drops off to like 50 percent, and then 25
percent, but there's adequate --
CHAIRMAN KRESS: But they are using the
same numbers for the AP1000?
MR. THROM: We think they are. It's one
of those phase 3 things where we get to it -- I'll
cover some of the issues that are going into phase 3.
They basically parallel phase 2.
The thing is at this particular point the
analyses that Westinghouse first presented back in
December were with a one-node model, which we hadn't
seen before. It gave us a lot of difficulty.
They came back in with a revised
calculation in September which went back and used the
109-node model that they used in the AP600. They
couched those as unverified results. However, as
scoping calculations, those were the numbers that --
the calculations that they used to go back into the
experimental database to assure themselves that they
weren't creeping up on any ends of the data where,
when the final calculations are done, you would expect
to have an issue with needing more information.
That particular comment leads into one of
the early concerns we had with the larger height of
the containment and the potential for more complex
mixing patterns that could influence the mixing and
the buoyancy.
We expected less homogeneity of the
environment and potentially higher temperatures in the
dome, and with the one-node model we didn't see how
they could address that issue.
They have gone back now, and they are
using the multi-node approach. So we believe that the
calculation they get will address these changes and
also still lead to the conservative pressure
calculation.
The scope -- As I said earlier, the
scoping studies performed by Westinghouse show that
the mass and heat transfer correlations are being used
within their acceptable ranges. We don't -- other
than the issue with the Rehle number and the chimney.
That's already addressed through the Ichida. But when
you look at the mix convection or the assisted
convection correlations that are being applied, the
data is applicable to the scale of the AP1000.
Well, in summary, the phase 2 review,
we've not identified any new phenomena that we think
need to be incorporated in the models. The current
rankings remain unchanged, and they have some heat
transfer correlations that are being used within their
applicable range.
Using the approved model and methodology,
we believe that the WGOTHIC is applicable for the
AP1000.
As I pointed out earlier, when we get into
phase 3, we need to confirm most of these findings.
The calculations we have seen today still have not
applied the evaporated flow model, and this is an
iterative process that Westinghouse uses to assure
that they don't have numerical instabilities in the
code.
That basically is a methodology for only
applying as much water as you can basically evaporate,
and that was approved as part of the original
calculation, but using the evaporated flow model in
combination with the Chun and Seban correlation, one
of the restrictions we put on the evaluation model to
address wavy flow.
The standard review plan mass and energy
release is one thing that is not yet done in the
scoping analysis. In the standard review plan, for a
pressurized water reactor after blowdown, there is a
30, 90, two minute period of time called refill at
which time the accumulators are basically condensing
the steam that is in the reactor vessel, and there's
no real release into containment, although in the real
world you would expect that the heat structures would
be condensing the blowdown steam, in the standard
review plan the expectation is that you collapse that
region.
So you don't get any credit for heat
removal while there is no mass and energy going into
containment. The calculations they have done haven't
included that, but they should not be any significant
impact on the application of the mass and heat
transfer models. It's just that we haven't really got
the committed to design basis action evaluation done
in accordance with the standard review plan, which is
the expectation.
Also the calculations we have seen to date
for the AP1000 are still using basically the ADS-4,
IRWST and sump injection and mass flow rates that
were developed during the AP600. If you remember,
it's the ADS-4 counteracted by the IRWST that leads to
and turns around the second peak.
So, you know, those numbers need to be
developed in the final analysis. While we don't
expect that any of this will lead to exercising the
mass and heat transfer correlations outside their
range, we want to look at it again, particularly.
We have them in the safety evaluation, a
requirement that when the mass fluxes on a containment
shell get much larger than the AP600, Westinghouse has
to revisit the Kline numerics to make sure that there
aren't any instabilities in that. So that's part of
the final review.
At this particular time -- In the
background package there's two pages of all of the
conservatisms that are in the evaluation model.
There's a lot of them, almost point by point. The
input is taken to be conservative.
MEMBER WALLIS: Your design pressures are
psia in the way it's put here?
MR. THROM: Yes.
MEMBER WALLIS: So you've added 14.7 or
something?
MR. THROM: Yes.
MEMBER WALLIS: Can't really use the
correlate psig. Atmospheric pressure goes up and down
a bit. There's a difference.
MR. THROM: Well, yes, but when you -- If
you want to use a percented ranking, you really should
use absolute. That's the only -- I guess in today's
environment we should be using some type of Pascal
anyway
MEMBER WALLIS: Do you have predictions --
they have predictions yet for the AP1000 max pressure
in the containment?
MR. THROM: Yes, we have two sets of
calculations from Westinghouse. We have the scoping
calculations which were not done with the model we
expected to see, but for the main steam line break
they were calculating on the order of 70.7 psia.
MEMBER WALLIS: So it's pretty close.
MR. THROM: Close, yes. And in
containment you will basically see that it is always
very close, but understand that the calculation is
extremely conservative.
For the LOCA, in the scoping calculations
that were presented to us back in December of 2000,
those again didn't really comport with the standard
review plan methodology. What they used there was a
five-hour time period to remove the sensible heat from
the primary system, where in the SRP methodology you
basically do that over an hour.
So that calculation showed that the peak
pressure was about 60.7 psia, but that's not a fair
marker. In the calculations they provided in
September where they went back in and started going to
the detailed model, they went back in and used the
one-hour time period, and I believe they are
calculating just about the same number now for the
LOCA, about 70, in that ballpark.
So there is a chance that things might
change a little bit, once all of the final boundary
conditions are input into the code.
MR. BOEHNERT: What's the design limit?
MR. THROM: Excuse me?
MR. BOEHNERT: The design limit, Ed?
MR. THROM: 73.7. That's along about
slide 3. There is margin. I mean, in the AP600 we
cut even a lot closer than that.
(Slide change)
MR. THROM: We have developed a contain-2
model. I've got a picture of it up here for those who
are interested. It's got 26 nodes in it. What's not
shown is the environmental node that's node 26.
It's similar in nature to the WGOTHIC
model, but we don't have as much detail in the model.
The progression of Westinghouse's development of the
AP600 model into more noding than they probably really
needed, because they were looking at a different
approach early on.
We have done calculations using the
scoping data, which is not final data, and for the
main steam line break we calculated 69.4 psia as
compared to the 70.7 that Westinghouse got.
In the contain model that we developed it
was an offshoot from the AP600 model, which was
developed very early in the process. In the model we
have modeled all of the heat structures. We've
modeled the heat structure for the upper deck.
So when we did the scoping evaluation, we
calculated 54 psia as a comparison to the 60.7 that
Westinghouse got for the same type of calculation.
What we did do -- and the problem with contain is it
meets everybody's expectation that there are no dials
in the code that make the analyst's job easy to turn
things on and off.
What we were able to do was mimic the mass
and heat transfer penalties. When you apply that to
the calculation, you see a 2 to 3 psi increase. So
the calculations are fairly comparable. The main
steam line break is showing to be the one, I think,
from the pressure point. We are getting about the
same calculation in most of the PCS characteristics.
The heat structure characteristics are
really not impacted by the main steam line break,
basically because of where you release the steam in
the model. You release it at an elevated location.
The lumped parameter model has a difficult time of
getting the steam back down into the areas where you
have the concern about how you address potential
stratification and mixing.
So we expect that WGOTHIC is applicable to
the AP1000, basically, within the context of the
evaluation model and the conservativeness of the input
and the look at the international database and the use
of the lumped parameter model.
What we've seen to date suggests that,
when the design calculations are done, the mass and
heat transfer correlations are going to be used within
their applicable ranges, and there is no need to
broaden the database.
That's all I have.
CHAIRMAN KRESS: Are there any comments or
questions from the members before we hear from
Westinghouse response on this. Seeing none, I guess
we will see if you have any comments you would like to
make.
MR. CUMMINS: This is Ed Cummins. We are
in basic agreement with the staff's conclusion. We
would like to respond to an earlier ACRS question
relative to CFD modeling, and show you what we have
calculated. Rick Austin is going to do that.
CHAIRMAN KRESS: Great.
MR. CUMMINS: Thank you.
MR. AUSTIN: Okay. My name is Rick Austin
from Westinghouse Electric. I was asked to present
some work that we did to qualitatively compare the
mixing in AP600 to AP1000. We did this when we were
looking at the PIRT and the scaling issues, and it's
contained in the WCAP-15163.
(Slide change)
MR. AUSTIN: We looked at -- WE built
models of the AP600 and AP1000 above deck operating
region in two dimensional CFD code. Star 3-D it's
called. Those results, as I said, are presented in
15613. We showed those results to the NRC in, I
think, the Thermal-Hydraulics -- ACRS Thermal-
Hydraulics Subcommittee.
Dr. Wallis asked us if we could look at
that in 3-D. So Dr. Milorad Dzodzo -- he's our CFD
code expert -- put together a 3-D pie slide model with
the same CFD code.
MEMBER WALLIS: A 1.5 degree pie?
MR. AUSTIN: 1.5 degree --
MEMBER WALLIS: Pretty thin sliver of pie.
MR. AUSTIN: -- pie slice. The 1.5 degree
pie slide model we built has approximately a million
cells. So the computing time was excessive, and I
think it took him a week to generate the results that
I'm going to show here. I think he might have had a
dual processor type computer, too.
MEMBER WALLIS: Now you said you had --
Well, you don't have a steam plume in there?
MR. AUSTIN: Yes. This is air only. We
couldn't model the condensation of the steam.
MEMBER WALLIS: This is air only?
MR. AUSTIN: Right.
MEMBER WALLIS: But there is a heated --
MR. AUSTIN: There's a heated plate at the
bottom. It simulates a hot wall at the bottom on
here, and there's a cold wall here and a cold wall on
top.
MEMBER WALLIS: In reality, you are
injecting steam rather than having all plate.
MR. AUSTIN: That's correct.
MEMBER WALLIS; Why didn't you inject --
Oh, but you couldn't remove mass, I guess. In the CFD
model, you didn't know how to remove the mass.
MR. AUSTIN: That's right. Couldn't
condense the steam.
I guess when Dr. Dzodzo looked at that, he
suggested that the steam, because it was lighter than
the air, would have probably provided even a little
better mixing --
MEMBER WALLIS: Because of more buoyancy,
it tends to go up more.
MR. AUSTIN: Right. A little better
mixing than what we would see with the CFD analysis.
So this first slide just shows the basic
model that we put together. I don't know if this is
going to come out very well. That's --
MEMBER WALLIS: I think it's just the
noding. The whole thing is full of nodes.
MR. AUSTIN: That's the noding structure
in the corners of the pie slide. It's highly detailed
on the edges. I think he was trying to resolve the
boundary layer. Those are like one millimeter size
width cells there.
(Slide change)
MR. AUSTIN: This is a 2-D model
prediction for the temperature distribution for
AP1000, and I'll put up -- I probably could use two
projectors.
(Slide change)
MR. AUSTIN: This is the 3-D model that
results. I guess you can imagine both sides of that.
MEMBER WALLIS: You're going to have to
tell us something. Everything looks a kind of uniform
blue here. So I'm not quite sure. The other one
showed more --
MR. AUSTIN: The colors are --
MEMBER WALLIS: The other one showed more
clearly a circulation pattern. The other one shows a
transition on the wall in color, which I don't see in
the 1000. It may be just the way that the --
MR. AUSTIN: Yes, it's very thin -- a very
thin boundary layer on this one. This one, you can
see. It just may be the way the colors came out here.
MEMBER WALLIS: Well, the plume is more
evident in that one, and the wall boundary is more
evident in the righthand one.
MR. AUSTIN: Right.
MEMBER WALLIS: So does this say it's well
mixed or does it show that it is a sort of circulating
pattern with a not very well mixed region in between?
MR. AUSTIN: It shows a circulating
pattern with a hot plume rising up the center and the
colder plumes falling down along the outer walls. In
the center portion -- There is a velocity profile plot
here. The center portion or that little donut region
-- this is what the 2-D predicted, and here's the 3-D
prediction, same type of behavior.
The donut region around the middle there
is -- That is somewhat stagnant. It doesn't have much
velocity.
MEMBER WALLIS: So it indicates that you
don't have a well mixed -- So then you are going to
argue the well mixed is conservative, I guess. That's
the Westinghouse position, isn't it?
MR. AUSTIN: Yes.
MEMBER WALLIS: This is not well mixed,
but the well mixed assumption in the code is
conservative?
MR. AUSTIN: Yes.
CHAIRMAN KRESS: I would have called this
well mixed.
MEMBER WALLIS: Well, that blue region
isn't particularly well mixed.
MR. AUSTIN: The blue region doesn't
circulate as much. There's very little circulation in
that region.
MEMBER WALLIS: You have this sort of
mixing in your breadmaker, you would get lousy bread.
I don't know how to describe it to you.
CHAIRMAN KRESS: Does the CFD model have
a turbulent term in it?
MR. AUSTIN: Dr. Dzodzo would be the
expert on that. I'm not a --
MR. BROWN: Bill Brown from Westinghouse.
I think it does, but I think you have to look at the
range of numbers here, too, before you get carried
away with making -- I think --
CHAIRMAN KRESS: That's what I was looking
at.
MR. BROWN: Yes. The differences are very
small, Dr. Wallis. They are very small.
MR. AUSTIN: The temperatures are real
small.
MR. SIEBER: Ten degrees or so.
MR. BROWN: You are only talking a few
degrees. So over the type of height -- we're talking
about, you know, hundred feet plus.
MEMBER WALLIS: I think we have to look at
the change in temperature of that blue annular
region, whatever you call, Taurus sort of region. Did
that actually warm up during the transient or not
much?
MR. AUSTIN: Yes, this one shows -- Well,
the temperature there is probably on the order of 370
degrees.
MEMBER WALLIS: So even the coldest bit
warmed up a lot. Is that what we conclude? I can't
really tell. What was the initial temperature
compared with the temperatures we are seeing here?
MR. AUSTIN: The initial temperature --
The cold all is at 366k, and the hot wall is at 394k.
MEMBER WALLIS: And the initial
temperature of everything was?
MR. AUSTIN: 385k. It's the average. We
use an average temperature.
MEMBER WALLIS: When you use k, it's
difficult to see the differences. I guess what I
would want to know is how much did that cold region,
the one affected region, change its temperature. You
can't tell just looking at it. That dark blue blob,
that elongated donut thing -- it doesn't mix very
much.
CHAIRMAN KRESS: I think this is safe
calculation.
MEMBER WALLIS: Does it go up in the
middle and down the wall in that thing?
CHAIRMAN KRESS: This is steady state
calculation?
MR. BROWN; Yes. Bill Brown. This is
steady state. This is after many, many iterations.
That's why it took a week to get here.
MEMBER WALLIS: So now you have to do the
transient.
CHAIRMAN KRESS: What this indicates to me
is that with those kind of temperature differences
inside the containment that you are very likely to
have a well mixed containment.
MEMBER WALLIS: I think what we have to
look at is the velocities here and see how rapidly a
chunk of fluid goes around compared with how rapidly
things are changing with time.
CHAIRMAN KRESS: Yes, that would be
something --
MEMBER WALLIS: How close are we to
Fauske's steady state? This is useful.
MR. BROWN: Bill Brown again. Don't
forget that one of the original intentions you had
here was just some interest in the 2-D versus 3-D
result. So that was originally.
MEMBER WALLIS: Also 600 versus 1000, too.
MR. BROWN: Yes. That was originally why
we did it, was to address the L/D type of aspect ratio
difference. One thing we also notice that Rick may
not have mentioned is that in the 3-D -- I believe,
Dr. Kress, maybe you said this at one of the last
meetings, that you expected maybe a little improvement
perhaps in the results.
What we noticed is that, if you look on
the picture on the left, that the plume actually
wasn't quite -- from what we could tell in the details
and looking at all the iterations, wasn't quite
actually hitting the top of the containment, but now
in the 3-D it actually is hitting the top of the
containment. So we actually get a little improvement.
MEMBER WALLIS: First I will also tell
you, heat transfer coefficients on the wall. Now this
isn't condensation, but there's probably a way to go
from the heat transfer coefficients you get here to
what they should be with condensation in some way that
you can sort of verify that heat transfer behavior is
conservative or whatever you want to show. You should
be able to get some heat transfer information on the
wall from this sort --
MR. BROWN: We can get dry heat transfer,
right.
MEMBER WALLIS: Yes, but then there's
probably a way you can show, you know, by analogy or
something. You can scale it.
CHAIRMAN KRESS: I'm pretty sure the CFD
code has it modeled in it for the heat transfer. With
that many nodes that small, it probably uses the
conduction equation.
MEMBER WALLIS: Then you can compare it
with some-- Well, you presumably used some kind of
Rehle number correlation or something for the heat
transfer coefficient?
MR. AUSTIN: Inside containment, yes. We
used free conduction correlations.
MEMBER WALLIS: It would be interesting.
This isn't just free convection on the wall. It's a
driven circulation pattern. So it's somewhat
different from just free convection on the wall. It
would be useful for you to compare the two. You could
probably show that neglecting the circulation is quite
conservative.
Just use free convection on the wall. You
may get a heat transfer coefficient, say, half as much
as you predict here for the air case. That might help
you. I don't want to do your work for you.
It would be useful if you could show,
because of circulation, there is really more heat
transfer, say, in air than you would predict using the
kind of assumptions you use for your heat transfer
calculation.
MR. AUSTIN: And we agree.
MEMBER WALLIS: And then that would be
reassuring.
MR. AUSTIN: That's all I had.
CHAIRMAN KRESS: We appreciate that. That
was very useful, I think.
MR. BOEHNERT: A question for
Westinghouse. Those slides are labeled proprietary.
Do you really mean that?
MR. AUSTIN: No.
MR. BOEHNERT: We can cross that off.
MR. BROWN: That was from my proprietary
presentation. Those slides there aren't proprietary.
Sorry about that.
CHAIRMAN KRESS: In the full Committee
meeting in March we only have two hours to cover all
the stuff we covered in a day and a half -- well, I
guess one day, counting both of them. So we need to
have some idea of how to condense all this.
My feeling is on yesterday's presentations
on the DAC and the exemptions, I think you can just
use one slide to tell what the exemptions were. I
think everybody is in full agreement on those almost.
For the DAC, I think the comment that
Westinghouse is going to request for the seismic
structure a hard rock side makes things a lot easier
in terms of what we say, because that sort of takes
those out of DAC space.
So given that, I think you just clarify
what Westinghouse intentions are there, and then that
means we focus the issues of DAC on the piping. I
think that may be one you want to really focus on,
because I think that's one the Committee would have to
make some sort of judgment on, if we want to give our
take on it, our thing sort of.
I would focus most of the talk on the
piping DAC. Now you have to do something about the
codes, how to get all this -- I like the thought of
the staff showing -- giving us an impression of what
the depth of their review was.
You had a couple of slides that -- for
talking about the code applicability and the data
applicability. So if you could just give the full
Committee an impression of the depth of your review
and the basis for your conclusions on that.
With Westinghouse, I particularly liked
that stuff you did to show from the entrainment that
it's self-limiting, and I think that would be useful.
A lot of this on the code and the data, I
think, is going to come down to the entrainment issue.
So I would focus on that as much as I could during the
two hours we have. This is the thing that I think is
going to give us problems.
I think with the containment and GOTHIC,
just a very quick overview on that. I didn't see very
much contentious. Everybody seems to be in agreement
on that. So I wouldn't do a lot with that. I would
just point out what it is.
So if the members have some other thoughts
on what the two hours might consist of, I would
welcome comments at this point.
MEMBER WALLIS: Well, I'd like to see the
scaling of the summary applied to the scaling that
showed some of the inserts for red indicating the few
areas where scaling needed to be investigated further.
CHAIRMAN KRESS: Oh, yes, definitely I'd
like to see that, too.
MEMBER WALLIS: I think we need the
picture showing entrainment in the two locations. I
think we need sort of the summary picture of is there
or is there not a possible issue with entrainment. I
mean, the numbers, actually Jg*, and there's just one
slide. I think there was a clear message there.
MEMBER SIEBER: I think you probably don't
want to go through all the math and logic in the
scaling.
MEMBER WALLIS: No. I don't think we need
all Marino's equations.
CHAIRMAN KRESS: Although that was good
stuff.
MEMBER WALLIS: I think there's a bottom
line for Marino, is that if you make different
assumptions about the void fraction, you get different
rates of loss of inventory. That's sort of the bottom
line. Maybe that can be done in five minutes.
CHAIRMAN KRESS: Yes.
MEMBER WALLIS: That sort of ties in with
what Steve is saying. I think also in the RELAP you
don't need anything like as much as we had this
morning. What's sort of the bottom line with RELAP?
What are you doing with it? Where is it going, maybe
a couple of slides there.
MEMBER POWERS: Tom, may I make a
suggestion, that for the staff presentation you invert
the order and begin with Mr. Throm's presentation so
that he can remind the members who are not here about
AP600, because most of those members that are not here
were not part of the AP600 review.
CHAIRMAN KRESS: That would be a good
suggestion.
MEMBER POWERS: And he has a caricature
that I think would allow him to point out some of the
-- He used it to point out some of the key containment
features, and I would encourage him to go ahead and do
that.
He might just also note a few of the
critical features for the discussion of the in-vessel
phenomena.
CHAIRMAN KRESS: There was a -- in the
chart with the table on it, comparing notes. That
would be a good thing to have up there.
MEMBER POWERS: I would use a picture.
CHAIRMAN KRESS: Well, a picture is going
to look just like AP600.
MEMBER WALLIS: A realistic picture, not
with a lot of content.
MEMBER POWERS: Well, I mean, sort of the
plant.
CHAIRMAN KRESS: But that chart showing
the --
MEMBER POWERS: What I'm thinking of is
those members that are not here -- I mean, nearly all
of them are --
CHAIRMAN KRESS: It would be a good idea
to have a picture.
MEMBER POWERS: -- relatively new.
CHAIRMAN KRESS: Yes, and then the chart
showing the differences between AP600 and AP1000.
MEMBER WALLIS: When words like CMT are
used, there needs to be, you know, a very quick
tutorial.
CHAIRMAN KRESS: Re-explain what those
devices -- where they come into play and when they are
used. That would be easy.
MR. CORLETTI: Would you like Westinghouse
to do just a comparison, a five minute comparison of
those two plans?
CHAIRMAN KRESS: Yes, I think that would
be a good idea. Let Westinghouse do that.
MEMBER POWERS: The members that are not
here, almost none --
CHAIRMAN KRESS: Yes, they are all new and
weren't here for AP600.
MEMBER POWERS: -- were present for the
AP600, and nearly all of them are thermal-
hydraulically averse. I like these guys a lot, by the
way.
CHAIRMAN KRESS: Yes. What would be good
would be a little short tutorial on the philosophy of
the passive systems, where they come into play, and
why, and that would be useful for those members, I
think. A good comment, Dana.
Any other thoughts?
MEMBER WALLIS: Now we are supposed to
write a letter on this?
CHAIRMAN KRESS: We have in mind a letter,
because I think the staff plans to go to the
Commission at the end of March with their feelings on
both the DAC and the exemptions and the code
applicability.
So I guess the Commission would appreciate
our thoughts on those.
MEMBER WALLIS: So we would be both sort
of endorsing some of the preliminary conclusions or
something or saying that they are on the right track?
There's no real conclusion yet, is there? So we got
to be careful about it.
CHAIRMAN KRESS: Well, we'll be careful.
MEMBER WALLIS: Saying they are on the
right track.
MR. CUMMINS: Maybe I could make a
comment, add comments. What we asked the staff to
conclude was that the AP600 tests were valid for the
AP1000, that the AP600 codes were applicable to the
AP1000, and their position on DACs and exemptions.
So we didn't ask them to conclude the
safety of the AP1000 plant, but we did -- There are
some conclusions for which we would like to achieve.
MR. WILSON: This is Jerry Wilson. So
what we are going to do is send a letter eventually to
Westinghouse answering those questions that Mr.
Cummins just summarized, but prior to sending the
letter, decided we wanted to run it by the Commission.
So as you say, I'm sure the Commission
will want to hear your views on it as they consider
that.
CHAIRMAN KRESS: We'll comment on those
four issues, I think.
MR. DZODZO: And if I may add a couple of
comments, first I'd like to thank you for very
insightful questions, and if there were any
shortcomings in answering those questions, it was
because a project limited to the otherwise unbounded
curiosity of reviewers into a scope of phase 3.
My second comment is also that the
reviewers also limited to design basis here. The
applicability of some codes may resurface when I go
into severe accidents, and in particular into source
the calculations, since thermal-hydraulic conservatism
sometimes is counterproductive to source the
calculations.
As we know, the position mechanism are
sensitive to local thermal-dynamic conditions.
CHAIRMAN KRESS: And we will look forward
to reviewing those during the phase 3 part of the
thing.
MEMBER POWERS: It strikes me, Tom, that
we are going -- One of the central issues arises in
connection with the DAC is the rule in Part 52
concerning the completeness of the design information.
CHAIRMAN KRESS: I think you're exactly
right. That is the issue.
MEMBER POWERS: And the Commission that
endorsed that rule is different than the Commission we
have now, and we are going to have to discuss for them
a little bit on why anyone would put such a silly rule
in.
CHAIRMAN KRESS: I think you are exactly
right. In fact, in mentally thinking what might be in
our letter, that was one of the things I had in mind
discussing. I think you're exactly right. That's the
central issue.
MEMBER SIEBER: Yes, it is.
CHAIRMAN KRESS: And particularly those
two criteria, you know. So that, I think, will
definitely end up in our letter, some discussion on
that, and how it impacts on the question of the DAC
for the piping. I think you're exactly right.
That's one reason I wanted a bit of
concentration on the piping DAC issue.
MEMBER POWERS: It's a good object lesson.
The other object lesson, of course, is the
instrumentation and control, because that's a case
where nearly everybody says, yeah, that's a good idea.
CHAIRMAN KRESS: Everybody agrees on that.
Yes.
MEMBER POWERS: Yes, and so you got two to
compare, and you need to compare up sides and down
sides.
CHAIRMAN KRESS: Yes. I think that should
be part of our discussions on the full Committee.
MEMBER POWERS: So this is going to be a
difficult letter, isn't it? I mean, it's a lengthy
letter.
CHAIRMAN KRESS: I don't think so. I've
already got it written in my head. But maybe it will
be.
MEMBER POWERS: The challenge that we face
is translating from Tennessee to English.
CHAIRMAN KRESS: Oh, that will be
difficult.
MEMBER SIEBER: I can't be any harder than
New York.
MEMBER WALLIS: Well, I'm glad to hear
that you folks understand what's going on with the
DAC. I mean, you said there was a problem with new
members understanding.
CHAIRMAN KRESS: Oh, that's right. You
weren't part of the DAC.
MEMBER WALLIS: I haven't a clue what this
DAC business is about. So you guys better understand
it well.
CHAIRMAN KRESS: Yes, we think we know
what DAC and ITAACs are, and Tier 1s and Tier 2s.
MEMBER POWERS: But we have to translate
from English from Tennessee to English and then to
academe use.
CHAIRMAN KRESS: Yes. You're right, it's
going to be difficult.
Okay. With this comment, have we given
you enough guidance? Are there any closing comments
anybody wishes to make?
Seeing none, I am going to declare this
Joint Subcommittee adjourned.
(Whereupon, the foregoing matter went off
the record at 10:58 a.m.)
Page Last Reviewed/Updated Monday, July 18, 2016