Thermal-Hydraulic Phenomena - July 17, 2001
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
Thermal Hydraulic Phenomena Subcommittee
Docket Number: (not applicable)
Location: Corvallis, Oregon
Date: Tuesday, July 17, 2001
Work Order No.: NRC-325 Pages 1-322
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
THERMAL HYDRAULIC PHENOMENA SUBCOMMITTEE MEETING
NRC-RES T/H RESEARCH PERTAINING TO
PTS RULE REEVALUATION
(ACRS)
+ + + + +
TUESDAY, JULY 17, 2001
+ + + + +
CORVALLIS, OREGON
+ + + + +
The ACRS Thermal Hydraulic Phenomena
Subcommittee met at Oregon State University,
Richardson Hall, Room 313, Corvallis, Oregon, at 8:15
a.m., Dr. Graham B. Wallis, Chairman, presiding.
COMMITTEE MEMBERS PRESENT:
GRAHAM B. WALLIS, Chairman
THOMAS S. KRESS, Member
WILLIAM J. SHACK, Member
ACRS STAFF PRESENT:
PAUL A. BOEHNERT, ACRS Engineer
VIRGIL SCHROCK, ACRS Consultant
NUCLEAR REGULATORY COMMISSION REPRESENTATIVES PRESENT:
STEPHEN BAJOREK, RES/SMSAB
DAVID BESSETTE, RES/SMSAB
NILESH C. CHOKSHI, NRC/RES/DET/MEB
JAMES T. HAN, RES/DSARE/SMSAB
JACK ROSENTHAL, U.S. NRC/RES/SMSAB
ROY WOODS, NRC RES/DRAA/PRN3
A-G-E-N-D-A
Agenda Item Page
Opening remarks and welcome from Dean. . . . . . . 4
Overview of OSU Nuclear Reactor Research . . . . .35
NRC Staff Report on Intentional Use of
OSU Work . . . . . . . . . . . . . . . . . 208
Pressurized Thermal Shock Research . . . . . . . 211
OSU PTS Test Facilities and Palisades
Operations . . . . . . . . . . . . . . . . 228
Summary of Integral System Overcooling
Test Results . . . . . . . . . . . . . . . 249
Numerical Simulation for APEX-CE MSLB and
SBLOCA Tests Using RELAP5/MOD 3.2.2
(Gamma Version). . . . . . . . . . . . . . 267
P-R-O-C-E-E-D-I-N-G-S
(8:15 a.m.)
CHAIRMAN WALLIS: The meeting will now
come to order. This is a meeting of the ACRS
Subcommittee on Thermal Hydraulic Phenomena. I'm
Graham Wallis, the Chairman of the Subcommittee.
ACRS Members in attendance are Dr. Thomas
Kress, Mr. William Shack. And also in attendance is
ACRS Consultant Virgil Schrock.
The purpose of this meeting is to discuss
the status of:
Item 1, the NRC Office of Nuclear
Regulatory Research -- big topic -- experimental
program being conducted at the Oregon State
University, OSU, APEX-CE facility, pertaining to
thermal hydraulic phenomena associated with
pressurized thermal shock, PTS, in support of the NRC
PTS Rule Reevaluation Program; and,
2, the RES Program underway at OSU to
investigate phase separation phenomena in support of
proposed model upgrades for the RES TRAC M and RELAP5
Codes.
The Subcommittee will gather information,
analyze relevant issues and facts, and formulate
proposed positions and actions as appropriate for
deliberation by the full Committee.
Mr. Paul Boehnert is the designated
federal official for this meeting.
The rules for participation in today's
meeting have been announced as part of the notice of
this meeting previously published in the Federal
Register on July 6, 2001.
A transcript of the meeting is being kept
and will be made available as stated in the Federal
Register notice.
It is requested that speakers first
identify themselves and speak with sufficient clarity
and volume so that they can be readily heard.
We have received no written comments or
requests for time to make oral statements from members
of the public regarding today's meeting.
Now I'm ready to start. I wonder if
Professor Adams is here.
MR. REYES: Mr. Chairman?
CHAIRMAN WALLIS: Yes.
MR. REYES: If I might, this is Jose Reyes
from Oregon State University. Our department head,
Andrew Klein is here. He could give a talk about --
CHAIRMAN WALLIS: Maybe he can, yes, give
us a few words, and then I'll go ahead with the
meeting. Thank you very much. Very appropriate.
MR. KLEIN: My name is Andrew Klein.
THE REPORTER: Could you go to the podium
so we could record you?
MR. KLEIN: Sure. My name's Andy Klein.
I'm Department Head of Nuclear Engineering here at
Oregon State University. And we're very pleased to
have this Committee, Subcommittee meeting here at
Oregon State University to review some of the work
that Drs. Reyes and Wu have done over the past few
years. It's been very important to the Department, to
the College, and the University.
This is one of the key programs in the
College of Engineering and certainly one of the key
programs in the Department of Nuclear Engineering.
And we're very pleased that the NRC has supported the
work here, and the DOE has supported work here over
the years. And we look forward to continuing to work.
If you have any questions, Teresa Culver,
who is in the back there, will be able to help us out
on logistics. If you have any technical questions,
I'll defer those to Dr. Reyes. Thank you.
CHAIRMAN WALLIS: Thank you very much.
Now we've gained some time. Let's
continue doing that.
Jack, are you ready for the opening
remarks for RES?
MR. ROSENTHAL: My name is Jack Rosenthal.
I'm the Branch Chief of the Safety Margins and Systems
Analysis Branch in the Office of Research. And I was
just asked to make some opening remarks.
We see the APEX facility as very important
at this point, that there's little else of places to
do pressurized water reactor experiments.
And so we're using the OSU, both the
experimental facility and their analytic ability, to
work on issues such as pressurized thermal shock,
after which we would do some work on bore and
dilution, after which -- my sequence may not be right
-- we have some plans for work on -- to answer some
steam generators to the accident-related issues
involving flow mixing that the ACRS has already
reviewed.
And then after that to work on AP1000 in
some sort of yet-to-be-defined collaborative mode with
the Department of Energy, similar to the arrangements
that were made for AP600. So that's a continuing
long-term involvement.
In preparation for the meeting we were
going over some of the work on two-phase flow
separation. And we intend to put those models into
TRAC M within the next 12 months and probably sooner.
So the work is just very important to our overall
efforts.
Later Nilesh, Dr. Chokshi, will be talking
about an overview of the pressurized thermal shock
effort. And then the staff has just a slide or two as
topics come up.
CHAIRMAN WALLIS: Are there any other
remarks from NRC?
(Comments off the record.)
CHAIRMAN WALLIS: Yeah. Would you like
to? Yes. Thank you.
THE REPORTER: When they speak from out
there, we can't get them on the recording machine.
CHAIRMAN WALLIS: Do you have a portable
mic or something, or do they have to go up to...
(Comments off the record.)
MR. CHOKSHI: Good morning. My name is
Nilesh Chokshi. I'm Chief of the Materials
Engineering Branch. Your agent actually is Mike
Mayfield, but I think he's at the Excelon meetings.
So in place of Mike, I think I'll give you a brief
status and overview.
Let me start by thanking the Committee for
having the meeting here. Without this meeting I would
never have come to this place and see the facility or
understand a little bit -- improve my understanding of
thermal hydraulics.
So I'm looking forward as much as anybody
to this visit and seeing the facilities. So it's
going to be an interesting two days for me personally.
Since the focus of this meeting is the
thermal hydraulics and research of the facility and
discussion of projects, I'm going to limit myself to
just a brief overview of where we are and discuss
briefly the status of activities in two other areas of
probabilistic fracture mechanics and PRA.
And a number of people are here from the
staff, including Roy Woods. So if you have some
questions on PRA and so forth, he's here to answer.
But I'm going to take a very few minutes, and it won't
take me half an hour, as shown on the Agenda.
We have been briefing ACRS Committees on
the progress of this PTS Evaluation Project, so I
think my remarks are -- basically it does not go into
any background and introductory material, with a very
brief introduction, so...
I'm going to take a few minutes to put in
perspective this particular activity. The current 10
CFR 50.61 pressurized thermal shock rule was
promulgated in 1983.
And between the '83-to-'86 timeframe work
we conducted detailed studies called integrated
pressurized thermal shock on the three -- three
plants, I believe -- four plants, to develop
methodology when somebody cannot meet the PTS
screening criterion.
And as a result of this study, I think
that Reg. 1.154 was published in 1987. That's -- my
recollection is 1987.
And then during the '89-l990, the Yankee
Rowe application, attempts were made to use the Reg.,
and subsequently it was discovered that there were a
number of issues in applying that Reg. guide.
Since then a number of improvements in
area of probabilistic fracture mechanics, some data as
well as methodological. So in April '99 we initiated
this project. And I think that's sort of a
background.
And I think one of the features of this
program -- some of the features are, one, this is the
risk-informed application. There has been extensive
industry and public involvement. We are working with
a number of industry groups, as you know, Materials
Lab Program, EPRI, and utilities, particularly in the
PRA area to make sure that our models reflect the
actual plant conditions.
And, as I mentioned earlier, we have been
coming to ACRS frequently.
In the program the four plants that are
being looked at, as you know, are Oconee, Calvert
Cliffs, Palisades, and Beaver Valley.
With this, a little bit of introduction,
I'm going to jump into the status of the program.
MR. KRESS: You were using INEEL to do
these four plants?
MR. CHOKSHI: We started with INEEL, but
--
MR. ROSENTHAL: Yeah. I'll just put on my
army voice. We're doing the Palisades' calculations
inhouse.
The Oconee, Calvert, and Beaver Valley
calculations are being done at ISL under Dave
Bessette's supervision. But I want you think of this
as a joint staff and contractor effort, because there
isn't a day of the week that we don't interact with
them.
CHAIRMAN WALLIS: What is "ISL," Jack?
What "ISL"?
MR. HAN: Systems -- I can't remember.
(Simultaneous talking.)
MR. HAN: Information System Laboratory.
MR. ROSENTHAL: Information System
Laboratory.
CHAIRMAN WALLIS: It's a Scientec
derivative?
MR. ROSENTHAL: It's a Scientec derivative
because there were conflicts of interest or potential
conflicts of interest, so they spun off ISL.
CHAIRMAN WALLIS: Where are they?
MR. ROSENTHAL: Down the block from us.
They're in Rockville.
MR. BESSETTE: Yeah, they're two blocks
down Rockville Pike.
MR. CHOKSHI: Okay. I think the current
status, we are making progress in all three areas of
the major technical studies: the thermal hydraulics,
fracture mechanics, and the PRA.
One of the significant activities
underway, particularly in the fracture mechanics area,
is to finalizing the FAVOR Code inputs. The goal is
to have all the uncertainty models and all the
improvements by September. And then we will be going
into doing plant-specific analysis.
I won't say anything about thermal
hydraulics. I'll wait for the next two days, too, you
know.
And in the progress on PRA, is -- I'll
have more details, but you could have --
CHAIRMAN WALLIS: But you say it's
completed, so if we find anything that's not
completed, then --
MR. CHOKSHI: Well, that's why I sort of
skipped over, because --
(Laughter.)
MR. CHOKSHI: -- because I think it's
probably premature for me to say it is completed until
I hear. But I guess what was planned has been
completed.
CHAIRMAN WALLIS: "Completed" means
they've submitted the last bill; is that what it
means?
MR. BESSETTE: The money ran out.
MR. SCHROCK: Does this involve some new
research, or is this massaging old data to find ways
of improving its application?
MR. BESSETTE: No, it's new. This is
David Bessette from Research. It's new experimental
testing.
MR. SCHROCK: Done where?
MR. BESSETTE: Here, here. It's done in
the APEX --
MR. SCHROCK: Oh, here?
MR. BESSETTE: -- facility.
MR. SCHROCK: Okay.
MR. BESSETTE: And you'll hear -- this is
one of the main topics, is to hear about the testing
that was done at -- the next two days.
MR. SCHROCK: I guess I was surprised at
materials research. I --
MR. BESSETTE: Well, you're not going to
hear anything about materials --
MR. SCHROCK: Well, we'll hear what it is.
CHAIRMAN WALLIS: Are we going to hear
about this last one, the uncertainty in key variables
in thermal hydraulics?
MR. BESSETTE: Not too much in the next
two days. It's not --
CHAIRMAN WALLIS: Is that part of the work
here, is to look at uncertainty?
MR. BESSETTE: Of course it feeds into
uncertainty. You know, the experimental results give,
you know, us much more of a feeling about uncertainty.
And plus the analysis.
CHAIRMAN WALLIS: So who's going to do
that work, the formulation of uncertainty?
MR. BESSETTE: Well, officially the
uncertainty work is being done at the University of
Maryland by the Almenas, and Mosleh, and Modarres.
CHAIRMAN WALLIS: So they're using some of
the results from here to assess uncertainty?
MR. BESSETTE: That's right.
MR. KRESS: Before we take that off,
Nilesh.
MR. CHOKSHI: Okay, sure.
MR. SHACK: You're saying FAVOR, reeval-
-- revision of the PFM Code FAVOR will be done in
September?
MR. CHOKSHI: We only -- we are in the
process of implementing changes to the code.
MR. SHACK: Okay. Is that -- would that
be a good time for you guys to come back to the
Subcommittee, ACRS, and Materials Metallurgy
Subcommittee?
MR. CHOKSHI: Oh, I think so, yeah. Yeah,
we'll be -- and talked about uncertainty modeling and
different things, what improvement --
MR. SHACK: Um-hum. Yeah, because I know
there's been some questions about that; they wanted to
see that when it was completed.
Okay. Thank you.
MR. CHOKSHI: Yeah. Right now we are at
sort of documenting a number of things we are doing,
so we should be ready by September.
MR. SHACK: Okay.
MR. CHOKSHI: In fact, just a last
question regarding the uncertainties. And as part of
the PRA work, and as David mentioned, the University
of Maryland is looking uncertainties modeling in all
areas of the program.
MR. KRESS: How are they dealing with
epistemic uncertainties?
MR. CHOKSHI: We are considering. I mean,
all right. Now how is -- I think in each of the areas
is somewhat of a different question.
For example, in the materials area, on the
toughness and the flaw distribution, you know they
each -- on the flaw distribution, for example, we have
an expert -- expert elicitation as well as the ND
examination of actual vessels, trying to get, you
know, the different uncertainties.
On the toughness side there has been a
joint effort with industry in developing epistemic and
both aleatory uncertainties. So --
MR. KRESS: But specifically it's
generally by expert opinion?
MR. CHOKSHI: In part, not always. You
know, because on the flaw distribution, we have both
the data simulations as well as expert opinion. But
epistemically more -- most part, I think I would say,
an expert.
MR. SHACK: Nilesh, just on the PRA, I
thought you also had some additional work going on on
sort of containment performance under --
MR. CHOKSHI: Yes. Yeah. We --
MR. SHACK: That's too new to make it, or
--
MR. CHOKSHI: We are -- well, I think
that's sort of butted into this second bullet, for
acceptable risk figures of merit. And I would say
right now -- and, Roy, correct me -- that we are sort
of still developing some concepts before -- and we had
looked at some studies, actually, what, at Santa
Barbara?
MR. KRESS: Is that Theophanus?
MR. BESSETTE: Yes. Well, we had a small
effort with Theophanus where he looked at the effect
of vessel failure on containment.
MR. SCHROCK: That's an application of his
rho M process?
MR. BOEHNERT: Yes. He wrote up some
paper on it. It's a -- I think it's at least out in
draft form.
MR. CHOKSHI: But I think we are still far
from, you know, settling on that issue in any sense,
so...
MR. SHACK: I mean it will have
substantial impact on your acceptance criteria,
presumably.
MR. CHOKSHI: And I think when we start
doing some plant-specific analysis it will sort of
start. You know, how much we need to worry about that
will come into focus, I think once we get some
research and --
MR. KRESS: But my impression of the rho
M process is that it gives a bound that is such that
you can rule out the particular issue or sequence
because of the low probability.
In here you're supposed to be doing a best
estimate. I'm wondering how you are going to
reconcile that sort of difference with the rho M
process.
MR. CHOKSHI: Let me ask David.
Did you hear Dr. Kress' question? The
calculation of what work is being done by Dr.
Theophanus?
CHAIRMAN WALLIS: Well, I think maybe we
should move on to what's being done at OSU.
MR. KRESS: Yeah, we'll hear more about
this later. Yes.
MR. BESSETTE: Yeah.
CHAIRMAN WALLIS: We'll get to you
somewhere and ask you about Theophanus' work, but
maybe we should --
MR. BESSETTE: Yeah. And we also did some
inhouse work as well.
CHAIRMAN WALLIS: Yeah. We should spend
time here on OSU work, --
MR. KRESS: Yeah.
CHAIRMAN WALLIS: -- because that's going
to take a long time to go through, I think, anyway.
MR. KRESS: Yeah. Good thought.
MR. CHOKSHI: Okay. I think on the -- the
next viewgraph is a little bit more details on what
exactly is happening in the PRA area currently.
CHAIRMAN WALLIS: Nothing exact ever
happens in PRA.
(Laughter.)
MR. CHOKSHI: I will just give you what I
read. We'll be complete -- I think by this, shown
here on the last couple of bullets, that Oconee and
Beaver Valley PRA models will be revised by September
2001. And we are working with utilities to make sure
that all updates reflect the current plant operations.
And then the Palisades and Calvert Cliffs
by mid-November. And, you know, that that -- so by
September then we'll start applying FAVOR to Oconee
and start doing those calculations.
And I think, as it's shown here, that we
are developing the PRA models for Oconee and Beaver
Valley and basically updating Calvert Cliffs and
Palisades.
My next two viewgraphs are on the thermal
hydraulics. And I think I would say it's -- you know,
you have the -- and I will just plan to skip that,
because we were going to talk about this. And let me
talk about something I have more familiarity, so...
And the next major piece of the area is
the probabilistic fracture mechanics Code. And, as I
mentioned, that there have been significant
improvements in number of areas since the mid-'80s
when we completed the IPTS program.
The flaw distribution was found to be one
of the biggest area of uncertainties in the
application of 1.- -- record 1.154.
Since then we have looked at a couple of
pressure vessels through both nondestructive
examination and destructive examination as well as, as
I mentioned earlier, expert elicitation and simulation
of -- through the codes like prodigal codes at --
directing flaw distributions for the plate, weld, and
heat-affected zone materials. These are being right
now programmed to the FAVOR Code.
At Pacific Northwest National Laboratory
Steve Doctor and Fred Simonen have been doing most of
the work.
The other area has been the crack
initiation and arrest fracture toughness. There has
been also significant improvement in the modeling as
well as the data.
And one of the key issues here has been of
the uncertainty. And we have been working with the
University of Maryland to develop the epistemic as
well as aleatory uncertainties.
Significant amount of new Smoglie data for
the embrittlement correlations, and a new database as
well as better correlations are being incorporated
into the -- we also have a plant-specific material
properties.
The fluence maps. And I think the
beginning of this year we came and talked to the
Committee about the Regulatory Guide. And that
methodology has been applied to now the four plants to
make -- address the mix of the fluence.
There has been also the -- in the fracture
mechanics itself improvements, things like better find
element modeling, the treatment of residual stresses.
So all of this sort of has been incorporated into the
PFM analysis.
Early in 1998 we did incorporate some of
these improvements and did a test case, going back to
one of the IPTS plants. And at least based on that it
looked like there could be some reduction, reduction
of counterweightism in the screening criteria.
MR. KRESS: In this context what do you
mean by the "risk-informed model"?
MR. CHOKSHI: I didn't hear that.
MR. KRESS: Your first bullet, I was
wondering in the context of this what do you mean by
"risk-informed model"?
MR. CHOKSHI: Well, because this -- all of
this area has been integrated through the
probabilistic risk assessment in making sure that the
ultimate screening criteria and whatever we come up
with has the, you know, the framework of Reg. Guide
1.174 or similar, you know, risk.
CHAIRMAN WALLIS: So you're revising a
code using risk-informed methods?
MR. KRESS: Yeah. That's the part I
didn't understand.
CHAIRMAN WALLIS: Will you --
MR. CHOKSHI: Well, I think that because
of including the uncertainties in all the --
MR. KRESS: That means best estimate with
uncertainties, is what you're saying, right?
MR. CHOKSHI: With uncertainties, right.
Yeah, that's right. Exactly.
Because we're at the end of the -- when we
make a run we'll have a distribution of the through-
wall, correct. And then it will have been to involve
-- become involve with the initiating events and plant
logic and --
CHAIRMAN WALLIS: So it's not like sort of
looking at some phenomena and saying, "Well, this
phenomena is not important in its influence on the
answer; therefore, we might as well as not use an
equation. We'll just assume a coefficient of 2," or
whatever, "in order to get on with it," because it
doesn't matter. It's not that sort of level of
risk-informing.
MR. CHOKSHI: Right. It's more, I think,
a simulation, you know, Monte Carlo type --
CHAIRMAN WALLIS: Because I haven't heard
of people actually trying to look at the details of
codes using risk information. It would be interesting
to do someday.
Where is it that your actual uncertainties
in modeling have a big influence on the answer from
the point of risk; that would be very interesting to
do. But I don't think that you folks are doing that
yet.
MR. KRESS: That was the nature of my
question. I didn't understand it.
CHAIRMAN WALLIS: Yeah. That would be
very interesting to do, though. So now that you've
opened the door, maybe we should ask you guys to do
it. Think about how to do it.
MR. CHOKSHI: I think I already talked
about my next viewgraph. So let me just go to the --
give you the overall schedule and -- where we are. I
think in the last three bullets, it sort of summarizes
the --
CHAIRMAN WALLIS: Are you talking about
OSU here, or are you talking about much more general
stuff?
MR. BOEHNERT: No, just in general.
CHAIRMAN WALLIS: You're talking about
much more general stuff, aren't you?
MR. CHOKSHI: In terms of the schedule?
CHAIRMAN WALLIS: No. When you say things
like "good progress" thus far, that's a very general
statement about --
MR. CHOKSHI: Yeah. That's some -- I'm
talking about all three areas.
CHAIRMAN WALLIS: That says nothing about
OSU, does it?
MR. CHOKSHI: No.
CHAIRMAN WALLIS: Okay.
MR. CHOKSHI: Because I think at the end
of tomorrow you can judge.
Just to, I think, summarize. We will have
the fracture mechanics analysis of all four plants by
end of this year, December 2001. And then we will be
putting through, you know, combining with the PRA.
And by February, next February, we should
have that -- all of the plant-built specific analysis
done.
And the idea is to have the
technical-basis activity completed by July 2002. And
then we -- at that phase is the question of whether we
go, move towards -- move forward with the rulemaking
and probably some develop- -- about that time we will
be sending a second paper to the Commissioners and
seeking their advice based on what's the results we
got.
CHAIRMAN WALLIS: Well, you didn't say
much about thermal hydraulics. My understanding is
that thermal hydraulics are presently in the
calculations, but if OSU experiments give surprises,
then you would have to revise the whole analysis?
MR. CHOKSHI: Yeah. Well, I --
CHAIRMAN WALLIS: And that's a very key
thing to know about, is the --
MR. CHOKSHI: Well, I thought --
CHAIRMAN WALLIS: -- is how much of this
stuff that you say you're going to do all this on the
schedule and you're going to achieve things by certain
days. How much of that depends on getting the right
answer out of OSU?
MR. CHOKSHI: Obviously a great deal, but
-- but that's -- I thought since in these two days
you're going to look at this in much more detail, --
CHAIRMAN WALLIS: You know, I'm just
trying to put the thermal hydraulics in perspective,
but I didn't quite see how much -- it does play a
pivotal role, doesn't it, in your --
MR. CHOKSHI: Yeah. From --
CHAIRMAN WALLIS: -- development of
whatever the rule is that you want to come out of
this.
MR. CHOKSHI: Right. From my
understanding, at least I -- I am not expecting, you
know, that list as -- I don't know that there is
anything right now which is going to adversely impact
the schedule, but I -- I'll find out along with you if
there is something.
MR. ROSENTHAL: Let me just make the
comment that it's very easy to put a Gant chart
together. And Gant charts are terrific for building
office buildings where you pour the concrete and then
erect the steel.
And here we're more in a design process,
an iterative process. And so we don't know quite what
surprises will come to pass.
One of the early and crucial issues is
because FAVOR is a 1D Code was the effect of thermal
plumes. And I think we have some answers for that
that you'll be hearing about, because that was like a
make/break issue.
A preliminary discussion from the
University of Maryland says: Well, you know maybe we
can be very, very precise in our calculations, but we
really don't know these temperatures to within 25 C.
And -- but that's not a fatal flaw,
because we can pass that information on. And it's
sequence dependent with its associated probability to
the fraction mechanics people.
And then the fracture -- and then having
done the fraction mechanics calculations, we'll then
see if we can live with those -- that lack of
knowledge, that state of lack of knowledge, or if we
have to loop through and attempt to refine methods.
So we're trucking down the announced
schedule, but I think the fatal flaws haven't arisen
yet.
MR. CHOKSHI: Yeah. And I think to some
extent the schedule has built some of the
administrative process that -- you know, it's all
recognized that we --
MR. KRESS: The FAVOR Code uses the
thermal hydraulics as an input.
MR. CHOKSHI: That's correct.
MR. KRESS: Is the plan to have a sample
from that input in a Monte Carlo way, or is there some
other way you're going to deal with these
uncertainties?
MR. CHOKSHI: Yeah. You get the
temperature, the pressure and temperature
distributions and --
MR. KRESS: Pressure, temperature, and
heat transfer coefficient is, I think, --
MR. BESSETTE: Yeah. You know FAVOR takes
FAVOR samples from things like flaw size and those
kind of distributions.
MR. KRESS: Um-hum. Yeah.
MR. BESSETTE: But it takes as a single
value the thermal hydraulic input, but --
CHAIRMAN WALLIS: There are no flaws in
the thermal hydraulics.
(Laughter.)
MR. BESSETTE: But, of course, you could
input different --
MR. KRESS: You could -- you could --
MR. BESSETTE: -- you can change your
input, of course, for the thermal hydraulics, so you
can put in different temperatures. So within your
range of temperature uncertainty, you can vary that --
CHAIRMAN WALLIS: Yeah. You probably
should do that, at least a little bit.
MR. BESSETTE: But it just handles it at
a different form of input.
MR. KRESS: Well, the good thing about
that is you know how -- pretty much how it affects the
outcome, --
MR. CHOKSHI: Right. That's --
MR. KRESS: -- so you can choose the
bounding values, or something.
MR. CHOKSHI: Right.
CHAIRMAN WALLIS: I think that's part of
the Maryland approach, is to see how uncertainty in
one thing affects the outcome and develop some sort of
influence coefficient, or something. And...
MR. CHOKSHI: That's the end of my
presentation. So just I wanted to give you an
overview.
CHAIRMAN WALLIS: Thank you very much.
It's very helpful.
MR. CHOKSHI: Thank you.
MR. KRESS: Thank you. It was.
CHAIRMAN WALLIS: I note that Professor
Ron Adams is here, and we'll give him a chance to say
a few words a few minutes before we hear from
Professor Reyes and get down to work.
MR. ADAMS: Well, good morning, everyone.
Welcome to OSU. I've got to tell you that the weather
pattern today is unusual. It's normally sunny this
time in July, but you won't notice, as I can see.
I am pleased that you're here today. And
also this building that we're in is one of the special
places at OSU because it houses the nation's best
Forestry Program. And with that I want to talk about
what our plans are for engineering and how the work of
Dr. Reyes and Dr. Wu are impacting what we're trying
to accomplish.
OSU is building an engine for the
knowledge economy. And this is being driven by
national needs as well as local needs here in Oregon.
One of the things that's happened in
Oregon is that the technology sector is now the
largest employer. In fact, the employment in the
technology sector in Oregon now exceeds the
combination of all of the natural resource industries
combined.
Oregon State University has, in the last
century, built great programs in natural resources, in
forestry and agricultural and oceanography. And now
we're on a quest to build one of the best engineering
schools in the U.S.
It's been done before. We have
benchmarked other schools, like North Carolina State
University and the University of Cal San Diego. And
we're following some of the things that they have done
to build excellence in their programs.
We have a plan. It -- our plan is to
achieve our goal by 2010. And we're making progress
towards that plan. We have a tremendous amount of
momentum right now, and I'll talk to you a little bit
about that.
Part of our plan is to build our Research
Program. And one of the things that we've discovered
is that we have a very strong competitive advantage in
interdisciplinary research. We have received a number
of multimillion dollar grants this year because it's
easy to work across departments here at OSU.
Now I know that from being inside here and
I also know that from observations from colleagues
from Stanford and other places, who have become
familiar with what it's like here at OSU.
The Thermal Hydraulic Program that we have
and you're reviewing today is one of those examples of
a very strong program, a national asset. And that's
why you're here.
I want to talk about our momentum. We
have been growing tremendously in student enrollment
for the past four years. And today we're the twenty-
third largest engineering program in the U.S.
Our Nuclear Engineering Program is also
very strong and highly regarded, both for its
undergraduate program as well as for the research
here.
So that's our growth from a student-body
standpoint.
Since 1999 our research has grown 40
percent. This past year, 27 percent. Inside of that
27 percent are several multimillion dollar contracts,
including today OSU will be a center for tsunami
research. That entails a partnership between Computer
Science and Civil Engineering.
We're not just building the world's
largest and most powerful tsunami simulator, we're
going to put it on the internet. That's an example of
our ability to work across departments.
There are several other awards like that.
We landed $18 million in new grants and contracts this
year.
Our plan calls for an investment of $180
million in public and private funds. So far we've
raised over $50 million in private funds. And the
Legislature closed last Sunday -- a week ago Sunday at
5:00 a.m., the State Legislature, and they
appropriated $30 million to OSU.
That 30 million includes 10 million in
operating expenses and 20 million to help us build a
new engineering facility.
We also have a matching 20 million in
private gifts for that building. So we will be
starting on that project soon.
The other thing that's happened here at
OSU, since we've started our path to build this great
engineering college, the faculty have gotten very
excited and focused, and that's why we're getting the
results.
We also have the backing of the leadership
of private industry, as well as the public sector here
in Oregon -- again, because this is important to
Oregon's future and they see also the importance of it
from a national standpoint.
So we have that leadership backing.
Oregon State is focused on results and a tremendous
amount of momentum.
So today you're going to be reviewing one
of our strongest research programs. And again I hope
you enjoy your stay here at Oregon State. I also hope
that when you depart from your hotel tomorrow morning
you'll see the sun or when you leave here this evening
you'll see the sun because, again, this is unusual for
us.
Again we're pleased to have you here and
thank you very much.
CHAIRMAN WALLIS: Thank you very much.
Well, Jose, are you ready to tell us about
the work that's going on here?
MR. REYES: While that's warming up, I'll
begin by saying welcome also. And I appreciate Dean
Adams' comments. He's given you an overview of the
college. I'm going to give you a little bit of a more
parochial view of our Research Program. I want to
talk about our thermal hydraulics in general, what
types of research we're doing.
It works.
OVERVIEW OF OSU NUCLEAR ENGINEERING
THERMAL HYDRAULICS RESEARCH
MR. REYES: So I'll tell you a little bit
our Advanced Thermal Hydraulic Research Laboratory and
what our program mission is. And this is kind of an
umbrella laboratory that does a lot of
interdisciplinary research, and includes the work that
we're doing of course here and that we'll be
discussing today and PTS and in base separation.
I'll mention who are researchers are and
talk a little bit about each program, so this will be
a fairly quick overview of the research that we're
doing.
Our goal from the beginning was to develop
and maintain world-class research capabilities for
assessing thermal hydraulic behavior in nuclear
reactor systems and components. And that was an
overarching goal for our mission.
We have five primary areas of research
that we have been focusing on: Integral system
research; separate effects/component research;
fundamental phenomena research and model development;
advanced instrumentation; and advanced thermal
hydraulic computer analysis research.
So over the years we've developed in these
different areas. I think we've done quite well in
developing that.
We consist of a team of professional staff
augmented by graduate students and undergraduate
students. And during the academic year we have quite
a few undergraduate students who work with us. We've
had at least 30 or 40 students, undergraduate
students, work with us over the years on APEX-type
design issues.
And today we have a couple of
undergraduate students, Ian Davis and Chris Linrud,
who have graduated and are moving on. They just --
for some reason they just keep graduating and leaving,
I don't know.
And there are a couple of names I haven't
listed there, Dan Wachs, who's helping us out today;
and Ben Ralph, who is with us.
Integral System Research Programs. We'll
be talking about the PTS Program in detail, so I won't
go over that at this point.
We do have some other test facilities that
we're working with on the Integral System. We have
the Multi-Application Small Light Water Reactors.
This is a program we're doing for DOE jointly with
Bechtel, Nexant, and INEEL.
We also have the AP1000 work which has
just recently been funded by a DOE NERI Program.
The Multi-Application Small Light Water
Reactor, which is funded by DOE, is going to be a
one-third scale test facility. This is a
high-pressure facility. It will be operating at about
1500 PSI and at full temperature.
This is a very simple design. The concept
is basically a small -- I won't say portable -- but
relatively small design. We're talking about a
40-foot tall containment section or a reactor section
in a 60-foot reactor vessel or containment vessel. So
it's a vessel within a vessel. A helical tube steam
generator. There's no cold legs, no hot legs, so
nothing to break in terms of loops.
So we're going to be -- we've done the
design jointly with Bechtel and with INEEL. We're now
in the process of constructing this facility. So
probably next time you come you'll see a high-pressure
facility in the bay where we'll be taking you for a
tour later today. So that will be -- we're looking
towards December having that facility operational. So
when we go do the tour we'll talk a little bit more
about that.
So that's one of our Integral System
Tests.
Of course, the AP1000 you're familiar
with. We are looking to do a testing of these new
larger passive safety systems. We hope to do some
design basis accident test and probably beyond design
basis accident test. And this program is scheduled to
begin sometime in September.
CHAIRMAN WALLIS: Well, this will look
very much like your AP600 work, won't it?
MR. REYES: Correct. Correct.
We will be modifying the facility
extensively. We will be -- this includes a brand new
data acquisition system. And a lot of the components
will -- well, several of the components will change
because of the scale.
In the Separate Effects area, today you'll
be seeing the ATLATS facility, which has been used for
the base separation work. We'll also be looking at
the APEX-CE transparent loop, which is what we've done
-- we've used to do some of our visualization of the
mixing in the Palisades geometry cold legs.
Fundamental phenomena. We do quite a bit
of different fundamental phenomena research. Dr. Wu
has certainly had a lot of capability in this area
when he joined us.
Fractal enhancement of transport
processes, bubble shearing-off rate, bubble-bubble
coalescence rates, annular entrainment mechanisms, the
natural frequency of attached bubbles, bubble
condensation in subcooled liquid flows, fractal
measurement of slug flow in vertical test loops.
So there's a lot of -- some of this
research we'll be doing jointly with the Mechanical
Engineering Department.
And the students have set up a whole bay
full of displays for you. So when we go over you'll
be able to see that.
This one in particular, fractal
enhancement of transport processes, again it was
actually kind of a mix of a program: Mechanical
Engineering and Forestry. So we had one of the
engineers in the Forestry Department who was
interested in the way leaves work. And we worked with
Mechanical Engineering. We came up with a design.
We've gotten a provisional patent already
on it, which was issued on the 5th of year 2000. And
then the final patent application was just submitted
here 2001. And we've gotten some interest from Intel
and HP. So we're kind of branching out. And again
it's an example of how we do some interdisciplinary
research here at OSU.
MR. KRESS: Is that just a process to
maximize the surface area of transport?
MR. REYES: Essentially -- that's correct.
Essentially it's an effective way of providing cooling
by maximizing a surface area.
MR. KRESS: Um-hum.
MR. REYES: But we're looking primarily at
an internal cooling process. And so there's a whole
range of products that can come from that.
We're doing fundamental phenomena
research: Two-fluid model improvements, interfacial
area concentration modeling. So we've done some --
you'll see some work later on on the coalescence of
breakage of droplets and bubbles in what we're doing
there.
We're also looking for advanced
instrumentation. You'll see some impedance probe
techniques. We're looking at MRI applications. We've
done some work in the past. That figure on the upper
right-hand side of the screen there is an image of
slug flow. It's an MRI image of slug flow. And so we
find that we're able to get some good resolution of
the images using MRI.
And this was with a one-and-a-half Tesla
magnet. We are looking at working with Argonne, Dan
Wachs is here today, and hopefully using a 9 Tesla
magnet to get much better imaging. So we think we can
get some good imaging of two-phase flows using MRI
techniques. It's completely nonintrusive. And we get
some good pictures.
Neutron and gamma radiography. Dr. Wu
will be talking, showing that a little bit later on
with one of his students.
Imaging processing and some double-sensor
conductivity probes. So for void fraction
measurement, bubble-size measurements, velocity
measurements, we're using that technique here also.
We also are working hard to try to develop
our computer abilities. We have been using RELAP5
Systems Analysis Codes. We have been using the
RELAP5-Gamma. I guess it's mod 3. something, point
something, Gamma, which is the NRC version, the latest
NRC version. And we've used that for the PTS work.
We also have the DOE version of RELAP5-3D,
which we've been using for the multi-application small
light water reactor work. So we're getting familiar
with both, both versions of the code.
We also have the GOTHIC Containment Code.
We're using that for the DOE work.
And then the CFD Code we're using
currently is STAR-CD, and we've used that for the PTS
work.
MR. SCHROCK: What's the origin of that
DOE RELAP5-5 to 3D Code?
MR. REYES: This is from the Idaho
National Lab.
MR. SCHROCK: Idaho?
MR. REYES: Right. Right.
So just in summarizing as an overview,
we'll be talking primarily at this meeting of two of
our programs within our research umbrella in thermal
hydraulics. We will be talking about the phase
separation of Tee work and the pressurized thermal
shock work that we have done.
We do -- we've done a good job, I think,
in developing our integral system capability, not just
the physical machines, but the infrastructure that's
required to operate these complex facilities.
We're doing a good job, I think, in the
separate effects area. We've built -- we're building
model developers. And I like that. So I think Dr. Wu
brings some good skills as far as model development to
our program.
We are, because of the advancements in
computers speed, we are able to run a RELAP5
reasonably well here. STAR-CD is a lot of work. And
we will see some presentations on that later. But
these Computational Fluid Dynamics Codes eat up a lot
of computer time.
CHAIRMAN WALLIS: What's the origin of
STAR-CD?
MR. REYES: That's a good question.
MR. HAUGH: It was built by Adapco. It's
their -- they've developed the code. Their offices
are in New York.
MR. REYES: You need to say your name.
This is Brandon Haugh.
MR. HAUGH: Oh, this is Brandon Haugh.
I'm a graduate student.
STAR-CD was developed by Adapco, which is
a private company. It's a commercially-available CFD
manufacturer.
CHAIRMAN WALLIS: It's like most of the
other CFD Codes?
MR. HAUGH: Yeah. It's pretty -- pretty
much the same, a little more graphical use interface,
just modern software engineering techniques, but
pretty much the same.
MR. REYES: We continue developing in the
area of instrumentation, so we've got fairly creative
work going on. And you'll see some new instruments.
Actually on the ATLATS you'll see a device that was
developed by the students for measuring level, which
is fairly unique.
And then we continue developing, doing our
fundamental model research. That's more in the area
of the fundamentals of two-phase flow, coalescence,
breakage of bubbles, and the transport equation.
So with that I think I'd like to turn it
over then to start talking about the two different
areas of research that we've been working on
primarily, which is the phase separation Tees and then
that will be followed by the pressurized thermal shock
work. And that's the presentation.
CHAIRMAN WALLIS: Thank you very much.
Now we have a break scheduled for around
10:00, so whenever it's natural for the speaker to
break around that time then we'll have a break,
because I noticed that the program just goes on for
two hours. But there's probably a natural break point
in those two hours when we can have a break.
MR. WU: My name is Qiao Wu, Assistant
Professor of OSU. Welcome.
My presentation today will be about phase
separation in tees. And currently the focus of our
project is about the entrainment in a vertical branch
of this on the horizontal main line. And my
presentation is divided into six parts.
The first part, the introduction on the
modeling improvement and the future efforts, will be
given by myself.
And the second part, test facility and
test results, model evaluation, is going to be given
by Mr. Kent Welter.
For the introduction part, before we go to
the details, we would like to show you a cartoon to
see what is the phase separation effect.
CHAIRMAN WALLIS: Now this phase
separation is very dependent on the flow conditions in
the hot leg, is it not?
MR. WU: Yes.
CHAIRMAN WALLIS: And so this is -- so
that rate is dependent? It's very much tied in with
the particular design of AP600, AP1000.
MR. WU: Exactly.
CHAIRMAN WALLIS: It might not be portable
to a different situation.
MR. WU: That's what we're going to
evaluate. And I'll show you why we're doing and --
CHAIRMAN WALLIS: So one might have to be
careful about putting it into NRC Code for some other
situation and using it?
MR. WU: We're going to show you how we
develop a model in the general sense and see how we
can apply it to this situation.
CHAIRMAN WALLIS: Did you do separate
effects tests, too, with other, other end conditions,
and things like that?
MR. WU: No, because we have some data
from Schrock and Smoglie. And we are going to use
that as a general case with simplified or idealized
unit condition, the outlet condition. And for our
purpose --
MR. SCHROCK: I'm having difficulty
hearing you. Could you speak into the mic a little
better?
CHAIRMAN WALLIS: Is there a PA system
here that --
MR. REYES: No.
CHAIRMAN WALLIS: There isn't. So it
doesn't help to speak in --
(Simultaneous talking.)
CHAIRMAN WALLIS: It doesn't work.
MR. KRESS: Just have to speak louder.
CHAIRMAN WALLIS: It doesn't help to speak
into the mic. You just have to speak up then.
MR. WU: We would like to use this
facility to generate the data because we don't have
such data available, and then to evaluate the existing
models. If we can find the efficiency of the models,
then we try to improve it, because we don't know if
it's practicable for this situation or not.
So the entrainment the process basically,
so when the liquid level is below the off-take, you
could still continue the work. Under the process,
it's similar like this, we show the experiment
process.
CHAIRMAN WALLIS: So the experiment
doesn't look quite like the picture?
MR. WU: No. This picture is ideal. That
--
CHAIRMAN WALLIS: So you can develop a
nice model for the picture, but the experiment is very
intermittent?
MR. WU: Yeah.
CHAIRMAN WALLIS: I think the waves depend
upon what's happening at the end of that hot leg.
MR. WU: That's correct.
CHAIRMAN WALLIS: So it's -- you know, if
you did it in a long pipe, it might be quite
different.
MR. WU: That's correct. I show you the
effect of the length.
MR. BESSETTE: And I think our ultimate
intent for us to be generalistic, I mean, so that we
had different end-point conditions, you know, from a
closed and to -- it's a symmetric condition, with
different development lengths as well.
MR. SCHROCK: I'm concerned about the sort
of mixing together of so many different physical
problems in what you've referred to as your database.
I know you have that report which has been sitting
dormant for a couple of years now that summarizes the
data.
I think the concept originally was: Let's
assemble what is known about this problem and lend
some clarification to that. I don't think that data
report did that. But I guess I'm looking forward to
hearing what's being changed in relationship to that
report that's going to guide you.
It seems to me you're pretty far down the
line for it to be in that uncertain state. And what
you've just shown is, as Dr. Wallis has pointed out,
indicating that there are many different
circumstances.
There are distinctly different physical
processes that are important at different times and,
to some degree, dependent upon the geometry.
If I look at the list of cited references
in this database, it covers the waterfront. And many
of them have no relationship whatsoever to the problem
that initiated this research program, which is the
difficulty in calculating ADS flow in AP600.
So I think you need to focus a bit more in
what you're telling us about why you're doing what you
have and are doing and what --
MR. WU: Yeah, the first --
MR. SCHROCK: -- what you know about what
has gone before you, because that's not very clear.
Okay.
MR. WU: Yes. The cartoons show you the
process. And you already pointed out the complexity.
So the introduction. It's obvious, it's very
significant. It's a high-ranked phenomena in the
OSU-CE meeting for the thermal hydraulics and u-sonics
coupled code of development.
And also RELAP5 could not have predicted
the core heat-up in the APEX of the NRC-25 series test
which pinpointed the code deficiency for the vertical
off-take entrainment process.
So for the database for entrainment model,
if we build the database, we found that all these
models were developed for breaks of relatively small
sizes. So there is a need for the new data for the
larger breaks. And it's scaled to prototype
conditions.
And using these data to reevaluate the
existing model, if we can, I think identifies a
deficiency of the model, then we improve it or perhaps
develop the new model.
MR. SCHROCK: So this is, again, a little
unclear to me. What you're saying is there may be
deficiencies in existing correlations that arise as a
result of a lack of experimentation on an adequate
range of geometries.
Now you're going to or are doing tests
with larger ratios of the break diameter to the
main-line diameter. And now you're saying as a result
of what you learned from those experiments, you'll
reassess deficiencies in the previous experimental
results or their correlations.
It's unclear to me how you can accomplish
that unless you redo the experiments using the same
diameter ratios.
MR. WU: We treat your data and Smoglie's
data as one of the targets that we're going to compare
with. And because you already -- already small breaks
of data, or the data, we are not going to repeat it.
So the best look at the database, we don't
have this larger break data. So that's the
motivation. We say, "Well, we're going to do larger
break data," and that --
MR. SCHROCK: I think you missed the
point. The point is that you've said that you're
going to reevaluate model deficiencies in the prior
correlations or experimental results.
You're going to do this on the basis of
data that are collected for larger ratios of diameters
for the break line to the main line.
It's unclear to me how you will assess
anything if it turns out that those correlations do
not scale -- previous correlations do not scale to the
range-of-diameter ratios that you're experimenting
with, that you'll have anything to say about how good
they are for the smaller diameter ratios.
I think that's what I heard you say you're
going to do.
MR. BESSETTE: I think they did put a lot
of effort in to collecting and, you know, digitizing,
and whatever, all the database in order to make sure
that whatever they came up with encompassed or, let's
say, was applicable to the range of conditions of, you
know, the off-take diameter.
MR. WU: I --
MR. BESSETTE: But -- go ahead.
MR. WU: Thank you.
The model deficiency, what I therefore
hear is, say, when the model is being applied to the
larger breaks and that obvious -- and these models
were or correlations were geared with respect to the
small-break data and we're already being evaluated
thoroughly with the existing data.
And what I say are called the deficiencies
is when we forcefully apply these models, correlations
to the larger breaks, what effect and what other kinds
of discrepancies we can get.
I hope I answered your question, sir.
MR. SCHROCK: No, but go ahead.
MR. WU: Okay.
MR. HAN: Can I say something? This is
James Han from NRC.
Let me just add one quick comment. I
thought initially we conducted this research was
because the existing database was not quite sufficient
in the sense that it has a small-branch diameter over
the main pipe diameter ratio. And also the L over D
is different. So I saw that Professor Wu, do you want
to show your review of --
MR. WU: Yes.
MR. HAN: -- of the existing database --
MR. WU: I'm proceeding to there.
MR. HAN: -- to start with?
MR. WU: Um-hum.
For the database we collected, we wrote
letters, emails to the researchers. We collected 20
sets of experiment facility on the data. It ranged
from 1980s to 1993.
All the data and the test facility and the
analyses, preliminary analyses, are being ready in
NRC's report and some first version was submitted --
submitted to NRC for comments.
And some of the data were published in a
product form. We actually digitized, bought us --
purchased -- purchased the software and digitized this
data in an Excel format. So it's convenient for
further analysis.
When we look at the -- focus on the
vertical branch on horizontal main line, the models
were developed in two steps. First is entrainment
onset modeling.
The top figure shows when the liquid level
below certain point, there's no liquid being drawn
into the off-take. That's called the entrainment
onset condition.
And Smoglie, Schrock, and Maciaszek, all
of them related that onset level to a Froude number
based on the off-take gas velocity and the off-take
size.
And the second step, if the level is above
the onset entrainment level, then all of the liquid
had been pulled into the off-take. And Schrock's
correlation, Yonomoto's correlation, and Smoglie's
correlation basically correlate the branch quality to
the real -- the real gas chamber height to the onset
height.
CHAIRMAN WALLIS: This must depend on the
flow in the main tube. And if you have a large liquid
flow in the main tube, I would think you would carry
that wave away. There must be quite a dependence.
And then the gas flow in the main tube is going to --
it has to go over that wave. It's going to do a lot
to its stability, or whatever.
MR. WU: That's -- we found --
CHAIRMAN WALLIS: So the flow rates in the
main tube must have a big -- it can't just be the flow
rate in the branch that matters.
MR. WU: That's what we found in our
experiment.
CHAIRMAN WALLIS: Yeah. That's what you
found?
MR. WU: Yeah.
CHAIRMAN WALLIS: Yeah.
MR. SCHROCK: Let me try to clarify for
you a little better what my problem is.
In your database you have a number of
references, notably the work by Lahey and his
students, that deal with the bubbly flow in the main
line. And the question then of what is the phase
separation as a result of the turning of the flow when
it is a bubbly pattern upstream.
That's totally different physics from what
you've shown a picture of here and totally different
physics than the case of the quiescent stratified
fluid, which is in the proximity of a take-off line
that either sucks in the vapor phase when the break is
submerged under the liquid or it may suck in the
liquid when it's above.
And it's that latter case that you're
addressing here that seemed to be a problem in the
application of the existing correlations in RELAP5 for
the ADS flow in AP600. That's one isolated thing.
But in the broad range of things that are
covered by all of these things that you have here,
there are many different distinct two-phase flow
problems that ought not be confused with the one that
you're addressing here in these experiments. So --
MR. WU: The database --
MR. SCHROCK: -- I'm really puzzled by why
you want to do that.
MR. WU: The database was built for all
type of entrainment, like vapor pull-through, side
branch. And like at the beginning I pointed out
currently our project phase is for the vertical or
entrainment on the horizontal branch.
And the database itself is more generous
and covered all type of branch with phase separation.
And we picked the data for the vertical branch on the
horizontal for this analysis. That's part of our
work.
MR. SCHROCK: Putting together numbers, a
database, for a broad range of different physical
problems is not a service unless you do something
about telling the user of that compilation what
physical problems each set of data addresses.
And there are different problems being
addressed by this collection of prior works.
MR. BESSETTE: Yeah. I guess -- I can
only say that we agree.
MR. SCHROCK: But I hear the story coming
out that we're going -- we're going to reassess
deficiencies in all of this collection of things on
the basis of, --
MR. BAJOREK: It is --
MR. SCHROCK: -- again, one isolated -- I
mean you could say the criticism of all of these
previous things is that they didn't cover the
waterfront. None of them were either funded at such
a level that they could cover the waterfront, nor was
it the intent either of the researchers or the
sponsor. So --
MR. BAJOREK: This is Steve Bajorek.
I think there's two issues that are
involved.
First, there is a lot of data that was
taken, all sorts of different flow regimes, all sorts
of different physics. Many of the conditions in
geometries were nonprototypic of the AP600 or AP1000
design.
MR. SCHROCK: Yeah.
MR. BAJOREK: What Dr. Wu's been doing at
this point is trying to group all of the experimental
data that has taken a look at off-take and Tees.
Now the next step has to be to segregate
that.
MR. SCHROCK: No, the --
MR. BOEHNERT: Only a couple of those
datasets were --
MR. SCHROCK: -- the very first step
should include critical comment about these results.
And that is lacking in this data report. There's just
no --
CHAIRMAN WALLIS: Now you're saying that
because you've read the report and you've seen it.
MR. SCHROCK: I've said it because I've
read the report.
CHAIRMAN WALLIS: Well, I think that
you're making very good points. I think we will
return to them as you make your presentation, because
now we've sort of set the stage --
MR. SCHROCK: Okay.
CHAIRMAN WALLIS: -- for what we're
looking for.
MR. SCHROCK: I'll back off.
CHAIRMAN WALLIS: And we'll see if we find
it.
So I'd like you to continue the
presentation, please.
MR. WU: When we found the data for the
vertical or off-take on the horizontal branch, it's
several sets of data available, like Dr. Schrock
pointed out, as Schrock, and Smoglie, and Anderson's
data.
And when we compare, compare with the
prototype condition, you'll see the prototype
condition for the D over -- the branch size over the
main pipe size is relatively large. It's about .3.
However, all the test data is like .1, around there,
and below .1.
And also the -- for the inlet condition,
inlet length over the main pipe, it's very far from
the inlet. It's about over 20.
So for the real case, the inlet is very
short. So we think it's necessary for the -- from the
horizontal pipe inlet to the branch location.
CHAIRMAN WALLIS: You've got two inlets,
one from the steam generator, one from the reactor
vessel. They're both short.
MR. WU: No, from the offstream. The
inlet side.
CHAIRMAN WALLIS: Well, it could be
flowing both ways -- either way.
MR. WU: Yes. Yes. But the other side is
much longer, so we focused on the shorter side.
CHAIRMAN WALLIS: So whenever you see a
short L over D like that you say the inlet conditions
to the big pipe must be very important?
MR. WU: We would like to look into it.
And for the data sets, again, we covered
all different branches of different orientations. And
we found only two sets of data were used for the
horizontal entrainment to the vertical branch. That
is Smoglie's data and Schrock's data.
And all of them, except Smoglie's, data
goes into a slug flow, but at a very low gas flow
rate. Smoglie's data is in the horizontal flow
regime. This flow regime map is a traditional flow
regime map.
CHAIRMAN WALLIS: So these are fluxes in
the main pipe? Is that what the jgs and jfs are here?
MR. WU: Yes.
CHAIRMAN WALLIS: They're based on the
main pipe?
MR. WU: Yes. Yeah, that's superficial
velocity. That's a traditional flow regime map for
horizontal pipe over two-inch size.
So the conclusion for this introduction
was a correlation --
CHAIRMAN WALLIS: Excuse me. You didn't
show AP600 on this plot?
MR. WU: It's in the prototype because
originally, when I prepared the transparency, I
decided this is sensitive information, so I put a big
spot there, about there so you can at least tell
exactly what the number is. It's .3 for the off-take
size to the main pipe.
So the correlation is based primarily on
this stratified flow data. And branch size is
relatively small compared to the prototype off-take.
And the inlet and the downstream conditions were
simplified.
So for the objective of this project, the
database construction and the design gave the facility
for experiment, the investigation, conduct the test to
generate the onset entrainment of the data and the
entrainment to read the data.
And then using this data to evaluate the
existing model correlation, see, for it's -- they are
able to be applied for this situation.
And if they can, that's good news, and we
don't need to go do further work. If there is room
for improvement, then we're going to do model or
correlation improvement or development.
That's our logic of this project.
So for the second part and to the fifth
part I'm going to hand to Kent Welter, and I'm going
to come back to talk about the model improvement.
MR. WELTER: Thank you, sir.
My name is Kent Welter. I'm a Ph.D.
candidate in the Department of Nuclear Engineering.
Before I go on to this, I'd like to take
one minute and address Dr. Schrock's question since I
am, I guess, the person who wrote the database that
he's speaking about.
When we first constructed the database we
had several things in mind. And I've considered it as
two parts.
The last half of the database which
actually contains the experimental descriptions,
facility descriptions, and data uncertainties, is a
collection of phase separation experiments. And those
are very different.
Through them I've reviewed all the papers
by Saba and Lahey and their models. They consider,
you know, a full set of Navier-Stokes equations. It's
different than what we are looking at here.
The phase separation is a larger set, so
you can consider liquid entrainment as a subset of
phase separation.
The first three chapters of the database
is adding an analysis as applicable only to the AP600
prototypic conditions.
The second half is a database that could
be used more generally. It could be used for
different applications as a starting ground.
If I want to know about phase separations,
there's no one place to go. And that was the purpose
of the second half of the database, which is why it
includes a large collection of stuff. The first half
includes analysis only pertaining to the AP600 and
liquid entrainment of vertical branch.
MR. SCHROCK: Well, what is the status of
that report? The one that I've read --
MR. WELTER: The one that you've read --
MR. SCHROCK: -- has been said to be two
years old and --
MR. WELTER: Exactly.
MR. SCHROCK: -- I don't know the extent
to which you've --
MR. WELTER: Revised that, --
MR. SCHROCK: -- made a revision on it, --
MR. WELTER: -- correct, sir.
MR. SCHROCK: -- but it had serious flaws
in it.
MR. WELTER: It did, sir.
I wrote that report three years ago and
about three months when I came into the program here.
And it was mostly a collection, a collection of
basically leave-no-rock-unturned. Okay.
I looked through everything, found
everything, looked at it, reviewed it, and that's what
I submitted. And for the last two years, as I've
increased my research, I've realized, well, that's not
what I want to submit. We've revised it.
The revision that we've now sent the NRC
has condensed the experiments into the last section.
And we've added several chapters on the analysis.
MR. SCHROCK: And so that exists?
MR. WELTER: Yes.
MR. SCHROCK: And when was it submitted to
NRC?
MR. WELTER: That was submitted to the NRC
when we submitted our ACRS stuff several weeks ago.
MR. REYES: That batch is still under
review.
MR. WELTER: And it's still under review
actually. So I am -- I am hoping that that will help
clear up a lot of discrepancies that the first one
saw. And you are correct --
CHAIRMAN WALLIS: This is under review by
the NRC?
MR. WELTER: No, not yet.
CHAIRMAN WALLIS: Not yet.
MR. REYES: It's being reviewed now.
MR. WELTER: It's being reviewed now.
MR. SCHROCK: Did NRC ever review the
draft that you've had for two or three years?
MR. WELTER: I received one comment back,
--
MR. SCHROCK: I'm asking the NRC that.
MR. WELTER: -- several comments back on
it. And then we revised it. It's still in the
original process.
MR. SCHROCK: I didn't hear your answer,
David.
MR. WELTER: Oh, I'm sorry.
MR. BESSETTE: The answer -- the answer is
yes, but we most -- mostly we relied upon your review
of it for --
CHAIRMAN WALLIS: What you want to avoid
is a situation we sometimes get where all this stuff
goes through and the NRC thinks it's great and it's
the basis of a rule. Then it comes up to the ACRS and
we don't like something about the whole basis of the
analysis. That's too late in the process to have much
influence.
MR. BESSETTE: Well, in this case you saw
the first draft.
MR. SCHROCK: Well, it was never reviewed
by the ACRS, as far as I know. I saw it in February
19- -- or 2001, which is very late in the game. I
don't know how...
MR. KRESS: Well, we only get into the
picture and if there are intentions to use it for
basically decisionmaking or rulemaking. And then we
look at the basis for that, but I mean we wouldn't
review a document like that just to review it.
CHAIRMAN WALLIS: No. No. That is part
of the problem, --
MR. KRESS: Yeah.
CHAIRMAN WALLIS: -- is that we don't see
it until it becomes important. By the time that
happens it may be too late to do anything about it.
MR. KRESS: Yeah. Well, I --
MR. BESSETTE: In this case you did -- you
did see an early draft, or it was distributed at
least. The early draft was distributed.
CHAIRMAN WALLIS: But it doesn't mean to
say that we worked on it. We work on it when it's
part of our schedule to work on it.
MR. BESSETTE: And -- and --
CHAIRMAN WALLIS: We're not your
reviewers.
MR. BESSETTE: No. And we don't count on
you as being the official reviewers.
CHAIRMAN WALLIS: Well, again we should
probably proceed with the presentation.
MR. WELTER: Okay. Thank you, sir.
CHAIRMAN WALLIS: And maybe we'll be
acting as reviewers today.
MR. WELTER: Thanks.
I'd like to talk, take off where Dr. Wu
left off, and speak about the scaling involved when we
develop our separate effects test facility.
It includes considering the hot-leg flow
condition using flow transition criterion developed by
Zuber. And if we determine the superficial gas
velocity in the main line, we can get an appropriate
HL over D, or a hot-leg liquid level to hot-leg
diameter.
To preserve the geometry of the AP600, we
also considered the main-line diameters of the hot leg
and of the vessel and also of the inlet from the
reactor vessel to the branch over the main-line
diameter.
MR. SCHROCK: So the problem that you're
scaling is a quiescent stratified flow; is that
correct?
MR. WELTER: For Zuber's flow condition,
that's correct.
MR. SCHROCK: Oh, I'm not talking about
Zuber --
MR. WELTER: I'm sorry.
MR. SCHROCK: -- or anybody else. I'm
talking about the problem that you are presenting
scaling analysis for. You have to define your
problem. Your problem is stratified quiescent.
CHAIRMAN WALLIS: By "quiescent," you mean
it doesn't have big waves --
MR. SCHROCK: It doesn't have waves on it.
MR. WELTER: Sir, --
MR. SCHROCK: And the picture you showed
us a few moments ago with waves is a different
problem. So --
MR. WU: No, that's not -- not what we --
MR. SCHROCK: Not true?
MR. WU: This flow regime is like a -- you
can say, well, we preserve the flow regime phenomena
by guaranteeing the Froude number on the left side is
the same. So whatever you -- your run, you say from
a stratified to slug or stratified to annular, if you
keep your Froude number the same as is this prototype
condition, then you preserve the phenomena of flow
regimes. So we didn't say we keep that Froude number
as under the transition line. That's what you refer
to the quiescent stratified flow.
CHAIRMAN WALLIS: Now I'm trying to think
here. The liquid -- is the liquid actually flowing up
into the steam generator?
MR. WELTER: In through the steam
generator?
CHAIRMAN WALLIS: Is this flow going out
of the reactor up into the steam generator?
MR. WELTER: To the lower plenum, but it
does not make the loop.
CHAIRMAN WALLIS: It doesn't make it? So
any liquid which comes in the pipe has to go out the
break?
MR. WELTER: Exactly.
CHAIRMAN WALLIS: Nowhere else to go. So
this -- all the entrainment a hundred percent.
MR. WELTER: In terms of injection flow,
that's correct. The correlations are developed on
determining level in the hot leg and how that relates
to the entrainment rate.
CHAIRMAN WALLIS: So you've got down to
the point where the level is so low that there's no
entrainment going up into the steam generator?
MR. WELTER: We've reached that point in
experiments, yes.
MR. WU: It's dry.
MR. WELTER: It's dry.
CHAIRMAN WALLIS: So that's just the exit
condition? You have to say something about the exit
condition, which --
MR. WU: Yes.
CHAIRMAN WALLIS: All right. And so this
is one where there's no way in which liquid can get
carried out at the end --
MR. WELTER: That's on my next slide, --
CHAIRMAN WALLIS: -- into the steam
generator --
MR. WELTER: -- which is the onset
criterion we used.
One of the things that we also considered
was the inlet flow condition. And we used the void
fraction from the vessel to properly scale the decay
heat.
CHAIRMAN WALLIS: Now what does that mean?
I didn't quite understand alpha vessel.
MR. WELTER: Alpha vessel. You have
decay, and there's of course boiling in the core. And
we wanted to make sure that we had the appropriate
void fraction from the AP600 going into the hot leg so
that we preserve the inlet condition from the reactor
vessel.
CHAIRMAN WALLIS: So you're going -- this
is a bubbly flow sort of thing in the vessel that's
going to be the same?
MR. WELTER: A bubbly flow that, as the
fluid is draining, will be going into the hot leg, or
when you're -- sir?
CHAIRMAN WALLIS: Okay.
MR. WELTER: Yes.
MR. WU: Basically if we run this test,
the different void fraction in the vessel, then we'll
get a different -- again from the vessel to the hot
leg inlet, there is a phase of separation. So if you
don't guarantee the void fraction's the same, then you
get a different level in the --
CHAIRMAN WALLIS: So you have to maintain
the vessel geometry?
MR. WELTER: Correct.
CHAIRMAN WALLIS: Correct. Not just L
over D hot leg. You've got to have D vessel.
MR. WELTER: We also maintain the diameter
of the vessel --
CHAIRMAN WALLIS: Yeah, okay. Okay.
MR. WELTER: -- on the right of their D
vessel --
CHAIRMAN WALLIS: Are there all kinds of
internals in the vessel? Are there internals in the
vessel that --
MR. WELTER: There are no reactor
terminals on top of the vessel, no.
CHAIRMAN WALLIS: But there are in
reality?
MR. WELTER: There are in reality, that's
correct.
To preserve the onset criterion, which
would make sure that we are at the correct flow rates
at which entrainment begins, we use the onset of
liquid entrainment developed by Zuber, Smoglie, and
Schrock, where if we know the gas velocity in the
branch, then that will give us the onset of liquid
entrainment height, hb. So it's a ratio of
gravitational to inertial forces. If the inertial
force is greater than the gravitational force, onset
will begin.
CHAIRMAN WALLIS: Now hb is something you
have to calculate?
MR. WELTER: hb is onset of gas
entrainment height. So it's a gas chamber height.
It's the opposite of liquid level.
CHAIRMAN WALLIS: It's not an independent
variable? It depends on all the other things you're
doing?
MR. WELTER: That's correct. It depends
on -- in this sense right here, it depends on the gas
velocity in the branch.
CHAIRMAN WALLIS: Well, it depends on the
amount of liquid. Again, I get back to what -- how
much --
MR. WELTER: In the onset entrainment
experiments there's no liquid injection flow. So you
can consider it as a small pool in the hot leg; that's
correct.
CHAIRMAN WALLIS: So it's just -- there's
no liquid flow, okay. So you can control it simply by
the --
MR. WELTER: Yes.
CHAIRMAN WALLIS: -- the void fraction in
the vessel, I guess. If you bubble through the vessel
you raise the level of everything, including hb?
MR. WELTER: That's correct. Which gives
the effect of the inlet.
From these scaling parameters we
constructed an integral -- I mean a separate effects
test facility. I'd like to go through and explain the
different components.
The critical complaints of the test
section is a clear PVC with a horizontal hot leg, the
vertical branch. We have a stainless steel reactor
vessel. We have a steam generator connected to the
downstream, which is appropriately scaled for the
friction factor.
We have clear tigon tubing, and you'll see
these in the tour also, for the steam generator tubes.
Injection flow is provided by a water pump
from a large 5,000-gallon water tank. Injection flow
for the air is provided by an air compressor that goes
through a 100-PSI air receiver.
We have 25 channels that record
temperature, pressure, flow, and catch tank max, along
with the mixture level and the hot-leg level. And I'm
going to introduce later how we determine the hot-leg
level using instrumentation.
CHAIRMAN WALLIS: Now where does the air
go in this experiment? It comes in through the bottom
of the vessel?
MR. WELTER: That's correct.
CHAIRMAN WALLIS: And some of it goes out
the water tank, I guess?
MR. WELTER: The air goes -- I was just
going to -- the next slide shows a cut-away of the
reactor vessel, which talks a little bit more how the
air comes in.
CHAIRMAN WALLIS: Well, I'm ahead of you
then, I guess.
MR. WELTER: Yeah. So if we take a closer
look, this is a cut-away of the inside of the reactor
vessel. It's approximately a one-quarter length scale
compared to the AP600. It has air, water, and
atmospheric temperature and pressure.
We use seven porous tubes in a shown
configuration to simulate decay heat boiling. Air
flows through the bottom of this reactor vessel and
through the porous tubes.
CHAIRMAN WALLIS: Well, I guess I was
looking at the previous figure. The air has to decide
whether it's going to go out through the catch tank
and the drain line or to go out through the steam
generator.
MR. WELTER: Exactly.
CHAIRMAN WALLIS: I don't quite see where
it goes when it gets out through the steam generator.
It's not clear to me there's any outlet from the steam
generator for air.
MR. WELTER: Everything goes through the
catch tank. So air coming from the steam generator --
CHAIRMAN WALLIS: Everything has to go
that way?
MR. WELTER: -- is going to come this way.
CHAIRMAN WALLIS: So nothing goes out
through the steam generator?
MR. WELTER: We have a return line, a
one-inch return line to equalize the pressure on the
reactor head. And so --
CHAIRMAN WALLIS: But all the air that
comes through the reactor goes out the catch tank?
MR. WELTER: Well, yes.
CHAIRMAN WALLIS: That's not reality,
though, is it?
MR. BESSETTE: Well, yes.
CHAIRMAN WALLIS: Well, it's a very
limited reality, isn't it?
MR. BAJOREK: Sort of. Yeah, it's kind of
real. Where else can it go? I mean there's only one
opening.
CHAIRMAN WALLIS: The steam generator --
there is no flow-through steam generator in the
accident?
MR. KRESS: As long as you don't build up
much back pressure in that tank.
MR. SCHROCK: It goes the opposite way.
MR. KRESS: It's like --
MR. WELTER: There's flow through the
steam generator, but it's backwards.
MR. BESSETTE: Generally speaking, it can
only go out the one place.
MR. BAJOREK: It's going in the opposite
direction here in the test. In the AP600 and for the
ADS-4 it was split through the intact loops, go
through the steam generator, and you would have a gas
velocity approaching the branch line from both sides.
CHAIRMAN WALLIS: From both sides, right.
MR. WELTER: From both sides. Yeah, it's
going the opposite way.
MR. SCHROCK: See, this doesn't look like
a clean-cut separate effects experiment. I had
thought that that was the objective for the OSU test,
but --
MR. WELTER: We varied the downstream --
I'm sorry, sir.
MR. SCHROCK: -- what you have is a system
here which is not representative of any reactor system
that I know of. And so I don't know what the value of
the results would be as a systems test. But as a
separate effects test, it misses the mark.
You have a variety of conditions entering
the test section that result from system effects. If
you look at the KFK experiments, the Berkeley
experiments, which I think are the main database for
the phenomenon with a quiescent interface, what you
see is that in those separate effects experiments
pains were taken to smooth the flow, to ensure that
there would be a smooth stratified flow.
What was sought was the conditions for the
onset of entrainment and then the phase distribution
following the onset of entrainment, those factors, for
that specific condition at the onset of entrainment.
Your system has these system effects,
which are atypical. And I don't understand then how
clarity is going to be brought to the problem if
separate effects are addressed via some kind of
randomly-put-together systems.
MR. BESSETTE: Well, I think --
MR. SCHROCK: It just doesn't follow.
MR. BESSETTE: I think the system -- I
mean, in fact, one of our objectives was to include
some system effects because, as he's just pointed out,
the system effects are -- see, you're correct that
that Zuber -- the initial Zuber formulation was for
smooth stratified flow, smooth stratified conditions
--
MR. SCHROCK: David, we're talking about
the distinction between separate effects and system
effects, okay.
MR. BESSETTE: Yes. But --
MR. SCHROCK: And the code has separate
effects models in it.
MR. BESSETTE: Yes.
MR. SCHROCK: And what you set out to do
is to improve on the code's separate effects models so
it could calculate AP600 ADS flows better. Isn't that
where we started?
MR. BESSETTE: That's -- yes. But what we
have seen at least is that the system effects are at
least as important to the --
MR. SCHROCK: Of course they are.
MR. BESSETTE: -- or more important --
MR. SCHROCK: If you have a wavy flow, as
the previous --
MR. BESSETTE: Yeah.
MR. SCHROCK: -- cartoon showed us with
animation here, it had a tremendous effect --
MR. BESSETTE: Yeah.
MR. SCHROCK: -- on the result that you
get. It would be naive to believe that it wouldn't.
But that's a different situation.
MR. BESSETTE: But that's what we wanted
to include. We wanted to include the --
MR. SCHROCK: Well, then do it in such a
way that you have control over what the upstream phase
distribution is. And --
MR. BESSETTE: But --
MR. SCHROCK: Your code is going to have
to know --
MR. BESSETTE: That's correct.
MR. SCHROCK: -- what the upstream phase
distribution is --
MR. BESSETTE: That's correct.
MR. SCHROCK: -- in order to properly
calculate what the branch flow rate.
MR. BESSETTE: But what we have seen is
that the code has to be able to calculate the flow
regimes in the upper plenum in order to calculate the
correct conditions in the hot leg.
So the code has -- the code has to be --
you have to back up. It's both in the experiments and
in the code. You see that you have to get the upper
plenum conditions correct in order to get the right
conditions in the hot leg.
So you have to feed the right flow from
the upper plenum to the hot leg. And we've seen that
both in the code calculations and in the data.
CHAIRMAN WALLIS: I think we'll have to
accept this as not the totally separate effects test,
not at all, but it seems to the --
MR. SCHROCK: Well, I think --
CHAIRMAN WALLIS: -- hybrid separate
effects, system effects --
MR. SCHROCK: -- it's a basic problem in
the thinking of how you can improve what the codes are
doing.
The codes attempt to calculate, using
correlations, for a wide range of different physical
phenomena. And unless you have adequate flow regime
characterization, you can't begin to come up with a
set of correlations that are going to correctly
calculate --
CHAIRMAN WALLIS: Well, the version --
MR. WELTER: But that's correct.
MR. SCHROCK: -- those flows.
MR. BESSETTE: Yeah, I think we agree. It
seems to us that the conditions, the model for smooth
stratified flow is quite -- quite transparent, let's
say, and adequate, good. There's nothing you could
improve upon. So --
MR. SCHROCK: No, and I don't think that
was your initial argument for starting this program.
It was that the choice of diameter ratios
made in those earlier experiments, before there was
any knowledge of what the AP600 geometry was going to
look like, didn't envision that there would be need
for data with such large -- large break diameters.
MR. BESSETTE: Yes. That was one of the
motivations, yes. Yes.
MR. SCHROCK: Right. All right.
MR. BESSETTE: But there was no -- we did
not see any obvious problems with the stratified flow
off-take modeling, other than the range -- the
diameter ratio.
The other thing was that we did not
believe that it adequately covered these conditions of
stratified -- wavy flow and --
MR. SCHROCK: Well, it doesn't. And you
need -- you need experiments for wavy flow, --
MR. BESSETTE: Yeah.
MR. SCHROCK: -- if that occurs in the
real reactor systems.
MR. BESSETTE: Yeah. And that was, of
course, one of the motivations.
CHAIRMAN WALLIS: Well, I think we have to
see the whole presentation and then maybe come back to
these issues in a discussion later.
MR. WELTER: Thank you, Mr. Chairman.
CHAIRMAN WALLIS: But --
MR. SCHROCK: I think the objective --
CHAIRMAN WALLIS: -- these questions are
going to emerge later --
MR. SCHROCK: -- needs to be set out more
clearly in the beginning, Mr. Chairman.
CHAIRMAN WALLIS: But if we spend all the
time on the prologue we'll never see the play, so I
think we have to go on.
MR. SCHROCK: All right.
MR. WELTER: Thank you, sir.
The test facility, let's take a closer
look at the test section geometry used. This is the
PVC test section.
Two test sections were constructed to
enable three different inlet lengths for testing.
They were constructed by welding two PVC pipes
together. This enabled us to save a tremendous cost
on test section. Each one of these test sections is
approximately $150 plus parts and labor, compared to
casting acrylic which costs 4,000 to $7,000.
CHAIRMAN WALLIS: You weld PVC?
MR. WELTER: Weld PVC. The investment was
a $400 PVC welder.
CHAIRMAN WALLIS: It doesn't electric use
arcs, though, does it?
MR. WELTER: Not usually, no.
MR. ROSENTHAL: I'm sorry. I know it's a
divergence.
MR. WELTER: Please.
MR. ROSENTHAL: I'm just curious. I'm
used to gluing PVC together. So what is a PVC welder?
MR. WELTER: John.
MR. GROOME: My name is John Groome. It's
basically used on a hot air gun.
THE REPORTER: Would you come to a mic,
please?
MR. GROOME: Good morning. My name is
John Groome.
And on the question of welding PVC,
basically you use a hot air gun. And you have a
filler rod. And you actually melt PVC to do the
welds. So it's kind of like tape welding PVC, but you
don't -- you don't actually melt the base material.
And you'll see some examples today of that when you
look at the test sections.
CHAIRMAN WALLIS: It's like mending many
holes on the base of your skis.
(Laughter.)
MR. GROOME: I couldn't tell you anything
about skis, but --
MR. ROSENTHAL: Says the man from New
Hampshire.
MR. WELTER: Thank you.
I'd like to take a moment to introduce the
hot-leg measurement instrumentation. It is a
half-ring-type conductivity probe. In this
illustration there are two stainless steel semicircle
wires placed within a PVC ring.
This PVC ring is then bolted between two
flanges, the hot leg. There are two of these, these
types of instrumentation: one on the reactor side of
the test section to give -- measure inlet hot-leg
level, and one on the steam generator side to measure
out-leg hot-leg level.
These wires are connected to signal
conditioning. We have a 100-kilohertz sine wave
oscillator. We use AC power to make sure there's no
iron migration that you'd encounter with DC.
It goes through a current driver that's
amplified, rectified, and then filtered. And then we
go ahead and measure the voltage. And --
CHAIRMAN WALLIS: This is just -- this is
just conductivity, --
MR. WELTER: A half-ring type --
CHAIRMAN WALLIS: -- or is it impedance?
Is it --
MR. WELTER: It's an impedance probe.
CHAIRMAN WALLIS: It says "conductivity,"
but -- so it measures actually --
MR. WELTER: Impedance of the air and
water, basically air.
CHAIRMAN WALLIS: It measures capacitance,
or does it measure...
MR. WELTER: Resistance -- impedance.
MR. WU: Impedance.
MR. WELTER: Yes.
CHAIRMAN WALLIS: So you could control the
chemistry of the water pretty closely to measure
resistance?
MR. WELTER: We calibrate -- we calibrate
the test section every test series --
CHAIRMAN WALLIS: Every day, okay.
MR. WELTER: -- to account for the
impurities in the water.
CHAIRMAN WALLIS: So is it mostly
resistance, or is it mostly -- or it's a hybrid of
some sort?
MR. WU: Mostly --
CHAIRMAN WALLIS: Mostly resistance?
MR. WU: The AC current -- this is Qiao
Wu.
The AC current put inside just want to
avoid as the iron accumulated to one electrode, so
cause kinds of drifting. But we didn't raise to that
high frequency. Run just the probe, we -- in the
capacitance mode.
MR. WELTER: This -- this work was done by
the Department of Oceanography to measure wave height
in their wave pools. And we've borrowed it, their
circuit, and modified it for our case.
So before each test a calibration curve is
run. That's a picture of the calibration curve.
There's an output voltage on the bottom. And we
calibrate with the DP we have in the reactor vessel.
So I flood and then drain the reactor
vessel and get output versus height in the hot leg.
And I run this before each test series.
CHAIRMAN WALLIS: So during the test
you've got this continuous output from this probe and
it shows waves and things?
MR. WELTER: At a one-second scan rate.
CHAIRMAN WALLIS: A one-second scan rate,
so it doesn't show waves?
MR. WELTER: There is some oscillation
involved.
MR. SCHROCK: The calibration curve is
done with static conditions?
MR. WELTER: That's correct. No gas flow,
just water draining and filling.
MR. SCHROCK: Oh, the water is moving?
MR. WELTER: That's correct.
Okay. So I went and I described the test
facility. The --
CHAIRMAN WALLIS: Excuse me. Where --
MR. WELTER: I'm sorry. Yes.
CHAIRMAN WALLIS: Where is the probe in
the circuit?
MR. WELTER: There are two -- yes, there
are two probes. One is on the reactor side. I'll
back up a bit.
CHAIRMAN WALLIS: Are they both ends of
the test section?
MR. WELTER: Yes.
CHAIRMAN WALLIS: Okay.
MR. WELTER: That's correct.
Can you back up about three slides on the
test facility?
(Comments off the record.)
MR. WELTER: This is break time; is that
what you wanted?
CHAIRMAN WALLIS: No, you're going to
finish before break time.
MR. WELTER: Okay. This is --
CHAIRMAN WALLIS: No, wait a minute.
You're going to finish this topic.
MR. WELTER: Yeah, okay. That will be
pretty fast.
Here are both of the impedance probes.
One is on the inlet side and the other one is on the
outlet side. That's correct.
CHAIRMAN WALLIS: So are you finished
describing the facility?
MR. WELTER: That's correct. I'll be
moving on to the tests, results.
CHAIRMAN WALLIS: Do you want to take a
break now?
MR. WELTER: Great.
CHAIRMAN WALLIS: Does that allow us time
to finish the rest of the --
MR. WELTER: Yes, I believe so, enough
time, yes.
CHAIRMAN WALLIS: You're going to do
sections 3 and 4 this morning, or you're going to do
all of this. Okay. Let us take a break for 15
minutes.
(Recess taken from 9:54 a.m. to 10:10
a.m.)
CHAIRMAN WALLIS: Go ahead.
MR. WELTER: Okay. Thank you.
Just before the break I finished speaking
about the test facility and instrumentation. I'd like
to go on and describe the two groups of tests that we
ran, one for the onset of entrainment, determining
that; and then one for determining the rate of
entrainment through the AF 4 line, or the off-take.
The first will be the onset of
entrainment. I'd like to describe the test procedure
that we went through to achieve the onset of
entrainment.
At first, from this figure, the hot leg is
flooded. Then gas is throttled to a specified flow
rate at constant value. And when that happens, from
this animation, entrainment will begin, and you will
lose primary inventory and the level in the hot leg
will drop.
CHAIRMAN WALLIS: Well, it's not
entrainment to start with. It's just flowing out,
because it has to go somewhere.
MR. WELTER: There's no liquid injection.
CHAIRMAN WALLIS: Yeah, but there's gas.
MR. WELTER: Which is pulling the liquid
with it.
CHAIRMAN WALLIS: It displaces the liquid.
MR. WELTER: There's pulling the liquid up
through the vertical branch.
CHAIRMAN WALLIS: Well, in the first
picture it was all full of liquid, so...
MR. WELTER: It was flooded initially. So
initially -- so we get the same -- the accurate -- we
start at the same spot every time, a flooded hot leg.
After a certain amount of time liquid
entrainment will stop, and basically there will be
only gas flowing through the ADS-4 line and you will
receive a constant level in the hot leg. This is the
point at which onset of entrainment begins.
And we go ahead and run a test series for
different gas flow rates to get the liquid level for
each gas flow rate.
CHAIRMAN WALLIS: Now in reality there
might be some boiling in the steam generator because
this secondary is a heat source and things. There's
all kinds of scenarios where things happen in the
steam generator.
MR. WELTER: That's correct.
CHAIRMAN WALLIS: Here it's just a deadend
for you.
MR. WELTER: That's correct.
And in this figure, it's not shown, there
is actually an air line connected to the bottom of the
porous tubes.
The test scope of the onset of entrainment
includes determining the effect of the inlet length.
We want to know what the effect of the inlet length
in regards to the effect of the reactor vessel and the
void fraction that we scaled.
Also the effect of the steam generator.
We have a valve on the three-inch cold-leg return that
can be opened and closed so that we can simulate. A
close would simulate a filled loop seal in the cold
leg.
CHAIRMAN WALLIS: There's a challenge here
of oscillations imposed between the cold leg and the
react- --
MR. WELTER: The oscillations occur in the
entrainment rate tests with a steady injection flow.
No oscillations occur for the onset of entrainment
with zero injection flow.
CHAIRMAN WALLIS: Nothing's happening?
MR. WELTER: Nothing's happening.
MR. SCHROCK: Is --
MR. WELTER: Sir?
MR. SCHROCK: Is the condition with a
voided steam generator and dome but a flooded hot leg
a condition that's calculated for AP600?
MR. WELTER: It's part of the -- sir.
Sir, go ahead.
MR. BESSETTE: It's -- of course, the
situation when ADS-4 opens is that the generator is
voided and the hot leg is full.
MR. SCHROCK: What's the condition of the
steam dome -- or of the vessel?
MR. BESSETTE: The vessel is -- when ADS-4
opens the vessel is filled to about the top of the hot
leg.
MR. WELTER: May I go on?
Okay. Thank you.
We also are curious to the effect not only
of the steam generator and its influence, mainly the
lower plenum, but also the effect of gas flow
direction. Meaning that if we open the three-inch
cold-line return, how much -- what is the effect of
gas flowing through both the cold leg and the hot leg
so that you get gas from both directions up through
the branch, not just a flow from a single direction.
Flow through. The cold leg is -- is
smaller than the hot leg, of course, because it has to
travel all the way through the cold leg into the other
side of the steam generator. So the majority of flow
is still from the hot leg.
Brandon, can you have the next one for me?
Okay.
Some test results. This is a plot that we
have used the same convention that Zuber used to
classify his onset data. It is a flow regime map with
the Froude number in the main line based upon the
superficial gas velocity in the main line.
I've plotted against hl, which is the
liquid level in the hot leg over the main line
diameter. The regions you see here are in the bottom
left are stratified, plug, and slug, and annular, and
dispersed.
Our onset data for this case, which we're
trying to determine the effect of the inlet condition,
falls within the stratified flow regime.
The different test series and the
different dots are for different L over Ds, from 2.71
up to 4.75. We can see that there is not a
significant impact due to hot-leg inlet length on the
onset of entrainment level. So this is a case where
we have taken into effect the inlet conditions.
Effects of the steam generator. We are
interested in knowing what happens when we basically
put a static pressure boundary on the exit. So we
have col- -- the previous onset data, there was no
three-inch cold-line return.
In this case we opened the three-inch cold
line, cold-line return. And the new -- the data for
with the steam generator has brought the onset level
to a slug transition line. So --
CHAIRMAN WALLIS: So you're saying that
the onset of entrainment is the same as the onset of
slug flow?
MR. WU: Yeah.
CHAIRMAN WALLIS: So it's nothing to do
with droplet entrainment. It's the formation of the
big wave and --
MR. WU: I think physically later you will
see it's the same argument. And the surprise --
that's not a surprise. But the surprise is you don't
need to -- so this coefficient that you get is exact
in the line for the larger breaks.
CHAIRMAN WALLIS: So what is the steam
generator doing to make the data different when it's
attached?
MR. WU: It's --
MR. WELTER: Flow direction. This is the
effect of the steam generator. One of the things is
that there is flow coming from the cold leg. So
basically you're not just flowing past, you're flowing
from both.
CHAIRMAN WALLIS: Flowing air?
MR. WELTER: Air, that's correct, sir.
There's no injection.
MR. SHACK: It would be more accurate to
say loop seal or no loop seal, right?
MR. WELTER: Exactly. One, one is where
the loop seal is filled and the other is where the
loop seal is blown out.
CHAIRMAN WALLIS: I don't understand how
the air gets around the other side in your experiment.
MR. WELTER: The cold leg is above the hot
leg, and so air flowing from the reactor vessel can go
either to the cold leg or the hot leg since the --
CHAIRMAN WALLIS: In your experiment?
Your experiment has a cold leg down the bottom of the
--
MR. WELTER: Oh, I'm sorry. The cold leg
in the AP600 is above the hot leg. In the
illustration I have connected it just back to the
reactor vessel.
CHAIRMAN WALLIS: The illustration, oh, is
wrong then?
MR. WELTER: That's correct, sir.
CHAIRMAN WALLIS: Oh, it's probably
misleading.
MR. WELTER: Okay. Thank you.
CHAIRMAN WALLIS: Okay. So now really the
cold leg is attached to the proper place?
MR. WELTER: The cold leg is attached
above the hot leg.
CHAIRMAN WALLIS: But it's drawn below the
hot leg in the diagram.
MR. WELTER: Another figure. That's
correct.
CHAIRMAN WALLIS: Gee whiz.
MR. WELTER: That is misleading. I
apologize for that.
CHAIRMAN WALLIS: Okay.
MR. SCHROCK: So is the wavy condition
from one set of these data and not the other, is that
the distinction?
CHAIRMAN WALLIS: No, it's coming from
both directions. What's the J in the Froude number?
MR. WELTER: The J we used is the inlet.
CHAIRMAN WALLIS: The inlet.
MR. WELTER: That's correct.
MR. WU: Now the J is totally calculated
from the branch of the take, and we consider it's only
coming from one side for this, for regime transition.
MR. WELTER: It's only coming from one
side, okay.
CHAIRMAN WALLIS: So this Wallis
transition is sort of --
MR. WU: For the one side.
CHAIRMAN WALLIS: -- entirely fortuitous
because this fellow Wallis, --
(Laughter.)
CHAIRMAN WALLIS: -- whenever it was, over
30 years ago, didn't have the prescience to realize
that you were going to connect a cold leg at the other
end of the pipe.
MR. WU: Yeah, but that's the surprise
here, you see. That can use a --
CHAIRMAN WALLIS: This is invocation of a
correlation which doesn't really apply to the
situation.
MR. WELTER: That's correct.
CHAIRMAN WALLIS: So, well, --
MR. WELTER: An interesting fact.
CHAIRMAN WALLIS: The impression, though,
that you get some authority by quoting this fellow,
but I'm not sure --
(Laughter.)
CHAIRMAN WALLIS: I'm not sure that's
true.
MR. WU: No. It's...
(Laughter.)
MR. SCHROCK: I'm still trying to
understand what you're saying about the two sets of
data. I don't understand. Is it --
MR. WELTER: The first set of data --
MR. SCHROCK: Is it that there are waves
on the surface in one case and not the other case, or
--
MR. WELTER: What does the flow regime
look like when I look at my experiments; is that
correct?
MR. WU: Quiet.
MR. WELTER: Quiet. They're both calm.
So the effect you're seeing is the flow direction.
MR. SCHROCK: The flow direction?
MR. WELTER: Right. Through this case
with a steam generator, your gray dots are for a
blinded outlet, which means that we have placed a
blind, a physical blind, where the steam generator is,
so it just smacks into a wall. For that case it is
calm.
For the case where we have a steam
generator attached, the blind is removed, it's also
calm, but the level --
MR. SCHROCK: Do you -- do you --
MR. WELTER: Yes.
MR. SCHROCK: -- see it or are you
imagining this?
MR. WELTER: No. We have a clear PVC
pipe, and we've recorded the flow regime.
CHAIRMAN WALLIS: So the difference isn't
only that you've got some flow in the cold leg, it's
that you removed some sort of a plug at the end of the
pipe.
MR. WELTER: Exactly.
CHAIRMAN WALLIS: An open-ended pipe
instead of a closed pipe.
MR. WELTER: Exactly, yes.
CHAIRMAN WALLIS: So I guess what you're
showing us is what we've been saying all along, the
end conditions make a difference.
MR. WELTER: Exactly, yes.
MR. BAJOREK: Do you have an idea of what
the flow split is, how much of the gas is coming from
--
MR. WELTER: We are currently --
MR. BAJOREK: -- the vapor side versus the
steam generator side?
MR. WELTER: In terms of actual figure,
no. We are going to install a meter on that side to
meter that. In terms of just considering friction,
there's at least a hundred times more length to go
through on the cold-leg side. So we, of course,
expect a lot less flow.
CHAIRMAN WALLIS: The difference in flow,
it's an h over D of a half is a factor of about three,
log paper. It's a big difference.
MR. WELTER: Yes. This is a log on the
horizontal axis; that's correct.
MR. SCHROCK: That's why I have a hard
time believing that it's a flat interface in both
conditions.
MR. WU: It's no entrainment. There's no
liquid that has been pulled out of the branch, so it's
calm. The key part is to --
MR. SCHROCK: The data for the onset of
entrainment.
MR. WU: -- see which level is higher when
it's become too quiet. So for the waves of the steam
generator case, it's a -- the higher level, then it
becomes quiet. And without the pressure boundary
there, your plant flooded. Then it's a lower level
and the level becomes quiet. So that means when
you're bring flooded, there is a kind of wave bouncing
back from that pressure boundary, and that will
entrain more liquid out. Then that causes the
entrainment --
MR. SCHROCK: So the open circles, --
MR. WU: -- onset a level lower --
MR. SCHROCK: It's -- the open circles,
it's picking up liquid off the tops of waves; is that
right?
MR. WU: That's -- that's right. And at
the end there's no more liquid being pulled out. The
liquid level becomes quiet.
CHAIRMAN WALLIS: So you have a plug at
the end of the pipe so waves can reflect from it,
right?
MR. WELTER: Exactly.
MR. WU: That due course of the
entrainment.
CHAIRMAN WALLIS: Because one thing that
happened in this, where you quote here, which had a
wave-absorbing thing at the end of the pipe, so it
wasn't a reflection.
MR. WU: It's because -- if you open the
return line, it becomes quieter.
CHAIRMAN WALLIS: So now I'm a bit happy
of it, because there is something -- you know, we took
care to have no waves reflected from the end of the
pipe in these experiments that you quote here. Okay.
MR. WELTER: Thank you, sir.
So that was a test series that was ran to
determine the effect of a steam generator blinded or
open --
CHAIRMAN WALLIS: Wait a minute.
MR. WELTER: Yes, sir.
CHAIRMAN WALLIS: My colleague's asking me
how do you know hl. Is hl something you measure
before you turn on the --
MR. WELTER: Hl is the hot-leg level at
which entrainment stops.
CHAIRMAN WALLIS: And this is as
determined by your probe?
MR. WELTER: It's determined by the
impedance probe.
CHAIRMAN WALLIS: Whatever the probe is
measuring. And it's an average of --
MR. WELTER: It's an average over time.
And we determine if it's constant, if it approaches
some moving average.
MR. BAJOREK: Is it based on both the
vessel side and the steam generator side conductivity
probes?
MR. WELTER: For the onset of entrainment,
there is -- the same. For the onset of entrainment,
the levels are the same.
When you encounter the entrainment rate
levels we'll show that that's different when we do the
entrainment rate tests.
We ran a test series to better understand
the effect of downstream condition. We also installed
a one-inch return line to the top of the reactor
vessel to give us a little bit of refinement to the
effect of the downstream condition.
What we have seen here is with the steam
generator installed no return line, which means both
the three-inch and the one-inch line are closed, which
gives the effect of the lower plenum of the steam
generator on the onset level.
Then we have the three-inch line open.
And then we have the one-inch line open. And of
course the three-inch closed.
CHAIRMAN WALLIS: I just have to ask you
something else, too.
MR. WELTER: Yes, please.
CHAIRMAN WALLIS: Once you get this
entrainment, it goes up into the ADS line.
MR. WELTER: Yes.
CHAIRMAN WALLIS: And I assume that you
have enough flow rate to carry it up that line, --
MR. WELTER: Yes, yes.
CHAIRMAN WALLIS: -- because if the line
is too big, it's not going to go over the line; you
have a different condition where you actually entrain
it, but it runs back down into the pipe again.
MR. WU: Yes, we guarantee its annular
flow. It's over 14 --
CHAIRMAN WALLIS: So you have enough flow
--
MR. WU: The minimum is a 14-meter per
second JG.
CHAIRMAN WALLIS: So it doesn't actually
go up like a jet in the middle. It actually splatters
onto the wall, or something?
MR. WELTER: That's correct.
I assume you're referring to the
illustration of the jet in the middle. Okay.
CHAIRMAN WALLIS: Most of the flow is
coming from the vessel so that entrainment is probably
carried to the left as it goes around the corner into
the branch pipe, or something.
MR. WELTER: Yes, that's correct.
CHAIRMAN WALLIS: It doesn't go off the
middle.
MR. WELTER: Yes.
Okay. This shows the case with the effect
of the downstream condition flow, which means
basically I am changing the amount of flow that comes
in from one side.
So with the return line closed there, of
course, is only flow from one direction from the
reactor vessel to the hot leg through the ADS-4 line.
And then I open up the cold leg, and so I get a varied
amount of flow rates from the other direction.
MR. KRESS: Does the gas flow spiral as it
goes up the tube?
MR. WELTER: Spiral?
CHAIRMAN WALLIS: Spiral.
MR. KRESS: Spin.
MR. WELTER: Spin.
MR. WU: We can't tell.
MR. WELTER: I can't tell exactly. I wish
I could measure that.
MR. SCHROCK: What is the return line
that's referred to here?
MR. WELTER: Which one, the one-inch or
the three-inch, sir? Both of --
MR. SCHROCK: Gosh, I don't know. I'm
asking you --
MR. WELTER: Okay. The return -- the
three-inch --
MR. SCHROCK: -- to tell me what is the
return line.
MR. WELTER: Okay. The return -- the
three-inch -- okay. The three-inch return line goes
from the outlet of the steam generator. And it's
basically a model of the cold leg, which returns to
the top of the vessel head.
MR. SCHROCK: And can you relate it to a
picture you've shown us of the system?
MR. WELTER: I can relate it.
MR. WU: System 1.
MR. WELTER: The system 1.
Brandon, can you go back about 12 slides?
This is not an elevation view, so this
doesn't dip this far down. The cold leg, the
three-inch return line comes off the exit of the steam
generator. It comes back around. We have a valve
there. It comes back into the top of the reactor
vessel.
CHAIRMAN WALLIS: Now why should there be
any circulation in that loop at all?
MR. WELTER: Circulation?
CHAIRMAN WALLIS: There's no pump in that
loop. Why would anything flow around that loop?
MR. WELTER: If there's flow from decay
heat boiling, there's vapor flow from here. Air is
being supplied through the porous tubes up through
here. The flow can choose.
CHAIRMAN WALLIS: Oh, it could go the
other way around?
MR. WELTER: Right, exactly.
CHAIRMAN WALLIS: Okay. Okay.
MR. WELTER: It can choose which way,
depending on the friction. Exactly.
CHAIRMAN WALLIS: It's on its way to the
break, okay.
MR. WELTER: Exactly. We're trying to
determine basically how much goes the other way. So
we're either cutting it off, opening it, or opening a
small one-inch return line that isn't shown here,
which is basically the same as a three-inch.
CHAIRMAN WALLIS: Okay. Okay.
MR. SHACK: The one-inch return line has
the same geometry?
MR. WELTER: As the three-inch return
line?
MR. SHACK: Yeah.
MR. WELTER: In terms of geometry? No.
MR. SHACK: No.
MR. WELTER: It's smaller and just goes
straight across.
MR. SHACK: Okay.
MR. WELTER: Yeah.
MR. KRESS: If you have a valve in the
line, why did you need a one-inch line?
MR. WELTER: Because --
MR. KRESS: Couldn't you simulate a
one-inch --
MR. WELTER: -- we didn't -- we don't --
it's a gate valve, as you'll see. And so --
MR. KRESS: Oh, you can't set it very --
MR. WELTER: No.
MR. KRESS: Okay.
MR. WELTER: You can't set it, but open,
close.
MR. KRESS: Okay.
MR. WELTER: So this shows the effect of
the gas flow direction. All of the data is still well
behaved, and so there is little effect of the gas flow
direction, meaning there is not very much flow going
through the cold leg or going through the return
lines. Most of the flow is still from the reactor
vessel into the hot leg through the ADS-4 line.
CHAIRMAN WALLIS: I notice there's a lot
of data scattered. It doesn't seem to be consistent.
If you make the return line size bigger, that's sort
of a consistent trend.
So these are some sort of data points, but
presumably if you repeated the experiment you wouldn't
get quite the same point?
MR. WELTER: Are you asking about the
repeatability of our experiment?
CHAIRMAN WALLIS: Well, it just seems that
if you look inside the trends, when you have no return
line, one-inch, three-inch, there's no sort of obvious
trend. And so --
MR. WELTER: Sure. There is an effects --
CHAIRMAN WALLIS: -- I conclude this is
just scatter that you're showing.
MR. WELTER: There is no significant
trend.
MR. WU: Yes, he said that.
MR. WELTER: Yeah, there's a scatter.
MR. WU: He already shows that the --
MR. WELTER: The dots are above.
MR. WU: -- symbol, circle is above --
CHAIRMAN WALLIS: They're sort of above,
yeah.
MR. WU: Yeah. We choose this plot to
represent our experiment data, as we want to leave the
correlation comparison later, because this was
originally initially using this -- you noticed that
before.
When we calculate the Froude number, we
use the gas, all the gas for -- to one side to
calculate it there. So basically you see that
shifting I think is because of the flow direction.
CHAIRMAN WALLIS: So this Froude number is
based on the total gas flow?
MR. WELTER: From the injection, exactly.
MR. WU: Yes.
MR. WELTER: From the meter.
CHAIRMAN WALLIS: It's the only thing you
know. You don't know how much is going each way. So
--
MR. WU: That's right. So that means when
that pipe is getting bigger, like you said, maybe it's
going to shift up because one side of the gas flow
rate is not that much.
So if we can't -- right now we installed
-- we are going to install a flow meter. Maybe we can
bring that down, we hope. But this is not the final
correlation or model we are going to use for this
entrainment answer --
CHAIRMAN WALLIS: So I guess I have to --
MR. WU: -- test --
CHAIRMAN WALLIS: -- ask: What's the
purpose of showing this picture?
MR. WU: Just to see the effect of gas.
CHAIRMAN WALLIS: Just to show that having
a return line doesn't make much difference?
MR. WELTER: Exactly.
CHAIRMAN WALLIS: But, you see, in the
code you'd have to actually calculate the flows in the
return line and use some kind of a correlation. I'm
not quite sure how that captures what's shown here.
MR. WU: Originally we -- in this figure
we expected a scatter, and -- like you already said.
And we -- my intention was to say, well, this is for
regime map -- for regime transition criteria; it
shouldn't work for this case, and -- but right now
it's very -- grouped like that give you maybe false
information.
You'll say, "Well, this is -- this is
correlation can't work for that." I apologize for
that.
CHAIRMAN WALLIS: Is that the correlation
that's in the code?
MR. WU: The correlation of that is for
our transition.
CHAIRMAN WALLIS: Is that what's in RELAP?
MR. WU: It's not the one we are going to
use. We are going to use the one --
CHAIRMAN WALLIS: You're going to use
something else?
MR. WU: -- in the code or improve it,
trying to compare with other correlations like
Maciaszek's correlation.
MR. SCHROCK: You have a horizontal solid
line. Is that part of the slug transition? What's
the meaning of that?
MR. WELTER: That's the difference between
annular, dispersed, and a plug and slug flow regime.
CHAIRMAN WALLIS: That's a Dittus-Boelter
transition criteria?
MR. WU: Yeah.
MR. WELTER: Yeah, okay.
So I summarized the test results on the
onset of entrainment. And the second group of tests
was to discuss -- or take a look at was the steady
state entrainments. And the major difference here is
that we have a steady injection flow.
CHAIRMAN WALLIS: Are you getting onset of
entrainment by extrapolating back from finite amounts,
or something? How do you know onset? How do you know
zero?
MR. WELTER: You mean when does it stop,
when does it start?
CHAIRMAN WALLIS: When does it start,
yeah. Sometimes --
MR. WELTER: We take a look at the data
and we take a look at the liquid level. If it
approaches a moving average, then we assume that no
liquid is being pulled out and the level does not
drop. That is the level at which onset begins.
CHAIRMAN WALLIS: Do you extrapolate it
then back from when it is dropping? Measuring zero is
not very easy.
MR. WELTER: There is still a level in the
hot leg, but there is a level at which entrainment
does not drop the level any farther. Basically --
CHAIRMAN WALLIS: So it has come to an
equilibrium level?
MR. WELTER: It -- yes, it comes to a
constant level --
CHAIRMAN WALLIS: So you're extrapolating
to equilibrium?
MR. WELTER: It just stops.
CHAIRMAN WALLIS: Okay. So you do --
MR. WELTER: That's correct, yes.
CHAIRMAN WALLIS: -- have an extrapolation
of the thing going on.
MR. WELTER: Okay. Yes. I was confused.
I'm sorry. Thank you.
MR. SCHROCK: The onset as reported in the
previous literature is dependent to some extent on the
method of observation.
MR. WELTER: That's correct.
MR. SCHROCK: In the KFK experiments, for
example, they used an acoustical method for measuring
the onset. We looked at it visually.
And in both cases there were circumstances
in which you would get intermittent lifting of the
liquid at a certain level. And then at a slightly
higher level you get continuous flow of the liquid.
And so you have to make a decision what is
it that you're using as the basis in your correlation,
because you need that hb or hl, as you've designated
it, in your correlation for the flow in the break line
after the onset of entrainment as a function of the
level in the stratified zone.
So I'm simply mentioning that, because I
don't hear any significance attached to the method of
observing the level for the onset.
CHAIRMAN WALLIS: He's measuring the --
MR. SCHROCK: Are you --
CHAIRMAN WALLIS: -- the opposite. He's
measuring the stopping of entrainment.
MR. SCHROCK: Are you giving the value for
continuous entrainment, the value for intermittent
entrainment, a value that's seen visually, or a value
that's detected by some instrument, such as the KFK
experiments using an acoustical detection in the
branch line?
MR. WU: For the entrainment onset level,
like Dr. Wallis' point, we measured when it stops so
that level is quiet --
MR. SCHROCK: When it stops?
MR. WU: Yeah.
CHAIRMAN WALLIS: So -- could you draw on
the thing here --
MR. SCHROCK: So you're not giving the
onset; you're giving the cessation?
CHAIRMAN WALLIS: Draw us what you
actually -- what you actually measure.
MR. KRESS: They're actually using those
probes on each end, I think they said.
CHAIRMAN WALLIS: You measure hl versus
time, or something?
MR. WU: Yeah.
MR. WELTER: Okay. So I'll draw -- I
think you were wondering what the data looks like,
draw a picture of what --
CHAIRMAN WALLIS: Yeah. How -- what
actually -- when do you say it stops and that sort of
thing.
MR. WELTER: Okay.
CHAIRMAN WALLIS: Could you use a thing
that shows up, not the plain red one.
MR. WELTER: Yeah.
Okay. So this is the time during the
test.
CHAIRMAN WALLIS: Right.
MR. WELTER: And this would be with the
calibration curve then, the level in the hot leg, hl.
CHAIRMAN WALLIS: Right. That's hl.
MR. WELTER: That's correct.
You would see -- at the beginning of the
test you would see a full hot leg right there. And as
you throttle the gas flow, you would, of course, see
a sharp drop in the level in the hot leg. And after
a period of time --
CHAIRMAN WALLIS: You leave the gas flow
constant now?
MR. WELTER: Leaving the gas flow as
constant. There's no injection rate. After a period
of time this level will go like that, the level in the
hot leg.
CHAIRMAN WALLIS: Okay. So --
MR. WELTER: This is the level --
CHAIRMAN WALLIS: It's its last gasp. In
fact, it's the level at which the last little piece of
wave comes off.
MR. WELTER: Exactly. And so we take a
look, and we say that this level right here, we take
an average of time and we compare it, of course, the
time's average between this one, average there, a
moving average. And we compare what does this
approach to, what does that value approach to. That's
where onset stops. And we have then said that's when
onset begins.
MR. WU: So for --
MR. SCHROCK: Well, how do you know
there's not a hysteresis involved in the phenomenon?
MR. KRESS: You would expect some.
MR. WU: We tried to bring the liquid
level up. But once you overbring it, it's going to
put it out. Because of our branch size, we guarantee
the branch's velocity is overly, a full regime
transition for the annular flow. So anything being
put to that branch exit is going to pull out. So it
mustn't pull out.
So if you go for -- we go -- we went from
the bottom up, bring liquid there, and to come back to
-- if we overshoot it, it will bring up to the
stopping point. That's the same result.
MR. BAJOREK: But can you tell, when you
bring that level up, whether you're getting
entrainment and then it drops back down to the level?
You could have been getting entrainment at a lower
level and you won't see it until you've entrained a
whole bunch of it --
MR. WU: Well, --
MR. BAJOREK: -- and drop back down. So
--
MR. WU: No, no. There's no drop back,
back down significant. We bring it up to the -- when
the entrainment occurs, you -- you obviously, when you
say "entrainment," that's already overshooting the
level, right?
Then you bring down a little bit, and
actually it's the same condition as what we -- we are
talking about.
You have a minimum in the gas count
constant, and then the entrainment stops. Here is you
just overshoot a little bit, and it just finally stop.
For our case it is from the top to the stop. It's --
MR. BAJOREK: But it's a question of
whether there's a hysteresis if you get entrainment at
a lower case in g- --
MR. WU: We didn't find that. For our
test we didn't show that. We didn't find that.
CHAIRMAN WALLIS: I'm uncertain about the
time now because you're describing a test and you've
had about four or five different correlations and
analyses to go through. And if we ask as many
questions about each one of those you're going to be
here until about three o'clock before we get lunch.
But we may -- you know, it may be
worthwhile asking those questions. We just don't have
the time.
MR. SPEAKER: You have to do what you can
to keep us going.
MR. WELTER: Okay. Thank you, sir.
I didn't describe those, okay.
For the steady state entrainment test,
this is for constant liquid injection and a constant
gas flow rate. We will go ahead and the reactor
vessel will start basically dry. And I will throttle
the flow rate of the liquid to fill the reactor
vessel, at the same time throttling the air at a
specified flow rate.
And this will go ahead and will raise --
the two-phase mixture will raise. At a certain point
the hot-leg level will start to entrain.
And we will then take this data for a
period of time, approximately four to six minutes.
And this is the time the h, the level at which there's
steady state entrainment.
And in this case, since flow cannot go
around the steam generator, the injection is equal to
the entrainment rate. We're trying to determine what
the hot-leg level is for those flow conditions.
The test scope, the matrix that we did is
the effect of the steam generator, close it, open it
-- the blind. I'm sorry. And the effect of the gas
flow direction on that entrainment rate.
CHAIRMAN WALLIS: Now when you do a
theory, are you going to use different theory in the
different places in the flow regimes in the different
-- in the next figure?
MR. WELTER: Theory?
CHAIRMAN WALLIS: You've got plug, slug,
stratified, wavy. I'm just following ahead.
MR. WELTER: Oh, yes.
CHAIRMAN WALLIS: Are you going to use
different theories in the different parts of the
picture?
MR. WELTER: Oh, okay. I'll go ahead
there.
CHAIRMAN WALLIS: Are you going to use
different theories in those different flow regimes?
MR. WELTER: I'm sorry. Could you clarify
"theory," what you mean by "theory"?
CHAIRMAN WALLIS: On the right you've got
four flow regimes.
MR. WELTER: This is classic flow regime
map.
CHAIRMAN WALLIS: Are you going to use the
same theory for all points?
MR. WELTER: In terms of the model
development, sir?
MR. WU: This one, in this case you have
liquid flow. For the previous one you don't have
liquid flow.
CHAIRMAN WALLIS: Yeah, but I'm just
saying you're going to develop a theory for liquid --
MR. WELTER: All of them, he wants to know
that.
CHAIRMAN WALLIS: Is it the same theory,
or different?
MR. WU: For this flow regime map we use
-- if you -- you don't have a JF you cannot do it,
right? That's for this stratified, okay. So without
the liquid flow in the main pipe you cannot use this
for a regime map.
The one Zuber proposed for that one is a
-- said that VF equal to zero. We'd use your
transition criteria. We said we have equal to zero.
Okay.
MR. WELTER: Okay. I think it would be
more explained when he talks about the model
improvement, when he talks about the actual model
part.
The test matrix includes gas flow rates of
up to 300 standard cubic foot per minute. We can --
our compressor is capable of at least three times
that, but with corresponding pressures we, since it is
PVC, rate at 20 PSI, we maintain low, so we maintain
integrity of our test section.
Similar with that, our liquid flow will go
up to 60. Our pump is capable of 600. We maintain it
low so the test section does not break or leak.
These are the data points we ran. We
wanted to get a good full spectrum as possible. And
when we look at a classic flow regime map, this is
different in the fact that this is not the flow
regimes that we see in our test section because we do
not have a developed flow. We have a short inlet.
But if it were, this is where the data would fall.
This is an illustration or a visualization
from the separate effects. We would like to see
what's happening, and so I have a clear PVC test
section.
This is a visualization data that we
recorded. I have another animation. And it
illustrates the oscillatory phenomenon that you see
when the three-inch return line is closed or the loop
seal is filled.
CHAIRMAN WALLIS: Are you going to show us
movies? I guess you are.
MR. WELTER: I'm going to show you a
movie. So that is the oscillatory nature and that is
approximately real time.
CHAIRMAN WALLIS: I would think that no
theory is going to predict that.
MR. WELTER: That -- okay. It comes to a
point that previous studies, of course, what liquid
level are you going to use for h. There are two
levels. And since we have two probes that measure a
hot-leg level, we have two distinct different levels.
CHAIRMAN WALLIS: And this stuff about
potential flow out to a sink or something isn't going
to be relevant to that picture, is it?
MR. SCHROCK: Let's see. This is what I
was trying to point out earlier. What you're showing
here illustrates that the phenomenon that you're
studying has nothing to do with the physics of the
flow of the gas creating a low-pressure zone that
lifts liquid off of a smooth interface and entrains it
into that branch flow.
CHAIRMAN WALLIS: It doesn't matter,
Virgil.
MR. SCHROCK: These --
CHAIRMAN WALLIS: The theory's going to
work anyway.
MR. SCHROCK: Huh? I mean --
MR. WELTER: This is -- I'm sorry.
MR. SCHROCK: -- how you could imagine
you'd fit this into the format of the correlation --
CHAIRMAN WALLIS: Well, it may well --
MR. SCHROCK: -- that represents the onset
of entrainment as a function --
MR. WU: No. We hope --
MR. SCHROCK: -- of a liquid level. There
is no definable liquid level in this thing.
CHAIRMAN WALLIS: There is whatever's
measured.
MR. WU: There's an average in the --
MR. SCHROCK: No. There is not --
CHAIRMAN WALLIS: It's whatever's measured
by the probe.
MR. SCHROCK: -- even a definable average.
Try to define it. See how far you get.
MR. WU: So that means we may need to do
some more modeling or --
CHAIRMAN WALLIS: No, no, no, no. You
finished the program, we heard. I think we've got to
go on, --
MR. WELTER: Thank you, sir.
CHAIRMAN WALLIS: -- but obviously there's
some skepticism.
MR. WELTER: Thank you, sir.
So we are concerned with, of course, the
real case. There are two levels. How to determine,
when we're using a model, which level to use, average
the reactor side, the steam generator side.
And part of our test scope I wanted to
illustrate what we just spoke about which is the step
phenomenon, and that is these data points. The square
dots are for the test series with a closed return --
I mean -- I'm sorry. This is mislabeled.
(Presenters Mr. Welter and Mr. Wu confer
off record.)
MR. WELTER: So with the closed line there
is not a large difference in the levels when we open
the steam gen- --
(Presenters Mr. Welter and Mr. Wu confer
off record.)
MR. WU: So the upper circle is the --
within the same -- oh, yeah, it's over it.
MR. WELTER: It's opposite, yes. I'm
sorry. The graph is incorrect. These data points
should be switched in terms of the squares are with
the three-inch line open. And the circles are with
the three-inch line closed.
So when the three-inch line is closed,
then the oscillatory behavior is seen. And you can
tell that by the difference in level. One level is
the react- -- this is the steam generator side,
hot-leg level. This is the reactor hot-leg level.
And when the case -- when the three-inch
line is open it's a much more calm surface and the
levels are much more similar.
MR. SCHROCK: What is the meaning of "step
phenomenon"?
MR. WELTER: "Step phenomenon" is meaning
-- I'm sorry -- that there is a step in your level.
You're basically seeing a lower level in the inlet to
your test section and a higher level on your steam
generator side, so the level is stepping.
MR. SCHROCK: Well, the thing that you
refer to as steady is shown, in three different views,
something that's very unsteady and you imagine some
average condition about it. But can --
MR. WELTER: We have taken a time average
of that condition; that's correct.
MR. SCHROCK: Yeah. But for step
phenomenon, what do you imagine, that you have a level
that suddenly changes as the liquid progresses?
MR. WELTER: I'm sorry. I was not clear.
MR. SCHROCK: I want you to explain to me
what the term "step phenomenon" means.
MR. WELTER: The term "step phenomenon"
means to me, in the way that we have described it, is
that there is a difference in levels between the
reactor side and hot leg, if we look at an average.
CHAIRMAN WALLIS: It seems to me what's
happening is that you have a plug of liquid in the
steam generator and then the -- everything's clear for
the --
MR. WELTER: Exactly.
CHAIRMAN WALLIS: -- hole, so the gas goes
out, the plug runs back. As soon as it comes back to
the hole, it blocks the hole, the pressure goes up, --
MR. WELTER: Right.
CHAIRMAN WALLIS: -- and it gets shot back
up into the steam generator. You're generating
oscillation of a slug of water.
MR. WELTER: And the plug does not reach
the inlet side.
CHAIRMAN WALLIS: Right. This is very
system dependent. So what you're studying is
entrainment. When you have an oscillating plug going
into a steam generator, it's completely different from
Professor Schrock's experiment --
MR. WELTER: That's correct.
CHAIRMAN WALLIS: -- and Lahey's, and so
on. It's a different thing altogether, but it may
apply to AP600.
MR. WELTER: Yes. Thanks.
So this is also the --
MR. SCHROCK: It may not, too.
MR. WELTER: This is also the effect of
the steam generator, again similar to the onset, with
a blind compared to with the steam generator. You see
with no steam generator there is going to be a higher
liquid level, similar to what we discussed with the
onset of entrainment tests.
So the steam generator is also important
when we're considering entrainment rate phenomena at
steady-state conditions.
CHAIRMAN WALLIS: How is this going to get
fit into a code, the fact that the steam generator is
important?
MR. BAJOREK: At this point we've got to
be very careful on what and how we would apply this
model. The problem in RELAP is that the horizontal
stratified model was grossly underpredicted in the
total entrainment.
I think what we see in the movies here is
something that's more flow-regime dependent rather
than something that's giving us entrainment off of a
horizontal stratified. The flow regime is sitting
there quiescent.
I think that at this point we should be
taking the correlations, Maciaszek, Schrock,
potentially the new one, and looking at the
sensitivity of entrainment that we might be getting,
but I don't know if we're far enough along that we
would say that this is a great model and we should
drop this in and replace what we've got there.
I think what we are seeing based on their
work so far is that they are seeing higher rates of
entrainment than the previous model that had been in
RELAP.
And that puts it more in line with the
no-reserve tests, some of the other -- I guess there
was another one of the tests that was showing much
higher entrainment than what they've been getting out
of the existing correlation.
We're going to need something like that
for AP1000, where the gas velocities coming out of the
core are going to be substantially higher and we're
going to expect more entrainment.
So I think it's headed in the direction of
increasing the -- being able to predict higher
entrainment. But I wouldn't say that this is a model
that we can say is completely adequate for all
situations.
CHAIRMAN WALLIS: I'm not sure yet that
the model represents the physics. I'm sort of with
Virgil. Perhaps we need to get to the model.
MR. SCHROCK: Well, I think this is the
problem always with the codes, is that -- I said this
in our private discussions -- you make comparisons of
the code predictions against integral system
performance. And you get an impression that the
reason that the code doesn't predict the experimental
data well is one of hundreds of correlations that are
embedded in that code isn't right for that situation.
Now what you've just said focuses very
clearly on what the difficulty in their thinking is.
It's not that the correlations were
entrainment from a stratified region are the problem;
it's that the code is telling you you have a
stratified region when, indeed, it's not stratified.
And, therefore, you're getting something altogether
different.
The problem is in the characterization of
the flow regimes, which is excessively simplistic in
the codes and leads to all kinds of difficulties and
confusions in interpreting these comparisons of
integral test performance and code prediction.
And you're never going to improve that
following the course that this is taking. You're
dealing here with different phenomena than the
references taken from the literature addressed.
MR. KRESS: It does -- it does --
MR. SCHROCK: You have to address the
phenomena that are occurring in that system.
MR. KRESS: It does point you to where you
need to work on your code, because basically the
RELAP-type formulation won't predict this oscillatory
behavior because of the way it's set up. And that's
what you need.
You'll need something that predicts when
you get this behavior, and then perhaps your results,
or whatever correlation you put out of them, could be
used if the code could predict that behavior.
MR. SCHROCK: Right.
MR. KRESS: But you need to work on that
part of the RELAP.
MR. BESSETTE: Yeah. I think we agree --
I think -- you know, see, what we have in RELAP right
now is -- simply invokes the off-take model when the
code says you have stratified conditions at that node.
So if we have -- so you have to invoke the
model at the right time. So if we have, you know, one
particular model for slug flow, one for wavy, one for
stratified, of course, as a starting condition, the
code has to get flow regime right, you know, --
MR. SCHROCK: Right.
MR. BESSETTE: -- in order to invoke the
model at the right time.
MR. SCHROCK: That's precisely my point.
MR. SHACK: The physics are different in
each one of those regimes in what it does. And right
now it just gravitates from regime to regime and makes
it very simplistic.
CHAIRMAN WALLIS: I'm just wondering where
you're going to go with this presentation. I see
you've got what looks like three models that we sort
of agree don't apply.
Are you going to just simply say here are
three lousy models that don't apply and go on to the
one that works that you developed, or are you going to
spend a lot of time going through something which
doesn't apply?
MR. WU: No. We want to say the two
models we compare are the two asymptotical condition.
And then we want to say bridge them together.
CHAIRMAN WALLIS: Well, we don't need to
worry about Schrock's sink flows and things like that,
do we, because it doesn't apply?
MR. WU: Well, --
CHAIRMAN WALLIS: Are you going to drag us
through --
MR. WU: The case is for this project, we
are trying to extend for small breaks. And Schrock's
correlation obvious for small breaks should work for
that. And then we want -- we don't want to abandon
these --
CHAIRMAN WALLIS: For very small breaks.
MR. WU: Yes. So we say, well, we cannot
develop a model say just for the larger break. We
have to consider some model that already worked in the
past. So we want -- so we figure out from the
theoretical reasoning we found it is Maciaszek's and
Schrock's work. It's actually two asymptotic
conditions. And then we try to bridge them together.
So if you have a small break, it works.
Basically it is approaching to Schrock's work. And
then larger break approaches to Maciaszek's work.
That's what our intention. We didn't try to abandon
what is the previous work.
MR. WELTER: Thanks, sir.
I think that's my last slide in the test
results section, determining the importance of the gas
flow downstream, the gas flow direction.
In the onset tests we saw that there was
a little bit of scattering compared to the entrainment
rate tests. When I open or close a three-inch return
line, there is a large difference in the liquid level
in the hot leg.
CHAIRMAN WALLIS: Well, how are you going
to use these data? Do you know the flow rate in the
return line?
MR. WELTER: In terms of where -- we're
installing a metered flow line.
CHAIRMAN WALLIS: Oh, you don't know it
yet?
MR. WELTER: That's correct, sir.
CHAIRMAN WALLIS: So we're simply
observing there's a difference, but --
MR. WELTER: That's correct.
CHAIRMAN WALLIS: So the data aren't
usable yet until you've done some more measurement --
MR. WELTER: That's correct, sir.
CHAIRMAN WALLIS: -- to know what's really
going on?
MR. WELTER: That's correct, sir.
This is showing just the effect.
CHAIRMAN WALLIS: But it's an effect which
--
MR. WELTER: It's important.
CHAIRMAN WALLIS: -- can't be reflected in
a theory because you don't know the flow rate split.
MR. WU: Not yet.
MR. WELTER: Not yet.
CHAIRMAN WALLIS: You don't know the flow
rate split going in the two directions?
MR. WELTER: Yes, that's correct.
MR. WU: We know one is closed --
CHAIRMAN WALLIS: So you can't have
finished -- you can't have finished the work.
MR. WU: The closed case, we know. That's
-- that presented the real case of the loop seal case.
That's the major, and --
CHAIRMAN WALLIS: It's a major effect,
yes.
MR. WELTER: In this presentation we
haven't presented an entrainment rate model. We've
presented an onset model. The entrainment rate model
is continuing work at this point.
CHAIRMAN WALLIS: Well, you've explored
some things which influenced the entrainment.
MR. WELTER: That's correct.
And so this shows the effect of the return
line or the flow direction.
Noticeably, with the three-inch line open
there is no oscillatory behavior. So when the loop
seal is filled and the three-inch line is closed, the
oscillatory behavior occurs. When it's open, it's
much more calm.
I'd like to take a few minutes because,
like Dr. Wu said, the new model proposed is asymptotic
conditions of Schrock and Maciaszek's work, so take a
moment to look at those models.
That's Schrock and Smoglie using a
potential flow formulation. The stream lines go to a
sink. The break is modeled as a sink. The stream
lines intersect the interface, so there's in effects
of the interface, gas leak or interface on the
potential. And hb is far from the break, or the other
way to say that is the break size is very small, can
be considered a point sink.
For noise equate --
CHAIRMAN WALLIS: There's a flat ceiling,
too, isn't there? There's no curvature to the pipe
and all that.
MR. WELTER: That's correct. Exactly how
strong it gets.
Hb again is considered as the gas chamber
height at which entrainment begins.
CHAIRMAN WALLIS: What does he do with his
interface then? He's got another sink reflecting an
interface, or something? How does he --
MR. WELTER: In the model? Without
improvement? I'm sorry, sir?
CHAIRMAN WALLIS: The stream lines come
out of the interface like that?
MR. WELTER: The stream lines --
CHAIRMAN WALLIS: Magically come out of
the interface?
MR. WELTER: That's correct. It's a sink.
MR. SCHROCK: I don't know why you've put
my name on there, but --
MR. WELTER: Smoglie. I'm sorry, go
ahead.
MR. SCHROCK: -- apart from the fact that
it's misspelled.
(Laughter.)
MR. SCHROCK: The history here is that the
program at KFK involved extensive experimentation
under the direction of Dr. Reimann -- Reimann and
Kahn, Reimann and some other people.
Smoglie was a student who did a
theoretical thesis employing potential flow to make a
prediction of the value of hb for a stratified
upstream condition.
I don't recall her having stream lines
intersecting the interface, but it's been more than 10
years since I last looked at that. Maybe she did, but
I kind of doubt that.
But, in any case, it's not something that
I suggested or that any of my co-workers suggested.
MR. WU: We put Smoglie --
MR. SCHROCK: Also in your reporting of
the KFK data you refer to that as Smoglie data. She
had no data. I mean she -- in the sense that she did
experimentation, she was not an experimentalist. She
was a theoretician -- is a theoretician I presume now.
MR. WU: Thank you. We should represent
it as KFK data later.
And for this Smoglie, they arrived at
that. That's true. The interface, the stream line
goes to the interface. I mean there is no interface
there.
For your name put there is because your
correlation. It's almost identical with that, except
that the gas density --
MR. SCHROCK: Well, that's kind of a loose
description. I mean --
MR. WU: So the two correlation basically
the same, --
MR. SCHROCK: Yeah.
MR. WU: -- to the fifth power, so we
didn't see this derivation is yours. We should have
cleared it up before. And yours is based on the
Froude number. And the Froude number is based on the
branch velocity.
CHAIRMAN WALLIS: One of the units here,
W is a flow rate?
MR. WU: Mass flow rate.
CHAIRMAN WALLIS: Mass flow rate. I don't
understand how the units of the final correlation work
out. It doesn't even make sense to me, but -- three
in -- oh, well, maybe it does. Okay. Maybe it does.
MR. WELTER: So Bernoulli's equation can
be written along the Z axis from the interface to the
point sink. You take the derivative in terms of that
with respect to the distance away, then you can
develop a criterion based on the pressure gradient at
the interface. If --
CHAIRMAN WALLIS: We're beginning to pick
up the interface.
MR. WELTER: That's correct.
So at some condition if this pressure
gradient is greater than or equal to the gravity
potential, then entrainment begins. So it's an onset
entrainment criterion.
CHAIRMAN WALLIS: So something happens.
MR. WELTER: Exactly. Something happens.
CHAIRMAN WALLIS: Is there any
confirmation that this works?
MR. WELTER: Oh, yeah. The next slide
will show data and how well this correlation -- yes.
Hb is then -- you can get that from this
correlation. And hb is a function of the gas mass
flow rate to the one-fifth power or squared to the
two-fifths power.
CHAIRMAN WALLIS: So the Smoglie data,
compared with theory, spans about two orders of
magnitude?
MR. WELTER: Yes, that's correct. But
this -- the KFK data --
CHAIRMAN WALLIS: Worst case.
MR. WELTER: Yeah. The KFK data has large
uncertainties in the determination. There are several
-- you see several gas flow rates for the same liquid
level.
This graph is a W3g squared, which is
what's in the brackets. So this is raised to the
one-fifth power.
CHAIRMAN WALLIS: So this is theory versus
experiment; is that what it is you're saying?
MR. WELTER: That's correct. So it shows
the degree of collapsing of the data, how well the --
so the exact -- the exact data would lie directly on
this line. Left side versus right side.
This is the correlation by Smoglie, and it
shows -- the red is ATLATS data for large -- or for
small data. I'm sorry. And for Smoglie and Schrock
data.
CHAIRMAN WALLIS: The only thing that
works well is the Schrock data.
MR. WELTER: Yes. That is right there.
It's beautiful. It has --
CHAIRMAN WALLIS: You deny --
MR. SCHROCK: Absolutely amazing.
CHAIRMAN WALLIS: -- having created this
data?
(Laughter.)
CHAIRMAN WALLIS: I thought you were
denying having created anything.
MR. SCHROCK: No, no. I created a lot of
data.
CHAIRMAN WALLIS: Oh, you created it, but
you didn't have any theory.
MR. SCHROCK: Smoglie had the theory only.
MR. WELTER: But it has difficulty
predicting where D over D is small.
CHAIRMAN WALLIS: Yeah.
MR. WELTER: The large breaks, that is.
Next we'll take a look at Maciaszek, who
used a formulation by Wallis. Considering -- Wallis
considered a branch or just basically a tube on top of
a large pool with gas flowing over it so gas flows
from all directions into an entrains liquid into the
branch. It considers an interface condition with a
two-bump sort of phenomenon, wave phenomenon, where
the wave crest height is determined and defined as
delta.
And you can write a simple continuative
equation here, where velocity of the inlet or velocity
from all sides in a virtual cylinder here, so it's rho
v pi times number, which is diameter of the sphere --
of the cylinder.
And the cylinder is from hb minus delta,
so it's the cylinder right here (indicating). And
that's show in this here. And that is --
MR. SCHROCK: Is this --
MR. WELTER: Yes.
MR. SCHROCK: -- a cylindrical off-take,
or what --
MR. WELTER: Yes, this is a cylindrical
off-take. So this is a cylindrical cylinder that the
gas is flowing into.
MR. SCHROCK: And so it's in cylindrical
geometry in that sense, and so these bumps represent
a ring of --
MR. WELTER: A ring, that's correct.
MR. SCHROCK: Um-hum. Why does it do
that?
CHAIRMAN WALLIS: I have no recollection
whatsoever of any of this.
(Laughter.)
MR. SCHROCK: Why would the liquid deform
in that way? It implies that the -- that there's a
ring of low pressure lifting it into that format.
CHAIRMAN WALLIS: I guess there has to be
a stagnation point in the middle, must be the
argument.
MR. WELTER: That's correct. Velocity at
this point is zero. There's a maximum velocity at the
wave crest. Work done by Dr. Wu in the model
improvement section I think delves a little bit deeper
into that question.
CHAIRMAN WALLIS: This leads to a theory
--
MR. WELTER: Yeah.
CHAIRMAN WALLIS: -- which compared on the
next figure.
MR. WELTER: Yes. So this leads to a
theory which is surrounded in this box here. The
difference being it has a different
experimentally-determined coefficient. And it uses
the break size diameter d and it's different as to the
one-third power inside of the one-fifth.
Here is how well the correlation -- it
brings our data down well, but it's skewed, the
Schrock and Smoglie data you can see, slightly skewed
compared --
CHAIRMAN WALLIS: Now why are you calling
it W2 over d5?
MR. WELTER: Okay.
MR. SCHROCK: That's this divided by h.
MR. WELTER: It's divided by d5 for
nondimensional. So the Maciaszek correlation, the
horizontal axis is hb over d.
MR. SCHROCK: What's on the axis of that?
MR. WELTER: I'm sorry. What?
MR. SCHROCK: What is being plotted on the
--
MR. WELTER: Okay. Yeah, the Maciaszek
correlation. So this is experiment -- this is
experimental and this is your theoretical. So this is
Maciaszek's correlation. It's the other --
MR. SCHROCK: No. What -- what quantity
--
MR. WELTER: Hb is on --
MR. SCHROCK: -- is on -- is on the
abscissa? It's unlabeled.
MR. WELTER: Okay. Yes. Hb over small d
to, in this case, the third power.
MR. SCHROCK: We should never have to
imagine that, you know.
CHAIRMAN WALLIS: Oh, is that what it is?
MR. SCHROCK: Even undergraduate students
know that.
CHAIRMAN WALLIS: I didn't think that. I
thought it was theory versus experiment. I guess in
a sense it is, but --
MR. WELTER: It is.
CHAIRMAN WALLIS: -- it's essentially flow
rate versus height, is what you're plotting. Flow
rate squared versus height.
MR. WELTER: That's correct. Can I go on?
Okay. To summarize, the model --
MR. SCHROCK: I'd just --
MR. WELTER: Yes, sir. Please.
MR. SCHROCK: -- finally like to tell you
that I saw this at close range many, many times and
never did I see a ring of liquid pulled up, never.
Always a symmetric --
MR. WELTER: Like this. Right here, yeah.
MR. SCHROCK: Right, yeah. Single little
thing coming up and drops coming off the top of it.
CHAIRMAN WALLIS: This happened so often
in two-phase flow.
MR. WU: Yeah. Maybe -- maybe --
CHAIRMAN WALLIS: The theory is based on
the physics which is utterly different from reality
and yet the correlation works.
MR. WU: Maybe the instability taken one
point then break the other symmetric.
MR. WELTER: To summarize the evaluation
of the entrainment onset models, Smoglie's data is a
large scattering with large uncertainties. The
Smoglie model is effective. As we saw, it predicts
that da- -- Schrock's data, and it's very effective
for small break sizes or when the interface level is
far from the break.
Maciaszek's model, which takes into
account the break size, is valid for large breaks and
-- which we saw why it pulled the ATLATS data down to
that line, or when the liquid interface level is close
to the break.
MR. SCHROCK: I guess you haven't shown us
any data yet that would tell me that you ought to plot
those on the same piece of graph paper. The data that
you have is for conditions upstream that are not
stratified for the most part, so far as I can tell.
MR. WELTER: Onset data -- the onset data
is a calm surface. The entrainment rate is the
oscillatory. Am I confusing that?
MR. SCHROCK: I just heard a lot of
discussion about the fact that you don't even measure
the onset of entrainment. What you measure is the
cessation --
MR. WELTER: Cessation of entrainment.
MR. SCHROCK: -- of entrainment.
Cessation of entrainment in relationship
to what kind of phase distribution you had upstream
was left quite unclear. But you've not shown us any
evidence of the fact that you have cleancut
measurements of the onset of entrainment from
stratified upstream conditions. I've not seen that.
If you have it, show it to us.
MR. WELTER: Okay. Sir.
CHAIRMAN WALLIS: Well, presumably the
bubbling in the reactor vessel sets up some sort of
wave motion in the pipe.
MR. SCHROCK: Well, I'm not arguing that
all these complications don't exist out there in the
reactor systems. But what I'm saying is you're not
going to improve your computer code by this kind of
pursuit of what is wrong with what the code is
currently doing.
What's wrong with what the code is
currently doing is, predominantly, it has no idea what
the upstream flow regime is.
CHAIRMAN WALLIS: Okay. We're getting
close to the break, Jose, are we? We're supposed to
go to 11:15 and then we're supposed to do a tour.
What would you like us to do?
MR. REYES: It would be valuable at this
point possibly to take a break and go look at the test
facility.
CHAIRMAN WALLIS: We're going to come back
and see this in the afternoon.
Then I think the interesting will be what
you have done to get better agreement with data. As
I understand, you have a better model, but it seems to
be based on these somewhat iffy past models rather
than a new model that really reflects what's actually
happening; is that the case?
MR. REYES: I don't believe we've gotten
to that point yet in the --
CHAIRMAN WALLIS: Can we discuss that
after lunch when we feel happier?
MR. REYES: Quite likely.
CHAIRMAN WALLIS: Is that good? Okay.
Thank you very much. Very interesting
subject.
So we're going to take a break now. We
don't have a recorder at the inspection of the test
facility. We don't --
MR. BOEHNERT: No.
CHAIRMAN WALLIS: So we're no longer in
session. And when are we going to come back? When do
we reassemble here?
MR. REYES: We would go to lunch from the
demonstration and then --
CHAIRMAN WALLIS: And then we'll return to
where we were here.
MR. REYES: At 1:30, I believe.
CHAIRMAN WALLIS: Shall we have a short
lunch; can we try and get back early? How soon can we
be back?
MR. REYES: I think a short lunch would be
--
CHAIRMAN WALLIS: Can we have a quick tour
and get back at 1:00?
MR. REYES: Let's do that.
CHAIRMAN WALLIS: Let's meet here at one
o'clock. We'll meet here again at one o'clock.
(Tour and luncheon recess taken from 11:15
a.m. to 12:58 p.m.)
CHAIRMAN WALLIS: And we'll continue with
the presentations by OSU.
MR. WELTER: I hope you enjoyed a good
lunch at West Cafeteria. I know I'm ready to fall
asleep now, but I don't get to do that.
We left off, I finished summarizing the
onset of entrainment model evaluation. I wanted to
then step into and discuss entrainment rate models
that we're evaluating also, the first one being the
Schrock correlation. It's based on a curve fit of
entrainment rate data based on your actual gas chamber
level, h, and then divided by your onset gas chamber
level, hb. That determined the quality or the rate to
your branch.
CHAIRMAN WALLIS: X is a mass flow rate?
There's nothing here about third properties at all?
MR. WELTER: It's predicting the quality,
which gives you a --
CHAIRMAN WALLIS: It's called a mass
fraction.
MR. WELTER: Exactly. It gives you the
fraction of liquid.
CHAIRMAN WALLIS: There's nothing about
densities or anything when -- if they were both water,
that wouldn't make any difference.
MR. WELTER: Hb.
CHAIRMAN WALLIS: It's remarkedly
substantive.
MR. WELTER: Smoglie also developed a
correlation based on the dimensionless h over hb.
CHAIRMAN WALLIS: She has densities,
though, so...
MR. WELTER: And there is a rho f and rho
g in this model. It's based on the right-hand term.
One minus that term is based on a vapor pull-through
through a down branch.
And then she went and said that for upward
is one minus the down, and then she modified the
experimental coefficient 2 over hb to match the data.
Yonomoto developed the correlation model
for determining x3. It has A and B, which are
experimentally-determined constants based on the void
fraction in your main line.
CHAIRMAN WALLIS: In the main line. It
says on branch void fraction.
MR. WELTER: Oh, I'm sorry. Branch.
CHAIRMAN WALLIS: Do you mean the main
line --
MR. WELTER: Thank you. Branch void
fraction.
CHAIRMAN WALLIS: So it's kind of funny
because x3 is a submeasure of the void fraction, isn't
it, indirectly.
MR. WELTER: Quality, yes. You can relate
quality in void fraction.
CHAIRMAN WALLIS: So it depends on itself
--
MR. WELTER: Yes.
CHAIRMAN WALLIS: -- in a sense. You have
to know the void fraction to predict the quality. You
might be in a little bit of trouble.
MR. WELTER: They ran a -- we ran
experiments to determine the void fraction in relation
to that.
The correlation is based on determining a
sphere of influence and within that sphere of
influence all of the liquid in the hot leg will then
be sucked up the branch.
CHAIRMAN WALLIS: If you plot these three
curves as versus h over hb, you're doing that
somewhere?
MR. WELTER: Yeah.
CHAIRMAN WALLIS: They're actually -- on
the same piece of paper?
MR. WELTER: That's correct. Smoglie is
not plotted on this one. H average, when we consider,
when we look at our data, we have oscillatory
behavior. So we have to have some way of placing that
on this graph. We use h average, which is the average
between the reactor side and the steam generator side,
liquid levels in the hot leg.
CHAIRMAN WALLIS: What shall I conclude
from this figure?
MR. WELTER: You can conclude from this
figure that the data -- that we ran ATLATS predicts
significantly higher entrainment rates than the
correlation. Higher h, yeah.
CHAIRMAN WALLIS: It's lower, isn't it?
X3 is lower?
MR. WELTER: Oh, I'm sorry.
MR. WU: Yeah, lower rate -- well, higher,
higher rate.
MR. WELTER: Well, the quality is lower,
so there's more liquid going in for the same -- for
the same level -- I'm sorry -- for the same level, if
you have the same level here, our data is a quality --
CHAIRMAN WALLIS: Oh, quality, okay.
MR. WELTER: -- less than .2.
CHAIRMAN WALLIS: Quality means --
MR. WELTER: And then theirs would be all
the way over here --
CHAIRMAN WALLIS: Okay. Sorry, that's
right. That's right.
MR. WELTER: -- up to 9, so there would be
entrainment rate, --
CHAIRMAN WALLIS: That's right.
MR. WELTER: -- higher entrainment rate.
CHAIRMAN WALLIS: X is a measure of vapor
fraction, not liquid fraction.
MR. WELTER: That's correct, sir.
CHAIRMAN WALLIS: Yeah. That's right.
MR. WELTER: I need to go back here. I
pressed the wrong button. Okay. So the different
graphs, Yonomoto is the higher one. Schrock's
correlation is the one in the center.
CHAIRMAN WALLIS: H ab over h break? Why
-- what's the --
MR. WELTER: Hb is the inception. H over
hb. This is the real hb that we find in our
experiments. So this is the hb from our new model.
CHAIRMAN WALLIS: So there the
correlations never go beyond 1 for h over hb, but
yours does?
MR. WELTER: Exactly.
CHAIRMAN WALLIS: Is that some physical --
MR. WELTER: Yes, there's a significance
in that, --
CHAIRMAN WALLIS: -- peculiarity?
MR. WELTER: -- in that our level in the
hot leg is below the onset level.
CHAIRMAN WALLIS: You get more entrainment
below the level which gives you the onset?
MR. WELTER: Yes.
CHAIRMAN WALLIS: It doesn't make sense.
MR. WELTER: Exactly.
CHAIRMAN WALLIS: When you're actually
entraining you have a lower level than at the onset?
MR. WELTER: Yup. That's what we see in
our data.
CHAIRMAN WALLIS: It seems to be
backwards.
MR. WELTER: Yeah. Because what happens
-- well, a physical meaning, when the oscillatory
behavior is set up, the level in the inlet side is
actually being -- I'd like to draw it. Can I draw
that? A better illustration.
If you look at the mixture level in the
reactor vessel, during steady-state entrainment it's
actually higher than the hot leg. So what's happening
is it's pushing down the level on the inlet. Then the
oscillatory behavior begins there.
CHAIRMAN WALLIS: So how does --
MR. WELTER: So this level is actually
being pushed --
CHAIRMAN WALLIS: So how does the gas get
out of the -- squeezes through out of the hole --
MR. WELTER: It squeezes through there and
it pushes this level down lower below the onset.
That's what our data is saying.
CHAIRMAN WALLIS: Okay. That's a huge
difference. I mean if you've got this factor of two
and a half. It's --
MR. WELTER: Yes.
CHAIRMAN WALLIS: There's very high data
points there.
MR. WELTER: Yes.
MR. SCHROCK: I didn't understand what h
average means.
MR. WELTER: H average is the average --
because we have the oscillatory behavior, we have this
step. There's two levels between -- we have these two
different levels. Average is the average between
these two different levels over time. So if this was
h2, this is h1, h average. H1 plus h2.
MR. SCHROCK: That presumes pretty
detailed knowledge of the shape of that interface in
a horizontal pipe. That's a complicated thing to come
by. How did you get that number?
MR. WELTER: Sir, h average?
MR. SCHROCK: Yeah.
MR. WELTER: Oh, the measurement. We
measure the liquid level here. We have a ring probe
that can measure this liquid level, and we have a ring
probe that can measure this liquid level.
MR. SCHROCK: You mean at -- at --
MR. WELTER: The inlet and the outlet. We
have a measurement --
MR. SCHROCK: Certain axial locations,
both of which are away from --
MR. WU: About 2D.
MR. WELTER: Oh, okay.
MR. WU: But 2D downstream and upstream.
MR. SCHROCK: But why would you think
there would be a correlation of what's happening for
two-phase flow going into the vertical off-take pipe
that depends on the average of those two. It depends
on the local conditions where it comes off.
MR. WELTER: We're not necessarily
presuming that there is a relation between those. We
want to evaluate the model with our data. And this is
the only way to determine some sort of an h, is to use
what we consider an average --
CHAIRMAN WALLIS: What happens if you --
MR. WELTER: What happens if.
MR. SCHROCK: Well, I don't see how it
relates to the correlation that we proposed or to our
experimental data. Our experimental data --
MR. WELTER: That's true.
MR. SCHROCK: -- were for a level h which
is seen visually at -- at the axis of the take-off
pipe, --
MR. WELTER: Correct.
MR. SCHROCK: -- not upstream, downstream.
MR. WU: So when you --
MR. SCHROCK: So how can you plot your
data against --
MR. WELTER: Your level is right here; am
I correct?
MR. SCHROCK: -- our correlation, on the
one hand, or how can you compare your data with our
data when the --
DR. WU: My question --
MR. SCHROCK: -- data-reporting scheme is
totally different?
MR. WU: My question is when entrainment
occurs, you see the liquid level jumping under there.
How can you observe by research just a crack in it to
determine the right under the off-take is a liquid
level. Your visualization window is both downstream
and upstream. And there are small windows there, two
windows.
MR. SCHROCK: Yes, you're right. That's
true. That's true.
MR. WU: So it's not right under -- well,
right under the off-take you cannot get it, that
level.
MR. SCHROCK: Yeah.
MR. WU: So we tried the different ways.
We use the --
MR. SCHROCK: But, on the other hand,
never was there a situation such as is depicted here.
MR. WU: You don't have the jump.
MR. SCHROCK: No.
MR. WU: You don't have the different --
difference. That's a difference of our data -- of
your -- under your data. Your levels, both the inside
and the outside, is the same -- are the same. So --
MR. SCHROCK: But you're comparing apples
and oranges, is what it amounts to.
MR. WELTER: Okay. The purpose of this
slide, I think, is not to necessarily show that the
correlation does not predict our data, but shows the
inappropriate application. The correlation is fine
for predicting the data of your case.
MR. KRESS: How do you determine the
quality? Is that the ratio or the average flows --
MR. WELTER: Yes.
MR. KRESS: -- averaged over the --
MR. WELTER: That's correct.
MR. KRESS: -- time period?
MR. WELTER: That's correct, sir.
MR. KRESS: Okay.
MR. WELTER: That it would be liquid mass
flow rate over the total.
MR. KRESS: Yeah. You know, --
MR. WELTER: And that's injection. So
basically our flow meter is what we inject from the
water pump, what we inject from the air compressor;
we use that.
MR. KRESS: So that's a -- um-hum.
That thing varies with time, but it would
-- as an average it averages out.
MR. WELTER: The quality for the steady
state is quite, quite steady, because we have a steady
injection flow, steady air flow, and they're both
pretty steady over time.
MR. KRESS: Yeah, but there's capicitants
in the system that would mess that up.
MR. WELTER: Okay. We do take an average.
MR. KRESS: Yeah.
MR. WELTER: Yeah, a time average.
CHAIRMAN WALLIS: Well, the Schrock theory
looks pretty lousy compared with the Schrock data,
too. I mean it doesn't predict the trends. And have
you had a --
MR. SCHROCK: Well, I don't understand
that. I mean our data didn't look like that against
our correlation, but --
MR. KRESS: Because you just fitted it to
your correlation, hum? I mean you code-fitted it,
right?
MR. SCHROCK: That's what they're saying
it did, but I --
CHAIRMAN WALLIS: It doesn't look like a
code fit at all, especially in low x3s.
MR. SCHROCK: It doesn't ring any bells
for me.
CHAIRMAN WALLIS: Anyway, we should
probably move on. This just shows that nothing works
very well so far.
MR. WU: That's right.
And also we tried to use the different
levels. The front level and back level, it doesn't
work.
MR. WELTER: At this point I'd like to
turn the model development and conclusion, Section 5
and 6 of your presentation, over to Dr. Wu.
Thank you very much.
MR. WU: As you may have noticed, the
entrainment rate test has a lot of irregularities in
the slug and oscillation. But the entrainment onset
data is grouping very well. At least our data, the
Maciaszek correlation predicted our data reasonably
well.
And also Schrock and Smoglie's
correlations also was predicted their small branch of
data very well.
So we think we have the hope here to do
some more research modeling to bridge these two models
and to try to find the physics behind it.
What we did here is we followed
Maciaszek's approach. Basically what his approach is,
which is the wavy ring, the diameter, you go to the
off-take diameter. And we think the liquid level goes
far away from the break. The wavy ring is getting
bigger and bigger, so we try to use importation flows
theory to find that ring change.
So what we did is we used a mirrored
distributed sink using potential flow to find the
velocity distribution, allowing this X line.
Like Dr. Schrock pointed out, if we can
find that somewhere the velocity goes to maximum, that
means the pressure there is minimum, then we say,
well, the bumps are supposed to -- the wave will crest
-- is supposed to be at that location. So that's our
rationale.
So we write -- we wrote this velocity
distribution using potential for theory. And we tried
to -- well, we kind of get the analytical solution.
So we went to a numerical solution to check it.
The upper corner of the figure is the
velocity at the -- the velocity on the interface
versus over the velocity in the average velocity in
the branch. And this is so a distribution, allowing
the interface, go away from this interface.
We found the crest can never move into
this break. That means when they -- this varies.
This -- that should be -- all you can say is a liquid
very approach to the break, then that mandates you go
to the D. But when the level go away from the break,
then maximum point is drifting away.
So we're trying to find the maximum point,
velocity maximum point. That's the dotted i on the
right side of the figure. And you say, well, that's
when -- that's the limiting case is you go to wall.
That means the diameter of that wave crest, you go to
the diameter.
And the way it's drift away, the
asymptotic condition matches the point sink, mirror
point sink, sink condition. That's one point -- the
square root of 2 of hb.
So that's -- basically you say, well, the
wave crest diameter is equal to the -- A is
proportional, directly proportional to the gas chamber
height.
So if we put this number into the
correlation originally Maciaszek developed, we found
that the number is this term. When the hb over d,
that means the -- a gas chamber height versus the
break side is getting very big, then this correlation
approaches to what the Smoglie and Schrock's
correlation, as it goes to the fifth power.
And when this hb over d approach to zero,
that means the level approach to the break, this term
goes away. Then you get the one here. That means the
correlation approach to Maciaszek's correlation
gathers the one-third power and the diameter of the
bottom.
So we thought that this is a nice approach
to this bridge this small-break correlation and
large-break correlation. And both of them proved
right.
And then we -- we just have one adjustable
coefficient, similar like what they did. With this
coefficient, .5, and the theoretical value of this
coefficient is .4. So we were very, very satisfied.
It brings them together and both are satisfied.
However, when we look at this we still
have a scattering. Again, make -- Kent, Mr. Welter
pointed out, KFK data has a lot of scattering.
Schrock's data and our data has risen in the group
very well.
So if we take out these blue squares, I
think this correlation is -- and we did the
sensitivity analysis. The standard deviation of this,
only Schrock's data and our data, it's like 30 percent
off. By the way, Smoglie's data then, the error is
standard. It goes to like 15 percent.
Well, this approach, like Dr. Wallis
pointed out, is for the flat -- flat top without the
confinement of a side confinement. And our approach
didn't consider the side confinement.
So basically our approach, you say, well,
the infinite place, the liquid velocity is -- the gas
velocity is supposed to be zero. Then for our case
gas is confined in the main pipe. And the infinite
place is so that the gas velocities are supposed to
not be zero. So we modified that.
Then we got a new correlation, is the
same. But it has two adjustable parameters. I don't
like it because if you get an x for parameter, you can
fit everything. So -- but, nevertheless, what is
necessary what's is a collapse of -- it's like a 20
percent standard dev- -- very well the data.
So as a summary of entrainment on the --
before we go further, I still want to go back to visit
this flow regime transition.
And the red line, solid line is the symbol
of Wallis' slug flow transition. And we use a --
since we don't know the -- which flow direction come,
we say, well, all the gas flow in the branch coming
from one side of.
So it's all this in this plot, our data
follows this very well for all regime transition.
And also Schrock's data -- one group of
Schrock's data also follow this line very well. And
the way I checked it, it's the -- the diameter is the
-- the break diameter is about 17 --
CHAIRMAN WALLIS: So I'm trying to figure
this out. Which of these is this slug transition?
MR. WU: The red and solid line.
CHAIRMAN WALLIS: That one that goes
through your data?
MR. WU: Yeah. And then I have a 20, 20
percent. And then a bracket dashed line is the
Smoglie and Schrock correlation.
CHAIRMAN WALLIS: They seem to be just two
different families that are completely unrelated --
MR. WU: You see here you have some...
CHAIRMAN WALLIS: -- on this plot.
MR. WU: Yeah. This plot doesn't have the
diameter effect inside. Well, I show this plot as the
-- one purpose is to say, well, if we treated this as
a small break and this as a large break, I think
anything should be between these two.
So if you run a sensitivity calculation in
your code, you can treat this as two asymptotic
conditions, like we just discussed for the
theoretical. So anything else should happen between
these two.
CHAIRMAN WALLIS: So the mechanism of a
small break is this sort of potential flow sucking out
from the surface. And the mechanism for the big break
is sort of similar, but it's really a civility of the
big-wave criteria.
MR. WU: Interface, because we -- we based
on the interface wave a gross, that delta gross, so
it's -- it --
CHAIRMAN WALLIS: The momentum of the gas
and the -- it's a Froude number in both cases.
MR. WU: Yes.
CHAIRMAN WALLIS: But it's a different --
MR. WU: Yes, sir.
MR. SCHROCK: Now the coordinates on this
graph seem to be the same as in an earlier slide where
you showed the Berkeley data.
MR. WU: That's right, yeah.
MR. SCHROCK: And somehow magically now
it's separated into two groups, which seems strange.
I don't understand how you managed that.
MR. WU: No.
MR. SCHROCK: Coordinates are unchanged,
but now the data seem to plot as two distinct groups.
MR. WU: No. It changed -- this to the
Froude number in the main pipe based on the velocity
in the main pipe, the superficial velocity in the main
pipe, and the Froude number.
In the previous one, your model and --
MR. SCHROCK: Well, you don't put numbers
on your pages, but I go back to one that's got a big
-- three test results, onset of entrainment, -- that's
quite a ways back -- has exactly the same coordinates
as this one.
MR. WU: Yeah, that's right. We used the
two group of --
MR. SCHROCK: FR1, the square root of rho
g1 over delta rho. It's the same thing, but --
MR. WU: Yeah, this figure. Is that what
you...
MR. SCHROCK: That's the one.
MR. WU: Yeah, that's the same -- same
coordinates. Using this Froude number is based on the
superficial velocity in the main line, not based on
the velocity in the branch.
MR. SCHROCK: Well, look, you've got to
define your terms and use notation to convey what you
mean. You can't expect we're going to understand
different interpretations for the same notation.
MR. WU: Well, this is a different
approach. The one we go through -- went there the
model development, that one, is following you and
Maciaszek.
This one is just to show you the finding
we found, to say, well, it's basically -- the
horizontal 9 has a Froude transition for the larger
break. It matches what our argument is, an
entrainment happens, so for the larger break is like
an interface, a wave,, instability on the interface.
And for the small-break case, maybe a potential for
the large going up.
So it's -- eventually I want to say this
is the two-boundary condition. One is to emphasize
what I say, for small break and the larger break, it's
a true asymptotic condition. And the real data should
lie between these two. That's from --
MR. SCHROCK: Let me try one more time.
MR. WU: -- a larger point of view to
argue my point.
MR. SCHROCK: Let me try one more time.
In engineering communications we have
certain principles that have to be followed. And one
of them is that you define your terms clearly. You
set down the notation and define what the notation
means physically.
MR. WU: Um-hum.
MR. SCHROCK: And then you don't use the
notation redundantly. And I think what I've heard you
explain is that FR1 on one graph is different than FR1
on the other graph.
MR. WU: No, no. We use only one FR1
here. We didn't use any other FR.
MR. WELTER: Sir, this is Ken Welter.
These are -- this is for the onset of
entrainment. The graph we showed you previously was
for entrainment rate, so those are different datasets.
This is for the onset of entrainment, this graph. The
graph that we were previously discussing is for
entrainment rate. So they're different datasets, but
it is the same FR1.
MR. SCHROCK: This -- this one is onset of
entrainment.
MR. WU: Yes.
MR. SCHROCK: The one that began this
discussion is number 5, model improvement.
Entrainment onset criterion.
MR. WU: Yes.
CHAIRMAN WALLIS: That's where it's the
same FR1.
MR. WELTER: Could you go to the last
slide that we were at?
CHAIRMAN WALLIS: I think we're mixed up
here.
MR. WELTER: One more.
MR. SCHROCK: Well, I'm -- I'm trying to
resolve in my mind what you've done that produced the
result that --
MR. WELTER: Okay. That was wrong --
MR. SCHROCK: -- that our Berkeley data --
MR. WELTER: Is that right?
MR. SCHROCK: -- separated into two clear
and distinct groups, which I never saw in our data.
CHAIRMAN WALLIS: Where are the two
groups?
MR. SCHROCK: Well, it's the --
MR. KRESS: On this curve you've got --
MR. SCHROCK: It's the sort of ghosty
dots.
CHAIRMAN WALLIS: "...ghosty"?
MR. SCHROCK: Yeah. Light colored gray
dots. There's a set of them on each of those lines.
MR. KRESS: Down here and also down here.
CHAIRMAN WALLIS: Oh, those are Schrock's
up there?
MR. SHACK: Yeah.
CHAIRMAN WALLIS: Well, I think that's a
mistake.
MR. KRESS: There must be some mistake.
MR. SCHROCK: Well, is there a mistake?
MR. WU: No.
CHAIRMAN WALLIS: I'm really puzzled by
these ghostly data. The Schrock data lie exactly on
both curves, so that's pretty well. That's really
strange.
MR. WU: You mean these lines?
MR. SCHROCK: Well, they always lie on
whatever curve you choose.
CHAIRMAN WALLIS: Something is very
peculiar.
MR. WU: Well, this -- this is a different
--
MR. KRESS: It's a quantum effect.
MR. WU: This is based on the main line
superficial velocity. It's not based on the branch
line super- -- I just tried to present this from a
different perspective, from the flow regime transition
perspective. It's different from what we just
discussed about the entrainment from the vertical.
CHAIRMAN WALLIS: Oh, I see what you mean.
MR. WU: Yeah.
CHAIRMAN WALLIS: It's these strange gray
things, though. And Smoglie doesn't have any data.
So those are the Schrock data, those square things, or
those are the Reimann data?
MR. SCHROCK: No. Those are -- those are
KFK data, Reimann and Kahn.
MR. WU: Square KFK data.
CHAIRMAN WALLIS: So the mystery is why
there's some Schrock data on the Wallis line. That's
the thing which is the mystery.
MR. KRESS: There you go.
CHAIRMAN WALLIS: And why this?
MR. SCHROCK: We never would have expected
that.
(Laughter.)
MR. WU: Well, sometimes it has to agree
with you again.
CHAIRMAN WALLIS: On the wrong line.
MR. WU: And this is -- we weighted -- we
just -- what we did is your branch gas flow rate we
are showing as coming from one side. And this data is
a relatively larger break, and it's coming from --
that's -- we didn't do anything.
It's the same abiscus for the previous
sets of data, but we just changed the perspective. We
changed the velocity for -- from the branch to the
main line.
What I would like to say this figure is,
again, I want to say is one-fifth and one-third all
here is for the flow regime transition. It's
represented to boundary condition. I think the --
anything should happen between these two, and that was
what we did to bridge these two together.
So as a short summary of this model
improvement for entrainment, the onset criteria, for
Smoglie and Schrock's correlation, it's based on --
well, Smoglie did -- single-point sink, no interface
effect, and the effect was far from a break or small
break case.
For the Maciaszek, based on Wallis'
interface instability argument, he uses a wave crest
interface instability kind of option and chooses a
crest, wave crest, as the basing and as the break
diameter. And it is effective for the larger break or
the liquid level is very close to the break.
For the new model we proposed, wave crest,
this basing is a function of the onset height. And
you -- and valid for both and it can be reduced to
Maciaszek's situation and Smoglie and Schrock's
correlation's case.
So that's what -- it was a challenge to us
because we cannot come up with something without
considering the previous contribution.
So I think we have the physical
interpretation here, and it's -- the data is -- not
much irregularity there because -- so we think this is
a good model.
For the further improvement, improvement
data, based on what we say is the velocity in the --
gas velocity in the pipe is different from the open
kind of case of flat pipe. So we made a further
improvement.
However, that made the correlation more
complicated. And we have two adjustable coefficients.
And for slug flow transition, I don't know
if we can use it a whole lot. We need to go do a
further analysis. Until we know which side, how much
gas is coming from which side, then we can revisit
that kind of argument.
But for previous KFK data and the Smoglie
data, they -- I show them as coming from one side, but
the aperture is -- the break is so small, so basically
the gas flow velocity in the main line doesn't
contribute too much.
So we -- it is suggested one of the
logical, we just jump -- tried to jump out of the loop
seal, is there any other simpler option for us to
take. That's it.
For the model improvement of entrainment
rate, again we -- we have -- we follow the similar
approach, argument of h over hb, the actual gas
chamber height versus the entrainment onset.
The rationale, as I say, any excess of
this kinetic energy of a gas or that contribute to the
pressure difference from the interface to the break,
overtakes the gravity. That excess of kinetic energy
is going to the liquid kinetic energy.
And when this liquid velocity and mass
velocity is equal to zero, left aside, this equal to
zero, that gives us the entrainment onset condition.
Using this argument, we derived -- the equation
quality in the branch is equal to this function. It's
a function of density ratio. And the function for the
h over hb, plus there's another one, is the diameter
effect of the break.
MR. SCHROCK: Is that derivation available
to us?
MR. WU: Yes. Yes. It's simply, just put
that k in front of this group, the right-hand group.
Then we can straightforward again.
CHAIRMAN WALLIS: Then you must use some
kind of a one-dimensional theory, or something,
because --
MR. WU: Yeah. The --
CHAIRMAN WALLIS: Or does the k take
account of two dimensionality, or...?
MR. WU: No. The only thing coming from
this part, that's the -- say were the -- we are shown
the -- a liquid of void fraction in the branch is a
function of h over h -- 1 minus of h over hb.
CHAIRMAN WALLIS: But these Vf3s, Vg1s,
these are averages across the whole area. It's a
one-dimensional --
MR. WU: Yeah, that's right.
CHAIRMAN WALLIS: -- approach.
MR. WU: That's right.
What --
MR. SCHROCK: What I asked is, is the
derivation available to us? Can you tell us, are we
going to have that derivation? Is it in a report that
we're going to get?
MR. WU: I can do it right now here, if
you --
MR. SCHROCK: Hmm?
MR. WU: I can do it right on the
blackboard, if you prefer.
CHAIRMAN WALLIS: Now is this compared
with data somewhere?
MR. WU: No.
CHAIRMAN WALLIS: The new --
MR. WU: It cannot solve that jump. So
what we compare with Smoglie -- Schrock's correlation,
for the D -- this -- this 9 can be treated as you have
fixed the gas flow rate. And as you change the liquid
flow rate to change the quality.
In such a case the hb is a constant
because that's based on the gas flow rate. And you
see the diameter ratio is when this is about
one-hundredths of the hb should be, that's very close
to Smoglie's situation.
The KFK correlation actually exactly is
shaped like this. Unfortunately, we didn't put on
that figure because it has the densi- -- has several.
I don't know. Previous, last year we put on a figure.
This year we -- I can't -- in a final one,
I will show you how this -- and also for Schrock's
data, it's about 17 to 30. So this is a tenth of it.
So it's -- the correlation goes through it.
When that break is getting bigger, then
it's going up like this. It's more like our data
case.
CHAIRMAN WALLIS: You said there was
Smoglie correlation down the bottom there that we
can't see?
MR. WU: Smoglie, no, we didn't put it
here.
CHAIRMAN WALLIS: You said he was close to
the bottom curve?
MR. WU: Yeah, that's right.
CHAIRMAN WALLIS: The .01h?
MR. WU: Yeah. It's the --
CHAIRMAN WALLIS: Smoglie is down there?
MR. WU: They have a data only in this --
in this shortened mix, very high quality data. And
their correlation, all of the way we extrapolate it to
zero quality. That was in Schrock's report, too, to
mention to the KFK.
And this correlation doesn't express one
thing -- let me -- okay. In our data we have a jump
like this. What this give us the trouble is for one
h liquid level you have two qualities. That means
using there's -- like you said, before you mentioned,
maybe it's related to liquid flow rate or gas flow
rate explicit and beside the hb.
And otherwise, only use the information of
h, we can't gather this bump like that. And that's
what we are working on, trying to see averages.
You will see later a behavior, see if we
can -- and amazingly we -- we found these -- out of
the turbulence here you will see -- in this region is
actually -- I will say the ration occurs (phonetic).
And the way it's flattened out is like a high gas,
very -- no liquid flow rate. The oscillation
disappear.
So the bump itself, it's coming from that.
It was initiation behavior. And we try harder to get
an average parameter to represent it. And we are
still working on.
CHAIRMAN WALLIS: Are you sure there's a
curve. Earlier where h over hb was bigger than one.
MR. WU: Yes.
CHAIRMAN WALLIS: And here it's less than
one. Are these different data or something?
MR. WU: No, no, no. Then we use the
average h -- h in that previous --
CHAIRMAN WALLIS: In the other one. No,
this is the h on the --
MR. WU: We are trying to say this is the
high side.
CHAIRMAN WALLIS: So this is the h on the
reactor side here?
MR. WU: Or 4 times vessel size.
CHAIRMAN WALLIS: Vessel size.
MR. WU: Oh, no, the steam generator size.
That's the higher part, because it was originally --
when you see -- when you saw the experiment at lunch
time, there was initiation actually occurs downstream
of the branch.
CHAIRMAN WALLIS: So if you're going to
use this in a system-solving computer model, you'd
have to somehow predict h over hb, then you'd predict
h -- x3 from that; is that what you'd do? And the
problem with yours --
MR. WU: This is the traditional approach.
CHAIRMAN WALLIS: The problem with your
curve is you don't know which one to pick. And the
fact that it's going to curve up again, you get three
xs for the same h or hb.
MR. WU: Yes. So that's got to be related
to the gas velocity or liquid velocity. That's what
we are -- we are trying. Right now we only use this
simply -- simple representative. If you don't
consider the other, then you've got this -- you cannot
find this bump, and you can't -- and that means use h
over hb and x as a correlation, you miss something
important.
CHAIRMAN WALLIS: Well, my comment in all
of this is what I saw in the experiment. It seems to
me you need a dynamic analysis, so the rate of
build-up of liquid in the plug which goes to the steam
generator, when it comes back, you sweep some out the
pipe, how long it takes to sweep that out depends upon
sort of the length of the pipe to the -- to the
air-water separator, or something.
All these things are very system
dependent, aren't they? So you really need a system
model in order to predict the entrainment.
MR. ROSENTHAL: Yeah. And why don't we
let him finish the presentation, and then Steve can
make some comments about our intent, how we can
ultimately use this and track that.
MR. BAJOREK: One question on that last
figure. The bump there, --
MR. WU: Yes.
MR. BAJOREK: -- does that include the --
both the -- with the block steam generator and without
the steam generator? Is that all the data together?
MR. WU: Yes. Waves of -- but waves of
the steam generator with a returning line, it's
calmer, but it still have a bump.
MR. BAJOREK: Would you still get that
bump --
MR. WU: Even Schrock's data has a small
bump there. If we go back. Please, go back. Go
back. I think go more, just go ahead. Go. Go. Go.
Okay.
You see this, you called it ghostly. It
has some...
CHAIRMAN WALLIS: This is the one where
the Schrock correlation has no relationship to his
data, or not much.
MR. SCHROCK: I haven't gotten it figured
out yet.
MR. WU: It's actually published in the
left --
MR. BAJOREK: Well, you see the hump even
more when you plot it in what you call h2, in the
other one. That is when you don't use the average the
hump is even more.
MR. WU: That's in your --
MR. BAJOREK: Yeah.
MR. WU: -- actually we tried to see which
side we're going to use. When we use h2 we brought it
down. But in this one we didn't know it.
In your handouts actually have a lot --
has a lot -- another figure. It's -- I use a steam
generator side of the gas chamber, and it can bring it
to about below one.
Well, we can go back.
MR. SCHROCK: Well, my recollection of our
data is that they did not look like this against our
correlation. I'll have to dig that out and refer to
it to understand what's wrong here.
MR. WU: Shall I proceed?
As a summary, we built our database that
actually covers all the branch separation cases. And
we picked out the material related to the vertical
branch entrainment. And we found the data is -- the
correlation works for small breaks, but it doesn't
work for the larger break.
We build our percentages relatively
complicated in that like Dr. Schrock and Dr. Wallis
pointed out. And it's amazing, we run a test of
entrainment now, said the test, the correlation
actually two ends meet -- it's -- there's not much
irregularity there for entrainment onset case.
And in that length we varied the form to
.7 to 4.7. It doesn't have much impact within the
test range. And also the downstream structure or
steam generator, with a steam -- with a steam
generator and without a steam generator. That means
we're being flooded.
It does have the effect because of the
wave bouncing back and so of course the entrainment
and that lower. But the gas flow rate direction from
the downstream in this case has negligible effect, but
we cannot quantify it. We didn't put a flow meter
there. And that's what we are doing right now after
this.
And for the steady-state entrainment case,
a downstream structure or steam generator affects the
entrainment rate. And the gas flow from downstream
changes so the entrainment rate is substantial. And
also we need to quantify how much a gas flow from the
downstream side and get a better data.
And for the model evaluation, the model of
Smoglie and Schrock, this is used in RELAP5, is
effective for relatively small breaks that was
evaluated from their experiment data.
Maciaszek's model works well for larger
breaks. That's what our case.
And for the entrainment rate model, model
of Schrock that is used in RELAP5, it seems it does
not collapse our test data. Like Dr. Schrock pointed
out, what hb is going to use, we don't know what hb is
going to use because in the new phenomenon you have
two levels there. And we use either one of them and
we use the average of them, still can work.
So we think there is room to do some more
work if we want to predict this phenomena using our
system code.
In the model improvement for the
entrainment onset model, we did a potential flow
analysis. We found Maciaszek and Schrock, Smoglie's
correlation can be interrelated if we consider a
wave-crest size as a function of the liquid level
height. And it's two asymptotic condition anything
have, and it should be covered within this -- within
this range. So it's one --
MR. SCHROCK: I'd like to make a comment
about one of your conclusions. And that is that the
correlation form that evolved earlier is okay for
small breaks but not for larger breaks.
I think that that conclusion is misguided.
And I say that because having seen the experiment in
operation now I'm convinced that you're dealing with
a completely different phenomenon that had been
addressed in our work and in the work at KFK, totally
different phenomena involved.
I do believe that if we tested with a
larger diameter break on our apparatus that we would
get consistent results with those already taken in
that apparatus.
So I would not conclude from the
combination of what you've learned from what we did,
what was done at KFK, and what you've done in these
experiments, that it is the diameter, the larger
diameter of the break line that causes disagreement
with the correlation.
The fact is if -- if you had the kind of
surging, pulsating slugs of liquid moving back and
forth in the test section with the smaller diameter
breaks, you would not expect the results to agree with
the correlation that we've developed for the
stratified flow case.
MR. WU: So, well, the oscillatory effect
only happened --
MR. SCHROCK: So your conclusion I think
is wrong and for the reasons that I just stated.
MR. WU: The oscillation for the
entrainment onset, you don't see oscillation. When
you went to see the experiment for the onset, it's a
little bit wavy interface. There's no oscillatory for
onset correlation. There's some -- no such a
complication or irregularity.
Only for the entrainment rate tests we
observed this step phenomena. So you try to mix these
two together to justify our conclusions, I don't agree
with that.
MR. SCHROCK: Well, maybe you can make
that apparatus produce a smooth stratified interface.
What you showed us did not include that kind of
interface, did not.
MR. WU: No. It -- for the entrainment
onset we don't have --
MR. SCHROCK: Yeah.
CHAIRMAN WALLIS: Even for onset it was
not smooth.
MR. SCHROCK: No.
MR. WU: And it's a little bit wavy, but
you don't see the jump like that. One side is
substantially higher than the other side. And we
measure, we use our probe with the measure to --
MR. SCHROCK: The onset that you
demonstrated was distinctly pulsating. It was not a
more or less continuous two-phase flow into the break
line. It was a highly-pulsating flow.
MR. WU: No.
MR. SCHROCK: That's what I saw.
MR. WU: Onset, there is nothing happened.
What we called the onset is the pulsating, everything
stops. There's nothing being drawn into the branch.
If you see the pulsing, that's still being entraining.
That's in the process. It's not stopping.
So if that's the case, then we -- it's not
our entrainment onset and measurement yet. Only when
it stops, that's what our value of entrainment onset.
When you say something's been drawn into
the branch, the entrainment is still there. That's
not our entrainment onset condition.
So when you say, well, say later and
there's some pausing and being pulled out, that's not
the entrainment onset condition yet.
MR. KRESS: Well, you do have -- as you
approach the condition of no entrainment, you do have
wavy surfaces.
MR. WU: Yes, that's right.
MR. KRESS: And what he's saying is even
then your entrainment is possibly not the same
mechanism as his was. So when you stop that, --
MR. WU: Yeah.
MR. KRESS: -- you're stopping something
different than what his is stopping.
MR. WU: If you have like 10 meters --
five meters of gas blowing over a surface you don't
expect that surface is calm. There is a capillary
wave which is being developed there. So that one, if
you say that's the case, that we cannot make it a --
CHAIRMAN WALLIS: Well, I think the waves
are coming from the reactor vessel.
MR. WU: Yes.
CHAIRMAN WALLIS: And you have this
bubbling and frothing, and there's sort of a big plume
--
MR. KRESS: That's right.
CHAIRMAN WALLIS: -- of stiff arising
which is stirring up the surface. And that goes into
the pipe, which is much bigger than these capillary
waves.
MR. SCHROCK: You said you've simulated
AP600, but in fact -- you say it's scaled to AP600 --
but in fact I think these big disturbance waves that
are entering that horizontal pipe from the vessel
depend significantly on the geometry of that entry.
You've joined two cylindrical surfaces
with sharp edges. That doesn't exist in the reactor.
MR. WU: I agree.
CHAIRMAN WALLIS: Now when you're
correlating your rate of entrainment, are you doing it
with this one-inch bypass, and so on? You're not
doing it with the closed end, because you get
different answers.
MR. WU: We did close it. We opened it,
opened --
CHAIRMAN WALLIS: But your correlation is
for the open end with the bypass?
MR. WU: Yes. We did open. We did close
it. And we did it with one-inch --
CHAIRMAN WALLIS: Yeah, but your
correlation that you're offering --
MR. SHACK: No, but it's for the steam
generator. Sometimes you have the loop seal and
sometimes you don't, right, but don't include the
blind data.
MR. WU: Oh, we close the -- with the
steam generator there, we close the three-inch and the
one-inch. That's what these -- that's with the
loop-seal case.
CHAIRMAN WALLIS: Yeah, but that's not
what's being correlated. The data for entrainment do
not include the one where you shut off the --
MR. WU: Can you go back? Go back to the
--
CHAIRMAN WALLIS: You can't, because
they're two different groups. I mean you can't
correlate the same thing --
MR. WU: No. That one is blind flooded as
dif- -- if we blind flood it, it's then coming from a
different group. For the other case with the steam
generator, we have three case. One is a three-inch
and line open, one is one-inch line, and all of the
line being closed. That's all the line closed --
CHAIRMAN WALLIS: But these data here are
for the three-inch line open, or the one-inch line
open, or something?
MR. WU: Go back.
Both -- all of them group with --
together. There's no effect --
MR. SHACK: Well, the peak -- the peak
there is for the closed-return line, right?
MR. WU: No. You are talking about an
entrainment onset or entrainment rate?
CHAIRMAN WALLIS: Apparently this is a
phenomenon even with an open line.
MR. WU: So this case is --
MR. SHACK: But for -- are we talking
about entrainment rate, because --
CHAIRMAN WALLIS: Yeah, this is
entrainment rate here. Yeah.
MR. WU: Are we talking about the
entrainment rate. If it's entrainment rate, this bump
will come --
MR. SHACK: When I look at the one with
the data on it, I mean I see that huge peak with the
--
MR. WU: That's for the --
MR. SHACK: Isn't that really with the
closed line?
MR. WU: Yeah, that's for the closed line.
Yes.
MR. SHACK: So he gets two -- he does get
distinctly different results with the line --
CHAIRMAN WALLIS: Well, the steam -- he
even has this awkward one with the open line.
MR. WU: I got a little bit like that, but
what I -- when I did the conclusion I was the first to
mention the entrainment onset correlation first.
CHAIRMAN WALLIS: So when we saw the
experiment this thing was chugging like a steam
engine, the thing there.
MR. WU: Yeah.
CHAIRMAN WALLIS: That was with the closed
-- something was closed?
MR. WU: Closed the returning --
CHAIRMAN WALLIS: But it's not -- it's not
ended. There's not a plug. You actually can go into
the steam generator and come back again?
MR. WU: That's right. That's right.
CHAIRMAN WALLIS: So it simply means you
close the valves in the three-inch and one-inch line.
MR. WU: That's right.
CHAIRMAN WALLIS: It's not as if you --
MR. WU: No, see -- okay.
CHAIRMAN WALLIS: -- block the end.
MR. WU: No. No.
CHAIRMAN WALLIS: So there's a big
difference between having those valves closed and
having them open.
MR. WU: That's right.
But that doesn't affect the entrainment
onset data.
CHAIRMAN WALLIS: No. No, no. We're
talking about rate.
MR. WU: Yeah. That was my conclusion
when --
CHAIRMAN WALLIS: We should probably go
on.
MR. WU: Yeah. For the entrainment com-
-- go to the previous page, please.
So for the entrainment onset model, again
the model of Smoglie and Schrock uses the RELAP. That
is used in -- RELAP5 is effective relatively for small
breaks. And Maciaszek models works well for the
larger break.
And for the entrainment, we already
covered that.
And the model improvement, we see the
entrainment onset model. What we did is we bridge
these two together. And we try -- we figured out a
way to sink that wave crest. The maximum pressure
point, it's a ring type. And that is a function of
the liquid level. And the --
CHAIRMAN WALLIS: I think you should --
excuse me. You should do an experiment where you
don't put the air in through the bottom of the reactor
and bubble it up, but you put it in through the head.
You have the same flow rate, but you wouldn't be
bubbling it through the surface and disturbing the
surface. You'd get a very different answer probably,
or a different answer.
MR. WU: You mean...
CHAIRMAN WALLIS: Depending on how you put
the air in.
MR. SHACK: Just do a genuine gas flow.
CHAIRMAN WALLIS: I mean the gas flow's
more like -- it's just more like what Schrock did.
MR. SCHROCK: Yeah.
MR. SHACK: A separate effects test.
MR. WU: Yeah, we can run that. That --
yeah, we were -- originally we run this to simulate
the prototypic. Yeah, we can run that. We put up a
-- bring on that --
CHAIRMAN WALLIS: And you'd probably get
a different group of data.
MR. REYES: Jose Reyes of Oregon State.
Just a brief comment.
I think one thing that is happening with
this data and what we've seen in the APEX test this
morning during the 25 uncovery series is that this
data matches our test.
And so we're seeing the same kind of
dynamic behavior that's giving us this over -- an
increase entrainment rate in their facility as we see
in ours.
CHAIRMAN WALLIS: What you're -- he's
correlating something which is very much like AP600 in
terms of end condition.
MR. REYES: That's right. That's right.
CHAIRMAN WALLIS: Right. It really
doesn't apply to a separate-effects type of thing.
MR. REYES: That's right. It basically is
-- geometry is -- it's a realistic geometry to try to
predict what's going on in a very specific case, the
AP600.
MR. SCHROCK: Well, it's superficially
realistic, but I mentioned the differences in the
entrance into the horizontal pipe. Sharp edges versus
a well-rounded entrance. It makes a lot of
difference. And I don't think you want to sweep that
under the rug.
MR. REYES: No, no. I think you're right.
There are some geometry differences
between our facility and even in the air warp
facility. We have upper internal structures, for
example. So I know that changes some of the
entrainment rate behavior.
MR. KRESS: Yeah. The real question is
for this kind of phenomena of entrainment, have you
used the right scaling parameters with respect to
AP1000.
MR. REYES: Right. And --
MR. KRESS: You may have scaled the wrong
things, because you weren't thinking of this
phenomena.
MR. REYES: Yeah. What we measure -- so,
for example, we do see a hydraulic jump in our -- in
our hot leg, just like they see, because we do measure
level on both sides. So we're seeing familiar
phenomena.
Now the question is let's look maybe more
closely at that to determine: Is that what we'd
expect to see for the AP600.
CHAIRMAN WALLIS: Well, in your results
for your APEX facility, you did see oscillations in
that --
MR. REYES: Absolutely.
CHAIRMAN WALLIS: -- that relief line.
MR. REYES: There was a chugging behavior.
CHAIRMAN WALLIS: ABS-4 relief line.
MR. REYES: In fact, we put a transparent
line in a portion of that.
CHAIRMAN WALLIS: And isn't that -- isn't
that because of the slugs, and the pulses of liquid,
and all that stuff?
MR. REYES: Right. You see -- you see --
you don't see a stratified smooth interface. You see
a very wavy energetic interface.
MR. SCHROCK: Well, given that, it seems
unlikely that an analysis based on the cartoon in page
number 5 model improvement should succeed, that that
would be very surprising and yet --
MR. REYES: Yeah. I think what we're
seeing --
MR. SCHROCK: -- you're saying essentially
that fortuitously it does succeed.
MR. REYES: Yeah. Yeah. I think what
we're seeing is that we're hitting a limit like a
slug. In essence, he's looking at a transition limit.
And I think that might be what's allowing us to
collapse the data at that higher point, which was a
surprise, I think, even when Dr. Wu looked at --
MR. SCHROCK: I want to ask, once again,
to have a copy of the derivation. Don't take the time
to put it up there, but I'd like to see that, please.
MR. WU: In consideration for future
effort, it's the test of liquid gas flows through the
main line is more general case or more general or --
okay.
A test for the smaller main line for
probably a noninnerflow case, because we are running
only the intermittent flow case. A test of ward down
-- downward break, branch and break, and also will
have an effect of gas flow from the downstream and the
effect of gas flow through the main line. And so we
just think about the possibilities. It's not to say
we are going to do it.
And one more phenomenon we would like to
point out is the pool entrainment. In our case when
the mixture level is way below the hot leg in that,
it's about, say, six inches below the hot leg in that.
Under the entrainment, pool entrainment, pool droplet,
and since this hot leg is the ADS-4 line is very close
to the inlet, some droplets go in through this ADS
line and have been transported to the upper plenum.
So we think the pool entrainment there is
one of the mechanisms, those with our inventory. And
we have one still. Right now it's trying to use the
existing model to predict this flow rate we measured.
And APEX was also run the data in last
year and also found the entrainment down continues so
when the mixture level is below the hot leg.
That concludes my presentation. Thank
you.
CHAIRMAN WALLIS: You have no theory to go
with this? This is just an observation?
MR. WU: We have some models selected from
publication and we are trying to simplify it, because
it's --
CHAIRMAN WALLIS: It looks like a very low
flow rate, what, .01 kilogram a second. Is that --
that's about a pint a minute, or something.
MR. WU: Yeah. Yeah, that's right. But
as it now running it goes... It's a mechanism.
MR. KRESS: Is that --
MR. WU: Those -- those are water
inventory.
MR. KRESS: Is that kind of entrainment
very important for the nuclears' side?
MR. WU: I think so. I think so because
that carry the droplets is not vapor going -- going
through the break. So you actually know the water
inventory faster than you just depressurize the vapor.
So that actually should be considered, I think.
MR. KRESS: Yeah, that looks like it's
such a low rate of liquid being lost that it wouldn't
--
MR. WU: That's a per second --
MR. KRESS: -- wouldn't -- it wouldn't
impact the rate at which you lose inventory by
vaporization. It seems like it's very, very
negligible compared to the vaporization rate in terms
of mass. So it may not be important for accident
sequences, it seems to me like.
MR. WU: We need to calculate the numbers
to this 1,000, like a kilogram per thousand. You're
right.
MR. BAJOREK: I think the gas flow right
there may also be low.
MR. WU: Um-hum.
MR. BAJOREK: We've run into this type of
a problem before at Westinghouse at hot-leg
switchover, where you have a period where you may not
be putting as much liquid into the vessel as you
normally would and you have a period of time where
boil-off and potential entrainment could drop that
level. It's only two or three feet down to the top of
the core. I think the gas velocities are also fairly
low there. If that goes up the liquid entrainment --
MR. WU: Yeah.
MR. BAJOREK: -- would also go up quite a
bit.
MR. KRESS: In the business of how
suppression pools extract aerosols from steam rising
up through it, one of the problems is how much liquid
carryover you get, because it carries over part of it.
And there's some data in that field. And a lot of it
has to do with the bubble size that goes up through
and --
MR. WU: That's right.
MR. KRESS: -- it's the --
MR. WU: A burst.
MR. KRESS: -- it's the film -- it's the
film that breaks, that kicks up -- kicks up the stuff.
MR. WU: That's right. The burst.
MR. KRESS: And there's basically two
populations of droplet sizes that's seen. But you
might want to look into that field --
MR. WU: Thank you, sir.
MR. KRESS: -- to see.
MR. WU: Thank you.
And also I think if we want to do a
thorough investigation, maybe we need to add the
internal, because droplets, or some of them are
deposited down in the re-entrainment, and so we need
to see.
Thank you, sir.
CHAIRMAN WALLIS: Thank you.
Are we ready to move onto pressurized
thermal shock? Any more questions or points about
this program?
MR. ROSENTHAL: Yeah. I thought that the
staff ought to make a summary statement about how we
intend to use the work and --
NRC STAFF REPORT ON INTENTIONAL USE OF OSU WORK
MR. BAJOREK: Sure. I guess first this
program initiated a couple of years ago. It was in
response to, I think, largely an ACRS concern and also
a co-development concern that the entrainment model,
the face separation models in RELAP and other codes
were deficient.
It was a problem in AP600 that was
resolved primarily because there was so much other
water in the system for the conditions, the power
level of AP600, you wouldn't suspect entraining so
much to challenge the top of the core.
It remains an important problem. With a
high uncertainty in AP600 it will become more of a
concern and problem in the AP1000 when the gas
velocities go up.
Our long-term intent is to use this data
and similar data to develop better models for TRAC and
RELAP. Clearly we're not there yet.
In looking at the data, what we saw in the
lab, there's still a lot of problems in our
understanding of what are the true physics of
entrainment and what's going on in this system right
now.
It's not solely entrainment off of a very
steady interface that is dominating the physics, but
it's clearly related to system effects, dynamic
oscillations in the hot leg.
They may be affected by geometry and
scaling of the facility itself. The split between
what goes down the hot leg from the vessel versus the
steam or air that goes through the rest of it and the
size of these waves relative to the size in the pipe.
It's not clear that we've really addressed
that total scaling issue at this point.
Our long-term intent, however, is to
continue to use this data, to look at the data, to try
to develop better models out of it; potentially to
refine the facility; and to deal with and better
understand the system effects before we get a model
that's going to be put into TRAC and/or RELAP.
As Dr. Wu mentioned, the long-term goal
does go beyond AP600. We've talked about that and the
AP1000.
Your limiting small break usually is
determined from a small branch line at the bottom of
the cold leg. That type of an orientation.
If we're ever going to be successful with
code consolidation and improving the codes, we're also
going to have to understand the side-oriented branch.
Practically that's important in several plants.
There's not a whole lot of those that have side
orientations, but there's a lot of experimental tests
primarily in the ROSA facility that would be
absolutely vital for co-development.
That unless we can get the side branch
correct we'll be forever dealing with the
compensating-error issue. So we need to be able to
get the top branch, the side branch, and the bottom
branch right in the long-run.
MR. SCHROCK: Well, we have data for the
side branch. But I think that, again, you'll find
that if the upstream conditions in the actual system
looked more like in the OSU mock-ups of the system,
that the data would not be applicable.
I certainly wouldn't want to argue those
data for anything that looks like this sort of
churning, pulsating flow in that horizontal pipe.
CHAIRMAN WALLIS: Okay. Do you want a
break?
MR. ROSENTHAL: I think it would be fair
to the presenters, et cetera, if we could take a short
break, if you wouldn't mind. I mean it's your
meeting.
CHAIRMAN WALLIS: Well, we just got back
from lunch. So I was thinking of having a break at
about the time that was originally anticipated, --
MR. ROSENTHAL: It's your meeting.
CHAIRMAN WALLIS: -- halfway through,
otherwise -- see, we're a little bit -- we are late.
I mean you can get up and walk around, whatever. We
don't have to sit here all the time.
MR. ROSENTHAL: I will.
CHAIRMAN WALLIS: Okay. I think that
Jose's been dying to get up there.
PRESSURIZED THERMAL SHOCK RESEARCH
MR. REYES: Okay. Now I'll get into the
pressurized thermal shock research. This presentation
is to give you a -- I've actually combined three of
the presentations to save a little time.
That includes the research objectives, a
little bit of the prior work that's been done on PTS,
and then a discussion of what test matrix was
performed.
So I'll to jump straight into some of the
-- does this go backwards.
So we'll talk about overview; give you a
little bit about our research program, what's been
done; and then I'll mention what tests have been
performed and what calculations have been performed,
and just a brief summary. So this is by way of an
introduction to what you'll be seeing in more detail.
So each of these areas will be discussed in detail by
the different presenters.
CHAIRMAN WALLIS: And a lot of this stuff
we won't see 'til tomorrow, right?
MR. REYES: Correct. Correct.
Now on the overview there are 10 slides on
what's been done in the past. And I think this is an
optional area, and I'll leave it to the discretion of
the Chairman. We --
CHAIRMAN WALLIS: Does it matter? Does it
matter to the present?
MR. REYES: I think we can jump 'til -- we
can jump to slide 12, and we'll gain some time.
Basically -- what it does is discuss what's been done
in the past.
And then beginning on slide 12 there it
discusses the results of my review of the previous
work. Does that sound fair?
CHAIRMAN WALLIS: Okay. That's fine.
Let's do that. Let's do that.
MR. REYES: That'll advance us --
CHAIRMAN WALLIS: I think that we were
most interested in what you have done.
MR. REYES: Okay.
CHAIRMAN WALLIS: And then at the end we
might see what everybody's done and tells us about the
conclusions we should be drawing.
MR. REYES: Right. So let's -- we'll jump
to page number 12, please. And so basically what I've
done is I've gone through and looked at what's been
done in the past. And of course I already had some
familiarity. And this would be the results of my
review of the previous research.
So I looked at integral system research
that had been done previously. Well, the integral
system work really was related to calculation. So
what we had were calculations performed by TRAC and
RELAP for the Oconee and the H. B. Robinson and the
Calvert Cliffs plants. So all we had were
calculations. There was no integral system test data
available specifically for the PTS scenarios of
interest.
So as a result of the review of the prior
work there was a need to benchmark the system analysis
codes to determine their ability to predict loop
stagnations, train the system pressure and downcomer
temperatures. So we'd like to have some benchmark
data for those codes.
We'd like to be able to integrate the
separate effects test results with the integral system
behavior. What we had were two sets of experiments.
We had -- well, actually all we had were the
separate-effects experiments. We didn't have integral
system tests.
And I think what you'll see is that they
do link together and that some of the separate LOCA
behavior affects loop stagnation, which is an integral
behavior. And so we'll be talking about that later
on. So that was one of the motivations for the
research.
Also there was an effort to examine the
effect of core heat, heat transfer on downcomer fluid
temperatures. The pre-separate-effects tests were --
essentially modeled the cold leg, loop seal,
downcomer, the lower plenum, and then they had a stand
pipe. So there was no core sitting behind the barrel
wall to heat up the core barrel wall.
So heat transferred from the core barrel
to the fluid is going to be examined also in this
study. So that was another thing that we noticed that
was missing from past research.
In the area of separate-effects testing we
need to obtain some fluid mixing data for low-flow
HPSI operation in a side injection cold-leg geometry.
So we didn't find in the existing literature any
low-flow HPSI for a Palisades-type geometry. So we
actually have done some flow visualization testing,
and we'll demonstrate some of that for you tomorrow.
We also saw the need to develop a
criterion for the onset of loop seal cooling. In
these plants you actually have a -- as John will show
in a little bit -- you have a reactor cool pump with
a lip on it that prevents the -- that keeps a layer of
cold fluid on the bottom of the cold leg essentially.
And when that spills over it has a
particularly important effect in cooling the loop seal
and it affects the loop behavior. So we need a
criterion for that.
Also we want to assess some of the advance
CFD Code capabilities. We've seen a tremendous
increase in speed. Back in '85 when we were trying to
run SOLA-PTS and some of the other CFD Codes, it was
painstakingly slow.
And the nodalization, we were able to get
maybe 4,000 nodes in the downcomer, and it was taking
10 hours to run maybe 10 seconds of transient. So it
was -- and of course at the time we needed to perform
amounts of 200 transients for 7,200 seconds apiece.
And you do the math and you figure, well, not in my
lifetime, I don't think.
So -- but what we've seen now is the CFD
Codes have been -- are much more robust now. They are
-- the computational speeds are much better. And
you'll be seeing some fairly-heavily meshed systems
later on in the presentations.
Okay. So now we set up this research
program here at OSU. We want to perform some integral
system and some separate-effects tests to address the
earlier research limitations. We've got -- we've had
a very good cooperation in place. It's worked very
well. From the NRC Research, Dave Bessette, Gene
Rhee, Sarah Colpo, Chris Boyd. Those folks have been
very supportive of what we're doing.
They have allowed us to -- it's allowed us
to be able to work with Oak Ridge and with Consumers
Energy -- I was misspelling it. Oak Ridge of course
is had interest in where we put our thermocouples to
measure our temperatures. And they have been helpful
in providing input in that area.
Consumers Energy, the Palisades Plant,
they've been very helpful in providing plant data.
And they've also done some other work in having us
have discussions with their operators so we know how
operators realisticly respond to main stream line
breaks and small break LOCAs. And we'll talk about
that a little bit more later.
So this is by way of the structure of our
research plan. And it might be hard to read on there,
so I'm just going to talk about the big, the overall
headings.
We essentially started off with a review
of what had been done. Did a scaling analysis to see
whether or not our APEX facility could be modified to
give us behaviors or to produce a geometrically
similar system to the Palisades Plant. That was the
plant selected.
We found that it was geometrically
similar, remarkably so, that the CE 2-by-4 design is
very, very similar to the AP600 design.
So in scaling that cross-section flow
airs, the volumes, we're essentially using a constant
factor all the way around the loop, so it was very
nice.
We did some facility modifications, and
then we were able to perform our testing. We've had
two types of tests, as I mentioned before, integral
system tests, and then we also did some
separate-effects test.
In the area of thermal hydraulic analysis
we're using RELAP5, the NRC version, the Gamma
version. And we're using REMIX and STAR-CD to do more
detailed LOCA separate-effects types of analyses.
So what we plan to present to you is the
summary of all this work in each of these areas and
try to provide you with some result in each of these
areas that will be helpful to the overall PTS
reevaluation.
So I mentioned the scaling analysis. I
won't get into a lot of details. We established that
the degree of geometric similarity between Palisades
and APEX was -- the plants are similar geometrically.
We developed a detailed scaling basis for
looking at the -- for assimilating the onset of loop
stagnation. So our flows, and our injection flows,
and our cold-leg flows under natural circulation
conditions are maxed so that we're able to get the
onset of loop stagnation under the same conditions.
The onset of thermal stratification, the
cold legs. This has been interesting. We've learned
a lot from our tests here. Both the flow
visualization and the APEX-CE test. We'll talk more
about that.
We have also did some scaling in the area
of thermal fluid mixing. And that also required doing
some separate-effects tests in APEX. We did some
very, very simple benchmark tests on those.
Now we developed scaling bases for
comparing the main steam line break and a small break
LOCA in our facility to Palisades. And we also
identified which of the PTS PIRT phenomena would be
adequately simulated in APEX-CE.
Facility modifications. The APEX design
was modeled, of course, after the AP600. What we've
done is we've added loop seals to this design to
simulate a Palisades plant. We've changed the
configuration of the cold legs to simulate the
Palisades Plant.
We've eliminated all the -- in essence,
all the passive safety systems of the AP600. And
we've changed the logic. We've isolated all the logic
of the AP600 to come up with a design that was similar
to Palisades.
We've also added four safety injection
lines on the cold legs in prototypic geometry,
including the check valves on those lines, to simulate
the inactive emergency core cooling system.
The types of integral system tests that we
performed, we -- in all the tests we measured
downcomer fluid temperatures and of course the
corresponding system pressures. And we did a series
of small break LOCAs and what are called excess steam
demands on main steam line breaks, a stuck-open
atmospheric dump valve test.
And in these tests we've tried to identify
the conditions that lead to primary loop stagnation.
And we've nailed those down pretty well, I think.
This is our test matrix just for the
integral system test. We start off very simply with
just a natural circulation flow benchmark test, making
sure that all of our loop resistances were similar to
the Palisades.
We did a stepped inventory reduction test
which was reminiscent of the semiscale tests done back
in the, I guess it was, late '70s, early '80s. We
were able to duplicate that work. Again we're trying
to characterize single-phase and two-phase natural
circulation in our loop. And I'll show you those
results later on.
And then, of course, we did the small
break LOCA tests. We had a -- it's actually -- we did
a 1.4-inch hot-leg break from full power conditions.
And actually number 8 was a two-inch hot-leg break
from full power conditions. So that test -- that's an
old test nomenclature. So that was actually a
two-inch break that was performed, number 8.
We did the stuck-open pressurizer 4 from
full power; 10 was revised also. We did stuck-open
pressurizer power with a combination stuck-open
automatic atmospheric dump valve. So some of these
have changed, and I guess we didn't catch it on this
slide. Sorry.
We did do two main steam line breaks, one
from full power and one from hot and zero power. And
then test number 13 we still need to perform. That's
one where we have an opportunity to do something a
little bit different there.
We're looking at HPSI injection in a
partly-voided downcomer. Now you've got level in a
downcomer. And we're essentially pouring cold water
into it. So it's a little bit different than what's
been done in the past. And we do need to get a little
bit of guidance on the best way to perform that test.
But that's all that's left from our original integral
system test matrix.
And we also performed some
separate-effects test in APEX, APEX-CE. And these
were steady-state HPSI tests. And these were similar
to what was done at Creare in their half-scale
facility. And we also simulated those in our
transparent loop. We wanted to see the thermal
stratification in the cold legs.
We wanted to study the plume development
interaction in the downcomer. And we wanted to look
at plume heat transfer downcomer walls. And so this
was an APEX-CE, so we were at full pressure. We were
injecting -- essentially it is a constant HPSI flow
rate.
And these were done with no core power, so
it was very similar to what was done at Creare. And
we'll show some of the differences that we've seen
from our data and what was done at Creare in the past.
Our test matrix, Tests 3, 4, 5, and 6, 3
was essentially a parametric study. And that was
safety injection under natural circulation fluid
mixing conditions. So we've looked at 16 different
conditions there.
And 4, 5, and 6 were just like the Creare
half-scale type test. We did it with one HPSI
injection and then with four at two different flow
rates. And the big difference there is we do see some
of the effect of the core barrel stored energy
release, and it plays a significant role.
And then in the flow bridge relation
series, we're still continuing to do some of those.
We're looking at onset of rear wall spillover and how
to go about modeling that.
In terms of analyses, we're doing two
types, of course. We're doing integral system
analyses and we're doing separate-effects analyses.
We're using RELAP5, the NRC, the Gamma version to do
the integral system analyses.
We've performed actually five analyses
there. One was a steady-state natural circulation.
The other one was Test Number 2, which was a reduced
inventory.
Then we've done two main steam line break
analyses and one small break LOCA analyses. So Dr.
Lafi will be presenting that later this afternoon.
In the area of separate-effects analysis,
we used STAR-CD to do some analyses there. And this
was kind of a very good experience for our students.
What we did, we said, here we have some
folks with some good basic engineering background. We
hand them the code and say, "You have a year to learn
this code," and to try to get -- to benchmark it
against existing data to see if it works. "I want you
to report back to me all your difficulties, all your
experiences, and how difficult it was to learn this
type of code to get to a point where you feel
proficient and able to use it to come up with
predictions for a new geometry."
So they've got some feedback and some
lessons learned there.
And we also used the REMIX Code for
predicting some of our separate-effects test. That's
one that had been used in the past in the previous
effort. It's a regional control volume type of
analysis technique that's used and had been used in
the past to predict the downcomer temperatures and
actually the -- all the way through the downcomer and
out the exit of the downcomer. And we've got some
calculations to show you there.
This mentions the tests that had been
performed or analyzed using RELAP. Again, we have
four there. And there was an additional one which was
basically our benchmark -- a benchmark case. And that
has the right nomenclature there with the two-inch
hot-leg group.
So you'll be seeing -- you'll be seeing
some of those calculations later today.
For the STAR-CD calculations, we did
benchmark the code against the Creare half-scale test,
MAY 105. You'll see that result.
And then we have also looked at
OSU-CE-0003 as one of the parametric studies that
included flow in the cold leg, a natural circulation
flow rate.
For the REMIX calculations we'll be
showing three of those today. OSU-CE Number 4, 5, and
6. And those are similar to what had been done in
Creare, except now we are doing an integral loop with
some of the heat transfer from the core barrel going
to the plenum. And we see that it does make a -- it
does impact the results quite a bit.
Okay. So this is an outline of what
you'll be seeing. Okay. I wanted to show you the
overall structure. We had integral system testing.
We had separate-effects testing. And then we had
modeling approaches or analysis approaches for each of
those areas. RELAP for the integral system, STAR-CD
and REMIX for the separate-effects test.
So we performed those analyses, and you'll
see those today. We were able to modify the facility,
and we think it scales quite well. We've got eight
integral systems tests that have been completed. We
have one test, number 13, lucky number 13, which
remains to be performed.
We have four separate-effects tests which
have been performed, and it includes one parametric
study which has actually 16 separate conditions. And
then we're right now continuing to do a series of flow
visualization tests in our transparent loop.
And that's that. Any questions on what
you're about to see?
So that's an overview of our research plan
and what's been done. I think the next thing on the
Agenda is to start describing working through this.
So the facility, I think, description is next.
CHAIRMAN WALLIS: Now we're due to have a
break after you stop talking; is that still
appropriate?
MR. REYES: I'm sorry?
CHAIRMAN WALLIS: We're due to have a
break after you stop your series of presentations. Is
that still appropriate?
MR. REYES: Sure, that will be fine.
We'll take a short break. So I have stopped talking.
What I've done is I've combined those
three presentations into one. So I've talked about
test matrix, what was performed.
CHAIRMAN WALLIS: Yes. Right.
MR. REYES: I talked about our research
plan. And what I skipped a bit of was the prior work
and the PTS research that's been done. But I did
summarize the results of that or the areas that we
felt we could contribute.
CHAIRMAN WALLIS: Are you going to give --
are you going to give John Groome's presentation?
MR. REYES: So now John Groome is ready to
present.
CHAIRMAN WALLIS: Okay.
MR. GROOME: Would you like me to present?
CHAIRMAN WALLIS: Sure.
MR. GROOME: Okay.
CHAIRMAN WALLIS: And then what do you
have to do after that, before --
MR. REYES: Well, then it's a separate
presentation -- oh, I see. The break's after my
second, okay. Yeah, the second presentation.
CHAIRMAN WALLIS: Your second
presentation. We'll see how long John Groome talks.
We may --
MR. GROOME: I'll try to go real fast.
OSU PTS TEST FACILITIES AND PALISADES OPERATIONS
MR. GROOME: Good afternoon. My name's
John Groome. I'm the Director of Facility Operations
on the APEX Test Facility.
I've outlined what we're going to talk
about. And I'm just going to go fairly fast. And if
you have any questions you can certainly slow me down
to ask.
I'm going to talk about the APEX-CE Test
Facility, some of the modifications that we performed
to the facility, a basic description of the facility.
I'll also show some slides of the flow visualization
loop.
I'll talk briefly about the NRC meeting at
Palisades. We traveled with NRC to Palisades, and we
actually observed the operators perform main steam
line breaks and small break LOCAs at their simulator.
And we learned -- what we tried to do --
the AP600 philosophy was hands off, the plant logic
takes care of the plant during an accident scenario.
Palisades is more like a traditional plant
where the operators actually interface with the plant
during the accident. And so we tried to incorporate
some of those operator actions into our tests.
And I'll talk briefly about some
conclusions.
CHAIRMAN WALLIS: Are we only doing this
work for CE plant?
MR. GROOME: Well, the facility that we've
-- the APEX facility was modeled after the Palisades
CE plant.
CHAIRMAN WALLIS: All right. But the
conclusions are going to be applied to pressurized
thermal shock in some other plants, or what's the NRC
going to do?
MR. BESSETTE: Yes. Well, the reason we
-- you know, the scaling was done to compare APEX to
CE plants because the APEX configuration falls very
close to the CE configuration. The other plants we're
doing analysis on were Oconee and the Westinghouse 3
Loop Plant, Beaver Valley.
So -- but in terms of, you know, looking
at -- some of these phenomena, of course, should be
generic, like the interruption of loop flow, you know,
flow stagnation.
CHAIRMAN WALLIS: Well, the method of
injecting ECC is different.
MR. BESSETTE: The method -- the method of
-- you know, the injection is similar.
CHAIRMAN WALLIS: It's different in
different plants.
MR. BESSETTE: Yeah.
CHAIRMAN WALLIS: A side injection is
rather unusual.
MR. BESSETTE: Yeah. You know, the plants
will have any -- all configurations, from top to side
to bottom.
CHAIRMAN WALLIS: I thought some CE plants
have top injection, don't they?
MR. BESSETTE: Possibly.
CHAIRMAN WALLIS: I think so, yeah.
MR. BESSETTE: But basically you can find
every configuration.
Now the other aspect is the general
question of code assessment is part of the objective.
So we look at different kinds of secondary side and
primary side transients with steam line break and
small break LOCA.
So, you know, we picked, let's say, the
most PTS-significant types of transients to run as
integral to the experiments.
Again I think, say, for a steam line
break, phenomena are similar. Small hot-leg breaks,
again the phenomena are going to be similar.
CHAIRMAN WALLIS: But you're going to
rewrite some PDS rules, aren't you, for all plants?
Isn't that in the offing?
MR. BESSETTE: Yes. Yeah, because the PTS
rule applies to all the plants generically.
CHAIRMAN WALLIS: So you're going to have
to show some sort of generic applicability of this
work.
MR. BESSETTE: Yes, -- well, you know, the
--
MR. KRESS: These methodologies, though,
would show up in a Reg. Yeah, they would show up in
a rule, wouldn't they?
MR. BESSETTE: So I think, you know, the
-- what we do is we say, well, the phenomena are going
to be -- the dominant phenomena are going to be the
same for different plant designs. We're doing some --
we're also incorporating -- let's say part of the
objective here is to assess RELAP. We've got the same
down on phenomena, so presumably the assessment is
going to be valid for those different plants.
Like I say, when we did the scaling eval-
-- when, you know, Jose did his scaling evaluation, he
did it looking at the geometric similartude of APEX
with respect to CE plants.
Was that an answer?
CHAIRMAN WALLIS: But RELAP doesn't model
this stratification, does it?
MR. BESSETTE: Some of these phenomena,
you know, of course RELAP can't do.
CHAIRMAN WALLIS: That's right. So how
can you test RELAP on phenomena you can't model?
MR. BESSETTE: Well, if we can show that
the downcomer temperatures -- if it turns out that
downcomer temperatures calculated by RELAP are similar
to the experiments, then we can argue that -- if
that's so, then we can argue that the phenomena it
can't calculate it don't seem to be that significant.
If there are differences, I think we have
to supplement the RELAP analysis with the CFD and/or
REMIX.
CHAIRMAN WALLIS: Okay.
MR. GROOME: The APEX-CE Test Facility is
geometrically similar to the Palisades Plant. It
includes a reactor vessel with a 48-rod
electrically-heated bundle. We have to two hot legs,
four cold legs with reactor cold pumps, and we added
high-pressure safety injection nozzles on a side
orientation, similar to Palisades.
One pressurizer. Two inverted U-Tube
steam generators, a feed-water pump. And we have a
programmable logic controller that we actually use to
model the Palisades Plant logic.
This is a brief summary of the
instrumentation. We have thermocouples. We added
approximately 50 thermocouples to the downcomer to
measure the plume temperature distribution for the PTS
work.
We have pressure and differential pressure
detectors. Some of our differential pressure
detectors are actually used to measure level and pipes
in various tanks.
Primarily for our liquid flow we use
magnetic flow meters. We use vortex flow meters for
measuring of steam flow. Some of our tanks have load
cells to actually get a mass.
And we installed Coriolis flow meters,
mass flow meters, on our four injection lines that we
added to the plant.
Testing capabilities in the CE
configuration include hot-leg breaks, cold-leg breaks,
main steam line breaks, stuck-open pressurizer safety
relief valves, and stuck-open steam line atmospheric
dump valves.
OSU modified the APEX to simulate the
Palisades' 2-by-4 PWR. Again we added four cold-leg
high-pressure injection lines. We actually modified
our cold legs to include a loop seal.
We looked at the Palisades Plant, and I'll
show you a slide here in a minute, and we added a weir
wall in our cold leg to simulate the Palisades'
primary-cooled pump housing lip.
And what that does is it prevents the cold
leg from completely draining. And we added again
approximately 50 additional thermocouples and 12
loop-seal thermocouples to our loop and four mass flow
meters to the injection nozzles.
This is an elevation view of the
Palisades' loop. And you can see the -- I'll just
point. This is the primary cooler pump lip here we're
talking about. The cold-leg nozzle would actually
inject -- would actually come off the screen here
perpendicular that prevents the cold leg from
completely draining.
CHAIRMAN WALLIS: So there's no loop seal
in this --
MR. GROOME: Well, I'll show you a side
view. So this is an artist's rendition of the APEX-CE
configuration.
One of the things that's different, this
is a planned view here showing the similarities
between the two plants. Here's an elevation view.
One of the things that's different is our
reactor coolant pumps were made specifically for the
APEX, the AP600 configuration, and they mounted
vertically to the bottom of the steam generators.
Typically on a PWR they're mounted
vertically upright on the top of the loop seal. What
we did is we dropped the pump vertically at the bottom
of the loop seal. And then there at the flange right
on the discharge on the vertical section or the
horizontal section of the cold leg, we added a weir
wall in the cold-leg pipe to simulate that lip of the
Palisades' primary coolant pump.
We also built a flow visualization loop to
help us understand the mixing in the cold leg. Again
it was constructed of clear PVC pipe. The test loop
includes a single cold-leg piping geometry
representing APEX-CE, a high-pressure safety injection
nozzle on the side with a check valve, a weir wall in
the cold leg, and we have a 50-gallon salt water
mixing tank that we use to simulate the difference
between the hot and cold streams.
Our high-pressure injection pump is
capable up to 20 gallons per minute and our cold leg
flow pump is capable of 500 gallons per minute.
Typically we ran the flow visualization tests at much
reduced flows.
CHAIRMAN WALLIS: This salt water mixing
is only for a separate-effects test?
MR. GROOME: Correct. It was just for
visualization.
And here is a side view of the test loop.
The green dye is actually in the salt water when we
inject it. This is actually a post-test. You can see
the weir wall right there. Right there's the weir
wall in our clear test section.
If we had a pump here that simulated the
APEX-CE, the pump would actually be installed down
here. But we just took a -- we have a test pump there
in the facility, and we just took a flow off of the
pump that's mounted on the floor there to simulate
natural circulation flow rates. And we could vary
that parametrically up and down.
CHAIRMAN WALLIS: Are you going to show us
this, or are you --
MR. GROOME: Well, you kind of saw it if
you --
CHAIRMAN WALLIS: Yeah, it was there.
MR. GROOME: -- when you were there, but
we didn't actually put on a demonstration. But we can
do that for you.
MR. SCHROCK: Now what does the weir wall
simulate?
MR. GROOME: The weir wall simulates the
loop, the lip seal.
Can you make me go back there, Brandon, to
where that picture was? One more maybe. Two more.
There you go. There's the mouse.
So this is the lip here that I'm talking
about that exists in the primary coolant pump casing.
And right here projecting perpendicular from the
screen is the cold leg nozzle.
And so what it does is it prevents you
from draining, completely draining this cold leg if
you were to open up a drain valve. In other words, in
order to spill over into the loop seal you have to get
by this lip.
And since our pumps are at the bottom of
the loop seal we added this lip here on the horizontal
section of our cold-leg pipe to simulate this lip of
the Palisades' primary coolant pump housing.
MR. SCHROCK: In what way does it simulate
it? It's got the same height restriction or --
MR. GROOME: The same -- it's the same
elevation. It doesn't simulate necessarily the
roundness of the geometry. It's just a vertical plate
that's welded on the inside of the pipe.
And again this is a top-down elevation of
the flow visualization loop showing the side injection
nozzle there.
We also traveled to Palisades to -- we
found out where Kalamazoo is. If everybody ever wants
to know where Kalamazoo is, it's in Michigan. And --
CHAIRMAN WALLIS: It's a Covert operation.
MR. GROOME: Yeah, it's in Covert. The
plant's actually in Covert, but you fly into Kalamazoo
to get to Covert, which we had to look at a map for
quite a while to figure out where that was at.
But we actually talked with the Palisades'
operators trying to understand the logic that they do
for their accident scenarios and when they train on
the simulators.
We actually observed, I think, about four
tests, two main steam line breaks and two small break
LOCAs. And based on discussions with the Palisades'
operators and our observation and understanding of the
scaling limitations of the facility, we developed a
set of test procedures for the APEX-CE tests that we
performed as part of our NRC work.
And I'm just going to briefly just kind of
-- Dr. Reyes wanted me to talk a little bit about just
to give you kind of a feel for what Palisades'
operators would do during tests.
We looked at some of their emergency
operating procedures, their standard post-trip
actions: Loss of coolant, accident recovery
procedure, an excess steam demand, their functional
recovery procedures, and some of their EOP supplements
when we were developing our test procedures.
And this is just a general outline for
Palisades for a small break LOCA, their general
sequence of events that you would see.
They normally would manually scram the
reactor whenever their second charging pump was
started. They have three charging pumps that can vary
the flow from anywhere from 30 to 150 gallons per
minute.
They manually tripped two reactor coolant
pumps at 1300 PSI, one on each loop. And they tripped
a second pair of reactor coolant pumps whenever they
approach less than 25 degrees subcooling margin. And
typically that will happen about five minutes.
Depending on the scenario that they're
running, typically will happen about five minutes
after tripping the first reactor coolant pumps.
And the operator determines this by
looking at T hot and temperature to core and
pressurize the pressure. And they actually have a
screen, and I don't have a viewgraph, but they
actually part -- I do have a little bit of a viewgraph
showing the operator trends as they depressurize.
And typically the reactor coolant pumps
are not restarted during a small break LOCA, even if
subcooling is regained. And that's because -- well,
this assumes -- for a steam break inside a
containment, they lose cooling water to the reactor
coolant pumps.
And so whenever they're isolated they have
about five minutes to regain cooling to the reactor
coolant pumps, otherwise they have to go to a warm-up
procedure that takes them about 30 to 40 minutes to
warm up before they can open up the cooling water back
to the reactor coolant pump. So they just go ahead
and lose the second pair, and they don't immediately
worry about restarting the reactor coolant pumps.
Monitor. They monitor the pressurizer and
reactor pull, reactor pressure level. The small break
LOCA does not adequately remove decay heat, so they
use the steam generators to remove decay heat via the
turbine bypass valve or atmospheric dump valves. The
turbine bypass valves are limited to five percent.
And then they manually control aux feed water to
maintain the steam generator level.
There's no automatic steam generator
isolation. And when the T hot approaches about 550
degrees Fahrenheit, they secure the main feed pumps.
And note this is approximately after the test T hot or
after the small break LOCA, the temperatures are
approximately 530 degrees Fahrenheit immediately
following the reactor trip.
The turbine bypass valve is -- the set
point is 900 psia, which is approximately Psat for
532. So the turbine bypass valve will open up and
start to dump steam also.
The main steam line break. They get their
signal either on a containment pressure greater than
four PSI inside a containment. They will isolate the
main feed pumps and main steam. Auxiliary feed water
and atmospheric steam pump valves are available.
They manually trip the reactor or allow it
to trip on a set point, but they will get a
power-to-flow set point scram.
They'll determine the affected steam
generator and isolate on excess steam demand or tube
rupture. So they have a procedure where they try to
determine if it's a U-tube rupture or a main steam
line break. They'll isolate the aux feed water.
And for our tests we were not
conservative, and we assumed that it took them about
10 minutes to isolate aux feed water to the affected
steam generator.
And they'll also isolate the atmospheric
dump valve, turbine bypass valve, and the safety
isolation valves. And they'll allow the affected
steam generator to blowdown.
They'll stabilize the plant. They'll cut
back on the charging pumps when the pressurizer level
is gained to 40 percent if it's a main steam line
break in the containment; and 20-percent pressurizer
level if the break is outside the containment.
They'll maintain the normal steam
generator -- the normal steam generator level on the
intact steam generator by using aux feed water.
When the steam generator pressure is less
than 800 PSI, they'll close the main steam isolation
valves.
The reactor coolant pumps are on unless
the excess steam demand is inside the containment.
And again what happens there is that the containment
is isolated and they lose coolant water to the reactor
coolant pumps. And the reactor would most likely trip
on their power-to-flow set-point.
Palisades has a set of pressure and
temperature limit curves that are available to the
operator on a CRT screen. These curves are basically
designed to help them avoid a PTS event.
This is a look at the pressure and
temperature curves from one of their emergency
operating supplements. There's three curves on the
bottom there. You could perhaps read a little bit
better from your notes.
The bottom curve is the saturation curve.
The curve right above it here is the 25-degree
subcooling curve. So if they got down to this curve
here they would secure the sector of reactor coolant
pumps during a small break LOCA.
And this third curve that's right above it
that actually crosses over the 25-degree subcooling
curve is their minimum-pressure temperature for
reactor coolant pump operation. And this is just an
abbreviated screen. This actually goes -- there's
another page that actually goes all the way out to 50
degrees Fahrenheit.
The upper limits of these curves, there's
a 200-degree subcooling curve. And there's this VLTOP
curve which is their variable limit for temporary
overpressure protection. I like to refer to this as
the brittle fraction prevention curve here, so that
when they're shut down this is the set-point that
would actually open up their primary power-operating
relief valves to avoid exceeding the pressure for the
temperature that they're at.
Normally this curve is not -- the VLTOP
curve, when they're operating, is not in play. They
use a 200-degree subcoolant curve until it comes down.
And the operator -- I'll show you a screen
here. This is from the actual main steam line break
simulation. And this is just -- there's a wide
variety of information on this CRT when you're
actually looking at it. And we just pulled one image
off. But this is their pressure and temperature limit
curves.
Number 2 here and 3 are the hot and
cold-leg temperatures. So what they're doing, as the
main steam line break is progressing, the operator is
trying to control pressure and temperature to maintain
the plant within the two limits, the subcooling curve
here and saturation curve, and their 200-degree
subcoolant curve here on the top or the VLTOP curve.
The APEX plant operating procedures and
the plant actions were generally realistic with the
following very important exceptions. We did not allow
-- for our test procedure we did not allow throttling
of the high-pressure safety injection. So we just
started our high-pressure injection safety, and we
modeled the full flow during the whole entire test.
Isolation of the feed water flow to the
broken steam generator was assumed to take 10 minutes.
And no effort was made to keep the plant
within the pressure and temperature band, scaled, as
required by Palisades' emergency operating procedures.
So we essentially just let the plant behave as it
would with no operator actions once we started the
scenario.
The APEX-CE Test Facility includes all the
key components needed to simulate the Palisades'
thermal hydraulic overcooling behavior.
The transparent loop provides
visualization of the fluid-mixing behavior in the
APEX-CE cold legs.
The NRC meeting at Palisades in March and
the emergency operating procedures provided valuable
insight into operator actions during the main steam
line breaks and small break LOCAs.
The Palisades' emergency operating
procedures are designed to avoid a PTS event. The
APEX-CE test procedures were based on Palisades'
emergency operating procedures, discussion with plant
operators, and actual observation of plant
simulations.
The APEX-CE operator actions were
generally realistic with a few very important
exceptions designed to produce PTS test conditions
needed to benchmark codes.
And I believe that's my presentation. Are
there any questions?
CHAIRMAN WALLIS: Not at this time
probably.
MR. GROOME: Thank you.
CHAIRMAN WALLIS: Thank you very much.
So you train your operators to simulate
Palisades or something, or do you put it in the
computer ahead of time, or...?
MR. GROOME: Actually we did both. For
the most part, for repeatability we programmed it in
our programmable logic controller.
But actually during the performance of
this test we had an equipment failure where we lost
our main feed pump. And so we had to use our pump
that we use for high-pressure safety injection for our
main feed pump. We have that capability. And so we
actually performed operator actions because of that
equipment failure at this plant.
MR. KRESS: The operators, their actions
are intended to keep you away from pressurized thermal
shock --
MR. GROOME: Correct.
MR. KRESS: -- or at least minimize it.
But your test, you're trying to see what happens here
in a pressurized thermal shock.
MR. GROOME: Exactly right.
MR. KRESS: So you don't really want to do
everything they do.
MR. GROOME: Exactly. That's what we did.
MR. KRESS: Yeah.
MR. GROOME: We looked at the actions that
they performed during the test, because it was quite
a bit different than the AP600. You know AP600,
basically the operator's dead. Plant logic takes care
of itself. There's no operator actions at all.
Palisades, there's operator actions that
affect the outcome of the test. So what we did was we
-- to understand that, we traveled to Palisades, read
and reviewed their emergency operating procedures,
talked to their operators, and incorporated some of
those operator actions into the test procedures.
For example, you know, we assumed that it
took a minimum of 10 minutes to isolate aux feed want
are.
MR. KRESS: Um-hum.
MR. GROOME: But we didn't try to --
unlike Palisades' operators, they'll actually try to
control temperature and pressure to keep the plant
within the two bounds of the P and T curves. We did
not do that. We let the test just progress after that
point.
MR. WOODS: Could I -- I'm Roy Woods. I'm
in Research, NRC. I just wanted to point out that
what these gentlemen are talking about, we take the
PRA contractors and the HRA contractors, and we've
gone to each of the plants that we are analyzing.
And one of the main things we do when we
go to these plants is go through some simulator
exercises so our HRA people and our PRA people can see
what the important human actions are and see how the
training the plant operators get corresponds to the --
whether or not they'll actually do those actions, and
whatever.
So you gentlemen went on the trip because
you were analyzing Palisades. And we also have
already gone to two other plants for the same kind of
purpose.
So the fact that they saw these human
actions and then chose not to simulate them, you know
that's okay. I mean they were looking for one purpose
and we were looking for a different purpose.
MR. KRESS: Yeah, that's what I get.
MR. REYES: Any other questions?
No.
SUMMARY OF INTEGRAL SYSTEM OVERCOOLING TEST RESULTS
MR. REYES: Let's proceed. What I'd like
to do is just start off with our integral system tests
and just give you the -- we'll start off with kind of
the big picture.
For PTS we're interested in the system
pressure and the downcomer temperatures, so we'll talk
a little bit about that.
In the presentations that will follow
we'll talk more about the details of how we got to
those temperatures and what's happening specifically
in the downcomer. So let's start with just an
overview.
So we'll look at downcomer fluid
temperatures, at some of the temperature and pressure
extremes for all the integral system tests that were
performed. And I'll just give you some conclusions
based on the big picture.
So you'll be seeing several presentations
looking at the specifics of the downcomer fluid
temperature under different conditions. So that will
-- that's to follow.
But in general the fluid temperatures were
relatively uniform around the downcomer at about eight
cold-leg diameters down into the downcomer. So we saw
good mixing for all the integral system tests at 8D.
And for the most of the tests at 4D, four
cold-leg diameters down, we saw good mixing. So
that's just perhaps a rule of thumb that you can keep
in the back of your mind.
MR. SHACK: Where does that put me
relative to the beltline?
MR. REYES: The beltline for this
particular plant, their beltline wells were right
about five cold-leg diameters down. Their centerline,
beltline -- well,...
Several of the tests that did experience
axial thermal stratification in the downcomer.
So we saw a cold temperature on the bottom
stratified all the way up to the top. But radially
they were all relatively uniform.
Here's the six integral system tests that
we performed. And this is just kind of trying to pick
the extremes. The minimum downcomer temperature for
that particular test and then in the column next to it
what the minimum pressure was.
And again in terms of scaling, you might
want to think of it this way: 360 psig would
correspond to about 1200 psig in the Palisades Plant
on a pressure-scaling basis. Okay.
So what we saw was an minimum downcomer
temperature for the smallest break, the 1.4-inch small
break LOCA, which was off of the hot leg, the minimum
downcomer temperature is about 255 degrees Fahrenheit.
We initially were at 420 degrees. It came down.
And the minimum pressure -- temperature --
the pressure at that minimum temperature was about
131. You can look at all the numbers there.
In terms of the case which produced the
most -- the highest pressure at the lowest
temperature, we're looking at the main steam line
break at hot zero power.
And I think that's consistent with what we
saw in the previous analyses that were done almost 20
years ago -- about 10 years ago.
So the minimum downcomer temperature is
about 230 degrees Fahrenheit and about -- we
repressurized essentially on that test, okay.
And what I'll do now is I'll show you the
plots of each of those scenarios, the pressures and
temperatures, so we can do a little bit of a
comparison.
CHAIRMAN WALLIS: What's the last column
there?
MR. REYES: Oh, thanks --
CHAIRMAN WALLIS: You actually observed
stagnation for part of the time, or something?
MR. REYES: That's right, yeah. So if the
cold leg experienced stagnation anytime during the
transient, we identify which cold leg's that.
CHAIRMAN WALLIS: But they're not all at
the same time then?
MR. REYES: Right. And we did see some
asymmetric loop stagnation, and that was very
interesting phenomenon. We can explain that. There
will be a whole separate presentation just on loop
stagnation mechanisms. We've isolated those,
identified what causes loop stagnation for this
design.
MR. SHACK: What are we really talking
about when we say the "main steam line break" here?
MR. REYES: Right. In this test we were
doing, it was the equivalent of a one-square-foot main
steam line break.
MR. SHACK: Oh, one square foot.
MR. REYES: And the assumptions that were
involved in that, we had -- we assumed a
one-square-foot break on the main steam line. It was
assumed to be inside containment. And so we used the
operator or the -- the operator actions that would
correspond with a break inside containment. So that
requires isolating containment.
And when you isolate the containment what
happens is you lose your component cooling water to
the reactor coolant pumps. You lose your ceiling, so
you basically trip your pumps. And so we followed
that logic there.
So we performed two of these cases, one at
hot zero power which essentially assumed that the
plant had been scrammed about -- for a period of about
100 hours. And then full power, which assumes that an
immediate scram going to a decay curve.
For the hot zero power case we essentially
picked the -- we picked the power at 100 -- scaled
power at 100 hours. And essentially that power varies
very, very slowly at that point. And so for our test
we just used a constant low power.
For the full-power case we went to a decay
curve.
So we can see the two different scenarios
there. For the full-power case, we're looking at
pressure. That's in the solid black line there. The
red-dash line is the case for the hot zero power.
We see that because we're initially at a
lower power. We get -- so we wind up at a lower
pressure. What brings the pressure back up in this
scenario -- okay, the main steam break force drives
the pressure down in the primary side -- what brings
it back up is just the action of the high-pressure
injection system.
Unlike Palisades in their scenarios where
they would throttle and just keep bringing pressure
down, we just let the system run and, given enough
time, the plant would repressurize. So this would be
-- this is a good benchmark case for RELAP Code, to
try to see if it could match these curves. So again
it's more severe than what you would see in the real
plant.
Now this is for the same test, same pair
of tests, what we saw for the downcomer temperatures.
And this is at the eight-diameter location, 8 cold-leg
diameter down into the downcomer.
And we see that the full power case,
because we still have core decay heat being generated
at a fairly substantial rate -- actually not only did
it repressurize, but it also reheated.
For the hot zero power case we stayed
fairly low in temperature.
We also did some of the small break LOCA
tests. And so here's the three scenarios that we did.
We did a 1.4-inch hot-leg break, a simulation of a
1.4-inch hot-leg break.
Then we did a large break, a two-point --
a two-inch small break LOCA.
And then we also did a stuck-open safety
relief valve on the pressurizer, which was the
smallest of the breaks. And so we can see -- an
interesting phenomena for the 1.4 inch. That's the
solid line that has kind of this jagged behavior. In
that case the HPSI pumps are capable, can actually
keep up with the break flow.
And so we see kind of a pressurizing and
then we fill up the pressurizer volume. And then we
would sweep all the liquid out again and repressurize.
And so we saw kind of a isolated behavior for that
one.
So if your pumps are able to keep up with
the break-flow rate, you see kind of a
repressurization in this kind of a jagged behavior.
When we went to a larger break, a two-inch
break, we see the pressure basically just come down
and it just keeps coming down.
MR. SCHROCK: How do you simulate the SRV
of the actual plant? What valve do you have and how
do you go about qualifying it as a simulation of the
actual plant valve?
MR. REYES: Right. Yeah, that's a good
point. All we do is we have a flow nozzle which is
sized to the diameter, a scaled diameter. And that's
-- it's well characterized, so we know the loss
coefficient for that flow nozzle. So all we can do is
characterize that. It's not -- it doesn't represent
the geometry of the valve throat in the real plant.
So that's a good point.
MR. SCHROCK: So you assume that critical
flow behaves in the same manner as loss coefficients
for incompressible flow?
MR. REYES: Come again? You said the
critical flow?
MR. SCHROCK: Yeah. It's critical flow
through the SRV.
MR. REYES: Right, that's correct.
MR. SCHROCK: And that's unrelated to a
loss coefficient ordinarily. So --
MR. REYES: Right, right.
MR. SCHROCK: -- when you say you have the
same loss coefficient, what's the significance of
that?
MR. REYES: In terms of loss -- what we do
is we characterize the flow nozzle for a range of
conditions. So we look at critical-flow conditions,
but we also look at essentially a subcritical flow.
For critical flow, of course we do have a
-- what this allows us to do at least with a code like
RELAP is specify what the conditions are at the valve.
So we're not giving it a complicated structure at the
valve to analyze, in essence.
So we don't -- to make the long answer
short, we don't simulate the geometry of the actual
SRV for Palisades. But we have -- but we know what
our geometry is, and we can characterize it.
MR. SCHROCK: But the scaled leak flow is
somehow demonstrated to be related to the, or the same
as in the plant for their actual valve?
MR. REYES: The data that we have for
their valve, they give a -- for a given pressure
condition in their plant, they give a given steam-flow
rate, and so we scaled to that.
So -- but they only give us that top
limit, so we know that their maximum -- we're given
pressure in their plant and what the flow should be
for that plant, but it's only one value.
And then it does behave quite well as a --
it's very close actually to a perfect gas behavior.
For the same three tests the small break
LOCA is looking at the downcomer temperatures. Again
this is somewhere down below the beltline weld.
CHAIRMAN WALLIS: Are you going to show us
the 4D ones, too?
MR. REYES: The -- oh.
CHAIRMAN WALLIS: Well, you said 8D. I
mean this --
MR. REYES: Right. Right, yeah.
CHAIRMAN WALLIS: You measured the various
Ds presumably.
MR. REYES: Well, we'll have several
presentations looking just at the downcomer fluid
temperatures. So I picked one in particular just to
-- as a characteristic like --
CHAIRMAN WALLIS: So we're going to
revisit that?
MR. REYES: Oh, absolutely. Absolutely.
We'll spend quite a bit of time in the downcomer.
The downcomer temperatures for the small
break LOCA, again looking at which would be -- gave us
the lowest temperatures. A two-inch break of course
because you're depressurizing gave us the lowest
condition there, followed by the 1.4 inch, and then
the smaller -- the small -- so we're just following a
saturation curve there.
We did do one combination test, and this
was kind of interesting. It was a stuck-open
pressurizer and the safety relief valve followed
immediately by a stuck-open or single stuck-open
atmospheric dump valve on the main steam line. Okay.
So we basically have a break on both sides
of the plant. And this is what we saw. Again initial
immediate depressurization, and then it gradually
tapers off, and then it just flattens off at a fairly
low pressure.
In terms of temperature we see somewhat of
an exponential-type decay which would go with a
primary side depressurization. And then a relatively
linear plot after that or trend after that.
The SRV is fairly small compared to the
other breaks. They have a fairly small valve size
that they use for that.
MR. SCHROCK: These are test data and yet
you get rather sharp changes in the slope of the
pressure curve. Not in this graph but in a previous
one.
MR. REYES: In the previous?
MR. SCHROCK: Yeah.
MR. REYES: Right.
CHAIRMAN WALLIS: Presumably something
happened to that point.
MR. REYES: Right.
CHAIRMAN WALLIS: TMI is low, though.
Something happened. The valve opened or closed, or
someone did something. A very sharp change in
pressure.
MR. REYES: Right. Yeah. This test I
don't recall if we fed the steam generators at some
point in the test.
MR. SCHROCK: So you've got about three,
four, five, six of those.
MR. REYES: So we have the -- we have a
sequence of events for each one of these tests. We
have a lot of details we can provide you. But I want
to give you kind of the big picture, the big picture
being that --
CHAIRMAN WALLIS: And you're going to show
us that RELAP predicted exactly?
MR. REYES: Yeah. We won't go there yet.
And this is one that we did not use to --
we have not benchmarked our RELAP against this one.
This is a prechallenging test.
Okay. So what we saw overall was that the
fluid temperature's relatively uniform around the
entire downcomer at about the 8D location, and for
most tests at the 4D location we saw that. So we're
seeing good mixing up in those regions.
So plumes appeared to be relatively well
mixed by the 4D axial location.
Test Number 11 resulted in the lowest
downcomer temperatures while at repressurized
conditions. And that was the main steam line break,
one-square-foot main steam line break at hot zero
power.
So that kind of gives you -- of the
scenarios we performed, that gives you a feel for
which one was, in essence, in terms of PTS probably
the most limiting.
That doesn't mean that there are other
pressures that could be of importance, though. We
looked at one that repressurized. Others kind of
tapered off to a lower pressure. But in terms of PTS
I'm not sure what, what the limit is --
CHAIRMAN WALLIS: So these were mixed four
diameters. And four cold-leg diameters below the cold
leg everything is mixed out?
MR. REYES: Right. So we look at the
temperatures all --
CHAIRMAN WALLIS: So pretty rapidly.
MR. REYES: -- the way around the
downcomer.
CHAIRMAN WALLIS: It's a pretty
rapidly-spreading plume, is it not?
MR. SCHROCK: These are diameters of the
vessel?
MR. REYES: Cold-leg diameters. Cold-leg.
MR. SCHROCK: Cold-leg diameters.
CHAIRMAN WALLIS: You're never going to
get eight diameters down.
MR. SCHROCK: My gosh.
MR. REYES: Yeah, yeah. So -- and when we
talk about the plume, the characteristics of the
plume, you will see that the injection flow rates are
very low in this design. And so they're fairly --
they tend to be weak plumes that break up relatively
quickly.
MR. SHACK: Oh, yeah. Is that -- I mean
is that related to the wimpiness of the high-pressure
injection system?
MR. REYES: Correct. Yeah. And then
we'll look at some other conditions where you're able
to preserve the plume a little bit further.
MR. SHACK: A very low flow rate out of
these safety injection systems, right, relative to
other systems?
MR. BESSETTE: Especially a CE plant.
MR. BAJOREK: Especially a CE.
MR. SCHROCK: So how does that work out?
The circumferential spacing of the cold legs is how
many Ds?
MR. REYES: The circumferential spacing of
each -- what are their angles on our plant, John? Are
we on 90s?
MR. GROOME: I believe we're 90, but I'm
not -- I'd have to look to answer the question
correctly.
MR. REYES: Yeah.
CHAIRMAN WALLIS: It would be good if you
showed -- maybe you will -- show an unwrapped annulus
and --
MR. REYES: Right. Right. And that --
CHAIRMAN WALLIS: -- an unwrapped
downcomer with all the pipes and --
MR. REYES: Absolutely. Right, that'll
show the configuration.
And we'll see that we've used -- well,
we've used STAR-CD and we've used REMIX to try to
benchmark those codes to see how well they predict our
behavior with the hope that maybe they could be
extended to other, other plant designs and conditions.
With that do we have any other questions?
MR. SCHROCK: Well, you've got uniformed
fluid temperature, but what about the metal
temperatures?
MR. REYES: Right. We do -- we have a
region of the vessel which is measuring metal
temperatures. And it's an outside wall temperature.
And it's mostly in the vicinity of the upper
downcomer. And we do see a distribution of
temperatures there.
Now our vessel wall, of course, is very
thin. We're a half-inch-thick wall compared to
Palisades, which has a half-inch of -- or quarter-inch
of stainless steel with about six-inch of
carbon-steel-based metal. So it's a significant
difference in our wall behavior.
So we focused primarily on the fluid,
fluid temperatures.
MR. WACHS: To answer your question about
the separation, the peak diameter is that -- it's I
guess --
MR. ROSENTHAL: You need to speak into the
mic here. Identify yourself, too, please.
MR. WACHS: I'm Dan Wachs. And to answer
the question about the separation of the cold legs, I
believe it's about eight diameters between each one.
You have about a 25-inch radius. I guess that's going
to be 16. if you just pi times the diameter, divide
by three-and-a-quarter inches.
CHAIRMAN WALLIS: They're closer together
than you think, I think.
MR. SCHROCK: Well, what he's saying is
they're far apart in comparison to what the data show
produces complete circumferential mixing. Surprising.
CHAIRMAN WALLIS: It seems very rapid
mixing to me.
MR. SCHROCK: Yeah, right. That was why
I asked. It seemed very rapid.
MR. REYES: Now we'll look at some cases
where, in fact, part of it is because the flow rates
are so low in this plant. Even on these low-flow
rates there are some cases where we'll see that loop
flow, natural circulation loop flow, acts to preserve
the plume. And so we'll talk about that a little bit
later. That's kind of an interesting effect.
CHAIRMAN WALLIS: So you're keeping all
that exciting stuff for tomorrow, are you?
MR. REYES: Yes. That way I'll be sure
you come back.
CHAIRMAN WALLIS: I have a question. Is
this a good time to take a break?
MR. REYES: Sure, I think this is perfect.
CHAIRMAN WALLIS: Then let's do it. We'll
take a 15-minute break, and we'll start again at 3:25.
(Recess taken from 3:08 p.m. to 3:25 p.m.)
MR. LAFI: Shall I start?
CHAIRMAN WALLIS: Yes, please.
NUMERICAL SIMULATION FOR APEX-CE MSLB AND SBLOCA
TESTS USING RELAP5/MOD 3.2.2 (GAMMA VERSION)
MR. LAFI: My name is Abd Lafi, and I'm an
Assistant Professor at the Nuclear Engineering
Department at OSU.
My presentation will be focused on
numerical simulation for APEX-CE main steam line break
and the small break LOCA tests using RELAP5, Model
3.2.2, Gamma version.
The outline of my research will include
objectives, input modifications, APEX-CE model
nodalization, RELAP5 calculation matrix, RELAP5 run
strategy, and then mention and discuss some
comparisons between the tests that I analyzed.
These tests will include a one-foot-square
main steam line break from hot zero power. This is
OSU-CE-11.
And then the second test is
one-foot-square main steam line break from full power.
This is OSU-CE-12. And then two-inch hot-leg break,
OSU-CE-08.
I will end up with some conclusion after
each of these tests.
The objectives of all of these analyses is
to assess the ability of RELAP5, Model 3.2.2, to
predict transient overcooling behavior. Actually, in
particular, I will focus on the onset of loop
stagnation during the integral tests under
considerations; and then I will discuss some finding
about the general behavior or general trend of
downcomer fluid temperatures and system pressures.
We use our version of the input -- and the
reason I say "our version," is because the original
input deck was developed up by our new -- this input
deck was adopted by Science Tech to analyze some of
the NRC tests that was conducted at OSU.
Now this was input related to the original
APEX with the new APEX-CE geometry that's required and
necessitated a lot of changing.
Also after I explain the modification
briefly to the whole input deck, I adapted also some
modification, according to the operating condition and
the geometrical configuration of each test.
So the first modification was to isolate
its own APEX, AP600 passive safety system and DVI
lines. There were ADS systems, there were CMTs,
accumulator, IWST, PRHR. All of these components were
isolated.
And I isolated these components in
addition to all the input related to these components.
Also elimination of all APEX AP600 safety
system actuation logic. As you recall, with the APEX
Test Facility, there was, for example, ADS, which were
actuated based on the CMT level and sometimes CMT
level plus time, as the case with ADS-4. So there
will be no LOCA CMT, no ADS, too. So all of these
logics were eliminated. This is just one example of
these logics that we no longer use it.
Also with the new configuration we added
the new seals, and a lot of piping associated with it,
and also some heat structures, some tables, some --
also, for example, the loop seal and the changing of
dropping the pumps down.
It was originally connected directly to
the steam generator on one side and the other side
directly to the cold leg. In this configuration we
dropped it by almost 18.75 inch.
So this needed some changes in the pumps.
Also the other side was connected to the loop seal.
In addition to this we added the
high-pressure safety injection head curve. And the
nodalization diagram that I used. This shows just the
primary. The secondary is not shown in this diagram.
As you see, there are loop seals and there are four
injection systems.
RELAP5 calculation matrix, as I mentioned
in the introduction, we analyzed almost five tests.
But I want to present, since the topic was just focus
on main steam line break and the small break LOCA, I
will discuss three tests which is two actually for the
main steam line break and the second one for this hot
-- of the top of hot leg number one, a break which is
called APEX-CE-08.
I have the result, and I can also offer a
conclusion of my finding in the other two tests.
I adapted the same strategy that was
adapted by Science Tech. Actually this strategy
includes running RELAP, a steady-state case. And the
purpose of this run is just to establish the initial
condition for each test.
In these kind of runs, I introduced some
control volumes and control runs and some
time-dependent volumes. The purpose of this is just
to bring the facility fast to the initial condition of
each test. These additions will be dropped later.
Then I run usually each test for 1200
seconds until I see just some stable initial condition
that fits or close to the real test. Then I stop.
This is 1200. From experience I found that sufficient
to reach stable initial condition.
Sometimes you reach it within 400, 500.
It's just because I introduce what I mentioned just
before, introduced some time-dependent volume that
accelerate the calculation.
I run another -- after actually the
steady-state test, I initialize and I zero the time
just out in the steady-state case to make the 1200
second is just zero. So the initial point or the time
equal to zero will be the initial condition at the
1200 second.
Another -- I didn't mention this, but I
run another one which is called null transient. I
just run the transient for short time, just to see
whether whatever I adapted in the control of the
steady state, the time different in that I added, does
not affect the transient case. So to see whether the
transient case will hold. And that's why we call it
null transient.
The real transient comes after, which is
called restart run, sometimes runs, not run, because
sometimes -- actually all the time I monitor the kill
the condition. If I see some abnormal condition, I
stop RELAP, look to the problem, fixing it, and then
I run restart run to continue.
The first test is called RELAP5
calculation for all OSU-CE-011. And some brief
description of this test. It was simulated,
one-foot-square main steam line break conducted at a
constant power.
Steam generator number one, the
power-operated relief valve was open to simulate the
main steam line break. And then upon the initiation
of the break, the reactor coolant pumps were tripped
and the power brought from 100 kilowatt to the 54.8
kilowatt.
The pressurized heaters were allowed to
cycle on and off based on the pressure of the
collapsed liquid -- pressurizer liquid level. In this
test we turned it off when -- the test turned it off
when the pressurizer liquid level reached 16. And
then it turned on when the pressurizer collapsed
liquid level reached 26-inch.
The auxiliary feed water was maintained
for the broken steam line for 10 minutes, and it was
isolated from the intact steam generator.
This is a brief description of the test.
And the sequence of events will be discussed along
with the RELAP sequence of events when I come to the
comparison.
The steady state. As I said, each test I
run steady-state analysis to initialize the input deck
to start with the correct initial condition that match
the real test.
The code was run for 1200 second,
time-dependent for volumes and some controls that were
added, as I said.
I run it for 1200 second and the results,
the calculated versus much of initial condition, as
you see it, is almost similar exactly. The power 100
kilowatt. Pressurizer pressure 370 for both. The
hot-leg temperature almost the same. The cold-leg
temperature. The steam generator 1 and 2, 272 psig,
which is the case with the calculation.
Pressurizer level, steam generator water
level, and the steam generator feed water temperature,
the steam generator feed water mass flow rate, all of
these almost even. The cold-leg natural circulation
flow is almost comparable. There is an -- I put 15.3.
As I recall the test, it was 15.29.
So this is the transient case comparison
between the pressurizer pressure. The red line is the
test. The dotted line is the RELAP calculation. It
shows the comparison. The general trend is
acceptable, except faster depressurization with the
RELAP calculation. I would connect this, the impact
of this on other calculated parameters.
The general trend of the broken steam
generator was almost an accurate agreement. This is
the pressurizer level. And the high pressure
injection system is almost for the first part.
(Comments off the record re pointer.)
MR. LAFI: This chart comprises all the
four high-pressure injection system flow rate RELAP
calculation. As you see before this point is almost
comparable except with this, because here the
high-pressure injection system comes into play when
the pressure reach 360 psig.
That's why I initiated almost at 160
seconds at the beginning, but when it reach the 360 it
initiated when it drop below 360. When it reached
360, the high-pressure injection system flow rate will
be zero.
So the reason for this discrepancy is due
to the pressure prediction with RELAP reach the 360
later, different from the real test. But the general
behavior is acceptable.
CHAIRMAN WALLIS: I would think the
integral under the curves has to be the same because
this is -- maybe not.
MR. LAFI: What? What is it?
CHAIRMAN WALLIS: No, it's not. No, it's
not. Okay, that's all right. Forget that question.
MR. SHACK: The pressure's not the same.
CHAIRMAN WALLIS: No.
MR. LAFI: Shall I continue?
CHAIRMAN WALLIS: So we conclude that the
comparison is pretty good? Is that what you conclude?
MR. LAFI: I think so, yeah. The
comparison between -- this is -- because the
high-pressure injection system is pressure system
dependent. Okay
There is impact in this area on the end of
the temperature, for example, but for the most part
it's acceptable.
The break flow rate, the general one, is
-- looks acceptable. The maximum flow rate comes
immediately after you initiate the break. And then in
the test it seems to reach a cutoff area, which is
corresponding to almost 120 cubic feet per minute,
while in the test and the RELAP prediction it's almost
150. This cutoff, it sounds, the test they used
vortex flow meter. And this is -- reach zero and
reading.
I don't know whether the RELAP adapted
some cutoff in this area or not, but it sounds similar
behavior. It reaches a certain point and then at a
drop zero. I will make sure about whether there is
anything adapted.
MR. SCHROCK: Well, the difference is
quite large at certain times.
MR. LAFI: The difference in the --
MR. SCHROCK: And one has to wonder, is
there an error in the stagnation state that's causing
that?
MR. LAFI: Actually I will mention the
stagnation first, but it sounds to me the opposite.
The stagnation is affected by whatever discrepancy I
saw earlier.
So if -- because here, if you look to
this, -- by the way, the stagnation or the onset of
stagnation happened in the main steam line break -- in
this test happened in cold leg 2 and 4.
The stagnation didn't occur in cold leg 1
and 3, which are both in the same side of the plant.
And the reason for this, I tried to find some cause in
the calculation. I plot the hot leg number 1 against
the steam generator number 1, the cold side of the
steam generator against the hot leg, which is the hot
leg side of the steam generator.
If you look to the hot leg, the steam
generator number 1, the test data, the red color, all
the time it's below this hot leg number 1. That means
still there is natural flow from the primary to the
secondary. In other words, still the steam generator
is acting as a sink to the primary.
This is the same situation, even little
bit different. If you look to the hot-leg temperature
prediction for RELAP, this one, this scale, against
the steam generator RELAP, number 1 RELAP calculation,
all the time also -- even the difference is not as
significant as in the real test -- all the time you
see the hot-leg temperature is higher than the steam
generator temperature.
This is -- it makes sense that I can
conclude that because of not having the stagnation in
cold leg 1 and 3 -- by the way, this is just a plot
for hot leg 1 against steam generator 1, just to
represent the behavior of cold leg 1 and cold leg 3
corresponding to the steam generator number 1.
This make me convinced that the reason for
this not having a stagnation, still I have a flow from
the primary to the secondary. If I go to the hot leg
number 2 against the steam generator number 2, just to
find a reason why we had stagnation in cold leg 2 and
4, if you look to this, whether RELAP or whether the
test, you see the steam generator 2, the first part
here is almost -- at almost 100 or 200 second you see
the hot-leg temperature is higher than the steam
generator temperature.
After that you will see the steam
generator, whether the test or the steam generator in
the RELAP calculation, you see is higher than the
temperature of the hot leg number 2.
This make me feel that now at this moment
the steam generator is acting as a heat source to the
primary. That's why it develop some potential to
reverse the flow. That's why we conclude that the
stagnation is due to this situation.
And if you go to the plot of the flow rate
for the cold leg number 2, you see the stagnation
happened almost at 500 second or so. It's actually
subjective. For the test maybe you can consider it at
this point, while for RELAP you can consider it later.
But the general behavior and the
stagnation occurred really in the RELAP calculation.
So RELAP predicted the stagnation like what happened
in the real test.
This is cold leg number 2. I said
stagnation happened 2 and 4 -- and also on 4. This is
the test against the experimental data.
CHAIRMAN WALLIS: The stagnation means no
flow rate; is that right?
MR. LAFI: Yes. The stagnation --
CHAIRMAN WALLIS: So the red is never
really zero unless there's an error in the plot. It
jiggles around as a negative and it crosses -- I guess
it crosses very, very briefly there. Does that --
MR. LAFI: Yeah. We --
CHAIRMAN WALLIS: Is that stagnation 4
long enough to really make any difference?
MR. LAFI: I consider -- I don't know. I
consulted the experimental team. They think that the
stagnation or when the flow meter read like negative
value, that means is it reverse or...
MR. REYES: The flow meters that we have
now installed, the electronics do allow us to
calculate or measure reverse flow.
CHAIRMAN WALLIS: Yes.
MR. REYES: -- reverse flow. Whenever it
says --
CHAIRMAN WALLIS: You're worried about
stagnation because it leads to the maximum or the
minimum temperature of the cold fluid going into the
vessel; is that --
MR. REYES: That was the -- the original
assumption was that --
CHAIRMAN WALLIS: That was the idea.
MR. REYES: -- was that if you stagnate
the cold legs you get stronger plumes in the
downcomer.
We're going to show tomorrow a little bit
of -- that's not always true.
CHAIRMAN WALLIS: So there's nothing
really magic about stagnation. It's not necessarily
the worst case.
MR. REYES: Correct, not for this plant.
CHAIRMAN WALLIS: It looks as though
there's quite a difference here. That in the test the
flow is getting very low. This RELAP is giving these
other bursts of flow in the two directions.
MR. LAFI: Yeah. It's isolated. But if
you compare it to this, for example, --
MR. REYES: Excuse --
MR. LAFI: You can't compare the previous
one to this one. I say there is no stagnation in cold
leg 1 or cold leg 3. This is true; there is no
stagnation. While in this...
MR. SCHROCK: That spike is the test data.
MR. ROSENTHAL: What's the accuracy in
your flow meter? Are we just looking -- should we be
painting this with a paint brush, with a... You're
talking about being off by two, three gpm.
MR. LAFI: Yeah, but I think if you take
the average -- this is my thinking -- that RELAP
calculation isolated actually from positive to
negative.
CHAIRMAN WALLIS: Well, should we care
about it?
MR. LAFI: At least the stagnation
occurred, but at different time. Even if you consider
the stagnation here, but I think -- but this will
contradict the temperature against -- the hot-leg
temperature and steam generator temperature will plot.
So that's why I thought the mechanism
behind the stagnation is whenever the steam generator
-- actually not exactly, because even in the test the
stagnation occurred not at exactly when this steam
generator exceeded the hot-leg temperature, that there
is some time in order to develop some potential to
reverse the flow.
That's why actually almost when the
temperature in RELAP, the temperature difference
between the steam generator and the hot leg reach
almost 80 degree, then the stagnation occurred. This
is what I saw even in the test, almost there is a
60-degree difference then the stagnation occurred.
MR. ROSENTHAL: Can you flip back to slide
20; would you mind, 20, if you -- that's good. So you
see if you go out about 1800 seconds you will see that
steam generator 2 RELAP versus --
MR. LAFI: This is --
MR. ROSENTHAL: That's good.
MR. LAFI: This one?
MR. ROSENTHAL: Yes, please.
-- about 1800 seconds --
MR. LAFI: Yes.
MR. ROSENTHAL: -- is a difference of like
100F. So that's --
MR. LAFI: The difference --
MR. ROSENTHAL: Okay. Now -- I'm sorry.
MR. LAFI: You say that --
MR. ROSENTHAL: No, between the blue --
MR. LAFI: Yes.
MR. ROSENTHAL: -- and the top --
MR. LAFI: And the top.
MR. ROSENTHAL: -- is about 100F.
MR. LAFI: Yeah. This is the difference
between the RELAP -- that's why there is discrepancy
between the RELAP prediction of the steam generator
temperature in both calculation and the test.
But there is in both, the test and the
RELAP prediction, the case occurred when the steam
generator temperature exceeded the hot-leg
temperature.
There is difference between RELAP
prediction for the steam generator temperature and the
test. But this does not mean that there is a case
when the -- this difference, whatever happened, for
example, in real test, when the temperature difference
exceeds 60 or 63 degree or almost -- I think 63 degree
the stagnation in the test occurred.
A RELAP calculation, and this is almost
here, when at 500 something, between 420 and -- almost
80 degree between the steam generator and the hot leg.
So this is where I saw stagnation, or at least I
considered the stagnation.
But if you talk about the discrepancy
between the RELAP prediction for the steam generator
temperature, I say there is a discrepancy.
Again I'm talking about -- I'm looking for
the cause of stagnation.
And if you look to this, also the steam
generator -- RELAP prediction for the steam generator,
not as good as the test, but all the time during the
test the steam generator temperature did not exceed
the hot-leg temperature. That's why we couldn't get
stagnation in both the test and the calculation.
CHAIRMAN WALLIS: So we can go on now?
MR. LAFI: Yeah. This is just the
stagnation, and this is another stagnation in cold leg
1 and cold leg number 3. And actually this is
supported by the fact that the steam generator during
the stagnation was full. And this is the plot of the
liquid volume fraction --
CHAIRMAN WALLIS: It's always liquid.
It's full of liquid; is that right?
MR. LAFI: Full of liquid, yes.
CHAIRMAN WALLIS: Or is it full of vapor?
Which is which?
MR. LAFI: No. This is liquid volume
fraction.
CHAIRMAN WALLIS: Liquid?
MR. LAFI: Yes.
CHAIRMAN WALLIS: Okay.
MR. LAFI: In RELAP they call it voidf.
The portion regarding the downcomer
temperature prediction by RELAP. I saw within 2,000
second RELAP prediction for the downcomer temperature
is in good agreement with the test.
And also I didn't notice this is the case
at 8 diameter in the downcomer and the same position
in RELAP. I plot different spot at 1.3 diameter; 2
diameter; 3, 4, 5 diameter; 'til 8 diameter. And I
saw the temperature profile. There is no significant
-- or actually there is no stagnation -- no
stratification.
This is the cold leg number 1, and the
same case with cold leg number 2.
As conclusions, the trend of the
pressurizer pressure is similar for the OSU-11 test
and RELAP5 prediction, although the depressurization
RELAP5 calculation was faster. This is what was
indicated in the first figure that I showed you.
RELAP5 successfully predicted the general
trend of the broken steam generator, starting with the
maximum flow rate, gradually decreasing to the cutoff
area, similar to what we notice in the test.
The steam generator pressure, the broken
one, was almost in a very good agreement. The flow
rate through the cold legs, the main steam line break,
the high-pressure injection system, and the downcomer
temperature almost in good agreement.
RELAP successfully predicted the
stagnation. Noted the same exact point. If we have,
for example, the pressurizer pressure here for exactly
and the injection system, which we say is
system-pressure dependent, then we will have
everything exact, but this is the problem with the
RELAP prediction.
CHAIRMAN WALLIS: How big is the cold --
is the injection nozzle?
MR. LAFI: The injection nozzle? The
high-pressure injection nozzle?
MR. REYES: I think it's about 1.3 inches.
CHAIRMAN WALLIS: How big?
MR. REYES: About 1.3.
CHAIRMAN WALLIS: It's this size hole?
MR. LAFI: Yeah, it is.
CHAIRMAN WALLIS: It's dribbling in at .1
gpm. It's hardly got any velocity at all. This is
scaled from a real plant?
MR. REYES: So the maximum flow rate, it
goes from about 1.1 gallons per minute per cold leg to
zero.
CHAIRMAN WALLIS: What?
MR. REYES: The maximum is 1.1.
MR. BESSETTE: It comes in at a low
velocity in the plant. It's about a -- something like
a foot a second in the plant.
CHAIRMAN WALLIS: The figure we have here
is -- yeah. It's one foot a second in the plant?
MR. BESSETTE: Something like that, yeah.
CHAIRMAN WALLIS: I thought it came in
gangbusters, hundreds of feet a second. It came
really in. I mean it's got several hundred PSI
driving it, isn't it, or is this not?
MR. BESSETTE: You see, it comes in at a
fairly large -- it comes in -- in the plant it comes
in at about a seven-inch pipe.
CHAIRMAN WALLIS: Well, this must be a
very different plant from the kind we used to -- we
did -- you know, Creare did experiments with cold-leg
injection and stuff trying to simulate fast. I think
we had a fairly small scale, but the water still came
pouring in through that injection nozzle at a pretty
high velocity.
MR. BESSETTE: It will. In the B&W Plant
it comes in at about 20 feet a second or so.
MR. REYES: Right.
CHAIRMAN WALLIS: So it's a completely
different beast?
MR. REYES: Right.
CHAIRMAN WALLIS: Okay. All right.
MR. LAFI: The hot leg number 2 and the
steam generator number 2 temperature histories
indicate that the steam generator became a heat source
at almost 180 second into the test in both the test
and calculation.
Loop flow continued for another 320 until
we reached the thermal potential, enough to reverse
the flow. So the difference at that time it was in
the RELAP calculation 80 degrees between the hot leg
and the steam generator. And then the primary loop
stagnation occurred.
Hot leg number 1, steam generator number
1 histories indicate that all the time hot leg number
1 exceed the temperature of steam generator number 1.
And this varies in -- there was no stagnation.
It can be concluded that the action of
steam generator number 2 as a heat source was the
cause of stagnation in cold leg 2 and 4, given that
the steam generator was full as indicated by liquid
volume fraction, the one that I showed you.
By comparing the steam flow rate out of
the break we can find the following: That the maximum
is almost -- happened exactly at the time when you --
when we initiated the break.
The flow experienced some sharp drop at
120 cubic foot a minute for the test, while for the
calculation it was 150.
There was similar gradual decrease in
between. Further assessment for the parameters
controlling the break flow in RELAP5 is being
conducted because there are some parameters that
affect the flow out of the break.
CHAIRMAN WALLIS: This is just a steam
flow, is it? It's not a two-phase flow?
MR. LAFI: This is the steam flow.
CHAIRMAN WALLIS: So you would expect to
be able to predict it quite well?
MR. LAFI: Actually I am trying to play
with the -- because here there is some parameters --
this is what I expect for --
CHAIRMAN WALLIS: I'm sorry to -- when you
say "cfm," that means that's some condition? I mean
it's not a mass -- usually it's a mass flow rate that
you want.
MR. LAFI: This is part of --
CHAIRMAN WALLIS: Because I don't know
what the condition is and we're valuating the cfm at.
I guess you can do cfm. Was it standard cubic feet
per minute, or something, or is it... What is it,
velocity times area; without --
MR. LAFI: This is volume --
CHAIRMAN WALLIS: -- any reference to
density?
MR. LAFI: -- actually cubic foot per
minute, right?
CHAIRMAN WALLIS: So it's volumetric load.
MR. LAFI: Is it 120, it was.
MR. SHACK: But is it reduced to a
standard condition, or is it just v times a?
CHAIRMAN WALLIS: Velocity times area? It
must be.
(Comments off the record.)
MR. LAFI: Converted to mass flow rate?
MR. REYES: The flow meter rate, I
believe, is the standard.
CHAIRMAN WALLIS: It's just velocity times
area?
MR. REYES: The vortex. That's it.
MR. LAFI: Is the vortex flow meter --
MR. REYES: So the vortex flow meter is
standard.
CHAIRMAN WALLIS: It's kind of strange,
because I mean if you were dropping the pressure and
you can get sort of the same velocity, but a lot less
mass flow rate -- no. Maybe you can -- that's going
to be all clear when we read the report?
MR. LAFI: That test is 12, which is
similar to 11.
CHAIRMAN WALLIS: I think the ACRS has a
preference for never using cfm or gallons per minute
as a unit-to-flow rate, because different gallons and
different pressures and temperatures. If you use
mass, then it's clearer what you -- what's going on.
MR. SHACK: Sometimes gallons are nice,
but...
CHAIRMAN WALLIS: Well, gallons of cold
injection are quite different from gallons of hot
ejection in terms of mass flow.
MR. SCHROCK: In your detailed comparisons
you have some poor results for hot leg -- or for steam
generator number 2 temperature. And then you have
some poor results for the collapsed liquid level
beyond 2,000 seconds. But those poor predictions
don't seem to be reflected in your summary
conclusions.
MR. LAFI: Actually the poor prediction
for the pressurized level actually -- in the first
portion it was good comparison, while in the -- almost
1, 3, 16-inch in the test, that was -- after it
reached the 26-inch the pressurizer heater turned on
for --
MR. SCHROCK: It what?
MR. LAFI: In the real test, when the
pressurizer level reached 26 amps, the heater -- the
pressurizer heater turned on.
For RELAP this 26-inch -- and I can show
you.
MR. SCHROCK: I don't understand what the
heater on or off has to do with the level.
MR. LAFI: So you're asking me about the
poor prediction of RELAP for the pressurizer level?
MR. SCHROCK: Yeah. I just looked at your
comparisons of test data against predictions, and I
see it made poor predictions of collapsed liquid level
beyond about actually 1500 seconds. And it made poor
predictions of the steam generator number 2
temperature.
But, as I listened to your summary
descriptions, it seemed as though those rather poor
predictions are not reflected in your summary.
MR. LAFI: I -- my --
MR. SCHROCK: Are they regarded as
insignificant, or what's the -- what should be --
MR. LAFI: Actually --
MR. SCHROCK: -- interpreted from that?
MR. LAFI: -- I expect, my conclusion that
the general run of RELAP predictions and, of course,
all acceptable agreement with the first.
Consequently, I don't expect from RELAP to
match exactly what happened in the test.
MR. SCHROCK: Well, the difference in
pressurizer level of 70 and less than 50 is a
significant amount of water.
MR. LAFI: Yeah.
MR. SCHROCK: And so I would think that
would have some -- if I didn't know anything else
about the test, I'd suspect that there's something in
the calculation that needs to be made better in order
to get reliable predictions from that code.
CHAIRMAN WALLIS: Well, that's the problem
with all of these comparisons. We don't have a
criterion for saying what's good enough, and what you
have to do, and how you're actually measuring the
goodness of RELAP with all these various wiggles, and
squiggles, and lines, and curves, and things. I mean
--
MR. SCHROCK: Yeah. But I keep --
CHAIRMAN WALLIS: -- that's the always the
problem.
MR. SCHROCK: -- noticing these things and
I mention them when I notice them.
CHAIRMAN WALLIS: Yeah. Well, I think you
really ought to -- really the NRC should start off
with some kind of an intellectual roadmap which says
how do we make these comparisons, what are we looking
for, and --
MR. BESSETTE: The thing about pressurizer
level, and it may -- it may be due to this difference
in the HPI flow between the -- see, pressurizer level
in this case is an effect of something else. So this
--
CHAIRMAN WALLIS: The extra water has to
go somewhere.
MR. BESSETTE: Yeah.
CHAIRMAN WALLIS: So that's a good, simple
principle.
MR. BESSETTE: So it looks like it goes
back to this difference in the HPI calculations.
CHAIRMAN WALLIS: Because this is the only
-- pressurizer is the only place which can accommodate
extra water.
MR. BESSETTE: Yeah, right.
CHAIRMAN WALLIS: The rest of it's solid.
MR. ROSENTHAL: And let me point out that
50 percent versus 70 percent of the pressurizer volume
is a much -- it looks like a lot, but the pressurizer
is, what, 10 percent or something of the total system
volume, so it really isn't that big of a deviation.
MR. BESSETTE: So I mean it's got to
relate back to a difference in --
CHAIRMAN WALLIS: I think we have
something similar with the next one, too. We have a
similar difference with the pressurizer level and come
back to that.
MR. BESSETTE: It can only be due to a
difference in the RELAP calculation of the injection
or the system temperature.
CHAIRMAN WALLIS: As long as RELAP is
conserving mass.
(Laughter.)
CHAIRMAN WALLIS: Maybe it's conserving
gallons of something. We're in trouble.
MR. BESSETTE: There you go. Conserving.
CHAIRMAN WALLIS: Okay. We should go on,
I think.
MR. LAFI: The next test is CE-12, which
is similar to one-foot-square main steam line break
initiated from 610 kilowatt from full power.
Steam generator number 2 power-operated
relief valve was open to simulate the break. And then
upon initiation the break, the same thing, the reactor
coolant pumps were tripped. The power and state of
test 11, it was kept constant at low pressure -- lower
power. Here it converted or switched to a decay mode.
And this decay was included in the input.
The pressurizer heater were tripped upon
the collapsed liquid level of the pressurizer, were
turned off -- switched between on and off. Again 16
under 26.
The auxiliary feed water was maintained
for the broken steam generator, which is the case of
11, for 10 minutes. And it was isolated from the
intact one. And this is the sequence of events of the
real test. And again the RELAP Code was run for
steady -- was run for steady-state for 1200 second.
And we established the initial condition, which was
almost similar to the initial condition of the real
test.
The mass, the auxiliary -- the feed water
mass flow rate is higher in this case. And this is
almost -- the pressure is different. In the previous
test it was 272, this one 232. Almost -- the other
parameter is close to each other, so this make me
satisfied that I will start my transient case.
So the comparison, also the behavior of
the pressurizer pressure not exactly but similar to
what I saw before, the depressurization rate and RELAP
prediction is faster, especially in this area.
And the steam generator pressure of the
broken one, RELAP prediction just in good agreement
trend-wise with the test.
This is the pressurizer level.
CHAIRMAN WALLIS: So you got the same
problem as with the previous one?
MR. LAFI: The same thing.
CHAIRMAN WALLIS: Except the HPI
prediction is okay. Look at that one after this. It
looks as if you're predicting the injection rate
right. Where is the water coming from or going to?
Maybe it's a question of getting the temperature
right.
MR. LAFI: Can I -- could I just one...
CHAIRMAN WALLIS: You're not -- you're not
losing water, are you, from this?
MR. LAFI: Actually I'm not losing water
in this test.
MR. ROSENTHAL: Well, then something's
hotter.
CHAIRMAN WALLIS: Something's hotter.
This is the steam -- the water is more -- expanded
more somewhere.
MR. ROSENTHAL: Someplace.
CHAIRMAN WALLIS: Yes. Okay.
Well, maybe we should move along. But I'm
not sure we learned -- what did we learn from that?
MR. LAFI: This is --
CHAIRMAN WALLIS: What are you testing,
that RELAP conserves mass, or something?
MR. LAFI: The high-pressure injection
system RELAP predicted well, compared to the test
data.
CHAIRMAN WALLIS: Presumably, unless
there's one of these traces which is invisible. I
mean there are eight traces, and I can't see eight.
I assume they're all on top of each other.
MR. LAFI: Yes. This is the RELAP
prediction, the solid line. This one. You cannot
recognize, but this is RELAP.
The stagnation occurred in this test in
cold leg 1 and 3. In the previous test it occurred in
2 and 4. Now 2 and 4, no stagnation. This is 4, no
stagnation.
And 1 and 3, you will see some kind of a
stagnation and resumption of the flow in both test and
RELAP prediction. So RELAP in this situation, this
case, it predicted what we saw in the test,
stagnation, and then followed by resumption of the
flow.
And when I tried to look for the reason,
it sounds the same reason, the same mechanism, when
the steam generator becomes a heat source, you would
have stagnation. When it goes back to the normal
situation, the flow will be resumed or will resume.
This is similar to what I saw in cold leg
3, stagnation and resumption of the flow in both tests
and a RELAP prediction, although at different times.
And this is the reason again. This is supported by
the plot of hot leg number 1 against steam generator
number 1.
You will see that the steam generator
number 1 for the test. And this is the hot leg number
1 -- what is it -- for the test. This one.
So when the steam generator higher in
temperature than the hot leg, then you will have the
stagnation. Whenever you have the hot leg exceed the
steam generator temperature, then the flow will
resume. This has happened in this point and this
point with RELAP prediction.
So at this point the hot leg temperature,
also predicted by RELAP, exceeded the steam generator
temperature. That's why the flow was resumed.
And these spots is corresponding to the
time of occurrence of stagnation and resumption that
I showed you.
Even, again, the steam generator
temperature is not exact between RELAP prediction and
test, but all the time this condition hold. You have
heat sink, steam generator as a heat sink. You would
have stagnation. When it comes back it will go to the
normal flow.
For steam, for not having stagnation in
hot leg number 2 and number 4, this is the reason that
hot leg temperature all the time exceed the steam
generator temperature.
Again, again, the temperature of the steam
generator not as exact -- as exact or similar to the
test.
This is supported also by being in the
steam generator at that time. When the stagnation
occurred it was full. There is no voiding.
The temperature profile for the downcomer.
This is at eight diameter. And it match the test to
this portion exactly, while after that it match the
trend of the temperature profile.
This has happened also with the cold leg,
because here at RELAP the downcomer was divided into
sectors, so each -- when I say cold leg 2 downcomer it
is corresponding to the sector that is in the cold leg
number 2 side. So this is the temperature profile in
the downcomer corresponding to the cold leg number 2.
And again I tried all of the cold legs,
and it seems the cold -- the downcomer temperature is
well mixed and there is no stratification.
I plot some data at 4 and at 6 and at 8.
There was no stratification. But when I plot the
downcomer temperature, even it does not show, there is
some almost 20-degree or maybe 30- --
CHAIRMAN WALLIS: It seems to me the
conclusions from this experiment are much like the
conclusions you drew from the previous experiment.
MR. LAFI: Exactly.
CHAIRMAN WALLIS: So maybe we don't need
--
MR. LAFI: Except -- except the
resumption, and actually --
CHAIRMAN WALLIS: So maybe we don't need
to read through all the conclusions.
MR. LAFI: Okay.
CHAIRMAN WALLIS: Is there something new
in the conclusions?
MR. LAFI: I think there's no significant
--
CHAIRMAN WALLIS: It's very similar to the
last conclusion.
MR. LAFI: The only -- what? The only
difference is I think the resumption of the normal
flow. This is what happened in this test.
Now the last test is the OSU-CE-08, which
is a break. That is located at the top of the hot leg
number 1, which is two-inch break.
Again this test, not all the reactor
coolant pumps were tripped at the same time. Two of
them, called pump 1 and 4, were tripped. And then the
second and fifth were tripped based on subcooling.
The pressurizer heater also allowed to
cycle on and off based on the pressurizer low level.
The sequence of events for this test will
be explained in the comparison.
A steady-state test was run like before
for 1200 seconds.
CHAIRMAN WALLIS: You've got flow rate
stagnating all of the cold legs --
MR. LAFI: Yes. Yes.
CHAIRMAN WALLIS: -- at the same time?
MR. LAFI: No. At different times.
CHAIRMAN WALLIS: But they -- you don't
have it stopping. Does it -- so I guess that, in your
sequence of events, cold leg 1 stops. There seems to
be a long period when there's no flow in any of the
cold legs, right? You take away that thing, will you?
That thing between 1841 and --
MR. LAFI: This is not RELAP prediction,
by the way.
CHAIRMAN WALLIS: It seems to me between
4723 and 5326, we have no flow in any of the cold
legs, because none of them are restarted yet; is that
right?
MR. LAFI: For the real test, tell --
CHAIRMAN WALLIS: They all stagnated at
the same -- they're still all stagnated -- at 4723
they are all stagnated?
MR. LAFI: From 1841 until 4,723.
CHAIRMAN WALLIS: They are all stagnated?
MR. LAFI: Yes.
CHAIRMAN WALLIS: So there's -- what's
cooling the core, just the HPSI dribbling in? The
HPSI's over, too, isn't it then?
MR. LAFI: Let's see, HPSI at that time --
CHAIRMAN WALLIS: It's finished.
MR. LAFI: -- if it is injecting.
CHAIRMAN WALLIS: So there's nothing
happening. It's just sitting there. Where's the heat
going? It's going out the break? It's just boiling
it? Is the pot boiling it out the break and there's
no circulation through anywhere?
MR. BESSETTE: That's right. The break is
big enough to take out the decay heat.
CHAIRMAN WALLIS: Okay.
MR. BESSETTE: And the HPSI is on the
whole time.
CHAIRMAN WALLIS: HPSI's on the whole
time?
MR. BESSETTE: Yeah. Or starting at 40
seconds earlier.
CHAIRMAN WALLIS: So you don't want to
close the break and repressurize.
MR. LAFI: HPSI all the time except if it
reach above the 360 psig there's no HPSI.
MR. BESSETTE: If you did close the break,
at some point those generators would become active
again.
MR. SCHROCK: Well, why have they got all
these ADS things?
MR. LAFI: They're calculated versus
measured initial condition, as shown in this table.
The pressurizer pressure behavior trend was predicted
by RELAP but, as you see, there is some -- at the
first portion it was acceptable.
Here it reach a certain plateau for both
of them, the test, and the RELAP prediction. And then
it goes -- it decrease in both tests and RELAP
prediction.
CHAIRMAN WALLIS: Well, now the pressure
in the system is determined by the heat generation
rate and the flowing out the break, isn't it? And the
pressure it takes to drive that flow right out the
break to carry out the heat essentially.
MR. LAFI: So that's --
CHAIRMAN WALLIS: You've have a certain
amount of heat making a certain amount of steam and it
has to go out the break, so the pressure is big enough
to get that right out the break. I'm saying that it
looks simple enough. You ought to be able to predict
the pressure pretty well.
MR. LAFI: Actually even this -- this will
reflect, this discrepancy between RELAP prediction and
the test will reflect on the flow rate of the break.
You will see big difference --
CHAIRMAN WALLIS: Yeah, that's right. You
ought to be able to predict that pretty well, or is it
--
MR. LAFI: No, it's not very well.
CHAIRMAN WALLIS: Is it because you can't
predict the liquid carry over out the break just like
what we talked about this morning, or what is it
that's difficult about the break flow? It's a
two-phase flow out the break? It's a two-phase flow.
MR. LAFI: It should be two-phase.
CHAIRMAN WALLIS: So the difficulty is
because you can't predict the two-phase flow very well
out the break, but you have these deviations here?
MR. LAFI: Actually what I know, that
maybe the critical flow model at the break, this led
to this problem because, as I understand, Henry
Foskey's particular model is adapted by this version
of RELAP. This is -- so I am thinking to look for
that parameter that control, for example, the
discharged coefficient or...
I will show you the break flow later.
This is actually the feed water flow.
This is just -- I used whatever the test used. And
this is the feed water mass flow rate RELAP and the
test. And the injection, since the pressure is
different, the injection now funnels because, as I
said, the HPSI is system-pressure dependent.
So if you look to the top here is the test
results, while the lower is the RELAP prediction.
CHAIRMAN WALLIS: So those four RELAP5 are
all in that one bottom curve, are they?
MR. LAFI: Actually four for RELAP. This
is --
CHAIRMAN WALLIS: The fuzzy curve is the
four tests, right? That bottom curve is -- all the
RELAP5 are on top of each other; is that right?
MR. LAFI: Yes. The level one is four
RELAP prediction while the top is four HPSI. The test
--
CHAIRMAN WALLIS: This is presumably
because you have the pressure system pressure wrong?
MR. LAFI: Exactly, different.
CHAIRMAN WALLIS: Right.
MR. LAFI: So it behaves -- if you look,
the pressure goes like this and down, it's different
from the test. That's why this is a discrepancy.
CHAIRMAN WALLIS: They don't cross at the
same point, but they...
MR. LAFI: As I told you, the break flow
is not predicted except at the beginning, but the
behavior reached maximum, then go down and then
oscillate.
MR. SCHROCK: What is the test measurement
of break flow?
MR. BESSETTE: But how is the break flow
measured?
MR. REYES: The break flow --
MR. BESSETTE: It goes --
MR. REYES: -- on --
(Simultaneous discussion held among others
in the room off the record.)
MR. REYES: We are using a separator
system. This is all --
MR. BESSETTE: So he runs it through the
separator, and he measures the -- you measure the
liquid flow --
MR. REYES: The closed end and cfms.
MR. BESSETTE: And the vapor flow.
MR. REYES: Correct.
MR. SCHROCK: And you get these kinds of
oscillations.
CHAIRMAN WALLIS: I don't know which is
the test and which is the RELAP.
MR. SCHROCK: The liquid coming out of the
separator.
MR. KRESS: Well, I can tell you which.
This is the test and this is the RELAP.
MR. LAFI: The problem is the test --
CHAIRMAN WALLIS: So we have a question
now. We say that RELAP is predicting more flow out
the break, about twice as much as in the test, so I
gather from this curve. When I go back to page 58 --
MR. LAFI: Yeah. It's overpredicted --
CHAIRMAN WALLIS: Overpredicted the break.
MR. LAFI: Yes.
CHAIRMAN WALLIS: We go back to page 58.
RELAP is holding the pressure up higher.
Do you think if it's predicting more flow
out the break it would depressurize faster? It
doesn't seem right.
MR. BESSETTE: It seems peculiar.
MR. SHACK: Yeah, something's wrong.
Yeah.
CHAIRMAN WALLIS: Something seems strange.
I don't know if it's wrong. It's just strange. I
mean --
MR. LAFI: What's this conflict?
MR. BESSETTE: Could you flip back to the
pressure comparison?
CHAIRMAN WALLIS: This --
MR. BESSETTE: Page 58.
CHAIRMAN WALLIS: RELAP holds the pressure
up.
MR. LAFI: RELAP within 1,000 seconds --
CHAIRMAN WALLIS: In the early part.
MR. (SPEAKER): Oh, in the early part.
CHAIRMAN WALLIS: The early part up to
about 5,000, RELAP is predicting a higher pressure
than reality. And yet RELAP's also predicting a
higher break-flow rate, which is consistent with
having a higher pressure.
But if you look at -- you'd expect the
higher flow out the break to depressurize faster.
That's why it doesn't seem to make sense.
MR. HAN: How about HPI in comparison?
CHAIRMAN WALLIS: HPI is being predicted
pretty well.
MR. LAFI: Yeah. It follows this one.
Look, it follows this --
CHAIRMAN WALLIS: I guess this is the
first time we're able to see something we can latch
onto which we can get cause and effect and try to
figure it out. I'm not sure we really need to pursue
it, but it does look a bit odd.
So it's something for you to think about
this. Can we leave it at that.
MR. SCHROCK: I have a question about your
initial conditions. The RELAP prediction of initial
conditions seems to me to be impressively good with
one exception, and that is the steam generator number
1 and 2 water level, seems to be off by quite a bit,
not just in this particular experiment but in the
others, too.
Is there an explanation for that?
MR. LAFI: Actually I noticed this. Even
if I start with the same volume of the steam generator
level, like the test, it leads me to higher volume.
And I couldn't find out what's the reason, but this is
what happened in many cases.
For example, if I see the initial steam
generator 15 and I set it as 15, it run --
CHAIRMAN WALLIS: To 27.
MR. LAFI: -- for a few seconds and then
it goes to 25, or something.
CHAIRMAN WALLIS: And you can't even blame
Bill Gates for that.
MR. LAFI: No.
(Laughter.)
MR. LAFI: As I said, the stagnation
occurred in all cold legs, similar to the test but at
different times.
And even I can't conclude, because it
sounds to me subjective whether I consider the
stagnation in the -- for example, in the test, it is
clear for RELAP. I don't know what I --
CHAIRMAN WALLIS: You have to stand by the
mic or you -- we lose the transcription.
MR. LAFI: It sounds to me subjective to
determine which is as far as that. I will -- but I
thought without the test, I will choose some point,
but this is what happened. RELAP predict the
stagnation, but at different time.
Not predicting the stagnation at its exact
time is due to other discrepancy in other factors.
CHAIRMAN WALLIS: I guess what really
matters here is what difference does it make to your
assessment of pressurized thermal shock if you have
these kinds of differences. And I have no idea.
How accurately do you need to know this
sort of -- is it stagnated, or is it close to
stagnation, and all that, in --
MR. LAFI: Actually --
CHAIRMAN WALLIS: -- order to assess
pressurized thermal shock?
MR. LAFI: It seems the stagnation -- I
found no stratification in the cold -- in the
downcomer temperature after stagnation. And this is
what we noticed in the test. Also the temperature in
the downcomer is uniform during the entire test.
CHAIRMAN WALLIS: Everywhere?
MR. LAFI: Huh?
CHAIRMAN WALLIS: It doesn't stratify
vertically, or anything; it's uniform everywhere?
MR. REYES: There are some tests where we
see vertical stratification. And as far as the timing
of when stagnation occurs, in this plant we -- because
of the low injection flow rates, we see a relatively
good mixing under stagnant conditions.
In other plants stagnations are important
because if you have high injection flow rates under
stagnant conditions you may see more penetrating
plumes. So the timing becomes an issue.
Tomorrow I'll be presenting. We'll look
at this scenario again and go --
CHAIRMAN WALLIS: See, I have no criterion
for deciding is this good, or bad, or indifferent, or
what shall I conclude, or --
MR. REYES: Right.
CHAIRMAN WALLIS: -- why should I worry,
or should I.
MR. REYES: Tomorrow what we'll do is
we'll look at loop stagnation mechanisms. That'll be
the first talk. And we'll come back to this, what was
causing stagnation in this particular test. And we'll
relate that to some of the separate-effects tests and
what we were seeing there.
MR. LAFI: This is cold leg 4. And
actually what I discovered through a RELAP calculation
that the mechanism of stagnation in this test is
different from the mechanism of stagnation in your
tests, as I will show it to you. This is -- again
stagnation occurred but different time.
This is cold leg 3. And if you notice
this, the hot leg temperature number 1 against steam
generator number 1 temperature is almost higher, the
hot leg temperature higher than the steam generator
temperature, except it reach a point when they are the
same.
And at this point, when I discovered the
steam generator voided, empty. And that's why I
conclude that the mechanism of the stagnation
according to RELAP calculation for this test is the
voided of the steam generator. This is supported by
this figure.
CHAIRMAN WALLIS: Now this -- wait a
minute. This is TF 143 is a thermocouple in the hot
leg?
MR. LAFI: Yes. 143, yeah, in the hot
leg.
CHAIRMAN WALLIS: And it has these
enormous dives that go off the picture and the one
picture before that. Is that real, or is that a
glitch in the instrumentation? Why do those things go
down to the bottom of the graph there?
MR. LAFI: This one, this figure or --
CHAIRMAN WALLIS: Yeah. There's -- the
color I can't describe.
(Comments off the record.)
CHAIRMAN WALLIS: Whatever you call that.
MR. LAFI: TF 143?
MR. SHACK: Teal.
CHAIRMAN WALLIS: Teal. Teal, or
something.
CHAIRMAN WALLIS: Yeah, Teal.
Why does it go down to -- is that a real
thing, or is that a...
MR. LAFI: With the temperature profile?
CHAIRMAN WALLIS: The temperature plunges.
It comes back.
MR. SHACK: The measured thermocouple
response.
MR. SCHROCK: Right there.
MR. LAFI: This one?
MR. SHACK: Yeah.
CHAIRMAN WALLIS: And it's in a hot leg.
Why does it do that? Does it suddenly sees a slug of
cold liquid?
MR. LAFI: Is this the --
MR. REYES: I don't -- I don't believe
that's --
CHAIRMAN WALLIS: If you see something you
don't like, you don't believe it?
MR. REYES: No, no. There are -- there
were two thermocouples that we were looking at
earlier. One was giving us spikes in the high
direction. And, of course, we want to look at both of
those, but --
CHAIRMAN WALLIS: So --
MR. REYES: So the --
CHAIRMAN WALLIS: Give them a break and
let them have low ones, too.
MR. REYES: That's right. We think there
might be some noise problem with --
CHAIRMAN WALLIS: It's a noise problem.
It's not a real thing?
MR. SHACK: Just put a bigger averaging in
the circuit.
MR. SCHROCK: Somebody put a little ice
water in there.
MR. LAFI: This condition, again, occurred
with hot leg number 2 and steam generator number 2.
And this is when the stagnation occurred when you have
the hot leg temperature as equal to the steam
generator temperature, which is corresponding now to
the voided of the steam generator.
This is steam generator number -- what is
it? The -- this one 225, which is the steam generator
number 2, this one. And this one's steam generator
number 1.
So this is the time when the stagnation
occurred. I didn't, and I doubt RELAP can predict
what is happening in the other location. But this is
what I connected the stagnation cause because of the
voidage of the steam generator going from volume
fraction 1 to zero.
And the downcomer temperature profile is
acceptable, except RELAP overpredicted the downcomer
profile.
And this is the cold leg to downcomer,
which is the sector corresponding to the area when the
cold leg number 2 connected to the downcomer.
As the conclusion, RELAP predicted general
behavior of the system depressurization, the
high-pressure injection system, feed water, and the
break-flow rates. Just in general for the break-flow
rate. We see -- we saw some significant difference.
By comparing the break flow rate one can
notice that the maximum break-flow rate at the
beginning of the test was 12 ga- --
CHAIRMAN WALLIS: So this is a two-phase
flow, not the break?
MR. LAFI: Out of the break, I suppose.
CHAIRMAN WALLIS: It's a two-phase flow?
MR. LAFI: I suppose two-phase flow.
CHAIRMAN WALLIS: What's a gallon of
two-phase? I don't understand the break flow of
gallons per minute.
MR. REYES: You're looking at the liquid
flow as the --
MR. LAFI: This is the RELAP prediction
liquid.
MR. REYES: Liquid?
MR. LAFI: Yes.
CHAIRMAN WALLIS: The beginning of the
test it's all liq- --
MR. LAFI: Because the mass flow rate
for... However, RELAP5 overpredicted.
CHAIRMAN WALLIS: This break flow in the
two-phase is still measured in gallons per minute?
MR. REYES: Basically what we have is a
separator with a loop seal. And the magnetic flow
meter on that loop seal measures in gallons per
minute.
CHAIRMAN WALLIS: So you're measuring the
-- okay.
MR. REYES: So I think what he's comparing
there then is just that flow meter liquid --
MR. LAFI: Comparing not after leaving the
break.
CHAIRMAN WALLIS: So it's the liquid flow.
MR. LAFI: After separating the liquid
from the two-phase.
CHAIRMAN WALLIS: So this is the liquid
flow out the break?
MR. LAFI: Yes.
CHAIRMAN WALLIS: Then the steam flow
isn't counted?
MR. LAFI: No. This is what -- the data
I think -- after I think the separation, right? Yeah,
it depends.
MR. REYES: So in RELAP were you comparing
just the liquid?
MR. LAFI: The mass -- the mass flow rate
for the liquid RELAP. That's why I compared this to
--
CHAIRMAN WALLIS: So the steam isn't
counted in some way?
MR. LAFI: There is control volume. It
occurred in RELAP to calculate the mass flow rate for
the vapor, for the steam.
CHAIRMAN WALLIS: So you have a phase
separator?
MR. REYES: Right.
CHAIRMAN WALLIS: You measure the water
and steam flow rates?
MR. REYES: Separately.
CHAIRMAN WALLIS: And you measure them
both in gallons per minute?
MR. REYES: No. They're -- the magnetic
flow meter is in gallons per minute. The vortex flow
meter is in the standard cubic feet per minute.
CHAIRMAN WALLIS: Standard cubic feet per
minute? Standard cubic feet per minute?
MR. GROOME: Well, most flow experts do
not measure mass. Only Coriolis measured mass. So
the raw data is in volumes. If you want mass, there's
--
CHAIRMAN WALLIS: It drives me up the
wall. And graduate students get always very confused
by all flow meters because they have these stupid
measurements, like standard cubic feet per minute
which no one understands. People are always
misunderstanding.
MR. REYES: I agree.
CHAIRMAN WALLIS: Okay. Sorry.
MR. LAFI: RELAP5 predicted flow
stagnation in all cold legs at different times, as was
the case during the test.
The stagnation cause in this test is not
similar to that of the main steam line break tests.
It is believed that the cause of stagnation here is
the voiding of the steam generator tubes.
RELAP5 prediction for the downcomer
temperature profile is in good agreement with the
test. RELAP5 is one-dimensional cold. Therefore,
studying the stratification that occurred during the
test in the cold legs and the loop seals was not
possible.
I think this is the end of my
presentation.
CHAIRMAN WALLIS: Thank you.
That's the end of the day's presentations,
Jose?
MR. REYES: That's correct.
CHAIRMAN WALLIS: So we're finished ahead
of time. And we're going to go and look at the
experiment now?
MR. REYES: We can do that for a short
while.
CHAIRMAN WALLIS: Yes. I think that would
be very appropriate, while it's cold and we can get in
there.
MR. REYES: That'd be fine.
CHAIRMAN WALLIS: Then we're going to see
it tomorrow when it's hot and running?
MR. REYES: Hot and running.
John has guaranteed it.
MR. GROOME: Well, I don't know. We're
going to be in gallons per minute.
(Laughter and comments off the record.)
(Whereupon, the meeting was adjourned at
3:05 p.m. on July 17, 2001, to resume on Wednesday,
July 18, 2001 at 8:15 a.m. in Corvallis, Oregon.)
Page Last Reviewed/Updated Tuesday, August 16, 2016