487th Meeting - November 9, 2001
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
487th Meeting
Docket Number: (not applicable)
Location: Rockville, Maryland
Date: Friday, November 9, 2001
Work Order No.: NRC-102 Pages 357-596
NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers
1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005
(202) 234-4433. UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
+ + + + +
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS (ACRS)
487TH MEETING
+ + + + +
FRIDAY,
NOVEMBER 9, 2001
+ + + + +
ROCKVILLE, MARYLAND
+ + + + +
The committee met at the Nuclear Regulatory
Commission, Two White Flint North, T2B3, 11545
Rockville Pike, Rockville, Maryland, at 8:30 a.m.,
George E. Apostolakis, Chairman, presiding.
COMMITTEE MEMBERS:
GEORGE E. APOSTOLAKIS, Chairman
MARIO V. BONACA, Vice Chairman
NOEL F. DUDLEY, Member
F. PETER FORD, Member
THOMAS S. KRESS, Member
GRAHAM M. LEITCH, Member
DANA A. POWERS, Member
STEPHEN L. ROSEN, Member
COMMITTEE MEMBERS: (CONT.)
WILLIAM J. SHACK, Member
JOHN D. SIEBER, Member
GRAHAM B. WALLIS, Member
ACRS STAFF PRESENT:
JOHN D. SIEBER, ACRS
SAM DURAISWAMY, ACRS/ACNW
HOWARD J. LARSON, ACRS/ACNW
SHER BAHADUR, ACRS/ACNW
CAROL A. HARRIS, ACRS/ACNW
JOHN T. LARKINS, ACRS/ACNW
MAGGALEAN W. WESTON
MEDHAT M. EL ZEFTAWY
MICHAEL T. MARKLEY
ALSO PRESENT:
LARRY MATHEWS
STEVEN MOFFAT
WARREN BANFORD
ALLEN HISER
ED HACKETT
KEN BYRD
ALEX MERRION
JACK STROSNIDER
RAJ PATHANIA
DAVE GEISEN
ALSO PRESENT: (CONT.)
MARK REINHART
TOM KING
N. PRASAD KADAMBI
ERIC J. BENNER
JIM LYONS
EDWARD McGAFFIGAN (Commissioner)
TONY ULSES
STEWART BAILEY
. I-N-D-E-X
Opening Remarks by the ACRS Chairman . . . . . . 361
Circumferential Cracking of PWR Vessel
Head Penetrations . . . . . . . . . . . . 362
Licensing Approach for the Pebble Bed. . . . . . 476
Modular Reactor Design
Meeting with Commissioner McGaffigan . . . . . . 529
Staff Presentation on Power Uprate . . . . . . . 575
Adjourn. . . . . . . . . . . . . . . . . . . . . 596
. P-R-O-C-E-E-D-I-N-G-S
(8:31 a.m.)
CHAIRMAN APOSTOLAKIS: The meeting will
now come to order.
This is the second day of the 487th
meeting of the Advisory Committee on Reactor
Safeguards. During today's meeting, the committee
will consider the following: circumferential cracking
of PWR vessel head penetrations; licensing approach
for the pebble bed modular reactor design; future ACRW
activities/report of the Planning and Procedures
Subcommittee; reconciliation of ACRS comments and
recommendations; preparation for the meeting with the
NRC Commissioners; proposed ACRS reports.
In addition, the committee will meet with
Commissioner McGaffigan to discuss matters of mutual
interest.
This meeting is being conducted in
accordance with the provisions of the Federal Advisory
Committee Act. Mr. Sam Duraiswamy is the designated
federal official for the initial portion of this
meeting.
We have received a request from the
Nuclear Energy Institute for time to make oral
statements regarding circumferential cracking of PWR
vessel head penetrations.
A transcript of portions of the meeting is
being kept and it is requested of the speakers use one
of the microphones, identify themselves and speak with
sufficient clarity and volume so that they can be
readily heard.
I would also like to remind Members that
during lunch time we are scheduled to interview three
candidates for potential membership on the ACRS.
The first item on our agenda is
circumferential cracking of PWR vessel head
penetrations. Dr. Ford is the cognizant member and he
will lead us through the discussion.
MEMBER FORD: Thanks, George. In July, we
heard a presentation on the proposed bulletin relating
to the vessel head penetration cracking. At that
time, we wrote a letter to Chairman Meserve indicating
that the bulletin was both appropriate and timely.
Bulletin 2001-01 was issued on August 3, 2001.
In the response back to us from the staff,
it was indicated that the update on the staff
assessment will be given early next year. That's
2002. Because of the urgency and importance of this
issue, the committee decided to request an interim
presentation on the status of various issues that came
up in the July meetings. These issues included risk
assessment, prioritization algorithms, evaluation of
inspection methods, responses to the bulletin and
question of cracking initiation and kinetics, both for
the axial and circumferential cracks.
Therefore, the purpose today is to hear
generic discussion of these issues. There's no plan
at this stage to issue a letter.
As you heard from the Chairman, NEI has
requested -- I'm just looking for the Member from NEI
-- has requested to make a statement. We'll hear that
statement at the end of the industry presentation.
We'll begin with Larry Mathews from MRP.
MR. MATHEWS: My name is Larry Mathews.
I'm with Southern Nuclear Operating Company. I'm the
Chairman of the Alloy 600 Issues Task Group of the
Materials Reliability Program.
I'm just going to provide an update on a
few issues of where the industry is on some of these
issues with respect to the head penetration cracking.
Some of the topics I'm going to address
quite briefly is the crack growth rate in the annulus
environment, not the details of what we're doing in
that regard; the risk assessment, where we stand on
that; inspection impacts on susceptibility. We've had
several plants inspected this fall. Does that have
any impact on our rankings? Recommendations that we
may be making for the spring outages and the ASME code
activity that's going on.
In the area of crack growth rate, I'm not
sure if I told you this last time, but I think, I
believe I did. We were convening an expert panel to
try and assess what's the appropriate crack growth
rate to use when evaluating this phenomena. That
panel, which consists of members from all over the
world, convened for the first time in August, kind of
an introductory, get to know what they're going to do
meeting. Then they had a second multi-day meeting in
Airlie in October, the first part of October.
Numerous phone calls and discussions have taken place
amongst those Members since then and they produced the
first draft of a report on Alloy 600 on October 22nd.
As the industry reviewed that, we had
many, many comments on the structure and the content
of the report. Those are being incorporated and not
just us. Dr. Shack had comments on the thing too, so
those are being addressed. The second draft is
expected out next week and we're scheduling the report
on crack growth rate from the expert panel for
publication at the end of this month.
MEMBER FORD: In terms of communication,
ease of communication, has any of this information
from this expert panel been transferred to the NRC
staff?
MR. MATHEWS: I don't believe we've sent
the draft to them yet. Dr. Shack was on the committee
and I believe he's seen the draft of the report, first
draft of the report and so that's as far as it goes.
We haven't sent anything else yet, but we've gone back
and tried to incorporate and I'll get into some of
what we've tried to do.
MEMBER FORD: The reason for my question
is because of the urgency, I just want to make sure
there's plenty of backwards and forwards.
MR. MATHEWS: Yes. I guess the extent of
that is Dr. Shack has been given the draft and
commented on it.
MEMBER FORD: Sure.
MR. MATHEWS: The process they went
through was they established some data screening
criteria. There's lot of data out there on Alloy 600.
Not all of it is of the same quality so the expert
panel established some criteria for selecting the data
sets and then the data sets that were to be used were
selected and initially a best fit curve was put in the
report and that was it. Some of the comments from the
industry were we need to put in the kinds of curves we
need to address, what the data is going to be used
for. So some of the curves that are being
investigated at the moment are heat to heat
variability, is there some effect there that needs to
be accounted for; the widespread in the data, there is
a very wide spread in the data; what amount of
conservatism needs to be put into the curve if you're
using it for deterministic evaluations; is that the
right same curve or should you use a different curve
if you're doing probabilistic evaluations? And also,
the annulus environment.
One of the things -- the things the expert
panel has determined is that the environment that is
possible in the annulus after a leak established is
about the same as a primary water as far as the impact
on crack growth rate. And it doesn't matter that much
where the boiling transition takes place inside the
crack, at the exit. The only times there's potential
any impact, as I understand it, is when the boiling
transition is right at the crack.
MEMBER FORD: I recognize you don't have
very much time today and obviously we'll be having
another meeting with the Materials Subcommittee. We
have a whole day to go through all the ins and outs of
this.
Can you give us some feeling about where
you stand right now in terms of the choice of a
disposition curve for both axial and circumferential
cracks? Is it going to be a mean? Is that the way
you're moving?
MR. MATHEWS: I doubt that it will be a
mean for a disposition curve or for probabilistic,
we're still trying to decide. It may be a mean for a
probabilistic evaluation, but even there you have to
-- we have to look at what's the effect of heat to
heat, how do we really handle those kinds of
variations and I don't think it will be a mean curve
for disposition. There will be some conservatism in
a disposition curve, but then again we're not really
talking about dispositioning a circ flaw above the
weld. Those will typically be repaired and so we're
not going to find one and try and figure out how long
we can run with it. We can't run with it. So what we
do there will be different than if it's a short axial
flaw and we're trying to disposition that then there's
also the dispositioning of the hypothetical flaw and
we've got to figure out how to handle all of that.
MEMBER FORD: I was just looking through
your presentation here. This is the only slide there
is because there's no data and there's a lot of
technical discussion necessary.
When do you think you'll be ready to
discuss this?
MR. MATHEWS: We're intending to try to
publish this by the end of the month and we will have
discussions in our meeting that's scheduled with the
NRC at the end of the month on the 27th.
MEMBER FORD: Thank you.
MR. MATHEWS: The second draft is due out
next week. Hopefully, we will -- that will be the
final one or shortly thereafter we'll be able to
publish it.
One of the things we found out yesterday
or it appears that there may be some differences in
the data sets that have been used by the staff and the
ones that the expert panel used, so those things have
to be addressed very quickly I think.
MEMBER FORD: Okay.
MR. MATHEWS: Anything else?
MEMBER FORD: Well, there's lots and lots
of questions.
MR. MATHEWS: That is the only slides I
have on the crack growth rate and that's a very
important topic because it drives a lot of stuff and
we need to reach resolution on that area and we
recognize that and we're moving toward that as fast as
we can.
In the area of risk assessment --
MEMBER POWERS: Maybe I don't understand
quite. What is it that's so crucially dependent on
knowing the crack growth rate?
MR. MATHEWS: How long you have -- well,
the crack growth rate can have impacts on reinspection
intervals. It can have impacts on how long you can
run, once you find an acceptable flaw, how long does
it remain acceptable and those kinds of things. And
it also has an impact on the probabilistic fracture
mechanics that you fold into your industry risk
assessment. It drives and feeds into a whole bunch of
aspects of this issue.
MEMBER POWERS: Yes, as long as one is
willing to operate the flaws, I see that.
MR. MATHEWS: Yes. Well, even if you're
trying to assess the probability of core damage or the
frequency of core damage on the mere presumption that
there may be a flaw.
MEMBER POWERS: It will usually come out
three times the life of the universe or something like
that.
MR. MATHEWS: We hope so.
(Laughter.)
MEMBER POWERS: Doesn't mean I believe
them, but that's the way they always come out, so it
doesn't matter what number you put in.
MR. MATHEWS: Well, in the area of risk
assessment, the preliminary work was performed for the
various plant types, some preliminary work. We had a
meeting this week to finalize our approach for the
industry and how to bring these various approaches
that were put together.
What we're looking at is using industry
statistics for probability of a thru-wall flaw versus
time to get -- that gives you then the probability of
initiating a thru-wall flaw. Then we're using
probabilistic fracture mechanics to determine the
probability of crack propagation versus time from the
time of the leak to the thru-wall or the rupture.
The conditional core damage probabilities
for results of LOCA and rod ejection will be assessed
from either generic industry numbers or plant specific
numbers and those should include any effects of
collateral damage that may happen. And then finally,
the core damage frequency --
MEMBER POWERS: I don't understand your
language here. You have probability fracture
mechanics for probability crack propagation versus
time from leak to thru-wall or rupture.
I mean if it's leaking isn't it through
the wall?
MR. MATHEWS: There's a couple of ways to
get a leak and if you had a leak through the weld, the
thru-wall we're talking about here is thru-wall in the
direction that could lead ultimately, if it grew far
enough to an ejection.
You can get water on the outside in the
annulus region by a leak through the weld that's an
axial leak that does not threaten the ejection of the
rod. Then what we're talking about is using from the
time you've got that leak to the time that the crack
could turn circumferential and grow through the wall
and/or grow all the way around until it ejects.
MEMBER POWERS: It's really not thru-wall
so much as being circumferential.
MR. MATHEWS: Yes, thru-wall in the plane,
that is a circumferential flaw. You can get a leak
from an axial flaw which has no -- does not threaten
a LOCA or an ejection. Once you get that, how does it
grow back toward the potentially dangerous flaw of the
circumferential flaw all the way around. And the
thru-wall component, they're talking about there, how
does that crack propagate until it's a thru-wall
position. Well, it just goes -- we propagate it
probabilistically all the way until it gets to the
limit.
MEMBER POWERS: I'm trying desperately to
remember language of general design criteria, but it
seems to me once we have a leak, we are in violation
of having a leak type primary piping system.
MR. MATHEWS: Yes. We're not arguing
that.
MEMBER FORD: Larry, at the meeting we had
both to the Materials Subcommittee and then to the
Full Committee earlier this year on this particular
topic, you came under a fair amount of fire, I seem to
remember, for the assumption that you made on
initiating event frequency. Since that time I know
Oconee has given a public meeting here to the staff on
that issue.
Could you, for the members, give a kind of
bottom line conclusion from that meeting?
MR. MATHEWS: I'm not sure -- I wasn't
there, and I'm not sure I know the specifics of what
they presented at that meeting.
MEMBER FORD: Okay.
MR. MATHEWS: They were -- I have seen
numbers of -- in one of the approaches, one and a half
flaws for an assumption, but that's for a given set of
plants at a given point in time. Most of the other
plants were much, much lower on the susceptibility
curve and therefore the probability of having a leak
at this point in time.
And what we're trying to assess is that
probability starting now and into the future for each
and every unit and you can use the data that has been
gathered on the plants that do have leaks and the
plants that don't have leaks at this point in time to
come up with some distribution and I believe there's
a wide distribution that they're trying to use to come
up with a --
MEMBER FORD: But there will be a uniform
industry argument on this particular risk assessment
issue for discussion with the NRC staff?
MR. MATHEWS: Yes, there will. That's our
intent.
MEMBER FORD: And that will be presumably
by the end of this month, November?
MR. MATHEWS: No, no. That was the crack
growth rate. It's going to take a little bit longer
to get all of this pulled together. It's probably
going to be in the January time frame.
MEMBER FORD: Okay.
MR. MATHEWS: Before spring outages for
the most part, we should have some results.
MEMBER FORD: Okay.
MR. MATHEWS: We've seen this curve
before, except that I've changed it a little bit and
I put up there who has which units have leaks, which
units have had cracks, but have not discovered leaks
on their head, which units have done visual
examinations with no leaks and then the ones that are
later. It's hard to see up close, so I got it blown
up for the first 30 years. This was our ranking of
the units and when we put it together it was a quite
simple ranking. We all know time and temperature of
the head and if you look at the plants that have had
leaks to date --
CHAIRMAN APOSTOLAKIS: Could you explain
a little bit. Don't assume we know.
MR. MATHEWS: Okay, I'm sorry. This
bottom axis is the effective full power years that it
would take a unit normalize to 600 degrees Fahrenheit
head temperature to reach the same effective full
power years calculated for Oconee 3 at the time they
shut down and discovered their numerous flaws at
Oconee 3. So we use Oconee 3 as the base point and
then we normalized all the plants to that same, to an
effective full power year number that would be
equivalent to an equivalent head temperature of 600
degrees.
Using that then, we just ranked the plants
as far as how far away they were in time until they
would reach that same equivalent time at temperature.
CHAIRMAN APOSTOLAKIS: So if I go to EFPY
10 and I go up and I find 20 on the left --
MR. MATHEWS: Okay that means --
CHAIRMAN APOSTOLAKIS: What does that
mean?
MR. MATHEWS: What that means is that the
highest ranked 20 units in the country are less than
10 years from being equivalent to Oconee 3.
CHAIRMAN APOSTOLAKIS: The highest ranked?
MR. MATHEWS: Right, or if you go up the
left column, 20, the 20th ranked plant, if you rank
them sequentially, the 20th ranked plant would be 10
years away from Oconee 3.
CHAIRMAN APOSTOLAKIS: They would be
weighed in what sense?
MR. MATHEWS: It would take 10 more years
of operation, effective full power years of operation
at their head temperature to reach the same time at
temperature equivalent as Oconee 3 normalized to the
600 degrees.
CHAIRMAN APOSTOLAKIS: I see.
MR. MATHEWS: So it's just a way that we
could rank plants based on time and temperature, how
far away are they from being equivalent to Oconee 3
and the number up the side is just the rank order of
the units.
CHAIRMAN APOSTOLAKIS: Now what I see
there at that point is a circle, a green circle and up
there it says later.
MR. MATHEWS: And that means that that
unit has not had an outage done and inspection of
their vessel head to date since Oconee 3 and since we
put this February 2001 was our normalization point.
VICE CHAIRMAN BONACA: Now I see a 6, I
see one unit with a triangle there, a red triangle?
MR. MATHEWS: Right, and that unit has
done an inspection of their head. I believe it was
this fall and discovered a leak on their head, at
least one.
VICE CHAIRMAN BONACA: So although it
would take 6 years for it to get to the same
temperature --
MR. MATHEWS: Same time.
VICE CHAIRMAN BONACA: Oconee still has
leaks. Why wouldn't that plant -- I'm trying to
understand that the reference is to Oconee.
MR. MATHEWS: Well, we can change the
reference. If we change the reference all that does
is move Oconee to a negative number.
VICE CHAIRMAN BONACA: Okay, I see. All
right.
MR. MATHEWS: We just selected Oconee 3 at
that point in time because it was the worse cracking
that we had observed to date at the time we put the
initial --
VICE CHAIRMAN BONACA: It's logical, all
right.
MEMBER FORD: At the time this was thought
about, the temperature time algorithm was a reasonable
start.
MR. MATHEWS: Yes.
MEMBER FORD: As you stand right now, are
there any surprises from the algorithm telling you
that there's something physically wrong with it? In
other words, just glancing at it, it seems that you
don't have any red triangles all the way to the right.
Thank goodness.
MR. MATHEWS: That's true. There are no
red triangles to the right. We do have some blue
diamonds to the left, at least one or two and have
done visual exams of their head and not discovered
leaks.
MEMBER FORD: Right.
MR. MATHEWS: We have some other green
circles over there that haven't looked yet. We have
plenty of "haven't looked yets" right in here. So --
and on out. So we don't see anything, I don't see
anything here that really surprises me and says the
model that we're using is just a totally ineffective
way to address the issue. The plants that have
discovered leaks are the plants that are high in the
susceptibility ranking here at time and temperature.
There are plants, we're starting to get
into plants that don't have leaks or don't have
observed leaks that are intermingled here. It's not
a perfect model by any stretch of the imagination, but
we are starting to get to the point where we're seeing
some that don't have leaks.
MEMBER FORD: Is there anything in these
observations that intimate that, for instance,
different -- they're all hot head plants, but is there
anything that would intimate that plants of a certain
design would be more likely to crack than not at a
given EFPY?
MR. MATHEWS: Well, there are many leaking
plants here. Seven plants, I believe, here that have
red triangles.
MEMBER FORD: Right.
MR. MATHEWS: Six of those plants are BNW
design plants.
MEMBER FORD: Yes.
MR. MATHEWS: One of those plants is a
Westinghouse-designed plant. There's a wide variety
of materials that were used to make the head
penetrations throughout the industry, especially on
the Westinghouse units. So that's one of the things
we're starting to look at, is there some segregation
of the population that we could do. One of the plants
that did have a leak, their cracks were entirely
contained in the weld metal and so we have to say
well, do we know enough to talk about the specifics of
the weld metal for all the rest of these plants or is
time and temperature going to ultimately be the thing
we use all along. And we're not sure yet. We're
assessing the data continuously as it comes in.
VICE CHAIRMAN BONACA: Given that you have
this dependency evidently. I mean most of the red are
for one type of plant. Is it reasonable still to put
all these kind of different designs, etcetera on the
same plot?
MEMBER FORD: That's the reason for my
question is can you read out from that that there is
a --I don't know, a heat to heat or a fabrication
style?
MR. MATHEWS: Certainly, we hope to
ultimately to be able to figure out what those
parameters are and characterize them and start to sort
this data into bins. And we've already sorted -- not
this data, but we're starting to look at well, who has
what type of material, what type of tubes and as we
get more inspection data, a lot of the types we don't
have much inspection data on yet, but as we get more
inspection data, we'll be able to say well, this
doesn't look quite as susceptible as it should be, you
know. We should be seeing some indications here and
we don't. So we can maybe sort that out. But at this
point in time we don't think we have enough
information to start to do those discriminations.
MEMBER FORD: But based on the data you
have so far, the historical data you have so far, it
is real to suppose those three open circles run about
3 EFPY if they were inspected, you would expect to see
cracks, based on that data?
MR. MATHEWS: I would say it wouldn't
surprise me to see cracks there.
MEMBER FORD: Okay.
MR. MATHEWS: Because they're not --
there's material variabilities, etcetera there.
MEMBER FORD: Sure.
MR. MATHEWS: And it wouldn't surprise me
to see some cracks, but it's not going to surprise me
to not see some either.
MEMBER FORD: Thank you.
MR. MATHEWS: When we look around the
industry we already have several plants that have
decided that it's the right thing for them to do,
based on their ranking and their decision making
process, etcetera, that they should be doing some
under the head NDE inspections and part of that could
be tied to the difficulty that they would have doing
a visual inspection because of their particular
insulation package, but we have several of these
plants that have already told us that they probably
will be doing inspections this next spring and next
fall and we have a draft industry inspection program
under review where we're trying to say is this enough,
we're looking at the ones that are volunteering and
saying is this enough? If not, what do we need to do
as an industry and we're working on that right now and
that program is under review and when we get through,
if it's a published recommendation we'll be sharing
that.
MEMBER FORD: On that basis and I'm
referring back to the previous one, do you remember we
asked whether it was possible to improve this
prediction algorithm.
MR. MATHEWS: Yes.
MEMBER FORD: Are there -- where do you
stand on that right now or do you not see any reason
to improve it at this time?
MR. MATHEWS: We'd love to be able to
separate the whole rest of the fleet from the ones
with the red triangles, but do we have the data now to
do that? We don't. We need some more data to be able
to sort out and figure out which plants might be able
to be separated and which ones belong in the same
population.
MEMBER FORD: I recognize it's a complex
problem, but from a cost-effectiveness point of view,
is there any point in continuing with that effort to
try and improve on the --
MR. MATHEWS: Improve on the modeling?
MEMBER FORD: Yes.
MR. MATHEWS: Well, it's pretty simple to
do this model.
MEMBER FORD: I'm sure.
MR. MATHEWS: And as we get into it, you
have to look at the parameters you're looking at. If
you've got great lab data on a parameter, but you
don't know what it is in the field, there's no point
in chasing it and so we're -- that's like the welds.
If we knew exactly what properties caused the weld to
crack, that may or may not be of any benefit if you
don't know what the properties are for all the plants.
MEMBER FORD: Sure.
MR. MATHEWS: So that's the kinds of
things we're having to weigh, but right now we're
still going with the time and temperature and trying
to gather data. We know what the subsets are as far
as materials, etcetera, but right now we're still
having them write this way.
MEMBER FORD: Okay.
MR. MATHEWS: As far as the long range,
NRC has indicated that the inspection criteria for
head penetration area in the ASME section 11 code is
potentially inadequate or has been over the years,
although we haven't really had an action because of
that, but they feel that it needs to be modified. So
Section 11 has established a working group to look at
the Alloy 600 inspection requirements that are in the
Section 11 code right now. They met at the last
Section 11 meeting and they'll be meeting again at the
December meeting and I believe there's at least one
draft of something that they'll be discussing at the
December meeting.
I'm not sure how quickly the code moves,
but ultimately, long range down the road, I believe
Section 11 will have some requirements that they will
put into place that will be a long range inspection
program.
We need to work with them because some
plants are replacing heads and they're using much more
resistant material, so we need to figure is that the
right -- make sure that whatever gets into the code
for inspection doesn't overpenalize people who go into
proactive or necessary head replacements.
That's all I have. I have used my half
hour almost exactly.
MEMBER LEITCH: Just one quick question.
The visual with no leaks, the blue dots on your
histogram?
MR. MATHEWS: Yes.
MEMBER LEITCH: Refresh my memory, what
does that mean? You looked for boron and found none?
That was the extent of the examination?
MR. MATHEWS: Yes, that's what it means
but it was at least an effective visual which means
that you got a good hard look at metal to metal
interface where the penetration goes inside the
interference gap on the head. And saw no evidence of
leakage on all of those plants.
Now NRC's requirement was that we
absolutely demonstrate for plants less than 5 years if
you were doing a visual that it be a qualified visual
which means you demonstrate it would leak and for the
others out to 30 years they didn't require that
demonstration at this point.
Some of these plants have done some work
to demonstrate that they believe most, if not every
single one of their penetrations would leak if they
had a crack. Others have not done that to date.
MEMBER LEITCH: Okay, thank you.
VICE CHAIRMAN BONACA: Just before you
take it down, of these plants, how many of them have
exhibited this circumferential cracks?
MR. MATHEWS: I believe three plants had
circ flaws: Oconee 3, Oconee 2 and Crystal River.
VICE CHAIRMAN BONACA: So they are spread.
MR. MATHEWS: Two of them are real close
and then the other one is out.
MEMBER FORD: The uppermost triangle is
Crystal River, uppermost red triangle?
MR. MATHEWS: I believe.
MEMBER ROSEN: You said that the small
cluster is down between 0 and 5?
MR. MATHEWS: These three.
MEMBER ROSEN: If you found the plant in
that group that had a leak, that it wouldn't surprise
you, am I right?
MR. MATHEWS: Right. It would be
consistent with the data from the other plants.
MEMBER ROSEN: Now what about the next
open circle that you see up around 6 or the next one
after that, 7th. If they had a leak, would that
surprise you?
MR. MATHEWS: It becomes more and more of
a surprise the further to the right you go.
MEMBER ROSEN: That's what I'm trying to
figure out. What is your crack definition?
MR. MATHEWS: I don't have one. I don't
know. The data is out there. We're trying to gather
the data to figure out where it is.
We didn't try to use this as a predictive
model to say this plant will leak in X number of
years. What we were trying to do is say these are the
plants that -- on a time and temperature basis any
way, the highest in rank and they're the ones we
really need to be concerned.
MEMBER ROSEN: I know from listening to
you that the ones below 5 would surprise you and I
think if you saw a crack between 25 and 30 that would
surprise you. The ones between 25 and 30 would
surprise you. The ones between 0 and 5 would not. So
I have two points on the curve.
(Laughter.)
CHAIRMAN APOSTOLAKIS: Are you plotting
being surprised?
MEMBER ROSEN: I'm trying to figure out
when we should say oops, this isn't working.
MR. MATHEWS: I think a leak in the 30-
year time frame certainly would warrant, whoa, what's
going on, a whole bunch more, no leaks down in the 5
to 10 year range doesn't.
MEMBER ROSEN: Yeah.
MEMBER FORD: You are looking at how a
scatter band of depths is impacting over time and so
you expect if all the plants were exactly the same
design, manufacture, there should be a clear cut, but
they're not. Is there any reason -- the reason I'm
asking the question, between cracking and no cracking
on that curve. That's why I asked the question can
you put a normal cut between, for instance,
Westinghouse and Babcock on that?
MR. MATHEWS: Part of the unfortunate
thing about trying to do something like that is all of
the BNW units happen to be very high in the ranking on
time and temperature.
MEMBER FORD: Because of time primarily.
These are 600.
MR. MATHEWS: Just from a time and
temperature basis, they're very high in the ranking.
MEMBER FORD: Right.
MR. MATHEWS: They were all built fairly
close together in time. They all run with fairly high
head temperatures relative to the rest of the industry
and so when you couple those, it pushes most of the
BNW units --
MEMBER FORD: It is primarily time, isn't
it, Larry, because most of these are 600 plus or minus
2 head temperatures. Is that correct?
MR. MATHEWS: Well, some of these units
are probably even on down into the 590, 580 range.
MEMBER FORD: Okay.
MR. MATHEWS: Well, maybe not these.
These may not be down in the 580, but some of the
plants on the higher end up here are in the final --
MEMBER ROSEN: Some are even lower than
that.
MR. MATHEWS: Yes. If you look at the
curve for the entire fleet it goes all the way out.
Most of these plants are designed such that their head
temperatures are calculated to be very near the cold
lake temperature.
MEMBER ROSEN: Right.
MR. MATHEWS: And these are either hot
head or warm head plants.
MEMBER ROSEN: Right.
MR. MATHEWS: The Westinghouse design
diverts a little of the cold lake flow up there and it
varies from design to design. But most of these
plants out in here have what we call cold heads.
Their heads have enough cold lake flow diverted to the
head and they're operating at significant lower head
temperatures than most of the other fleet.
MEMBER ROSEN: Okay.
MR. MATHEWS: Now the BNW designs tend to
be even on the high end of the head temperatures, over
600 and most of the Westinghouse plants are a few
degrees under 600, NCE plants.
MEMBER ROSEN: Okay. Have you got all the
inspection results in for the fall outages that's
shown on this curve?
MR. MATHEWS: I believe North Anna 2.
Well, we've got all of them that are complete. And
we've got all of them that are -- I believe all of
them that have discovered any leaks, even if they're
not completely through with their exam is on there.
There's a few units that are down now that have not
finished their exam and I don't know if they've found
anything or not. So --
MEMBER ROSEN: I'm going to take your
answer to mean there may be some new data yet from
this set of fall outages.
MR. MATHEWS: Absolutely.
MR. BANFORD: There's one more green
diamond in the less than 5 year.
MR. MATHEWS: Green circle.
MR. BANFORD: That we just became aware of
about two days ago. Warren Banford from Westinghouse.
So one of the three open circles in the 0 to 5 is now
a green circle.
MR. MATHEWS: A blue diamond.
MR. BANFORD: Or a blue diamond.
MR. MATHEWS: Visual with no leaks. I
guess that that's within the last two or three days
that that came out.
Well, it won't change the color. Oconee
is down and I don't know how I'm going to plot this
when -- if they come down and don't find anything this
time, how do I plot that. Successful repair.
MEMBER FORD: I'd like to bring this one
to a close unless there's any urgent questions. But
is it my understanding, Larry, that by December if we
had a Materials Subcommittee meeting in December, you
would have a full technical data presentation on the
-- at least on the cracked assessment?
MR. MATHEWS: Crack growth rate?
MEMBER FORD: Yes.
MR. MATHEWS: I think we can --
MEMBER FORD: And the annulus would not be
a waste of time, I believe to have a meeting at that
time if that's what we decide.
Larry, thank you very much. I'd like to
call upon now Steven Moffat of Davis-Besse. I think
it's 10 minutes.
MR. MOFFAT: Thank you. I have a power
point type presentation that starts with Davis-Besse.
Again, I'm Steve Moffat from Davis-Besse. I'm the
Director of Technical Services.
With me is David Geisen and Ken Byrd who
are on our staff. We also have several of our subject
matter experts who contributed from an industry
perspective to our discussion this morning with us in
the audience. First off, I'd like to sincerely thank
the ACRS for allowing us this opportunity to give what
could be characterized as a case study of some of
these diamonds we just saw as relates to Davis-Besse.
Our objective today is to have ACRS
consider additional industry information as it
deliberates on the Control Rod Drive Nozzle cracking
issue. Obviously, this is an evolving issue. There's
information we recently got from the staff on
positions relative to this issue and we'd like to
include the additional information today in formal
correspondence to the NRC that may be related to
regulatory action.
As we heard Larry talk about consensus
documentation at this point in time, what we have done
is used the best industry data we could in order to
determine what would be appropriate action for us
based on the safe operation of our power plant. Our
reason for request is that Davis-Besse is considered
a high susceptibility plant as discussed in Bulletin
2001-01, specifically the time and temperature model
puts us within three years of -- actually, slightly
over 3 years of the adjusted Oconee 3 data that we saw
a minute ago.
MEMBER FORD: So you're one of those open
circles?
MR. MOFFAT: We are one of the "to be
inspected" --
MEMBER FORD: To be inspected open
circles.
MR. MOFFAT: Yes sir, that's right. And
as indicated here, we are the only high susceptibility
plant that's ranked by the Bulletin that will not
perform a visual inspection by December 31 of 2001 and
we've used these subject matter experts, as I
referred, to assess the impact of the continued safe
operation of our plant until our next scheduled
refueling outage which is currently scheduled for the
end of March of 2002.
As far as Dave is essentially a
deterministic engineer and Ken is a probabilistic,
some of the discussions will go in that area in my
brief presentation. As far as the deterministic, we
believe we have what could be characterized as
qualified visual inspection that was performed in 1996
and additional inspections in 1998 and 2000. My
reason for using that terminology is that we had done
inspections of the head for other reasons, however,
the people that did the inspections are available and
we, in fact, videotaped the heads which we have
subsequently reviewed in light of the information that
evolved as part of Bulletin 2001-01. So this provided
us an opportunity to put a new set of eyeballs, if you
will, on the historical issue at hand.
What we did with the 1998 and 2000 is
essentially utilize that data, albeit limited in our
probabilistic assessment of the risk associated with
this issue.
As Larry indicated earlier, the majority
of our nozzles through out own plant-specific finite
element analyses show that they would open which
essentially means the visual efficacy of the
inspection was adequate.
MEMBER POWERS: I guess, maybe I don't
understand that quite. I guess I have a couple of
questions about that.
MR. MOFFAT: Certainly.
MEMBER POWERS: If I go in and I look at
these regions and I don't see anything, what is the
likelihood that there is, indeed, a crack there I just
didn't see it.
MR. MOFFAT: The likelihood due to the
fact that we did specific finite element analysis is
extremely low because the analysis shows that should
there be a crack and pressure --
MEMBER POWERS: It has nothing to do with
my vision and your finite element analysis. If I go
in and look at your head and I don't see anything,
what's the probability that I just missed the crack?
MR. MOFFAT: That human error probability
Ken will discuss momentarily, Mr. Powers. We did
include that in our probabilistic safety analysis for
essentially the ability of a human to detect that.
What this is is more of a mechanistic assessment that
says should there be a flaw, would we be able to
detect that with on head leakage. That's what I'm
referring to by the 56 and 69.
MEMBER POWERS: Well, your finite element
analysis shows what?
MR. MOFFAT: We looked at the 69
penetrations and of those, 65 of them would open up
enough so that we would have a sense as characterized
as qualified that we could detect a flaw through
visible leakage boron deposit.
MEMBER POWERS: All right. Thank you.
MR. MOFFAT: Additionally, we perform our
own plant specific analysis that essentially would
look at the end point as it relates to safety factor
of 3 using the code we've discussed previously and
have a number that's slightly bigger than the industry
data to date as it applies to Davis-Besse. That
number is 302 degree flaw as safety factor of 3.
We also believe --
MEMBER FORD: Excuse me, physically, what
allows you to make that difference in material
properties? I mean what's the --
MR. MOFFAT: I'll let you take that, Dave.
MR. GEISEN: Well, we had the -- the
analysis was done by Structural Integrity Associates
and when they did their analysis they were looking at
the -- using the fracture mechanics with the nozzles
and looking at our worse case nozzle in our particular
strengths of materials what does that take us out to.
MEMBER FORD: So it's a larger qualifying,
larger than you mentioned than who? Industry average?
MR. MOFFAT: Nominally 270 plus or minus
several over here. Quite a body of information
yesterday, but that's the general characterization
between roughly 260 and 272.
MEMBER FORD: Okay.
MR. MOFFAT: Additionally, in order to
look back, as I said previously, with a new set of
eyes, assumed an initial flaw size that would have
occurred immediately subsequent to our 1996
retrospective, if you will, and believe that we would
still have a reasonable assessment that we can proceed
until our refueling outage based on that initial flaw
size.
And then finally the last deterministic
aspect I'd like to go over briefly is just the fact as
Larry was showing, we are over the three adjusted
effective full power years plus at Oconee Unit 3.
Essentially, we're at roughly 6 EFPY as far as the
EFPY, but as adjusted for time and temperature, we're
a bit over 3.
VICE CHAIRMAN BONACA: A question on the
1998 and 2000 inspections, you said that they were
limited.
MR. MOFFAT: Yes sir.
VICE CHAIRMAN BONACA: What was the extent
of the inspection?
MR. GEISEN: I'll talk to that. What we
did is recognize -- this is Dave Geisen. With regard
to these inspections, recognize that they were not
done looking for this particular phenomenon. They
were looking for other things. The two inspections
done in 1998 and 2000 were really looking for the
impact of boric acid leakage from leaky flanges that
we had subsequently repaired and what was the impact
to that. So the view that we got from those was in
many cases some of the drives you couldn't even get a
good view of.
There were many cases, the camera angle
was looking upwards because it was looking at the
structural material of the service structure on top of
the head.
When we looked at a 1996 data, you got
more of a downward look at these nozzles because we
were specifically following around a vacuum and probe
that was looking for head wastage as result of the
boron being deposited on head. So what really comes
down to it, the best video we have on this goes all
the way back to 1996.
MR. MOFFAT: So essentially those are some
of the characteristics specific to our power plant as
plant specific information, as we discussed, given
that the body of knowledge related to this issue, what
we've done is using Framatone and our own engineering
staff, constructed a probabilistic safety assessment
to help us gain an understanding of the significance
of the issue as it pertains to our specific plant and
for that, I would ask Ken Byrd to present.
MR. BYRD: Davis-Besse has performed a
plant specific risk assessment and we base this on the
method developed by Framatone, however, we have worked
with Framatone to make some improvements and we've
also investigated some of the sensitivities to a
greater extent with this method.
We acknowledge that there's a number of
uncertainties with regard to some of the phenomena
we're dealing with and in order to resolve those kind
of uncertainties we had to make a number of bounding
assumptions. Consequently, the results we see here
aren't necessarily consistent with our baseline core
damage frequency. These are actually numbers that
would have to be looked at as kind of a bounding case
for a core damage frequency.
I'll talk a little bit about how we did
that. Again, as more information is developed, we can
hone this more accurately.
Our results did come out. Our core damage
frequency, which again is a conservative number, came
out in the category which would be considered small.
Our large early release frequency came out in the
category that we considered very small and we did run
this through our level 3 PSA and we came out with a
negligible public health risk.
I might point out right now because
results seem somewhat inconsistent with each other and
we have a core damage frequency in one category and a
large early release frequency, a public health risk
that's much lower. That's consistent with our plant-
specific, what we would have expected. We're a large,
dry containment. We have a relatively large
containment relative to our core power. The
consequent that we placed on this was a medium LOCA
and we would not expect a medium LOCA to have a
significant impact on our containment.
For a large dry containment usually that's
containment bypass or interfacing system LOCA or it
could be kind of an event which affects a support
system such as a loss of cooling or a station blackout
and obviously this doesn't deal with any of those kind
of circumstances. It is not a significant release
issue and certainly not a significant public health
issue.
Going on to my second page, what I did
here was just listed a number of what I would consider
the most significant bounding or conservative inputs
that we use in order to deal with some of the
uncertainty that's currently not been resolved yet.
Starting out with a nozzle leak frequency, we did base
our nozzle leak frequency on information from the
other BNW plants, but we applied this very
conservatively so that our model estimates that right
now we should have nine leaking nozzles and I think
that bounds the rest of the BNW fleet. Or actually
that was the number that was found at Oconee 3 which
is the maximum. Obviously, that would be a lot of
nozzles and we wouldn't expect that.
The second thing we did was the
probability of the outside dent diameter crack
initiation. We assumed that for every axial crack
which gives leakage or every leaking CRDM nozzle, we
would have an outside diameter crack initiated.
Actually, if you look at the data from the BNW plants,
we've had 21 leaking nozzles. Five of those have gone
to circumferential cracking, so obviously not all of
them have initiated circumferential cracking. We're
assuming in every case, we will get circumferential
cracking.
MEMBER FORD: Can I please interrupt for
one second? This hasn't been discussed with the staff
yet, I understand.
MR. BYRD: We've given them our analysis
and presented and will discuss this afternoon.
MEMBER FORD: That's just great.
CHAIRMAN APOSTOLAKIS: We have this
analysis.
MEMBER FORD: Could I suggest, this is
really good input to the overall general thing which
we're discussing today. Maybe if you could just move
to the bottom line. You've obviously done a lot of
work. It's just fantastic. And it's relevant, but it
should be discussed with the staff. Could I ask you
to move to the end -- to your end conclusion maybe?
MR. BYRD: Certainly.
MEMBER FORD: Based on -- we read these
inputs.
MR. MOFFAT: I appreciate that. I'll take
it from here. As Ken said, we've submitted this PSA
to the staff and we are going to discuss it. It is on
the docket. Looking at this as a topic, realizing
that we are one of the ones within the three and where
we sit from our understanding it becomes an issue
essentially is our power plant safe to operate until
March, as far as the basis between December 31st and
March, as far as our best understanding of this issue
and the expertise that we've brought to bear as a
utility and a nuclear steam supply system, Davis-Besse
nuclear power station believes that the plant is, in
fact, safe to operate until March. We don't believe
that the distinctions between December 31st and March
are an issue with regard to reasonable basis of
safety. So that's our bottom line that you asked me
to get to.
MEMBER FORD: I thank you very much
indeed, sir, Mr. Moffat.
Are there any overall questions from the
Members at this time, bearing in mind that it hasn't
been discussed in detail with the staff.
Any questions?
CHAIRMAN APOSTOLAKIS: I wonder what is
the actual evidence of the plant regarding this issue?
I remember there were circuits there, but have you
seen -- maybe I missed it, leaks or --
MR. MOFFAT: What we were able to do, Dr.
Apostolakis, by going back through the video tape was
to do our best retrospective examination, knowing in
light of what is known now of the characteristic which
we not only call popcorn due to the nature of this
leak. You can tell, based on this body of evidence
that we've acquired as an industry what it looks like.
So we essentially went back as best we could at those
video tapes from 1996, 1998 and 2000 and that is the
evidence that we have for the current condition of our
plant. We also use the Peter Scott model to then
extrapolate what would bound that should something
progress in that interim since we've had those visuals
and that's some of the detail that we'll be discussing
with the staff when Mr. Byrd has the opportunity
today.
MEMBER ROSEN: I'm going back to your
second slide, your objectives.
MR. MOFFAT: Yes sir.
MEMBER ROSEN: One was to have us consider
the information that you presented here today and I
think that clearly is being done.
Your second one was to include, have us
include this additional information in formal
correspondence to the Commission.
MEMBER FORD: Well, we won't be doing that
until the staff has come to us with their --
MEMBER ROSEN: Yes, I don't think we're in
a position to do that.
MEMBER FORD: That is correct. Not at
this time.
CHAIRMAN APOSTOLAKIS: It can still be an
objective.
(Laughter.)
What we do with it is a --
MEMBER POWERS: As I understand it,
they're fighting over four months.
MEMBER FORD: Mr. Moffat, I thank you very
much indeed. I appreciate it.
MR. MOFFAT: Thank you.
MEMBER FORD: I'd like at this time to ask
the representative from NEI, Alex Merrion to make a
statement.
MR. MERRION: I'm Alex Merrion. I'm the
Director of Engineering at the Nuclear Energy
Institute and I just have a couple brief comments. I
want to say something about the susceptibility model
and my concern with -- and this is just an observation
that people may be reading too much into it. As Larry
clearly indicated, it's a simple model primarily
focused on two parameters, time and temperature, to
try to represent an estimate of who may be susceptible
to finding some kind of flaw should they do an
inspection.
It's very simple from the standpoint of
the kind of degradation mechanism we're talking about
which is very complicated. I think as we go through
this, we need to keep that in mind because I find --
I've seen where people in the industry have been
looking at that susceptibility model as if it was
founded in ground truth and try to make all kinds of
conclusions from it. I've also seen people outside
the industry reading to much into it, so I felt
obligated to make a comment about that.
There was a public meeting held yesterday
where the NRC staff provided their preliminary
technical assessment on crack growth rate and I have
to admit I thought it was an excellent meeting and
it's not painful for me to make that admission, but it
was very useful and I think beneficial from the
standpoint of industry understanding what the staff's
current thinking is and I think the NRC benefitted
from some of the comments and interactions from that
meeting.
And we will be providing comments to the
NRC at a public meeting that's been scheduled for the
27th of November on that. We'll probably follow up
with some formal communication with the staff and on
that meeting on the 27th we'll provide a further
update of the industry efforts of the material
reliability project and deal with some of these
questions that you raised regarding the chemistry and
crack growth rate.
MEMBER FORD: Excellent.
MR. MERRION: And that's all I wanted to
say and I thank you for the opportunity.
MEMBER FORD: Any comments? Thank you
very much indeed.
MEMBER KRESS: I have a question. You
mentioned that the susceptibility model is very simple
for a very complex phenomena. Am I to interpret that
to mean it could go either way?
MR. MERRION: Yes. Yes, it can and we're
in the process of collecting data. That's why Larry
was so cautious in making any kind of firm
representation of what we know today because it could
very well change with the next couple of inspections
and it could go either way.
MEMBER FORD: And I think it's also my
understanding, however, that this idea of having more
than just temperature and time, at temperature, in the
algorithm to include weld conditions, heat conditions
is very hard to do and many people have tried this
over the last 20 years and have failed. And this is
the reason behind my question to Mr. Mathews, is it
cost effective to try and improve on that? Are you
going to materially improve your prediction
capability? That was the reason. You're absolutely
correct, it's a question of whether we move forward.
I'd like to move on at this time to cover
the NRC staff presentation.
MR. STROSNIDER: I didn't have a lot to
say as an introduction and given the time constraints
we have, I just suggest that the staff start on the
presentation.
I guess I could comment that this is a
work in progress. You've heard a little bit already
with regard to need for some additional information,
but we've been working with the data that are
available and trying to pull that into some
comprehensive models.
I'm Jack Strosnider from the staff and
hopefully you'll get some flavor for that based on
this presentation.
MR. HACKETT: Good morning. I'm Ed
Hackett from the Office of Research and joining me at
the table are Allen Hiser and Mark Reinhart from NRR.
We're going to try, as a team here, to
give you a perspective for what the staff has been
doing. I guess especially since July as Dr. Ford
mentioned. So I was just going to start by
summarizing a few items. One was an independent group
of experts, two of whom are here. Your colleague, Dr.
Shack, Dr. Gary Wilkauski here who assisted us in
performing an independent evaluation of the cracking
occurrences, Oconee and ANO. We have also in the
Office of Research continued to support NRR
specifically in development of the technical
assessment that was just mentioned that's in a state
of development that's basically deterministic at this
time point. What I'm going to focus on briefly in
this presentation is aspects of the probabilistic
assessment in terms of probabilistic fracture
mechanics, but the disciplines that are noted there
are the ones that we're providing support on.
If I could just mention briefly that the
Office of Research has also provided inspection
support to NRR, I believe for two inspections at Three
Mile Island and also North Anna. The bottom line of
this presentation is that we're working towards and we
did discuss this in July. We have made some progress.
We're working on developing the elements of a
probabilistic fracture mechanics assessment, as is the
industry. Structural Integral Associates has had the
lead for the industry in that regard as far as I'm
aware. We had a conference call with Structural
Integrity Associates led by Dr. Pete Riccadella on the
27th of September. I think that was very productive
for both us and for the industry in terms of trying to
get a baseline as to what's going on. And Alex
Merrion mentioned yesterday the public meeting we just
completed which was also very beneficial.
Moving on, what we wanted to do is focus
on some of the key considerations. I think Larry
Mathews and the representatives from Davis-Besse have
covered a lot of this information. In our
deterministic assessment which the NRR deterministic
assessment which was released yesterday, I believe,
these are basically sort of the elements that went
into what is Section 6 of that report, but obviously
key to these types of assessments are the assumptions
you're going to making on crack initiation and crack
growth rates in specific. In that particular area,
Dr. Shack has had the lead for us in the contract work
he's done. There's also a lot of variability in the
analysis of the stress state for this particular
situation and the maintenance of the structural
margins. That's an area where Dr. Wilkauski and Dr.
Richard Bass from Oak Ridge have helped us out
significantly. And then sort of the bottom line of
this whole thing, that a number of the speakers have
already mentioned is where does this go to in terms of
inspection methods and maybe specifically timing is
sort of the crucial element at this point for several
plants.
CHAIRMAN APOSTOLAKIS: So what is the
ultimate product of this analysis? What is it going
to be used for?
MR. HACKETT: The ultimate product, at
least in my mind, is a probabilistic assessment that
would hopefully be able to give us guidance on exactly
this bottom line, inspection methods and timing. The
actual, I think the representatives from Davis-Besse
were heading there in their presentation and sort of
where does this end up in risk space ultimately. I
don't think -- we're not ready to discuss that in
detail today. I don't think we're there yet and
that's what I'm going to come to in the presentation,
but I think that's where it's going.
CHAIRMAN APOSTOLAKIS: So if I wanted to
keep the probability of risk initiating event or
frequency low, then this analysis will tell me how
often I will have to inspect?
MR. HACKETT: Yes.
CHAIRMAN APOSTOLAKIS: And what to do?
MR. HACKETT: Exactly.
CHAIRMAN APOSTOLAKIS: And it will include
things about the ability of the inspection method to
get an accurate picture --
MR. HACKETT: That would be factored in,
would have to be.
MEMBER KRESS: And what would the risk
analysis tell you if you calculated just the
conditional core damage frequency and the conditional
large early release, given that you had the failure of
these particular penetrations?
MR. HACKETT: That's a good question and
I might turn to my colleague, Mark Reinhart on that
one.
MR. REINHART: We've presented that just
looking at the IPEs and updates of PSAs to date, the
conditional core damage probability for the medium
LOCA that we're projecting would result from this type
of control drive mechanism ejection is in the 10-2,
10-3 range. There's not a lot of debate on that. The
issue that Ed's really addressing is the initiating
event frequency that we're having the problem with,
but if you just assume one on that initiating event
frequency, you end up with a CCDP.
CHAIRMAN APOSTOLAKIS: So you will be able
to say something about the frequency of a medium LOCA
in this mechanism? That's the whole purpose?
MR. REINHART: That's where we're trying
to go.
MEMBER KRESS: But from an overall risk
perspective, do your conditional probabilities tell
you that it's not that big of a deal?
MR. REINHART: I think 10-2, 10-3 is -- you
can't just say that's not a big deal, if you have the
event. I think we're not comfortable just going along
allowing this. I think we want to --
CHAIRMAN APOSTOLAKIS: Allowing this for
how long?
MR. REINHART: Right.
CHAIRMAN APOSTOLAKIS: I mean allowing
forever is --
MR. REINHART: We don't want it to go
undetected for --
MEMBER KRESS: I was trying to decide
whether there was a real urgency to the inspection and
timing.
MEMBER FORD: I was about to address that
specific point. This is really a program management
chart. And if you look at Larry's temperature time
plot, you can see that this is not a one off problem.
It's a generic problem and you're going to continue
doing it. So what is your timing on this to deliver
this capability, analytical capability to the staff?
MR. HACKETT: That's a good question. I
guess that one is mine.
CHAIRMAN APOSTOLAKIS: The only ones you
get here --
MR. HACKETT: I'm looking at Gary
Wilkauski at this one because we -- the short answer
is we haven't exactly worked that out. I think the
sort of on-line answer to it would be I think pretty
much we were hoping for around the same kind of time
frame that Larry indicated, somewhere maybe after the
turn of the year, but in all honesty, recognizing the
complexity of this problem and we talked about
surprises or not being surprised. It wouldn't
surprise me if we're still looking at some of this
well into 2002, some of the aspects of it. One of the
things I'll come to and Dr. Ford as well, I think to
really make some progress on this area Jack
Strosnider, when we met last time talked about the
dangers of the seductiveness of using a PFM analysis
in this case and I think they're readily evident here
that we're missing a lot of data is the bottom line
and being able to generate all that data in the next
six months or year, I don't want to be up here telling
you we're going to do that.
MEMBER KRESS: That's one reason I asked
the question. One way to prioritize would be the
conditional probabilities, the ones that have a high
condition probability you do first and you sort of
bypass all this difficulty.
CHAIRMAN APOSTOLAKIS: Well, there is
another issue that, I don't know, maybe it's a
non-issue, but it just occurred to me, why should all
these decisions be made on the basis of delta CDF and
delta LERF?
I mean the reactor oversight process tells
me that the staff worries about the frequency of
initiating events itself.
MEMBER KRESS: Yes, but only because
that's going to affect delta CDF and delta LERF.
CHAIRMAN APOSTOLAKIS: They have declared
that they don't want to see higher frequencies
independently of what happens.
MEMBER KRESS: That's a defense-in-depth.
CHAIRMAN APOSTOLAKIS: Yes,
defense-in-depth. So why don't we apply that here and
say okay, the 10-2 or 10-3 conditional core damage
probability is a useful input, but I really don't want
the frequency of the medium LOCA to be too large. So
on the basis of that, I will also make a decision. Is
that -- are we just using regulatory guide 1174 here
or are there more issues?
MR. STROSNIDER: This is Jack Strosnider
and I'd like to make two comments with regard to this
discussion.
First of all, the question on schedule.
And I think with the problem this complex, it's clear
that we can continue to do research and collect data
and try to refine our analysis for a long time.
The practicality and reality of the
situation is that we need to set some priorities and
we're going to be looking at that such that we can
have some analyses early next year that will help
inform what the appropriate inspection intervals and
timing is. And I put that in the context of
recognizing that the bulletin that's being discussed
is just a one time action, right? And the industry
and the NRC still have to develop and come to
agreement on what the longer term program is for
managing this issue. And that has to happen next --
in 2002. All right? Because people will have to be
making decisions in 2002, so we have to position
ourselves in early next year to have the most
important information and the best information we can
to inform that decision. And there may be additional
work that continues to refine this and to better
refine whatever that decision is, just as this fall,
we're making information based on the best information
we have available today.
And that gets to the second point which is
from a regulatory decision making process, how do we
deal with this and I would just point out that is it
just Reg. Guide 1174 and we have a tendency when we
think about 1174 to sort of focus on the quantitative
aspects of that framework, but in fact, that's only
one piece in there and if you look at the total five
elements that need to be considered you need to
consider traditional margins, defense-in-depth,
monitoring and those other areas and we clearly will
be taking those things into consideration with any
regulatory decisions we make.
But again, getting back to this
presentation, we need to get this technical work, we
would like to put this into a 11174 quantitative
framework as well early next year so that we can help
inform the decision, so that's our goal.
MR. REINHART: And maybe Ed, we have a
suite of regulatory guidance that is guiding us,
obviously, but we're certainly open to be thinking out
of the box and as we go along we're looking for
different ways to look at the data, look at the
information, what's important, what's not important,
so we're not tunnel vision, but we do want to work
within not just 1174, there's a whole suite of guides
that we have, and use that information as a place to
start from the probabilistic aspects.
MR. HACKETT: I think as a kind of
illustration of what Jack and Mark were just talking
about, you look at -- kind of two ways you could go
about looking at crack initiation. I've got them on
the slide. Two sort of processes that have been
discussed or postulated for how this would progress in
a mechanistic sense are really what those first two
bullets are up here. That would be the way you would
like to go about doing this in a more purely
scientific mechanistic way. I think, in fact, we're
probably going to be severely limited in that regard
by lack of data and other difficulties, so when you
come to the bottom bullet there, what Dr. Shack has
done is consider laboratory data, supplemented or form
of laboratory data for some of what we're looking at
in initiation space, sort of conditioned by
consideration of time to initiation from inspection
data which is more of an empirical approach for going
about this kind of thing. But that's a good example.
I think the first two bullets there are something we
could spend many master's theses and Ph.D. theses on
exactly how these should be progressing. And as Jack
pointed out, that's not what we're about.
CHAIRMAN APOSTOLAKIS: By the way, Dr.
Shack's name has been mentioned many times this
morning. This is Dr. Shack of Argonne National
Laboratory, right?
MR. HACKETT: That's correct.
CHAIRMAN APOSTOLAKIS: Okay.
MEMBER FORD: Ed, on just a definition.
Your definition of time initiation is the time it
takes to get an environment in that annulus, is that
correct?
MR. HACKETT: I might turn to Bill on
this. My own definition -- I don't know that we've
really gotten into discussion of that would be the
development, that plus the development of I guess what
I would call an engineering crack that would be
hopefully discernible through the --
MEMBER FORD: For circumferential crack.
MR. HACKETT: Right.
MEMBER FORD: Okay.
MR. HISER: I think also what you would
need to look at, is going back to the very beginning,
what is the crack that gives you the leak? Is it
through the J groove weld? Is it an axial crack that
goes through the wall?
MEMBER FORD: That's why I'm asking the
question.
MR. HISER: There are several.
MEMBER FORD: What do you mean by --
MR. HISER: I think there actually several
times that really are important. I'd include all of
those.
MEMBER FORD: Okay.
MR. HACKETT: The other one in this area
that's obviously already been discussed this morning,
that's been contentious to say the least is the
annulus environment. I think to sort of cut to the
chase, I think we're in -- largely in agreement with
the industry that this is hopefully a primary water
environment for the reasons I state here. We don't
know that for sure is the bottom line. But I think
this is an area that's a tight crevice. It's unlikely
to be oxygenated. Depending on where the flash point
is when the fluid is exiting the crack that exists, if
that's far enough away you're maybe not likely to
concentrate as aggressive a species as we might
otherwise and then the other thing is once these
cracks go through wall, they now have a communication
with the primary water environment that's hopefully
causing some flushing. This is key to the next slide
because we are going on with the assumption that this
would largely be a primary water environment. We
don't know that.
I'll just move real quick through --
MEMBER FORD: Hold on. Before you do
that, because this came up for a lot of discussion in
the meetings we had in July where it was arbitrarily
decided that there was a primary environment without
any justification.
From what I understand, you can have pH as
anything from 4.5 to 8.5 in that annulus.
MR. HACKETT: That's correct.
MEMBER FORD: And practically speaking,
there's really no way that you can look at a specific
CRDM housing and say the pH in that annulus is that.
There's no way you can do it.
You have data, I assume, that shows that
it doesn't really matter. Is that correct? On
cracking susceptibility, you can go from 4 to 8.5, do
you have data that shows, hey, this uncertainly
doesn't really matter.
MR. HACKETT: Without a large impact for
Canal 600, I'm caveating this, in a primary water
environment, that's correct.
MEMBER FORD: Okay, fine.
MR. HACKETT: But that's why, actually,
starting with the next slide I said the first
statement that I did there. The bottom line is we
don't know exactly what this environment is. To cut
to one of these, there is -- I don't know where I
guess I had on the previous slide, the -- one of the
considerations would obviously be to go find out. We
discussed that with the industry that came up at the
meeting yesterday to sample one of these environments.
MEMBER WALLIS: Can't you make some
calculations what happens in that crack in the
annulus?
MR. HACKETT: That has been done indeed.
Dr. Shack has done that.
MEMBER WALLIS: I mean in terms of the
fluid and chemical environment. Has he done that too?
MR. HACKETT: He's saying no.
(Laughter.)
So maybe not.
MEMBER POWERS: Why don't you get yourself
a new contractor.
MR. STROSNIDER: This is Jack Strosnider.
I think there's some industry calculations that were
done that were referenced in some of the work that Dr.
Shack provided to us and is documented in the
assessment report, but they were industry
calculations.
MR. HACKETT: This comes down to then the
next step is for crack growth. We're making the jump
from the first one to the second one as Dr. Ford
mentioned that the big jump there is that this is a
primary water environment largely.
MEMBER WALLIS: Intuition would be that it
can't possibly be, boiling off the water, leaving
behind whatever else is there.
MR. HISER: I think it's described in the
preliminary technical assessment, it may not be
primary water initially, but as you get cracked
opening and get additional leakage, there's enough
communication with the bulk primary water. I think a
lot of the --
MEMBER WALLIS: The water is continually
disappearing by flashing.
MR. HISER: Initially, yes.
MEMBER WALLIS: It always is. You're not
squirting out water into the environment. You've got
a very dry environment on top of this crack.
MR. HACKETT: I think what Allen was
coming to, Dr. Wallis is that once the communication
is there through wall --
MEMBER WALLIS: I'd like to see -- it
would be much more reassuring if someone would draw a
picture to show that something else hasn't been done
to justify these --
MEMBER FORD: I think what's becoming very
obvious, Graham, is that we are not going to have the
answers to all of these questions.
MEMBER WALLIS: It shouldn't be all that
difficult to do some simple fluid dynamic diffusion,
whatever is necessary, boiling analysis and get maybe
some boundary estimates of what's there.
MR. STROSNIDER: This is Jack Strosnider.
I just would comment that the technical assessment
report that we just provided a few days ago actually
has several page discussion referencing those sort of
analyses, so there's a lot more detail in that report.
MEMBER FORD: I think from the point of
view of this meeting, all we're trying to get is a
feeling that we are moving forward on this whole
issue, all the parts of the issue.
MEMBER WALLIS: I'm not feeling we're
moving forward if we don't have some sort of analyses.
MEMBER FORD: Well, put it this way. I'm
at the advantage, I've seen some of this data
unofficially, and yeah, they are moving forward.
If we have another meeting, then we will
be discussing all of these ins and outs.
MEMBER POWERS: I think we ought not
underestimate the complexity of doing a simple quote
diffusion analysis in a crack environment and it's
because not only do you have multiple species reaching
saturation, but you have an incredibly high
electromagnetic field in the vicinity of this crack
surfaces, that I'm not sure you would really have the
ability to calculate things like diffusion
coefficients and things like that in that relatively
narrow crack.
You could put a diffusion coefficient in
there for the ionic species, but I think you'd just be
dead wrong.
MEMBER FORD: Dana, you could well be
right. I'm trying to move this one on forward. Where
I stand, anyway, at this point is that all the
analyses I've seen you're between 4.5 and 8.5. That's
four orders of magnitude of hydrogen ion concentration
and you intuitively say that's going to make a big
effect on cracking susceptibility. If you have data
and I understand there is data to show that you can go
over that range and you're only going to change the
crack propagation rate by a factor of two, then forget
this problem for the time being. Forget it, and let's
move on.
That would be my technical view. I think
go back and revisit it later on. But if you're going
to move forward, scrub that out as a major issue at
this point.
MEMBER ROSEN: I'm having a great
difficulty believing this is a primary water
environment. Primary water is under 2000 pounds per
square inch pressure. Clearly, that's not what's
happening in this annulus. So just from that
standpoint alone, it's not a primary water
environment.
MR. HACKETT: I think that's what Dr. Ford
was getting to and I know you guys will want to move
on, but we talked about this yesterday for many hours
and in fact, we could probably devote an entire
meeting to this subject easily.
Maybe it's incorrect to be saying primary
water, but maybe largely more that way than more
concentrated, once the communication is there with the
thru-wall crack. I think that's where we're getting
to.
To move beyond that, one of the things
that's of interest here, if that is a problem, we have
no data for that specific environment. What we do
have a lot of data for and that's one of the good
aspects of this slide is there is a lot of data for
Canal 600 in a primary water environment from a lot of
sources. There's significant variability in that
data, but it's real variability. It's actually data
that we have as Dr. Ford has pointed out. It's also
appropriate, we think, to consider that variability on
a heat basis as opposed to an actual per data point
basis. The net effect of that is it expands your
uncertainty bounds when -- because you have fewer
numbers of heats than you have total data points.
The overall problem is complicated by
consideration for multiple initiation sites. You
could account for this in several ways. At least one
way is that you're multiplying your crack growth rate
potentially and we certainly have seen the crack
growth rates in excess of units per year are possible
for this type of phenomenon.
In terms of the stress state, I think this
is an area where we're still developing some details
on some of the inputs. I think the ones that are
fairly straight forward, the first bullet are stresses
due to primary pressure and crack face pressure.
However, the primary drivers in this particular
phenomenon are really residual stresses and I've tried
to summarize some of the sources there, the main
source being welding. Any time there's a welding
process, there's potential for leaving up to yield
level residual stresses from that process. This is
also a complicated manual weld which is done in a --
it's basically a helical sort of situation. The
fabrication processes and sequences can change the
residual stress state or affect it significantly. The
installation procedures for the penetrations
themselves can result in residual stresses from
straightening or bending that might have occurred when
they were put together.
Modeling details can be fairly critical in
how many passes are being modeled. If you're looking
at finite elements to try and figure out what the
residual stress state is, how many elements do you
need to go through the thickness there to get the
refinement that you need. And there's been -- the
industry has done some work in pointing out the
influence of thru-wall strength gradients that looked
to be a factor at Oconee, for instance, with shifting
the primary stresses, the larger stresses to the outer
diameter of the penetrations.
There's also -- these are the first two
bullets are ones that have been considered so far in
the analyses that we're aware of that we've done and
that the industry has done. Dr. Wilkauski has pointed
out the two bottom sources that really haven't been
considered thus far, that we're looking at, addressing
in the analysis that the NRC is sponsoring. One is
differential expansion at the root of the J-groove
welds. You know, we've talked about and the industry
I think has demonstrated pretty convincingly that
there's a separation when the plant is under pressure
and heated up between the penetration and the head
itself.
That's a good thing from the perspective
of leakage, but what it does for people who are
familiar with fracture mechanics, it induces basically
a crack tip opening displacement at the bottom where
the bottom of that annulus where it's joined to the
J-groove weld. We haven't really assessed the driving
forces from that.
There's also potential for a contribution
from cyclic thermal stresses. That was discussed
yesterday at the meeting that we had with the
industry. Dr. Scott made some comments on that in
terms of this may not be a principal driver for like
fracture mechanics, but it may have the effective
breaking of film in terms of the film rupture model or
it may have initiation impact on the initiation.
Anyway, these are considerations that we still need to
address.
MEMBER FORD: On the residual stress
aspect, I know you can do a whole lot of finite
analyses to determine what these could be. Are there
any qualifying measurements?
MR. HACKETT: Not that I'm aware of. I
turn that to --
MR. HISER: Maybe somebody on the industry
side could speak to that a little more.
MR. PATHANIA: I'm Raj Pathania from EPRI.
I'm working on MRP project. AS a part of this problem
back I think about 7 or 8 years ago, measurements were
done on CRDM nozzles in France and then those
measurements were checked against the finite element
stress residual stress models and there was reasonable
agreement. There's a published EPRI report on that
and we can provide that report if anyone is interested
in that. There was reasonable agreement both with the
residual stress measurements and the fact that when
you weld these things, the tube goes over because of
distortion and so the ovality measurements themselves
are a reasonable way to check out your models.
And in addition, I believe Westinghouse
has done some measurements also on the CRDM nozzles.
MR. BANFORD: Yes, this is Warren Banford
from Westinghouse. We made some measurements of the
ovality, of the tubes below the head. Because of the
weld, being on an angle, what it does it makes the
tube go from circular to oval and we used finite
element results that we got that we were using to
predict residual stresses and we used those results,
the predictions of the ovality and check with the
actual ovality that had been measured in the field of
tubes in that particular position and we found very
good agreement. So that was the way that we bench
marked our finite element work, in addition to the
French work which was further confirmation.
MEMBER FORD: And the stuff that you were
talking about was primarily x-ray?
MR. PATHANIA: No these are strain gauge
measurements.
MEMBER FORD: Strain gauge, okay.
MR. HACKETT: And moving on to address
structural margins. Obviously, there's a need -- one
of the bottom lines of the analysis is Professor
Apostolakis was getting to earlier is maintenance of
appropriate structural margins. Where we are with
this is obviously with a PFM analysis, the intent
would be to perform this on a best estimate basis. In
fact, what the second bullet goes is that at this
state of development we have enough uncertainties in
this that we're addressing that if we had to try and
assemble this right now we would need to be adding
margins in certain elements to address these
uncertainties. This is not the way you want to do
this type of analysis. That's where we would be right
now.
What we're hoping to get to is obviously
a greater level of refinement that we don't need to do
that. But in this case we had to do it, where to
apply the margins and what the magnitudes would be
problematic.
CHAIRMAN APOSTOLAKIS: Is this where you
disagree with the industry or do you disagree at all?
We heard earlier there was an estimate of delta CDF
and so on that you have not examined yet, the
calculations. But is it a disagreement here
somewhere?
We are being presented with programs that
will calculate this and that, difficult problems and
so on, but in terms of real decision making in the
next few months, is there a disagreement somewhere?
MR. STROSNIDER: This is Jack Strosnider.
Let me interject this thought and with regard to some
of the plant-specific information you heard earlier
and we received that evaluation within the last week
and it's under review by the staff now. I think the
plan is to meet on it next Tuesday and to go into some
of the detailed discussions, so that review is in
process.
I think from the generic point of view and
from the development of this technology I don't' think
there's disagreement. There's technical issues that
we need to share information on industry and NRC to
make sure that we come up with the most reasonable and
appropriate models to do this. That's really what the
purpose of putting out the preliminary technical
assessment document was, so that people -- the
industry and others could see where we're at in terms
of this assessment. And if there's information that
can be brought to bear, if there are some assumptions
or way we're interpreting information that people
think is incorrect we want to hear it and we want to
work that. So I think if you get down into some of
these specific technical areas, there's certainly room
for discission. If you want to call that
disagreement, maybe there's some issues that need to
be worked out, but that's the process we're working
through in terms of trying to develop this model.
MR. HACKETT: For the bottom line of
where we're heading for this and where the technical
assessment goes right now, is looking at the impact
and Larry Mathews covered some of this earlier,
inspection methods and particularly timing. This goes
to -- we had a lot of debate yesterday over when do
the cracks start and how fast are they growing.
Certain inspection methods for this type
of phenomenon have been quote unquote qualified. One
per the Bulletin is this qualification for leakage
that was discussed earlier by the industry with the
idea that you have the leakage path and you do have an
access to the bare metal that you can make the
determination.
The other one is the previous phenomenon
back in 1997. There was an approved methodology for
using Eddy currents inspection technology for looking
at ID cracks on the penetrations themselves.
Moving down further, there are other
methods that really remain to be qualified and I guess
clarification is probably needed there and
qualification in this case, I was probably looking at
as an ASME PDI type qualification performance
demonstration initiative, where you're looking at
qualification of both methods and personnel to be able
to detect and size some of these type of defects. And
in that case, you're looking at probably both surface
and volumetric exams through the J-groove welds, also
volumetric exams for the vessel head penetrations and
being able to shoot through that wall to look at OD
cracking.
Bottom line there is that the
determination of the appropriation inspection
intervals from PFM will obviously depend on the
reliability and effectiveness of those inspections to
at least some degree which is really -- remains to be
determined.
Allen is going to go into some more of
this in detail, but just in terms of the tech
assessment, but just to sort of summarize where we've
been going, this has really been -- this is RES
initiatives on reactor vessel head penetrations. It's
very much really been a team effort at NRC between RES
and NRR and the research contractors to developing an
integrated technical perspective on this, largely
using a probabilistic fracture mechanics construct.
As I mentioned, the conclusion at this
point is that the development so far indicates that
there is some key inputs that are lacking appropriate
data and/or are highly uncertain and some of that we
may be able to do some things about. Some of it may
be just what we're going to have to live with, that
the current state of development of these
uncertainties do limit the ability of this type of
assessment to do a lot for you. And we're hoping,
obviously, that the continued efforts in this area are
going to result in reductions in these uncertainties
and make this a more viable methodology for assessing
the overall impact of this phenomenon in the future.
MEMBER ROSEN: Is this a separate, or a
scientific search for truth or is it an attempt to
ultimately be able to apply these techniques to plant
specific situations?
MR. HACKETT: I guess we like to feel that
we're always looking for the truth in terms of the
research aspect of the thing. But I think the answer
is really more the latter, because I think we could be
searching for the truth for a long time, probably at
least the rest of a lot of careers that I can think of
and in this case, we have some real practical problems
to come to grips with near term as Jack was mentioning
earlier.
So I think it's really going to be more
the practical lean and there will obviously need to be
compromises along the way in that regard.
MEMBER FORD: But my earlier question, you
are planning by say early spring of 2002, you will
have a defensible disposition algorithm?
MR. HACKETT: That would be the --
MEMBER FORD: That would be either
deterministic or probabilistic.
MR. HACKETT: That would be the goal and
then to have -- I think what we have now is Dr.
Apostolakis asked about agreements, disagreements. I
think with the industry and the NRC, I think there's
fairly wide agreement on the elements that need to go
into this thing.
MEMBER FORD: Right.
MR. HACKETT: I think you've seen that
today. I think there's just really a question on some
of the key input varies in discussions over
uncertainties in those. But I think the construct is
there, and I think the answer is yes. Hopefully we'll
be in a position.
MEMBER FORD: And so that will have an
effect, not on inspection periodicities, but it will
also have a feedback into the completeness, if you
like, of the prediction algorithm that you're
currently using for privatization of inspections?
MR. HACKETT: That's correct.
MR. STROSNEYDER: This is Jack Strosneyder
and I'd just comment with regard to the practicality
of the work, and I think Allen's going to present, if
we get through, the material that he has. If we get
toward the end of that, there's some curves that are
in the technical assessment documents which were
intended to present the results of the work done so
far in a format such that it could be help to inform
decisions, recognizing that there's uncertainties and
it's not the best model for doing that because there
are still uncertainties we need to deal with.
But we're trying to put this work into
that framework so that it can be used to inform those
decisions.
MEMBER FORD: Well, we've got to see those
graphs. I love graphs and we're going to do it by
half past 10:00, so let's do it Allen.
MR. HISER: Let me ask your guidance.
MEMBER FORD: Get rid of all the pictures.
MR. HISER: No pictures?
MEMBER FORD: Yes, pictures. Okay, I'm
kidding.
MR. HISER: There are two parts of my
presentation. One is the preliminary staff assessment
that you've had some time to look at. The other is
more an overview of the inspection findings and some
photos of some of the visual exam results.
MEMBER FORD: Fine.
MR. HISER: Which one would you like me to
start with?
MEMBER FORD: Provided that you'll promise
me that we'll see some actual graphs.
MR. HISER: Okay, you'll see some actual
graphs. They may be short of background and we'll
jump to graphs.
MEMBER FORD: Right.
MR. HISER: The first thing I'll start with
is just an overview of the bulletin very briefly, the
status of the inspections and things, and the status
of our understanding at this point. When we spoke to
the committee, in I believe it was July, we were
talking about a proposed bulletin. The bulletin was
issued in early August. We did ask questions of all
plants related to the plant specific aspects of
susceptibility, the number of nozzles, things like
that.
We were looking at what is the type of
insulation on the head as it relates to the ability to
do visual inspections, and also that's more the
inspection side. We also were asking what are the
consequences and what structures, cabling, things like
that are above the head that could be damaged if a
nozzle were to become ejected.
For plant specific consideration, we did
have questions related to the susceptibility of each
plant, specific questions if they found cracking or
leakage, what the extent of that was, what their plans
and schedules for future inspections were, and then
also how those plans would meet the regulatory
requirements. That's just a brief overview of the
bulletin.
Within the bulletin, we speak of a graded
approach to doing examinations for plants that have
varying levels of susceptibility to cracking, or have
a history of cracking. We think that it is
appropriate to do different examinations. The
bulletin speaks of an effective visual examination,
and effectively what that is, is that you have access
to the bare metal where the nozzle intersects the head
and you can detect boric acid deposits. That assumes
there's no insulation in the way, no other
impediments, that there are no other boric acid
deposits in all honesty. It's hard to differentiate
one from another.
So that would be, I guess if you will, the
lowest level of examination. The next would be a
plant specific qualified visual exam that's been
mentioned before. That's a plant specific
demonstration that you have a leak path, such that if
you get a leak in the J-groove weld, that there is a
way for deposits from that leak to come up to the
head. Then you can detect through an effective visual
examination.
The last inspection that's discussed in
the bulletin is a volumetric exam and this one, again,
is focused on the outside diameter of the nozzles to
detect circumferential cracking at that point.
MEMBER WALLIS: Does what you see give you
any indication of how far the crack has progressed?
If you see great big balloons of popcorn, that's going
to tell you something that's different from just
little clusters of popcorn?
MR. HISER: Within the results that we have
to date, in particular going back to the Oconee 3
results, we had 165 degree through-wall crack
circumferential. There doesn't seem to be any
relationship to amount of deposit in extent of
cracking.
MEMBER WALLIS: Okay.
MR. HISER: Now, the bulletin responses
came in in early September. The bulletin does bin the
plants into four groups. The first is those that have
exhibited cracking or leakage. That's five plants all
together. The other three bins, if you will, are
based on the relative susceptibility ranking. In each
case, the bulletin provides some suggestions on what
would be an appropriate inspection for each bin.
In particular, for those plants that have
found cracking or leakage, the bulletin suggests that
a qualified volumetric exam should be performed by the
end of 2001. Upon looking at the responses and
additional consideration of things, what the staff has
accepted is a qualified visual exam at the last outage
as being a sufficient inspection method.
As you can see, as you go down in
susceptibility, there are sort of reductions in
intensiveness of the inspections. The staff has
addressed clarifications to some of the bulletin
responses with specific licensees, and numerous
licensees have provided either revised or supplemental
bulletin responses. We're still actually reviewing a
lot of that information. As Jack mentioned, some of
it just came in earlier this week.
Now looking at the plants that have
performed their metal visual examinations, in all
cases except one these are plants that are either in
the high susceptibility bin or have exhibited cracking
or leakage at some point. The one exception is
Crystal River Unit 3.
What I'd like to indicate first of all, of
the ten plants that have done their metal exams within
the last twelve months, nine of them have come up with
at least relevant visual indications and that really
is the case for North Anna Unit 2. It's the reason
that their number's in parenthesis. Those are visual
indications that they're doing additional examinations
on.
The other eight plants have confirmed
cracks in the nozzles, in some cases, circumferential
cracks. At Oconee Unit 3, Oconee Unit 2 and Crystal
River 3, a number of nozzles have been repaired.
There are a number of nozzles that have been left in
service with axial cracks at this point.
VICE CHAIRMAN BONACA: The bottom line here
is you say that those plants at the bottom, Beaver
Valley 1, et cetera, et cetera, have completed
effective visual examinations. How do they differ
from bare metal?
MR. HISER: These are the plants that are
in either high susceptibility --
VICE CHAIRMAN BONACA: Okay.
MR. HISER: -- or had previously had
cracked or leaking nozzles of the top, including
Crystal River 3 in that population. That's thirteen
plants all together, so ten of them have looked, three
have not. Nine of them have found at least relevant
-- eight have found cracking. Robinson is the only
plant at this point that has not found any evidence of
--
VICE CHAIRMAN BONACA: So the effective
visual examinations on the bottom means bare metal
visuals?
MR. HISER: Right. Right, they have not
done the analysis to show that leaks would put a
deposit on the head.
VICE CHAIRMAN BONACA: Okay.
MEMBER FORD: Allen, as you look at -- put
the next one up. As you look at those, do you have
any comments on the question I asked Larry? Can you
draw anything from this information about the
difference between Babcock and Westinghouse reactors
in terms of susceptibility, or is it purely a function
of time at 600 plus or minus 2?
MR. HISER: I think it's too early to say.
The one comment that I can make, one observation,
seven BNW plants exist in the PWR population. Six
have looked and found cracking. Three of the six have
found circumferential cracking. At this point, we
have not seen circumferential cracking in Westinghouse
plants.
But then again, the intensity of the
examinations has not been as strong as the BNW plants.
That is what the bulletin was trying to get to, trying
to increase the level of inspection so that we could
determine whether there are any particular groups that
are more problematic in this area.
MEMBER FORD: Now, I've heard it said that,
maybe comment on the accuracy of it, that Babcock
seemed to have been using more susceptible heat for
whatever reason, and we don't know how to define
susceptible physically. Could you comment on that?
Is there any basis for that or not?
MR. HISER: I would say that we really
haven't looked at it to that level of detail. We
have, the industry has identified cracking in at least
one Westinghouse plant. That plant is doing repairs.
We really have not assembled the data and been able to
try to look for those kind of conclusions.
MEMBER FORD: Okay.
MR. HISER: I think the one comment that
I'd make on the -- and this is very similar to Larry
Mathews' graph, just a little bit different colors I
think, different color scheme. The plants that are
highlighted in the red circles have identified
cracking recently. The one plant, North Anna 2 has
indicated maybe that will end up turning into a blue
square.
The bulletin made a distinction of 5 EFPY.
The plants to the left of that were high
susceptibility. Those to the right were considered to
be moderate susceptibility. I think that the results
generally support that. The one plant in this area
that the bulletin maybe didn't designate properly is
the next plant in line.
MEMBER FORD: Crystal.
MR. HISER: That's right. Crystal River
identified a circumferential crack. So the
circumferential cracks are here and here. It seems to
sort of span the --
MEMBER FORD: But the ISI 5 is purely
arbitrary?
MR. HISER: Absolutely. Yes, the 5 was
just a first cut at trying to provide some data that
would be useful to look at. At this point, one plant
in the high susceptibility zone has identified no
cracking. Two at this point have not performed a
recent ISI. But it does appear that these are the
plants that are more susceptible.
As time marches on, you know, red circles
will begin to populate out here. This one plants was
included in the first bin of plants that have a
history of cracking, because I identified axial
cracking in 1994.
So my conclusion from this, susceptibility
ranking is working as intended. I mean, we don't have
any plants that have identified cracking at 20 or 30
EFPY. So, that's a good sign.
MEMBER POWERS: At 20 or 30?
MR. HISER: EFPY from Oconee Unit 3.
MEMBER POWERS: Have they looked?
MR. HISER: Some plants have looked using
the effective visual exam. They've looked at the bare
metal on the head and have identified no relevant
deposits. These are two plants, Kewaunee -- the other
two plants, I believe the one furthest out is about 22
EFPY from Oconee Unit 2.
MEMBER POWERS: How long from the time a
crack exists to the time you can find deposits?
MR. HISER: I think an expectation is that
from the time that you get a through-wall crack in a
J-groove weld, that the time to see the deposit is
hopefully less than the time to initiate a
circumferential crack. We don't have any firm data in
that regard.
Unfortunately what we're, you know, the
bulletin was information gathering. We wanted to try
to find out the population of flaws that are out
there, how severe is the problem. What we're ending
up with is a population that's sort of one-sided. If
you find a leak, you do additional inspections. If
you don't see a leak, you don't do anything. And
we're not sure if that, you know that clearly doesn't
cover the full range of possibilities in terms of
behavior.
Moving on from there, I just wanted to
show some of the typical visual examination results
from this fall. In both of these cases, cracks were
identified in the nozzles and so these would be
classified as leakers in a classic sense.
For this one as well. I don't know if you
-- let me show this next one. This is sort of the
classic popcorn appearance that the industry speaks of
in this area. This is what we've seen at Oconee Unit
3 and Unit 2 and ANO 1. That's sort of the classic
deposits. In this case, I think this is easy to tell
that you have popcorn deposits. On some of these
others -- How do you interpret that? Is that a
relevant -- is that popcorn? Is that a relevant
indication?
CHAIRMAN APOSTOLAKIS: When you say "that",
that is that?
MR. HISER: These deposits.
CHAIRMAN APOSTOLAKIS: That's just -- okay.
MR. HISER: It clearly isn't just bare
metal with the nozzle coming into the head. I guess
we don't have any on that one. But the industry has
treated nozzles like this in general as relevant
visual indications, and they've done additional exams,
volumetric surface exams. I think they have in
general done a very good job of following up
indications like this.
The interpretation the industry has tended
to put on this is they do not call this a leak until
they confirm that through volumetric or surface
examinations. The difficulty comes in when you get
things that look like this.
MEMBER ROSEN: Would you go back to that
prior one.
MR. HISER: Sure.
MEMBER ROSEN: Because it just seems to me
so clear what's going on here. I'm sort of puzzled by
you not saying it. That looks to me like a very old
leak that may have somehow dried up. Clearly there's
damage there. It's not normal. It's not the way it
was built.
MR. HISER: One of the problems that many
of the plants have is they have leaking CRDM flanges,
CONO seals and canopy seals, and that provides boron
on the head, differentiating is it from the nozzle,
from under the head leaking out, or is it from an
external source, is part of the difficulty in
interpreting the visual exams.
What the industry has consistently done is
to say, this required more attention. They're not
sure if it's from a leak in a nozzle or from an
external source, so they've done additional exams to
try to determine the source.
This is the next one that I put up. That
looks similar to what we saw at Oconee and the early
graph that I showed. The problem though in a case
like this, is trying to find where the crack is so you
can fix it. And, in some cases like this and also
with this nozzle, this is the same nozzle. These
kinds of indications in that area are sort of
ambiguous.
In the case of this nozzle, the licensee
did extensive ultrasonic and surface exams, found no
leaks, no surface cracking, no source of deposit. The
conclusion that they reached and that we agreed with,
was that this came from an external source, in this
case a CONO seal leak.
What I wanted to do is just give you a
flavor for the kind of visual examination results that
we're seeing and having to interpret. I think
overall, the industry has done a good job in following
up relevant deposits.
MEMBER ROSEN: Would you go back to that
for a moment?
MR. HISER: Sure.
MEMBER ROSEN: Why would you say that's a
CONO seal leak? Wouldn't you expect to see some
evidence of it coming down from above? And yet, that
penetration looks perfectly clean to me above the area
that you indicated.
MR. HISER: And in the case of this
licensee, and I think again we're finding a lot of
variations from plant to plant. This licensee has a
very high air flow rate through the upper head area,
and the CONO seal leak deposits don't drop straight
down, but they're blown every which way.
Looking at videotapes of this head, you
can see on some nozzles where there are streaks along
the side of the nozzle, so the leakage comes down
vertically, gets caught up in the air flow and is
blended horizontally. There tends to be boric acid in
many placed on the head. It tends to be fairly well
localized, but it's not -- I think one could -- look
at this case.
MEMBER WALLIS: The only place we see it in
that picture is where you would see it if it came out
of a crack.
MR. HISER: The deposits back here?
MEMBER WALLIS: Yes, that bottom one.
MR. HISER: You really have to see the
videos to get a full appreciation for this, but there
tends to be a lot of boric acid. This does not look
like the popcorn appearance that we've seen
classically with the circumferential cracks in
particular. This has been followed up with additional
exams. No relevant cracking has been found.
MEMBER WALLIS: It could mean you don't do
a good job of finding cracks.
MR. HISER: The inspection methods have
been qualified with service cracks. Clearly, there's
a chance that that has happened as well.
MEMBER FORD: Could I suggest we move on
Allen? MR. HISER: Sure.
MEMBER FORD: I recognize that are
scuppering you here, but looking through here, there's
a lot of good information at the back here that I want
these gentlemen to be exposed to. Now, do I
understand that we are allowed to go to a quarter to
11:00? George?
CHAIRMAN APOSTOLAKIS: I'm sorry?
MEMBER FORD: Can we go until quarter to
11:00?
CHAIRMAN APOSTOLAKIS: No. I'm sorry. We
have an absolute deadline of 1:30 with the
commissioner coming.
MEMBER FORD: 10:30?
CHAIRMAN APOSTOLAKIS: So I can't move
things.
MR. HISER: How would you like me to
proceed. I can jump to the bottom line more?
MEMBER FORD: Why don't you jump to the
bottom line if we must leave at 10:30. It's obvious,
I think, that we're going to be having another full
day meeting sometime.
MR. HISER: Let me speak a little bit to
one of the areas of contention yesterday in our
meeting. The crack rate is an area that we don't have
alignment if you will between the industry and -- at
this point in time. This is some data for Alloy 600.
The top curve illustrated here is what would be called
a 95/50 confidence bound on the data. The curve at
the bottom I believe is a 50/50 analysis of a larger
bulk of data.
What the staff has used in its preliminary
assessment is a 95/50 curve as an upper bound. In
addition, we have defined what I would call a high
mean value. What has been seen and has been
hypothesized is that the heats of the material that
are going to crack are the ones that are most
susceptible.
So instead of looking at the full
population of data, one should really concentrate on
the upper half of the data. We've defined a high mean
as the square root of the product of the 95/50 and the
50/50 data, so it would be a curve that lies
essentially in between. In this case, I think it --
MEMBER FORD: From where you stand right
now Allen, if you were asked what disposition curve do
I use, would you choose the 95/50 or the 50/50, and
what would the industry want to do?
MR. HISER: I won't speak to what the
industry would like to do. My guess is that, well
what we recommend in the technical assessment is a
95/50 curve.
MEMBER FORD: Good.
MR. HISER: I do not believe the industry
would recommend the same curve.
MEMBER WALLIS: Where is K for your
situation here on the picture?
MR. HISER: For the nozzles, we are
generally in the range from 30 to 50 or 60 MPA root
meters.
MEMBER FORD: And I'm reminded there's a
second. There's a centimeter in 100 days. It's called
insignificant.
MR. HISER: The difference from the 50/50
to the 95/50 is about an order of magnitude.
MEMBER FORD: Yes, but the top line is not
an
insignificant crack. It's pretty rapid.
MR. HISER: No, it's about an inch, 30
millimeters per year on that order.
MEMBER FORD: That's right. So again to
come down to the bottom line where you stand right
now, unless there's other information that comes
available the next two months, you have been
suggesting an inspection periodicity based on that
10(-9) meters per second rate?
MR. HISER: Based on the 95/50.
MEMBER FORD: Yes.
MR. HISER: And that puts you in that sort
of a crack root.
MEMBER FORD: PWR environment?
MR. HISER: Yes.
MEMBER FORD: In the PWR environment.
MR. HISER: Now there was a little bit of
mention earlier about the annular environment, what is
correct? Is PWSCC really an accurate way to
characterize what's going on in the environment? The
technical assessment indicates that if there are any
upset chemistries in the annulus, the effect on crack
root may be up to a factor of 2. I think maybe the
95/50 is one interpretation of it, is that it would
account for things like that.
Looking ahead to Slide 17, the staff has
looked at deterministic and probabilistic assessments
and there's discussion of both of those in the
preliminary assessment. The parameters that we've
used are here. I think the numbers are not in
substantial controversy on critical flaw size, about
270 fora factor of safety of 2 or 3 on pressure; 342
is what our analysis indicates for nozzle failure and
possible ejection.
As I mentioned for crack growth rate,
we're using a 95/50 statistical bound. We're looking
at a temperature of 318°C for this conditions. You
get an A for the Scott Model as indicated there. For
the initial flaw size, we really don't know what the
flaw size is. Without inspection data, it's hard to
draw conclusions on a plant specific basis.
In lieu of that, our deterministic
assessment uses a range of initial flaw sizes to see
what the effect of that is. In terms of uncertainty
and sensitivity studies, we've looked at different
statistical bounds to the crack root data, the effects
of temperature and again, we're using an initial flaw
size as a parameter.
CHAIRMAN APOSTOLAKIS: I was just informed
that Commissioner McGaffigan is willing to come a
little later. So, you can have your fifteen minutes
of glory.
MEMBER FORD: Thank you George.
CHAIRMAN APOSTOLAKIS: Thank the
commissioner.
MEMBER FORD: There's a lot of stuff here.
MR. HISER: This is an illustration of the
K that we're using as a function of crack half length.
It peaks at about 60 KSI root inches. This area is
principally due to residual stresses, mainly from
welding on the tail end of the curve is due to
pressure stresses, and we've tried to model as best we
can at this point in time what the K range or K curve
would look like for the nozzles.
We talked about crack sizes. Thus far,
five circumferential cracks have been identified
through inspections. The first two were verified
through destructive examination. The last three have
not. In particular, I guess what I'd like to point
out, the one destructively examined crack of 165&, the
ultrasonic record indicated an inside diameter extent
of 59ø.
So clearly, there's a lot of uncertainty
on the ability of the ultrasonic method to size these
cracks. But it would be very helpful from an
analytical standpoint to have some sort of
confirmation of cracks that occur in the future,
instead of just the machining the cracks away, some
way to confirm the size of those would be very
helpful.
For the base case, again 318øC, 95/50
crack growth curve. That's illustrated here as a
function of crack length. The maximum growth rate
would be about 1 3/4" per year down to about ‹‹" per
year.
MEMBER FORD: Allen, could I presume just
to tell you what to do, because I want to make sure --
MR. HISER: Sure.
MEMBER FORD: -- all the members have time
to ask you questions. There's a tremendous amount of
work here. Could I suggest maybe you discuss Slide
#27.
MR. HISER: Okay.
MEMBER FORD: I'm really interested then to
hear the bottom line on your inspection and what the
next staff plans are.
MR. HISER: Okay.
MEMBER FORD: And then that would give time
for all the members to ask any closing questions.
MR. HISER: Now, as a part of the
uncertainty and sensitivity study, we have looked at
temperature and statistical bound to the crack growth
data as parameters. What is indicated here is the
operating time from an initial crack length to the
nozzle failure, nozzle ejection, crack length. We
have different crack growth curves. As an example,
the 318øC bound curve would be at that temperature.
The 95/50 curve, the curves marked as M are not the
50/50. It's a curve that's between the 95/50 and
50/50 curve, so it's more of a high mean value.
As you can see, there are affects of
temperature. If you go from 325ø, 318ø to 315ø,
reducing the temperature increases the operating time
to failure. More substantially, I think, is the
effects of statistical bound, if you go from a 95/50
to a mean curve, you get very substantial changes.
CHAIRMAN APOSTOLAKIS: When you say 95/50,
what do you mean?
MR. HISER: That means there's a 50 percent
probability that you've bounded 95 percent of the
data. That's 50 percent confidence that you have 95
percent bound. As an example, the 95/95 would be 95
percent confidence that you bounded 95 percent of the
data.
Using the 318ø bounding curve, the 95/50
curve for flaw sizes smaller than 120ø, you would have
at least four years before you would reach a critical
flaw size.
MEMBER FORD: Could you just comment on
what you call the industry crack growth rate?
MR. HISER: The industry crack growth rate
curve is 10 millimeters per year. I'm not sure the --
this came to us in at least one submittal and I think
is more of a mean curve or maybe below mean overall.
MEMBER FORD: So obviously as we move into
the future, there's going to be a lot of discussion
between you and the industry as to which one of those
curves is appropriate for a specific plant?
MR. HISER: Yes, clearly one of the ideas
is that depending on the level of inspection could
impact which of those curves would be appropriate for
the plant to use.
CHAIRMAN APOSTOLAKIS: Let's see, all of
them start at 300 and something degrees.
MR. HISER: Right, 324.
CHAIRMAN APOSTOLAKIS: What is that now?
MR. HISER: Okay, what this means, if I
have a 60øC to start with and I'm using the 318øC
95/95 curve.
CHAIRMAN APOSTOLAKIS: Oh.
MR. HISER: It means I would grow to
failure in about eighteen months.
MR. Leitch: 324 is four phase.
CHAIRMAN APOSTOLAKIS: Okay, and the
confidence you're referring to comes from the
percentage of the distribution on a number of
parameters of going to the mother, right?
MR. HISER: In this case, it's totally on
the crack growth rate.
CHAIRMAN APOSTOLAKIS: Oh, just the crack
growth rate.
MR. HISER: This only has crack growth rate
as a parameter. That's the only thing that we're
varying in any of these curves.
CHAIRMAN APOSTOLAKIS: Right.
MR. STROSNEYDER: This is Jack Strosneyder.
I'm glad that point came up because I wanted to make
it. There's one random variable in the plot you're
looking at here, and that is the growth rate for
circumferential cracks. All right.
When we talked briefly about some of the
other uncertainties with regard to what the stress
levels are and certainties in the inspection and what
might be left in service after inspection and those
sort of things, that's not reflected in this analysis.
Those are the things we need to develop in order to
come up with the full probabilistic assessment.
CHAIRMAN APOSTOLAKIS: The random variable
is the operating time to failure.
MR. HISER: That's right.
CHAIRMAN APOSTOLAKIS: And the others are
the uncertainty variables, and so far you have used
only the crack growth rate.
MR. HISER: That's the one variable that we
have sufficient data on that we feel we can do the
sort of analysis.
CHAIRMAN APOSTOLAKIS: So, if I pick one
value of the crack growth rates, say the 50th
percentile, the median, what kind of a distribution
have you assumed for the operating time?
MR. HISER: That's what this curve shows
you that the median's plotted on there and you can
look at --
CHAIRMAN APOSTOLAKIS: No, there's no one
curve. I mean, you're down horizontally.
MR. STROSNEYDER: Al, explain the 95/50
where that's at for a 315 center. Let's just pick
that because it's an easy color.
MR. HISER: This would be a 95/50
representation of the crack growth rate.
CHAIRMAN APOSTOLAKIS: When you say "this"?
MR. HISER: The green curve.
CHAIRMAN APOSTOLAKIS: Okay.
MR. HISER: The green curve.
CHAIRMAN APOSTOLAKIS: Okay, 95/50. So
this is the 95th percentile on the operating time,
using the 50th percentile for the crack growth rate?
MR. HISER: No, this is a deterministic
calculation. We take the crack growth rate that
correlates to a 95/50 bound on the crack growth data.
CHAIRMAN APOSTOLAKIS: Which is a 50
percent confidence.
MR. HISER: For each crack size, how long
it will take to grow to the failure, to the nozzle
failure crack.
CHAIRMAN APOSTOLAKIS: So this is not the
95th percentile of the operating time?
MR. HISER: No.
MR. STROSNEYDER: No.
CHAIRMAN APOSTOLAKIS: Oh.
MR. HISER: The 95th percentile on the
crack growth data.
MEMBER WALLIS: In the reactor environment.
MR. HISER: Assuming a primary --
CHAIRMAN APOSTOLAKIS: What is the 50 then?
MEMBER KRESS: It's the mean value of the
95 percentile. The distribution is about how well you
know the 95 percentile, and this will be the mean of
that distribution.
CHAIRMAN APOSTOLAKIS: Do you have a
distribution on the crack growth rate?
MR. HISER: Yes. It's humongous.
CHAIRMAN APOSTOLAKIS: That's one
distribution.
MEMBER POWERS: It is an estimated
distribution on that and yes on confidence on how
accurate his estimate is.
CHAIRMAN APOSTOLAKIS: So that's what my
second question was. It's not one distribution then?
You have a family of distributions, each one with a
different degree of confidence, is that it?
MR. HISER: Yes, that's correct.
CHAIRMAN APOSTOLAKIS: So you don't have a
-- okay, so you don't have any other uncertainty
coming into the calculation of the operating time.
MR. HISER: Right, and if you assume
effectively a mean value of the crack growth rate, you
would predict very long operating times to failure.
If you assume a statistical bound 95/50, then it
brings it down substantially.
CHAIRMAN APOSTOLAKIS: So the mean value
then of the crack growth rate is on the relatively low
side, right? That's what you're saying?
MR. HISER: Yes.
CHAIRMAN APOSTOLAKIS: And then you have a
long tail?
MR. HISER: Right.
CHAIRMAN APOSTOLAKIS: And the fun begins
when you move onto the tail.
MR. HISER: Absolutely. And the philosophy
that went into the assessment is, which nozzles, which
materials are going to fail in service? It's probably
not the mean values we're concerned about. The
materials that are more sensitive, where the cracking
would be more aggressive in the material. So from
that standpoint, like a 95/50 curve is more
appropriate from a philosophical standpoint.
MEMBER POWERS: I mean, I appreciate it.
CHAIRMAN APOSTOLAKIS:
I don't understand it.
MEMBER POWERS: The qualifying statement,
why not on a 95/95?
MR. HISER: At this point, I think that may
be too extreme.
MEMBER POWERS: What I'm trying to
understand is how do you decide?
MR. HISER: It's pretty much engineering
judgment.
MR. STROSNEYDER: I would interject two
thoughts on this. One is -- this is Jack Strosneyder.
One thing when we look at the data, and Allen had put
the plot up before and remember that it's on a log
plot, so there really is a lot of variability, right.
But the other thing that comes into play here on these
confidence levels is that when you do this by heat,
there's a relatively small number, I think it's 20, 18
or 20.
MR. HISER: 11, it's actually 11.
MR. STROSNEYDER: So of course the
confidence amount is driven by the amount of data that
are available too, so we want to capture what we think
is the real variability in these data, all right. But
on the other hand, if you recognize you're dealing
with a small amount of data and you use those high
confidence levels, you can drive yourself to some very
conservative values. So, that's one way of looking at
it.
MEMBER POWERS: I understand that. I just
am trying to understand what goes into the engineering
judgment, that 95/50 is okay but 95/95 is too extreme.
MR. HISER: I think what we've seen so far
is that the 95/50 operates as an effective upper bound
to the data.
CHAIRMAN APOSTOLAKIS: So what is up curve
now, 95/50? Is it there?
MR. HISER: It's the ones marked B.
MEMBER POWERS: But if I looked at the
data, I think in your 95/50 curve, I think I can find
a data point. I better be able to find a data point
that's above it.
MR. HISER: That's correct
MEMBER POWERS: So the 95/95 would be even
a better upper bound to the data.
MR. HISER: Correct.
MEMBER POWERS: I'm just trying to
understand your engineering judgment. I don't doubt
your engineering judgment. I'm just trying to
understand exactly how it comes out.
MEMBER FORD: If I could interject here.
Allen, I apologize for asking to put this slide up,
because it's started a whole lot of questions.
CHAIRMAN APOSTOLAKIS: That's why you got
15 minutes.
MEMBER FORD: But if the members just look
at Chart 31, you can just see overall what the staff
are planning and on that basis, I'd like to go around,
remembering that his meeting, the objective of this
meeting is merely to inform the full committee that
things are moving forward and it's very obvious things
are moving forward.
CHAIRMAN APOSTOLAKIS: I mean, that's an
interesting point. Let's go back to that. In
documents like Regulatory Guide 1174, there is some
guidance. It says "use the figure with the bounds and
the limits using the mean value" but as your mean
value approaches some limit, there is increased
management of tension, which means now you're looking
at the tail, how much probability there is above the
limit, and you know, you start sending RAI's.
Why can't we apply the same philosophy
here. Instead of saying we have to select the 95/50
or the 95/62, apply some consistent approach and say,
you know, maybe I'll go with the mean but then when
something happens there will be increased management
attention.
MR. STROSNEYDER: This is Jack Strosneyder.
I just point out an application of this information,
it's been pointed out there is only one random
variable that's considered in this, and clearly some
judgment needs to be involved in how you take this
information and apply it on plant specific
discussions. And recognizing, not only how we're
going to treat this random variable, but recognizing
that some of those other variables and uncertainties
are not yet included in the analysis process, which is
the work that remains to be done.
CHAIRMAN APOSTOLAKIS: Are you going to
include the model uncertainty as well? Are not these
equations --
MR. STROSNEYDER: I'm sorry, I didn't hear
the question.
CHAIRMAN APOSTOLAKIS: The equation that
gives you the operating time as a function of the
crack growth rate, is that cast in stone or there is
uncertainty about it?
MR. STROSNEYDER: That's two numbers moth
planned together.
CHAIRMAN APOSTOLAKIS: That's all.
MEMBER KRESS: That's all it is.
MR. HISER: But in terms of the K that goes
into determining what the crack growth rate is at each
crack, that's a parameter. I mean --
CHAIRMAN APOSTOLAKIS: My second comment is
that, you know, I realize there is a language that is
being used in this community, but again we're going to
try to make it consistent with the language we use in
the risk-informed approach. This 95/50 business, I
just don't like it. I mean, we have terminology.
MEMBER KRESS: George, that's good
terminology. I like the terminology because it's very
descriptive.
CHAIRMAN APOSTOLAKIS: It covers the data?
What is that?
MEMBER KRESS: Yes, but I think you have a
real good point George.
CHAIRMAN APOSTOLAKIS: There is no
probability theory that says that.
MEMBER KRESS: Oh sure.
CHAIRMAN APOSTOLAKIS: If you open up a
epistemic distribution with the percentiles and that's
the state of normal distribution, that's all you do.
MEMBER POWERS: Nonsense. This is well-
developed analysis for experimental data.
MEMBER KRESS: Yes, absolutely. Everybody
does that with experimental data.
CHAIRMAN APOSTOLAKIS: Everybody does use
it.
MEMBER KRESS: Yes, but --
MEMBER POWERS: Greenfield, developing
experimental data uses that terminology.
CHAIRMAN APOSTOLAKIS: That type of data
goes into the distribution.
MEMBER POWERS: I know PRA has never
recognized experimental data, but the rest of us
actually work with that.
CHAIRMAN APOSTOLAKIS: We're talking about
bigger people here.
MEMBER FORD: If I could suggest --
CHAIRMAN APOSTOLAKIS: Give me one
probability that talks about this.
MEMBER KRESS: Oh, I can show you two or
three of them.
MEMBER FORD: I think I've lost control of
this.
CHAIRMAN APOSTOLAKIS: Back to you, Mr.
Chairman.
MEMBER FORD: No, he's not ready yet.
MR. HISER: Can I just make one comment?
CHAIRMAN APOSTOLAKIS: Final comment, yes.
MR. HISER: The intent is to do
probabilistic fractured mechanics assessments. What
would go into that is this variety of crack growth
rates.
CHAIRMAN APOSTOLAKIS: I understand that.
MR. HISER: Along with other parameters.
So that would be considered under the PFM analysis.
MEMBER FORD: Okay. The objective of this
particular meeting was just to let the full, all the
members know that we are advancing on this problem
which is a generic problem in my view. We were
planning on having a material subcommittee full-day
meeting on this in February sometime. We may need to
have it beforehand if we have any added value to the
full process.
On that ground rules, I'd like to go
around to the members and just ask if they've got any
closing opinions, statements, on what they've heard so
far.
MEMBER POWERS: I'm just tempted to say
something abusive about Dr. Apostolakis' views about
the distributions. I'll do that on the fly.
MEMBER FORD: Jack? George?
CHAIRMAN APOSTOLAKIS: I have a lot to say
but I think it's too late.
MEMBER KRESS: One point. You know, we
were debating about the source of which confidence
level and which confidence on that we would use, and
if you're going to use it in an overall probabilistic
fractured mechanics, you ought to use them all to get
a final distribution. But that doesn't solve your
problem because then you're going to have to look at
final distribution and say which part of it is of
interest to me?
I'd like to point out that Dr. Apostolakis
has said many times that this is good place to use
formal decision theory in order to decide where to use
it.
MEMBER POWERS: He's just as wrong about
that as he is about unprobability distributions too.
MEMBER KRESS: Yes, so I just wanted to
make that comment.
MEMBER FORD: Steve?
MEMBER ROSEN: I'd like to make a comment
on the pictures being worth a thousand words. The
discussion that we had earlier seems to be
extraordinarily tortured in trying to say that, for
example, this picture is not in fact some evidence of
leakage from a crack, and, I'm left very unconvinced
of all that rationale.
MEMBER LEITCH: I'm a little confused by
initial crack size versus crack growth. I don't have
any idea -- I'm sorry. I said I'm a little confused
by initial crack size versus crack growth. Do we have
any idea that when these cracks occur, they go to some
particular depth immediately.
And, what we're talking about here is the
growth after that initial cracking? Until we know
what that initial depth might be, is there any thought
or discussion on that, or -- in other words, do they
grow at the rates we've been talking about here from
zero, or what I'm saying, isn't there an initial crack
and then what we're talking about is growth rate?
That's what I don't quite understand. I don't know if
there's any information in that regard.
MEMBER FORD: Quickly within the next ten
seconds.
MR. HISER: There is uncertainty on
exactly what is going on. Clearly the crack growth
rate that we're using is fracture mechanics based.
It has certain assumptions to it, and does not apply
from zero, from the incipient crack. That is the kind
of thinking that we need in a more phenomenological
model to be able to extend this in a more scientific
basis overall.
MEMBER FORD: Graham?
MEMBER WALLIS: All this is based on crack
growth and PWR environments. I think you need to do
more on crack growth in a crack environment with the
real, some assessment or enough assessment of chemical
flow or electromagnetic, all the things that are going
on there, and I don't have a good factor for it.
But conceivably, you could have fluid
coming out of there at Mach 5 or something, whatever
it is, you know, drifting away to it. I don't have a
clue. But you haven't really said anything about what
happens to flow dynamically, chemically and all of
that and that is part of the problem.
MEMBER FORD: I'd like to finish off just
by thanking all the presenters and apologizing for
pushing you so much. We had only a little time. The
main objective, as I said before, was just to let the
full committee know that we are moving forward on this
problem.
I think we have a material subcommittee
potentially planned for February sometime, and if
there's any value added, as I say, to our involvement
before then, we're open.
George, I pass it back to you.
CHAIRMAN APOSTOLAKIS: Thank you, Peter.
We will recess until five minutes after 11:00.
(Whereupon, the above-entitled matter went
off the record.)
CHAIRMAN APOSTOLAKIS: Okay? The next item
is licensing approach for the Pebble Bed Modular
Reactor Design. Doctor Kress will guide us through
this topic.
MEMBER KRESS: All right. There's two
parts to this. You say one of them talks about the
future plant design workshop that Dana Powers and I
have participated in on October 10th and 12th.
I don't intend to say anything about that
because Dr. Powers once again has issued a remarkable
trip report on this, and I recommend to read it to get
the summary of that. I don't know if Dana wanted to
say anything about that particular workshop more than
his trip report or not?
MEMBER POWERS: No, I do note that Dr.
Kress listed a set of 17 or 18 regulatory challenges
that he distilled out of the meeting that also should
be examined in conjunction with the trip report.
MEMBER KRESS: Yes, those didn't
necessarily all come from that one workshop, but --
yes. Also, I remind the committee that during our
October ACRS meeting, we had a presentation from
Exelon giving us their proposed licensing approach for
the next PBMR.
I won't have to remind you of what that
approach is because I'm pretty sure the staff plans on
reminding us of what it was. I thought it was a very
interesting approach and, in fact, I thought it was a
pretty good one. But today, we're hearing from the
staff what their view of this approach is and perhaps
what issues, policy and otherwise that it might raise.
Then I guess I'll turn it over to who over here?
MR. LYONS: I guess I'll start off.
MEMBER KRESS: Okay.
MR. LYONS: I'm Jim Lyons. I'm with NRR.
I'm the Director of the New Reactor Licensing Project
Office. Today, as Tom was saying, we want to come
back and talk to you a little bit about the licensing
approach that Exelon is proposing, and we've got a
presentation to go through how we are seeing this.
Actually, here's Tom. Do you have some opening
remarks Tom that you want to make?
MR. KING: You may hear a status report on
what we think of the Exelon proposal for fitting their
design, the Pebble Bed into today's set of
regulations. You're going to hear our thoughts on how
we're looking at that, what criteria, thoughts we're
using to try and make a judgment on whether that's
okay or not okay, and again recognize that there's no
final decisions.
There's probably some policy issues that
are going to come out of this that ultimately the
commission's going to decide. But at least it will
make you aware of what they're proposing and what our
thoughts are at this point on it.
CHAIRMAN APOSTOLAKIS: Okay.
MR. KADAMBI: Thank you. Good morning Mr.
Chairman, and members of the advisory committee. My
name is Prasad Kadambi. This is Eric Benner and this
is a joint NRR research presentation.
We had an abbreviated presentation last
month and we'll try not to repeat some of the things
that were covered in some detail. This presentation
is somewhat preliminary, only to the extent that we
are looking for feedback from the committee, and we
want to take into account whatever we can glean from
the questions and comments, et cetera, so we can
reflect it in the commission paper which is due this
month.
We are looking for your questions,
concerns about the proposed licensing approach, and
our assessment of the licensing approach. We'd like
to --
CHAIRMAN APOSTOLAKIS: Excuse me. We will
review that document you will send to the commission?
MR. KADAMBI: Well, a predecisional draft,
I believe, has been provided. I mean, we are looking
for this feedback during the meeting and a letter if
you feel it's necessary, but we are not necessarily
requesting a letter.
The presentation that we will make broadly
follows the format of the paper. The message that we
got last month was that the ACRS wanted to hear more
about the staff's overall approach, including any new
ideas that we may bring to the table.
Our overall approach is to use the
commission's directions, decisions, and policy
statements to build a foundation for the PBMR
Licensing Review. We found a lot of guidance in these
documents and we feel that there is enough to proceed
along the lines that we'll describe.
What we did not set out to do was, if I
may so say, build a better mousetrap where, you know,
we would start from a clean sheet of paper and try to
invent a regulatory system or framework. We believe
that a separate effort is underway. We don't know
very much about it, but we believe that NEI is going
to be undertaking something like that.
As we said, the licensing of Fort St.
Vrain more or less began our venture into using the
current regulations into licensing something that is
not a lightwater reactor. Eric, do you want to talk
about that?
MR. BENNER: Yes, and again to emphasize
what Prasad was saying, and I believe you heard some
of this yesterday, that there are activities going on
both with the DOE NERI Project and NEI and looking at
building on 3, to look at a more clean sheet of paper
approach.
There's a commission paper due on that
middle of next year to look at various options, and I
believe one of the options that's going to be
considered is, you know, how would you apply the
current regulations with some other sources of
information to see how they fit, along with the clean
sheet of paper to see the relative merits of the
different approaches.
Basically, the group that was accessing
Exelon's approach looked at the licensing of Fort St.
Vrain more to assure itself that there is a way that
you could use the current regulatory structure to
license a non-lightwater reactor, and the meat of that
was really the applicability of the GDC, which for the
most part are pretty general. And, at that time, the
staff also focused on some of these higher level
topics of defense-in-depth and multiple concentric
barriers to contain radiation.
This was sort of the stepping stone for
the staff's look at Exelon licensing approach just to
see whether there was merit to moving forward with
using the existing regulations as a licensing
approach.
The next thing the staff looked at was the
pre-application review for the MHTGR. The meeting
that took place in the mid-`80's, the MHTGR was a gas-
cooled reactor. At that time, it had a steam cycle
and recently General Atomics has somewhat modified
this design to make it direct cycle, and we're going
to be talking about the application review of that
sometime in the future with the ACRS.
But for this review, there was extensive
staff work and DOE work on a licensing approach that
was very similar to what Exelon is proposing, and at
that time, the staff looked at conformance of the
approach with the advance reactor policy statement as
well as, just like in the case of the Fort St. Vrain,
the NRC regulations, reg. guides and standard review
plan that they found to be applicable.
Staff provided assessments in a new reg in
1989. After that time, there were the design
certification reviews going on for AP600, ABWR, and
System 80 Plus, and during those design
certifications, whenever the staff distilled an issue
that they thought would be applicable to the MHTGR,
they kind of put it off to the side and tried to
address it in the communications that were going back
and forth between the commission and the staff.
The staff basically wrapped all those up
in a paper to the commission containing a draft new
reg in 1985 and kind of discussed where they were
applicable and how they were applicable to the MHTGR,
and essentially the gas-cooled reactor technology.
So, we have used a lot of the findings
that the staff provided at that time also as a basis
for our review.
MR. KADAMBI: What we have tried to do in
trying to identify the main considerations that would
be the basis for the staff's review is to more
specifically sort of identify the particular
commission directives and decisions and the policy
statements in the context of the agency's strategic
plan and performance goals.
So what I'll try to do over here is try
and describe the individual components that we feel
are relevant in doing this. The advance reactor
policy statement in a sense provides a guidepost,
although it's not a qualitative one, it does say that
as a minimum, an advanced reactor should provide the
same degree of protection as the current fleet of
operating reactors.
CHAIRMAN APOSTOLAKIS: How do you mention
that?
MR. KADAMBI: Well, in a sense that's a
separate question which, you know, at this point I'm
trying to in a sense define the problem areas.
CHAIRMAN APOSTOLAKIS: As a guidance?
MR. KADAMBI: Yes.
CHAIRMAN APOSTOLAKIS: Okay.
MR. KADAMBI: So, trying to just lay out
the framework, if I can use that word, the policy
statement also says that the commission expects that
using simplified inherent passive or other innovative
means that enhanced safety margins would be realized.
MEMBER WALLIS: How can you have the same
degree of protection and enhance safety margins at the
same time? No, it's at least the same.
MR. KADAMBI: Yes.
MEMBER WALLIS: Oh, at least. Okay.
MR. KADAMBI: And I believe that one way of
looking at it, at least this is the way I look at it
and some of these things are my own thoughts right now
is that you ought to have more confidence that you're
achieving a given level of safety on some distribution
when you're using these simplified methods, et cetera.
MEMBER WALLIS: Oh, you should go with the
95/95 one.
MR. BENNER: During the design
certifications that we did go through, some of that
expectation was realized through some of the severe
accident management expectations. So that may be
another way where we talk about enhanced margins for
this particular advanced design.
MR. KADAMBI: In fact, the decisions that
were taken on AP600, System 80 Plus, et cetera, do
show us how the advanced reactor policy statement and
the current set of regulations are brought together in
order to accomplish what the commission wanted.
In all, this is what we perceive as
essential to the role of defense-in-depth philosophy.
This can not be stressed enough.
MEMBER WALLIS: Sounds like a religious
statement, invoking.
MR. KADAMBI: In many ways --
MEMBER WALLIS: It is.
MR. KADAMBI: -- it has become that.
CHAIRMAN APOSTOLAKIS: Is what? I didn't
get it.
MEMBER WALLIS: Religious statement,
invoking virtue by appealing to the sainthood so the
sanctity of the defense-in-depth.
MEMBER POWERS: Yes, but it's --
MEMBER WALLIS: Sure, some of us.
CHAIRMAN APOSTOLAKIS: Raise the slide
please.
MR. KADAMBI: It's such a vague thing.
CHAIRMAN APOSTOLAKIS: Put it higher, the
slide itself.
MEMBER WALLIS: Like saluting the flag or
something.
MEMBER POWERS: Well, I don't know that I
would agree with you that the depth is vague.
MR. KADAMBI: At least if I may, I'd like
to go to what we have found from the commission's,
again various documents that we have seen to identify
the individual attributes that feed into defense-in-
depth, and to some extent this is where, you know --
we begin with the basic objective of finding a
defensible basis for making an adequate protection
finding, and because defense-in-depth is so key to
that, there are several sources where we can find
guidance on what we look for on defense-in-depth.
The white paper called Risk Informed and
Performance Based Regulation, which was issued in
March of 1999, you know, it was issued to the public
and to the staff with the commission direction that it
should be used. You know, it says that defense-in-
depth is the employment of successive compensatory
measures of prevention and mitigation.
It is something that does not wholly
depend on any single element of design construction,
maintenance or operation. It's something that if it
is done right would be more tolerant of failures and
external challenges.
MEMBER WALLIS: All of which needs to be
quantified in some way.
MR. KADAMBI: Certainly as we get into the
details of a given design, you know, these are the
sorts of things that we would seek to quantify.
Further in specific applications, relative
to the risk informing effort of Part 50, Option 3, the
staff further identified elements of defense-in-depth
that are dependent on risk insight and those that are
independent of risk insight.
Examples of those that are dependent on
risk insight are limiting initiating events, cold
damage frequency releases, et cetera, and achieving
the kind of safety function probabilities that would
be consistent with what needs to be achieved.
The elements that are independent of risk
insight are things like balance between prevention,
containment, consequences, the avoidance of reliance
on programmatic activities and, you know, adhering to
the principles of the general design criteria.
So, you know, these sorts of things that
we have done in the recent past give us, we believe,
a sufficient high level of guidance on how we can deal
with the concept of maintaining safety.
The next performance goal I'd like to
address is increasing public confidence. Again, I see
some significance to the vector that's associated with
the performance goal, that this is something that we
want to increase, and I believe that increase would be
relative to where we are on the current generation of
reactors.
MEMBER POWERS: Before you go on to public
confidence, could I come back to maintain safety?
MR. KADAMBI: Sure.
MEMBER POWERS: It seems to me that there
is an element missing here and it's born of the fact
that when you look for guidance from the regulations
that have transpired over the last 25-30 years, you're
looking at a fairly well established technology as far
as the material.
And yet, just minutes ago we heard a
presentation on where this relatively established
materials and technology is surprising those that have
tended and cared for it, because things break that
they didn't anticipate breaking.
Now, with some of these new concepts,
either the gas-cooled reactor, you're wandering into
an area of materials that you have relatively little
experience with, relative to the kind of support
infrastructure that you have with metals and things
like that. And I'm wondering if there isn't room in
this for some appeal to a generally conservative
design philosophy in your list of things that you're
looking at?
MR. KADAMBI: I think --
MR. BENNER: One of the things we have been
doing, and I can't speak for the staff in research,
but I'm sure they're doing this, the material staff
and NRR has been working with ASME to start talking
about what some of the materials challenges are going
to be for the licensing of future designs. And as far
as whether there will be, you know, that we'll try to
employ general conservatism overall or just in those
specific areas where we don't feel comfortable with
materials, I don't think we've gone that deep into the
infrastructure.
MEMBER POWERS: I think you need to
recognize no matter how much talking with ASME you do,
even they will not have the kind of database and
experiential base that you have with the metals that
you're using in the current fleet of reactors.
MR. BENNER: That's been a struggle is
trying to get materials data and Raj Pathania of
research has been working to get access to a graphite,
a radiated graphite materials database from IAEA as
one source, and obviously the staff's going to have to
look at the available data and determine.
MR. KING: I think it's important to
recognize that Fort St. Vrain was not the only
graphite reactor run in this country, that there was
a very large one up in the northwestern part of the
United States, and that they have quite a lot of data.
Unfortunately you can't get their kind of graphite,
so.
MEMBER POWERS: But I think what your point
is how do you compensate for the lack of data, whether
it's materials, graphite or anything else? In
principal, the way you do that is you develop a
monitoring program that maybe is more extensive than
you would if you had such data. I think it's part of
the research and part of looking at what we ought to
do in licensing these plants.
I think monitoring is one aspect to it.
I think holding people to a more conservative design
standard is another one. There's probably a trade-off
there someplace, but it's not a complete substitution.
A lot of what we're doing in the regulatory body
nowadays could be interpreted as a liberalization born
of the fact, you know, we have 3,000 reactor years of
operating experience with the current fleet and we
shouldn't leap to the belief that that liberalization
is applicable to some new design.
MR. KADAMBI: I think we agree.
MEMBER POWERS: I encourage you to
articulate that.
MR. KADAMBI: I hope we will as we go
through this. I think it's really part of what
underlies everything we have to say about how we will
proceed with the review.
Anyway, on the matter of increasing public
confidence, again we believe the advance reactor
policy statement did state that we should provide
stakeholders a timely and independent assessment and
that's one of the reasons we are trying to abide by an
aggressive schedule to bring out, you know, what the
current thoughts that the staff have are on this
matter. And also, we are of course being very careful
to provide opportunity for feedback and comment on the
part of all stakeholders at every opportunity.
Now in terms of increasing efficiency,
effectiveness and realism, we find many of the
elements that have been described in Reg. Guide 1.174,
again the white paper, and the Option 3 Framework to
be useful in proceeding. Reg. Guide 1.174 does talk
about using appropriately the exemption process.
Again, being consistent with defense-in-
depth philosophy, assessing safety margins very
carefully, paying attention to performance monitoring,
the safety goal policy, all these things come together
in terms of how the particular issues that were dealt
with in Reg. Guide 1.174 were dealt with, even though
that was really in the context of making changes to
the licensing basis.
In the white paper, there is reference to
performance based approaches and the staff has issued
high level guidelines for performance based activities
which in some matter may appropriately be applicable
in increasing efficiency effectiveness and realism.
The Option 3 Framework also stressed
defense-in-depth philosophy, and consideration of core
damage accidents. By the way some of these things,
you know, we just don't really know what the
equivalent would be in a Pebble Bed reactor, but at
least conceptually these are the things we would be
looking for, and of course, the stress on all the
different types of uncertainties that one should pay
attention to.
On the fourth performance goal of reducing
unnecessary regulatory burden, what we find is that
the commission has over the years stressed the rule of
the safety goal policy statement as defining -- I
mean, at least providing a basis for saying how safe
is safe enough? You know, of course this raises
policy issues some of which we will talk about later,
but in addition to this performance based approach,
where appropriate could mitigate some regulatory
burden by providing licensee flexibility, provided the
margins are maintained.
Now this picture you saw last month and we
wanted to make it available here just to provide a
basis for how we are viewing this effort of screening
of the regulations. What we'd like to stress is, we
will independently, you know, undertake the kind of
screening required, but the process itself, the logic
flow seems to be adequate to proceed.
CHAIRMAN APOSTOLAKIS: Now, let me
understand this a little better. I see under the
diamond that says "compare completely applicable,
partially not applicable, PBMR specific." So, where
is it that we're adding things, adding regulations to
replace the ones that aren't applicable, under PBMR
specific?
MR. KADAMBI: Correct, although they may
not be regulations. They may be regulatory
requirements through license conditions.
MR. BENNER: I think there are two things
that go in that box, and there are some places where
a regulation may not be applicable just on its face
because of the language, but there's an underlying
concern that drove the staff to move towards a
regulation.
CHAIRMAN APOSTOLAKIS: Which was my next
point. If you go up to the left, it says "function
level or intent of regulations." So when it says
"partially applicable" is it referring to the
regulation itself or its intent?
MR. BENNER: The Exelon approach talks
about partially applicable, as when you have a
regulation that has three parts and two parts are
applicable and one is not; whereas, there are like we
just talked about, applications that on their legal
face may not apply, but there's an intent there, and
how that gets handled as applicable or have to come up
with a PBMR specific license condition, we're still
working with OGC on that.
MEMBER WALLIS: I'm a little bit puzzled by
the logic flow here. I don't see how you can design
a PBMR if you don't know what the rules are. You sort
of assume the thing is designed and then someone
figures out how to regulate it?
MR. KADAMBI: Well, I mean the design, I
believe a case that's been presented to us so far,
which is at a relatively high level, it doesn't have
a lot of details in it, but it's based on certain
physical principles that will --
MEMBER WALLIS: So what can you change, how
it's operated or what?
MR. KADAMBI: I mean certainly at some
point, you know, if we don't have sufficient assurance
of some physical capability beyond the operational
aspect may have to be changed. But right now, we see
the PBMR specific box being populated entirely by, or
substantially by, you know, the right side of this
diagram. And so, you know, it's not that we don't
know anything about the PBMR, but we certainly don't
know enough to see --
MEMBER WALLIS: So a cynic might claim that
you're going to be confronted with an existing design
and you're going to be told "find a way to license
it."
MR. BRENNER: Well, part of the pre-
application review and part of the discussions we're
having with Exelon is to, as we get some design
assumptions, to determine what regulations are
applicable so that they can make changes to the design
now as opposed to when they come in during the
licensing phase.
MR. KING: That's the main purpose of the
pre-application phase to dig out these things up front
before they've gone through and spent a lot of money
to develop a final design.
CHAIRMAN APOSTOLAKIS: Part of the problem
with this approach though, which you have not said you
have accepted or whatever. I realize you're
presenting what Exelon has presented. It's something
I think Tom King also referred to yesterday. If you
just do it this way you may, you know, be gaining
perhaps efficiency in the short term, but in the long
term maybe not because now what do you do? Next time
you have an IRIS, then you do the same thing and now
you have IRIS specific, then something else.
Wouldn't it be better to try to think a
little more broadly and see what kinds of fundamental
principles and requirements we want to have for all
technologies, and then have the box that says this is
technology specific, not technology specific.
MR. BENNER: I think that's the goal of
where eventually we're trying to get to, because some
of the people on the working group looking at this
approach are also supporting and looking at Option 3
and advance reactor regulatory framework so that
hopefully in going through this, we're trying to keep
our eyes open for where, you know, because of
something that the staff and the applicant wrestled
with, does a principle come out of that that needs to
go up to the highest level. And also, have we
identified a gas cooled reactor specific requirement
that needs to go on the lower level, whatever it ends
up being, whether it be a regulation or a guidance
document.
VICE CHAIRMAN BONACA: It has to be high
enough and flexible enough that it doesn't prevent
innovation in designs. What I mean is I totally
agree.
MEMBER KRESS: Well, they have two
problems. One of them, Exelon has come in with this
as a proposed approach and they want to know what the
staff thinks about it, so they have to focus on that.
But how many times they've done, they can be thinking
about how to have general principles that focus their
attention on what they think about it.
VICE CHAIRMAN BONACA: Sure.
MEMBER KRESS: I think they're working both
sides of it.
MR. KING: And the two compliment each
other, because the PBMR illustrates a number of issues
that we're going to have to deal with on future
plants, containment issue, the emergency planning
issue, a number of these key issues. So, it's useful
to have the two going on in parallel because they do
cross fertilize each other.
MR. BENNER: We can go on. I think we've
really covered most of what was on the slide in the
discussion of the graphic. So, basically it's screen
the regulations, find out the applicable ones, apply
them, and final determinations will be made by the
regulator.
MR. KADAMBI: You have seen this before
also, and we just present it to refresh your memory on
it.
VICE CHAIRMAN BONACA: Yes. One question
I have on this. This clearly is not -- we discussed
it. It was not a frequency consequence curve in the
sense of integrating all of the results. It is
actually a very useful curve to determine initiators
and frequency that should be associated with them.
Would you plan to have also a frequency consequence
curve that provides the outcome? I mean, when you
integrate all of this?
MR. BENNER: In one of the rounds of
questions we had with Exelon we saw, as one of the
things they need to do, was to somehow sum up the
consequences of all the events, or at least show that
the events they've identified represent the majority
of the risk, and somehow compare that to the safety
goals.
CHAIRMAN APOSTOLAKIS: So how is this
curve going to be used? Can you explain to me -- I
mean --
MR. BENNER: We see this, really, as a
plot of the -- some of the high-level regulations that
are applicable to the PBMR. So it's useful to
illustrate what some of the bounds are.
MEMBER WALLIS: This is a continuous curve
and events are all discrete.
MR. BENNER: Discrete.
CHAIRMAN APOSTOLAKIS: But if I enter,
say, at .1 rem, right, on the horizontal axis, and I
go up and I hit the mean frequency of 2.5 10-2 per
reactor year, what does that say in terms of the
design? How am I going to use this? That's what I
don't understand.
MR. BENNER: I think the way the staff
sees this as being used as just -- we are going to
apply the regulations that they reference as they are
intended to be applied. This just provides a
graphical representation. Basically, we identify, you
know, events that need to be designed for in the
design basis region, and they meet -- need to meet
10 CFR 50.34.
So there -- you know, that's why we don't
really consider this a frequency consequence curve,
because that regulation was designed to apply to
discrete events.
VICE CHAIRMAN BONACA: I mean, they were
very clear that this was the equivalent of what was
used in the FSARs when you set the categories, you
know, anticipated transients and LOCA, and so on. I
mean --
CHAIRMAN APOSTOLAKIS: But that doesn't
make it right.
VICE CHAIRMAN BONACA: No.
CHAIRMAN APOSTOLAKIS: Because, again, you
know, I have a lot of flexibility in defining the
events. That's the problem. If you say -- you make
it workable if you say all the events that may lead to
.1 rem or greater must have a frequency smaller than
the bound, then it makes sense. So --
MEMBER ROSEN: All the events individually
or summed?
CHAIRMAN APOSTOLAKIS: All together.
MEMBER ROSEN: All of them, if you sum
them up --
CHAIRMAN APOSTOLAKIS: If you sum them up,
their frequency should be less than 2.5 10-2.
MEMBER ROSEN: Well, that's what this
curve says.
CHAIRMAN APOSTOLAKIS: No, no.
VICE CHAIRMAN BONACA: We're talking about
individual it will be the frequency. This is
individual. This is like saying, for example --
CHAIRMAN APOSTOLAKIS: Pick one.
VICE CHAIRMAN BONACA: -- assume you apply
this to the current set of reactors. You would say
that a LOCA, which has a frequency of 10-4 cannot
have, you know, a dose higher than -- and you go to
the right and you find the dose limit for the LOCA.
Okay? All LOCAs individually have a dose higher than
that.
MEMBER ROSEN: So if Exelon says it's --
you take it event by event, and George has just
postulated that different one, which is to sum all of
the events, it should be --
VICE CHAIRMAN BONACA: That's why I asked
the question -- the first question about a frequency
consequence curve that would, in fact, integrate.
CHAIRMAN APOSTOLAKIS: That's right. But
that was the original intent of the Farmer curve. It
was confused and the interpretation was not clear for
a number of years.
MR. BENNER: We're looking at that two
ways. One is --
CHAIRMAN APOSTOLAKIS: By the way, it will
not be -- it doesn't have to be the same curve you are
showing there if you interpret it correctly. You may
have --
MR. BENNER: And that's one of the things
we brought up with Exelon about the safety goals
dealing with the integral. The other thing we're
talking about is maybe integrating somehow the area
under this curve to see where that compares to the
safety goals, so that while these will be a
representation of the current regulations, it will
also give us some guidance as to, you know, maybe for
the future for developing a frequency response curve.
VICE CHAIRMAN BONACA: I want to say
actually that the existence of this curve -- it's
confusing.
MEMBER WALLIS: Do you rely on Exelon to
determine what the events are, or are you guys going
to tell them what the events are they must consider?
MR. KADAMBI: Well, we will take full
liberty to, you know, make sure that the events we
think are important are considered, along with
whatever that Exelon proposes.
MEMBER WALLIS: Are you going to imagine
all of the scenarios and then tell them they ought to
consider these?
MR. KADAMBI: In some ways, that's part of
what goes into it is to imagine the scenarios and to
-- to make sure that --
CHAIRMAN APOSTOLAKIS: Which means to do
a PRA, right?
MR. KADAMBI: Well, yes. Eventually,
that's what it will all lead to.
MEMBER KRESS: Basically, I think in your
Chapter 15 for LWRs, you have a set of design basis
events. And these are things like you have holes and
you lose coolant or you inject a rod or you have an
ATWS event or loss of power, and various sets of
events that have been chosen for -- to look at as
design basis events.
Now, I viewed this as an attempt to decide
what those events would be for the Pebble Bed Modular
Reactor. And you would start out by using similar
type of things, I think. Overall grand events like a
-- like --
CHAIRMAN APOSTOLAKIS: Let's not forget,
though, that even for the design basis events, after
we did the PRAs, we did extra things to mitigate, I
think, consequences, didn't we? The Level 2 stuff?
MEMBER KRESS: Well, you had --
CHAIRMAN APOSTOLAKIS: I mean, you can do
the same thing here.
MEMBER KRESS: You had principles like a
single failure principle, and then you had to meet
certain figures of merit. And then, you know, there's
questions about what --
CHAIRMAN APOSTOLAKIS: The station
blackout was not a design basis event, was it?
MR. KING: No. That came because of PRA
insights.
CHAIRMAN APOSTOLAKIS: PRA, that's what
I'm saying.
MEMBER KRESS: Yes. But now that we know
that it was an important thing, we might look at it
for this.
CHAIRMAN APOSTOLAKIS: No. The point is
that, again, curves like this -- I mean, we have a lot
of discussion in Regulatory Guide 1.174 about it
should not be taken as absolutes with bright lines,
and so on.
MEMBER KRESS: Right.
CHAIRMAN APOSTOLAKIS: And all of that
stuff. But I think that the question, how do you
determine the so-called design basis events, or
whether you actually need that, is a good one. It's
an important one. One way of doing it is what Dr.
Kress just suggested. But there may be other ways,
too, or a combination of ways.
VICE CHAIRMAN BONACA: One thing I want to
note is that to some degree that curve that we liked
last time at the meeting creates confusion, because
what I --
CHAIRMAN APOSTOLAKIS: No, I didn't like
it.
VICE CHAIRMAN BONACA: No. Because we all
understood -- I understood there was, at the
beginning, a frequency consequence curve and then we
discovered it wasn't.
CHAIRMAN APOSTOLAKIS: It was not.
VICE CHAIRMAN BONACA: Typically, the
expectation for it is to be that, and the fact that
this is already on paper leads one to believe that it
has been developed and it hasn't been developed.
So --
MEMBER KRESS: But I think if you use that
kind of curve to select design basis events in a
particular way, and then you go to frequency
consequence curves as your figure of merit --
VICE CHAIRMAN BONACA: No, I understand.
MEMBER KRESS: -- to see if -- see if
these things are -- the design part of these events
are acceptable.
VICE CHAIRMAN BONACA: I completely agree
with you. I'm only saying that the fact that it was
presented almost made everybody feel, oh, we got it.
And, well, we haven't got it.
MEMBER ROSEN: Well, I see the value of
this is that it avoids the arguments about when
someone postulates a new event, someone typically
reacts, "Well, that's well beyond the design basis.
We're not going to think about that."
MEMBER KRESS: Yes.
MEMBER ROSEN: And that -- this gets away
from that completely.
MEMBER KRESS: This -- yes, that gives you
a way to --
CHAIRMAN APOSTOLAKIS: But it has to be
right. It has to be -- the concept of having the
curve does that.
MR. BENNER: And I think that's a good
lead-in to our concern about licensing basis events,
that we do --
CHAIRMAN APOSTOLAKIS: By the way, have
you seen what the Dutch have done along these lines?
They had passed a regulation I think in the late '80s
where they actually used frequency consequence curves
in regulating nuclear and chemical plants. And then
a few years later somehow they got out of it or part
of it for some practical reasons that came from
experience. So that probably would be another source
of information for you.
MR. KADAMBI: We'll make note of that.
MEMBER WALLIS: To get back to Dana's
point about conservative design, in trying to license
water reactors there was a lot of experimental
evidence in terms of LOF tests, and so on, and scale
tests of this, that, and the next thing, because we
didn't rely just on calculating things.
And the impression I get is that this is
supposedly a more calculable system, so you don't need
to do a whole lot of these tests?
MR. BENNER: No, I think in -- in some of
the documentation back to Exelon that we're struggling
right now with what level of testing is going to be
necessary to validate any assumptions they may be
making.
MEMBER WALLIS: So is there data on
combustion of piles of graphite balls in various
configurations, and so on? All of that stuff is
there?
(Laughter.)
MR. KING: That's part of the -- part of
what we're trying to do now is gather what information
is out there that supports graphite behavior, fuel
behavior, high temperature materials behavior, and
then some judgment is going to have to be made on what
the licensee or applicant has to do to expand that to
apply to his design and what we, NRC, want to do to
have some confirmation of what the licensee is telling
us, or to push the margins and see where the cliffs
are.
That's the thing we talked about yesterday
when I said for the HTGR we're developing a research
plan over the next several months, and we're going to
come to the committee hopefully in February and talk
about that.
MEMBER WALLIS: But if it's something like
the LOF tests, then this thing is dead in the water
because there's not that much money around from the
agency and the licensees to do those LOF-like tests.
MR. KING: Yes. But maybe you don't need
a LOF-like test for a helium reactor. You know, maybe
you need some smaller scale phenomenology-type test,
fuel performance, graphite performance. But those
questions are still on the table and are going to have
to be worked out.
Part of the pre-application review is to
try and settle those things up front. So the licensee
or the applicant knows if we're expecting a large-
scale test on some phenomena, that he knows that now.
He doesn't know that, you know, two years after he
submits an application.
MEMBER LEITCH: This mean frequency curve
has units of per plant year. Do you know if that --
by that nomenclature is it meant a module?
MR. BENNER: No, it's up to 10 modules.
MEMBER LEITCH: Up to 10 modules.
MR. BENNER: That's how they're planning
on --
CHAIRMAN APOSTOLAKIS: So you are
automatically, then, doing what I suggested at the
workshop that Exelon didn't seem to like. You are
reducing the goals per unit. Are they aware of that?
MR. KING: It's their proposal. They came
in and said that's the way they're designing.
CHAIRMAN APOSTOLAKIS: Did they make it
clear this is the plant or the unit, the module?
MR. KING: No. Exelon said, "We are
considering a plant is up to 10 modules." And,
therefore, for an individual module --
CHAIRMAN APOSTOLAKIS: Ah, for an
individual module the curve --
MEMBER KRESS: -- risk goals will be, you
know, a factor of 10 less.
CHAIRMAN APOSTOLAKIS: But that's not what
Eric just told us. This is for 10 --
MR. BENNER: I thought they flipped it
around, that these would be applicable to --
CHAIRMAN APOSTOLAKIS: Ten modules.
MR. BENNER: -- 10 modules.
MR. KING: Yes, for common -- where
there's common cause failures, you know, that they
would expect all 10 modules to meet that. Right.
CHAIRMAN APOSTOLAKIS: As a group.
MR. KING: As a group.
CHAIRMAN APOSTOLAKIS: So, in essence, for
-- on a per module basis, we're going down, the goals
go down. Right?
MR. KING: Yes. Certainly things like
CDF, if we can define a CDF for this, it would go
down.
MEMBER ROSEN: It would seem to me
irrational to do anything else. I mean, you have 10
nuclear reactors there.
CHAIRMAN APOSTOLAKIS: That's right.
MR. KING: Yes.
CHAIRMAN APOSTOLAKIS: The problem is
still -- they can come back and say, "Look, we may not
build all 10 of them," so there may be a period of
many years where there will now be two. But, again,
you know --
MEMBER ROSEN: So you have two critical
reactors on the site. But you have to -- you can't
say that we really -- if you have seven, that you
really only have one. That's irrational.
MEMBER POWERS: Yes. But I guess that
raises the question, suppose I have one, but I may or
may not have 10 in the future. Does that mean while
I have one I can run that sloppy, and I only have to
clean my act up when I start adding other modules?
(Laughter.)
MEMBER KRESS: That's a rationalist
position.
(Laughter.)
MEMBER WALLIS: The tenth one has to be
absolutely perfect. Pristine.
MR. BENNER: I actually think we've
covered most of this slide.
CHAIRMAN APOSTOLAKIS: Yes.
MR. BENNER: To the concern about being
able to raise up additional --
CHAIRMAN APOSTOLAKIS: Is licensing basis
events a new terminology?
MEMBER KRESS: It's like design basis.
CHAIRMAN APOSTOLAKIS: I know. But it's
new.
MR. BENNER: And it's supposed to cover
the spectrum, both above and below design basis.
CHAIRMAN APOSTOLAKIS: Well, I mean, it's
everything.
MEMBER KRESS: That's right.
CHAIRMAN APOSTOLAKIS: You have already
assumed, then, that we need licensing basis events?
I mean, there would be other ways.
MR. BENNER: Well, I think some of that
gets into -- for licensing of the first PBMR using the
current regulations as -- as the scheme. That seems
to be an inherent part of how the regulations work.
CHAIRMAN APOSTOLAKIS: In any case, the
licensing basis is broader, so you have a home there
to add that in.
MEMBER KRESS: If you had an FC acceptance
curves you could use the whole spectrum of sequences
as your --
CHAIRMAN APOSTOLAKIS: On the other hand,
there is a lot of value to the --
MEMBER KRESS: Oh, yes.
CHAIRMAN APOSTOLAKIS: You make it --
MEMBER KRESS: It really helps --
CHAIRMAN APOSTOLAKIS: It helps everybody
focus.
MEMBER KRESS: -- focuses a lot of --
CHAIRMAN APOSTOLAKIS: Okay.
MR. BENNER: I think -- there is where we
talk a little bit more about what would the licensing
basis for the PBMR be, and we see it as the set of
requirements that apply to all of your safety-related
structure systems and components.
One of the concerns we see in trying to
assess the acceptability of the design is it relies a
lot on inherently reliable passive components. And
you get into the concerns of what special treatments
do you need to make sure that, you know, those
inherently reliable components retain the quality that
you're assuming of them.
Another aspect that came up in the design
certification reviews was the concept that, you know,
you would have a lot of these passive systems, which
would be what you would rely on to mitigate an event,
but you would also have active systems which would
provide a defense-in-depth function. And the
Commission provided some guidance on how the staff
should treat those -- those components.
One of the things that Exelon proposes,
which the staff is struggling with, is component-level
special treatment. And in the development of the
special treatment work done on South Texas Project,
the staff found that that was virtually impossible.
That you could look at functions, you could look at
system requirements, but to try and bring that down to
the individual component level was very difficult.
Now, we're not precluding that. It may be
able to be done for the PBMR. We're just raising that
as a potential challenge.
VICE CHAIRMAN BONACA: Although I think
the considerations that Dr. Powers mentioned before
very much apply here, that, you know, for a lot of
these materials maybe you have to be more than
conservative.
MEMBER ROSEN: Well, I think in the case
of South Texas, the argument was that we had a lot of
performance-based information about equipment. And as
Dr. Powers said, we have a lot of new equipment here,
and with no history or very little history. So the
conditions are different.
CHAIRMAN APOSTOLAKIS: What is the third
bullet, defense-in-depth also provided for non-safety-
related SSCs? What does that mean?
MR. BENNER: That was a concept that was
really brought up in the review of the AP600, where to
call it a passive design, they wanted to identify only
the passive systems as safety-related.
CHAIRMAN APOSTOLAKIS: So that was the
regulatory requirements for non-safety-related --
MR. BENNER: Exactly. That's exactly it.
CHAIRMAN APOSTOLAKIS: And we are keeping
the terminology of safety-related versus non-safety-
related?
MR. BENNER: I think at this point, yes.
MEMBER ROSEN: Why?
CHAIRMAN APOSTOLAKIS: But that's a
problem now. We are keeping everything from the
existing regulations that in another arena we're
trying to get rid of. I realize the problem you are
having, but at least say that it's something you're
going to think about. I realize that, you know, you
really have to start somewhere and proceed.
MR. KADAMBI: And also, this is part of
what Exelon has proposed. You know, they have kept
the terminology, and they are defining it in a certain
way, and we are looking at it in the context of, you
know, can it be applied in the context that they
are --
CHAIRMAN APOSTOLAKIS: Well, is there a
methodology in the existing regulations for defining
what is safety-related?
MEMBER KRESS: Sure. It's what's needed
to meet the figures of merit for the design basis --
MEMBER POWERS: But, I mean, this does
strike me as one of the areas I -- this is going to
pain me to admit -- where -- that's the strength of
the probabilistic-type analysis methods, were you can
look at the plant in some sort of an integrated
fashion and escape the problems that arise when you
look at it in train-by-train fashions that it was done
in the past. It was done by necessity in the past.
It seems to me that this is one area that
the -- where the rationalists and the structuralists
really agree on this is this -- in the categorization
of equipment is where the PRA really has a strength.
MEMBER KRESS: And like they say, there's
a problem doing that down at the component level.
MEMBER POWERS: Because it'll show up in
the PRA.
CHAIRMAN APOSTOLAKIS: I must say, though,
that I'm very happy that Dr. Powers has not entirely
lost his ability to reason rationally.
(Laughter.)
MEMBER ROSEN: He is a structuralist with
some rational tendencies.
CHAIRMAN APOSTOLAKIS: Flashes of
rationalism.
MEMBER ROSEN: Flashes, yes.
(Laughter.)
Whereas I am a rationalist with some
structuralist tendencies.
MR. KADAMBI: We have tried to identify
some potential policy issues. At this stage, it's a
little premature to present this to the Commission as
policy issues. We're still working on them. We have
to keep in mind that this is the review that we did of
a submittal that we got in August -- on August 31st,
and we are still, you know, developing information.
But the sorts of things that will
certainly lead to policy issues are the use of
probabilistic criteria, where maybe they have not been
used before.
MEMBER KRESS: Haven't we always done
that, though?
CHAIRMAN APOSTOLAKIS: For seismic, I
think we have. We've always --
MEMBER KRESS: I thought we've always done
that, even with the fully structuralist deterministic
process.
MR. KING: Well, we've done it
qualitatively. I mean, everybody has sort of had, you
know, something qualitative --
CHAIRMAN APOSTOLAKIS: But that -- with
one PRA, I mean --
MR. KING: But, remember, what's being
proposed are some quantitative frequency guidelines
that you would decide what's in and what's out for the
design, and that's going beyond what we've done in the
past.
Now, maybe we've done it probably in a
couple limited situations. We've put some numbers in
a Reg. Guide or a standard review plan. But this is
across the board, everything.
MEMBER POWERS: But we always defined
accidents and scenarios based on whether they were
credible or not. And some people I think define
credulity with a number, but the fact is we've always
used risk as a --
VICE CHAIRMAN BONACA: But one thing I
wanted to say is that it is also true that there are,
for example, events -- and I think it's important for
the presentation we had this morning, like rodejection
accident. At some point, it was a true proposal on
the floor for years to eliminate it because it cannot
happen.
Well, I mean, we are not looking now at
cracks in CRDM nozzles, and suddenly, you know, this
exercise in reactivity insertion becomes more
credible. So I think -- I think it's a dilemma
they're facing. Sure, I mean, because at some point
in time you might find something like this so unlikely
that you just don't consider it.
CHAIRMAN APOSTOLAKIS: But the thing is --
I think Tom is right. I mean, if you say go with
option 3, which gives you high-level frequencies for
core damage for LWRs and protection, and so on, then
what do you do with that? I mean, you have to keep
going down to the design, right, and have some
quantitative criteria or goals or guidance -- not
criteria -- guidance for function level on
availabilities, and so on. And I think that's fine.
I mean, it is something that the Commission should
explicitly comment on and decide on.
MR. KING: But see, option 3 is looking at
an existing set of regulations and hopefully approving
them. This is taking a clean sheet of paper and a
design and saying, "I'm going to use probabilistic
numbers to decide what we'll design for and what we
don't design for." That goes beyond where we've used
probabilistic information in the past.
CHAIRMAN APOSTOLAKIS: Yes. But option 3,
even for LWRs, I mean, it just says the core damage
frequency should be this. But, I mean, if you wanted
to design in your reactor -- a new LWR with that, you
will need some guidance how far down to go with these
numbers.
MR. KING: And you could take the option 3
guidelines and design a reactor to those. The policy
question I see for the Commission is, do we want to go
that far in applying probabilistic -- quantitative
probabilistic guidelines?
CHAIRMAN APOSTOLAKIS: Yes.
MR. KING: And what other factors do we
want to bring into this decision?
CHAIRMAN APOSTOLAKIS: I think it's going
to be a combination. There's no question about it.
It's not going to be just the numbers.
MEMBER ROSEN: I heard an answer from
someplace over in the rational corner here. But I
want to -- the question I have is about the process.
Are we going to weigh in as ACRS on these questions?
MEMBER KRESS: Oh, probably eventually,
maybe not right now. We're already weighing in a
little bit. But as I see it, Tom, the -- we -- these
policy issues come head to head with our -- what our
definition, and I consider it a limited definition, of
defense-in-depth.
That's where there is a collision here,
and the question is, if you just use pure
probabilistic criteria, you have to decide on where
defense-in-depth fits into that choice to making these
decisions. I think that's the issue, and I think it's
going to require a little bit different thinking than
on what defense-in-depth is, more than what's in 1.174
and more than what's in option 3.
MR. KING: Yes, I agree with you. And at
this point, we're not asking for the Committee to
weigh in on these policy matters, because we haven't
reached a position yet on it. Just for information,
you know, that these are coming down the road at some
point.
CHAIRMAN APOSTOLAKIS: I think the first
bullet, though, might scare people.
MEMBER WALLIS: I don't understand the
first bullet. Do you mean --
CHAIRMAN APOSTOLAKIS: Why don't you say
use the risk-informed approach to selecting licensing
basis events.
MEMBER WALLIS: But you first have to
analyze a whole lot of events. Then you --
CHAIRMAN APOSTOLAKIS: Well, the
implementation is somebody else's problem. But here
they talk about criteria, which I think is going to
scare everybody. It would be a risk-informed approach
to selecting the licensing basis events, right?
That's what you are talking about.
MR. BENNER: Well, and I think that's part
of our concern is that we see in the approach being
proposed that it -- that it does have the use of
probabilistic criteria --
CHAIRMAN APOSTOLAKIS: Yes.
MR. BENNER: -- where we're saying in the
risk-informed approach idea that, well, you need to
have some deterministic portion that --
MEMBER KRESS: Yes. And my point about
defense-in-depth needing a new interpretation had to
do with if you just look at the, say, frequency
consequence type thing to get a selected event that
you design against, I think you have to look very
strongly at the uncertainties associated with each
frequency, its contribution to the uncertainty. And
you can't just use the number as is, and that's where
you go a little deeper in your defense-in-depth
theory.
MR. KING: Yes, I agree. What confidence
level do you want? And, you know --
VICE CHAIRMAN BONACA: Well, and there has
to be the balance there, because, I mean, again, I
want to say that for extreme events that often times
one could say you can't rule it out, the -- the
assumption of those events, in the current generation
of plants, led to -- to the design of certain
parameters that served us very well for more probable
events.
I'm trying to say, for example, all of the
characteristics that we implemented in -- for negative
coefficient reactivity, and so on and so forth, there
were results of some evaluation of extreme events that
we thought were very improbable or unlikely and maybe
couldn't happen. But the fact that you assumed it
purely for the purpose of certain characteristics of
the reactor, that it served you well for less severe
accidents that are more probable.
So I think it's a very -- it's a very
critical area, this one here.
MEMBER ROSEN: But you will acknowledge
the trap in that as well.
VICE CHAIRMAN BONACA: Of course, I'll
acknowledge the trap. I'm only saying that in design
-- when you design something, you don't assume
necessarily only what is going to happen. You assume
you bound certain conditions so that you can define
coefficients and certain characteristics of the plant
that will serve you well for less severe but more
probable events, and --
MEMBER ROSEN: But never assume the vest
of invincibility simply because you have --
VICE CHAIRMAN BONACA: Of course.
MEMBER ROSEN: -- created some design
basis events, and, therefore, everything less than
that is okay. We know from bitter experience that
that's not true, that you have to go event by event
and think -- and think each event through.
VICE CHAIRMAN BONACA: Well, and say that
you do that, and --
CHAIRMAN APOSTOLAKIS: Right. Let's come
back to these fellows. Anything else you would like
to say?
MR. KADAMBI: I think we're about ready to
wrap it up.
CHAIRMAN APOSTOLAKIS: Very good. Have
you wrapped it up?
MR. KADAMBI: Well, unless there are any
questions, I believe that we are done.
MEMBER WALLIS: My feeling is all of this
sort of theoretical. It's an approach, and I think
that if you start looking at specific events which are
unique to this particular system, like if we're
getting into the compressor and there's a loss of
compressor accident, what happens, see whether you can
handle it using some of the ideas you've got here, or
whether you come up against something else which you
haven't thought of that you have to think about at the
fundamental level.
MR. BENNER: And that's what we've tried
to feed back to Exelon, that we may not need a
complete design to start exercising this approach, but
if we wanted to take some -- you know, some select
event and determine what --
MEMBER WALLIS: There's things having to
do with water reactors that you have to think about,
just see how far you can go. If you try to use --
apply some of these things, you might find --
MR. KADAMBI: The only thing I would add
is that this exercise was, you know, put into practice
for the HTGR to some extent. You know, they did go
through what they call bounding events which were not
proposed by DOE at the time. And, you know, it
provided sort of a check of how well did the design
accommodate or take -- tolerate these things.
So that's -- Mr. Chairman, that's our
presentation.
MEMBER KRESS: Well, I think --
personally, I don't see the need for a letter at this
time. I think we've made -- given some feedback with
our comments, and I don't know having them in a letter
would be any more useful to you.
So I look forward to continuing the
conversation with you as you get further along in
this. I personally think you're at least asking the
right questions. And so --
MR. KADAMBI: Thank you very much.
CHAIRMAN APOSTOLAKIS: Thank you. We will
start the interviews with the candidates half an hour
from now, 12:40.
MEMBER WALLIS: George?
CHAIRMAN APOSTOLAKIS: Everything has been
shifted.
MEMBER WALLIS: George?
CHAIRMAN APOSTOLAKIS: Yes.
MEMBER WALLIS: Could you tell the
membership that the staff is going to --
CHAIRMAN APOSTOLAKIS: Yes. Well, go
ahead. Go ahead.
MEMBER WALLIS: I had a telecon with
Exelon and GE about this business of the distribution
of flux and fuel, and so on. And it turns out that
it's a real jungle. It isn't the simple matter at all
to say there isn't really flux flattening. There's a
very, very complicated fuel management program
necessary in order to meet the regulations. It's not
-- there's no simple explanation that you can put in
a few sentences and put in a letter.
And the staff agreed to come at 3:00, and
in 10 minutes or so tell us how, with all of this
complicated stuff going on, can they assure themselves
that the core uprate -- this power uprate is okay.
CHAIRMAN APOSTOLAKIS: And also, I remind
you, the Commissioner is coming at 2:00. That's why
we have this flexibility.
MEMBER ROSEN: He is pushed back a half
hour.
CHAIRMAN APOSTOLAKIS: Yes. His assistant
was here. He heard us talk about it, and I said
absolutely. He called the Commissioner, and he said,
"I'll come half an hour later." So the ACRS will have
-- Dr. Ford will have more time to run over time.
(Whereupon, at 12:14 p.m., the
proceedings in the foregoing matter went
off the record for a lunch break.)
CHAIRMAN APOSTOLAKIS: Can we start before
2:00?
MR. MCGAFFIGAN: Is Graham Wallis still
here?
CHAIRMAN APOSTOLAKIS: He's coming.
MEMBER POWERS: Yes, George. You may
start before 2:00.
CHAIRMAN APOSTOLAKIS: We are back in
session. We are very pleased to have Commissioner
McGaffigan with us for an exchange of views.
So we'll let you go first, Commissioner.
MR. MCGAFFIGAN: Well, thank you. You can
see I've shown up with a blank piece of paper, so I
don't have long remarks to make. I'm here more to
play a reverse role. We get to sit in the Commission
meeting room and ask you questions, so I'll make
myself available so you can ask me questions.
Hopefully I will do as well as you guys do before us.
I will preface it. I just heard Dana
talking about I guess some of the discussions you had
this morning about cracking and control rod drive
mechanism housings and whatever. There are some
issues that are going to come before us I am not sure
you are fully on top of. That is whether we are going
to shut down some people early by order. Davis-Besse
was before you. They are probably the one with the
most at stake. There's one other plant that staff
hasn't reached resolution with. Earlier we were
arguing with some plants over ridiculously small
numbers of days, but I think in Davis-Besse's case we
are arguing about 100 days or so compared to when they
would like to go into their outage, and we might force
them into an outage.
I am not an expert about control rod drive
mechanism cracking. I do know that we have some
additional -- just orally, we have some additional
data from TMI in recent days. All the cracks were
axial, not circumferential. They were axial. They
are just going to be ground out, and that is going to
be it.
I don't know where things stand, to be
honest with you, except that the staff is working on
this sort of two-parallel path process. One, trying
to negotiate with these two licensees who remain in
their bullseye. The other, drafting contingency
orders for ordering early shutdowns to do inspections
that they feel are appropriate, and I think we all
would like to have inspection results.
It's a fairly profound question, whether
you order somebody down early. So if you all have any
comments for me on that or for the Commission as a
whole, nothing is going to happen until the end of
December. It would be interesting for us to hear
those comments and understand your perspective. This
is a sort of hard technical issue that's exactly up
your ally.
MEMBER POWERS: Certainly the Davis-Besse
folks came in and made a probabilistic argument to us
today on why they should not be. I guess I look at it
a little bit and say well, OConee ran for at least
half a cycle with these cracks. Why do I want to
worry about 90 days, or maybe it's 120 days?
MR. MCGAFFIGAN: As I understand, again,
I have heard these arguments. The staff argument is
that in an ideal situation, they would have liked to
have already had the data. So it is a matter of
giving them an extra 90 days beyond where they feel
comfortable.
As I understand it, the issue has to do
with the rapidity with which the circumferential
cracks might grow. There is a range of possible
numbers. The staff is taking a very conservative
number in the absence of data. They admit, and I
think it's just true that there isn't a lot of data.
I think they also, as I understand it, would like to
have some data. What licensees are doing is they find
anything, they are just repairing it and getting on
with it rather than stay in outage to do science
projects.
MEMBER POWERS: My initial reaction is to
say their number isn't all that conservative. It's
what I would have picked. Then I realized I am very
conservative so --
MR. MCGAFFIGAN: As I say, it's an issue
that I think as you guys think about this stuff
theoretically, there is this -- the way the staff
process works is if they don't get to yes, with the
two licensees, they will come to us with an order and
we'll have sort of five days of negative consent on
which to think about whether we want to go along with
that order, because an order is an enforcement action.
For high visibility enforcement actions, that is the
process that we -- so there isn't a Commission vote on
it unless the Commission chooses to weigh in. I doubt
the Commission would weigh in unless there were strong
technical arguments on the -- you know, that were
presented to us as to why the staff was being overly
conservative and we could wait the extra 90 days or
100 days, whatever number of days we are arguing
about.
I don't necessarily want you to tell me
the answer right now. It's just of all the technical
issues -- you know, most of the other issues honestly
before us at the moment start and end with security
and physical protection. Everything else is done on
the side as we have time, but we are fairly consumed
with the September 11th follow-up. I don't know
whether the staff has talked to you about -- and much
of that is appropriately classified and we wouldn't be
talking about it in this room. I think you obviously
all have clearances. I think there could be a role
for ACRS in thinking about design-basis threat
adjustments and other things. You know, obviously if
we were to do rulemaking with regard to enhancing
security at say spent fuel pools and ISFSIs, you all
see rulemakings and can comment on them.
MEMBER KRESS: We do have an interest in
this of course. We have one of our staff members
taking part in most of the activities that are going
on. He is going to brief us later today I think.
MR. MCGAFFIGAN: In a closed meeting?
MEMBER KRESS: In a closed meeting. Then
we will of course have some subcommittee meetings
later on.
MR. MCGAFFIGAN: Yes. We are trying to
figure out how much funding we have in this area. We
have made a submittal to the Office of Management and
Budget for some of the funds that are at the
President's discretion. We will need some significant
funding. There have been provisions.
One of the bills in the House, Heather
Wilson has a provision that would require us to do
something we would have done anyways which is,
provided we have adequate funds, and even if we don't
we'll reprogram some, to do some vulnerability
assessments using the National Laboratories. Sandia,
Los Alomos and Livermore, are probably the three labs
we'll primarily use for those analyses. They have
done work in the past for us, but they have also work
for other sectors.
Sandia did some interesting work for the
Gore Commission with regard to transportation, looking
at the transportation system as a whole and how
security could be -- airline transportation system,
and how security could be enhanced.
So we are -- I think that is an area where
-- I'm sure I'm not telling you anything, but that
takes up most days. We get about ye-thick set of
documents to read, some of which are classified, some
of which aren't, all of which are sensitive. We plow
through that. The Chairman is in charge if it's fast
moving, but if it's slow moving, the Commission as a
whole deals with the issue. As I say, we are looking
at everything. Everything is on the table.
Although I do want to say, and I'll say
publicly, I mean I have had debates with Paul
Levanthal sort of privately. I believe that I am
proud of the security system that we had in place on
September 11th. I think it provides the highest
standard of security of any private sector enterprise
in America and appropriately so because there are
significant risks. There is no chemical regulatory
commission that looks at the petrochemical plants and
has requirements for security that are inspected by
chemical regulatory agency staff, and there are no on-
force exercises, and none of the apparatus that we
have in place is in place for much of the rest of the
infrastructure. It is quite clear that you can get
catastrophic consequences in industries other than the
nuclear industry.
That said, that we had a very, very high
level of security on September 11th, and we have
maintained it and obviously enhanced it, we are going
to have to take into account the events of September
11th and looking at security in the future.
One thing that we have stressed since
September 11th is that we have to think in terms of
what's appropriate for private security forces to do,
and what is appropriate for the Federal Government to
do. If it makes sense to defend against air threats,
diving commercial airliners, and I'll get back to
that, then that clearly is not something that we are
going to ask licensees to do.
I do think we have to worry about
commercial airliners, but I think given that we have
been attacked by four of them, but the way to handle
that is with enhanced security on the airplanes
themselves, the cockpit door reinforcement that has
been talked about, enhanced security at the airports
to prevent people getting materials onto the planes
that could be used to attack the pilots and take over
the plane, additional air marshalls, all of those
sorts of security.
The notion of having -- I mean Paul
Levanthal for a month, until I talked to him on
October 18th, has been calling for the deployment of
surface -- not surface-to-air missiles, of guns, of
anti-aircraft guns. I informed him on October 18th at
our public meeting that the U.S. military doesn't have
guns. I mean the last guns I was aware of was in 42
dusters that were in the New Mexico National Guard
while I was working for Senator Bingaman. Div-ad was
cancelled. In 1985, I was working for Bingaman then
too. It would have been a hell of a gun. But all of
our air defense forces are designed for the clash of
tank armies on the battlefield in Germany as the
Russians try to pile through the Folda Gap and we try
to stop them. They were not designed for defending
the sort of 63 nuclear power sites.
So he has now amended it. He has figured
out that we only have surface-to-air missiles, so he
is now correctly saying that he wants -- really
correctly in the sense of what's available in the
American inventory. He wants surface-to-air missiles
at the 63 sites.
I try to explain to Paul the command and
control issues involved in having 63 surface-to-air
missiles. I also try to explain to Paul the mal-
deployment of scarce defense resources that would be
involved in trying to have -- having 63, you know,
taking essentially all the Hawks and Patriots we've
got and sticking them around a bunch of nuclear power
plants that aren't particularly vulnerable.
Now you get into this issue of -- you
know, we have not except at Seabrook we looked at FP-
111 going into Seabrook because at Pease Air Force
base nearby there were large numbers of FP-111s
stationed until they were taken out of the inventory.
We looked at Harrisburg, at TMI we looked at 707s. So
we have not done an analysis. We are clearly working
on it, an analysis of what happens when a large
commercial aircraft crashes into a containment dome or
elsewhere in a nuclear power plant.
But you know, there's at least a pretty
decent chance as I agree with Intergy, I think was in
today's New York Times, is a pretty decent chance that
these plants would survive famously. Number one, it
would take a hell of a pilot to hit some of these
plants. They are not the World Trade Center. Number
two, the main thing, the Germans and the Swiss do have
regulations. They looked at things like FB-111s.
What was it, the old, I think it was F-104s. Yes, the
F-104s Starfighters just routinely dropped out of the
sky all through the 1970s. People decided they better
look at what happens when Starfighters hit them at
high speed. They look at speeds like 700-odd
kilometers an hour. It's the engine, whether it's a
jumbo jet or it's a fighter plane that's the key thing
that might penetrate.
Even if it penetrates containment, it
isn't clear that you have an accident yet that you
can't recover from. So what has happened, thanks to
Paul Levanthal and others, is this notion that if you
hit anywhere, you have certainly got a catastrophe.
You know, they don't take into account the defense and
depth of the plants. You know, we can't rule out,
just we can not provide perfect assurance, perfect
safety to the American public. But we have provided
a very high level of safety. We will continue to
provide a very high level of safety. A lot of the
issues that some of these members of the public are
trying to trot out are in my mind the sort of worst
national security analysis one could possibly come up
with. You know, they are suddenly becoming national
security experts. If you follow their prescription,
you will do a disservice to American defense, in my
view.
But that is just a passing comment. I do
not really expect you guys to comment on that. I just
had to get, in case there's somebody from the press
here, I get it out and let them report it if they
want.
What else? Risk informed regulation, one
of you guys' favorite topics. We are making slow
progress, in all honesty. I don't think any of us
have dusted off the 50-46 paper we have in front of
us. I keep promising I am going to vote on the 50-44
paper because it's relatively straight-forward, but I
am under no pressure to do so because just about no
one else has. We are diverted at the moment.
So I think the staff continues to make
progress. The staff that's involved in the security
issues isn't the same staff that works generally on
risk informed regulation. The staff has made good
progress, as I understand it, at least they have a
couple options out there with regard to how the option
two would work, and I think are having this month
discussions with the public about these various
options for how they would deal with the special
treatment requirements at the nuclear power plants on
a generic basis.
The Commission has encouraged the staff,
as you know, to put drafts out even before they come
to us in order to expedite these rulemakings. We will
continue to do that. I think that is one of the big
changes that has occurred in recent years. We learned
from some of our rulemakings, the Part 70 rulemaking
on the materials side, that sort of going closed for
long periods of time and saying everything is pre-
decisional and doesn't really help the process. When
the information finally comes out, you tend to have a
big controversy.
So we are hoping -- we learned through the
50-59 revision process was very open and transparent.
The maintenance rule revision, the A4 revision, as we
called it, was open and transparent. Now we are going
to try to do that with these new risk informed rules.
But anything that requires a Commission
decision in the near term, we are a fairly diverted
and distracted Commission at the moment. So things
may get coagulated a little bit.
With that, and having taken about 15 of
the 60 minutes, why don't I open myself up to your
questions and see what you want to ask me about.
MEMBER POWERS: Can we come back to the
security issue? I think one of the areas that the
Committee could probably contribute on. I don't think
we can contribute to some of your more policy things,
but I think one of the areas that the Committee
probably could contribute in is looking at what the
staff is doing and the depth to what it's doing then,
especially if we had a better understanding of the
kind of information that you would like to have on the
threat posed by terrorists to the nuclear power
plants.
MR. MCGAFFIGAN: I don't want to get -- I
think you'd need to talk -- I mean I almost would
volunteer except I'll be in a briefing about other
things at 3:00, whenever you are going to get briefed
by the staff. But I could stick around some day in a
closed session with you and go through some of the
things that are in our minds as to where we might have
to adjust our security posture.
One thing I will say to you in open
session. The spent fuel pool study last year, you
know, the possibility that cannot be eliminated that
however infinitesimal it may be, that years after the
spent fuel is put in an ISFSI or left in the spent
fuel pool, I guess you were talking about a spent fuel
pool here, years after the stuff has been put in
there, there is still an infinitesimal chance -- I'm
using my words, not the study's -- that you possibly
would get a zircoloy fire is now of course being used
to say that there's a real vulnerability in spent fuel
pools.
You know, as you know, Commissioner Diaz
and I and others at the time had problems with that
because the assumptions where the staff went from
there being a zero probability of an accident to there
being some infinitesimal probability that therefore we
had to relook the exemptions we already have. You all
said that wasn't a -- I think they said that that
wasn't something that desperately needed to be done
quickly. We still haven't had any relook information
from the staff about the exemptions previously
granted. But it's sort of -- I wish we could have
brought that study further along to the point where we
really had some sense as to whether there's a
vulnerability there that really needs to be addressed.
My understanding was with the proper
earthquake and the proper misalignment of the fuel and
assuming that heat somehow doesn't get out, you could
somehow get a zircoloy fire, but it was even if the
stuff has been cooled for five years, I think it's
related to the point I made earlier. This sort of
search for perfect security and perfect safety, you
know. Is the remote, remote, remote possibility of
something, with significant consequences if you
believe it, something that we should expend in this
case regulatory resources on to try to prevent? I
don't know.
MEMBER KRESS: I think you have the right
take on that. I think when we relook at it, we might
ought to relook at it from the standpoint of how
quickly can we get that stuff out. Because it would
be much -- you would be a better risk status if you
could get the stuff out and put it in the dry cask
because I think they are much less vulnerable to a
sabotage event.
MEMBER ROSEN: Better yet, put it in the
mountain.
MEMBER KRESS: Dry cask first.
MR. MCGAFFIGAN: But you can't possibly
put anything in the mountain because moving any spent
fuel from here to anywhere is a mobile Chernobyl.
Right?
I am not going to get into putting it in
the mountain. Clearly, you would be better off -- I
think our view historically has been the spent fuel
pools and the ISFSIs are protected of public health
and safety. I don't know how -- one of the things we
will have to look at is the vulnerability of spent
fuel pools to various sabotage and whether -- there's
an additional argument now compared to before
September 11th to get things into ISFSIs.
But the pools themselves are pretty darn
safe. A plane diving into one, it strikes me, you
know, you have got a lot of time to add water if the
worst thing happens. I just don't -- you know, I
don't -- compared to what one could do at various very
soft targets around this country that aren't regulated
by us, I am not sure that it is something that we are
going to have to expend a lot of extra resources
trying to regulate. The Nevadans actually have tried
to use the September 11th event to say see, you can't
possibly move this stuff. It has to stay where it is.
I think your comment is closer to the truth, that we
clearly never envisioned having spent fuel in large
quantities forever at the sites. We would have been
better off moving it off some time ago. Not to
reprocess, because processing is not economic at this
time. It may not be economic for a long time, given
the price of uranium.
But the ideal world, we would have moved
this material off of these sites as quickly as it had
adequately cooled to make the transportation and the
casks easy to carry out, the campaign easy to carry
out. But I don't know.
MEMBER KRESS: I really like your view on
this issue that the best thing you can do is control
the initiating event frequency, but not dealing with
the security of the airplanes.
I wonder what your view might be on, as
part of that, having a no-fly zone over all the
plants?
MR. MCGAFFIGAN: I am skeptical about no-
fly zones, to be honest with you. A plane traveling
400 miles an hour travels six-and-two-thirds miles in
a minute.
MEMBER KRESS: That's not much warning
time.
MR. MCGAFFIGAN: At 12 miles, you have
less than two minutes warning time. You obviously
aren't going to recognize it instantaneously either,
so there's a delay time in the recognition.
We only in certain places have combat air
patrol aircraft. They will not be there forever.
Obviously they are there at the current time over the
east coast and other places, and we have Strip Alert
aircraft, but the Strip Alert aircraft aren't going to
get there in 30 seconds or a minute, whatever time.
MEMBER KRESS: I was thinking about
warning time to the reactor operators so they might be
able to do something to shut down the reactor scrams.
MR. MCGAFFIGAN: To scram the plant,
right. Well, that requires again, a tremendous degree
of capability that we some day could have there, but
I don't think we have today. In some places in the
east coast, you are well inside FAA radar areas. If
somebody dips inside the zone, a controller would
recognize it. But you are talking about sort of real
time communication from that controller to the
reactor. We don't have that capability today.
We have the event on October 17th at Three
Mile Island. We did shut -- the FAA did shut
Harrisburg Airport. This turned out to be a non-
credible threat the next morning, but one we had to
treat as credible that evening on the advice of the
intelligence community. But the airport was shut
down. That was announced actually, as I understand
it, at O'Hare Airport, you know, Harrisburg has been
shut down due to threat to Three Mile Island.
The combat air patrol aircraft were in the
vicinity. We had a specific time. Once it expired,
the airport reopened and the combat air patrol
aircraft went away.
MEMBER POWERS: Does the reactor being
operated or shutting down if it's hit by an aircraft
make any difference? I think it makes no difference
at all.
MEMBER KRESS: I don't think it would
either.
MEMBER POWERS: They certainly would shut
down by itself. They shut down by themselves pretty
easily.
MEMBER KRESS: It's assuming the control
rods can go in.
MEMBER POWERS: Even if the control rods
don't go in, it will shut itself down.
MR. MCGAFFIGAN: The other problems with
no-fly zones, as I say, it isn't clear -- you get some
seconds of warning time, which it isn't clear we could
utilize today. Some day we might be able to utilize.
But around many of the sites there are airports not
far away, you know, where people like to fly their --
MEMBER KRESS: You'll have to shut those
down.
MR. MCGAFFIGAN: Well, you tell the
airline owners and pilots association you would like
to shut down all the airports that are within X miles
of nuclear power plants. We were getting a lot of
calls, and the FAA was getting far more last week,
from Congressmen whose constituents were grounded.
They honestly don't feel that they are much of a
threat. These little pilots, you know, Lynchburg
Airport down near the Cat1 Fuel facility in the
western part of Virginia in the Shenandoah Valley.
That airport was shut down the week that FAA had the
controls in effect.
That's another piece of analysis we are
going to have to do. I suspect it's going to be a
straight-forward analysis, but people have suggested
publicly that we need to look at small aircraft diving
into the plant with explosives on board, sort of the
air truck bomb. If the analysis previously done about
the engine being the most important part of the
ability to penetrate containment, these plants aren't
going to be moving that fast. They don't have very
massive engines. The additional explosive may not
make much difference on the outside of the
containment. Most of the explosive force is going to
go out.
MEMBER POWERS: That's not where you want
to hit the plant.
MR. MCGAFFIGAN: You can think of better
targets, and that is what Paul Levanthal has been
saying publicly. He can think of better targets, and
I guess we all can.
No-fly zones just are something that it's
sort of an example -- it's like the surface-to-air
missiles. You know, it's an example of an easy
solution that people gravitate to. In this case, the
analogy early on was the first week we were protecting
football stadiums during football games, you know, to
the extent that no-fly zones protect them. So if it's
good enough for Oklahoma versus Nebraska, why isn't it
good enough for the nuclear sites was the basic
question being asked by the public.
The same sort of thing happened with the
Coast Guard coming around some of the plants. The
Coast Guard, just as it's on high alert and wanted to
know where the nuclear plants were, they put assets
around several of the -- or at least they were
patrolling near several of the sites. As one of the
staff -- one of the staff aphorisms around here is
that no good deed goes unpunished. When they actually
wanted to take those assets out because they figured
out that maybe they would be better utilized somewhere
else in some of these busy ports where you have these
large liquified natural gas tankers wandering through
and whatever, there was uproar and human cry as to how
dare you pull these Coast Guard assets from the
nuclear sites.
A lot of the security people, and I'm glad
Governor Hodges visited Catawba yesterday or the day
before. He came out of the plant very impressed with
the security. He visited with his director of state
security.
Most of the people who go to our plants,
you know, and see the capability that we have at the
plants, especially in the state they are in at the
moment, come away saying this is not the highest and
best place to assign my marginal resources or that
they are not particularly needy of National Guard
assets compared to other places in that state. That
we have a very mixed bag.
I saw in today's press clips that the
Governor of Illinois felt peer pressured into putting
the National Guard at his sites. I don't know whether
he shut it down today or not. Texas took their
National Guard out of Comanche Peak, feeling that the
capability was adequate. I believe that's probably
going to happen at Palaverde in Arizona. Governor
Hull had put National Guard assets in there, and I
think is probably going to pull them out. But these
are decisions governors have to make, given the
information that they have.
There are times when augmenting with
National Guard, clearly augmenting -- we have
encouraged since September 11th that there be these
protocols in effect between the licensees and state
and local. They are sort of there always. We have
encouraged dialogue with the governors. Some of our
licensees have asked for National Guard augmentation.
That is fine.
But in many cases, Governors have gone in
with their security folks and said, gosh. I don't
think it's needed. I've got some state police there.
That is enough. I don't need to put the National
Guard in.
In Florida, we had two different utilities
making two different choices. Florida Power at
Crystal River decided they didn't need the National
Guard help. Florida Power and Light accepted it at
St. Lucy and Turkey Point.
We probably at some point need to, you
know, as a Nation, think these things through. They
are clearly not decisions just for the NRC. We are
not the deployers of national security assets of the
country. But there is a real concern that we have
that by focusing too much on the nuclear facilities,
you are going to actually hurt overall security by
raising vulnerabilities at these software chemical,
petrochemical, et cetera, sites that would offer very
attractive targets.
MEMBER SIEBER: I was wondering, you spoke
earlier about doing vulnerability studies. It would
be my impression that to really do good ones to try to
redefine the design basis threat, it might take a year
or two to do that.
MR. MCGAFFIGAN: Well see, the
vulnerability studies, I'm not sure will inform the
design-basis threat. If I am postulating that I
personally do not think that we are going to add ever
air strikes to the design-basis threat. By
definition, the design-basis threat is the threat
against which licensees should have high assurance of
being able to defeat. Licensees have assets that are
constrained by law as to what they can have. I don't
think even Charleton Heston, who we'll have to pry
that gun out of his cold dead hands, wants surface-to-
air missiles in licensee hands at these sites. It
wouldn't be Patriots, it would be Stingers.
You solve one problem that we may solve by
other means and we create another. Proliferation of
Stingers around the country at a variety of sites, and
god help us if somebody gets their hand on a Stinger.
That's the real way to bring down a commercial
airliner.
So some of these vulnerability analyses
are going to be more of interest to the Pentagon and
the Office of Homeland Security in terms of trying to
decide how important it is that they assign assets
that are under their control to the defense of these
plants. Vulnerability analyses, the sort of analyses
that we're going to have to -- the design-basis
threat, we are going to have to decide about the
numbers of attackers, and we are going to have to
decide about the weaponry that the attackers are going
to be hypothesized to have available to them.
We are going to have to decide how large
the truck bomb. We have a certain sized truck bomb in
our design-basis threat at the moment. We'll have to,
as Mr. Levanthal requests, and I think we would do it
in any case, look at whether a larger truck bomb will
need to be protected against.
We will need to look at -- I mean there's
legislation that you all should look at pending. It
is attached to the Price Anderson Act, reauthorization
in the House, that would require the President to do
a study in which he would parse. It is the so-called
Tauzin, Dingle, Markey amendment, in which he would
parse threats into two bins, the design-basis threat
bin and the enemy of the state bin. Then the
provision would require us to do regulations within a
certain time period, I believe 275 days after
receiving the President's report, to adjust our
design-basis threat. Then presumably the enemy of the
state threat, that information would be used by the
Pentagon and Office of Homeland Security and others to
decide what they were going to do.
It also mandates that we will have
essentially an OSRE type program, but an expanded OSRE
type program, in that it defines sensitive nuclear
facilities for which an OSRE will be required to
include reactors, category 1 fuel facilities, the
gaseous diffusion plants, spent fuel pools,
decommissioned reactors, et cetera. So you would have
-- we have done OSREs in the past at the 63 or 64
sites, depending how you count them, that we have,
reactor sites.
We have not done OSREs -- we have force-
on-force exercises at the 2 Cat 1 fuel facilities as
well. We have not done them at decommissioned
reactors or spent fuel pool facilities that are
independent of reactors. So it envisions an expansion
of the OSRE program, and in a sense, the design-basis
threat analysis that they want us to do I think given
the definition of sensitive facility, would also
require that the Presidential study look at binning
design basis threat and enemy of the state threats for
this larger category of facilities.
MEMBER POWERS: What was their force-on-
force exercises? The problem you always have with
large-scale tests, you get to do one.
MR. MCGAFFIGAN: We got to do four. Each
OSRE has four drills that are gone through.
MEMBER POWERS: That's right. It gives
you a very limited view of what your actual defensive
capabilities are, and does not allow much in -- if you
find deficiencies in those defensive capabilities,
your choice is to amend them, and then to go retest.
You don't have the facility for looking at lots of
different options.
That is a problem that the Air Force
encountered in defending lots of its bases. There has
been a fairly well developed technology of developing
computerized scenarios, where you calibrate against a
test for the facility and what not, and then you use
these computerized capabilities to evaluate options
and designs and what not, and make your testing along
some sort of a progression where you improve actually.
MR. MCGAFFIGAN: Right.
MEMBER POWERS: I wondered if we shouldn't
be thinking about bringing those technologies to bear
here, rather than just going out and running lots of
OSREs at every facility around the country at God
knows what cost.
MR. MCGAFFIGAN: I think we understand
some of the limitations of force-on-force testing. I
would welcome any thoughts as to how to bring that
technology. Unfortunately, the dynamic that has been
set up is, you know, the factoid out there is that 47
percent of the time licensees over the last X number
of years, ten years or so, have failed OSREs.
MEMBER POWERS: I'd like Dick's response
on that one too.
MR. MCGAFFIGAN: I'm not sure what his
response his. The actual number is more like, you
know, in terms of the four drills per site, licensees
have actually succeeded 85 or 90 percent of the time
in these drills. When they have had problems, we fix
them, like you say.
Does that mean that the whole strategy has
been fixed? It isn't clear. It means that that one
hole in the strategy that was tested that day in one
of four tests has been fixed. I don't know whether
that -- I think we're better off having done OSREs
than the chemical industry that doesn't have anything
like this capability. I think we do give hard tests.
These tests that the staff uses in an
OSRE, they do a bunch of tabletops. Then they detect
what they think is a vulnerability. Then they test
against that vulnerability. So these are fairly smart
-- the four drills that are carried out, they can't
all test the same vulnerability, but if they see a
vulnerability in the tabletops in the licensee's
defense posture, we go try to probe that vulnerability
in the OSRE. As I say, if we find it, as we did at
Vermont Yankee or other places, we get it fixed
promptly with compensatory actions.
But I think trying to find a way -- you
know, I have come at OSREs a little bit from at Fort
Irwin in the desert in California, we do force-on-
force exercises at very large units. The red team is
pretty damn good, and they win. It is better for our
other units to learn in the desert in California than
to learn in the battlefield in Iran or Iraq. So there
is a value to force-on-force because you are getting
to make mistakes that don't matter, so that when it
does matter, you are better able.
MEMBER POWERS: I think when you get to do
repetitive tests that that's true. I think you are
running into the problem of diminishing returns
because in many respects you are playing got-you with
the licensee here, because they don't happen that
often at each one of them. I think there is room for
bringing an improved technology into this.
MR. MCGAFFIGAN: Well, I would love to
hear about it.
CHAIRMAN APOSTOLAKIS: Can we move onto
other subjects?
MR. MCGAFFIGAN: Sure. You guys are
falling into the September 11th trap that we all are
in.
CHAIRMAN APOSTOLAKIS: Let me come back to
something you mentioned in passing earlier and others
also mentioned it in the public forum. Risk informing
the regulations is proceeding at a very slow pace or
slower pace than anticipated.
I guess I am a little puzzled by that.
Can you elaborate? I mean what should we have done by
now to be able to say we are on schedule?
MR. MCGAFFIGAN: I don't know. In 50/46,
there's some fairly complex issues, that paper for us.
You guys have it. I forget if you sent us a letter on
that paper, I forget whether you have. You probably
have. It shows you how much I've read.
The problem is others are suggesting from
the industry that the strategy in the 50/46 paper is
not aggressive enough, and that we would do better to
carve out a few things and to get some early success.
I have not -- you could use the next few minutes if
you want to educate me as to why I should go back to
my office and just check yes next to the approved
column and move that paper along.
I thought that probably given all the
letters that we have gotten from members of the
public, mostly from industry to be honest with you,
that it was worth my looking at them, thinking about
them, and seeing whether the industry folks were right
that there's some parts of this -- they also are
disappointed, is my recollection, that we are not
willing to -- they have been arguing, as I understand
it, for an amendment to 50/46 that essentially would
say you can change the large break loca. We will
entertain changes to the current double guillotine
break large break loca design-basis accident. You
know, they fully expect that any change would have to
be done with NRC approval.
They would like to get that regulation
moving so that there's a regulation in place whereby
owners groups could submit arguments for another
design-basis accident, and we all could grapple with
it. As they see the staff proposal, it is going to be
many years before they can even have the argument with
us. I don't know what the right answer is there. So
I am trying to figure out what the right strategy is
on the 50/46, the various 50/46 options that are
before us.
CHAIRMAN APOSTOLAKIS: So it seems then
that --
MR. MCGAFFIGAN: On 50/44 it is going to
be straight-forward. When we all get around to voting
on it, we are going to basically, I think, endorse
what the staff, the revised approach the staff is
taking. I think that one will go reasonably rapidly.
50/69, or whatever the option two stuff, I think
depends how these meetings go and whether the staff
itself can come to consensus. But if you, the staff,
and the public were all to come to consensus that this
is the right 50/69 option, that could go fairly
quickly. It's the 50/46 is the place where I see a
problem.
CHAIRMAN APOSTOLAKIS: So as long as the
first 50/46, not the overall.
MR. MCGAFFIGAN: Not the whole thing. As
I say, I think a lot of people are working, the people
who do risk-informed regulation on a day-to-day basis
are out there trying to figure out how to do it. They
are having their meetings. But I think it is the
option three stuff that I see as going a little
slower.
MEMBER POWERS: I think we have run into
some technical challenges on 50/46 that made
consideration of going for some of the gimmes in 50/46
attractive. I mean things like, for heaven sakes, why
can't we update the decay heat curve. That is kind of
a gimmee.
MR. MCGAFFIGAN: Well, see what Oshuk
tells me in my office is well, not so fast,
Commissioner, because that is clearly an area of over
conservatism, but there may be a few places in 50/46
where we're not being overly conservative. By giving
up the over-conservative here, are we somehow
upsetting something. So my understanding is, and the
50/46 paper I think reflects this, that everything is
connected to everything else. So therefore, you can't
do the gimmees because everything is connected.
I am skeptical about that, to be honest
with you. If you all were to come in and say take the
gimmees, I think it would have a strong effect on how
the Commission would think about this stuff if there
was a consensus in this group.
But the staff's approach is that
everything is connected and everything will only come
together when everything comes together. That could
be some significant period of time.
MEMBER ROSEN: That sounds to me like a
prescription for not doing anything for a very long
time.
MR. MCGAFFIGAN: I worry about that, yes.
MEMBER POWERS: I think there was merit to
what the staff's approach was when it looked like you
could move forward. They have run into a real
technical barrier in one aspect of it. I think you
need to rethink your strategy here and look for the
gimmees in this thing.
Before, you know, they are right. There
are tentacles from 50/46 that go out and touch lots of
places, but now that you have run in and found a hard
spot, it is going to take some substantial work. Why
can't we go back and look for the ones where the
tentacles are few in number and limited in extent.
MEMBER ROSEN: The decay heat curve that
you mentioned is one of those places. It is merely
being used as a -- we need to trade off something with
it. I don't think that's the right kind of thought
process.
MR. MCGAFFIGAN: I tend to agree with you.
You know, the staff is in this sort of let's make a
deal mode. I'll give you that if you will give me
this. I am not sure that that's -- there is clearly
an awful lot of conservatism in 50/46. There may be
a place or two where there isn't, but there's an awful
lot of conservatism in it. It drives a lot of stuff.
It may not be the best place to expend resources. So
we need to think about it.
As I say, there is an opportunity for you
there. I will tell you, the number of commissioners
who have voted on the 50/46 paper is a null set. So
you would be still timely if you have any additional
thoughts you want to make on the 50/46 paper. You
have heard some of the comments, I suspect, from NEI
and others in your deliberations in recent months. If
you have now come to the conclusion that we should
think about gimmees, I think it could have an effect.
CHAIRMAN APOSTOLAKIS: The difficulty you
just raised I think is an important difficulty. In
the old traditional way of doing business, we put
conservatisms in various places. You should get them
as a package as a whole. You can't really start
removing here without thinking about what happens
somewhere else.
But the concern now, and I understand how
that evolved because risk-informed regulations came a
few decades after the original system was put in
place. What I am concerned about is that we are about
to create a situation that will be very similar to
this for the future reactors. Again, there are good
reasons for that, but maybe we should try to be a
little more vigilant to avoid it.
The reason why people want to use as much
of the existing system as they can, like Exelon, for
example, is that of course it is much faster. I mean
if you use something that is already in the books and
you modify it a little bit to accommodate your new
design, then you have a hope that sometime you will
see your license.
MR. MCGAFFIGAN: Within a glacial time
period.
CHAIRMAN APOSTOLAKIS: Well, on the other
hand, you are creating again a situation where we're
perpetuating philosophies and approaches to regulation
of the past. Then we're going to say again, 10 years,
15 years from now, my goodness now how do we remove
burden, how do we -- I mean we are stuck again with
safety-related and non-safety related components.
What do we do about it? The year will be 2020.
So I don't know what to do about it,
frankly, because I appreciate the difficulty or the
concerns that the applicants have. You know, we can't
wait until you guys come up with a new system. Right?
But it's really something I think that should be of
concern to everyone.
MR. MCGAFFIGAN: Isn't NEI talking about
giving us a new part 50 for --
CHAIRMAN APOSTOLAKIS: We have not seen it
yet.
MR. MCGAFFIGAN: Part 53, whatever they
call it, for new reactors?
CHAIRMAN APOSTOLAKIS: The last time the
representative was here, he didn't present anything.
MR. MCGAFFIGAN: This is a new. I mean it
doesn't exist, but I thought they wanted to come up
with some sort of risk-informed operating regime for
--
VICE CHAIRMAN BONACA: We have heard the
rumors.
MR. MCGAFFIGAN: I haven't seen -- I've
seen it in Inside NRC or Nuclear Next Week, reliable
publications. I assume they have sent it to somebody
in public.
I understand what you are saying, George.
I don't know that -- there is at least, my
understanding is that they are going to take an extra
year to make their pebble bed decision or at least
nine months. So there is some extra time for us to
try to put different regimes in place. But there
isn't a lot of -- that probably isn't the focus to be
either Exelon's or the staff's work at the current
time.
I see Graham Wallis. I should mention I
do like the work that you guys have been doing with
regard to reviewing some of the staff's work on
approving codes and whatever. I see some of these
hard-hitting letters. I commend you for those
letters. I think somebody has to keep the system
honest. You guys are clearly doing that, so at least
as one commissioner, I appreciate that.
MEMBER FORD: Can I ask you a question?
Going on from the risk-informed aspect. Come back to
the very first topic, CRDM housing. Specifically,
what do you want our advice on?
MR. MCGAFFIGAN: Should we issue an order
shutting down Davis-Besse 100 days earlier than their
normal outage because there is the risk of allowing
them to operate that extra 100 days does not provide
reasonable assurance of public health and safety. Or
should we not issue an order and allow the 100 days to
run and let them shut down at the normal time and do
the inspections at their normal outage date. That is
the issue before us. It is before the staff.
As I say, the staff is -- Davis-Besse is
talking to you. Davis-Besse is talking to the staff.
Davis-Besse is trying to argue -- I mean the undertow
of what you guys were watching this morning was this
dance between the staff and Davis-Besse.
If the staff isn't satisfied that they
believe there is reasonable assurance of public health
and safety during this 100 day period, they will
sometime in December come to us with an order. They
won't come to us. They will send us -- Ken Rogers is
in the audience. They will send us an email to one of
our staff saying we would like to issue -- not hearing
from the Commission otherwise, we will issue five days
hence, an order to Davis-Besse and perhaps to this
other facility telling them to shut down on December
31st and get their inspection done.
I am capable of a lot of stuff, but I am
not an expert in this. So this is exactly where this
group of people could tell us is the staff on the
right track or should we give Davis-Besse the extra
100 days. You don't see a lack of reasonable
assurance during that period.
If you were to send us that letter or just
even threaten to send us that letter, the staff --
CHAIRMAN APOSTOLAKIS: Actually, there may
be some procedural problem here.
MEMBER FORD: We really haven't had enough
data. I don't think we have really heard, we haven't
heard formally from the staff on their position on
this. We have had in fact remarkably little
information. We have heard a lot of plans of what's
going to happen. So any advice we give you would be
very, very --
MEMBER ROSEN: I would offer one
perspective, which is that when you're talking about
100 days in the timeframe of this whole thing and
saying it somehow goes against our analysis, it
ascribes through our analysis a degree of precision
that I'm not sure is really there.
MEMBER POWERS: That's right.
MR. MCGAFFIGAN: As I say, I'm open to
whatever you guys want to advise. But the issue comes
down to how conservative do we want to be and when do
we lose reasonable assurance.
If one of these circumferential cracks
grew and the whole thing was severed, is that an
accident that we would want somebody to have to endure
and recover from? My understanding is that that
accident is well within the design-basis of these
plants, and they should easily be able to handle it if
it occurred.
MEMBER POWERS: Provided it's just one.
MR. MCGAFFIGAN: Provided it's just one,
right.
MEMBER ROSEN: And the consequence of the
accident, it's a medium loca, are within the
containment. So the public's health and safety is not
at issue.
MR. MCGAFFIGAN: Right.
MEMBER POWERS: It's a medium loca with a
failure to scram is what it is.
MR. MCGAFFIGAN: That is the issue that is
going to be before us. There is a procedural problem
in probably the Commission as a whole. This one
commissioner hasn't asked you guys your opinion. But
if you would want to be asked, I could work on getting
you asked.
(Laughter.)
CHAIRMAN APOSTOLAKIS: Soft vote.
MR. MCGAFFIGAN: But that is an issue that
of all the issues that are currently sort of kicking
around that is so up your guy's ally and so not up my
ally, that I would ask for any help I could get on it.
MEMBER LEITCH: We haven't really had a
chance to talk about this yet, but after hearing the
Davis-Besse presentation this morning, this cracking
phenomena, and I haven't bounced this off my
metallurgical counterparts here, but the cracking
phenomena is extremely proportional to temperature.
A relatively small reduction in temperature gives
considerable relief from this phenomena.
I was just wondering -- chance to ask the
Davis-Besse people this morning, but I don't know. It
seems to me that plant may be able to be operated --
VICE CHAIRMAN BONACA: At lower power.
MEMBER LEITCH: At lower temperatures, at
lower power levels. So perhaps --
VICE CHAIRMAN BONACA: There's a
compromise there.
MEMBER LEITCH: A compromise position
might be a -- I don't know what the temperature of a
say, an 80 percent power might cause a reduction of
seven degrees or something like that. I mean I don't
-- it might be of that order.
MR. MCGAFFIGAN: And I don't know what the
TMI data. You know, the TMI data where they did not
see circumferential cracks when they inspected. We
had the Oconee data. These are all, what, BNW plants.
So we know we have the TMI data. Does that mean that
we should have more assurance now that we have the TMI
data and there weren't circumferential cracks? I
don't know how you handle all that.
But as I say, it is clearly an issue that
you are more capable than I of thinking it through,
although I am willing to hear from anyone on the
subject.
Other questions? I have almost used up my
hour.
MEMBER ROSEN: Well, I was interested in
your take on the whole question of advanced reactors
and licensing of them. Where do you think where the
Commission is headed on it?
MR. MCGAFFIGAN: I don't know. We clearly
are putting resources into it Congress has been happy
to give us. We did get 10 million dollars extra in
our budget to help us deal with advanced reactor
issues this coming year. What's most important to the
industry as a whole is what we need to work on, the
early site permits.
My understanding, Intergy now has just as
Exelon has lined up with the pebble bed, Intergy has
lined up with the modular high temperature gas reactor
General Atomics. Others are still looking at the
Westinghouse AP1000.
We just need to put adequate resources
into doing the things we need to do. We need to look
at the PAR52 rulemaking, to update it on certain
things that I think would expedite the process if it
starts. We're all a little skeptical, to be honest
with you.
I mean a year ago when the price was up
here, and it was easy to see these plants being quite
economical. We're about to go through a winter where
the price of gas is going to be a lot lower, and I
don't know what to predict for the long-term price of
natural gas, which is one of the key things that
utility executives look at when they try to make a
decision to invest in a nuclear power plant.
So we're trying to put adequate resources
into it. If we get early site permits, we will try to
run that process expeditiously as we have done the
license renewal process.
I think the September 11th events -- we
have 2 206 petitions before us. We have various folks
who are going to be trying to raise for anybody who
does come in for an early site permit, some of these
security issues.
Mr. Waxman -- there is a provision in the
House version that was adopted unanimously by the
House Energy and Commerce Committee that would require
us to consult with the Office of Homeland Security
before providing any Price Anderson indemnification to
any new applicant.
That will require, if it becomes law,
require us to put in place some procedures that we
don't have today to get -- I'm sure our licensee will
want to get that check off from the Office of Homeland
Security very early in the process. They are not
going to want to build the plant and say, okay, give
us our Price Anderson indemnity, and we say sorry,
Homeland Security vetos your plant.
So it's an extra little bit of
uncertainty. I think it's handleable. But if that
becomes law, and as I say it was adopted by voice vote
in the committee, and I don't believe the
Administration is objecting to that provision in the
statement of administration policy in the House bill,
we will just have to build that into our process in a
way that isn't there at the moment.
So I think if that is the only thing that
results from September 11th, that isn't much of a
burden and we'll handle it. But I can't fully predict
what's in the minds of utility executives at the
current time in light of the price of gas going down
and the climate being complicated by the security
events, and the undue focus on vulnerability of
nuclear assets.
General Atomics, and I'm not sure it's
good for the industry as a whole, you know, General
Atomics and Exelon have helpfully said that they could
bury their plants.
CHAIRMAN APOSTOLAKIS: And still produce
power.
MR. MCGAFFIGAN: And still produce power,
right. They can make these almost impossible targets
if indeed we have to worry about it. If they want to
bury their plants, great. I am not sure that as a
regulator I am going to require them to bury their
plants in order to take on diving commercial
airliners.
Is that enough of an answer?
CHAIRMAN APOSTOLAKIS: Any other questions
for the Commissioner?
MR. MCGAFFIGAN: Okay. Well, thank you
very much.
CHAIRMAN APOSTOLAKIS: Thank you for
coming down here.
We'll take a five minute break.
(Whereupon, from 3:00 p.m. until 3:15
p.m., the proceedings went off the record.)
CHAIRMAN APOSTOLAKIS: Okay. We are back
in session. We have an ad hoc presentation by the
staff on the issue of power uprate.
MR. BAILEY: Yes. This is Stewart Bailey.
I am the Project Manager for Quad Cities. We are here
to try to answer some of the Committee's questions
related to how Exelon is achieving the power uprate
out of their BWR course. So with that, I'll turn it
over to Tony Ulses.
CHAIRMAN APOSTOLAKIS: What's the name
again?
MR. ULSES: Tony Ulses. How to do a power
uprate in ten words or less.
CHAIRMAN APOSTOLAKIS: Good.
MR. ULSES: What I would like to start off
with is -- well actually I'd like to not put that on
right now, Stu.
I was kind of reviewing this material that
you all got yesterday from GE. I kind of see where
the confusion is coming from here. I would like to
make a brief comment on it, and then try and avoid
getting into it because I don't really understand all
the details of how the information was generated. I
understand I will probably have limited success at
that, but that is kind of where I would like to go.
Essentially what you are seeing here when
you compare this information from cycle 17 to cycle
18, what really is causing the confusion is that they
are introducing a new type of fuel into cycle 18. In
other words, they are going from a nine-by-nine fuel
in cycle 17 to a ten-by-ten fuel in cycle 18.
MEMBER SIEBER: Partially.
MR. ULSES: Right, partially. That is
going to be the new batch that they are inserting, is
going to be ten-by-ten. That fuel in fact can run at
higher total bundle powers than the nine-by-nine fuel.
That is how you can increase the average and not cause
the peak average to go down because you are in fact
increasing the peak maximum of bundle power, but you
are not increasing the peak rod power, kilowatts per
foot. Because basically you have more rods to work
with, you can make more power. That is basically what
they are doing.
Also, another feature of the ten-by-ten
assembly is that they have additional margin to
minimum critical power ratio, which allows them to get
this additional power out of the assembly due to
changes in the assembly design, mainly in the spacers
and the optimization of their axial location.
That is all I really wanted to say on
that, unless there are any questions.
MEMBER WALLIS: Well, ten times ten is 100
and nine times nine is 81. The difference is
something like 19 percent or something, which is what
they are asking for almost.
MR. ULSES: Well, I think that might just
be a coincidence.
MEMBER SIEBER: It's the surface that
counts there. The surface doesn't go up by 20
percent.
MEMBER WALLIS: I think they said this
morning that the fuel is also more enriched.
MR. ULSES: Yes. They are running at a
higher enrichment because they want to get more energy
into the core.
MEMBER WALLIS: And they play more tricks
with Gadolinium. They also put in pressure fuel more
often. So the more fuel, the more --
MEMBER SIEBER: More assemblies, yes.
MR. ULSES: Let me jump into the
presentation here. I have this stuff on the slides.
Let me go ahead and change this around, Stu.
All right. How to do a power uprate. So
essentially they have to do three things. They have
to get more fissile content into the core. They want
to burn more U-235. So they load more bundles. Also,
in this case the bundles happen to have a higher
bundle average enrichment, although I don't know that
that's necessarily generally true. That means they
withdraw more as well because they have a fixed number
of locations.
MEMBER WALLIS: The total level stays the
same, but in each load they exchange more.
MR. ULSES: Exactly. They have a fixed
number of locations. Essentially they extract more
and they add more fresh, which gives them a higher
fissile content and they can burn it.
Now with this new fuel, they have to keep
it within limits obviously. So what they do is they
use more changes in the Gadolinia loading, both
axially and radially, and they also do a lot of radial
enrichment and axial enrichment changes in the fuel as
well. These modern beauty bar fuel assemblies are
extremely complicated.
MEMBER KRESS: Tell me what the different
Gadolinium loading means. You add less of it in?
MR. ULSES: Well, what it means is that
within one bundle you could see different locations
that actually have different Gladolinia concentrations
radially and axially in order to shape the power.
Then within each type of bundle that they insert at
the beginning of core, you could actually see actual
different total Gladolinia loading as well.
MEMBER KRESS: It might put more in the
central channels and less in the --
MR. ULSES: No. What they are doing is
putting in what they need where in order to maintain
the power distribution. You know, the concepts of
doing like an outer middle, inner ring loading, that
is not done any more.
CHAIRMAN APOSTOLAKIS: But by fuel rod,
fuel rod by fuel rod.
MR. ULSES: Exactly.
MEMBER KRESS: Pellet by pellet almost.
MR. ULSES: Sure. Right. Well that's
actually the truth, pellet by pellet. You are going
to see radial and axial enrichment changes, and also
Gladolinia loading changes. They also will increase
the total load of gad in order to keep the reactivity
down as they burn the core, because obviously they are
going out to longer burnouts. Well they are going out
to higher -- the batch average burnouts are not
increasing the fuel burnup over the limits. I mean
let me just get that out there.
MR. BAILEY: Higher reactivity.
MR. ULSES: And essentially they have been
doing this for about 10, 15 years roughly. It
actually started with their later eight-by-eight
products. It is in the nine-by-nine products, and
it's in the ten-by-ten products that are out there
right now. This is also not just done by GE, it's
done by all BWR fuel vendors to a certain extent.
Now for EPU, which is basically the power
uprates that get up in the 15 to 20 percent range,
they are going to have to go to these newer fuel
designs because they have to get more maximum bundle
power. You just simply can't get it out of the nine-
by-nine assembly because mainly limits on the critical
power ratio. In other words, they are going to go
into dry out in these assemblies if they run them up
in power.
MEMBER KRESS: Where do they get this
extra space? Do they actually cut down on the spacing
of the fuel?
MR. ULSES: What they do is they make the
pin smaller.
MEMBER KRESS: That's what I meant.
MR. ULSES: Right. They are actually
physically smaller themselves. Then they will
increase the enrichment, they will offset any changes
in the actual physical geometry of the assembly. That
allows them to run these bundles out to higher
burnups.
Now like I said before, they are going to
have to get more CPR performance out of these bundles
or they won't be able to increase the maximum power,
but that has been achieved over the years by changes
in spacer design and by the physical changes of the
geometry themselves. But the key point here is that
it is confirmed with prototypical tests. They
actually design a prototypical bundle with
electrically heated rods. They put it into their
atlas test facility, and they do CPR testing in order
to confirm the continued applicability of the
correlation they use to predict critical power. That
is done for each fuel type that they manufacture.
Essentially just the last bullet there is
just kind of the point that they made incremental
design changes over the years, but if you look at the
modern fuel, basically what they have done is they
have grabbed everything that they have learned, and
they have put it all into one place, which is really
what is allowing them to get this extra power out of
these assemblies and maintain the actual local limits
on the assembly, which is how they are really
licensed.
MEMBER WALLIS: It's obviously a very
complicated fuel management program.
MR. ULSES: Extremely complicated.
MEMBER WALLIS: If you look at these, you
find 1.39 is beside .99, and then there's another
1.15. There's no pattern at all that makes any sense.
MR. ULSES: It is extremely complicated,
which is why if you look at the way they do the actual
reactor design these days, they are going to take the
core and they are going to do detailed calculations of
the reactor as it burns out on its lifetime because
they need to ensure that they are going to maintain
thermal limits.
Actually, I have got a lot of --
MEMBER WALLIS: The flux may not flatten.
The flux flattening is probably a red herring.
MR. ULSES: But that is more consequence
of what they are doing. They are going to load more
reactor fuel at the beginning of the life, and the
flux has to flatten simply because they are loading
more fuel and it's going to take up more locations,
and the flux is going to have to flatten. So I would
say it's more of a consequence than a means to an end,
myself.
That is even more true when you are
talking about inserting new types of fuel which can
run at higher maximum bundle power.
MEMBER WALLIS: Also the axial
distribution varies tremendously from beginning to
end.
MR. ULSES: What I was going to say is
that I have a lot of -- actually, I have a lot of
background information up here which is actually GE
proprietary. I would obviously rather not get into it
here, but if anyone wants to look at it, I can stay by
after we're done here and I could show it to any of
the members who would be interested in seeing
information about the design.
MEMBER ROSEN: The ACRS handles a lot of
GE proprietary information during the reviews. It was
an innocent that was asked to try to follow just a
little deeper in the proposal which said something
like we're going to flatten the profile and that's how
we are going to get a lot more power out of it.
So I asked for the profile. Let me see a
core map. Let me see a beginning a live core map.
Let me see an end-to-live core map, the pre-EPU and
post-EPU so I can get a sense of just taking it
another level down so we understand. That may have
turned out to be the wrong question.
What I would like to know is, because I
think we need to go another level beyond oh, we're
just going to flatten the power. That is all that was
really said about how we're going to get all this
power in this stack of documents this high. I would
like to go another level down below that, get a little
more sense of sensible information. I don't know the
right question. What is the right question?
MR. ULSES: The right question to ask is
will they maintain the bundles within their rating,
within the thermal limits.
MEMBER ROSEN: What are those? Show me
what your projections are and all that.
MR. ULSES: If we look at how we license
BWR fuel, it is very dependent on the local parameters
because of the fuel channels. The reactor fuel itself
doesn't really care what's around it because of the
channel. All it cares about is what it sees at the
inlet and the outlet.
So it is very local. It is very specific
on the assembly. That is one of the reasons why they
are able to do this, because they are able to using
basically the tool of the Gladolinia, if you will, in
this case it uses the tool, that they able to shape
the power in the assembly such that they can keep the
peaking within the assembly down as they burn the
fuel. They can stay within the applicable limits.
Let me go and jump to my next slide here,
which is how we license fuel. But as for your
question about what you need to ask --
MEMBER ROSEN: I am going to ask a
question on the next EPU that comes through, and I
understand there is going to be a lot of them.
MR. ULSES: There will be.
MEMBER ROSEN: I don't know exactly what
the question is but I will certainly want to zero in
on this. I would ask the staff to help me with that.
MR. ULSES: Well I would say without
thinking about it a great deal, the question that I
would ask is, just like I said, I mean essentially are
you maintaining the bundles within the design limits.
That information ought to be able to be provided to
the Committee. Essentially those limits are you have
to maintain the LHGR limits. You have to maintain the
MCPR limits, and you have to maintain the maximum
average planar linear generation rate.
What is in parenthesis here is what those
limits are trying to protect. Essentially you don't
want to melt fuel. You want to maintain good heat
transfer, and you want to meet the 10 CFR 50:46
exception criteria, which is what you have the maximum
average planar linear generation rate for. That is
going to be set by your loca analysis.
MEMBER WALLIS: Average planar is local?
MR. ULSES: It's an average planar, right.
That's the mapple hover.
CHAIRMAN APOSTOLAKIS: Just one comment I
have is that I think at least I was mislead in the
statement of flatten out the power distribution. You
are assuming that that meant that the peaking factors
go down. That is not the case. You are pushing it
up. Let me finish.
You are pushing it up axially at the top
and bottom because the same strategy is being used by
the PWR vendors.
MR. ULSES: But actually the axial power
profiles, but those have been used for many years.
That is not atypical. That is in use right now.
Essentially what they are doing is they
run the core early in cycle with a highly bottom
peaked power distribution. That allows them to
spectral shift the reactor. Then they start moving it
up at the top because they want to burn the fuel out
evenly. In other words, they want to use all the
uranium that's in the core.
If they went back to like what they used
to do was like a hailing concept, but that is not used
any more because it does not allow them to burn out
all the fuel. Essentially the utilities are spending
money on enrichment that they are not using.
So what they do now is they go to these --
is that they go to these management strategies that
allow them to move the axial power distribution around
a great deal during the cycle. That is how every
reactor that I am aware of currently operates right
now.
MEMBER WALLIS: What's LHGR? Like the
number here, .77. What's that? I am looking at these
printouts we got, to relate them to your criteria. It
says LHGR .77. That's not a temperature?
MR. ULSES: What we need to do is look at
the thing at the section called the thermal limits
summary. You look at maximum kilowatts per foot,
which is for the one I'm looking at, is 10.14.
Another one is --
MEMBER WALLIS: I think that's the second
one on the list.
MR. ULSES: It's actually the second to
last.
MEMBER WALLIS: What's rapid LHGR?
MR. ULSES: I actually don't know. That's
a value that they probably use --
MEMBER WALLIS: Same symbols as you have
up there.
MR. ULSES: Well, what they call a maximum
kilowatts per foot is what I am referring to here as
LHGR. Other than the limits of what they are going to
use is what are more than likely going to be used by
the operator in the control room. They try to come up
with parameters in the control room that are really
easy to understand. They try to ratio the parameters.
MEMBER WALLIS: This is less than 13 or
something like that?
MR. ULSES: The values vary from field
type to field type, but that is a pretty good average
number.
MEMBER WALLIS: Where do I find MCPR?
MR. ULSES: That is --
MEMBER WALLIS: That's the 1.79.
MR. ULSES: The one you are looking at,
1.79. Yes, that's going to be limited in the
technical specifications. The value is on average
typically I want to say 1.09, 1.1.
Ed, is that about right? In this
particular case, yes.
MEMBER WALLIS: Then the other one, MCPR,
is --
MR. ULSES: The next one is the mapple
hugger.
MEMBER WALLIS: That must be the APLHGR,
.77.
MR. ULSES: No. That is actually going to
be the one that is above the actual thermal limit
summary which is on the order of 9.14 in this case,
and on the other ones --
MEMBER WALLIS: Somewhere else?
MR. ULSES: Yes. You have to go right up
above the section that says thermal limits summary, to
something called maximum APLHGR.
MEMBER WALLIS: It says 9.11?
MR. ULSES: Right. In this case it's
9.14. I'm not exactly sure what those units are. I
assume they are probably kilowatts per foot. That
would make sense.
MEMBER WALLIS: So what you guys do is you
assure yourselves that all these numbers that are
going to be varying throughout the cycle and with
different fuel loads and all kinds of strategies,
never go over some regulatory level?
MR. ULSES: Exactly. Those are specified
in the fuel type and bundle-specific basis. They are
monitored continuously throughout the cycle. The
reactors, they are actually running online monitoring
which actually runs a three-dimensional solution of
the reactor all the time, comparing it to the in core
instrumentation. They are using that to ensure that
they are meeting all applicable thermal limits on the
fuel.
MEMBER ROSEN: Maybe you could help me
with the second question. The first question is are
you maintaining LHGR for fuel temp and for loca and
MCPR below the limits? The answer they give me is
yes.
MR. ULSES: Yes. That is the answer they
have given us. We have confirmed that through our
audits.
MEMBER ROSEN: My next question is what
should my next question be? Show me, right?
MR. ULSES: That would be my next
question. Show me.
MEMBER ROSEN: What do I ask for? What
should they provide that shows me that they are doing
that?
MR. ULSES: They can give you a map, I
suspect, just like you got with the normalized power
distribution which has the kilowatts per foot on it
for a bundle. But the maximum value per pin in a
bundle. That would be useful information.
You can get the information about the
minimum critical power ratio that's in the reactor.
You can also get the mapple hugger limits.
However, most PWRs are not going to be
limited by mapple hugger. They are typically limited
by MCPR values simply because they have so much ECCS
injection. Normally loca is not a limiting factor for
PWRs.
But those would be the questions that I
would ask if I wanted to convince myself, and those
are the questions that we do ask when we want to
convince ourselves that the power uprates are not
going to exceed any applicable licensing limits on the
fuel. That information ought to be readily available
to the Committee.
MEMBER SIEBER: Actually, the way all this
is licensed is a little bit misleading. When you go
for a change in license for a power uprate, you are
basically using a demonstration bounding core to show
that you can actually manipulate the fuel in order to
get the power output. Each time you refuel the
reactor though there is a design process that goes on
that specifies how each fuel assembly will be built,
how it is to be oriented in the core, and where it is
supposed to go, plus where all the other ones are
supposed to go because you've shuffled them around.
Each time they do that, they send in an
RSE, a reload safety evaluation, the licensee does,
that says I have followed all the procedures that the
staff approves and I have done all these calculations
and this is a good core. It's a 10-page document,
which is what they get. So the process is approved by
the staff, and then each reload is approved by saying
I followed the process.
MR. ULSES: Right.
MEMBER SIEBER: So that is the kind of
paper flow.
MR. ULSES: That is for a plant that's at
a given power level and they are just reloading it.
MEMBER SIEBER: Well, what will happen
here too.
MR. ULSES: Sure.
MEMBER SIEBER: How they have licensed a
plant to go to a higher power and they have changed
their machinery around to achieve that, that the next
reload that goes in is going to have an RSE that's
going to be reviewed by the staff using the same old
process as General Electric always used or
Westinghouse or Siemens or whomever. That is the
process. There is no change to the process and
there's no change to the analysis that they will do.
They will use the same tools.
MR. ULSES: The reason why they don't do
an actual calculation on what they expect the real
power uprated reactor to be is that when we're in the
review process, they are not going to know what target
they are shooting at because they don't know exactly
where the actual real core will be at the end of a
cycle.
So they try and do a generic analysis to
give us an understanding of what it is going to look
like, but then you are certainly right, that they will
use the standard reload process. We will get what
actually nowadays is called a core operating limits
report, but it's the same thing. It basically
describes the fuel that's in the reactor, the method
used, and a summary of a few key results which are the
thermal limits.
MEMBER SIEBER: Then during core
operation, you take flux maps or in core instrument
readings to determine how well core is reproducing
what the calculation showed in advance of refueling
the core?
MR. ULSES: Right. Exactly. Nowadays
that is done online continuously.
MEMBER SIEBER: So you get a map out of
the computer that looks like the map they gave us,
which is an analytical map as opposed to a flux map.
MEMBER WALLIS: Do we have enough to go on
those in Committee?
MEMBER SIEBER: The change I would suggest
is the same one I said yesterday. All they have to do
is change one word.
MEMBER WALLIS: I have concluded that this
power uprate is achieved by having a new fuel.
MR. ULSES: That's true.
MEMBER WALLIS: Ten-by-ten instead of
nine-by-nine.
MR. ULSES: That's correct.
MEMBER WALLIS: And by using new fuel
management techniques.
MR. ULSES: That's correct.
MEMBER WALLIS: Which are so complicated
in detail that there's no way that this Committee
should try to explain them in a letter.
MR. ULSES: I definitely wouldn't try to
explain them in a letter myself. It is an extremely
complex process that has evolved over several years.
CHAIRMAN APOSTOLAKIS: Have we achieved
what the purpose of this meeting was?
We certainly appreciate your coming down.
MR. ULSES: No problem.
CHAIRMAN APOSTOLAKIS: On such a short
notice.
MR. ULSES: I hope that we have
straightened this out. Information on the first slide
is basically a summary of what they are doing to
achieve these power uprates. You can have this
information if anyone is interested.
CHAIRMAN APOSTOLAKIS: Thank you very
much. You answered a lot of good questions.
Why don't we recess for 15 minutes. Then
we'll come back and do planning and procedures.
(Whereupon, at 3:36 p.m. the proceedings
went off the record.)
Page Last Reviewed/Updated Monday, August 15, 2016