480th ACRS Meeting - March 1, 2001
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
480th Meeting
Docket Number: (not applicable)
Location: Rockvill, Maryland
Date: Thursday, March 1, 2001
Work Order No.: NRC-097 Pages 1-234
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
480TH MEETING
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
(ACRS)
THURSDAY
MARCH 1, 2001
ROCKVILLE, MARYLAND
The Advisory Committee met at the Nuclear
Regulatory Commission, Two White Flint North, Room
T2B3, 11545 Rockville Pike, at 8:30 a.m., Dr. George
Apostolakis, Chairman, presiding.
COMMITTEE MEMBERS:
GEORGE APOSTOLAKIS, Chairman
MARIO V. BONACA, Vice Chairman
THOMAS S. KRESS, Member
GRAHAM M. LEITCH, Member
DANA A. POWERS, Member
ROBERT L. SEALE, Member
WILLIAM J. SHACK, Member
JOHN D. SIEBER, Member
ROBERT U. UHRIG, Member
GRAHAM B. WALLIS, Member
F. PETER FORD, Invited Expert
NRC STAFF:
RALPH CARUSO
TIM COLLINS
NOEL DUDLEY
CHRIS GRIMES
BILL HUFFMAN
RALPH LANDRY
RONALD LLOYD
JOHN NAKOSKI
BOB PRATO
HAROLD VANDERMOLEN
OTHERS PRESENT:
LYNETTE HENDRICKS
ROBERT HENRY
SCHEDULE AND OUTLINE FOR DISCUSSION
480TH ACRS MEETING
MARCH 1-3, 2001
THURSDAY, MARCH 1, 2001, CONFERENCE ROOM 2B3, TWO
WHITE FLINT NORTH,
ROCKVILLE, MARYLAND
1) 8:30 - 8:35 A.M. Opening Remarks by the ACRS
Chairman (Open)
1.1) Opening statement (GEA/JTL/SD)5
1.2) Items of current interest (GEA/SD)6
1.3) Priorities for preparation of ACRS
reports (GEA/JTL/SD)
2) 8:35 - 10:00 A.M. RETRAN-3D Thermal-Hydraulic
Transient Analysis Code (Open/Closed) (GBW/PAB)
2.1) Remarks by the Subcommittee
Chairman . . . . . . . . 8
2.2) Briefing by and discussions with
representatives of the Electric Power
Research Institute (EPRI) and the NRC
staff regarding the EPRI RETRAN-3D
thermal-hydraulic transient analysis
code, associated staff's Safety
Evaluation Report, and resolution of
issues previously raised by the
ACRS . . . . . . . . . . 9
[Note: A portion of this session may be closed
to discuss EPRI proprietary information.]
3) Subcommittee Report (Open)(JDS/GEA/MWW)
Report by the Chairmen of the Plant Operations
and Reliability and Probabilistic Risk
Assessment Subcommittees regarding the South
Texas Project Exemption Request that was
discussed during a meeting on February 21, 2001
. . . . . . . . . . . . . . . . . . . . . . . . .36
4) 10:15 - 11:45 A.M. Interim Review of the License
Renewal Application for Arkansas Nuclear One,
Unit 1 (Open) (MVB/GML/NFD/SD)
3.1) Remarks by the Subcommittee
Chairman . . . . . . . .88
3.2) Briefing by and discussions with
representatives of the Entergy
Operations, Inc. and the NRC staff
regarding the license renewal
application for Arkansas Nuclear One,
Unit 1 and the associated staff's
Safety Evaluation Report92
5) 12:45 - 2:15 P.M. Spent Fuel Pool Accident Risk
at Decommissioning Nuclear Power Plants (Open)
(TSK/DAP/MME)
4.1) Remarks by the Subcommittee
Chairman . . . . . . . 111
4.2) Briefing by and discussions with
representatives of the NRC staff
regarding significant findings and
recommendations of the final report on
spent fuel pool accident risk at
decommissioning plants, new
developments, status of developing
proposed options, and related matters.
Representatives of the nuclear industry
will provide their views, as
appropriate. . . . . . 112
6) 2:30 - 3:45 P.M. Management Directive 6.4
Associated with the Revised Generic Issue Process
(Open) (TSK/AS)
5.1) Remarks by the Subcommittee
Chairman . . . . . . . 188
5.2) Briefing by and discussions with
representatives of the NRC staff
regarding Management Directive 6.4
related to the Revised Generic Issue
process, results of the case study
performed to determine the
effectiveness of using the Management
Directive to implement the revised
Generic Issue process, and related
matters. . . . . . . . 188
. P-R-O-C-E-E-D-I-N-G-S
8:30 a.m.
CHAIRMAN APOSTOLAKIS: The meeting will now
come to order. This is the first day of the 480th
meeting of the Advisory Committee on Reactor
Safeguards.
During today's meeting the committee will
consider the following: RETRAN-3D Thermal-Hydraulic
Transient Analysis Code, Interim Review of the License
Renewal Application for Arkansas Nuclear One, Unit 1,
Spent Fuel Pool Accident Risk at Decommissioning
Nuclear Power Plants, Management Directive 6.4
Associated with the Revised Generic Issue Process,
and Proposed ACRS Reports.
I would like to note some changes to the
agenda. RETRAN-3D and ANO-1 license renewal
application were discussed by cognizant subcommittees.
As recommended by the chairman of the subcommittees,
there will not be presentations either by the staff or
by the industry groups on these matters. Instead the
subcommittee chairman will provide reports to the full
committee. Representatives of the NRC staff will be
present to answer any questions from the members.
In addition, the subcommittee report on the
South Texas Project Exemption Request scheduled
between 1:00 and 1:30 P.M. on Friday, March 2, will be
held today following the subcommittee report on
RETRAN-3D.
After completing the subcommittee reports,
the committee will discuss the proposed ACRS report on
the regulatory effectiveness of the ATWS Rule. I hope
these changes will not cause any inconvenience to the
meeting participants.
This meeting is being conducted in
accordance with the provisions of the Federal Advisory
Committee Act. Dr. John T. Larkins is the designated
federal official for the initial portion of the
meeting.
We have received no written comments or
requests for time to make oral statements from members
of the public regarding today's sessions. A
transcript of portions of the meeting is being kept
and it is requested that the speakers use one of the
microphones, identify themselves, and speak with
sufficient clarity and volume so they can be readily
heard.
I will begin with some items of current
interest. We received from Dr. Powers draft one of
the research report on February 26 and some additional
sections yesterday. The most recent version of the
research report along with assignments for reviewing
various sections will be provided to you this morning.
Members should not only review the sections
assigned to them but also should review the entire
report and be prepared to provide their views during
the discussion of the report this evening.
Representatives from the Office of Research will
attend the meeting to respond to questions from the
members.
I would also like to bring the members'
attention to this pink items of interest report, in
particular items referring to management changes.
There have been some senior management changes.
Also to the announcement that the NRC will
hold a workshop on the initial implementation of the
reactor oversight process on March 26. Also the
agenda and registration information for the NRC 13th
Annual Regulatory Information Conferences included in
this document.
I think we are ready to start the meeting.
The first item on the agenda is RETRAN-3D Thermal-
Hydraulic Transient Analysis Code.
Dr. Wallis, will you guide us through this
and give the report to the committee.
DR. WALLIS: Thank you, Mr. Chairman. We
met on the 20th of February, last week, with
representatives from EPRI and from industry, the users
of EPRI code RETRAN.
Finally, with the technical folks who are
consultants for EPRI who actually put together the
code. We actually this time had discussions of
technical matters which had eluded us two years
previously and for some time in between.
Mr. Swindelhurst from the users gave us the
familiar story, RETRAN is being widely used, ACRS
concerns have been addressed, and everything is fine.
We then turned to Mark Polson, the technical
man. He made a technical presentation. Before long
he realized that our critique had some merit. He said
things like, "Oh, I see where you're coming from," and
expressions like that which it was quite nice to hear.
After this had gone on for an hour or two,
the new program manager from EPRI who wasn't here the
previous time, Jack Prahl, asked to make a statement
and he essentially wished to admit that there were
problems with this code and EPRI had something to fix
up.
Then we went on with more technical details
and more discussion with Mark Polson and he saw even
more clearly some of the places from where we were
coming. It was decided from the results of these
discussions that it would not be appropriate for EPRI
to make a presentation today before the full committee
which was originally planned.
That's a brief overview of what happened
last week.
Now, I think it will be good for this
committee to discuss quite a few points raised or
lessons learned from all this. I'm sure you have some
other than the ones I wish to point out at this time.
First one is the ACRS role. It seems that
without the willingness of the ACRS to actually look
at equations and question them, none of this might
ever have happened. One wonders if it really should
have to come to the ACRS in order for this sort of
review of equations to occur.
The staff has issues with the SER
and one might wonder what the mechanism is now for
closing the loop on these issues. Will the
documentation have to be changed since the code
reflects the documentation?
And since the problem with the RETRAN
momentum equation is the supposed resolution, if the
momentum flux turns in arbitrary direction psi which
leads to peculiar results, this presumably is in the
code. Will the code have to be changed? If the code
has to be changed, the evaluations of the code will
have to be rerun and so on. There are a lot of issues
about what should happen now.
The shorter-term issue for us is what should
we do now. I understand the latest proposal is that
this committee write Larkins a very short letter --
you may even have the draft of it here -- and append
the two documents that I prepared, the tutorial on the
momentum equation and the detailed critique of RETRAN
which doesn't necessarily capture everything in there
which might need examination and change.
I think my colleagues might consider at this
point what we are aiming at as a resolution of these
matters. What do we hope to change? Do we hope to
change the way things are done around here? Do we
want to change the way reviews are performed? Do we
want to change substantially the standards required
for code documentation?
Do we want to change RETRAN code itself?
What is it that we hope to achieve by our actions
today and in the future on this code and other codes?
There are several questions. Maybe the chair would
like to decide which ones to take up first.
CHAIRMAN APOSTOLAKIS: I'm not sure this is
the appropriate time to do this. We will perhaps
revisit these questions when we draft this short
letter and decide what the attachments should be.
DR. POWERS: I wonder why you think it's not
the appropriate time.
CHAIRMAN APOSTOLAKIS: This is supposed to
be a short proposal. The questions that Dr. Wallis is
raising are require a lot of discussion.
DR. WALLIS: The staff is or are here,
depending on grammar. There are members from the
staff here and this is your chance to have a
discussion with them about perhaps how we got here and
where we go from here.
CHAIRMAN APOSTOLAKIS: It seems to me that
regarding the code we have to make sure it's correct.
But the other questions you're asking, where do we
want to go and whether we want to make proposals
regarding the review process, I mean, I don't think
this is the right time to discuss that.
DR. POWERS: It seems to me that the review
process itself is pretty good. I mean, I am impressed
at all the things that are going on, getting the code,
running it, very carefully going through all the
things like that.
Now the question of the documentation and
what not, it seems to me it's not a change. It's
simply exercising and reinforcing the standards in the
technical community in general.
The documentation simply has to accurately
reflect what is done in the code and has to be
technically correct. You cannot have scalar
quantities treated as vectors. That's just
unacceptable, or vice versa.
CHAIRMAN APOSTOLAKIS: There are bigger
issues here, though. I think that's what Graham
implied. Why did it come to ACRS having to check the
equations and find that they were not appropriate and
so on? That should have been done somewhere else in
the process. Is it the job of this committee to check
equations and find mistakes?
I think that is an issue that we certainly
need to discuss and maybe try to come up with some
recommendations to the commission that will correct
the process because we should be reviewing whatever we
like but it seems to me that, you know, this should
not be the place where errors of the type that Dr.
Wallis identified should be found.
In that sense I don't think it's appropriate
to discuss these bigger issues. But we definitely
want to make sure that there is documentation of the
concerns and so on. This is simply intended to
document these concerns.
Dr. Wallis has transmitted to us two reports
that are in the handout No. 2, the first two, Comments
on EPRI Response to RAI and Other Recent Submittals
Concerning the RETRAN Code dated February 25 of this
year. The second one is Tutorial on Momentum
Equations dated January and February of this year.
The thought was simply to write a short
letter to the EDO transmitting these two documents at
this time. That's all the action we're going to take
and wait for EPRI response.
Is that correct, Graham?
DR. WALLIS: Maybe you don't want to do it
now but I think we have to have some idea of where we
think things are going and where they ought to go
because if we wait, we don't know what we're going to
get. We may get off on some track which isn't going
in the direction we would like things to go.
CHAIRMAN APOSTOLAKIS: Like which way?
DR. WALLIS: I don't know. This committee
needs to decide what its role is. We could stand back
and say we've given our input. Now we'll wait and
see. Whatever comes back, we'll respond to that when
we see it.
Or we could give more indication of where we
would like things to go and what we would regard as a
suitable resolution of the issues so that when folks
come back to us, whenever they do come back to us, if
they come back to us, with what they think is a
resolution of the issues, that they don't find that
our expectation was something different.
CHAIRMAN APOSTOLAKIS: Would you raise these
issues and reflect our thoughts on the matter in the
letter?
DR. WALLIS: No. I think we should probably
discuss this at some other time, George, than in this
meeting.
CHAIRMAN APOSTOLAKIS: That's what I'm
saying. That would make a very difficult letter.
DR. WALLIS: Since we have two members of
the staff here, do we want to ask them about the
mechanism for closing the loop?
This is something that Virgil Shrock raises
rather strongly in his comments is that we go through
all these motions but the SER is out there and unless
somebody follows up on these things the issues may
just fade away and people get tired of them and
nothing will happen. What is the mechanism for what
one could call closure on these issues?
DR. LANDRY: Mr. Chairman, Ralph Landry, NRR
staff. As we discussed with the subcommittee, our
position at this point is we have prepared an SER
based on the documentation which we received on
RETRAN-3D.
Now, since we have been involved very
heavily with the subcommittee in the review of the
code, the errors in the documentation that was
presented on the momentum equation, we've expressed
our view that the approach that was taken in preparing
this documentation was very difficult to understand
and very difficult to follow through.
EPRI attempted to derive a momentum equation
from basic principles and in that process ended up
with material that was very hard to follow through
and, quite frankly, we would agree with Dr. Wallis
that it's highly suspect and there are errors in it.
We pointed out a number of errors to EPRI and their
consultants ourselves in addition to the errors that
Dr. Wallis pointed out.
Our suggestion in front of the subcommittee
was that the documentation should be retracted and a
presentation should be made of what is in the code
with regard to a motion equation, momentum equation,
however you want to term it, what are the terms in
that equation, what do they represent, and how can
they be justified rather than a derivation from basic
principles.
EPRI in their presentation to the
subcommittee indicated, as Dr. Wallis said, that they
recognized the problems in what they had in the
documentation. They were going to go home and do some
further work.
At this point the staff is waiting to see
what that further work is because we don't want to
dictate to them what they should put in the
documentation. It's their job to come up with the
documentation. We want to see what is in that
documentation, is it correct, and is it in acceptable
form.
At that point we would entertain the idea of
writing an addendum or a supplement to our SER. We
have done that in the past. In the years gone by in
code reviews there have been numerous SERs which had
supplements and addenda written to them which
explained further information or evaluated further
information that had been received. We would be more
than willing to do so should they provide information
that is reviewable that would correct what we see as
shortcomings in the documentation today.
That's the approach that we're on on RETRAN.
Some of the other points that Dr. Wallis has brought
out are a bigger picture discussion and those I think
we need to discuss more fully and how we approach
these reviews.
From the perspective of the staff, the
review that has been undergone in the code in the
recent two years with the TH subcommittee has been
very good and very fruitful. We've had a very good
relationship and very good interchange of information
with one another and this has been a great benefit to
the staff.
In this process we have been writing a draft
standard review plan and draft regulatory guide on
code and code review. This has been in itself an
educational process. We've learned a great deal from
these code reviews and from the interactions with the
subcommittee.
DR. WALLIS: Let me ask you about the code
itself. Just to pick one thing out of my critique
here, they have an analysis of a bend. If you throw
out the friction turn and it's a smooth bend, it turns
out the way they formulated it, there's a pressure
rise across the bend for no cause.
If you add these bends together, you've got
a pump with no moving parts and no energy input which
doesn't seem very desirable. If these features are in
the code itself, you have to after the code, not just
the documentation.
DR. LANDRY: That's correct and that is one
of the problems that we also pointed out and point out
in the SER, that you don't get an effect as their
equations would indicate. It would be a very creative
piece of equipment. It's almost a perpetual motion
machine that they have created in their derivation.
It would be very nice to see if they could build one
of these. That's a little sarcasm.
DR. WALLIS: They could have a dot com
company which would flourish for a while.
DR. LANDRY: Until they went bankrupt.
Typical dot com. That gets back, Dr. Wallis, to the
point I was trying to raise a few minutes ago. The
approach that we think would be far more fruitful and
beneficial would be to show what is in the code,
explain what is in the code, and why what is in the
code is correct and acceptable.
Right now we're going down one path with
documentation and that may not match up with the code.
We keep saying to the applicant, "You should come back
here and explain the code and why the code is
acceptable. What is in the code, not necessarily what
is in the documentation.
MR. LEITCH: Dr. Landry, could you explain
what is the status of the SER now and what use would
be made of the SER where we are at this point in time?
DR. LANDRY: The SER has been issued to the
staff and I believe the SER has been released into the
public sector by the project's office. That means
that those who would like to use RETRAN-3D can come in
and reference the SER.
But that doesn't mean that anyone
referencing the SER and RETRAN-3D is completely clean.
There are 45 conditions and limitations stipulated in
the SER on RETRAN-3D which puts a severe restriction
on anybody using the code in that they must come in
and justify every option chosen.
In most applications of the code provide
adequate assessment because the assessment is so thin
in the documentation for the application of the code.
That puts a great deal of onus on anyone who wants to
use the code in that they must completely justify what
they are doing. They must justify the code.
MR. LEITCH: But even with that
justification there would still be another cloud over
that work.
DR. LANDRY: At this point there is in that
we are very concerned about what is actually in the
code now. This issue has been raised and we are
flagged when anything comes in referring to this code,
that indeed we must understand what is in the code
first.
MR. LEITCH: I was just wondering if
licensees might be spending a great deal of effort
developing work in this regard only to find that it's
unacceptable.
DR. LANDRY: At this point we are not aware
of a number of our licensees using the code for a
licensing application. There is one licensee that has
submitted a reference to RETRAN-3D but that was to use
RETRAN-3D in a RETRAN-02 mode as a substitute for the
old version of RETRAN.
We have put very strict stipulations on how
that can be done within the SER. Only one who has
been approved for use of RETRAN-02 can use RETRAN-3D
in a RETRAN-02 mode and then we specified what that
entails.
The applicant in question is not a licensee
who is approved for use of RETRAN-02. So that raises
an issue in itself and we simply ask that licensee to
demonstrate how they satisfy all conditions and
limitations stipulated in RETRAN-3D SER. When they
get to the stipulation that they have to be approved
for RETRAN-02 to begin with, they are going to run
into a road block.
DR. BONACA: I had a question about what are
the genetic implications of these findings to other
codes such as RELAP-5, such as TRACK, that are being
used now for best estimate calculations? I mean, do
we expect to see the same kind of issues or problems
there?
DR. LANDRY: In this discussion a number of
issues have come up with the formulation of momentum
that point back to work that was done back in 1974 and
even before. The issues at that time that were
brought up pertained to the formulation of momentum
for the RELAP-3 and RELAP-4 codes which are the basis
for RETRAN family codes which actually goes even
further back. It goes back to FLASH. RELAP-3 came
from FLASH.
This issue so far after looking at the other
codes would not apply to the RELAP-5 and TRACK family
because those codes started from a different
derivation and different basis.
They drew on the work on RELAP-4 but the way
in which they constituted the continuity equations was
different than was in the older versions of the RELAP.
We have not gone back and checked exactly what's in
there but the formulation is different for the newer
versions of the codes.
DR. BONACA: I believe it would be
appropriate at this time to look back into those codes
and see if the same issues apply just because, I mean,
clearly, I agree with you, there is a totally
different formulation.
DR. LANDRY: We did raise that issue when we
were doing the Siemens S-RELAP-5 review for an
Appendix K application to small break LOCA. That
question came up because of typos and other errors we
found in the documentation.
One of the lead engineers that they now have
at Siemens came in and gave a cogent, very good
explanation of what is in the code and justification
for the way momentum is formulated in the S-RELAP-5
which is going to be the same essentially as the
RELAP-5 code.
Their explanation was far more justifiable
and indicated that far greater support for the
formulation of momentum that they have than we can
point to at this point in the RETRAN codes.
We don't know absolutely that what is in the
RETRAN family is wrong. What we have in the
documentation is not supportive of it.
DR. WALLIS: Apparently the code is based on
the equations and equations have this strange way of
resolving momentum fluxes which led to this pressure
rise around the bend which seems, since you have the
code, you could look at how they model bends they made
in piping, loop seals and things.
We had a discussion with them which was
inclusive of how they model the cold laid down comma
transition which is a bend in there, we're looking at
it, and they had some very strange terms in that one.
It is possible to look in the code and say
what does the code actually have. You may be
surprised. The code may have something else. But if
the code reflects the documentation, then presumably
these bends are doing the same sort of thing that the
bend in the documentation was doing actually in the
code.
DR. LANDRY: We would agree with you, Dr.
Wallis, and that's why we've said that our
recommendation to the applicant is that they explain
what is in the code and justify it. Generally when a
code of this nature is used, you don't actually model
bends and calculate angles and change in flow
direction but you but nodes together with junctions
where you have a farm lot.
DR. WALLIS: RETRAN makes a big thing about
not having that. They actually have size and
mysterious things which enable them to handle things
like bends.
DR. LANDRY: We've asked for this to be
explained.
DR. WALLIS: It would be very strange if
they have in documentation all this new theory about
bends and the code is still the old straight pipe
junction.
MR. CARUSO: Dr. Wallis, this is Ralph
Caruso from the Reactor Systems Branch. I think
listening to all this discussion I would like to
inject a little bit of, I'm going to say, brutal
honesty here. We don't really believe there's a
problem with the RETRAN code itself. We believe that
the problem is the documentation.
The RETRAN-3D is a transition code. It's a
new version of the RETRAN family and because the
RETRAN community is trying to take the user community
and bring it along to a new version of the code, they
had to come out with something that they thought would
be attractive to the existing users.
We believe that unfortunately in developing
the documentation they attempted to describe it in
terms which give it more credence than it necessarily
deserves. They oversold it a bit.
DR. WALLIS: This is a strange statement.
You mean the actual practice has no relationship to
the theory?
MR. CARUSO: We believe that the way the
code is set up -- as Ralph said, the problem is the
documentation does not reflect what's in the code.
The code and the documentation don't agree.
The documentation attempted to derive the
momentum equations from first principles and show how
they were applied in the code. Unfortunately, they
are not applied that way.
DR. WALLIS: It almost implies that the ACRS
should recode.
MR. CARUSO: Well, this goes back a little
bit further to, I guess you could say, the strategy
for doing this code review from the start.
We understood that the underlying structure
of the code was essentially the same as RETRAN-02 and
we made a conscious decision at that point that we
were not going to review that underlying structure and
those underlying equations.
The existing code, that structure, had been
reviewed, had been approved, and has been in use for
a large period of time and it generally seems to
produce reasonable results that can be used by people
to analyze the behavior of the plans.
DR. WALLIS: How long is this reasonable
comparative time?
MR. CARUSO: Oh --
DR. WALLIS: Is it 20 years?
MR. CARUSO: Something on the order of 20
years. The documentation that's cited in RETRAN-3D is
very much like the documentation. RETRAN-3D is 20
years old. It's an old report from INEL, I think. Is
that what you're referring to?
DR. WALLIS: No. What I'm referring to is
the documentation that you saw for RETRAN-3D as new
documentation.
MR. CARUSO: It's very much the same as was
in the 20-year-old document from Idaho. Same sort of
treatments of bends and things there. It hasn't
changed.
DR. BONACA: What concerns me is that there
was a departure from RELAP-4 when RELAP-5 was
developed. Unless the agency was totally wasteful
with its money, it was done intently because it was
recognized that RELAP-4 was not capable of being a
good prediction code. Is this correct?
I mean, that's the history of that time.
That's why there was a departure. My concern is that
whatever you do to patch up RETRAN, which is a
derivation of RELAP-4, you may not be able to achieve
what you want, achieve in a prediction fashion for the
very reasons that led the whole industry and the
agency to go to RELAP-5 to develop all new formulation
of these equations.
My concern is that here we have -- I'm
expressing this concern because this has been
discussed with EPRI for 20 years, this attempt to
bring RETRAN from RETRAN-1 to RETRAN-2 and now to
RETRAN-3D. Next maybe RETRAN will do neutronics or
who knows what.
I mean, is there something mentally wrong
about attempting to take this code and make it do
things it cannot possibly do? The reason why I say
that question is that Professor Wallis brought up some
fundamental issues there. I'm not sure that purely by
changing somewhat the algorithm psi and putting some
correction in terms will solve this issue.
DR. SHACK: Although I think it is true,
what's unique about RETRAN is the introduction of the
psi angle and the attempt to apply a one-dimensional
momentum equation to a 90 degree angle.
I think probably as long as the modeler
avoids that option, it really does essentially
adjunction the model. Those applications are probably
okay.
What you worry about is the occasion when he
actually tries to use that feature that he can take
the momentum 90 degrees to the angle to which he
thinks he's writing a one-dimensional equation. I
suspect that is largely why the code works is that
people by in large don't use that feature.
It kind of floats in there because, I mean,
it's wrong. They fixed up one set of terms but
they're not the worse because the cosine squared term
isn't the problem. It's either zero or one so whether
it's cosine or cosine squared doesn't make much
difference. They have a missing cosine.
DR. BONACA: The problem with that --
DR. SHACK: It's zero and one.
DR. BONACA: The problem with that is this
places the burden on the issuer and the issuer is not
typically an expert in the code.
DR. SHACK: I agree that is a problem, why
one set of codes is really different. I think that is
the unique feature of RETRAN is to introduce this
notion that you can apply a one-dimensional momentum
equation 90 degrees. That makes it different.
DR. BONACA: I knew that.
DR. SHACK: The reason it probably works
much of the time is you don't try to do it too often.
MR. CARUSO: It is also important to
understand that we know that this situation exist and
the question was asked, well, how does this get fixed
in the future? How do we know that someone doesn't
make a mistake?
The staff process for approving the use of
these codes has several steps. The first step we've
just gone through is to improve the generic topical
report but then each application has to be reviewed
and approved specifically by the staff. We do ask for
copies of the actual plant models as part of those
approvals.
The staff will actually see the models that
will be used by the RETRAN users when they want to
apply them to their plants. Now that we are aware
that this situation occurs, we can be alert to it and
say, "Well, wait a minute. How are you modeling this
1-D momentum through 90 degrees in your plant model?"
DR. WALLIS: But you still have a problem.
Which psi will you accept? No psi is really right.
MR. CARUSO: Actually zero.
DR. WALLIS: No psi is really right.
MR. CARUSO: From our understanding of the
way the RETRAN users actually use the code, they don't
use that factor. They just put in form losses at the
junctions.
DR. WALLIS: But they have to. You have to
put in something for your momentum flux terms and
something for your inertias terms. The L1s and L2s
themselves, there is a question about how they fit in,
too, when you go around a bend.
There are all of these questions about how
this fundamental equation is used for various
components. Are you going to examine every component
in the reactor to see if they use the psi and which
one did they use, if they used L1 or L2, and how did
they choose it and all that.
MR. CARUSO: Actually, the EPRI people have
also made a commitment to have the new users of
RETRAN-3D submit their models to peer review panels so
that there will be experienced users that look at the
models that are developed to make sure they are not
doing things in a too creative way.
DR. WALLIS: I guess there was a concern
that the consultants, particularly Novak Zuber, who
has been around us for four years or so, if you don't
go after some of these conceptual problems when you
review the codes, errors get embedded and they go on
for decades.
I don't know how you fix that but if you
back off or if you say we'll fix it when it comes to
a given utility using and so on, then this blemish
stays there and will resurface again. You have to
deal with it forever.
DR. LANDRY: That, I think, is one of the
points that Ralph Caruso was referring to a moment ago
that, yes, RETRAN-02 was reviewed and approved. The
way we approached the RETRAN-3D review initially was
we would look at the material that was new and
different from RETRAN-02.
Rather than expend resources on reviewing
the old code, we would only look at the new material.
During the course of this review in all these
discussions we've had, we found that we had to go back
and look at the old material also.
This is part of the learning process we've
been going through in these reviews in the past two
years. Perhaps what seemed like the expedient thing
for use of resources to only look at new material is
not the way we should approach the reviews.
We should be a little more sensitive to a
code of this nature looking at the older material, the
older version of the code also and take into
consideration that perhaps there are things in the
review of the old code that need to be re-reviewed as
we move into the new version of the code.
That's quite different than when we receive
a brand new code that we haven't reviewed because
there we would want to look in detail at the entire
code. This is a way we have approached reviews in the
past. We have continued doing this code looking at
only the new material.
Now we realize that perhaps that isn't the
best way and that we do have to look at old material,
too, so we don't perpetuate a problem from version to
version simply because it's been grandfathered in.
MR. CARUSO: And to be honest with you,
these problems exist in the plants. We have plants
now that were licensed back in the '60s and the way
they did things back then is not the way we would have
them do them now.
When we have new license issues come in for
those plants, we don't restart the entire review of
the plant from ground zero. We make a conscious
decision to limit the scope of our review.
I understand the problem of grandfathering
in decisions which seem to be a good idea at the time
but which in 20 years hindsight may not be appropriate
for current times. We don't have the resources to
review everything from ground zero every time we have
a change to it. We just can't do that.
DR. BONACA: But even for all the plants if
you find a fundamental problem, you reopen the issue,
right?
MR. CARUSO: The issue is what is a
fundamental problem? In the case of RETRAN it does
model the behavior of the plant reasonably well, well
enough to make a decision. The question is is it
doing for the right reason. Is it doing it for a
technically defensible rigorously defensible reason or
is it a simplification? How simplified can these
equations be before they become undefensible?
DR. BONACA: But isn't a determination you
have to make before you make a decision?
MR. CARUSO: But it's a judgment decision
and right now what we've been going on is do the
results look reasonable. Can somebody who is
reasonably knowledgeable use this code to produce an
analysis of the plant behavior.
Although there is a problem with the
documentation, we believe that the code as used by
those users still gives reasonable results.
DR. WALLIS: Maybe we've said enough at this
time on this issue.
CHAIRMAN APOSTOLAKIS: Now, regarding the
other issue you raised, which direction we want to go,
maybe we ought to discuss this at another time after
perhaps you give us some options based on your
experience. You obviously have thought about it.
It's always good to have a structured discussion,
especially among 10 people to have some structure,
some starting point. Would you be willing to do that?
DR. WALLIS: Do you think we would do it
this Saturday if we're still here?
CHAIRMAN APOSTOLAKIS: We may start this
Saturday because let's not forget we have huge task to
complete at this meeting, namely the review of the
research report. I'm not sure we will be ready by
Saturday. Again, all I'm asking is two or three
lines. It's not a major understanding.
Any other comments on this issue from any
members?
Thank you very much, gentlemen.
We can proceed now with the chairman's
report on the South Texas Project Exemption Request.
It was a joint meeting of the plant
operations and reliability and risk assessment
subcommittees. Mr. Sieber, chairman of the plant
operations subcommittee, will take the lead on this
and I will jump in as necessary.
MR. SIEBER: Actually, since these 10
members were present for the joint subcommittee
meeting, all this will be sort of a review as opposed
to new material.
I guess I viewed this whole process from an
operating standpoint as opposed to a PRA standpoint.
The meeting that we had on the 21st involved the
process, the element of categorization. We all got a
packet of material on February 8 which most of my
remarks are based on that packet of information.
My approach to doing these things is
actually to first look at the plan itself and try to
compare the numbers and logic that they use versus my
memory of how these plants actually go together. I
used the NRC database to look at the characteristics
of the plant.
There are two units there. They are 4 loop
PWRs. They are large united rated at 1250 MW
electric. It's owned by Houston Power and Light.
It's about 80 miles from the city of Houston. It's a
lake cooled plant and it does have some unusual
features that affect its risk profile.
One is that it has three safety trains
including three diesel generators which most plants
have two safety trains. The safety trains starting
from cold shutdown going up to the high pressure
systems includes three RHR systems, three low-head
safety injection systems, three intermediate head
safety injection systems and what you would ordinarily
think of as high-head safety injection there is
actually charging pumps.
There are three of those even though from an
accident standpoint I would discount one because it's
a positive displacement pump 35 gallons a minute which
I don't think help you much in an accident situation.
DR. POWERS: It's worse than that, Jack. It
would probably hurt you in an accident situation.
MR. SIEBER: It's there and drawing power
and doing nothing.
DR. POWERS: And it's putting the reactor
into hot clad.
DR. UHRIG: Jack, am I correct in
remembering that this is sort of a unique plant that
has a longer core than the standard Westinghouse 4
loop plant? This is different than almost all other
4 loop plants?
MR. SIEBER: I think it's another foot
longer.
DR. UHRIG: Yeah. Yeah.
MR. SIEBER: It has more elements, for
example, by about 32 elements than a 3 loop plant
which has about 157 elements in it.
DR. UHRIG: You know, the SNUPS design was
the standard plant of that era.
MR. SIEBER: Right.
DR. UHRIG: I believe this one was supposed
to be the next generation plant.
MR. SIEBER: Right. And it is lake cooled
which is not unique in the United States. There are
an awful lot of lake cooled plants. It has large dry
containment.
Now, the exemption request itself, as I see
it, it's purpose is to identify components that are
important to safety from a risk standpoint and
eliminate components not important to safety from the
requirements of Title 10 CFR, Part 50, Appendix B and
Special Treatment Requirements. I see 50 isn't on
there.
The other thing is that it is also designed
to identify components which are risk significant but
don't end up on the Q-list so that they can be added.
This process actually goes both ways.
I tried to look at the number of components
that they had. Of course, they listed their totals as
for both units and specifically for 29 systems. A
typical PWR might have anywhere from 50 to 60 systems,
but there is no point in trying to categorize safety
related or nonsafety, things like drinking water,
building ventilation and so forth.
The ones that have some significance at all
are the 29. If you look at this in a typical BWR, a
single unit, it will have about 17,000 valves, another
17,000 circuit breakers or electrical components,
motors and so forth, about 300 pumps and about six
other heat exchangers, and a myriad of other things
which for two units would be about 70,000 total
components in the plant that are assigned mark numbers
of one sort or another.
In the 29 systems there are 43,690
components in the two units. If you would look at
their Q-list, those items falling under the
requirements of Appendix B, there are 16,715
components listed there. These are the ones that are
initially identified either by the nuclear steam
supply system vendor or the architect engineer for the
interfacing systems.
CHAIRMAN APOSTOLAKIS: These are safety
related?
MR. SIEBER: These are safety related
components as originally determined when the plant was
built.
CHAIRMAN APOSTOLAKIS: And there are no
other safety related. This is it.
MR. SIEBER: This is it.
DR. WALLIS: This is the total for two
units?
MR. SIEBER: That's total for two units.
DR. WALLIS: So 16,715 is an odd number.
That means that something is gone?
MR. SIEBER: Yeah. For example, some
systems are shared and some are not. I know of no two
units regardless of how they were built that are
identical.
DR. SHACK: The Inside NRC article on this
made a comment that South Texas dumped more stuff on
the Q-list than the typical plant does. They sort of
hit this thing at the peak.
MR. SIEBER: Well, Appendix B come out in
the early '70s and ours was one of the first plants to
have to adopt Appendix B after construction was
underway and the design was done. Our Unit 1 had
something like 4,000 or 5,000 items on the Q-list.
Unit 2, which was built a year before South
Texas and went commercial, had the broad range and had
about 7,000 items. There was a growth in what ended
up on the Q-list of about, I would say, 35 percent
over that time period.
DR. POWERS: I believe South Texas in the
time it was in construction was one of those plants
that benefitted from increased management attention.
MR. SIEBER: Don't we all.
Now, based on what I have learned, a typical
PRA really covers about 2,400 components per unit.
That leads to some interesting things. When you try
to re-categorize all these items that are on the Q-
list, you actually find out that you can only do that
on the basis of PRA results for 5.7 percent of those
items.
If you want to do the remainder, the only
choice that you have is to do it by expert panel
elicitation which amounts to 94.3 percent. I guess
that makes my eyebrows go up a little bit when I think
about the fact that categorization is "risk informed"
based on PRAs. In my mind, less than 6 percent of the
items are based on the PRA. Everything else is based
on the expert panel.
CHAIRMAN APOSTOLAKIS: No, but I think that
is some additional elaboration that is required here.
MR. SIEBER: Okay.
CHAIRMAN APOSTOLAKIS: I don't think you
mean that because it's not just the numbers. I mean,
it's not the 5.7 percent of SSC in the PRA and the
remaining 94.3 are not. The 5.7 percent are there
because they are important to the CDF and LERF. I
mean, there is a reason why they are there and the
others are not.
MR. SIEBER: Well, you've got to go beyond
that because the others may, one way or another, be
implicitly a part of the ones that are specifically
listed.
CHAIRMAN APOSTOLAKIS: Very good point.
Yes.
MR. SIEBER: On the other hand, you can't go
back and do a Fussel-Vesely or RAW for an item that's
not there.
CHAIRMAN APOSTOLAKIS: Exactly. Exactly.
I think, in other words, we should not be talking only
in terms of the numbers. We should elaborate a little
bit on that.
The other point is that, yes, it does appear
that the remaining 94 percent are really categorized
not using risk information but it was pointed out by
the STP folks when they were here that the reason why
they called it risk informed is because the whole
context within which the characterization takes place
is risk informed.
The fact that these are not in the PRA is
already useful information to the panel because there
is a reason why they are not in the PRA. You're
right. I mean, it's not as formal as using the
importance measures, for example, because you can't do
it.
DR. POWERS: But, George, I think what he's
saying is something more important there. There is
not a case that there is a reason they are not in the
PRA. There are two reasons, two general categories of
reasons.
One, it's not important, and the other one
is that it's implicitly present and the PRA analyst in
order to simplify his model didn't call it out. I
mean, that seems to me that's a very significant
point.
MR. SIEBER: Well, and I think you have to
go a step beyond that, too. When we get finally to
the explanation of how the expert panel does its
business, there is actually risk information in the
questions that they ask and the weighting factors.
In a way it's risk informed but as I still
see it, it's less than 6 percent come directly from
the PRA.
CHAIRMAN APOSTOLAKIS: It's not risk
informed if you interpret risk informed using strictly
numbers.
MR. SIEBER: That's right. These are the
numbers here related as --
DR. BONACA: Just before you go past that,
all that I've heard here is true. The only thing I
want to point out is that there has been a focus on
two measures of performance and if some other measures
were used, probably some other components will have
ended up there.
DR. KRESS: Yeah, I'm glad you said that.
CHAIRMAN APOSTOLAKIS: That's true. That's
very true.
DR. BONACA: There is no doubt in my mind
some of the components we probably question. For
example, the assumption that since it is not an early
release, you don't have to worry about it.
Therefore, you know, small failures of less
than one inch penetrations could affect later releases
in containment and are not considered because that's
not significant to the public. That includes the full
characterization of penetrations.
DR. KRESS: And late releases in general.
DR. POWERS: It's also true that the crucial
systems for shut-down operations aren't going to make
this list here unless crucial equals well under normal
operations.
DR. BONACA: You mean for intermediate
targets?
CHAIRMAN APOSTOLAKIS: I think --
DR. BONACA: I am making this comment
because I believe that there is an issue, at least in
the generic fashion, for ranking we have to derive
which is the issue of having a well-reflected on set
of acceptance criteria. I mean, our CDF and LERF are
the only criteria to use. I mean, we have discussed
that.
CHAIRMAN APOSTOLAKIS: The question really
is, I mean, it is a legitimate question in a sense
but, on the other hand, you might say, "Well, gee, you
guys have approved regulatory guide 1.174 and all this
licensee does is follow 1.174 and that guide says LERF
and CDF." Shall we raise the issue of what is risk
informed regulation every single time there is a case
before us?
DR. KRESS: Yes.
CHAIRMAN APOSTOLAKIS: Then that throws the
process --
DR. KRESS: We've approved a lot of
regulations in the past that have proved to have flaws
in them. I view this as a flaw of 1.174. You see,
1.174 was meant for very specific things. I think
we've carried it well beyond what it was intended for
when we try to make it a generalized way to risk
inform the regulations.
CHAIRMAN APOSTOLAKIS: The next time the
issue of revised or updating 1.174 comes up, I think
this is a legitimate issue. Put yourself in the
situation of a licensee. We have these new regulatory
guides, they want to use them, and then the issue
comes back and they say, "No. Look."
DR. KRESS: Put yourself in the place of the
public and the concern of late releases and land and
sees that NRC is not dealing with that.
DR. BONACA: But they have an expert panel,
too. The expert panel makes judgments that remove
components from a list and may even add them.
All I've got to say is to make this
statement that has been made, that you're going to
have to consider late containment failure because by
the time evacuation has taken place, it defeats
everything we have done in this industry from day one
which is simply you are not going to mess around with
the public issues.
CHAIRMAN APOSTOLAKIS: And why when we look
at license renewal we say the regulations dictate that
we look at it in a deterministic way. All this stuff
about risk and PRA over the last 25 years is not
relevant, all of them.
Why don't we say, "Gee, if you've got that
much frequency above the goal, maybe you ought to do
something more." I would say no because the
regulations say this. I mean, at some point you have
to go by the rules.
DR. SHACK: In this case, George, we don't
have a rule yet. Option 2 is trying to figure out how
to do this.
CHAIRMAN APOSTOLAKIS: They are following
1.174.
DR. SHACK: Nothing says that has to be cast
in concrete.
CHAIRMAN APOSTOLAKIS: I'm not saying it
should be cast in concrete. The issue should be
raised but I don't think it's fatal because then
nobody is going to try these things. They are going
to say, "Wait for 10 years and until those guys in
Rockville decide what's important.
DR. KRESS: I think all we're asking for is
a question and an expert panel to look at it and say
does this particular SSC impact light containment
releases or late containment failure. If the answer
is yes, you give it a weighting factor on the scale
but you put that particular component in with the list
that you have. I maintain it would probably only add
about five or six. Maybe more than that but the
question ought to be asked is my point.
DR. BONACA: And the point again, the
latitude that the expert panel has is very large.
Clearly, they --
DR. SHACK: They don't drop things. If the
PRA says it doesn't get dropped --
DR. BONACA: If a system is rated
significant but has multiple trains to support it,
they are calling them, for example, a lower
significance because they have it on their system.
Now, they are taking quite a latitude.
CHAIRMAN APOSTOLAKIS: That's the same
issue.
DR. BONACA: No, no, it's not the same
issue. I'm saying that the expert panel has a
significant decision making they have established and
I support it. I can question the decision but I agree
that they have the capability.
I think they should also reverse capability.
I think that the issue with these guys here is one
that looking at the generic process we would have to
reflect on and understand.
CHAIRMAN APOSTOLAKIS: I'm having problems
with 1.174 myself. I think when we make comments like
this, we should be aware of the other guy's problems.
If we throw 1.174 out the window --
DR. KRESS: 1.174 has a statement in it that
in addition to the CDF and LERF you will comply with
all the deterministic requirement. Those
deterministic requirements deal with things like late
containment failure releases.
Here we have an exemption that says we don't
have to do light containment because it doesn't affect
CDF or LERF, but it is in all these other
deterministic requirements that you are supposed to
comply with.
CHAIRMAN APOSTOLAKIS: Well, if you put it
that way, I think it's a more legitimate concern in my
view because you're doing it in the context of an
approved guide.
DR. KRESS: I think that was the reason they
left that kind of statement in the 1.174 is to
recognize it wasn't just CDF and LERF.
CHAIRMAN APOSTOLAKIS: In that context,
though, I mean, when you talk about these kinds of
things, the question is whether you should limit
yourself to these big items like core damage and
releases from containment.
A lot of these other requirements are there
to really address the cornerstones of the oversight
process. I mean, we don't want to see initiating
events. We don't want the integrity of the primary
look to be compromised.
A lot of the requirements are there to make
sure that these cornerstones are satisfied. Now if we
come in with a risk approach that says we are going to
look at CDF and LERF and late containment failures,
are we consistent?
I don't think we are because there may be
some requirements there, you know, the staff has made
it very clear we just don't want to see initiated even
though they may not progress to something very severe.
DR. BONACA: The issue of late containment
failure, I don't think they use the PRA for that.
They use some of PRA regarding the fact that the
highest risk is LERF.
MR. SIEBER: When you look and see how they
classify based on RAW and Fussel-Vesely, that is
really CDF and LERF without those extended effects.
I could do it but I don't think as I read the
methodology that they have done it.
CHAIRMAN APOSTOLAKIS: Do you think, though,
that if there was an issue regarding one particular
component that came from the PRA or from the questions
that it was really in risk 2, category 2, and it was
important to late containment failures, do you think
the panel would not be aware of that and perhaps move
it to something else?
DR. KRESS: From what I read in the report,
yes.
CHAIRMAN APOSTOLAKIS: Which may be a matter
of documentation again, the same as it was with the
other thing. If you put it in writing, then the staff
will stop asking questions about that.
DR. KRESS: All I have to go on is what I
have in writing.
CHAIRMAN APOSTOLAKIS: I know.
DR. BONACA: They ask questions. We ask
questions and the answer was because it's a small
leakage so it is minor and there will be no impact.
I mean, it really undermines somewhat my
faith in that expert panel because although you may
rationalize that, dealing with issues that have to do
with the last barrier of the tail end of a major
accident is something that is totally new in this
environment.
DR. KRESS: And their statement, George,
that large early releases prompt fatalities dominate
the risk to me is an unproven assumption. When I say
that, what they mean is if you meet that goal, you
will also meet the latent fatality goal but it says
nothing about land contamination, total injuries,
total deaths. I don't know whether it dominates the
risk because we do not have appropriate risk metrics
for these other things to compare it with.
CHAIRMAN APOSTOLAKIS: But, Tom, we proposed
to the commission to do that and they said no.
DR. KRESS: I know, but I'm a persistent son
of a gun.
CHAIRMAN APOSTOLAKIS: There has to be --
DR. BONACA: All that we have to do is say
this stays in the list because they are significant
and they could affect releases. The whole issue of
performance measures from PRA would be moot if they
had made the call. They didn't. That's why I'm
questioning the call.
I'm not questioning the structure of the
regulation. I'm questioning the call. Maybe then on
a generic basis if those calls can be made, then there
has to be a need for more structured guidance.
CHAIRMAN APOSTOLAKIS: But it's pretty
clear, to me anyway, that when you consider affects
that go beyond CDF and LERF, that the expert panel
probably wouldn't know how that component actually
affected the late release or land contamination or
injuries, I think it's beyond what information the
expert panel would ever have.
CHAIRMAN APOSTOLAKIS: But in terms of the
cornerstones, though, I think the expert panel will be
very much informed. In other words, you know, on an
initiating event that's something that is within the
experience of people.
MR. SIEBER: Does the staff have a comment?
MR. NAKOSKI: This is John Nakoski. I'm the
project manager overseeing South Texas. I would just
like to remind the ACRS members that the staff shares
a concern regarding late containment failure. We have
an open item with South Texas on this issue. We have
asked them to evaluate their categorization process
and consider methods to address that.
One of the alternatives we suggested they do
was to look at their PRA specifically for conditional
containment failure probability with doing a
sensitivity study where they increase the failure
rates of those components important to protecting the
containment by a factor of 10, similar to what was
done for the broader sensitivity study.
For each component really, or system, come
up with an evaluation that says why it's not necessary
to protect a containment. We share the concern that's
being expressed here and we are working with the
licensee.
CHAIRMAN APOSTOLAKIS: Why only the
containment? Why not the other cornerstones? I
thought the whole idea of special treatment was to
make sure that this totality of the deterministic
regulations protect us from public unhappiness. It's
not just health and safety. It depends upon how you
interpret health.
DR. KRESS: What this process will do,
George, is -- what this process will do is focus on
only risk dominant sequences when they do what they
talked about.
If you look at conditional containment
failure probability, I think that's probably an
appropriate way to deal with this late containment
issue because you dealt with the other phase on your
CDF.
CHAIRMAN APOSTOLAKIS: I'm not clear. CDF
is in full sequence. I mean, now the initiating event
itself is something we don't want.
DR. KRESS: Yeah, but it gets involved in
the CDF and they've dealt with it to some extent.
CHAIRMAN APOSTOLAKIS: In some sense but it
doesn't get the same importance.
DR. KRESS: Perhaps. Perhaps.
CHAIRMAN APOSTOLAKIS: The objectives, it
seems to me, have not really been settled.
DR. KRESS: But, you know, if the thing is
not important to CDF and SSC, then it's likely not
real important to the initiating event frequency.
CHAIRMAN APOSTOLAKIS: No, because you may
have an initiating event that has been mitigated with
very high probability.
DR. KRESS: Of course.
CHAIRMAN APOSTOLAKIS: I think we are going
to come back to these things. Right?
DR. KRESS: Right.
DR. WALLIS: I had a question about that.
You have 5.7 percent in the PRAs but 8.78 percent turn
out to be safety risk significant. Presumably the
expert panel added quite a few.
MR. SIEBER: Right.
DR. WALLIS: I just wonder about the
overlap. Are there perhaps things that the expert
panel considers which are really more important than
are in the PRA? There's an overlap there.
MR. SIEBER: I think there are some things
in the PRA that are of low risk significance.
DR. WALLIS: So you might argue that --
MR. SIEBER: So not all 2,400 items that
were in the combined PRAs for those units necessarily
made it to the --
DR. WALLIS: I was just telling you
something about completeness of the PRA. The expert
panel adds things which really are more significant
than some of the things in the PRA. Perhaps those
things should have been in the PRA in the first place.
MR. SIEBER: When we get to the
classifications scheme that the expert panel used, you
can see how, for example, some components would have
ended up being risk significant as far as their scheme
is concerned and not necessarily been in the PRA when
we get to that.
DR. WALLIS: Maybe you could address that
later.
CHAIRMAN APOSTOLAKIS: The panel used
criteria out of CDF and LERF. They actually did.
DR. KRESS: If PRAs were complete and dealt
with uncertainties and dealt with all the modes of
operation such as shutdown or low power, then you
would expect PRA to kick out all the important things.
CHAIRMAN APOSTOLAKIS: That's right. On the
other hand --
DR. KRESS: It's not complete and there are
parts that are highly uncertain, then it doesn't deal
internally with shutdown and other things it doesn't
deal with very well so, you know, you would expect
other questions to be asked.
CHAIRMAN APOSTOLAKIS: It's not just
incompleteness. It's also --
MR. SIEBER: You could not write a PRA that
covered all of these components in my opinion. I
mean, that would be lifetimes worth of work to try to
model all of these subcomponents.
DR. POWERS: I'd like to point out that a
lot of people are making PRA their lifetime's work.
MR. SIEBER: I understand that.
The reason for me putting this slide up is
just to show that there's two different methods of
arriving at determination of risk significance. As
Dr. Shack pointed out, these are additive. You go
through the PRA portion of it to cover the 2,400
components. Then the expert panel does the remainder.
Interestingly enough, they also use the
expert panel as a way to check by doing some of the
PRA components also. It turns out that there was some
consistency there between when they were evaluated
both ways. One way by PRA and the other way by the
expert panel. I would like to talk about the PRA
components first and then the expert panel components
next.
From the PRA results, classification for the
ranking that they got was high, medium-R, which means
that they want to consider it as high so that sometime
in the future if it became reclassified as high, they
wouldn't be stuck without documentation, without
adequate maintenance, without inspections and
surveillance, and all the other things that Appendix
B requires because these things can shift as the plant
is modified. There is additional operating experience
as far as failure rates and so forth are concerned.
Then medium and then lastly low. These are
all based on risk achievement worth and Fussel-Vesely
criteria. That is one of the reasons why the PRA
subcommittee was a part of this to assist to the plant
operations subcommittee.
It's not clear in my mind. These look a
little arbitrary to me. It's not clear in my mind if
these are the right numbers and the right criteria or
not. Perhaps I could ask for comments on that from
anyone who feels --
DR. KRESS: Associated with that question is
if the RAW is 99 --
MR. SIEBER: Right. What do you do?
DR. KRESS: -- take it down to the medium
where the other things are met.
MR. SIEBER: That's where the expert panel
comes in again. We shouldn't take these as rigid
boundaries and so on. I mean, the expert panel does
evaluate the results of this, too.
I think the whole approach here should be to
put things in context. There is a decision that is
made by the panel. In order to make that decision,
they collect information from analysis. One is the
PRA with these kinds of things, the high, medium, and
so on.
They collect information from the five
questions that Jack will talk about in a little bit,
the rates and so on, doing it different ways. They
can decide looking at the individual categories and
then they deliberate. This is really a structured
deliberation. In that context if RAW is 99 is
irrelevant because they will look at it and they will
not say, no, it's not high because it's 99 and the
boundary was 100. The other thing is --
CHAIRMAN APOSTOLAKIS: What would they do
with 90?
MR. SIEBER: Well, they have to make a
judgment.
DR. SHACK: Once you've made the decision,
you do have to check with the sensitivity study. I
claim that's the real decision.
CHAIRMAN APOSTOLAKIS: Exactly.
DR. SHACK: This is the way to select a
group of components to examine that way. If you can't
meet the sensitivity study, then you have to go back
and you'll throw out components that hit 90. You'll
have to go back and keep throwing stuff out until you
can get through the sensitivity analysis.
DR. KRESS: RAW is the sensitivity study.
CHAIRMAN APOSTOLAKIS: Yeah, but the big one
at the end where they increase the federal rates by
10.
MR. SIEBER: I have some questions about
that, too, which maybe I would like to address. First
of all, this classification puts things on the list.
On the other hand, when the expert panel did evaluate
components, they evaluated not only the ones that
didn't show up in the PRA but also ones that did.
It could end up on this new Q-list more than
one way. It could end up there because of the PRA and
this classification scheme, or it could have ended up
there because of the expert panel which is independent
but serves as a check, one against the other.
CHAIRMAN APOSTOLAKIS: No, but this was
input to the panel. There was no categorization
independently of the panel.
MR. SIEBER: That's right.
CHAIRMAN APOSTOLAKIS: The panel has the
final word so this goes to the panel for evaluation.
MR. SIEBER: Right.
CHAIRMAN APOSTOLAKIS: Jack asked where the
numbers come from. It's really experience and
sensitivity.
DR. KRESS: Let me ask you a question. I
would expect that the value of, say, RAW or Fussel-
Vesely that's important would depend on absolute value
in the CDF and LERF. It doesn't show what they want.
Why doesn't it?
CHAIRMAN APOSTOLAKIS: It's relative.
That's one of the problems with these things, that
whether you are at the 10 to the -3 CDF.
DR. KRESS: I understand that. I'm saying
that the cutoff, the threshold ought to depend on the
absolute value and I don't see that reflected.
CHAIRMAN APOSTOLAKIS: It was in the paper
by Geoak, Perry, and Sherry that these numbers and the
actual delta CDF don't relate. Why should you have
the same cutoff value for all plants?
DR. KRESS: If I had a CDF 10 to the -6, why
would I worry about the RAW and the 100.
CHAIRMAN APOSTOLAKIS: Because it can still
make it 10 to -4 which is still acceptable.
DR. KRESS: That's why I think I worry.
These might be plant specific values and I worry about
getting them locked into a system for every plant that
we review. I might not like those numbers for some
plants but I might like them very well for, say, South
Texas.
CHAIRMAN APOSTOLAKIS: I still think we have
to have things in perspective here. I don't think
that any single method they use can withstand the kind
of scrutiny we are giving it here. We expressed a lot
of concerns last time at the subcommittee meeting.
Let's see what we're trying to do here.
Again, this is structured deliberation. The panel
puts them in categories. Then you have two major
things that I think save the day. One is the
sensitivity.
They say, "Okay. Forget about all these
things. We made mistakes. Let's raise all the
failure rates by 10 and see what happens." Then they
find that nothing much happens. That's a very
powerful argument.
DR. POWERS: I wonder how powerful it is,
though. The challenge you always have with these
things is they are one at a time kind of variation and
they are not really partial derivatives.
You tell me that you've raised all these
numbers by a factor of 10 it's a little difficult for
me to put that into perspective. Has anyone ever
taken one of these assessments for any plant, I don't
care which one, and looked at partial derivatives and
second partial derivatives?
CHAIRMAN APOSTOLAKIS: No. This is a very
new idea.
DR. POWERS: Why not? Why shouldn't
somebody do that?
CHAIRMAN APOSTOLAKIS: Because they haven't
thought about it. Nobody's thought about it.
DR. POWERS: You see, it all boils down to
the question of where did the factor of 10 -- I mean,
factor of 10 sounds big but it's not really big. I
mean, we're working in long space here.
CHAIRMAN APOSTOLAKIS: That's my point. The
sensitivity study is one. The second, let's not
forget what the decision is here. It seems to me you
are relaxing some of the special treatment
requirements. What's going to happen?
If you have an impact at all, it's going to
be gradual. You're not going to have a catastrophic
failure tomorrow to 15 components and they will have
a monitoring problem.
DR. KRESS: In reality what you're saying is
special treatment requirements are not very risk
significant in the first place.
CHAIRMAN APOSTOLAKIS: That's exactly right.
DR. KRESS: That's a saving grace made for
here.
CHAIRMAN APOSTOLAKIS: Exactly. That's my
point.
DR. KRESS: But that's an assumption,
George.
DR. SHACK: No. If you pick a different set
of components and change the failure rate, you get a
very different answer. This factor can only work
because you're doing it to a selected set of
components.
CHAIRMAN APOSTOLAKIS: That's correct. But
there are several issues here. First of all, to save
that relaxing the requirements will lead to an
increase in factor of 10 is ridiculous. It's utterly
ridiculous.
Second, you are increasing the failure rate,
not the event itself. It's not going to happen
tomorrow. You're not going to have a huge common
cause failure where all sorts of things fail. I mean,
if these things happen, they will catch them. They
will have a monitoring program.
DR. SHACK: But, George --
MR. SIEBER: This is one of the elements of
this classification scheme, the feedback system, which
comes from the corrective action program. On the
other hand, I guess when I thought about this, I think
of different kinds of plants with different risk
profiles and how this sensitivity study would reflect
itself in those plants.
For example, the South Texas project has a
pretty good risk profile and it comes about because of
the three safety trains. When you increase a
competence failure rate by a factor of 10, is it
really going to show up as being significant in the
profile for that plant?
I would think it would not be as significant
because of the redundancy that is already built into
that plant with the three trains. But if you had two
trains, it may be more significant. For a different
plant, you may reach a different conclusion.
The other kind of plant that I consider is
there are some plants that have relatively high but
acceptable risk profiles and are dominated by a
particular sequence.
If you change the failure rate of a
component not involved in that sequence, it gets
swamped out by the dominate sequence so you may not be
able to draw a conclusion from that either. I think
sensitivity works better for some plants than for
other plants.
DR. KRESS: I'll tell you what bothers me
about the whole process is I have this intuitive
feeling, like George said, that this is not very risk
significant, but my intuition has been wrong a lot.
I don't see a coherence to this process where you
start from the top level.
Our objective is to meet these regulatory
limits on certain things. They are going to be things
like containment failure, total deaths, maybe CDF.
I'm not even sure I would include that. Most people
would but we want to achieve certain frequencies which
you exceed land contamination. Those are all
regulatory objectives. Those are what we're trying to
achieve by the systems and components we have in the
design.
I don't see starting from those things we're
trying to achieve looking at how the plant already
meets those, and determining how each system and
component affects that and whether or not if I put one
in one category or another, whether or not I step over
the balance or get too close to the balance depending
on the uncertainty.
That coherence is just not there for me and
that's what bothers me. It just doesn't hold together
because, you know, you look at this and I don't know
why RAW of 100 is a good number for this plant. Why
is it a good number?
DR. SHACK: This is just a preliminary
screening value. There is a misplaced precision here.
We're talking about numbers that just don't have that
kind of exactitude.
DR. WALLIS: I think they should have some
justification.
DR. SHACK: What you really look at is you
get to these numbers and do you get a change in CDF
and LERF that is significant by the standards of
1.174. It may well be that for other plants when you
go through that final assessment you'll have to use
different numbers. Maybe you could have changed these
numbers and still met that assessment in South Texas.
DR. KRESS: That's the part that's missing.
DR. SHACK: No. That's the consistency
part.
DR. KRESS: It doesn't say how these numbers
were derived from the 1.174 requirements.
DR. SHACK: The question is is it good
enough that when you use these numbers, you meet the
1.174 requirements?
DR. KRESS: I don't know. That's the part
that's missing.
DR. SHACK: No. They do. That's what they
check at the end. They mask their categorization.
Then they do their sensitivity analysis to make sure
they meet the 1.174 requirements. Could they have set
the numbers at 110 and still met it? Maybe. Could
they set them at 90 and still met it?
DR. WALLIS: Maybe for some plants it should
be 1,000 or 10 or something. Maybe it's really
different for some plants.
DR. SHACK: The answer is as Jack said,
you'll get different answers for different plants.
Maybe if you use these numbers and you go to a plant
with two trains, when you make the sensitivity
analysis you'll find out that you don't meet the 1.174
requirements. You'll have to come back and change
these values. You'll have to be more restrictive.
MR. SIEBER: In fact, I see this as an
interactive process. If you applied this methodology
from one plant to another, you would have to go to the
end, do the sensitivity analysis to determine whether
you picked the right numbers in the first place.
As Dr. Apostolakis said, it's basically
experience. I don't have enough experience to say
whether 100 or 110 or 90 is the right number for the
upper boundary. On the other hand, the proof of the
pudding comes from the sensitivity analysis as long as
you understand what that really means because
different plants are going to respond in different
ways to the outcomes of that analysis in my opinion.
DR. KRESS: So we're being asked to believe
that a sensitivity of 10 varying one component at a
time --
DR. SHACK: No, no. All together.
DR. KRESS: All together.
DR. SHACK: All together. They raise them
all by a factor of 10 all at once. It's not one at a
time. Bang, all the non-risk significant components
go up by a factor of 10 all together. To do what you
want to do, you would really have to know how the
special treatment affects the failure.
DR. KRESS: Which I agree is impossible.
DR. SHACK: If we want to stay here until
hell freezes over, we can do it.
MR. SIEBER: There is a more subtle question
buried in that. If you maintain surveillance and have
a good corrective action program and so forth, it
probably doesn't change the failure rate very much.
When I think about it where you don't have
diversity, you may change the common cause failure
rate which I think is perhaps more significant than a
single failure or an increased probability of single
failure. That's my intuitive feeling as opposed to
any proof that that would occur.
It seems to me if you eliminate certain
portions of a consideration for a group of identical
components, if they're going to fail, they're all
going to fail in that mode sooner or later. So that
may have a barring on it. On the other hand, their
treatment of common cause appears to be conservative
in the way they have approached it.
In any event, this is the --
MR. SIEBER: There's a typo in there
somewhere. It cannot be .001 on the top because the
Fussel-Vesely of .002 and a RAW of 1 would be both
high and low at the same time.
DR. WALLIS: There probably is and I'll look
that up and tell you what it is.
MR. NAKOSKI: This is John Nakoski again.
It is on the high value Fussel-Vesely greater than
equal to 0.01.
DR. WALLIS: There are too many zeros.
MR. SIEBER: Okay.
Well, this takes care of the 5.7 percent.
Let's take a quick look at what the expert panel does
with the 94.3 percent. They ask five critical
questions and they rank each component by the
component's sensitivity to frequency of occurrence,
which is demand, and/or the perceived risk impact.
Let's take a look at the five questions.
These are evaluated basically two different
ways. These are the questions that they chose to ask.
I guess one of the observations one could make is that
there is some overlap from one question to another.
It's not totally clear as to how great the answer is.
On the other hand, these seem to be reasonable
questions in my own mind to ask for the purpose of
categorization.
Does the loss of this function cause an
initiating event?
Does the loss of this function directly fail
another risk significant system?
Is the function used to mitigate accidents
or transients?
Is this function directly called out in EOPs
and ERPs?
Does this function directly affect safe
shutdown or mode changes?
Now, they have assigned a specific weight to
each of these questions. If you want to make notes,
"Does the loss of this function cause an initiating
event?" is weighted as three which seems to me a
little odd but that's the way they weight it.
"Does the loss of this function directly
fail another risk significant system?" is weighted as
four.
"Is the function used to mitigate accidents
or transients?" is weighted as five or most important.
"Is this function directly called out in
EOPs and ERPs?" is also weighted as five. At least
for the confidence of the operator, it would be nice
if he knew that everything that was in the EOPs or the
ERPs was operable and would work.
"Does this function directly affect safe
shutdown or mode changes?" is rated as a three.
Now, for each of the questions the component
is rated basically two ways. One is what is the
demand and what is the risk significance in the
component.
Then it is weighted by a scale of one
through five with five being the most risk
significant. You multiply the five times five points
for the question itself times by the weighting factor
which is five and you end up with a maximum 25 or a
minimum of five.
DR. WALLIS: So if they weighted them one,
two, three it would have been just the same.
MR. SIEBER: Well, they --
DR. WALLIS: They all got three for writing
their name on the paper.
MR. SIEBER: That's right.
DR. UHRIG: Jack, this isn't as a yes, no,
zero, one which is then multiplied by three and then
multiplied by five?
MR. SIEBER: No. Actually, the --
DR. UHRIG: The question is is it or is it
not.
MR. SIEBER: The expert panel is actually
instructed by their procedure to rank. Okay? And
that's on the basis of frequency of occurrence or
demands and risk significance. Is that not correct?
It's not a zero one proposition.
For example, and let's go back, if I asked
the question, "Is the function used to mitigate
accidents or transients." When the demand is high and
the risk significance is high, I would rank it as
five. Five times the weighting factor of five is 25
so you get 25 points. Okay?
On the other hand, does the function
directly affect safe shutdown or mode changes, the
weighting factor is three. Even though it may be
important and risk significant, the total score of
five times three is 15. Okay?
DR. WALLIS: Who fills this out? Does the
STP fill this out or does the expert panel fill this
out?
MR. SIEBER: The expert panel who is
employed by STP.
DR. WALLIS: They have to answer all their
own questions?
MR. SIEBER: That's right. You end up, by
the way, as part of the process a different feudalist
than the original one which is part of the submittal.
DR. WALLIS: So they have to do all the work
of finding out if this function is called out in EOPs
and all that?
MR. SIEBER: That's right. It's pretty easy
to do. The EOPs are on the computer and all the mark
numbers are in there. All they have to do is a word
search and out comes all this --
DR. WALLIS: Yeah, but if it's a kind of
secretarial job, we really don't need to have an
expert panel do it.
MR. SIEBER: No. The clerical function of
arranging all this I'm sure is done by clerks. The
panel actually has a pretty demanding qualification
requirement as I see it.
MR. LEITCH: Jack, are there two answers,
one based on frequency and the other based on
perceived risk impact or are they somehow merged
together?
MR. SIEBER: They are merged together so
that you end up with a single number. These are the
risk impact and the frequency and this is the way it's
phrased. If you look in your package, there would
have been -- you weren't there but there is a document
called "ACRS Backup."
If you look at that, and these pages aren't
numbered, but about halfway through where it says
"weighting scale," it explains how the questions are
asked, how the match is done, and how the scores are
determined. They are actually determined two
different ways. One of them is you determine the
total score based on all the questions.
As it turns out, the combination of two
fives, a four, and two threes when multiplied by five
equals 100. That's where the weighting factors
actually came from as opposed to getting 25 points for
putting your name on the paper.
So then they look at the ranges in which
these answers came out and they said if it's between
71 and 100 it's high-risk significance. If it's
between 41 and 70 it's medium-risk significance. If
it's 21 to 40, it is low-risk significance. Zero to
20, it is not risk significant at all.
This is one of two methods that they use to
categorize. The other method actually looks at the
answers to individual questions. If you get an answer
for an individual question with this weighting factor
that is greater than 20, then any one question
automatically high-risk significance.
If it's between 12 and 20 it's medium-risk
significance. If it's between six and 12 for any
single question, it's low-risk significance. If it's
below six, it's not risk significant at all.
These are additives. You can either achieve
the score this way or the answer to a single question
could put it into a category, the components up in the
highest category of whatever method is used.
Now, I have to ask myself a few questions
when I think about this whole process. The question
that come to my mind is when we just stick with CDF
and LERF, which to me is implied when you use RAW and
Fussel-Vesely, are these -- that's not true?
DR. KRESS: No. You can do a RAW or Fussel-
Vesely on anything.
MR. SIEBER: Okay.
DR. KRESS: But the RAW and Fussel-Vesely
they use were for --
MR. SIEBER: For CDF and LERF. Well, the
question is are these the right criteria and are they
the only criteria that should be used which, in fact,
CDF and LERF --
DR. WALLIS: Let's go back to what's
happening here. There's the PRA results which form
one package. Then there's the 94 or 96 percent.
MR. SIEBER: Right. Another box.
DR. WALLIS: This is the other box.
MR. SIEBER: And they overlap.
DR. WALLIS: The experts don't evaluate the
stuff that's in the PRA using their matrix?
MR. SIEBER: Yes, they did.
DR. WALLIS: Ah, so you can compare one
versus the other.
MR. SIEBER: In fact, that's one of the
checks used during the process.
DR. WALLIS: Okay.
MR. SIEBER: That's one of the checks.
DR. SHACK: That's how you decided that the
binning was reasonable.
DR. WALLIS: Is there some evaluation of the
reasonableness of the binning when you look at this
comparison?
MR. SIEBER: Yes.
DR. WALLIS: Okay.
MR. SIEBER: And to me that's one of the key
saving graces of this process, at least from the
standpoint of what the staff has to deduce out of the
process to say, "Yeah, this is reasonable." Or, "No,
it is not." That is one of them. The sensitivity
studies is another one. To me I think it's pretty
important that they did that overlap and came up with
a reasonably consistent answer because that tells you
something about the effectiveness of the panel.
DR. SHACK: I seem to recall numbers like
PRA gave me 800 and the expert panel on the same set
of components gave me 840 so they were somewhat more
conservative which you would sort of expect.
MR. SIEBER: Okay. The next question that
I asked in my own mind, which we have discussed at
length here, is are RAW and Fussel-Vesely the correct
measures of importance of the component in this
context and also the numbers.
DR. WALLIS: You shouldn't use the term
correct. You say appropriate or something.
MR. SIEBER: Appropriate.
DR. WALLIS: Correct implies some sort of
absolute standard which is the reason for these
things.
MR. SIEBER: That's right. Okay. And the
third question, I think, that I asked of myself, which
I came away with based on the outcome of the
comparisons as being okay, is do these deterministic
questions and the weighting factors make sense.
I guess you can ask any questions that are
pertinent to risk and assign any weighting factors.
The proof of the pudding is when you compare that to
the PRA studies, do you end up with consistency? The
answer is yes. These are reasonable questions to ask
except, in my opinion, there is some overlap
associated with them.
I scratch my head. For example, does a
failure of this component create an initiating event,
and they weighted it only as three. I thought, gee,
if you don't have any initiating events, your risk
goes way down. It wasn't clear to me why that was the
case.
On the other hand, there's a lot of
initiating events that don't proceed beyond the fact
that the plant shuts down safely and 9,999 out of
10,000 is probably the right number for that because
we've only had in commercial plants one accident.
DR. BONACA: You are close to the end,
right?
MR. SIEBER: Yes, I am. We'll move rapidly
to the end. I just have two more slides to do.
The process of doing this comes up with a
two-by-two matrix which looks at safety related and
risk significant components, non-safety related but
risk significant which covers the two categories that
I listed first as the purpose, and then safety related
non-risk significant and non-safety related non-risk
significant.
Of course, they end up with 8.7 percent as
compared to 5.7 which is one way or another identified
by the PRA as being important. There is some overlap
so the expert panel actually added approximately 1,200
components to the process. Non-safety related and
risk significant, 372.
Now, the question here is these were not on
the original Q-list but it turns out that they are
important from a safety and risk standpoint. So the
question is do we now have a safety question that
perhaps STP has answered by using this process.
But maybe there are other plants out there
that have similar original classifications schemes
where they haven't gone through this process and
perhaps there are components in other plants that have
more risk significance than is reflected in the
application of Appendix B. To me, that's a site
benefit to STP but a question for the staff to think
about in the process.
Safety related and non-risk significant is
12,905 which would be the items where special
treatment requirements could be relaxed to one extent
or the other. And non-risk significant and non-safety
significant is all the remainder of the components in
the 29 safety systems that were analyzed.
DR. POWERS: Jack, I never understood
exactly why in the safety related non-risk significant
category we don't just treat them the same as the non-
safety related, non-risk significant. I relax it down
to industrial use or whatever it is that you specify
for --
MR. SIEBER: I think that when you go back
to the slide, and I may not be able to remember it,
but there was four classifications from high, medium
or --
DR. POWERS: I understand that.
MR. SIEBER: So in some cases components,
the application of special treatments, was not fully
relaxed so that in the event that --
DR. POWERS: I know what they've done. What
I don't understand is why they've done it.
MR. SIEBER: Why they did it the way they
did? I think it is a conservative approach in my
opinion.
MR. NAKOSKI: This is John Nakoski if I
could address Dr. Powers' question. In option 2 one
of the restrictions that we have is that we need to
maintain a design basis of the plant, which is what
the plant was licensed to.
Completely relaxing all the controls without
having any confidence that these components would be
able to perform their functions would essentially be
a change in the design basis which would be a change
in the licensing basis which is not where we wanted to
be in option 2. That's the short answer, sir.
DR. POWERS: See, I come from the viewpoint
for these non-safety related non-risk significant
items when they acquire them, they basically acquire
things that actually work and do their job so your
confidence here, if you did the same thing for these
things in the lower left-hand corner, it's not that
you would have zero confidence. You would not have
maybe as much as you would for the upper left-hand
corner but it's not zero.
I think there's a nice term or phrase for
industry practice or something like that for the kind
of confidence you have. It just strikes me as
timidity for the reason of being timid. That's what
it strikes me as.
MR. SIEBER: I think the kinds of things
that are relaxed are some of the pedigree
requirements. Is that not the case?
DR. POWERS: Sure.
MR. SIEBER: You buy a valve and the valve
cost you $1,000, but the bill you get is $10,000 and
the paper that you get weights three times as much as
the valve.
DR. POWERS: It should because --
MR. SIEBER: You have to ask yourself how is
that used for safety?
DR. POWERS: It's nine times more expensive.
I mean, the paper is nine times as expensive as the
valve so it should weigh more.
MR. SIEBER: That's right.
DR. POWERS: It's a trouble I have with
option 2 to begin with.
MR. SIEBER: I think what you're telling us
is it's as much a legal requirement as anything else.
MR. NAKOSKI: Specifically for South Texas
and the exemption space, yes. In rule making there
may be other alternatives.
MR. SIEBER: Now, just to finish up here, on
February 8 we got a package which listed basically
open items. I think there were 18 open items -- or
22. In any event, when we actually had the
subcommittee meeting, we only talked about three open
items.
The difference is because at the
subcommittee meeting we are only talking about the
categorization process. The list that we had on
February 8 included all open items on the option 2
process. A lot of those have gone away.
Let's see. Actually, on that list there
were 16, four of which were closed, one of which was
confirmatory, one which is before the risk informed
licensing panel for some kind of a confirmation or
final resolution or approval, and seven still remain
open for the whole process. Is that correct?
MR. NAKOSKI: I can give you some more
updated information. We met with South Texas on
February 15 and 16. There were currently five open
items that were closed without exception based on the
licensee's response.
Three open items that with some editorial
changes that were agreed to during the meeting would
be closed. Six have some level of success path
identified and agreed to into varying levels of detail
and agreement.
Three require, I think, further interactions
between the licensee and the staff and those deal
primarily with the seismic and environmental
qualification issues. There is one on controlling
changes to the processes that the staff has not yet
finalized its position on.
MR. SIEBER: Okay. Thank you.
That concludes the presentation if anybody
has any comments. This is where we stand at this
point. I was sort of under the impression that what
we ought to do is wait until the process is completed
before we write a letter but we may want to reconsider
that because, I guess, in my opinion this is a pretty
complex subject and to leave everything until the end
might cause a setback from the staff's standpoint on
their timely resolution of things. That's something
we have to decide this week.
DR. BONACA: What are the thoughts of the
staff regarding the report at this time?
MR. NAKOSKI: I think your insights on
categorization at this time would be valuable for us
to move forward recognizing that you haven't gotten
any substantial feedback on where we are with
treatment. I think there would be value added now to
get this behind us.
MR. LEITCH: I have one question regarding
weighting. If I understand correctly, zero to 20 they
call non-risk significant.
MR. SIEBER: That's right.
MR. LEITCH: That would then put it in the
lower left-hand box, safety related.
MR. SIEBER: If it was on the Q-list and it
was zero to 20 and confirmed by the expert panel as
belonging there, it would be in the lower left-hand
corner.
MR. LEITCH: I can understand how it would
get to the very low risk-significant but not non-risk
significant. In other words, if you ask these
questions, say is the function used to mitigate
accidents or transients, and even if it's a three as
far as risk, I tend to get a 15.
MR. SIEBER: I think what happens is a lot
of times in the original classification of what
belongs on the Q-list and what does not, they would
take it either as functions or systems.
There are things in a system that might
require some pedigree because it originally fell under
the requirements of Appendix B who really doesn't
serve any function whatsoever as far as accident
mitigation.
It's not called out in the EOPs. It can't
cause an initiating event. It's just there. It's in
that system because of the way it was classified the
first time around. I suspect there are a fair number
of items that are like that.
MR. NAKOSKI: Mr. Leitch, if I could answer
that. An example at South Texas, for example, a gauge
in a safety related system that's just used to collect
data. It doesn't perform any function. Answer does
it initiate an event, you're going to say no so it's
zero. There's a lot of times when you answer those
five questions you can have a zero.
DR. UHRIG: Then you do have a zero one type
thing multiplying.
MR. NAKOSKI: It's a zero if it's no and it
can be one through five if it's yes based on --
DR. UHRIG: Okay.
MR. SIEBER: Mr. Chairman.
DR. BONACA: With that, any other questions?
I think we will have to make a decision later. I
think we should have also the chairman here to make a
decision on whether we should write a report. We
heard the request and that may be appropriate at this
time.
If there are no further questions, at this
point we'll take a break for 15 minutes and resume
again at 20 of 11:00.
(Whereupon, at 10:26 a.m. off the record
until 10:40 a.m.).
DR. BONACA: Let's resume the meeting now.
I wanted to start on time because we have Mr. Grimes
here who came to help us and Mr. Prato who is the
present manager for the Arkansas One License Renewal
Application.
The intent here for me was to provide you
with a summary of the meeting that took place last
week on this subject. We decided not to have a full
presentation to the committee because, you see, this
application is very similar to the Oconee applications
and we felt there were no issues that deserve at this
time to have a full presentation from the applicant
and the staff or the full committee.
The intent right now is not to write an
interim letter at this time and distribute to you a
two-page summary that I put together for my own use to
keep a memory for the final report we'll have to write
when the open issues are closed.
This summary that you have in front of you
does not contain information on the open items. I
will provide it to you as I walk through these
paragraphs.
Also, this two-page summary. On the second
page at the bottom has Jack Sieber written in. For
some reason his name got into it but he doesn't belong
there so disregard it. As you can imagine, I was
surprised when I saw that but somehow it got there.
This is to do with some of the intricacies of
computers I guess.
DR. SHACK: Of all the random things to type
Jack Sieber seems pretty far down on the list.
DR. BONACA: So let me just walk through a
little bit this summary.
On February 22 we met with the
representatives of the applicant from Entergy for
Arkansas One and presented to the staff to review the
Arkansas One license renewal application and the
interim SER.
The SER we just call interim because the
open items are now closed. Arkansas One is a B&W-type
PWR designed to generate 2568 megawatt thermal or
about 836 megawatt electric.
Now, the reactor is very similar to the
Oconee units that we recently reviewed and for which
have approved those who participated in the approval
of the SER and of the application.
Because of the similarity Arkansas One has
utilized a lot of the lessons learned from the license
renewal of Oconee. In order to benefit from these
similarities, we asked the staff to provide us with a
presentation that would highlight the differences
between the applications for Oconee and for Arkansas,
as well as the differences in the solutions which
means specifically age and management programs that
they have chosen if there are differences.
The reason is this will allow us to benefit
from previous experience. I would rely on your
judgment for future applications if that's the right
approach. I believe it is the right approach because
it allows us to keep our memory of where we're going,
even for the BWRs we are going to review with the
exceptions of the reactor vessel and other components
which relate to that.
There is so much similarity in the
applications from PWR and BWRs simply looking at
passive components so the staff provides us with an
informative presentation which was really based on the
formative comparison.
The second observation I would like to make
is that the application which appeared at the
beginning quite condensed was quite effective, I
think. I'm giving this feedback because it really was
easy to review it for a number of reasons. One is it
contained in the back a number of appendices which
condensed the information we needed.
For example, Appendix B contained a full
summary of all the problems that are being credited
for a license renewal and also segregated the first
seven problems and new problems. The rest were
existing problems. That really helps understand where
the new issues are, where the new problems are. That
was, in my judgment, a very good format.
Appendix C described the approach that was
chosen to manage aging effects. Also that was very
helpful because, again, you have a full dedicated
appendix where you can go to look for those solutions.
I just bring up these issues because I don't
know to what extent the next applications will reflect
this format but maybe there is some chance because of
the NEI.
MR. GRIMES: Dr. Bonaca, this is Chris
Grimes. I would comment that I think you'll find
Arkansas is very close to the standard form and
content that we are recommending in the standard plan
and the NEI guide that we would endorse with the
regulatory guide.
DR. BONACA: Thank you. With that, in
general the subcommittee had the following
observations or questions regarding scoping and
screening. The scoping and screening methodology
devised by the applicant identifies components appear
to be well structured and comprehensive.
This methodology we know is consistent
within the I-9510 and also with the NRC SRP. The
Arkansas One FSAR was facilitated in many ways because
the definition that Arkansas has used for safety
related is the same definition that the license
renewal rule uses for safety-related components.
Also, the Arkansas One Q-list include all
the support systems of the safety-related component or
those systems which are not safety-related but whose
failure would cause safety-related systems not to be
effective. Therefore, because of the definition that
they have used for actual list, it was easy for them
to pull those lists out and say these are the
components which are in the scope of license renewal.
Actually, in addition to that, the Arkansas
application included a number of systems and
components which were included in the Q-list purely
because they could have interference with safety-
related systems by physical interaction, for example.
That expanded somewhat the scope and the
feeling you get when you look at the application is
that the scope in general is pretty conservative. It
went beyond the requirements of the rule in my
judgment.
DR. WALLIS: Can I ask you about electrical
cables?
DR. BONACA: We'll get there.
DR. WALLIS: This may look like Oconee but
the cables can be quite different. We know that there
is degeneration of cables. I was looking at the hatch
which is a different one all together. I couldn't
quite figure out why some of the cables got screened
out and some of them were considered because we know
the cables do deteriorate. Do they pay proper
attention to the cables?
DR. BONACA: I think for the EQ medium
voltage cables they are going to be subjected to the
requirements that result from the generic issue
resolution.
MR. GRIMES: Dr. Bonaca, I would like to
clarify there are three points that are raised by Dr.
Wallis. The first is with regard to scoping and
whether or not particular cables are screened out
based on function.
As I recall Arkansas uses a spaces approach
so they would only screen out cables if there aren't
any in the space. We were confident that the scoping
will capture all of the requisite cables whether they
are subject to EQ under 5049 or not. We rely on the
process for compliance with 5049 to maintain the
qualified life for EQ cables. For non-EQ cables the
applicant has proposed to --
Is this an open item?
MR. PRATO: An open item on medium voltage
cables -- this is Bob Prato -- that are inaccessible
and that can be exposed to underground conditions.
MR. GRIMES: And we would expect the same
form of resolution that we achieved on Calvert Cliffs
and Oconee.
DR. BONACA: I never thought about the
medium voltage cable because it's an open item.
Regarding scoping and screening again, I
said before that it seemed to me it was quite
comprehensive and went somewhat beyond the narrow
interpretation of the rule.
There are two open items on that. They have
to do with the flow orifice that brings in the sodium
hydrazide and the question is why is it not included
in the scope. The other issue is why are a number of
fire protection systems and components not in the
scope.
My sense is those are good questions. I
understand the reason for the resistance on the part
of the applicant about his issues. It's simply that
they need to provide more information to bring closure
to these issues. We felt that we agreed with these
questions and they need to be addressed.
These are the only open issues on scoping.
Now, the subcommittee also raised a number
of specific questions on scope. We made an effort of
raising questions regarding systems that have the
appearance of having been in the scope and they were
not.
To all those questions we raised during the
subcommittee we got answers from the licensee on the
staff that said that either there was a good reason
why they were not scooped and we accepted the reason,
or they indeed were in scope.
Often times the reason why there was a
disconnect in the understanding is that the
application included in scope, the SER included in
scope because there had been already communication
between the staff and the licensee and that brought
the component in scope.
Now, we asked questions regarding these
issue. Would the application be modified to include
those changes which were negotiated in the
federalization of the SER and the answer was no but
the FSAR update will include those commitments.
Regarding the process used by the applicant
for defining aging effects requiring aging management,
that process appears to be comprehensive and
effective. They used a new approach where they have
a set of tools in what they call Appendix C.
In our review we found that the application
has considered the aging effects we have seen in
previous applications. They really have applied the
lessons learned from previous applications. That's
pretty obvious.
There are some open issues regarding some of
this aging management programs. They have to do, all
of them in my understanding, with additional
information to better understand why they are
effective enough to deal with the aging management
issues. Is that correct?
MR. PRATO: This is Bob Prato. What they
need to do is they need to add additional description
to the FSAR supplement. There are 11 of those items
where the description that they provided in their
application was inadequate and they needed to provide
more information in the supplement itself.
DR. BONACA: So I understand in that sense
there is no contention there except you want to have
more information and detail on what they are
committing to.
MR. PRATO: That's correct.
DR. BONACA: In the section that has to do
with limited aging analysis, there also seems to be a
pretty comprehensive inclusion of all the issues
they've seen for other plants. There are a number of
open issues on this.
One has to do with the -- well, one has to
do with additional information also regarding TLAA in
the addendum to the FSAR.
Second has to do with buried medium voltage
cables for which the staff is contending that similar
cable not in similar environment is not indicative of
the status of the one which is buried and, therefore,
cannot be used as an indication and they are
requesting a program for that and we fully agree with
that kind of perspective and we are going to see that
there is closure on this issue.
There is another TLAA which is still open
regarding the specific criteria to be used for forces
of the Arkansas One containment. That's also, it
seems to me, reasonably similar to the Oconee
questions that we had. I would expect to have a
similar closure on that.
There is finally an open issue of the aging
of boroflex. During the presentation we are told that
the applicant has agreed that is a problem even for
the current life and, therefore, they will provide a
solution that doesn't address specifically the
extended life but specifically boroflex now.
I guess the question I have for the staff is
are you looking for a commitment at this stage?
MR. GRIMES: Dr. Bonaca, this is Chris
Grimes. The controversy evolved primarily because the
applicant chose to challenge the definition of a time-
limited aging analysis to put this into the context of
a corrective action.
We don't want to argue about whether it's a
time-limited aging analysis. We believe it is. What
we're trying to focus on now is establishing
confidence that there are program attributes
associated with the corrective action that can be
relied on in a programmatic way.
We don't necessarily need to know exactly
how the life-limiting aspects are going to be fixed,
but we want to know that there are the 10 program
elements in place that will ensure that before there's
a loss of function corrective action is taken. I
think we'll be able to work that out with the
applicant.
DR. BONACA: The only surprise I had
somewhat there is that for initial the nature I know
that other licensees have already developed plans to
deal with what is the criteria which you've cited at
some point you cannot operate any further so some of
them assume that they have certain split open spaces
or assume that you have large gross formation of
boroflex. I was somewhat surprised that Arkansas did
not have a problem with the nature.
MR. GRIMES: Actually, I think Arkansas was
also surprised. I think they had envisioned that this
was an issue that they could deal with in the future
and they had originally, and Bob can correct me if I'm
wrong, it originally said that, "This is a time-
limited aging analysis and we can manage it."
Then they were surprised to discover that
the inspection results did not support the current
licensing basis. They were trying to get positioned
so that when they deal with the future of their spent
fuel capacity and the maintenance of that facility,
that they would decide how to deal with it. I imagine
that it was for timing more than anything else for
Entergy.
DR. BONACA: Okay. I already spoke about
the effectiveness of Appendix B. Having this summary
listing of problems has allowed us to really get a
sense of the extensiveness of the problems.
Appendix B is formatted in a way where you
have this seven new problems. Then you have all the
other problems which already exist but some of them
are modified to deal with some of the issues which
will be raised in the context of license renewal.
In my review of the SER, it was apparent
that the staff had performed an effective review of
the Arkansas One application. We asked questions
regarding the process that was used and there were two
processes.
Certainly the first one is a lesson learned
also for the subcommittee licensing ACRS. They
specifically take systems of components which are not
in scope and test why they have been left out. Now,
that's a guidance which is also given in this SRP, but
I think this time I use it personally and I found I
had a lot of questions.
Each one of us has a sense of what really
should be a safety system. For example, the Fussel-
Vesely measurement device does not have any other
function than the safety function. The question of
why it's not in scope is a good question. The answer
is provided and so it was acceptable but I think that
was an effective review that the staff performed.
Another question we asked was regarding the
staff visits to the site. They were performed by the
staff and clearly they had done a reasonably extensive
process of V&V, validation and verification. I
believe there were two trips to the site involving
several days and several people.
MR. PRATO: This is Bob Prato. There was
actually three. There was an audit for the
methodology review. Then there was a scoping
inspection to verify the implementation of the
methodology. Then there was a two-week aging
management review.
The last two, the scoping inspection and the
aging management review, involved seven spectors and
the scoping methodology, the week that we spent on
site, involved three engineers from the site -- from
headquarters. I'm sorry.
DR. BONACA: The second visit was for what
you said?
MR. PRATO: For verifying that they
implement the scoping methodology correctly. It was
a scoping inspection. During that inspection we
actually looked at systems that were excluded and
verified that they had good justification for
excluding them. Then we looked at the structures and
the components individually and made sure they
included them correctly as well.
MR. GRIMES: This is Chris Grimes. Just to
make sure that we keep the process clear, we refer to
an audit of the methodology. That is the team from
headquarters that goes down and gathers information
first hand that they use to prepare their safety
evaluation of the methodology.
The other two pieces are the scoping
inspection and the aging management program
inspection. Both of those are conducted under
inspection manual chapter 2516.
Those inspection reports then support the
recommendation by the regional administrator. That is
when we go back to the path that shows all the
contributors to the evaluation findings. We try to
keep the product lines distinguished.
DR. BONACA: Okay. The subcommittee noted
that a number of new problems and one-time
inspections, seven in total, have decreased
significantly from the first application. The first
application had 30 odd one-time inspections and this
one has two.
We asked questions of the staff and the
reason clearly is that a lot of the open issues have
been addressed now. I would like Mr. Grimes to
describe the reason.
MR. GRIMES: Dr. Bonaca, as we've reflected
on our ability to explain to the ACRS the consistency
and the treatment of our review process, we discovered
as time as gone on we've learned some lessons. We are
going to explain those to the committee when we
present the generic aging lessons learned report.
One part of this is evidence that lessons
have been learned and applied. A second part is that
the numbering system, the accounting system has
changed from plant to plant. If you recall, for
Calvert Cliffs we counted some 436 programs. We were
actually counting individual procedures.
On Oconee we had roughly the same number
that Arkansas reports and that is about 30 programs.
Also, there is a reflection here that Arkansas was
much more aggressive than Calvert Cliffs or Oconee or
even the industry in general in their GALL approach.
Arkansas has leaned forward and they are taking on a
number of these routine inspection activities.
I think the best example is by going to a
risk-informed service inspection program they captured
small bore piping which is a one-time inspection in
GALL and it is still a controversy with the industry
in terms of whether or not license renewal should be
solving that problem or some industry initiative
should be credited for that problem.
I do think that is appropriate that we
should give recognition to Entergy's aggressiveness in
treating these areas. That accounts for part of the
reduction, too.
I have also committed that when we come to
explain GALL to the committee, we will provide a
cross-reference of what was done for Calvert Cliffs
and Oconee and GALL in order to show the evolution of
the learning.
DR. BONACA: I believe that's really very
useful to the committee if we can get this perspective
of how those one-time inspections have evolved. I
looked at some of the one-inspections which were in
Fussel-Vesely terminals and they are, in fact,
included in the program here and there was no specific
identification separately of the one-time inspection.
It was simply folded in the problems as some
of the other applicants have done, too. I did it not
under duress but still under some kind of negotiation
and had to choose how to do it.
Clearly, Arkansas came in and I believe that
as the industry accepts this kind of resolutions,
these will facilitate the next applications and
reviews.
During the subcommittee meeting we noted a
number of apparent inconsistencies between the
application and the SER information already assembled
when we were talking about scoping but we found it
mostly in the problems.
Typically, again, discrepancies were tied to
the fact that the applicant proposed some program, for
example, visual inspection of some piping. The staff
said, "Visual is not enough. You should have
ultrasonic examinations."
The applicant agreed so the SER documents
ultrasonic testing as the program used to deal with a
particular issue while the application still quotes
visual so there wasn't really a discrepancy there.
The discrepancy had been either solved.
Typically the discrepancy resulted in an
augmented program in the SCR than was presented in the
original application. I understand that the addendum
to the FSAR will contain all the commitments anyway so
there will be commitments as reflected in the SER.
In conclusion, the feeling we got as a
subcommittee was that the staff has performed an
effective review of the Arkansas One application. The
Arkansas One application is an aggressive application
that went, from what we can see, beyond the minimum
requirements of the license renewal rule.
Therefore, we felt confident that we
understood enough to stay with the process right now.
We recommend the committee that we do not at this time
write a letter.
We also would not conduct a subcommittee
meeting to review the closure of the open items
because there are very few. Are intent is they want
to bring in now the staff back the applicant for a
full committee meeting when the open items are closed.
Hear a presentation by the applicant at that time and
hear a presentation by the staff and we will write a
report at that time.
Any questions from members or staff?
DR. WALLIS: The CD that we have with the
application, nothing has changed from that so if I go
back to that to look at things, that is the key
document to review before our next meeting?
MR. GRIMES: That's correct. The CD
contains the application as submitted and it also
provides the FSAR. Dr. Bonaca has made the point that
when you review the safety evaluation, the safety
evaluation will articulate the paper trail from the
application to a resolution. All the correspondence
that has occurred in the intervening time should be
clear in the safety evaluation.
MR. PRATO: This is Bob Prato. All that
correspondence is identified by dates and each of
those letters that were provided by the staff and the
applicant are on the docket. There is a complete
paper trail on the docket. If you need anything
specific, though, feel free to call us and we'll make
sure you get a copy of whatever you need.
DR. BONACA: Tomorrow, I believe, we will
also talk about the Hatch application. The SER is
coming to come for our review. Well, you probably
already received it at home.
With that also we have two subcommittee
meetings, one that will lead us to review the guidance
documents, the final changes to those. A second
meeting on the BWR VIP which support in some form the
Hatch application.
We have talked about having a presentation
of the Hatch application and the SER in the same
format by emphasizing the similarities with even PWRs
given the fact that there are so many classes of
components or commodities that are similar
irrespective of the type of reactor that is being
used.
MR. DUDLEY: This is Noel Dudley. I do have
those documents in house. I can send them out either
express mail if you want to start working on them
Monday, or I can send them regular mail and you'll get
them Wednesday or Thursday next week.
DR. BONACA: These are the SERs?
MR. DUDLEY: These are the SERs. These are
four BWR VIP reports, associated SERs, and all the
proposed final draft of guidance documents.
DR. BONACA: Guidance documents. Okay.
MR. DUDLEY: Well, say the March 1 draft of
the guidance document.
DR. BONACA: So all the members of your
choice. You can take it with you.
MR. LEITCH: It's 1,700, 1,800 pages.
DR. BONACA: What we thought of doing was to
send to you only those sections that you are asked to
review.
MR. DUDLEY: I think the document is small
enough. The actual GALL report now is only a couple
inches.
DR. BONACA: It's small enough you can
memorize it.
CHAIRMAN APOSTOLAKIS: Is anyone dying to
have it in his hands by Tuesday or Wednesday?
DR. WALLIS: I would love to have a CD
rather than a big pile of paper. That means someone
has to scan it in presumably which is a pain.
MR. DUDLEY: Disks for the guidance
documents are not available yet.
MR. GRIMES: This is Chris Grimes. We had
envisioned putting them together on a compact disk
after they are approved. I'll explore the possibility
of having the files loaded onto a CD-ROM.
You wouldn't have the benefit of the
electronic book features with tables of contents and
so forth but if you're more comfortable in working in
electronic forms, we can have Word Perfect files
assembled on a CD for portability.
DR. SHACK: How about PDF?
MR. GRIMES: I hesitate to say that because
we would have to pull the PDF files out of ADAMS and
I would rather not.
DR. SHACK: When I get Word Perfect
documents from Paul my computer thinks they are PDF
files anyway.
DR. BONACA: So we'll do that. I just want
to ask if there are anymore questions from members.
No further questions. With that, Mr. Chairman, I give
you this 32 minutes of time.
CHAIRMAN APOSTOLAKIS: The Chair expresses
deep gratitude.
We were hoping to go over the ATWS letter
because we have a new version of it. No, we don't
need the description now.
(Whereupon, at 11:14 a.m. off the record for
lunch to reconvene at 12:46 p.m.).
A-F-T-E-R-N-O-O-N S-E-S-S-I-O-N
12:46 p.m.
CHAIRMAN APOSTOLAKIS: We are back in
session. I neglected to mention this morning that we
have Dr. Peter Ford sitting with us at the table as an
invited expert.
The next session is on Spent Fuel Pool
Accident Risk at Decommissioning Nuclear Power Plants.
Dr. Kress, you are the leader on this.
DR. KRESS: Yes. It's a simple,
noncontroversial subject.
CHAIRMAN APOSTOLAKIS: We should be done in
five minutes then.
DR. KRESS: As you all recall, there was a
technical study on this issue intended to give
guidance on how to develop a rule or exemptions to
relax requirements at spent fuel pool, requirements on
emergency preparedness, and perhaps insurance
requirements and security requirements.
We reviewed that technical study and give it
fairly good grades, I think. They determined that the
risk after a certain amount of time was low enough
that you could do without the emergency preparedness
and still meet the safety goals and this risk was done
a fairly conservative basis.
They also noted that previous exemptions for
these sort of things were based on the concept that
you couldn't have a zirconium fire after a certain
time because the heat generation rate was too low. It
was balanced by the cooling rate so that you weren't
hot enough to start a fire.
The staff to some extent backed away from
that concept and said they couldn't exclude the
zirconium fire, whatever that means. I think there
was some differences of opinion between the industry
and the staff on these things. There have been some
further correspondence and some discussions.
I think today I view this as more or less a
status report or an update on where we are with
respect to publications on possible rule making
options. I don't think we are ready for a letter
again this time. I don't know. It depends on what we
hear.
With that as a sort of vague introduction,
I'll turn it over to Tim Collins of NRR to get us
started.
MR. COLLINS: As a status report this will
be a real quick meeting. The status report is that
the study is done and we're starting to do the
thinking on the policy options.
There are a couple of things, though, I
would like to provide a little clarification on. When
we started out the study, the charter was not just
aimed at EP and insurance. It was to provide a broad
basis for rule making relative to decommissioning
plants as a whole.
There was some emphasis in the report on EP
because of the number of exceptions that had been
granted in the past and the most recent actions in
decommissioning were requests for exemptions in
decommissioning or insurance.
Another thing I would like to make a
clarification on. The finding in the report with
regard to not being able to preclude a zirconium fire.
The finding in the report was really that we couldn't
define a generic time without numerous constraints,
okay?
The original exemptions were granted on the
basis of an unobstructed airflow calculation, those
previous exemptions which said after a certain number
of years you wouldn't get to a temperature that would
lead to a fire.
When we originally tried to do the study, we
started out with the unobstructed airflow cases. In
the course of public comment there was questions
raised with regard to partial uncovering of the fuel
which would obstruct airflow.
We also in trying to do our own calculations
we were trying to decide how much airflow we should be
using. We ran into problems with different rack
configurations, different spent fuel pools, the
relationships between the building airflow and the
flow in the racks.
Then when we looked at our results, we said
the most likely events that could get you into trouble
were major seismic events and major cast drops, all
catastrophic events. We finally threw up our hands
and said we can't define the geometry which will allow
us to do as calculation which will give us a generic
decay heat time.
I think the characterization that the study
concluded that you could never preclude it is a little
bit of an alarming characterization. It's more a
matter of, well, there are so many uncertainties and
so many unknowns in trying to do a generic analysis
that we couldn't come along and say after five years
we're sure there's not going to be a fire.
As far as where we go now, I mean, as far as
any technical work goes, it's at a standstill. We
believe that the results of the report were -- the
staff is comfortable that we are below the safety
goals. We think that additional technical work could
be quite expensive if it's going to really
significantly reduce uncertainties further.
We're not sure what goal we would be
shooting for if we started to do that work so we're in
the position of developing policy actions for the
commission. The policy actions are aimed at questions
like how important is factors like public confidence
in your decision making process.
I mean, we believe that the risks are very
low and below the safety goals for reactors, but there
is a very significant question in areas like emergency
preparedness, how important of a factor is public
confidence and how does that weigh into decision
making. These are the types of things we are going to
address in our May paper to the commission. There is
also questions of how do we use risk in security
related --
CHAIRMAN APOSTOLAKIS: I'm sorry. This
issue of public confidence, maybe we can clarify it a
little bit. You see, you believe that the risk is
low. When you say that, you mean the whole
distribution is below the goals or the mean value is
below the goal but there is a tail that goes perhaps
above?
I'm trying to understand the meaning of the
statement and why some other groups might differ, I
mean, with the issue of confidence. When the staff
says the risks are low --
MR. COLLINS: When we did our analysis, we
tried to use a range of sensitivities. For example,
in the seismic analysis we used the Livermore curves.
We used the EPRI curves. We did source term
sensitivities to include large amounts of ruthenium
release, what we thought was a reasonably large
fraction of fuel finds. Using those bounds, we
believe that the risk is still below the safety goal.
CHAIRMAN APOSTOLAKIS: You did not quantify
the uncertainty?
MR. COLLINS: We did not quantify the
uncertainty.
CHAIRMAN APOSTOLAKIS: Just sensitivity
studies?
MR. COLLINS: That's correct.
CHAIRMAN APOSTOLAKIS: And what you thought
were bounding analysis?
MR. COLLINS: Yes.
CHAIRMAN APOSTOLAKIS: Okay.
DR. POWERS: It must surely be if the risks
are low and you use high source terms, that is, just
the frequencies were low. That's the only reasons.
MR. COLLINS: Sure. I mean, the frequencies
were very low.
CHAIRMAN APOSTOLAKIS: The frequencies of
what?
MR. COLLINS: The frequency of -- what we
calculated in the report was the frequency of uncovery
of the fuel. We didn't try to do a calculation of the
conditional fire probability. We went from uncovery
of the fuel to consequence analysis.
DR. KRESS: And it was driven by seismic.
CHAIRMAN APOSTOLAKIS: So why would someone
else then have less confidence in what you did?
MR. COLLINS: Let's find that someone else
and ask them.
CHAIRMAN APOSTOLAKIS: I mean, how can you
take into the decision making process the fact that
others have less confidence in your results unless you
understand why? You raise the issue of how do we make
a decision if the public doesn't agree with us. Who
is the public anyway?
DR. POWERS: I think we have to go back and
look at what some of the other speakers had to say.
We had several speakers at our meetings and some of
them -- the universal sentiment was that the issue
deserved more attention.
When you looked at the more attention that
they were asking for, in general each person said that
there are design specifics at each site that make a
generic conclusion difficult to draw. You have to go
look at those design specifics and they make a
different.
Now, the difference they were asking for, of
course, depended on point of view a little bit. So
one of the questions that comes out of that is does
the generic analysis give you the answer or is it
always the site specific analysis that you have to do?
The other distinct point of view was that
looking at this strictly from an accident probability
is the wrong way to do it. In fact, there is a
security element of this as well so you have to take
into account both misadventure and deliberate actions
here in making decisions about these pools. Those
were the alternatives.
The one I think the staff is in a position
to address is the one can you get any mileage out of
a generic analysis or are all things so site specific
in the phenology affected by that site specificity
that you just don't derive any answer.
MR. COLLINS: I mean, certainly in the
development of a rule if we were depending upon
generic analysis, one of the elements of the rule
would have to be demonstrating consistency with the
generic analysis, or that you were bounded by the
generic analysis, or you would do your own plant
specific analysis. That's typically what we do with
rules anyway.
We review topical reports all the time where
generic analysis are submitted and then we get a plant
specific submittal which references the generic report
and just demonstrates why it's founded by the generic
analysis or where it's not and why it's still okay.
The rule would have to be structured in such a fashion
if it's going to depend upon --
DR. POWERS: You can do that in a rule but
I would think you would come into the rule saying, "My
generic analysis is going to be pretty good or 90
percent of 90 percent of the site specific things."
Do you have any feeling for that?
MR. COLLINS: This analysis?
DR. POWERS: Yes.
MR. COLLINS: I think this analysis would be
applicable to most facilities, yeah.
DR. POWERS: Okay. That gives the answer
the chairman was looking for. It's contrary to what
speakers on both sides of the issue have said.
MR. COLLINS: Well, I understand that. In
some cases on one side of the issue speakers would say
that the risk is so much lower. That's fine. That's
okay. If they want to do analysis which shows it's
lower, we'll find that just as acceptable.
CHAIRMAN APOSTOLAKIS: Are you done?
MR. COLLINS: I think I'm pretty much done.
I mean, there's not much more to say. We just started
developing the policy paper.
DR. KRESS: That's going to the
commissioners in May?
MR. COLLINS: Yes.
DR. KRESS: Near the end?
MR. COLLINS: I expect it will be about May
31st.
DR. KRESS: We could probably here a draft
version of that in our May meeting, you think?
MR. COLLINS: Well, a different group is
responsible for the development of that paper. I
don't want to commit them.
DR. WALLIS: So have you concluded the
better understanding of the physics, chemistry, and so
on of fires is not to be sought because the risk is so
low?
MR. COLLINS: Staff is not recommending that
we do additional analysis at this point. If the
commission decides that we need it, we'll do it.
DR. KRESS: I think there are broader
applications or needs for such stuff. For example,
with respect to pressurized thermal shock, which may
be an iron ingression type accident also. There may
be other reasons other than for decommissioning for
such research but you're not excluding that in
particular?
MR. COLLINS: No.
DR. KRESS: Just for the decommissioning?
MR. COLLINS: I'm talking about for
decommissioning rule.
DR. KRESS: I think you probably have enough
for a decommissioning rule maybe.
DR. POWERS: I guess I have a couple of
questions on that. One of them I would like to come
back to is the statement that things are dominated by
seismic.
I have been given a sheet of paper which,
unfortunately, I don't have right here with me, in
which initiating events for fires in the pool were
listed down and the percentage contribution was
provided. That list of seismic only is 13 percent.
It wasn't even top on the list and there were several
comparable to it.
Naturally enough, I can't remember what the
others were but they certainly involve station
blackout, loss of cooling capabilities.
Is it true that this thing is totally dominated by
seismic?
MR. COLLINS: We believe the seismic clearly
dominates it. We took into account events where you
had loss of cooling to the pool. You can look at it
as two basic types of events. You have a catastrophic
draining of the pool or you have a slow boil off or a
very slow leak, those two types.
Now, the second type of event, the slow one,
is dominated by human error and there's hundreds of
hours for recovery actions in the secondary. We
looked at that very carefully. We had several back
and forths with the industry and I think with the
committee even on the human error assessment. When we
were finished, we found that the seismic events were
dominating.
DR. WALLIS: Well, human intentional error
like deliberately turning on pumps which would drain
pools or something like that?
MR. COLLINS: Errors of commission as
opposed to errors of omission.
DR. WALLIS: It might well be that your risk
levels are so low that the unexpected wayward
performance of one individual might have --
MR. COLLINS: Still the recovery time was
important more than the initiating event frequency if
it was started by someone turning on a pump.
DR. WALLIS: So they cannot drain the pool
rapidly? It takes many days or something? I don't
know.
MR. COLLINS: It depends on how big the pump
is, I supposed.
DR. WALLIS: That's right. That's an
obvious statement. How long does it take?
MR. COLLINS: I'm not sure if we looked at
someone deliberately pumping the pool out.
DR. WALLIS: But you may have risk levels so
low that that sort of event is the thing you have to
worry about.
DR. POWERS: You are ham strung. You have
no way of estimating the probability of that
initiator.
DR. KRESS: That's right.
DR. POWERS: That's the fundamental problem.
The ground rules on any kind of risk is that risk is
going to be taken out. That, of course, means that
somebody has to say those kinds of risks are handled
some other way or are small enough that I don't need
to worry about them.
We have the same problem with sabotage,
somebody from the outside attacking the pool. We try
to handle that by putting fences up and a few guards
and things like that.
Similarly the kinds of people that you hire
have some sort of screened background and don't have
a predilection for sticking pumps into spent fuel
pools or something like that. Those kinds of measures
are taken. You can't put it into a probalistic frame
work.
DR. KRESS: You cannot put it into a
probalistic frame work?
DR. POWERS: People certainly haven't found
any way. What they have found is I can produce an
estimate of the probably of an error commission. What
I can't do is produce an estimate that I can persuade
Tom is correct.
DR. KRESS: That's correct.
DR. POWERS: I can do it.
DR. KRESS: It can be done.
DR. POWERS: But I can never persuade you.
Now, one could imagine that you could sit down and
have a panel of experts persuade each other what it
is.
The problem is no one ever felt like they
could take that product and put it forward and
convince anybody that these people were so profound in
their expertise on people sticking pumps in spent fuel
pools that their estimate was better than anybody
else's.
MR. LEITCH: In terms of error by omission,
it seems to me, and my memory is a little fuzzy, but
in Dresden about four or five years ago there was a
freeze up and I think a line had ruptured in an
attempt to drain the spent fuel pool. Did you think
about things like that?
MR. COLLINS: Yes, we tried to look at all
the operating events that we were aware of that could
have led to a pool draining. It's considered in the
likelihood of the initiating event.
DR. KRESS: We went over all these questions
when we reviewed the technical study. We convinced
ourself that they did a pretty good job.
CHAIRMAN APOSTOLAKIS: Do you want to say
something about the options? Please identify
yourself?
MR. HUFFMAN: My name is Bill Huffman. I'm
with NRR and I'm the project manager for the policy
paper. I think your question was would it be ready
for a briefing the first week of May. I would hope it
would be in a draft stage at that time.
However, I would say that it would certainly
have to be a closed meeting. It's predecisional.
It's not something that we want to have the public
privy to before we went to the commission on. Plus,
there would probably be safeguard information.
CHAIRMAN APOSTOLAKIS: I don't understand.
Don't we always review things that are predecisional?
DR. KRESS: Yeah, but he also brought up the
safeguards.
CHAIRMAN APOSTOLAKIS: Okay. Okay.
MR. HUFFMAN: My schedule right now did not
factor in briefing ACRS and I'm not sure exactly what
a lead time you would want on the draft.
DR. KRESS: About a week.
CHAIRMAN APOSTOLAKIS: Thirty days.
DR. KRESS: In this case we'll make an
exception. Two weeks.
CHAIRMAN APOSTOLAKIS: How big is it going
to be?
MR. HUFFMAN: Fifteen pages.
CHAIRMAN APOSTOLAKIS: Two weeks then is
reasonable.
MR. LEITCH: Are there not certain
decommissioned plants now that have spent fuel pools
where they have backed off on emergency preparedness
and security?
MR. COLLINS: Yes. We've granted exemptions
to several plants for emergency preparedness,
insurance, and security.
DR. KRESS: And those were generally based
on the problem that they couldn't have a fire after a
certain amount of time?
MR. COLLINS: It seems each exemption was
granted for a different reason. Generally, though,
often a part of the basis was the fact that you
couldn't have a fire anymore based on an assumption of
unobstructed airflow calculation.
DR. KRESS: Then you don't feel like you
need to revisit those because the risk is low.
MR. COLLINS: No, we intend to go and
revisit them. We believe that the risk is low enough
there's not a safety concern with those. We were
maybe in a situation where the basis is not correct
for the exemption. We plan to revisit those. In all
cases, I think the most recent -- the freshest fuel is
almost four years old in the facility with the hottest
fuel.
DR. KRESS: Long time.
MR. COLLINS: A long time.
DR. KRESS: I think we are also scheduled to
hear from the industry.
Lynette, are you going to take the lead on
this?
This is Lynette Hendricks with NEI and our
old friend Bob Henry with Vaski and Associates.
MS. HENDRICKS: We appreciate the
opportunity to revisit this issue with you. We see
it, I guess, in maybe a little more of an evolutionary
stage than maybe the staff views it.
We would like to basically talk about two
issues today. One are some of the little touch on the
phenology questions that were raised last time. Some
information on the basis for the cask drop. Then
finally with great boldness I would like to have a
short discussion and get some input on the seismic
question.
With that, I'll turn it over to Bob.
MR. HENRY: As Lynette said, we would like
to offer some suggestions because we think there are
some issues that can be dealt with a little more
crisply in the report and take advantage of a lot of
the experimental data that has been acquire by both
the NRC and the industry over a number of years.
I would like to be constructive in that
regard and offer some suggestions of things that could
be incorporated in the report. As Lynette said, at
the end she has some comments on seismic. The issues
I would like to particularly address to start with
would be the experimental basis that we could
subscribe to catastrophic events to get a somewhat
better perspective of the potential damage that could
really cause.
The last time we had the opportunity to
visit with you we talked a little bit about fission
product release, particularly ruthenium under those
conditions where the pool has been assumed to be
drained rapidly.
The at the end also talk perhaps a little
about suggested peer review to make sure that all of
the data that people have at their disposal gets input
into these kinds of documents that do get used for
policy making.
I should also say we're talking about the
cask drop here just as Tim was just saying. We are
focusing on the likelihood that could be a mechanism
whereby the pool would be rapidly drained. There is
some data that I'll share with you here that I think
the study could benefit from by incorporating and
suggest that this is a pretty difficult thing to do.
To start with, we feel that the status is
that this provides a good start for quantifying the
risk for significant fission product releases. We
think it's certainly a good basis. Tim was just
talking about all the field information they went
through to provide quantification of the likelihood of
losing pool cooling.
We also believe it should incorporate these
experimental results, I was mentioning, that one could
use to evaluate the likelihood that a cask drop could
indeed cause rapid draining of the pool.
Also, we believe that there is a technical
basis to be incorporated into the report to at least
give a best estimate in addition to the bounds that
are already in for fission product releases and,
therefore, health consequences.
I think if we do a little bit more than just
provide the bounds, we provide some additional
insights on how people might be using this to make
judgments.
To start with, let's start with the
experimental basis for assessing cask drop. I've
listed four references I was able to dig up. The
first two being with full-size casks dropped onto
concrete pads where they were principally there to
measure the damage to the cask, but they also recorded
the damage to the concrete. We can certainly use that
to assess our ability to determine how tough the
concrete really is.
The third one is an NRC study using steel
billets dropped onto concrete surface that is very
useful. However, the first two I'll use this
afternoon because the information for the compression
of the concrete in the locality impact is reported.
The last one are some experiments that were
done quite a while ago for high velocity impacts that
really relate to tornado missiles but from a practical
point of view they are just as usable, as we'll see,
as the first two in terms of assessing what the
implications would be for impacts on concrete.
DR. WALLIS: When you talk about a pool,
what part of the pool is being hit by this cask?
MR. HENRY: Conceptually, just think that
the cask has been lifted up and is somewhere around
the top of the pool, the rigging breaks and it comes
down through the pool.
DR. WALLIS: Through the pool of water?
MR. HENRY: Goes through the water.
DR. WALLIS: Doesn't that slow it down quite
a bit?
MR. HENRY: It does a little bit. We'll
talk briefly about that. To give you a feel for it,
the terminal velocity of water is maybe in the range
of 20 meters a second if you just use a drag
coeffision of one. You'll see this has fins on the
side and maybe that slows it down a little bit more.
The 20 meters a second --
DR. WALLIS: In your picture, it goes
through the pool and hits the bottom of the pool.
MR. HENRY: Correct.
DR. WALLIS: It's not knocking off a piece
of the sidewall or anything?
MR. HENRY: In what I present today, no, but
this fourth set of experiments here do have
experiments where the projectile was at a 45 degree
angle also. I didn't include those here because this
is a fairly quick thing but you could certainly use
those. What they did observe in those is that it
principally just grazed along the side and didn't do
anything to cause a large rupture of the wall.
I apologize for the simplicity of this.
There are correlations for this. I chose not to get
into correlations but to just use it from a very
fundamental point of view in terms of the mass and how
far it's going to fall, plus the strength of the
concrete and the dent it will make in the concrete
delta and the kinetic energy.
All this does is equate the change in
kinetic energy to the work done and the work is just
the force of compressing the concrete times delta.
This delta is the dent that it would make in the
concrete or how far it has to go into it before we
finally can get something that actually opens up a
hole that could drain the pool very quickly.
DR. WALLIS: The biggest uncertainty is A.
I mean, how does it fall. Does it fall in a corner.
We have to put in an A here. That's the biggest
uncertainty.
MR. HENRY: Again, I didn't focus on that
today. I wanted to make sure the database was
available to everybody and understood and discussed.
Again, going back to the tornado missiles, they used
As which were very small like rebar.
The part that I'm going to give you here is
a very simple approach I'm going to use. As you start
making A smaller and smaller, this thing starts giving
you far too deep of a penetration.
If you think of a corner going in first,
then it's very quickly going to spread to something
which is, let me say, is just half of the A of the
total cask. Again, we're not close to any kind of
cliff where you would almost break the pool as we'll
see here.
DR. WALLIS: You are assuming that energy is
absorbed by the concrete where it's hit.
MR. HENRY: Right.
DR. WALLIS: I think sometimes when you hit
concrete on one side the concrete comes off on the
other side but the shockwave goes through the wall,
hits tension, comes off the far side of the wall. It
hits the wall here and the plug of concrete goes out
into the next room.
MR. HENRY: Right.
DR. WALLIS: But that wouldn't be reflected
by this kind of mechanism. Would it?
MR. HENRY: It is in the way I'm going to
use it as you'll see because we're going to go back to
data where actually what did it take for something to
fly off the opposite side.
DR. KRESS: The compressive strength, that's
force per unit area it takes to compress concrete a
certain distance?
MR. HENRY: Yeah, if you want to use a
simple kind of thing, it's like the yield point. Once
it starts giving, it essentially has almost the same.
Also there is a rate of strain that gets involved in
all these, of course, but that's also not in here.
When I talk about the calculated delta, it's
going to be this very simple thing of just equating
the kinetic energy to the work done.
DR. WALLIS: That's an interesting one, too,
because if you want to bust concrete, it depends on
the size of your sledge hammer. Having the same
amount of kinetic energy with a sledge hammer that
weighs a ton isn't the same as having one that weighs
nine pounds. It makes quite a difference.
MR. HENRY: That's correct. That's
obviously all in the equation. If you do go back to
this part of it and if you want to scale it then
you've got this thing and here is the mass and the
area that you've been focusing on.
If you want a scale from one to the other,
my mass and area might be different and can I make up
for it with a different velocity to get the right kind
of energy. We'll get back to that in a minute.
These are the full scale tests and the
results of these tests for the cask drop experiments
that were done by BNFL at Sandia and also at AEA
Winfrith. As you can see, this is a pretty big
hammer, 64.5 tons.
The concrete's compressive strength was 22
MPAs with something in the range of like 3,500 or
3,600 psi. The drop itself -- excuse me for a second
because I do have color photographs that aren't in
what you have. I have six of each here if you want to
pass these around and share a little bit with each
other. There's three different photographs.
DR. WALLIS: This is a big flat hammer.
Isn't it?
MR. HENRY: Yeah.
DR. WALLIS: You're really spreading the
load.
MR. HENRY: That's why it's also --
DR. WALLIS: It's a pretty expensive hammer.
MR. HENRY: This is the apparatus. It
weighs 64.5 metric tons. In this case it's held 60
inches above the surface of this reinforced concrete
block and about to be dropped.
When it is dropped, this is the dent, the
impression that's made on the concrete from the
highest drop of 60 inches. Here you can see the ring
that's left in the concrete.
Then also this is the measurement of the
deepest part of the ring which shows that the
impression is eight millimeters. As we talk through
this, it's that impression I'm talking about.
Obviously for the first set of tests --
DR. WALLIS: Great care has been taken to
spread the load as much as possible here.
MR. HENRY: I think that's why you have to
look at all the databases and not just this one.
On the first case it was dropped 18 inches.
If you go through the simple analysis you calculate a
value of eight millimeters and it measured at the
deepest point four millimeters of imprint and
obviously some cracks in the concrete. But nothing
was -- I mean, this is sitting on soil so nothing was
broken off the other side.
In the third test it was dropped 40 inches.
Here you can see that the simple way you look at
things begins to fall apart a little bit but it still
gives you a perspective. You calculate that you would
make a dent about 17 millimeters deep or 1.7
centimeters and they measured six millimeters.
The last one, which is the picture I showed
you, they dropped it 60 inches, they calculated 26
millimeters and they measured, as we saw in the
measurements in the figure, eight millimeters.
DR. WALLIS: Doesn't the whole business of
impact impedance come into this, that if the concrete
mass is -- it's not infrared mass so you actually set
it in motion when you hit it. It's a fairly
complicated problem. It's not just a question of
absorbing energy and a distance.
MR. HENRY: Obviously the basic thing you
have to do is absorb the energy somehow. How it all
gets absorbed is more complicated than what this
simple explanation shows.
DR. WALLIS: When you hit a base ball, it's
different from hitting a wall. The baseball moves.
MR. HENRY: Right. From a practical point
of view, the same thing is true in a plant. I mean,
if it does hit the wall, the wall will bend and push
it back up again.
This, at least, gives us a perspective. We
can go to these other tests where it is much more
focused in terms of the load. The first thing here is
you see it takes a big wallop to put a hole in the
concrete.
The next one, I have taken three of those
tests and shown them here. I've taken the tests which
are the same missile all the time which is a 12-inch
pipe and propelled at different velocities.
They also have a three-inch pipe and a one-
inch pipe but there's only a couple of tests of each
where this gave us a number of tests to look at
different velocities, three of which I've just showed
you here.
If we go through the same kind of analysis,
now the missile is at much higher velocity, as you can
see, upwards of over 200 feet per second when it hits
the concrete. I put three of them in here because the
first one is with 12 inches of concrete and this is a
velocity which is big enough to drive a hole right
through the whole thing.
If you look at the next sheet, that's a
picture of the front and backside. The frontside
still has the pipe sticking in it and the backside you
can see the concrete that's blown off the back.
You want to make sure you understand what
does it take to do this because these experiments tell
you that it takes a certain amount of kinetic energy
for a particular thickness of reinforced concrete.
The first one I have on test No. 10 is
sufficient to penetrate the entire wall. Test No. 12
then is 18 inches. This is 203 feet per second, same
kind of missile. Now you would calculate something
that's in the range of almost a foot of penetration.
The real penetration is about 7.5 inches but it is
enough to start pushing some material off the back.
This again now is what the frontside looks
like. You can see it's removed all the concrete right
down to the first row of rebar. The backside you can
see that it has spalling or scabbing off the back and
you can see the exposed rebar. Maybe you would have
a reasonable leak through that. You wouldn't know for
sure.
Then the last one is 18 inches at 143 feet
per second. This one I use, and we'll come back to it
again, because there's no spalling off the back face
at all on this one. Here you would calculate a
penetration depth by the simple approach of 18
centimeters and the actual measured value was more
like five inches. Reasonably close.
This one is important because it gives you
a frontside which looks like this again, very similar
to the others, but the backside -- I apologize. This
is very dark but it's the best copy at the time --
there's just some minor cracks. This is not something
that would drain the pool rapidly.
Again now we can work backwards and say what
is the criteria now that we ought to be using to
determine whether or not we have an impact that can
push something off the backside and open up a whole.
I took all the information that was
available on the 12-inch pipe so each one of these is
the same missile. I oriented them in terms of
increasing velocity here but you have to realize there
are also some other things changing.
A minor thing that changes is the strength
of the concrete because that depends upon the pore and
what they were trying to do for a particular test.
The concrete thickness is also changing. That goes
all the way from 12 to 24 inches.
I've listed here the measured penetration,
the calculated penetration, and then the ratio of the
calculated penetration to the total thickness. Then
the results over here on the right-hand column. These
two I've highlighted halfway through the table and at
the end so test No. 10 and 18 had complete
penetration.
Here you can see that the calculated value
is over half of the thickness of the wall. That's
when you can begin to think that, just as Graham said,
you've got enough reaction on the back surface that
you would open that up and then you would not be able
to convince yourself that you did not have a complete
path to drain the pool.
But these others that are very slight,
0.20 - 0.25, thirty percent of the calculated
thickness, hardly anything was observed on the
backside at all.
DR. WALLIS: If I really wanted to drain the
pool, I would drop something on the pipe.
MR. HENRY: Most of them have pipes that the
suction is on the inside.
DR. WALLIS: Penetration is on the bottom,
right?
MR. HENRY: I don't know about all of them.
The older ones may. Some may have that.
This is now what I use to formulate a basis
to say, okay, if we have enough energy that we could
penetrate something that begins to approach half of
the wall, they would have to think that we could force
a leak through the entire wall. That's just
summarized on this.
It says first we have the large-scale tests
which give us an idea of how well the simple
representation characterizes what was observed in the
concrete within a factor of two or three so that's a
good start. As I say, there are correlations for
these things. I don't want to confuse it with
correlations and get right down to the gut physics of
control.
DR. WALLIS: You would be extrapolating. If
you dropped a cask, it could well drop more than 60
inches.
MR. HENRY: Yeah, I'm going to get to that,
too.
DR. WALLIS: You've got to use some kind of
correlation to go up to that.
MR. HENRY: That's also why those high
velocity pipe tests give you kinetic energy a specific
loan that is greater than what you have even if it was
the terminal velocity in the cask.
The observations from the high velocity
missile tests, I think, are quite important and they
give you an idea of how you can scale up to the things
that are of interest because we do have to consider
dropping from things in the range of nine meters.
We find that only relative small cracks
appear on the backside as long as the calculated
penetration is less than half of the wall thickness.
If we don't have spalling off the backside, we
wouldn't expect any large leakage from the pool.
Therefore, we ought to have something that
is relatively easy to make up and slow drainage of the
pool, which I think are some important insights to put
into the physics side of the study because that is, as
Tim said earlier, one of the mechanisms for rapid
draining of the pool. It's in the study.
Now, what does this mean for the actual
cask? The height that has to be used in the spent
fuel pool is something like nine meters. I use nine
meters here. Maybe it's 10 meters or whatever the
particular event is.
It should include obviously the buoyancy and
the drag of the water and the buoyancy reduces the
acceleration by about a meter per second. If you just
take a dry coefficient of one, as Graham was asking
earlier, you get to something in the range of 20
meters per second is the terminal velocity.
The impact through water of a nine meter
drop gives you a velocity of about 12 meters per
second. That's only twice what you observed in this
one that I passed around. Kinetic energy wise, you're
only talking about a factor of four up from that
particular experiment.
If I go to the experiment and say I might
expect something that's about four times as deep as
what I saw, I'm only talking about maybe three
centimeters whereas the calculated value would be in
the range of 12.5.
DR. POWERS: Bob.
MR. HENRY: Yeah, Dana.
DR. POWERS: Maybe some stupidity on my
part. You drop the cask and it's going at some
velocity and hits the water. How long does it take
for that velocity to full up to the terminal velocity?
I presume that the terminal velocity in the air is a
lot higher.
MR. HENRY: Oh, terminal velocity in the air
is a lot higher.
DR. POWERS: Yeah, so that when it hits the
water, it's going faster than the terminal velocity.
MR. HENRY: It's barely above the water pool
DR. POWERS: Okay. So you're saying it has
almost no velocity when it hits the water?
MR. HENRY: Has almost no velocity.
DR. POWERS: Okay.
MR. HENRY: The acceleration is principally
through the water.
DR. POWERS: Okay.
MR. HENRY: Just to answer your question to
the extent that you asked it, you're at 12 meters. It
would take quite a bit longer obviously to get to 20.
At this point the drag is roughly half of the
acceleration. So now if --
DR. WALLIS: It also has an added mass.
MR. HENRY: You mean the divirtual mass?
DR. POWERS: Yeah, but, Graham, give him a
break. Yeah, Bob, correct that in your calculations.
MR. HENRY: Sure.
DR. POWERS: He doesn't have any decimal
points.
MR. HENRY: Everything we have comes out of
one dimensional and two-faced.
DR. KRESS: The added mass is taken care of
in the terminal velocity.
MR. HENRY: When you get to terminal
velocity it's just equilibrium.
DR. POWERS: Tom, you're correct if he
measured the terminal velocity on the cask but since
he's just calculating it, it doesn't take the added
mass.
MR. HENRY: The chairman told me to move on
here. The pool itself -- the bottom of the pool is
anywhere from 1.5 to 2 meters thick. My point here is
even the calculated value is an order of magnitude
less than the full thickness so putting all the
database together whether -- I apologize.
I took the rebar out because in essence it
has even less penetration than the pipe does for the
same kind of specific impulse. I could have put that
in also as a low.
When you put in the total database, you come
to the conclusion that it's going to be extremely
difficult for dropping a cask the full height of the
water pool to end up with something that's able to
open up the backside of the pool even if it's not
sitting on soil.
Some of these are sitting on either bedrock
or soil. Others are elevated. This tells you, I
think, that this concrete is very, very tough against
these kind of impacts.
DR. KRESS: What contribution did the staff
have for cask drops and draining the pool in their
technical study? Was it significant enough to worry
about?
MR. HENRY: It was significant enough to
worry about.
MR. COLLINS: Two times 10 to the -7, the
likelihood of a uncovery of the pool due to a cask
drop, two times 10 to -7.
DR. WALLIS: Bob Henry is saying it's zero?
MR. HENRY: I've been accused of saying that
in the past. I think what's really important when we
do these risk studies is to make sure we represent the
available technical basis. I would like to see this
kind of information in there.
DR. WALLIS: What did the staff use for the
mechanical probability with this scenario of the whole
development with most of your risk number probability
of the cask dropping at all?
MR. COLLINS: Most of the number was the
probability of the cask dropping at all.
DR. WALLIS: If it did drop, you're assuming
it went through the bottom?
MR. COLLINS: No.
MS. HENDRICKS: It was factor 1.
MR. COLLINS: No, it was not. It was 1 in
10.
MS. HENDRICKS: One in 10 was for the wall.
It was a factor of 1 if it hits the floor.
DR. WALLIS: So you and Bob disagree on the
maximum amount possible. You say it's 1 and he says
it's zero.
CHAIRMAN APOSTOLAKIS: So what's the factor
of 1 in 10?
MS. HENDRICKS: It's .1 if the cask is
presumed to hit the wall and it's a factor of 1
probability of failure if it hits the pool floor. At
least that's what in the appendix.
DR. KRESS: So you're saying instead of this
being 2 times 10 to -7 it ought to be 2 times 7 -8?
CHAIRMAN APOSTOLAKIS: Neglectfully small.
DR. WALLIS: It ought to be 2 to -14 or
something like that.
MR. HENRY: I would say the conditional
probability ought to be less than 1 in 100 for sure.
CHAIRMAN APOSTOLAKIS: Because in this
analysis that you have done something that might be
wrong or why not zero?
MR. HENRY: Every time I used zero in the
past you guys jumped all over me.
CHAIRMAN APOSTOLAKIS: That's the fun in it.
If I look at this and I don't have any other
information, I would say zero. Why shouldn't I say
zero?
MR. HENRY: I would say zero.
DR. KRESS: But that doesn't change anything
because it was already low enough that they didn't
have to worry about it.
MR. HENRY: There's a couple things. I
won't speak for the conclusion of the study but I will
say in writing the study that cask drop failing the
spent fuel pool shows up a number of times as a way
that you could rapidly drain the pool.
DR. KRESS: Okay. It's a perception.
MR. HENRY: Plus the fact I would like to
see these things referenced so that we know the
database that is used in the physical part as well as
the probability part has got a good strong foundation.
CHAIRMAN APOSTOLAKIS: So it's a matter of
confidence.
MR. HENRY: Yeah.
MS. HENDRICKS: I think what it also does,
too, is you end up going from it's dominated by
seismic to it's only seismic. I think that makes you
want to look a little more closely at what you're
doing to the seismic.
CHAIRMAN APOSTOLAKIS: I'm a bit confused
now. Dana, you keep telling us there is this table
where seismic appears as --
DR. POWERS: I showed the committee the
table.
CHAIRMAN APOSTOLAKIS: Yeah, I remember that
so why is it only seismic?
DR. KRESS: I don't remember the source but
it came out of AEOD.
DR. POWERS: The table was given to me. It
was repeated to be part of the staff study.
DR. KRESS: I don't know what that means.
DR. POWERS: It was part of the staff study
and it got corrected later or it was part of the staff
study and nobody believed it or what, but clearly it
would be erroneous to say that it is only seismic.
CHAIRMAN APOSTOLAKIS: If the table is
correct.
DR. POWERS: No, no, no. I don't have to
say if the table is correct. I don't have to put that
codicil in because there are clearly things that cause
pools and concrete things to fail other than seismic
event. There are clearly drain-down events that
occur. They are just small compared to the seismic is
what the staff is saying.
MS. HENDRICKS: E to the -8 to E to the -9
if you go through the report. The question is whether
you keep adding up a lot of small numbers.
DR. POWERS: You can add either the -9 for
a long time before you get any change of probability
here. I mean --
DR. KRESS: I can buy the argument for the
need to be consistent and have the right perceptions
and important stuff even though it probably doesn't
make any difference to the bottom line on how you
write the rule. I think there is some value in having
a technically sound argument.
DR. POWERS: One of the things is you would
have to worry about it seems to me in thinking about
these pools is they thermally cycle and they are going
to thermally cycle a lot during recommissioning.
If you don't have stress relief for that,
then we're cycling concrete dose fatigue. If you do,
then you have to worry about compression of the stress
release on them. I mean, there are lots of things you
can worry about.
DR. KRESS: Thermally cycled because of the
outside temperature change?
DR. POWERS: Yeah. Actually, it's not the
outside temperatures. It's the ground temperature
that's going up and down.
DR. WALLIS: Bob, did you do the seismic
calculation too?
MR. HENRY: No.
DR. WALLIS: I'm kind of intrigued about the
mechanism of failure under seismic loads of such a
massive concrete.
MR. HENRY: No, I didn't.
DR. WALLIS: You can shake a big mass of
concrete quite a bit without busting it.
DR. POWERS: You can look at the news
pictures of the concrete abutments from the 1994
earthquake and see it doesn't take much to bust up
concrete.
DR. WALLIS: It depends what it's connected
to and a lot of things.
MR. HENRY: One of the things I should
mention here that I didn't is in all these things I
didn't credit the liner strength which obviously all
the new pools have a significant liner on the bottom.
Last time we were here we talked a little
bit about fission product release so I won't belabor
this point.
DR. POWERS: I wish you would belabor it
enough to tell me why you didn't put the Chernobyl
incident on your list here. And explain to me a
little bit why we got so much ruthenium release so
early in the Chernobyl accident.
MR. HENRY: Which was metallic.
DR. POWERS: Say again?
MR. HENRY: Which was metallic.
DR. POWERS: It was metallic afterwards.
The release itself almost surely had to be as a oxide.
There's just no way to do it any other way. By your
own calculation the vapor pressure is diddly squat at
temperatures two times what Chernobyl ever had.
MR. HENRY: Yeah, but the temperature of the
fuel at the time of the actual --
DR. POWERS: Even at that, Bob --
MR. HENRY: It's pretty hot.
DR. POWERS: If we had, we would have been
boiling U02. I mean U02 will boil off before your
ruthenium will boil off.
MR. HENRY: One of the reasons I left
Chernobyl off of here was just because of these I see
as a lot more technically scrutable that we can get in
and exactly better understand the releases and the
relationship to zirc.
DR. POWERS: There is a little tiny test,
Bob, that don't allow the zirc to go up to
temperature, melt, and drain away.
MR. HENRY: I'm aware of that. I'm not
saying these are the final answers. I'm only saying
these are an important part of the technical basis.
DR. POWERS: Things can happen.
MR. HENRY: These type of things can happen.
It's just the rate at which it happens. As you know,
nobody really knows what the temperature was of the
fuel observed at the Chernobyl event.
Plus the fact that nobody is still quite
sure what the initial event actually looked like in
terms of how it was released because obviously there's
an explosion. It's a nuclear explosion and that also
scattered the fuel. All those things would influence
the rate at which things could be released.
These hit home to the issue of having air
there and steam which is an important part which is
particularly this Oak Ridge test, VI-7, and the CANDU
test because they tell us the relationship with these
cases where we have oxidation ongoing over a long
period of time, what's the role for competing of
oxygen with all these reactive metals.
DR. POWERS: That's not even close. I mean,
they are not even close. The reactive metal is so
reactive it will suck the oxygen out of anything
before you get to the ruthenium. I mean, I don't
think that's an issue. I mean, I don't anybody doubts
that ruthenium is an excellent getter.
MR. HENRY: Ruthenium or zirc?
DR. POWERS: I'm sorry, zirc is an excellent
getter.
DR. WALLIS: Tell me about the melting of
the cladding. Why doesn't the cladding flow?
MR. HENRY: I was going to get to that in a
second but certainly the upper part of the cladding in
this kind of an event is the thing that oxidizes
first. The more it oxidizes the stronger it gets in
terms of these events because it has a higher melting
temperature. The zirc oxide could sit around for a
much longer time and even support the zirc on the
inside.
DR. KRESS: Yeah, but it's apparently
brittle.
MR. HENRY: It is.
DR. KRESS: It cracks.
MR. HENRY: If you give it any kind of --
DR. POWERS: You want to be careful about
drawing experiences from steam because you get a much
higher energy input for unit of oxygen reacted and you
get a much less compact oxide.
MR. HENRY: I understand. Realize this is
steam and air in this.
DR. POWERS: A little bit of air wins the
battle every time because it's the nitrogen component
that is causing the problem.
MR. HENRY: I understand.
DR. POWERS: It doesn't take much.
MR. HENRY: My only point here was we
finally get back to analyzing it in a pool especially
for those systems that are only partially boiled down
so you've got a "blockage" at the bottom.
That's also steam and air because you've now
cut off your air supply except for whatever small kind
of curve flow you have from the top. That's why I put
both of these on here. I think they are very relevant
to the database and they are in the report that the
staff wrote. They didn't forget about these.
Our only point here is that we think it
would be very helpful instead of -- I shouldn't say
instead of -- in addition to the two boundaries that
they have for what the ruthenium release would be
let's use this information and also put a third curve
on that gives a best estimate.
When people look at these two boundaries
they have some idea of these two orders of magnitude,
whereabouts we think things are most likely to be.
DR. KRESS: It generally takes a lot of data
to get a best estimate. I'm not so sure we have
enough data to call anything a best estimate.
MR. HENRY: I would always use the data for
something because that's what you know. The bounding
parts you've kind of more or less pitched in and said
it's got to be between zero and 1. I realize that,
Tom, and that's why the two things we're going to come
down to.
First off, these are recommendations. And,
more importantly, we think this would be a study that
should have a peer review because it's not my opinion
and it's not any individual opinion around here. We
ought to make sure that what's known in the technical
community gets shown in this report.
DR. POWERS: I guess what I'm struggling
with is you're saying let's use these data for a best
estimate. I think what you're saying -- I think I may
agree with you.
MR. HENRY: It's time to go home.
DR. POWERS: That what you're saying is that
the staff merely needs to model the dynamics of the
clad because the data show that dynamics is of
overwhelming importance.
If it's there and it can oxidize, you not
only are not going to get any ruthenium release,
you're not going to get a decrepitation release. If
it's not there, then you've got another problem so you
have to mode the dynamics of the clad. That's the
best estimate you're talking about.
MR. HENRY: In essence, yeah.
DR. POWERS: Okay. I'll go along with that.
I mean, you can't argue with that.
DR. KRESS: I can't argue with that either.
Clad dynamics is a very tough problem.
DR. POWERS: It will attract your attention,
yeah. On the other hand --
DR. KRESS: I wouldn't mind doing some
experiments.
DR. POWERS: Well, on the other hand, I
think you may have the experimental base to do it
because they did a test in which they put enough
specific energy input into them to get the clad to
flow between the two oxide crust, one on the outside
and one between it and the fuel so that you would have
enough information to give yourself a criterion for
when the clad would flow down those things.
You probably would struggle with when the
clad would rupture and allow flow but, for an
unruptured case, I think the data exist.
MR. HENRY: A lot of things I think the clad
does that is shown on here and certainly the geometry
is influenced by the details of the pool, whether they
have boroflex or bural for PWR systems or nothing at
all. They just chose to control it with borax acid.
Obviously, for BWRs fuel assembly cans.
If you get to this issue of where the system
is partially drained down, it begins to look, except
for the decay power and the fact of atmospheric
pressure the whole time, just like the kind of
analyses that have been done for large break LOCAs for
BWR systems.
You can go do that calculation to see just
how much oxidation you're going to get with that
because it's being limited by how much water you have
in there. You get in the range of 10 or 15 percent
and from then on it's just accumulating molten metal.
Obviously where that goes is down to the bottom of the
pool and after a long period of time you start having
concrete attack, etc.
But the cladding itself controls how the
material begins to relocate because the first part
that melts is actually inside the cladding because, as
Tom said, the zirc oxide could be brittle but unless
you give it some kind of a privation, the molten
material drains down in the inside of that Zr02 to
begin with and it starts dissolving U02.
All those things are relatively complicated
but what they tend to do is give you melt relocations
and start blocking everything off. Especially when it
finally breaks through the cladding there's a lot more
molten material to go out than just the cladding by
itself.
All those things are part of what you have
to be concerned with if you go to detailed
representation. But I believe if you look through the
various things that have been done. Dana just
discussed the CODEX experiments. That's part of the
technical basis.
TMI is part of the technical basis. It's
not exactly what we're talking about here but it has
all the issues related to cladding dynamics and melt
relocation and even having the potential for some of
the fuel to be declad from the top part of the fuel
assemblies that's left on top of the debris.
These tests we just talked about certainly
you need to consider the fact that there can be a
small fraction of the material left on top of the
debris and that should also be assessed in terms of
it's temperature because it by itself is hard for it
to get very hot because it is cooling.
Unless the debris bed gets real deep it's
cooling by radiation. That has a very long -- if you
keep temperatures below 1,000 Kelvin it takes quite
awhile to release the material. Those are only
recommendations for expanding the technical basis.
DR. POWERS: Would you go over that debris
bed a bit for me? I just didn't follow you. I mean,
debris beds get hot pretty easily actually.
MR. HENRY: Debris beds can get hot pretty
easily depending upon their decay power. Of course,
we are dealing with things that are fairly small here.
If this material that collapses down and
accumulates a continuous mass and it's having a hard
time getting energy out of it, which it will, then all
the particulates sitting on top isn't receiving much
from below and it's only going to reach a temperature
that it by itself is able to power.
You have circulation through that bed and
you have radiation off the surface. An example, 10
percent of the material is going to cool very
effectively in the range of about 950K.
You can translate that back to what the rate
of release is that you again get from experiments that
have unclad fuel. That's a very slow release rate
given from ruthenium exposed to the air.
DR. POWERS: 1,000 degrees?
MR. HENRY: 950K.
DR. POWERS: You might want to go back and
look at the Oak Ridge disk. Their top temperature in
their test series was 950, I believe, and they got
quantitative release.
MR. HENRY: You mean this Oak Ridge test
here, VI-7?
DR. POWERS: No, no, the Lorentz tests that
were done back in the '60s. They ran a series of
tests that -- Tom, correct me if I'm wrong -- 450,
650, 750, 850, 950 and the 950 they got quantitative
release in less than 20 minutes.
DR. KRESS: That meets my recollection also.
DR. POWERS: They also got the tellurium and
a couple of other things were high at that
temperature. I can't remember what they were but
nothing is important except the ruthenium and the
decrepitation release.
Ah, that's a point. At that temperature
they did not get decrepitation release at that
temperature because the U308 that was forming was
centering almost as fast as it was spalling.
DR. KRESS: Decrepitation happened at lower
temperatures.
DR. POWERS: Yeah, they got decrepitation at
low temperatures but not at high temperatures.
DR. WALLIS: Dana, I think you should keep
all these results up your sleeve and have him do his
analysis and then see if it works.
DR. POWERS: He just has to do the Chernobyl
calculation.
DR. WALLIS: I know but I'm just suggesting
that to verify or validate his approach.
DR. POWERS: I'm going to give him every
parameter in the world that he can adjust. He can
adjust the flow rate through it. He can adjust the
temperatures.
I mean, I'm giving you all kinds of fudge
factors here, Bob. Just calculate Chernobyl for me.
MR. HENRY: I'll be happy to go back and
look at Lorentz' data. I based mine on the data that
the Canadian people did. That's all why we should
have some kind of peer review here to make sure that
not only is the database known but used in a
consistent manner.
In conclusion, the evaluations for the cask
drop event we think should incorporate this database
which is significant, which is full scale. It also
takes advantage of things that are done with very high
specific loadings.
We think that if you use that quantitative
approach, in essence that one is virtually impossible
or zero, George, because, again, that's why you have
a peer review, to make sure you get a cross section of
opinions.
I think you could take that one off of the
list. We think the risk that is shown in here should
also represent a third curve -- should include a third
curve to give some idea of what we think the best
estimate is because we did these experiments.
While we may not have the kind of database
there we have with other parts of the analytical
spectrum, it sure would be nice to give some
perspective in the integral sense what this really
means.
Lastly, we think that things like this, that
peer review cuts across the board of both industry and
academia as well as the regulator is essential because
then we make sure that whatever the technical basis is
gets surfaced, gets at least reported so people know
what is sitting out there they can use to make some of
these decisions.
DR. KRESS: Let me ask you about the second
conclusion. The technical study did use a relatively
high ruthenium releases. Yet, they found the risk to
be acceptable. Why should they go any further if they
already have acceptable risk?
MR. HENRY: A lot of it is the perspective
that comes from it. Then there is also some
conclusions drawn in the back about what that means in
terms of issues related to EP and others as well and
that comes from those studies where the risk is
acceptable.
DR. KRESS: Do you think there are other
considerations that might come into play eventually
because ruthenium has melted and land contamination
might be an issue if you use these bounding
calculations as opposed to best estimate?
MR. HENRY: I guess I'll beg on that one
until I knew exactly how they were going to be used in
the land contamination evaluations. I mean, it's hard
to compete with cesium in contamination.
DR. KRESS: I agree with you.
MR. HENRY: After five years ruthenium is
not much of an issue.
DR. KRESS: I agree.
MR. HENRY: Some of these pools are pretty
full.
DR. KRESS: If the risk was already found to
be acceptable, I was wondering if you thought maybe
when they get down to plant specific considerations
that they might find some that weren't acceptable by
the bounding calculations. Therefore, you might need
this best estimate as a basis for specific plants.
MR. HENRY: Lynette probably wants to say a
few things about this. I think the best estimate also
gives you a good idea of where you need to focus your
attention in the future as you do come across other
issues. That's why we would really like to see it.
MS. HENDRICKS: I would like to pick up
there. By definition, if you have studies out on
different aspects of the plant operation and one is a
bounding estimate and the other is a best estimate, in
doing your plant PRAs and all this stuff how are you
going to treat this risk?
There's no basis other than best estimate or
mean estimate with the understanding of the
uncertainty to apply this in a risk informed
situation. I think you are going to be hard pressed
to do that. Yet, it's on the books as something that
is by its bounding nature implies a lot of risk. I
think the best estimate is really critical.
Another reason it's really critical is
because, and this was actually captured in the study,
when you say the risk is acceptable, they actually
went so far as to say, "We're not thinking about
saying that you need a containment for the spent fuel
pool."
That is different than asking those
questions about do you need EP financial protection.
Those questions, I think, you may need to look more to
what is a negligible risk. For that, again, you need
a best estimate. That is the tool the commission
needs.
Ultimately it will be a policy call because
there is no magic number associated with financial
protection or EP. But certainly a best estimate would
allow them in a more absolute sense to say, "Okay, we
have EP over here for the plant and should we
determine on the basis if this risk is somewhat
negligible compared to this that we can justify
terminating those requirements."
Although the staff mentioned the stuff is
intended to be broader than just looking at those
requirements, the gist of the study was to address
three ongoing rule makings for these requirements.
The reason those rule makings are predominate is this
is the only opportunity to save money or conversely to
spend a lot of money unnecessarily in the
decommissioning phase.
DR. KRESS: When people say best estimate,
I'm never quite sure what they mean. I'm wondering if
you could tell me what you mean by best estimate?
MS. HENDRICKS: What the safety goal says is
you use mean values with a clear understanding of what
the uncertainty is.
DR. KRESS: So a best estimate is a full
distribution because in order to get a mean you have
to have a distribution. Sounds like a tough job to
get a best estimate for this particular issue.
DR. WALLIS: Could we put this in some
perspective? We heard from the staff that no further
physical chemical studies are needed because the risk
is so low anyway. What's to be gained by learning
anymore about this phenomena?
MS. HENDRICKS: I think what's to be gained
is the commission has to make harder decisions. You
know, is it safe enough compared to the safety goals.
Obviously it is but the harder decisions to
make are do you need the extra protection, expensive
protection, very expensive especially if it goes on
forever because you can't determine a configuration at
which point you can determine a heat removal that
would imply that you don't need to worry anymore.
Twenty years or so of EP and financial
protection are going to be grossly expensive compared
to how we look at funding for decommissioning today.
DR. WALLIS: So you're worried about the EP
cost and sort of custodial cost. You would like to
just leave the thing after a while?
MS. HENDRICKS: You don't leave it.
DR. WALLIS: Close it up or something.
MS. HENDRICKS: You don't leave it. There
are still people there.
DR. WALLIS: Not so many people.
MS. HENDRICKS: Stuff to be done. We have
an operator on site 24 hours a day with nothing to do
but focus on this pool.
DR. WALLIS: It must be the most boring job
in the world.
MS. HENDRICKS: Well, it may be boring.
DR. KRESS: But you don't feel you could
make those reductions and requirements on the basis of
risk alone?
MS. HENDRICKS: Not with the bounding
estimate. Another thing that concerns me about the
bounding nature of the study was we talked about what
will it mean. Okay, it's bounding because we just
couldn't do much better with all the conservatism
stacked up on a generic basis, but we imply that we
can do more on a plant specific basis.
But if it's done within the constraints of
the study, you're not going to get a different answer.
You're going to have it driven by seismic. You're not
going to be able to predict the configuration and you
are going to assume maybe avaticia conditions. I
don't see much relief going from this bounding to a
sight specific, unless I'm misunderstanding the way
the study is put together.
Do we have time to talk a little about
seismic?
CHAIRMAN APOSTOLAKIS: There's 10 minutes
left.
DR. KRESS: Ten minutes.
MS. HENDRICKS: Given the depth of my
understanding of seismicity, I don't think we have to
worry about this going too long.
This is a curve that we shared at the
commission briefing that shows the distribution of the
risk by peak ground acceleration. I'm going to flip
a couple up here quickly just to show that --
CHAIRMAN APOSTOLAKIS: What is this figure
now? Let's understand the figure. If you put it up
there, you have to understand it. You have the peak
ground acceleration on the horizontal axis. The
percent contribution. What does that mean?
MS. HENDRICKS: The percent contribution at
the different seismic bands of damage to the pool.
This represents convuling the hazard of the pool
fracturing on top of the seismic hazard on a plant
specific basis.
CHAIRMAN APOSTOLAKIS: This is conditional
on this peak ground acceleration so given that I have,
say, .9G, right? Or .8G, I go up and I see that there
is the probability of .2 of causing damage. That's
what that means.
MS. HENDRICKS: I don't know that it can be
interpreted that way. It's a percent of the total
contribution. It's more a way to show a distribution.
What percent of the seismic failure.
CHAIRMAN APOSTOLAKIS: It says condition to
spent fuel pool structure or failure probability.
That's what it says.
DR. WALLIS: There must be a frequency in
there somewhere because you would expect 2G to be more
effective than 1G.
MS. HENDRICKS: Well, the probability goes
down.
DR. WALLIS: That's right so probability is
in this, too.
MS. HENDRICKS: Right. Yeah. When you
convulve the risk on top of the seismic hazard.
CHAIRMAN APOSTOLAKIS: On the horizonal axis
you count accelerations.
MS. HENDRICKS: Right.
CHAIRMAN APOSTOLAKIS: Not frequency. This
is .8G, for example. If I go to the left, this is a
percent contribution to failure of the pool. That's
what it says.
MS. HENDRICKS: Right.
CHAIRMAN APOSTOLAKIS: So 20 percent of the
failures are due to .8G. Is that what it means? Then
I have to multiply by the frequency of .8G to get the
absolute frequency of the damage. That's the way I
understand it.
MS. HENDRICKS: No, that's already in there.
You took the hazard curve where you've already
convulved the risk of failure on top of the
probability of the event as well as the magnitude of
the event. This is just taking that curve and
parceling it out to show you the distribution.
I think Gary Hollihan's comment on this was
it's not surprising. You have a very robust structure
that even though the frequency of the seismic events
are larger, the probability of failure is small.
CHAIRMAN APOSTOLAKIS: So let me rephrase it
then to make it consistent with what you said. .8G
and I go up and find .2. .2 times 1.4 to the -6 will
give me the failure frequency of the pool that's
caused by a .8G acceleration. It's unconditional,
right? It's unconditional. It says it includes the
frequency of .8G. Who came up with this diagram?
MS. HENDRICKS: Our EPRI seismic experts
did. I was looking for a way. We all say --
CHAIRMAN APOSTOLAKIS: A way to confuse us.
MS. HENDRICKS: No, no, no. We all agree
it's all driven by uncertainty and that we just go on
and use the curve. I think this will help you
understand what it means when you use that curve.
I'll show you the other soon or you can just flip.
DR. SHACK: this is basically telling us
that most of the risk is coming from this far tail
with the big acceleration.
MS. HENDRICKS: Right. The median is in
excess of -- the median is at 1G so more than half
this risk that we're applying. The question obviously
is does this make sense.
CHAIRMAN APOSTOLAKIS: It's the same with
reactors. I mean, the seismic contribution comes from
accelerations of three or four times a safe shutdown
earthquake which is what you're saying here. The safe
shutdown earthquake is .15G and you're stuck seeing
the significant import from .5 and so on. It's
consistent I think.
MS. HENDRICKS: It's consistent to a point.
It's consistent to a point. What makes this different
and in some cases worse is just to look at damage for
this very rigorous structure we extended the curves.
The Livermore and EPRI curves that we use for plants
stop at a lower return frequency. They stop at, I
think, 10,000 years. We specifically took this out to
a million years which is going to influence the
results.
The rest of the curves, as you can see, you
get the same basis distribution and the same basic
conclusions which are that --
DR. WALLIS: This structural failure
problem, doesn't this thing leak before it breaks?
MS. HENDRICKS: This is looking for a
catastrophic failure because leaks you can replace.
Then you get your human error.
DR. WALLIS: Doesn't it leak? Even after an
earthquake it's full of rebar. Doesn't it just leak
in a few places? It doesn't just fall apart.
DR. SHACK: I think it's a question of
whether it leaks or it leaks like a sieve.
DR. WALLIS: That's got to come into the
analysis.
CHAIRMAN APOSTOLAKIS: I think we understand
it now.
MS. HENDRICKS: Okay. So the point is the
mean is about 1G. At this level between about .5 and
.7 if you go to the next curve is where you pull off
at least for surry. We'll go surry to surry numbers.
What this curve shows is that at the 50
percentile to get into this range where you get into
the real risk contribution between here and here, the
frequency is about E to the -6.
You are really reaching out to grab very
improbable events. You may do it in the context of
reactors but it isn't going to have the same effect.
I think you need to ask questions about how
appropriate it is to do here and potentially in other
areas of regulatory space where it's going to be the
sole contributor.
Are you going to drive all protection
requirements, all costs based on this seismic event?
It will be the issue for passive plant designs and it
could be the issue for the new plant design.
CHAIRMAN APOSTOLAKIS: I guess I don't
understand quite what the issue is.
MS. HENDRICKS: The issue is --
DR. KRESS: The issue seems to me is you
don't believe the seismic hazard risk.
CHAIRMAN APOSTOLAKIS: Why not?
DR. KRESS: I don't know. I mean, that's
what --
CHAIRMAN APOSTOLAKIS: The issue is you
don't want to use it.
DR. POWERS: I would say that she absolutely
believes the seismic hazard risk studies. She thinks
that they tell you that this is something beyond the
pale.
MS. HENDRICKS: Exactly.
DR. POWERS: I understand. Can I ask you a
question about your slide?
MS. HENDRICKS: Sure.
DR. POWERS: Which really has nothing to do
at all with spent fuel pools.
MS. HENDRICKS: No, it has everything to do
with spent fuel pools.
DR. POWERS: The figure does but the
question doesn't.
MS. HENDRICKS: Oh, right.
DR. POWERS: The question is you plotted the
15th and 85th percentiles. About the mean, why those
particular ones? There's nothing devious about the
question.
MS. HENDRICKS: Right. I don't know. It
wasn't the 5th and 95th.
DR. POWERS: Those are just numbers you had.
MS. HENDRICKS: That's what was provided.
DR. POWERS: Nothing devious.
MS. HENDRICKS: No. I know you're not
devious.
CHAIRMAN APOSTOLAKIS: Aren't you saying
though --
MS. HENDRICKS: I would never say that.
CHAIRMAN APOSTOLAKIS: Is this the -- I
mean, are you expressing different words the old
argument that if I have designed a thing against .15G
SSE, and I see that my damage occurs four times that,
three times high earthquake, I shouldn't just do
anything and just say it's good enough.
Is that what really you're saying which is
the argument that why to use a PRA. PRA doesn't
recognize this design basis thing and just goes all
the way until it fails the thing and what really
matters is the frequency. Is that the same argument
you're bringing up?
MS. HENDRICKS: It's real close but I think
what I'm trying to say is part of doing PRAs is
understanding the uncertainty when you understand that
the uncertainty is really completely driving you.
Another thing I wanted to point out on this
curve, and I don't have it marked, but if you look at
the delta or the decrease in probability of occurrence
between .1 and .3 because the air bands are fairly
narrow, you get a factor of 10 decrease.
Between .3 and .6 because your bands are
diverging, you have to go that much further to get a
factor of 10. And to go from .6 to 1 you don't get a
factor of 10. You get a factor of 5 decrease in the
probability of exceedence.
What that tells you is even if you were to
say it makes sense to design at these higher levels,
you wouldn't even get credit for it in the
uncertainty. It makes you ask what basis is that for
doing what this agency is supposed to be doing which
is determining what is appropriate to apply in terms
of additional requirements.
If you're looking at a curve that wouldn't
give you any credit for extreme redesign of your
plant, does that really seem like a logical basis to
regulate with?
CHAIRMAN APOSTOLAKIS: Well, I don't
understand this credit business. I mean, the
frequency of occurrence of this acceleration would not
go down significantly but you would certainly get a
hell of a lot of credit because you have built a
stronger facility. This is not the probability of
failure. It's just the frequency of exceedence of the
acceleration which is modern nature. I think we are
getting into a debate here.
MS. HENDRICKS: Yeah. Yeah. I think --
CHAIRMAN APOSTOLAKIS: Your point that we
should really understand better the details are going
to the analysis is well taken. Beyond that maybe we
shouldn't debate it more.
MS. HENDRICKS: Okay. Well, let me kind of
close out here.
CHAIRMAN APOSTOLAKIS: Okay.
MS. HENDRICKS: I think the way the study
progressed is nobody completely ignored this
phenomena. They came up with a number for the
seismic. They said seismic predominates. Very, very
conservatively is 3 or 2E to the -6.
Everybody recognized that was highly
conservative because of the conservatism in this and
the conservatism in the furgility estimates. They
even went so far, the experts, to say the risk is
acceptable and it's much lower probably than E to the
-6.
I think it begs a fundamental question if it
gets that close to negligible, does it make sense from
a public communication point to go ahead and do the
math. You know, risk times consequence and show these
health effects.
I mean, we say that negligible probability
is in the E to the -7 range and we are probably very
close and we have qualifiers and we have the ability
to show significant capacity beyond earthquakes that
would even be expected on the east coast.
I mean, it seems -- and what I'm leading to
is the question has come up in discussions with the
staff and it came up in discussions with the
commission of, well, then should we go back since we
have this process and resolve the differences between
the EPRI and the Livermore curves.
I think there is even a more fundamental
question here of how to treat seismic risk than just
can we bring the experts together and get them to
agree.
CHAIRMAN APOSTOLAKIS: The difference
between Livermore and EPRI is not that great anymore.
I understand Livermore updated their curves in 1993.
MS. HENDRICKS: But it makes for good
agreement in areas where you expect earthquakes but
the tails diverge significantly. For the spent fuel
pool study it made a factor of 10 difference which is
only a factor of 3 if you look at cask drop but if you
take cask drop out, you're back to --
DR. SHACK: And your last slide says we
should just truncate these suckers.
MS. HENDRICKS: I think there should be some
consideration of truncating. There should also be
consideration to come up with analysis which looks
more deterministic.
DR. POWERS: If my objective is solely to
look at the bottom line risk in these things, I think
I agree with you since you're trying to communicate to
people. Taking outlandishly high numbers and then
claiming very low probabilities to them is probably
not really communicating. I mean, taking an
infinitesimal probability with a big high spike
doesn't communicate.
If, on the other hand, I was doing this to
say, now, what kinds of things should I be doing and
what things have risk achievement worth and risk
reduction worth, then don't I want to go ahead and do
this?
MS. HENDRICKS: You may want to do it for
that reason. I think NUREG 1150 did it for large
seismic events. They looked at core damage frequency
but they showed some restrain and didn't go ahead and
do the multiplication to show these consequences. The
maddening thing about --
DR. POWERS: It was a lack of money. I know
those guys.
MS. HENDRICKS: They're cheap.
DR. POWERS: No, their sponsors are cheap.
They're profligate.
CHAIRMAN APOSTOLAKIS: Who is communicating
with whom here when we say communication?
MS. HENDRICKS: With the public. With the
public. When you go --
CHAIRMAN APOSTOLAKIS: This agency is
supposed to be doing good technical work so, I mean,
they have to communicate it to the public. I don't
know what else can they do.
MS. HENDRICKS: I think it's inflammatory to
take events of very, very low probability and multiply
times consequences.
DR. KRESS: I don't understand that. That's
what risk is, frequency times consequence.
DR. POWERS: Yeah, Tom, but let's be
practical. I'm very sympathetic to this point of
view. If you come out and tell me there is a
probability that 100,000 people are going to die in
Russia as a result of the Chernobyl accident, that
gets the headlines.
Now, the fact that the probability is 10 to
the -8 somehow doesn't ever make the headlines or
anybody's reading. If it was just the headlines, I
would probably say that's not the only thing the
public --
DR. KRESS: But I don't want to cook the
numbers.
DR. POWERS: I think she has a good sound
point here depending on what you're going to do. If
what you're going to do is look at the risk for this
pool, then I think truncation has its merits.
If I'm going to do it to then derive
something from the risks based on differentiating
them, risk achievement and risk reduction worth,
should I have a guy come by and check the pool once a
week, once a month, once a year, that kind of
question, then I think you shouldn't truncate it.
CHAIRMAN APOSTOLAKIS: Anyway, I wouldn't
want the staff to come here and tell us we did this
calculation but it would scare the public so we're not
going to tell you about it. I don't think that's
where you're going but --
MS. HENDRICKS: At some point, though, we do
that.
CHAIRMAN APOSTOLAKIS: That's nature.
MS. HENDRICKS: We don't do the
multiplication. I mean, your number that you gave at
the commission briefing, Dana, was 10 to the -7.
Maybe there's a point where you don't do the
multiplication, not that you deny the risk.
DR. POWERS: I think you've got good sound
reason to pick that number because what did we do with
the VANRs? (A) We don't put them in the PRAs and the
reason we don't is 10 to the -7. Hence, we don't even
put them in.
This "I don't want to scare the public" sort
of argument has merits to it. I understand that sort
of thing. I understand the purists. What I worry
about is when we say there's a probability of 100,000
people dying in Chernobyl -- or the Ukraine because of
the accident at Chernobyl, even though that's 10 to
the -8 probability, it does provoke people to do
things.
We get massive studies of radiation effects
that can't possibly ever yield a useful number because
of the background chance of deaths but they are in
response to these kind of flamboyant numbers. I
appreciate the point that's being made.
MS. HENDRICKS: I think, too, to look at it
from the reverse perspective, it's inflammatory, one,
but then also I think it sets up an expectation of a
level of protection that's unreasonable. I mean, are
you telling the public that they should expect the
next facilities to be built and to withstand --
CHAIRMAN APOSTOLAKIS: No. I don't.
DR. WALLIS: It's like asteroid collisions.
MS. HENDRICKS: Exactly. Everybody knows
about the big asteroid in the back of their mind but
if we did the numbers and showed how many people were
going to die, you may end up in a situation where
people would demand research into how to protect us
from the asteroid.
DR. WALLIS: And it will happen twice in the
age of the earth.
MS. HENDRICKS: As a result, money would be
taken away from our real risks in things that we can
more readily mitigate.
CHAIRMAN APOSTOLAKIS: This is a much bigger
issue than can be resolved in the next few minutes.
Are there any other comments you would like to make?
MS. HENDRICKS: We always end up on the same
note. Peer review. Maybe we could do some sort of a
peer review on the seismic. I'm not sure it's the
seismic experts because I don't think these curves
will ever change but maybe they shouldn't but peer
review about the appropriate use.
DR. POWERS: You know, I guess I realize you
are kind of insulting the committee. We were asked to
do a technical review and apparently what we did was
inadequate, I guess, in your mind, but we are not
peers. I mean, some of us think -- at least one of us
thinks he's a lord.
CHAIRMAN APOSTOLAKIS: I don't think we
should get into that. Any other comments?
DR. POWERS: How much more peer review do
you want?
CHAIRMAN APOSTOLAKIS: The commission has a
history of establishing external peer review. When
they did the 1150 review they did not intend to insult
the ACRS so I don't know why -- in this particular
case if they want to have a peer review, they want to
have a peer review. I mean, if they weren't happy
with an ACRS review, they wouldn't probably ask for
it.
MS. HENDRICKS: Nor does ACRS have the
resources to look at -- I mean, this study portrayed
many questions that would take a lot of resources to
complete.
CHAIRMAN APOSTOLAKIS: Any other comments
from the staff? The public? Thank you very much.
MS. HENDRICKS: Thank you.
CHAIRMAN APOSTOLAKIS: We'll recess until
2:45.
(Whereupon, at 2:26 p.m. off the record
until 2:46 p.m.).
CHAIRMAN APOSTOLAKIS: The next subject is
Management Directive 6.4 Associated with the Revised
Generic Issue Process. Dr. Kress again.
DR. KRESS: I'm busy today.
CHAIRMAN APOSTOLAKIS: Boy. Go ahead.
DR. KRESS: Well, just to remind the
committee, the ACRS has had misgivings about the whole
generic issue process for some time and had expressed
it in a series of letters. The staff decided to look
at the GSI process and see how to make it better, I
guess.
They came to us back in '99 with the revised
GSI process. As far as I can tell from reading your
letters and my recollection is we liked what we heard.
It sounded like a comprehensive way to do it and an
improved way. I think what we asked was could they go
out and do a pilot assessment of it and tell us how it
worked in practice.
I think that's what they did and today I
think we're supposed to hear about the results of that
pilot assessment. I'll turn it over to Harold
VanderMolen, I guess.
MR. VANDERMOLEN: Thank you, Dr. Kress. My
name is Harold Vandermolen. I have sometimes the
dubious distinction of being manager of the generic
issue program. On my left is Mr. Ronald Lloyd who is
our person in charge of management directive 6.4.
Yes, we're going to tell you about our
experiences with the trial application of the draft
management directive.
MR. LLOYD: Our first slide kind of recaps
some of the things that Dr. Kress has alluded to. We
initially had a draft version of 6.4 that was issued
back in early '99. We had an ACRS presentation at
that time talking about what was in that particular
document.
We did go through it by the end of July as
is shown on the slide. We actually produced an entire
version of management directive 6.4. There was a very
minor change to it that was proposed by OGC to add in
some lawyerese to the document on October 21 of '99.
Then we are here today, March 1, to provide some
lessons learned on what we actually found and
discovered as we tried this out on some reactor issues
and also some material issues.
We also have a tentative schedule that will
be on another slide. The purpose of our being here at
this time is to seek approval to go through and update
that management directive based on the lessons learned
we have to date.
DR. KRESS: Are you looking for a letter
from us then?
MR. LLOYD: We would be looking for a letter
probably at your convenience in the April/May time
period to tell us to proceed.
DR. KRESS: Okay.
MR. LLOYD: We'll go over that schedule
which is on a slide further back in the presentation.
Our next slide, please.
The process that existed earlier was
referred to as RES office letter No. 7. It basically
had three different steps in it: Identification which
was basically what is the issues. It had a
prioritization phase where we would go through and do
an assessment, a PRA type assessment.
We would then categorize so there's a high,
medium, low or a drop situation. Then comes the
resolution phase. I know this caused some problems in
the past with ACRS, you know, what does resolution
mean, what does resolve mean, and so on.
Beyond coming up with basically an available
solution which was the resolution phase of the old
procedure, nothing was really procedurized which was
one of the concerns, I believe, of several people.
The draft management directive, as you can
see on the left side of the table, currently has eight
stages to it which takes it through from the very
beginning to complete close out which is
proceduralized to verify the corrective actions have
been taken by licensees on some sort of audit basis
and a closeout inspection that would have to be
documented to do that with several steps in between.
To date we have experience of going through stage one,
two, and three of the management directive.
Next slide, please.
The next one here shows the issues that we
actually tried it out on. We had six candidate
generic issues, three reactor, and three materials
issues. The material issues we'll go through kind of
briefly now and then we'll talk about the reactor ones
on subsequent slides.
The material issues were basically all
identified in the October 2000 time period. As you
can see by the current status on the block to the
right of that, these issues were received by the panel
and an in depth discussion as to what the issue really
was and its risk significance and what should be done.
We are subsequently dropped from any further review by
the generic issue program in January and then again in
February of 2001.
They were basically dropped because of a
couple of reasons. They ended up being isolated
cases, i.e., not generic, where their risk
significance was lower, or there was already existing
regulatory guidance that was sufficient to maintain
whatever needed to be maintained as far as inspections
and verification that things were being done
appropriately.
MR. LEITCH: Were these dropped at step 2,
that is, the initial screening level?
MR. LLOYD: Yes. That's correct.
Next slide.
MR. VANDERMOLEN: Now, we are going to speak
about our experience with each of these issues in a
moment. In addition to those six, we felt that we
should have one issue that we sent to the old process
just for comparison purposes for this use, generic
issue 185 which had just come in.
The old process has been in place for about
20 years now and it has had over 800 issues processed
through it. Although it's been modified a few times,
it has essentially been unchanged in all of that time.
It was one of the first uses of
probabilistic risk techniques in agency decision
making. We did not feel that we could really evaluate
experience with the new process unless we had at least
one that we sent through the process with some
examination and oversight to be able to compare the
two.
Now, getting into the specific issues, we'll
start out with 186. Ron.
MR. LLOYD: 186 was one that I was
personally involved in as far as gathering data and
information. This was one that was proposed by NRR
and they had worked it for some period of time. They
figured that we needed -- they needed to have
additional technical basis for making decisions so
they wrote a letter and forwarded it over to research.
Then it came in a time period when we were
just starting to work with management objective 6.4 so
it got picked up under the new procedure. The panel
met successfully. We had a very good panel that did
get together. We had a couple of different meetings
with the panel to discuss additional information.
There are a few lessons learned out of that.
Initially when the panel looked at all the
information and the data it was classified as a
compliance issue and the recommendation would be that
the issue should be dropped from any further
processing through the generic issue program.
At that NRR requested that we actually do a
risk significance and gather some operating data that
they could use. It was decided we would continue on
with the generic issue program and continue to process
that issue.
After which time NRR then complained so much
that too much time was being burnt up by their people
coming to the panel meetings and so on and they
actually didn't budget sufficient amount of time in
their own budget for the entire year. Harold will
talk about that a little bit later. These were some
of the lessons learned that we came up with.
We actually ended up going out and visiting
eight different facilities of different design types
to get a good broad spectrum of what would be out
there from a risk perspective. We hit all the various
kinds of BWRs with different containment type designs.
We also did different PWR designs by different NSSS
ventures and so on.
DR. KRESS: Is there a record of drops that
end up in the LERs or somewhere?
MR. LLOYD: Yeah. What we did was we went
through new docs and went back to the beginning of
time and looked at all the different drops or problems
that had been recorded some place, either by vendors
themselves, crane vendors, licensees, inspection
reports, where we could find them.
Then we went out and actually gathered data
going back to the time that NUREG 0612 was generated
which was 1980 which then required licensees to kind
of beef up their crane program and come up with a lot
of different sorts of procedural requirements and
training requirements, electrical interlocks and so on
to make them more reliable, I would guess.
We got that data and we extrapolated to the
other different kind of design types and looked at the
number of refueling outages that they had had and came
up with a number of problems and also the number of
lists that you had. We had a frequency of failure to
start.
We've got some good data on that. That has
been put into a couple of different databases and we
expect to turn out a report on this probably within
the next month or so. That's 186.
187 is one that Harold will cover.
MR. VANDERMOLEN: Yes, 187 is not as far
along as 186. 186 has actually gone through initial
screening and is into technical screening, the third
stage of the process. The next two are newer and
haven't gotten quite as far. 187 is nearing the end
of initial screening.
This is one on cesium concentration. All of
these are interesting in their own light technically
but I would like to concentrate on the experience we
had with the procedure.
We learned all the lessons that we learned
in 186 and a few more. Difficulty encountered in
arranging panel meetings. Well, what did we learn?
What we learned was that the panel members that you
really want are people who are very much in demand.
It's not always easy to get their time. They are
often already busy and booked up.
This contrast would be management directives
requirement that we try and get initial screening done
in 30 days. This particular one, one of the
principals was called out to testify at an ASOB
meeting right in the middle of that period and we were
pretty well stuck. Not an impossible problem but it
did mean a delay.
We also learned that it is very wise to give
guidance early in the process on how the panel is
going to decide things. The management directive is
silent on this but the question is should a panel come
to its decision by unanimous consensus. Should it be
by a majority vote? If you do go by majority vote,
you have to talk about whether or not you are going to
allow descending opinions to be written.
These are not new questions for this
committee I'm sure. But in this particular case the
panel decided right at the beginning that it would try
and achieve the full consensus. Then what we
discovered was that even if everyone agrees on the
conclusion, it is possible to disagree on exactly how
you are going to get there. We are still resolving
this one, although we are pretty close to getting it
out.
DR. KRESS: Suppose you had -- how many
members are generally on the panel?
MR. VANDERMOLEN: Anywhere from six to --
what was the biggest one we had?
MR. LLOYD: Five to seven or eight.
Something like that.
DR. KRESS: Suppose you had four of them
that said drop the issue and two that said go on with
it? Do you have a binary system that says pick the
highest, the most problematic one? That is, send it
on and not drop it?
MR. VANDERMOLEN: Well, we really don't have
an answer to that one yet. This is one of the things
we have to resolve. There is always the oath that is
administered to a jury here in Montgomery County.
If any of you live in this area, you may run
into it, where the jury is sworn by the judge to keep
the jury with neither meat nor drink until a decision
is rendered. Fortunately, we're not too serious about
it. Although, I am tempted on some days, I don't mind
telling you.
We'll talk more about this when we come to
our recommendations.
Next slide, generic issue 188, also in the
initial screen stage. This is one on resonance
vibrations of steam generator tubes following a main
steam line break event.
Again, this is a very complex issue
involving inspection of disciplines that wound up with
a somewhat larger panel still. We had all the same
difficulties of getting an expert panel together.
Then when we got the panel together, the staff member
who raised the issue was unavailable because he was
involved in still other activities, some of them
involving the ACRS.
DR. KRESS: Is the issue that you might fail
the steam generator tubes by these vibrations?
MR. VANDERMOLEN: Yes. That's exactly it.
Strongly related to similar issues that I know you've
been considering.
DR. KRESS: Yes.
MR. VANDERMOLEN: Also, the principal person
that we wanted to talk to wound up having some
significant medical problems at the time and was
unavailable. We have to allow for these things.
One thing I should point out is when we
started this issue and, again, I think people around
here will be sympathetic to this, it is amazing how
much briefing material you sometimes have to provide
to committee members. This was a stack that was about
six inches high.
What did we learn from the process?
Obviously the panel preparation is not easy. Also we
learned in the discussion that it was not easy to tie
down the scope of the issue.
In any of these generic issues you have to
at a very early stage just decide where the scope is
and not change it once you start because otherwise you
will never come to consensus. We went through all of
this on this issue.
Having said that, I want to make a few
overall comments. When I say that there was
difficulty in panel preparation, I did not mean that
there was anything inadequate about the technical
discussion of the panel.
What this translated into was that people
had to spend a lot of time preparing for the meeting.
I have to say having been on these panels and feeling
a little skeptical when the whole thing started, the
technical discussion that I observed was of extremely
high quality.
I've been here for a while and I've been on
a lot of committees and panels and things like that
and I've observed many more. These were very
professional. There was a lot of good discussion not
only at the meetings but in between meetings as
members would discover new facts or documents that
were relevant.
They were sharing them with the entire panel
by e-mail and so forth and people arrived at every
meeting well prepared having read all the material.
It was quite a good focused discussion.
DR. KRESS: Who selects these panels?
MR. VANDERMOLEN: We recommend panel members
to our management but ultimately the office director
sings a memo after negotiating appropriately and
actually nominating the members.
DR. KRESS: You picked the panel and their
candidates by their expertise related to a specific --
MR. LLOYD: It could be based on their
expertise and also be different depending on whether
it was a reactor issue or a materials issue, whether
Research would make that move or whether NMSS would
make that move.
DR. KRESS: When you decide I would like to
have this guy here, do you check with him to see if
he's willing to serve?
MR. VANDERMOLEN: Yes. And we have to check
with his boss, too. The management sometimes has
strong opinions about this.
I should also say that we had some
difficulty in this issue and some of the others. Once
the people got going they were all set to go ahead and
charge in and try and solve the issue. We had to keep
it just on the purposes of the initial screening.
Next slide, Ron.
We also, as we mentioned, had an experience
with the generic issue we processed under the old
system. This is generic issue 185, control of
recuracality following small-break LOCAs and PWRs.
Again, this is a rather complex technical issue
requiring quite an in depth review.
Now, let me explain something here, the
difference in procedures which is why we were doing
this, of course. In the original procedure, the one
that we've been using for 20 years, there is no
initial screening panel. What happens is usually a
single analyst here sits down and investigates and
then writes up the issue doing a probabilistic
analysis and puts a package together describing all
the findings.
Then it goes out for a concurrence review.
Now, concurrence is nothing new to anyone here but
this is a little bit more than usual office
concurrence. It does go through our management, yes,
then the write up under the old procedure.
Parallel copies are sent. One copy is sent
to whatever person or group originated the issue.
This person may not agree with the analysis and
usually gives it a pretty thorough looking over.
Another copy goes to whoever, be it a single person or
a group, usually group, is going to have to work to
resolve the issue. That person may have very
different opinions from the first one.
Thirdly, we would send one to an independent
analyst, usually a PRA expert, just as a quality check
on the work. And there may be more. If you have
special technical areas we would try and get a review
by an expert in whatever technical discipline was
involved. They have all been collected together. The
comments are resolved and then it goes back to the
management review.
In this particular case the prioritization
write up was completed in six rather intensive weeks.
The concurrence review then lasted 197 days. I might
add also that this is not 197 days of benign neglect.
This is 197 days with gentle reminders, not so gentle
reminders, sometimes more forceful reminders.
I was often reminded many years ago when I
was in college and I worked in a public library and
had to remind people, sometimes professors at a local
university, that it was time to bring back the books.
It's not always easy to get this.
Well, why did it take so long? We all
agreed this was too long a period. Speaking to people
in retrospect it was probably pretty obvious. This is
a 20-page write up, one with a lot of meat in it, well
marbled with figures, tables, drafts, and equations.
It was not the sort of thing that you could just read
and pass on.
In fact, I think this is characteristic of
any generic issue write up. It's usually not the sort
of thing you can read a few pages one day and as time
goes by the next day pick up a few and so on.
This is something where you have to set
aside a few days and read it, ponder it, and
understand it, which people wanted to do. They were
very well motivated but these are busy people so in
retrospect I can't say that it's that surprising.
Clearly you've got to do better than this.
Moving on to materials issues which Ron will
discuss.
MR. LEITCH: But if I'm reading the data
correctly, it looks as though using the new system you
would be pretty pleased with 197 days.
MR. VANDERMOLEN: These thoughts have gone
through my mind. We'll talk about that in just a
moment.
MR. LLOYD: The next slide, No. 9, has to do
with the candidate materials that were presented
through the materials area, NMSS. As I mentioned
before, none of the candidate issues had specific
comments as they went through each one of those issues
but we did have some generic sorts of comments on the
process itself.
Recapping a little bit, going back to that
table that was presented earlier on slide No. 4, each
of the issues I thought NMSS did a really outstanding
job. These were brought up in, like I said, October
of 2000. There were panel meetings that came up very
soon after that.
I think they were very well prepared. They
looked through the issues and came to the appropriate
conclusion, each of those being dropped because of
situations that led them outside of the generic issue
program.
The leaking pools which was the first one
was initially a B&W issue with casks where they had
radioactive material in it, where they had water that
was leaking, and there didn't appear to be any
regulatory requirement to go out and track water that
was seeping out and measure the radioactivity and so
on.
They found subsequent to that that this was
an isolated case and the water never really got
outside of the bounds where it would do any problem
anyway so that one was eliminated.
The second one, unlikely events, I guess
would be parallel. You would look at kind of a duel
train system and a reactor situation. The ANSI
standard 8.1 allows them to take unlikely events and
say, hey, that's not going to happen and that sort of
fills the second train criteria.
Therefore, you don't have to worry about it.
Inspections realized that even though certain
licensees were taking advantage of this classification
of unlikely events, they actually had failures in
those systems where they were saying this was an
unlikely event. This was the key that got them into
maybe they had other problems and other sorts of
situations where we have unlikely events and different
licensees.
I looked at that one and that came back and
it was determined that once again this was an isolated
case. It was with the Portsmith gaseous diffusion
plant that the issue was at.
There were some changes made that came out
of that so there was a positive part of this. This
was the subpart H of 10 CFR 70 that got changes. That
was also a letter that was sent out to licensees
reminding them they should look into these areas and
fix those things.
The third one that was tracked through NMSS
was the gammaknife. That's the gamma stereotactic
radio surgery. There were several misadministrations
where they actually got the coordinates of where they
wanted the dose distributed incorrectly. In some
cases they got the two axis backwards.
Out of this rather than continuing on with
the generic issue process and calling it a generic
issue, they then processed it in a lower level format.
It was IEN generated. It was 2000-22 which told all
those types of licensees of the kinds of problems that
were generated, the human errors that were generated,
and so on.
This was an example of how the new procedure
would tend to go. You would have a quick analysis of
where's the risk and is it generic. Go through your
panel and come to a consensus with the panel hopefully
on whatever that decision would be.
If the answer is not let's process it as a
generic issue but let's call it some other form of
generic identification, then let's do it with the
easiest possible pathway. That's what NMSS did here.
It was done through the IEN or the genetic
communications process.
Next slide, please. I have some positive
things about the process itself. Several of these are
kind of interrelated as you look down through the
bullets. Most of them have to do with saving staff
resources at various stages of in the game.
The first one would be to save resources
obviously for those issues that would be proposed that
were of low risk that would never meet the thresholds.
Therefore, why waste your time doing analysis if
you've already got a very good idea from a group of
experts that would say, "If we don't need to do that,
we're smart enough to determine that we're not going
to meet that threshold so let's drop the issue
entirely or possibly go and do it under some other
sort of a format like the generic communications
process."
When you get down to the compliance issue,
if you actually determined there was a compliance
issue and you did some analysis on it, at least you
would have some sort of a technical basis to give back
to NRR or give back to NMSS or whoever is working the
issue that would help them in that compliance arena.
NMSS felt that the formality of the process
gave it visibility. At their meetings I was very
impressed that their panel was quite large. There
were a lot of people that attended the meeting, not
only the panel people themselves but other people who
were interested.
I think a lot of discussion took place at
that time so there was a lot of visibility given to
the process so they got some respect, I guess, is what
we're saying here.
They also thought that the flexibility of
the handbook, which is a guidance document, was
written not to be a verbatim compliance document where
you had to do step one, two, three in order and check
off all the blocks. It is a guidance document.
They use it as a guidance document and took
those things out of the handbook that best fit them in
addressing the issue at hand. In that way, they had
flexibility to do what they did based on what the
generic safety issue was.
The next bullet down, the processing time
may be shortened. I already mentioned that, that you
could eliminate unnecessary analysis because of
whatever the thresholds might be that you're not able
to meet them.
They also thought there was a consensus on
the scope of the generic issue early on. This was
perceived by many as something that the old process
possibly needed some fine tuning on and that was to,
as Harold mentioned earlier, too, try and define what
the scope of the issue is quickly so it's something
that everybody can get arms around that you can
define, that you can see what the analysis should be,
and then go ahead and work it as opposed to something
that is foggy and too broad.
Next slide, please.
I did all the good stuff. Harold can do the
rest.
MR. VANDERMOLEN: I get to do the
shortcomings and limitations which also exist. The
first one I think we've already discussed quite a bit.
It's been administratively cumbersome. That's party
because, as I said before, it's not hard to get
people. It is very hard to get certain people. The
people you really want for these things are often
people who are very much in demand, very heavily
committed.
What makes it even worse, particularly for
reactor issues, you're dealing with two offices and
all the implications that would imply. It's not easy.
You have to get memos back and forth on a high level.
It takes a while.
Also, the initial screening stage, which is
the panel meeting, may not provide a sufficient basis
for decision making. At this point you haven't done
a quantitative or any semi-quantitative analysis, just
looking at the issue.
One of the outcomes the panel can vote on
according to the management directive is to drop it
based on it having very little -- I'm trying to quote
it as exactly as I can -- very little chance of
meeting the threshold criteria.
That's not so easy to do in practice. In
fact, it's not always easy to make conclusions based
on an actual quantitative PRA analysis. When you're
trying to do it before you even do the analysis it
gets a little bit more difficult still.
The threshold for processing candidate
issues is not clearly defined for materials issues.
What that means is there is an Appendix C attached to
the handbook of the management directive that gives
the criteria for reactor issues in terms of LERF and
the usual PRA parameters. We don't have an analogous
one for materials issues. We really need to develop
something like that.
The documentation of "closed" issues for
materials issues could be enhanced. The existing
process is in an RES office letter. It only applies
to RES but there is a very definite process we go to
when we finally decide we're done with a generic issue
how do we close it out. The answer is a resolution
package is written and, as I'm sure you all know, it
comes down here for review.
After that, assuming all you gentlemen give
us a positive letter, the letter is attached to the
package, a cover letter goes on top. There is
definite guidance on who concurs on it but it's signed
by our office director and goes to the EDO. A similar
thing had not been developed for NMSS and that's not
really specified in the management directive.
Finally, we need a clear link between
management directive 6.4 and GIMCS. GIMCS stands for
generic issue management control system. It's our
tracking system for all of the generic issues.
There is no requirement to use that
specifically in the management directive. It just
says that you want to have quarterly reports, although
it's no problem when you're doing everything within
research because research administers the system.
In fact, we are upgrading the system event.
It used to run for many years under quarterly
publications and we are trying right now to put it on
the World Wide Web as well. It is public. Having
gone through all that, we would like to keep GIMCS as
our agency-wide tracking system and have everything on
all generic issues in one place.
Moving on to the next slide, these are more
observation. The last slide had shortcomings. These
are observations. Not all of them are problems. The
issues are complex. They do result in a significant
amount of review time and some conflicts with other
priorities.
The fact of the matter is nobody -- nobody
puts a simple problem into the generic issue process.
If you run across a simple question, you just go ahead
and solve it. You don't go through all this. The
ones we get are virtually guaranteed to be thorny.
I think if I did this over again I would
strike that word often. If you have a generic issue,
count on it. Actual practice says that it's not going
to be simple or straightforward. It's going to take
a little bit of effort to investigate and make these
decisions.
DR. KRESS: What's the incentive or
motivation for staff to serve on this panel? Because
it's their civic duty?
MR. VANDERMOLEN: For reactor issues we
provide them free donuts.
DR. KRESS: Okay.
MR. VANDERMOLEN: Other than that, they have
to get all of their work done. Even with the free
donuts we didn't always get everybody quite as eager
to serve as we want.
Issues can involve several disciplines. In
just the issues we've talked about, we've had to
consult people who had expertise in metallurgical
sciences, expertise in reactivity, thermal hydraulics
and thermal hydraulic phenomenon, post-accident
phenomenon maintenance practices, the engineering of
motor operated valves.
All of these things and you inevitably are
going to wind up with a fairly large number of panel
members. The higher the number of panel members, the
harder it is to get everybody together.
I say that in the context of the next four
bullets which manage to pull us in four orthaganol
directions, hard to do in a Euclidian space. The key
is that third bullet from the bottom.
It's difficult to establish a panel and
complete initial screening stage within the required
30 days. Now, for one thing, it can take you at least
a week and more likely two weeks just to get the metal
out establishing the panel. Going through all of this
in 30 days, it's just not realistic.
Backing up to the bullet right above it,
greater commitment from NRC staff will be required to
establish panels. What we ran into here is that we
discovered some of the offices were budgeting
something like $100 for the year for these panels.
That would be fine if they gave us
perfunctory things to do but you give us stuff where
we're going to have briefing packages that are this
big and then have a combined total of $100 for all
issues. That's not realistic. I'm not complaining.
The offices had to budget something and we didn't have
the experience back then to find out.
When we finish this up we are going to check
how many hours were spent on these panels and we'll be
able to budget more appropriately.
Going on, there is still a desire by NRR for
a more in depth risk based evaluation prior to
drafting an issue from the generic issues program.
When you get a generic issue somebody really believes
in it. Now, some of these come from DPOs and we don't
discourage this. This is how we handle a lot of them.
Most of them actually come with the full backing of
NRC management.
If the panel votes the thing into a drop,
possibly that makes some people happy but someone
thought enough of this issue to send it and it just
doesn't work just to have a panel vote. You have to
have some basis. The panel has to document it as well
and it's clear we have to have good guidance on how we
do that documentation. It's just not going to wash
just to say they voted.
Finally, as Ron mentioned earlier, if you
have an issue that is voted to be a compliance issue,
in theory that was supposed to say, okay, it's a
compliance issue. It's not a generic issue. Give it
back to NRR or NMSS. You've got the tools you need
already to fix it. Just tell them to go do what they
are supposed to do. Tell your licensees to do it.
Well, the people who have to enforce them
want to be risk informed as well. What happened in
reality with the heavy load issues, they came right
back and said, "Tell us how important it is." That's
an honest question and deserves an honest answer so we
wind up doing the problemistic evaluation anyway.
Going on to the next slide, a few other
observations. The previous generic issue process did
not work so well either. I think it's pretty clear
that we can't just go back to the old process. We do
need some improvements. But we do feel that this
whole process we've gone through comparing the two
issues, although it had its frustrations, and I think
Ron and I have more opportunity to be frustrated than
anyone, we do feel it was worthwhile.
I do want to add one more caution before we
get into the recommendations and that is these lessons
learned are not all inclusive. We haven't gotten
anything all the way through the process. We probably
have gotten most of the differences by going through
the first three stages. At this point we haven't had
any issue get past stage 3 of the management
directive.
Stage 4 when you get into technical
assessment, then you are going to contractors and
spending big money. You are really working the issue
and you'll have task action plans and all this sort of
thing. That takes a lot longer but that's what needs
to be done to fix these issues.
Going on, Ron.
MR. LLOYD: The next slide has to do with
the recommendations based on our lessons learned from
taking a look at both reactor and materials issues.
The first thing kind of going through in chronological
order as to really clarify the information that's in
Appendix A.
Right now Appendix A is basically a table
that list a whole bunch of items. It says to the
person that wants to submit one of these things
whether it's outside the agency, with the industry, or
inside the agency, "Here are a whole bunch of things
that you need to put down which would include what you
think the issue is, what you think the basis for the
issue is, whether there's a regulatory problem.
Also, what you think might be a solution to this
problem."
When it gets to the panel, they would have
a good idea of what the scope should be of that issue
and that there should be some sort of direction for
the panel to take from which to go out and do
something.
I think Appendix A, or the documentation
that would surround A, ought to be kind of flushed out
a little bit. I think there would be much time, I
think, spent in trying to figure out the scope. I
think that needs to be clarified so everybody knows
what the scope is. I think the appendix could be made
more user friendly so better information could be put
down. Another thing would be to actually enforce the
fact that whoever wants to submit a generic issue and
actually fill out Appendix A and provide that
information rather than saying, "Yeah, I've got a
bunch of information for you. Go do your thing," and
then not follow the procedure in its entirety.
Another one was the initial screening stage.
As Harold mentioned, it was difficult sometimes to
limit the scope there. If you've got a bunch of
engineers together, they want a lot of data upon which
to make a decision.
Consequently, the initial screening state,
which was really supposed to be kind of a basic look
at the background, what was on Appendix A, and some
basic kinds of information, really got into doing the
kinds of things that were in subsequent stages within
the management directive.
We had proposed here to limit some of the
scope and make sure that people that were on that
particular stage understood exactly what they were
supposed to do.
Which gets us down to the third bullet. We
felt that collapsing either stage three and four into
one stage or stage two into three, and most likely
stage two and three together would save some time and
cut back on the amount of administrivia you would have
to go through in order to process an issue.
Right now stage one is identification which
is basically here's my background information as to
why this might be a generic issue. The second stage
is the initial screening which is basically to review
the preliminary information.
When you get to stage three, it's technical
screening and that is supposed to be sort of a quick
look based on expert opinion and analysis of what the
situation is.
Then by the time you get to stage four, you
are more of an in depth look. The in depth look would
include things like going out to licensees, gathering
data, doing PRA studies, getting a contractor, and
other kinds of things to get more hard data upon which
to make a decision.
We felt there were some similarities in
here. Most likely the best thing to do would be to
collapse stages two and three and make one that would
best benefit the needs of everybody. It would save
time and cut back on the administrivia that really
isn't necessary.
Another issue that we had a problem as we
did our very first one which was on the heavy loads
area. We had our panel together and there were some
questions as to what adequate protection really meant
and what substantial --
CHAIRMAN APOSTOLAKIS: Did you answer it?
MR. LLOYD: You guys have answered that,
right?
MR. VANDERMOLEN: I have every confidence in
all of your deliberations.
MR. LLOYD: We went through and I took a
look at all the documentation and everybody trying to
figure out what adequate protection means. Of course,
we are still trying to figure that out today. We did
have a guest speaker come. Joe Murphy came and talked
with us and he gave us his best interpretation of what
all that meant. At least we had some input there.
As far as where the thresholds are, where
your safety goal fits in, we tried to explain what
that was. What we ended up doing was basically using
the guidance that we have in Reg Guide 1.174 and that
was just basically copied right into Appendix C of the
management directive. We used the best guidance that
we have to date.
If somebody here in the ACRS panel or in the
agency is able to come up with a good definition of
what adequate protection means with thresholds in
there, we would gladly put that into the management
directive and use some better information. Right now
we're using the guidance that is provided by Reg Guide
1.174.
Next slide. For the materials issues, as
Harold brought out, we really don't have any
thresholds that are out there that would give them a
good idea as to whether this is an adequate protection
issue or whether it's a safety enhancement issue or,
to a lesser extent, a burden reduction issue.
We ran into the same problem when we were
going through the inspection program trying to figure
out under each one of the basic categories how much
should we inspect, when should we inspect, what the
frequency should be, what the impact on core damage
frequency is.
If you're looking at an inspection issue
from a reactor standpoint, how should you interpret
that when you get over into the HP areas, the
safeguards areas, the materials issues, the
irradiators, those other kinds of things where you
have different consequences of did things go bad.
We really need to come up with something that NMSS and
others could live with that would provide better
guidance than what we've got right now.
The documentation, as Harold mentioned, we
really need to add some additional information in
there on actually how to close things out as we come
to conclusions from the panels and other things that
we know what we should do.
That means what is the proper format for
closing these things out. What's the proper
distribution for closing these things out. What is
the level of detail that is really needed. Can you do
this in a three or four-page memo or do you need
additional information that would attach some sort of
small report in order to make it go on to the next
step or to make the decision to drop that issue
completely. Those things could be clarified.
The other one is on GIMCS. It's not
mentioned in management directive 6.4 right now. At
the time we wrote the draft we thought that we
wouldn't be using GIMCS and so what's in the
management directive is NUREG XXX.
What we've really decided to do, I believe,
is to continue on with GIMCS but we are going to
update it and it will be put on the Web so people will
be able to get access to it. That change will have to
be made.
Clarify the level of technical analysis that
would be done within the scope of the MD. I think
this should be more explicit at some of the early
stages, once again to eliminate wasted time at
addressing some of the issues that are of low risk
significance and wouldn't meet any threshold.
The next slide we've got is a tentative
schedule for the next few months.
DR. KRESS: In your recommendations I didn't
see anything to do with the problem of getting a panel
together.
MR. LLOYD: Those kinds of things, as Harold
mentioned, those would exist with the old system.
DR. KRESS: Those exist no matter why.
MR. LLOYD: Those exist no matter why.
DR. KRESS: No way to help that process? Do
you have your own separate budget for generic issues?
MR. VANDERMOLEN: No, we do not. Generic
issues process is no contract dollars whatsoever.
DR. KRESS: It's a stepchild.
MR. VANDERMOLEN: And a fairly small team,
exactly half of which is sitting up here at the table
at the moment. The other two, I see one in the
audience. That's it.
MR. LLOYD: We're a small group. I think
that was one of our issues from the administrative
standpoint. We need to really raise this to a level
of where people could be made available and that they
would also realize what the approximate time
commitment would be.
I think that ought to be put in the memo.
I think offices should then be encouraged to stick
that in their budget. Assuming if we get the same
amount of generic issues coming in, if we got four to
six of these things in the period of a year, that
equals X number of hours and then the different
offices would just budget that amount.
DR. KRESS: Just getting that guidance to
the offices might help.
MR. LLOYD: We should be able to provide
guidance to the offices as to what to expect so it
wouldn't be a surprise to them. Good point.
Slide 16 shows what we would like to do is
to make some basic revisions based on lessons learned
by the end of this month. That would then go through
management review and research for a couple of weeks
and then try to get it from the Office of Research by
April 10.
We would also notify the EDO as to what
we're doing in a memo that would basically forward our
lessons learned. Then we would give everybody
approximately a month to go through and do their peer
review, get comments back to us by about mid-May, and
then take the following six weeks to address those
peer review comments, make the changes, and then get
it once again out of Research by the end of June.
It's an aggressive schedule.
DR. KRESS: It looks like May 11th time
frame might be a good time for us to look at your
changes and maybe make our comments then.
MR. LLOYD: Sure. You bet.
MR. LEITCH: I have a couple of questions.
MR. LLOYD: Sure.
MR. LEITCH: On the average how many of
these issues are being identified per year?
MR. LLOYD: I think we are probably looking
at right now maybe four to six.
MR. LEITCH: Four to six. And what is the
average age of the open issues?
MR. LLOYD: It depends on what time period
you're looking at. Some of these if you go back in
time, we were looking at 15 to 20 years on some of
these issues.
MR. LEITCH: I mean currently the ones that
are open.
MR. LLOYD: Currently that ones that are
open we've got --
MR. VANDERMOLEN: We've had some that go
back 20 years. Having said that, I realize also that
they go through a priority order based on these
quantitative estimates, not on their chronological
age. The ones that tend to be left are the ones that
were either very difficult to do or weren't of top
safety significance. I'm not trying to apologize that
they are that old.
MR. LEITCH: How many are open at the open?
MR. VANDERMOLEN: Ron Emerette, how many are
open right at the moment? How many generic issues are
open right now?
He'll tell you in a moment.
MR. EMERETTE: About 12.
MR. VANDERMOLEN: About 12 right now. We
are also working these generic issues and we're
getting for or six of them done.
DR. SHACK: Are you working them all with
the new process or half?
MR. VANDERMOLEN: No, the older ones, the
ones that are already in process, are still under the
old process. What really makes them move is that we
report to Congress every month on our progress and the
end date of our task action plan goes to the Hill.
They don't pay too much attention to our
intermediate milestones but if we don't make the final
one, we don't like to think about that possibility
because we try making deadlines.
MR. LEITCH: Are the goals of the new
process to reduce the time or improve the quality?
What was the problem with the old process? Is it a
quality issue or a timeliness issue?
MR. VANDERMOLEN: That's a little bit
difficult for me to answer for two reasons. One is 20
years ago Tom Cox and I together wrote the old process
so I guess I have a vested interest in it. The other
reasons is that this management started based on about
three years ago.
After the first draft had completed the
office reorganized and I came in one day and
discovered that I was once again in the generic issues
process and the other people were not. The day
ironically was April 1 of 1999 which I thought was a
most appropriate date.
I can't completely answer the question. I
think candidly people were having problems getting the
probabilistic analysis done. I think part of it was
they weren't following the old process that closely
either. It wasn't that the old process was so bad but
it was sufficiently difficult to use it and people
were not following it.
MR. LEITCH: Is the prioritization step in
the old process? I don't see a similar step in the
old process.
MR. VANDERMOLEN: It's the technical
screening stuff.
MR. LEITCH: I see.
MR. VANDERMOLEN: At this step it comes in
and we do a probabilistic analysis of it using what
resources we have in house. That is, although we once
had a contract on it, we don't go to licensees for
information. We don't spend major dollars on it.
What we found is that at least 80 percent of
the time based on the information we have in house we
can say that this is a drop. If there's any doubt,
that is, if we have unknown information, we put down
a conservative figure.
This is how it differs from most PRA
calculations that you'll see here. PRAs are supposed
to be completely realistic. Then if there's doubt, we
continue with it and then we go over to the next stage
and spend money and do it right.
DR. BONACA: Just as an answer, if I
remember, one of the major concerns was that because
the first screening was not -- didn't have sufficient
probabilistic analysis at the time, an issue was
classified as potentially generic significance, would
get there and then sit there for a long time and then
years later would be evaluated and then dropped. You
had a lot of potential GSIs. If I remember, that was
one of the issues.
MR. VANDERMOLEN: There was a significant
backlog. It depends. When you get a reactor event
happening, even one that is a precursor and wasn't
really directly any kind of major threat, but we try
and learn as much as we can. Every time that happens
a flurry of generic issues comes. These do not come
in on a regular basis. I'm almost afraid to say this,
but they tend to be somewhat stochastic in their
occurrence. You never know if you'll get none for a
long time and then you've got three or four together
all at once. That's one reason why the backlog
develops.
I have to say also that doing the
probabilistic analysis we read them after the fact and
they look very straightforward but they look very
differently when you start out with what I call NRC
Form 0, a blank sheet of paper.
It's not always easy. Sometimes it takes a
bit of thought, consultation, and work to actually do
the issues and we can't do all of them. We can do
most of them but we can't do all of them.
Other questions?
DR. WALLIS: You have a new process in draft
form and you want to assess it but nothing so far has
got beyond step three, technical screening. The real
work is done when you do technical assessment and
develop regulations and guidance. No one has done
that yet. It's a new process. How can you come up
with a well-developed process when you haven't tested
it yet in the main part?
MR. VANDERMOLEN: It is quite possible that
we would have to go back and revise it again. The two
aspects that I can say in partial reply, one is that
when you get into the later stage of the issue,
there's not that great a difference between the
existing and the new process. They run much more in
parallel. It's more that we track them further down
rather than at a certain stage turn them over to
another program.
The other is most issues don't make it that
far. It's a fairly rare occurrence but a very
significant occurrence when an issue makes it all the
way through. In most cases, even when we go into
complete technical assessment and really investigate
the issue, it becomes a major research program and we
may well find when we have all the information that
this does not meet the criteria for any kind of
regulatory action. It's pretty rare that things go
all the way through. We have to allow for that
possibility, of course, and we have to do it with some
vigilance.
MR. LLOYD: I would say that the old process
went through those same set of phases as far as if you
had rule making to do or some other kind of thing to
do. You had corrective actions to come up with.
You had to figure out whether or not
licensees actually implemented those corrective
actions and verified that they were acceptable, that
they would, in fact, solve the problem at hand. Those
kinds of things were done under the old process. The
difference was is there wasn't a procedure. They
tried to track it.
I think all we really did in that whole
formulation process was to go through NMSS, go through
NRR, and get that process that was being used, give it
our best shot at how to make it smooth, and go as
smoothly as possible, and then we put it down on a
piece of paper.
That's what you see under stages five, six,
seven, and eight, the backend of the entire management
directive. The frontend, as Harold mentioned, is
really the part where you try to resolve the issue.
After you've made a decision whether or not you're
going to drop it, most of these would end up in the
drop category.
It was perceived, I think, by the agency and
by Arthur Anderson, that helped us out on this, was
that if you had a committee of experts looking at it
initially, that you could come up with a pretty good
fix on whether or not this would pass risk thresholds
and, therefore, if we can get a real good fix on that,
why should we go through and expend all of the
resources to do that which we could do in a much
shorter period of time.
If, in fact, we decide to move it on to
another stage and to go out and do a real in depth
analysis from a PRA standpoint, then we would go ahead
and do that and there would be very little difference
between what we would do under this procedure than
what was done under the old procedure.
You might have one person doing it under the
old procedure as opposed to a committee doing it under
this procedure. We were looking basically for
efficiencies and then also some staff reduction time
because you've got to realize that we are cutting back
staff.
DR. WALLIS: Do you have measures of those
efficiencies and staff reduction times?
MR. LLOYD: We don't.
DR. WALLIS: Do you have measures of these
efficiencies?
MR. LLOYD: As Harold mentioned, we are
going to try to go back through and based on the
timekeeping situation look at what was actually spent
on these issues because we do have codes to charge
against and we'll take a look at that.
CHAIRMAN APOSTOLAKIS: Okay. Any other
comments? Thank you very much, gentlemen.
(Whereupon, at 3:46 p.m. the meeting was
adjourned.)
Page Last Reviewed/Updated Monday, August 15, 2016