482nd Meeting - May 10, 2001
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
482nd Meeting
Docket Number: (not applicable)
Location: Rockville, Maryland
Date: Thursday, May 10, 2001
Work Order No.: NRC-206 Pages 1-172
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
+ + + + +
482nd MEETING
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
(ACRS)
+ + + + +
THURSDAY,
MAY 10, 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.
PRESENT:
GEORGE E. APOSTOLAKIS Chairman
MARIO V. BONACA Vice Chairman
F. PETER FORD Member
THOMAS S. KRESS Member-at-Large
GRAHAM M. LEITCH Member
DANA A. POWERS Member
WILLIAM J. SHACK Member
JOHN D. SIEBER Member. PRESENT:
ROBERT E. UHRIG Member
GRAHAM B. WALLIS Member
STAFF PRESENT:
JOHN T. LARKINS Executive Director
ACRS/ACNW
SAM DURAISWAMY ACRS
ROB ELLIOTT ACRS
CAROL A. HARRIS ACRS/ACNW
HOWARD J. LARSON ACNW
JAMES E. LYONS Associate Director for
Technical Support
MICHAEL T. MARKLEY ACRS
ALSO PRESENT:
RAJ AULUCK NRR
PATRICK BARANOWSKY NRR
TOM BOYCE NRR
BENNETT BRADY RES
J.E. CARRASCO NRR
BOB CHRISTIE Performance Technology
EUGENE COBEY NRR
JIM DAVIS NRR
BARRY ELLIOT NRR/DE/EMCB. ALSO PRESENT:
JOHN FAIR NRR
HOSSEIN G. HAMZEHEE NRR
STEVE HOFFMAN NRR
TOM HOUGHTON NEI
RANDY HUTCHINSON Entergy Nuclear
PT KUO NRR
STEVEN E. MAYS NRR
HO NIGH OCM/RAM
DUC NGUYEN NRR
ROBERT PRATO NRR
DEANN RALEIGH LIS, Scientech
MARK RINCKEL Framatome-ANP
MARK SATORIUS NRR
PAUL SHEMANSKI NRR
JENNY WEIL McGraw-Hill
PETER WILSON NRR
STEVEN WEST NRR
TOM WOLF RES
GARRY G. YOUNG Entergy Services
BOB YOUNGBLOOD ISL
. I-N-D-E-X
AGENDA PAGE
Opening Remarks by the ACRS Chairman
Opening Statement. . . . . . . . . . . . . . 5
Items of Current Interest. . . . . . . . . . 6
Final Review of the License Renewal Application
For Arkansas
Briefing by and Discussion with. . . . . . . 7
Representatives of the NRC Staff and
Entergy Operations, Inc. Regarding the
License Renewal Application and for ANO,
Unit 1 and the Associated Staff's Safety
Evaluation Report
Risk-Based Performance Indicators
Briefing by and Discussion with. . . . . . .69
Representatives of the NRC Staff Regarding
the Staff's Draft Document Entitled, "Risk-
Based Performance Indicators: Results of
Phase I Development," and Related Matters
. 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 482nd
meeting of the Advisory Committee on Reactor
Safeguards. During today's meeting, the Committee
will consider the following: Final review of the
license renewal application for Arkansas Nuclear One,
Unit 1, risk-based performance indicators, discussion
of South Texas Project Nuclear Operating Company
exemption request, and proposed ACRS reports.
In addition, the Committee members will
attend the Commission meeting on the Office of Nuclear
Regulatory Research Programs and Performance, which
will be held at the Commissioners' Conference Room
between 10:30 and 12:30 this morning.
This meeting is being conducted in
accordance with the provisions of the Federal Advisory
Committee Act. Dr. John T. Larkins is a Designated
Federal Official for the initial portion of this
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 that they can be
readily heard.
I will begin with some items of current
interest. I'm very pleased to announce that the Board
of Directors of the American Nuclear Society has
elected Dr. Tom Kress, a Fellow of the Society. This
honor recognizes Tom's outstanding efforts in the area
of nuclear health, safety, and regulation. It is
certainly a well-deserved honor, and our Committee is
fortunate to have members of this caliber.
(Applause.)
CHAIRMAN APOSTOLAKIS: Because of the
unavailability of staff documents, Committee review of
the South Texas Project exemption request and spent
fuel pool accident risk of the Commission in plants,
which was scheduled for this meeting, has been
postponed to future meetings. As a result, there will
be no Saturday meeting this month, and the meeting
will be adjourned around 4 p.m. on Friday, May 11.
I hope the staff recognizes the impact on
ACRS resources of dropping items from the ACRS meeting
agenda at the last minute. The ACRS Executive
Director has been discussing this concern with EDO.
I'd like to draw the members' attention to
the items of interest, the pink cover. The three
speeches, or comments, by commissioners; comments by
Commissioner Dicus at the Texas Women's University
Honors Convocation on April 19 where she was honored
as a distinguished alumna of the University; the
opening statement of Chairman Meserve at the press
conference that he held on April 26, and remarks, or
a paper, that Commissioner Diaz gave at the meeting in
Germany of the Internationale Lander Kommission
Kertechnik on April 26.
Finally, a fourth item of interest is the
testimony by Mr. Lochbaum of the Union of Concerned
Scientists on Nuclear Power before the Clean Air,
Wetlands, Private Property, and Nuclear Safety
Subcommittee of the U.S. Senate Committee on
Environment and Public Works.
And the first item on our agenda is the
final review of the license renewal application for
Arkansas Nuclear One, Unit 1. Dr. Bonaca is a member.
Mario, it's yours.
DR. BONACA: Thank you, Mr. Chairman. Our
Subcommittee on Plant License Renewal met with the
applicant and the staff on February 22, 2001 to review
the license renewal application. At the time, we
noted two things: One, is that the application was
quite clear and easy to follow on the part of the
members that facilitated that review. The second
issue was that there was only a few open items
remaining between the staff and the applicant to be
closed.
Because of those two circumstances, we
recommended to the Committee that we would not have an
interim meeting, and therefore we did not have that.
We are here now to discuss the review of the final SCR
with open items closed. Therefore, this is really the
final report regarding license renewal. And with
that, I will let the staff and -- actually, I would
like to let the staff, first of all, initiate a
meeting.
MR. KUO: Thank you, Dr. Bonaca, and good
morning to the Committee. My name is PT Kuo, the
Chief of Engineering Section of the License Renewal &
Standardization Branch. The staff is ready to report
to the Commission its review of Arkansas Unit One
license renewal application. The presentation will be
made by Mr. Robert Prato this morning. He will first
give you an overview of the project, followed by the
applicant's presentation on its license renewal
application. And then Mr. Prato will summarize the
results of staff's detailed technical review.
I would like to just make one observation
since Dr. Bonaca already mentioned that from this
review I know of no open items left unresolved. And
the remark I would like to mention is that this review
is about eight months ahead of schedule. It's
remarkable; it's very impressive. I also was told
that Mr. Hutchinson, the Senior Vice President for
Entergy Nuclear, would like to make a few remarks
after I finish my remarks. And after Mr. Hutchinson's
remarks, I will then turn the presentation to Mr.
Prato.
MR. HUTCHINSON: I'm Randy Hutchinson,
Senior Vice President for Entergy Nuclear. We're
pleased to be here today and to be a part of this
review of ANO Nuclear Unit One's license renewal
process.
We, as you know, followed just behind the
Oconee application, which is a sister plant, and we
incorporated a number of lessons learned. In between
incorporating those lessons learned from the Oconee
process of what's been done in the industry and the
guidance provided by the Nuclear Regulatory Commission
in terms of license application format and that sort
of thing, we're able to put together a license
application, and as a result of that, one that had
very few open items, a substantially reduced number of
requests for additional information.
So, to us, this was really a pretty
pleasant experience. We found the license renewal
process to be stable and predictable, and it worked
very well for us. And Mr. Garry Young, the Project
Manager for our ANO project, will be making our part
of the presentation when we get to that. Thank you.
MR. KUO: And with that, Mr. Prato?
MR. PRATO: Good morning. I'm Bob Prato.
I work in License Renewal Branch in NRR. Before I get
into the overview, I'd like to inform the Commission
that we used Oconee as a benchmark for our
presentation, as we did for the Subcommittee. We did
that for a number of reasons. First of all, Oconee
and Arkansas Nuclear One are sister plants on the NSSS
side. The other reason is they used the same topical
reports that was used in the review for the Oconee
license renewal application. And the third reason is,
is that ANO incorporated a lot of the lessons learned
from the Oconee application. All of the open items
that Oconee had, most of them at least, were resolved
in the application for ANO 1.
So as I go through my presentation, I'm
going to be identifying some items. Those items are
the items that ANO 1 captured in their application
without any concerns as there were for Oconee. It's
not an intent to comment on Oconee's application.
Oconee did a great job. They were the first one up at
bat -- one of the first ones up at bat. And I just
wanted you to know they took advantage of the lessons
learned from the Oconee application.
To begin with the overview, the unit
description for ANO 1 is ANO 1 is a two-unit site
consisting of a Babcock and Wilcox Pressurized Water
Reactor and a Combustion Engineering Pressurized Water
Reactor. And it's located in Pope County in Central
Arkansas on Lake Dardanelle.
On January 31, 2000, the applicant
submitted a license renewal application for Arkansas
Nuclear One, Unit 1, the 2,568 megawatt thermal
Babcock and Wilcox Pressurized Water Reactor. Unit 1
construction began in 1968, and it went commercial in
1974. The current facility operating license expires
in May of 2014. The facility is similar to Oconee in
NSSS design.
ANO 1 site compared to the Oconee site,
Oconee is a three-unit Babcock and Wilcox facility.
It has a standby shutdown facility, which is unique to
the industry, which ANO 1 does not have. And they use
the Kiwi Hydroelectric Dam as their emergency source
of power. ANO 1 uses diesel generators as their
emergency source of power, and they have an emergency
cooling pond as an alternate source for the ultimate
heat sink.
Comparing the two applications, Oconee's
application was developed before the standard review
plan was issued. Therefore, it was broken down
basically into five sections. There was an
introduction, a scoping, an aging effects section, an
aging management review section, and a time-limited
aging analysis section. The ANO 1 application is more
consistent with the standard review plan in which they
combine Section 3 and 4 of the Oconee, so there's only
an introduction, scoping, aging management review, and
time-limited aging analysis.
In addition, they added an Appendix C.
And Appendix C are the aging effect tools. One of the
concerns with the Oconee application was applying
consistently the aging effects for the different
components that were inside containment and outside
containment. In the Appendix C, the tools that they
used resolved that concern.
As far as the safety evaluation report,
ANO 1 only had six open items. They included a sodium
hydroxide orifice, scoping question -- fire protection
scoping question, FSAR supplement additional
information needed in the FSAR supplement for, I
believe it was, a total of 11 different items. There
was some concern with the Medium-Voltage Buried Cable
Aging Management Program; there was some concern with
the Boraflex, and there was some concerns with the
trending of the tendon pre-stress forces. We will get
into all of those specifically as we go through the
aging management review presentation.
At this time, I'm going to turn this over
to Garry Young, of Entergy, who will cover the
application.
MR. YOUNG: Thank you, Bob. My name is
Gary Young, and I was the Project Manager for the ANO
1 license renewal application for Entergy.
The first thing I'd like to go over with
you is on slide 4, which is what we call the document
hierarchy for our application. The top item on this
slide shows the actual application itself, which was
the package that we submitted to the NRC for review.
Below that you'll see a list of several documents
here, which are what we call our on-site documentation
that was a backup, or supporting documentation, that
supported the statements that were made in the
application. And at the very bottom of that slide
you'll see the basic breakdown of the different types
of aging management reviews -- scoping and aging
management reviews that were done.
We broke them into categories. We had the
class 1 mechanical reviews, which were based on the
B&W topical reports. These are the same reports that
Oconee used in preparing their application. The
second grouping is the non-class 1 mechanical. These
are the systems that were not covered generically by
the topical reports, and we had to review those on a
site-specific basis. The third grouping is the
structural aging management reviews. Those were based
on some industry guidelines that were prepared by the
B&W Owners Group. And then the next one is the
electrical grouping, and these were based on Sandia
aging management guideline documents that were made
available to the industry. And those are the major
categories.
In addition to that, we did a TLAA review,
which was separate from the aging management reviews,
although closely related. We also did an
environmental review, which was part of the Part 51
review requirements for license renewal. And then we
summarized in one document all of the aging management
programs that were identified in all of these various
aging management review reports.
So, total, there were probably around 50
engineering reports, individual reports that were
generated to support the application that was
submitted to the NRC for review.
Okay. Then on the next slide, on page 5,
I'd like to go into the -- I'm going to talk through
each one of the areas of the application, a little
quick summary on how we did the review that went --
the results that were documented in the application.
And the first part of that is the scoping. And the
scoping is based on the rule requirements that
identify what is to be in scope for license renewal.
We used the guidelines from NEI 95-10 to prepare this
portion of our application.
There are three major categories of
scoping. The first category is safety-related
equipment, which is in Part 54.4(a)(1). There's a
definition there of what is safety related. For ANO
1, we had a site-specific component levelfic component level Q-list. And
this Q-list uses a definition for safety related that
matches the definition in the 54.4(a)(1). So we were
able to go right to our component level Q-list at the
site and basically print out a list of the equipment
that was in scope that met the (a)(1) requirement.
DR. BONACA: I have a question. I would
like a clarification. During the Subcommittee
meeting, you indicated that the scoping and screening
for mechanical class 1 components was done using the
B&W --
MR. YOUNG: Yes.
DR. BONACA: -- Owners Group topical
reports. Could you expand on that? Is it the whole
class 1 components, the mechanical ones were done from
those topical reports? Or did you have to use the Q-
list really to include also the Bechtel components?
MR. YOUNG: We did use the topical report
as the core of our review, and then we did a site-
specific comparison against the topical to ensure that
we were enveloped. We did have some areas where we
were different, and so we documented that in our site-
specific documentation.
DR. BONACA: Because you had a number of
Bechtel components.
MR. YOUNG: Yes.
DR. BONACA: I believe that they would not
be identified by the -- or would they be identified in
the topicals?
MR. YOUNG: No. The B&W components --
now, Mark Rinckel is here from Framatome, and he
helped us with that. Go ahead, Mark.
MR. RINCKEL: Yes. This is Mark Rinckel
of Framatome. We did include in the RCS piping report
Bechtel-supplied or AE-supplied piping. And so what
we had to do for Arkansas was to show how we're
bounded, and so we had to reference site-specific
information. It was included in the topical.
DR. BONACA: So also the Bechtel
component.
MR. RINCKEL: That's correct.
DR. BONACA: Thank you.
MR. YOUNG: Okay. The second category is
the non-safety-related structure systems and
components that are part of the 54.4(a)(2). These are
non-safety-related components that could prevent the
accomplishment of a safety function. For ANO 1, we
had very few components that fall in this category,
because of our definition of Q or safety related would
include most of these support type systems that are
sometimes classified as non-safety related.
We did have a few, though, that did fall
in this category. For example, our category two over
one seismic supports were in this category and a few
others. So we identify some additional equipment that
fell into this (a)(2) category.
And then (a)(3) was the final category for
scoping, which included what we sometimes refer to as
the regulated events -- fire protection, EQ,
pressurized thermal shock, ATWS, and station blackout.
And here we used our site-specific documentation for
each one of these reviews and identified the
structures and components that were relied upon to
accommodate these regulated events.
DR. BONACA: Just a question: On the
seismic two over one, you included not only the
supports but also the piping segments.
MR. YOUNG: Yes. Yes. When we did our
aging management review for the structural, we
included -- the way we did the program for evaluating
the aging effects on the supports and the piping is
the Maintenance Rule Walkdown Program. So when they
do that walkdown, they include both the hangers and
the piping that the hangers support.
DR. BONACA: Yes. Because I know it's an
issue that is being disputed on a different
application, and I just wonder -- in fact, I don't
know where the industry is on this. I mean is it --
you didn't have any objection to -- just your program
actually included the segments.
MR. YOUNG: Right.
DR. BONACA: So you didn't have to --
MR. YOUNG: The existing program.
DR. BONACA: -- make an exception.
MR. YOUNG: Yes, right.
DR. BONACA: Okay. Thank you.
MR. YOUNG: Okay. And that's the summary
of the scoping section of the application.
On the next slide, on page 6, is the
screening activities. Once we had completed the
scoping, we went through the screening process to
determine which components in those systems and
structures that were in scope required an aging
management review. Again, we used the material in the
rule itself and the guidance document that was
provided by NEI in 95-10.
The first effort was to identify the
passive structures and components that had an intended
function that required an aging management reviews.
And the definitions for passive and the intended
functions are covered in the rule. We applied those
definitions. We also then identified those passive
structures and components that were not subject to
periodic replacement. In other words, they were long-
lived and passive.
The screening for the mechanical
components was, again, done for -- the class 1 was
done using the B&W topical reports, the same reports
that Oconee used. And then for the non-class 1
mechanical, we did a site-specific review using the
guidance in NEI 95-10. For the electrical and the
structural components, these were also performed on a
site-specific basis using the guidance of NEI 95-10.
Okay. And that's a summary of the screening process.
Then on the next slide, on page 7, we go
into the actual aging effects identification. At this
point, again, it's all split up by discipline. We use
the guidance of NEI 95-10. The aging effects were
identified for the class 1 components using the B&W
topical reports. The non-class 1 was done on a site-
specific basis. At this point, we did rely on another
B&W guidance document, which is sometimes referred to
as the mechanical tools. This is the information that
we summarized in Appendix C of our application.
The mechanical tools was a document that
was created to help ensure consistency when we did our
aging effects review. And this document was used for
-- we had about 25 non-class 1 mechanical systems that
we had to perform an aging management review on. So
we relied upon this B&W guidance document to help us
go through that process and make sure that we
consistently identified the same aging effects for
each system. That document --
DR. BONACA: I'd like to ask a question.
This must have been a pretty time-consuming portion of
the effort.
MR. YOUNG: Yes.
DR. BONACA: If the GALL report had been
finalized by the time you were preparing the
application, would it have been much more efficient to
use that or would you have used that?
MR. YOUNG: Yes.
DR. BONACA: I'm trying to understand how
much the process would have been helped by the
existence of a generic document like the GALL report.
MR. YOUNG: This portion of the process I
don't think would be any shorter with the GALL report,
but we would definitely have used it. It would have
helped validate the conclusions that we came to. I
think the overall intent of the GALL report is for the
utilities to use to validate the work that is done,
and then be able to then, with some confidence, go
forward and say, "Well, this already has been reviewed
by NRC, so we don't have to worry about new issues
coming up."
DR. BONACA: It would certainly minimize
a number of questions.
MR. YOUNG: Yes. That's the area where I
think the benefit is, is that once you go through the
process and use the GALL to validate what you've done,
then you have some confidence going in with your
application that that's essentially been pre-reviewed,
and you know what the questions might be for that.
DR. BONACA: Thank you.
MR. YOUNG: Okay. The next area was the
electrical review. Here we used the Sandia aging
management guidelines and what's the spaces approach.
This was, again, to help ensure consistency. The
Sandia guideline was a basis for that. This is the
same type of review that Oconee did and Calvert
Cliffs, so we were following the examples that had
already been set for the electrical.
And then for the structural and structural
components, there was another B&W guidance document
that's sometimes referred to as the structural tools.
And that document was used, again, to help us ensure
consistency as we went through all the various reviews
of the buildings that were in the scope of license
renewal.
There were several -- at this portion of
the review, and in our application, there were several
lessons learned from both Calvert Cliffs and Oconee
that we applied in our application, and we feel like
that was a big part of the reason for the reduction in
the number of requests for additional information was
our efforts in this particular area to deal with
issues that had come up previously on the first two
applications in our application.
And in addition to that, we got some
guidance from the NRC in the form of a standard
format. And in that standard format for the
application, we also got some guidance on how to
present the material based on the review results that
came out Calvert Cliffs and Oconee. And we tried to
apply that lessons learned from the NRC staff, and we
think that, too, was a big benefit in reducing the
number of RAIs.
Okay. Then on page 8 of the slides is the
aging management programs. Once we had identified the
aging effects, we looked at the aging management
programs that were needed to manage those effects. We
identified a total of 29 aging management programs, or
actually major groupings of programs. Some of these
program titles you see here are actually a collection
of programs. Out of that, only seven of the 29 are
new programs. There were 22 that were existing
programs that were already in place at A&O.
The new programs included such programs as
buried piping inspection, electrical component
inspection, some pressurizer examinations, the vessel
internals program, spent fuel pool monitoring and some
others. And they're listed in a later slide.
For the existing programs, there were 22
of those, and those included such things as our ASME
Section 11 In-Service Inspection Program, our Borax
Acid Corrosion Prevention Program, chemistry control,
which included primary and secondary chemistry, our
Preventive Maintenance Program, and this is one that
included a large number of preventive maintenance
activities. So even though it's only listed as one
program, it includes a large number of individual
preventive maintenance activities. And, again, there
were a total of 22 of those.
These 22 programs probably represent aging
management programs for 95 to 99 percent of the
components that were in scope. The seven new programs
are actually very limited in scope as far as the
number of components that they cover. So the existing
programs cover the majority of the equipment.
DR. BONACA: Mr. Young, one of those
existing programs is this CRDM Nozzle and Other Vessel
Closure Penetration Inspection Program.
MR. YOUNG: Yes.
DR. BONACA: I'm sure this is a question
you were expecting to come today. And it clearly
gives you an opportunity to see how effective the
problem that you reference in the application would be
in light of the recent findings of the colony and
those at Arkansas, I believe. And I have a question
that you would comment on that. And also that comment
on possible changes you may have to make to the
program --
MR. YOUNG: Okay.
DR. BONACA: -- to deal with the findings.
MR. YOUNG: Okay. Yes, the cracking
that's been identified at Oconee and at Arkansas in
the CRDM nozzles was found using our existing aging
management programs. The Boric Acid Corrosion
Prevention Program was probably the lead indicator, at
least at Arkansas, when we went into the inspection --
the beginning of the refueling outage, we found the
boric acid crystals on the head of the vessel, and
that led to subsequent investigations that identified
the cracking that had occurred in the CRDM nozzles or
in the weld.
From that, we initiated our corrective
action program, which is also part of our aging
management programs to do a root cause evaluation and
to look at the extent of condition, and to look at any
needs to modifications to existing programs. And the
two programs that may be affected -- or actually will
be affected by those findings are the Alloy 600
Program and the CRDM Nozzle Inspection Program.
So those activities are currently ongoing.
They're part of our existing aging management
programs, and we expect some modifications to those
existing programs based on the operating experience
that we've gained recently.
DR. BONACA: Is this the first indication
of cracking that you have seen at Arkansas One?
MR. YOUNG: Now, Mark Rinckel is here from
Framatome. He's our expert. I'll let him answer that
question.
MR. RINCKEL: Yes. Mark Rinckel, from
Framatome. Actually, it's the second; the first CRDM,
but the first Alloy 600 issue was in the pressurizer
nozzle. It's a partial penetration nozzle that I
think failed back in 1991. So it's actually the
second occurrence at Arkansas.
DR. BONACA: Now, your program, if I
remember, was referencing inspections of Oconee, and
then you would perform the inspection based on the
findings from the Oconee inspection, correct?
MR. YOUNG: Well, I think on the CRDM
nozzle, we are doing inspections in addition to the
inspections at Oconee. We're sharing information --
DR. BONACA: Okay.
MR. YOUNG: -- but we're not dependent on
Oconee in this particular case. There are some other
programs where we are dependent, but in this case we
are doing our own inspections and then comparing those
results with Oconee to see if either one of us need to
change our programs.
DR. BONACA: So you have a commitment to
inspection at every shutdown for refueling?
MR. YOUNG: Mark, do you know the
frequency on the inspections for the CRDM nozzle?
MR. RINCKEL: I think that's still being
determined, but initially, and what's stated in the
application was that ANO was amongst the least
susceptible and was not predicted to see any cracking
until after 48 fpy. Once the incident at Oconee Unit
1 happened, that changed everything, and, as Garry
said, the program has now changed. But I think that's
still being determined what the inspection frequency
will be. That hasn't been determined.
DR. BONACA: Because I have an exhibit
from some presentation from Framatome that shows
Arkansas to be the one with an inspection at every
cycle. That's what I thought. That's why I asked the
question.
MR. YOUNG: Yes. There have been some
very recent changes, and all of this is being
coordinated through B&W Owners Group. So the ultimate
solution for the inspection frequency, both at
Arkansas and at the other B&W plants is coordinated.
There have been meetings with the staff on that
specific issue. The long-term resolution will be the
findings from the B&W Owners Group effort, and we'll
incorporate those into our aging management programs.
DR. BONACA: How difficult are these
nozzles to access for inspections at Arkansas?
MR. YOUNG: They're fairly difficult, yes.
You have to get --
DR. BONACA: You do not -- I mean so many
of the other PRWs have difficulty because they have
insulation, and it makes it very impossible to see
from outside unless the full installation is removed.
MR. YOUNG: Well, I believe these
inspections are on the inside -- the welds themselves
are on the inside of the head, so I know the
inspections that we did and the weld repair were done,
obviously, with the head off the vessel on the
headstand, and they had to work on the inside of the
head. Mark?
MR. RINCKEL: Again, Mark Rinckel from
Framatome. The control drive service structure that
we have at the insulation is not an issue. We're able
to see really all the CRDM penetrations with a visual
inspection. And I think we differ from Westinghouse
and CE in that regard. So being able to see the boric
acid from the outside is not an issue for us. And
we've done safety assessments to show that the cracks
are predominantly axially-oriented; this is not a
safety concern for the B&W design plants. So we
should be able to see these.
DR. BONACA: Now, just a question I have
is regarding Oconee 3 since --
MR. RINCKEL: Yes.
DR. BONACA: Oconee 3, when was the last
inspection they had prior to the February 2001
inspection?
MR. RINCKEL: As far as visual from the
outside, I can't answer that. The initial integrated
program had Oconee Unit 2 as the lead indicator or
would be the lead plant, and that inspection included
a volumetric from the underside of the vessel. And I
believe that was somewhere around 1996.
DR. BONACA: I'm trying to understand.
This inspection comes and there are up to nine nozzles
--
MR. RINCKEL: That's correct, yes. At
Oconee Unit 3 there are nine.
DR. BONACA: What is the rate of
development of these cracks? That's what I'm trying
to understand. And to understand that rate of
development I have to understand the period that went
between the two inspections.
MR. RINCKEL: Yes. I think the EPRI model
that was used to rank the CRDM penetration is being
re-looked at and has been completely revised. And
they're really looking at Oconee Unit 3. Everything
is being normalized now to ONS 3, and I think all of
the NW plants will be inspected -- TMI and Crystal
River 3 as well.
MR. ELLIOT: This is Barry Elliot, NRC.
There are two issues here: CRDM nozzle cracking, and
there's a susceptibility model which was used to pick
the worst plants. There's a new issue that has just
occurred, which is Alloy 600 weld cracking. That is
a problem we're having now. And that's a separate
issue. It is being addressed by the staff currently.
As far as the susceptibility model that
inspects the 600 nozzles, that was used -- that model
was used by this plant in an expanded scope beyond the
CRDMs, used for other components. And they have
identified other components that need inspection. The
susceptibility is in question because, as Mark said,
ANO 1 was not one of the limiting plants, and yet it
had the cracking.
The cracking is probably also related to
the weld problem, and that weld problem -- the problem
is that once the crack goes through the weld, the
reactor coolant now is not -- it is no longer under
priority chemistry control. It is now outside the
confines of the reactor coolant pressure boundary, and
it doesn't have the same chemistry anymore. So the
rate of crack growth is going to change from what we
-- which all the models predict. This is today issue.
It is being evaluated today, and we recently put out
an information notice on this.
DR. SHACK: But I think Mario's question
was in the context of the license renewal application.
When you have a new phenomena here where you do have
the weld cracking, you now have the potential for
cracking from the OD of the nozzle, the
circumferential cracking, which is really different
than what people -- the safety evaluation was looking
at axial cracking and the conclusions. But that's
incorporated into the license renewal process in the
sense that you're doing this experience update, and
that's why the staff feels that it can go ahead with
the approval, even though you really don't know what
the answer really is going to be at this point.
MR. ELLIOT: Yes. Our work is through the
current license and whatever occurs during the current
license, whatever inspections are going to be
required, will be carried forward into the license
renewal period.
MR. PRATO: Part 54 requires that.
DR. BONACA: Although if an issue of this
nature would come during the extended license period,
you would have the same ability of working with the
licensee to develop changes to the program. So I mean
this is a -- okay.
MR. KUO: Yes. That is exactly right, Dr.
Bonaca. The regulatory process carries forward into
the license renewal period. Whatever the resolution
here in today's space will be carried into the license
renewal space.
DR. SHACK: It just seems a little strange
at the moment that you're approving an aging
management program for the drive nozzles when at the
moment you don't have an acceptable, or you don't know
whether you have an acceptable aging management
program.
MR. ELLIOT: Well, I don't think we do
have an acceptable aging management program simply
because the cracks went right through. But we will,
and that's -- you know, over the long-term that's what
the goal is, and that's where we're headed.
DR. SHACK: Okay. But that's a today
issue, and it will be addressed and will just carry
over.
MR. ELLIOT: Yes.
DR. BONACA: I'm not sure whether I would
characterize it as not an acceptable aging management
program for license renewal, I mean. Today it is.
MR. ELLIOT: Yes.
DR. BONACA: For license renewal, all I
need to see is you're flexible enough to incorporate
promptly changes that result from the findings that
you have. I mean we cannot expect that there will be
no issues arising over the next 40 years of operation
or whatever. The important thing is that there is a
program in place, and it is flexible enough to
accommodate and to incorporate changes.
MR. ELLIOT: Yes.
DR. BONACA: So you would conclude, too,
that --
DR. SHACK: I conclude that you're right.
(Laughter.)
DR. BONACA: -- for license renewal that's
an important issue.
MR. KUO: I might also use this
opportunity to mention that there are other technical
reviewers here sitting in the audience that they are
ready to answer any questions you might have later on.
DR. BONACA: Yes. No, I think it would
inappropriate for us to expect a solution to this
issue right this minute. We are not expecting that.
But, certainly, an understanding of how, from a
perspective of license renewal, the extent to which
the programs which are committed to in the LRA are
able to accommodate the findings. And that's really
proof to us that the programs are effective.
MR. YOUNG: You know, we have the
enveloping aging management program of our corrective
action process, our Non-Conformance Program, and that
applies to all of our individual aging management
programs, including the CRDM Nozzle Program and the
Alloy 600 Program. But if we were to have some
problem with one of our other programs in the future,
they too would be subject to that Non-Conformance
Program, which would include an evaluation of the root
cause of the problem and corrective action, which
would possibly include changes to those programs,
either in frequency or inspection methods or scope.
So all of our aging management programs are subject to
that adjustment as we get additional operating
experience.
DR. BONACA: Any other questions on this
issue? Thank you.
MR. YOUNG: The next slide, on page 9, is
the time-limited aging analysis. And, again, this was
done as somewhat of a separate activity from the aging
management reviews, but it was also done in
conjunction with those reviews. We had a list of the
TLAAs, which were evaluated. This list is very
similar to Oconee. It included such things as the
reactor vessel neutron embrittlement, metal fatigue,
EQ, reactor building tendon pre-stress, and boraflex
in the spent fuel racks, in addition to some others.
So, again, this list was consistent with the previous
applicants, and we performed our evaluation and
documented the results in the application.
Okay, the next slide, I would just like to
conclude on the application itself. We, again,
utilized a number of the lessons learned from Oconee,
from Calvert Cliffs, and from the rest of the
industry. The number of NRC requests for additional
information was reduced relative to the Oconee
application. We had approximately 265 RAIs for
Arkansas versus about 350 or so for Oconee. Again, I
think this, at least in some sense, reflects the
application of lessons learned. We took the RAIs from
Oconee and tried to address as many as we could in our
application. Obviously here there's still room for
improvement. We'd like to get that number even lower
than 265, and I think subsequent applicants will be
able to do that.
On the number of SER open items, we had
six and Oconee has approximately 49. Again, we
applied lessons learned from Oconee to assist us in
reducing this number and the lessons learned from the
NRC review of the Oconee application.
In summary, the license renewal
application is stable and predictable, and we
appreciate the efforts of the NRC staff to help us
reduce the schedule for the review of this application
from the original 30-month schedule, which we started
out with in February of 2000, and we're now down to a
17-month schedule. So we really appreciate the
efforts that went into accomplishing that.
And in particular, we'd like to
acknowledge the effective management of this review by
Mr. Bob Prato on the safety reviews and Mr. Tom Kenyon
on the environmental reviews. Both of these
individuals were a great contribution to this process,
and we appreciate their efforts. And that's all I had
on the application. Thank you.
DR. BONACA: Thank you. Mr. Prato?
MR. PRATO: Okay. On the safety
evaluation, again, I'm Bob Prato. At the end of each
of the major topics -- scoping, aging management
review, and time-limited aging analysis -- there is a
slide on the open items that were identified at the
end of the first safety evaluation. The last four
pages of this handout has the summary of the open
items and a summary of the resolution of each of those
open items. So as I go through this, we'll stop and
we'll talk about the open items that we found in each
of these sections. And both myself and Mr. Young will
try and answer any questions you might have as to the
resolution.
I'll begin with scoping. If you remember,
the Oconee application had a number of questions on
the scoping. Both plants, Arkansas Nuclear 1 and
Oconee Nuclear Station were originally designed to
barriers to release of fission products. However, in
1987, about that time frame, ANO 1 performed a design
basis reconstitution. As part of this design basis
reconstitution, they revised a Q-list to criteria that
is consistent with 54.4(a)(1) for safety-related
components and 54.4(a)(2) for non-safety-related
components, which can effect safety-related functions.
They used the accident analysis in the US
FSAR. They used the environmental and exterior vents
in their design basis reconstitution. They used site-
specific and applicable industry operating experience,
and they also used generic communications. The
applicant also incorporated lessons learned from the
Oconee scoping review. The chilled water system,
skid-mounted equipment, structural sealants, ANO 1
ventilation sealants, water stops, expansion joints,
electrical cables, fire-detected cables, and buried
pipe were all not excluded from the aging management
review in the original ANO 1 license renewal
application.
ANO 1 aging effects discussed and accepted
by the staff were consistently applied by the
applicant based on Appendix C of the license renewal
application, as discussed previously. And corrective
actions, ANO 1 committed to 10 CFR Part 50, Appendix
B for all license renewal corrective actions, safety-
related and non-safety-related both. That includes
corrective actions, the confirmatory process, and
document control activities.
As far as the open items for scoping,
initially the applicant did not identify a flow
control orifice -- I'm sorry, the applicant did not
identify flow control as an intended function of an
in-line orifice that controlled the injection of
sodium hydroxide for pH control. In resolution to
this item, the applicant did include the flow control
function. And because the orifice is made of
stainless steel and is subject to cracking, the
applicant added the orifice to the inspection program
used to manage other stainless steel components within
the sodium hydroxide system as their resolution.
The second item was fire protection.
There were five sets of components that the staff was
concerned about. They were the fire protection jockey
pumps, the carbon dioxide system, fire hydrants, the
water supply to the low level radwaste building fire
protection system, and the piping to the manual hose
station as being within the scope of license renewal
and subject to an aging management review. The
applicant took the position that it was never part of
the current licensing basis, these components. And
the staff felt that it was necessary to include them
based on the rules under Part 50.
We had a number of meetings on these
items. What the final resolution was was that the
applicant realized that even though it wasn't part of
their initial current licensing basis, that the fire
protection jockey pump and the fire hydrant should be
included within the scope of license renewal. And
they did include it, performed an aging management
review, and identified aging management programs for
those components.
MR. LEITCH: When you refer to the fire
protection jockey pump, are you speaking specifically
of the casing?
MR. PRATO: Just the casing; yes, sir.
MR. LEITCH: Just the casing. Okay, I
understand. Thank you.
MR. PRATO: As for the other three items,
based on the applicant's presentation to the staff,
the staff found that these components were not
required to be included within the scope of license
renewal, and therefore this item was close.
Initially, when we started this review and we
identified these differences, we thought we had
potentially a Part 50 item, because it wasn't part of
the licensing basis. But based on the resolution,
because both the staff and the applicant agreed what
should have been included and what shouldn't have
been, it did not even end up as a Part 50 item.
As for the aging management review, aging
effects, the applicant addressed void swelling in the
reactor vessel, reduction in fracture toughness of the
reactor vessel internal task components by thermal
embrittlement and irradiation embrittlement, cracking
and loss of material of letdown cooler tubings, loss
of material for external Ferritic surfaces due to
boric acid wastage, irradiated-assisted stress growths
and cracking for baffle bolts, and cracking of reactor
vessel internal non-bolted items as applicable aging
effects.
As for intended functions, the applicant
did include heat transfer as an applicable intended
functions for heat exchanges. These things were
already included in the aging management program in
the initial license renewal application as lessons
learned from Oconee.
As for the aging management review, they
performed an aging management review on all the
service water piping, including the copper, brass, and
ductile iron, et cetera, all the materials that are
within the scope of the license renewal. But they did
not perform an aging management review of the tendon
gallery in the license renewal application, consistent
with the staff's conclusion on the Oconee application
review. They did not perform an aging management
review of the pressurized spray head, contrary to
Oconee, which did end up performing an aging
management review of the spray head. ANO 1 does not
use it for their accident analysis at all, and
therefore it was not within the scope.
As for aging management, the applicant
used performance monitoring consistent with Generic
Letter 8913 for managing filing in the service water
system. Cracking of Alloy 600 and Alloy 82/182 will
be monitored during the period of extended operation.
And aging of small-bore piping will be managed by
risk-informed methods used to select reactor coolant
system piping welds for inspections. These are all
differences between ANO 1 and Oconee.
DR. BONACA: This is an existing program?
MR. PRATO: Excuse me, sir?
DR. BONACA: Is this small-bore piping
management risk-informed --
MR. YOUNG: Yes. It's a fairly recent --
it was a change. We just, in the last couple years,
switched to the Risk-Informed In-Service Inspection
Program, and that was when we included the small-bore
piping at that point.
MR. PRATO: That's been reviewed and
approved by the staff as well --
MR. YOUNG: Right.
MR. PRATO: -- independently of this
effort.
DR. SHACK: Okay. So you had the small-
bore piping when you did go to the risk-informed
inspection. You included it rather than as part of
the license renewal, it was actually --
MR. YOUNG: Right. Right. Right. We had
already gone to the small-bore piping inspection as a
result of the risk-informed ISI, which was prior to
doing our license renewal review. So we're able to
take credit for that.
DR. BONACA: Why would you do that, I
mean, technically? Some other applicants claim that
they don't need to inspect small-bore piping.
MR. YOUNG: Well, if you haven't gone to
the risk-informed ISI, then you would not include the
small-bore piping under Section 11 requirements. They
currently do not require you to do a volumetric-type
inspection, just a visual inspection. But during the
risk-informed review, and that's very plant-specific,
we did identify some locations of piping welds that
met the criteria for both risk and susceptibility that
we did include them for doing volumetric inspections.
So I don't think very many plants have gone to risk-
informed ISI yet is part of the reason for the issue.
DR. BONACA: But given your findings,
wouldn't that suggest that maybe one-time inspection
for other applicants is not sufficient?
MR. KUO: Well, there's -- as you know, in
the GALL report right now, that we do require one-time
inspection for the small-bore piping, but this issue
is continually under review. And I believe that in
the industry they have also MRP Program that also uses
the risk and they have concluded that something should
be done. And they are about to make recommendations
to code body. So if this materializes later on, the
staff will certainly incorporate lessons learned from
these activities.
DR. BONACA: Thank you.
MR. PRATO: Okay. As for the open items
identified during the first safety evaluation, there
were two for the aging management review. The first
one was a summary of 11 different aging management
programs that needed additional information to be
included in the FSAR supplement. Each one of those 11
items are identified in the attachment on the back and
the additional information that they agreed to put
into the FSAR supplement.
If you get an opportunity to look at the
operating license, we do not have a license condition
for the FSAR supplement. The reason is, is based on
the findings of this Committee, at that point, the
applicant has agreed to incorporate the supplement
into the FSAR prior to the Commission decision. So an
open item license condition wasn't needed at that
point. So that supplement will be part of their FSAR
prior to the Commission making their decision and
issuing the new license.
The other open item, the applicant did not
identify an aging management program for buried,
inaccessible medium-voltage cables exposed to
groundwater that are within the scope of license
renewal and subject to an aging management review.
When we identified this, the applicant looked at their
aging management review and incorporated it. As a
resolution, they offered something a little bit more
unique than Oconee. They offered to do either what
Oconee did, which is to do some sort of a measurement
on the cabling to try and identify if the installation
is breaking down and to monitor the water that these
cables are exposed to. Or they will do a periodic
replacement of those cables.
The reason they chose to take that second
option is because they've had three failures on-site,
and each time they did do Megger testing not too long
before the failure had occurred. And if something is
not developed that would accurately identify
degradation of the installation far enough in advance
so that they could prevent the failure from happening,
they agreed to just go through a periodic replacement
based on plant-specific and industry experience.
Did I explain that accurately, Garry?
MR. YOUNG: Yes. Right now we're
evaluating basically the qualified life of this buried
cable. It's non-EQ, obviously. It's outside of the
EQ Program because it's not in a harsh environment
relative to EQ. And we have had some failures. So
we're looking at now determining whether or not we can
come up with a qualified life based on operating
experience that would warrant just doing a periodic
replacement or do the inspection. As the inspection
results get better, we may choose to use inspections.
Or if they don't get better, we may choose to do
periodic replacement.
DR. BONACA: This issue, too, will have
some generic implications?
MR. KUO: Yes, sir.
DR. BONACA: As to the adequacy of just
simply doing a measurement?
MR. KUO: Yes. We certainly would take
note of that, and we will incorporate any lessons
learned from this later on.
DR. BONACA: The reason why I'm raising
this issue is that we see a number of applications
coming through with different Project Managers. It's
not clear how these lessons learned are shared among
the different project reviews.
MR. KUO: Well, in fact, there is -- we
have an office letter 805 that describes or detailed
all the procedures that we have followed. So we hope
that these kind of lessons learned will be
incorporated into the official reviews rather quickly.
DR. BONACA: Clearly, for us, it would be
more difficult in the next application to accept just
the measurement of the buried cable as a means of
identifying --
MR. KUO: Well, these issues, like a one-
time inspection for small-bore piping and the buried
cables, are all really issues of contention. It's
constantly under review, and we certainly will take a
continuous look at it.
DR. BONACA: And GALL certainly applies
for this will be documented.
MR. KUO: Yes, sir.
DR. BONACA: And that's why we've asked
for frequent updates.
MR. KUO: Yes, sir; we agreed to that.
DR. UHRIG: Would these cables be actually
replaced or would there just be a new cable put in
parallel and the old one left in place?
MR. YOUNG: They'll probably be replaced.
They're in conduit underground, so they would just be
pulled out.
DR. UHRIG: They can be pulled?
MR. YOUNG: Yes.
DR. UHRIG: Okay.
MR. PRATO: During the inspection process,
shortly before we did the aging management review
inspection, they had their third failure. And they
had the cables out on the grounds, and we took a look
at it. We also found out at that time that they tried
to do analysis to find the root cause of the previous
two failures without any success. The root cause
analysis, the laboratory analysis, was unable to
identify the specific mechanism that failed.
DR. UHRIG: Was there moisture in the
pipes when you pulled the cable out? Was there
evidence that there was moisture in there?
MR. YOUNG: There was evidence of
moisture, yes. Yes. Part of the problem we're having
is that the inspection of the cables is not conclusive
as to the reason for the failure. It could have been
a manufacturing defect that was originally in the
jacket or it could have been some sort of aging
mechanism. But by the time they get them to the
laboratory for inspection, they haven't been able to
conclusively identify the root cause.
MR. LEITCH: The testing program you're
referring to is the Megger Program; is that right?
MR. YOUNG: Yes. The industry currently
is evaluating options for testing, but right -- what
we used was a Megger test. But through EPRI and
through some industry efforts, they're looking at some
other options for maybe other ways to test.
MR. LEITCH: Right. And it was shortly
after the Megger, if I understood you correctly, that
these failures occurred?
MR. YOUNG: Yes. Probably within 12
months or so of the previous inspection we had the
failure, the most recent failure.
MR. LEITCH: Thanks.
MR. KUO: Dr. Bonaca, for the record, I
just want to correct what I said earlier. I was
informed by Mr. Paul Shemanski that the issue actually
has been copied in the final version of the GALL.
I'll let him explain it to you.
DR. BONACA: Okay.
MR. SHEMANSKI: Well, basically, we took
the information -- actually, this issue started back
in October of '99, I believe, with the Davis-Besse
event where medium-voltage cables on the service water
systems catastrophically failed due to moisture
intrusion. These were cables that were in four-inch
PVC pipes underneath the turbine building floor and
somehow -- we believe it to be groundwater -- got in.
And over time, that water actually migrated through
these 4160 volt cables into the insulation, resulting
in ultimate dielectric breakdown.
And as such, we took that information and
the information from Arkansas. We have incorporated
that into GALL. It's in there under aging management
program for medium-voltage cables, subject to
significant moisture and voltage. And we do even
recommend several tests that might be considered.
These are actually used by Davis-Besse -- the partial
discharge test and power factor test. We found those
are more sensitive. Megger is too gross a test to
detect insulation degradation. So I think we've
captured the operating experience in GALL -- well, I
don't think we have, so we're comfortable licensees,
future applicants, will be aware of this issue.
DR. BONACA: Thank you.
MR. KUO: And I also would like to mention
that the April 2001 version of the GALL has been
released to the public.
MR. PRATO: Okay. Time-limited aging
analyses fatigue. The applicant considered cumulative
effects of fatigue for the containment liner plate in
penetrations, and the reactor coolant system
environmental assisted-fatigue, consistent with GSI-
190 in the license renewal application initially.
As for fractured toughness, the applicant
considered fractured toughness related to the
acceptability of reactor vessel internals under loss
of coolant and seismic loads in its reactor vessels
internal aging management program, consistent with the
topical report, BAW 2248.
For flaw growth, the applicant considered
flaw growth in accordance with the ASME boiler and
pressure codes, Section 11 of ISI requirements in the
license renewal application, consistent with the
topical report, BAW 2248.
For neutron embrittlement of the reactor
vessel, the applicant performed analysis to evaluate
the impact of neutron embrittlement on reactor vessel
integrity.
DR. BONACA: I have a question regarding
the specimen for the vessel. It wasn't clear to me
reading the application, you have specific specimens
for your vessel, Arkansas One.
MR. YOUNG: Yes. I may need to get with
Mark here. I think the specimens for the Arkansas
vessel I don't believe are in the Arkansas vessel
anymore. I think they're in another --
MR. RINCKEL: That's right. Mark Rinckel,
Framatome. Yes, they are being irradiated in Crystal
River 3 and Davis-Besse Unit 1. And they're part of
the integrated program, which is a MIRVP.
DR. BONACA: Thank you.
MR. PRATO: Pressurized thermal shock.
The applicant performed an analysis to the criteria in
10 CFR 50.64 and Sharpy upper shelf energy analysis to
Appendix K of the ASME code for the end of the period
of extended operation.
Containment pre-stress tendons. Concrete
reactor building tendons pre-stress will be managed
during a period of extended operation using ASME
Section 11, IWL In-Service Inspection Program.
DR. BONACA: Was this an open item?
MR. PRATO: Yes, sir.
DR. BONACA: Yes, it was.
MR. PRATO: Yes.
MR. YOUNG: Yes. The issue here was in
the original application we provided just the
description of the ASME Program, but there was some
additional monitoring that the staff wanted to see the
results or the information regarding. And it was
really, I think, more of a miscommunication. We were
misunderstanding what the question was, and the by the
time we got to the open item, we finally got down to
the details and were able to provide the needed
information.
DR. BONACA: Yes. You needed to develop
curves, if I remember.
MR. YOUNG: Yes. Right.
DR. BONACA: Okay.
MR. PRATO: For reactor building liner
plate fatigue analysis, the applicant demonstrated
that the original fatigue analysis is valid for the
extended period of operation. For the reactor vessel
underclad cracking, fracture mechanic analysis
indicated that the reactor vessel will have adequate
fracture resistance through the period of extended
operation. And for the reactor vessel in-core
instrumentation nozzles, flow-induced vibration on
reactor vessel in-core instrumentation nozzles have
been projected to the end of the period of extended
operation.
DR. UHRIG: Is that a movable system or is
that a fixed system?
MR. YOUNG: Mark?
MR. RINCKEL: Mark Rinckel of Framatome
again. The nozzles that they're referring to are
fixed and attached to the bottom of the head. We
don't have a system like Westinghouse does with the
thimble tube. Our in-cores are actually exposed to
the reactor coolant, and they move within the guide
tube and through the nozzles and up into the fuel
assembly.
DR. UHRIG: It's not like the Crystal
System.
MR. RINCKEL: Crystal River is a B&W
plant. It is, yes, yes.
DR. UHRIG: Is the in-core instrumentation
essentially the same?
MR. RINCKEL: Yes. The in-core
instrumentation is, but there's not a separate thimble
tube or pressure boundary. I mean the in-core itself
is exposed to the reactor coolant, and it's made of
different material. The stainless steel guide tube
goes from the seal table to the bottom nozzle of the
-- and the nozzle is attached to the vessel. And then
it runs from there up through the internals and up
into the fuel assembly.
DR. BONACA: Do you inspect these nozzles
on a periodic basis?
MR. RINCKEL: The nozzles will be
inspected from the outside in accordance with Section
11. It would be a VT-3 -- I believe VT-3 or VT-2
inspection. And then from the internal, it would be
during when they pull the reactor vessel internals
out. So you would look at both from the outside and
the inside.
DR. BONACA: That would be once every --
MR. RINCKEL: That is correct, yes.
DR. BONACA: And I guess they're less
acceptable?
MR. RINCKEL: Yes, they are. In fact,
those things, if you remember from your history, they
were repaired. They initially broke off at Oconee
Unit 1, and then they were beefed up and repaired at
all of our plants.
DR. BONACA: Thank you.
DR. SHACK: The wall thing isn't
explicitly included at a time-limited aging analysis
here; is that correct? It's not treated as a time-
limited aging analysis?
MR. YOUNG: Right. We went back and
evaluated whether or not we had any corrosion
allowances or wall thinning that was based on time-
limited aging analysis, and we did not find any in our
documentation that took credit for that. So those
were not identified as TLAAs for Arkansas.
DR. SHACK: So in your Flow-Assisted
Corrosion Program, you have no measurable thinning in
your feedwater piping?
MR. YOUNG: No. No, no. Okay. That
falls in the category of being an aging effect, so
that is included -- that was identified as an aging
effect when we did the system reviews. And we did
identify the FAC Program as being the program that
manages that. The TLAAs were strictly the analytical
evaluations that were done in the original safety
analysis to determine the safety of the plant. So if
we had had an analysis that showed that we had a
corrosion or an erosion/corrosion allowance that was
valid for 40 years, then we would have evaluated here
to extend it to 60 years.
DR. SHACK: But doesn't the flaw growth
TLAA include flaws that you would find after -- that
weren't considered in your original design and then
you project that life?
MR. YOUNG: Yes. You're right, yes. For
flaws, any time we identify a flaw then we do an
evaluation for the remaining life of the plant. And
those, too, were identified as TLAA. So, you're
right, those get identified after the original design.
DR. SHACK: Why wouldn't wall thinning be
in the same category as the flaw that you find?
MR. YOUNG: We didn't do any analysis to
project that the walls would remain in tact for the
life of the plant. When we did the evaluation for the
FAC Program, we determined that we in fact needed an
aging management program, not an analytical analysis,
to show that it would go the life of the plant,
because in fact it won't.
DR. SHACK: Okay. But you mean you do an
analysis to show that it will go till the next
inspection.
MR. YOUNG: Yes. Right. But those are
not classified as TLAAs because -- right.
MR. PRATO: One of the criteria for TLAAs
is that it's projected to the current operating term.
MR. YOUNG: Right.
MR. PRATO: And that brings us to our open
items. We talked briefly about pre-stress tendons.
There were a number of different graphs that needed to
be developed, and the applicant provided that prior to
the final SE. And the staff found that acceptable.
And the second item was the Boraflex Monitoring
Program. This is kind of interesting in that the
applicant initially provided a program similar to
Oconee.
From the time they submitted their
application to the time that the staff developed a
request for additional information, they took some
additional data on that monitoring program. And they
found that the -- when they plotted that data, they
found that the boraflex would not last through the
current operating term. As a result, it ended up
being a TLAA.
Under Part 50, they're required to
maintain a sub-critical margin, and if they can't
maintain that sub-critical margin, they have to submit
a plan to the staff for their review and approval. So
they felt that it did not belong under Part 50. And
the staff reviewed the definition under Part 54 for
TLAA and concurred, because it is supposed to be for
analysis that are projected to year 40.
Design Engineering Management did not feel
comfortable in that resolution, removing boraflex as
a TLAA. As a result, we spent some time with OGC, and
OGC concurred with DE Management and said it does not
necessarily have to be eliminated just because recent
analysis shows it's not going to make it to the 40
years. So what the staff requested is that the
applicant keep the program in place, the monitoring
program in place until the resolution has been
identified and that the boraflex life and the ability
to maintain sub-critical margin can be established out
through the period of extended operation.
DR. BONACA: You do have boraflex only in
one region of your pool.
MR. YOUNG: Yes, that's right.
DR. BONACA: And in the other regions, you
have Boral or some other material?
MR. YOUNG: I'm not totally up to speed on
the details of our spent fuel pool, but we --
DR. BONACA: But there's no boraflex.
MR. YOUNG: Right. We do have some
regions that have the boraflex and some that do not.
DR. BONACA: Do you already have a plan on
how you're going to get rid of the boraflex?
MR. YOUNG: We're developing that plan
right now. As Bob mentioned, the finding was fairly
recent, and there are several options to correct the
situation, and those are being evaluated. And
probably within the next two years, we're going to
wind up with a recommendation to take some action with
either a different material or --
DR. BONACA: So you still have flexibility
in your pool to move those assemblies in some
different location as you --
MR. YOUNG: Yes. We still have some room
in the pool for moving the fuel around, yes.
DR. BONACA: Okay. All right.
MR. PRATO: The next slide, slide 20, is
just a list of the aging management programs. If
anybody has any particular question on any of the
aging management programs, I'll be glad to answer them
at this point.
DR. BONACA: I would like to go back a
moment to the CRDM casings. And the question I have
is -- I know that Oconee has already committed to
repairs. Essentially, the repairs include re-welding.
The question I have is, is the material being used for
welding over? And I'm sure that Arkansas has some
plan of that nature too. Is it going to be less
susceptible to the same kind of failures? I guess
what I'm driving at is are these steps that are being
taken now to repair those cracks going to be -- are
they being viewed as a permanent repair that should
not be affected anymore by this phenomenon or is it
going to be simply another time-limited repair?
MR. YOUNG: Well, I think the answer to
that is it's still being evaluated. And I know the
repairs that were done at Arkansas were different than
the repairs that were done at Oconee, but I think it's
part of this evolving process and analysis of what is
the correct solution, where do we need to go from
here. I think in the case of Arkansas, the repairs
were done with the information that was available at
the time, which was just within the last couple of
months. And they're continuing to do the analysis on
the findings to determine if -- well, first of all, it
will change our inspection program.
DR. BONACA: Sure.
MR. YOUNG: So that's definitely a change.
And then it may require some subsequent repair actions
or preventive actions based on the results of those
analysis. But that's still being evaluated.
DR. BONACA: I guess what I'm driving at
is that ultimately the measure of success of the
program is going to be the ability of preventing an
occurrence to happen again. And so right now really
we don't know if these kind of repairs are going to be
effective to do that. I mean we don't, I guess.
MR. YOUNG: That's my understanding. Now,
there may be some people here from the staff or from
Framatome that know more about the details of the work
that's been done so far. But I know at Arkansas we're
fairly early into the analysis. And like I said, we
just finished the outage in which we found the
problem, so I know there's still a lot of work going
on in that area.
DR. BONACA: I understand that some of the
materials are being changed, so there is some
expectation that those changes in materials should
lead to a different kind of performance, although we
cannot right now estimate whether or not they will
prevent these kind of failures from occurring again.
And so you have to rely on future inspections.
MR. YOUNG: Right. And I think, as
mentioned earlier, the Materials Reliability Program,
the industry program, is looking at this as well to
see what changes are needed throughout the industry
relative to this type of problem.
MR. PRATO: Before I go into the
conclusion, are there any other questions?
MR. LEITCH: Earlier there was an
indication that there were 22 existing programs, and
here there are 28 listed. Is that just a different
bean count or is there some significance to the
difference in those numbers?
MR. YOUNG: Again, the way we count the
programs is somewhat difficult, because it's a bean
count issue. We all have the same list of programs,
but in the application itself we would have a section,
and then it would have an A, B, C part. So it depends
on whether you count the A, B, C part or just the
headings.
MR. LEITCH: Okay.
MR. YOUNG: That's really where we're at
on that.
MR. LEITCH: Thanks. Just one other
minor, very minor, comment. In the SER Chapter 5,
there's a section about presentation to the ACRS, and
the number of the ACRS meeting at which those
presentations occurred is incorrect.
MR. PRATO: I'll verify that.
MR. LEITCH: It's just a nit.
DR. BONACA: That's a good point. I mean
this is the first application for which we have not
had an interim full Committee meeting. And, of
course, as I mentioned before, there are good reasons
for that. One was the low number of open issues
identified, and we agree with the staff that there
were no additional ones.
Second, the fact that there was a lot of
lessons learned, and we actually asked the staff to
articulate the presentation on the basis of comparison
to the previous ones so that we could understand
whatever we accepted the program for Oconee, then the
program should be acceptable for Arkansas, unless
Arkansas presents a better program, which in some
cases did.
And the reliance on the standard
application format, actually striving for it. The
work that Arkansas did with the NRC I think was very
helpful, and the reliance on the guidance of NEI 95-10
made the application, I think, much easier.
And I point it out because we have been
trying to have some demonstrations from repeated
applications that in fact ultimately the guidance
documents and the endurance of the guidance documents
and previous experience will facilitate the review and
improve the applications. And we, I think, have proof
here in front of us.
MR. PRATO: Any other questions?
Okay. In conclusion, on the basis of the
staff's review of the license renewal application and
the applicant's response to the request for additional
information and resolution to the open items, as
documented in the safety evaluation report, the staff
found that, one, the applicant has appropriately
identified the aging mechanisms associated with
passive, long-lived structures and components, as
required under 10 CFR 54 and 10 CFR 54.21(a).
Two, the applicant has instituted the
programs needed to manage age-related degradation of
these structures and components such that there is
reasonable assurance that ANO 1 can be operated in
accordance with its current licensing basis for the
period of the extended license without undue risk to
the health and safety of the public.
And three, the applicant has analyzed the
time-limited aging analysis associated with ANO 1,
consistent with the requirements of 10 CFR 54.21(c).
On the basis of these findings, Region 4's
verification of these activities, and the Regional
Administrator's recommendation, the staff requests
that the ACRS provide the Commission with a favorable
recommendation on the renewing of the ANO 1 operating
license for an additional 20 years of operation. And
that concludes our presentation for today.
DR. BONACA: Okay. Any questions from the
members? Any perspectives you want to share regarding
the application and the SER? If none, I would like to
thank the staff, Mr. Prato and Mr. Young, for well-
informed presentations. And I would like to also,
again, recognize Arkansas for an application that
facilitated that review. And I think it's been quite
effective. And with that, I thank you very much, and
I --
MR. KUO: And this concludes the staff's
conclusion. And what I would take back, I think,
there are three points here that we're going to check
SER Section 5 and correct, if possible, the
discrepancy in the numbers of the ACRS meetings. And
the second one is we will monitor the progress of
aging management for both the CRDM cracking issue and
the small-bore piping issue.
DR. BONACA: Small-bore piping, yes.
MR. KUO: And with that, of course, we
will recommend that ACRS write a letter to the
Commission for approval of the --
DR. BONACA: We will write a letter.
MR. KUO: Thank you.
DR. BONACA: Okay. Thank you very much.
And with that, Mr. Chairman --
CHAIRMAN APOSTOLAKIS: Thank you very
much. We were told that the review of the application
was completed eight months ahead of schedule?
MR. PRATO: Yes, sir.
CHAIRMAN APOSTOLAKIS: And Dr. Bonaca
completed his presentation half an hour, actually --
half an hour before schedule. There must be something
going on with license renewal issues.
(Laughter.)
We probably overestimated what it takes to
review those. Thank you very much, gentlemen;
appreciate it.
MR. YOUNG: Thank you.
CHAIRMAN APOSTOLAKIS: As the members
know, we will meet again at 10:30 in the
Commissioners' Room to attend to the Commission's
meeting on nuclear research with Dr. Powers and Dr.
Wallis leading the charge on behalf of the Committee.
Thank you very much, and we'll see you
back here at 1:30.
(Whereupon, the foregoing matter went off
the record at 9:55 a.m. and went back on
the record at 1:30 p.m.)
. A-F-T-E-R-N-O-O-N S-E-S-S-I-O-N
(1:30 p.m.)
CHAIRMAN APOSTOLAKIS: We're back in
session. The next item on the agenda is a
presentation on risk-based performance indicators.
Mr. Mays, the floor is yours.
MR. MAYS: Thank you, George. Good
afternoon. It's a pleasure to be back here before the
ACRS to discuss our work on risk-based performance
indicators. This presentation will be an abbreviated
version of what we presented to the Subcommittee last
month. The Subcommittee asked us to concentrate the
proposed shutdown, performance indicators, the
validation and verification, including comparison with
the current reactor oversight process PIs, and the new
alternative approaches for risk-based performance
indicators that we've developed in response to
internal and external stakeholder comments.
So as we did at the last meeting, our
counterparts from NRR are here to briefly explain the
relationship between the RBPIs and the reactor
oversight process. And the rest of the presentation
will be our summary of the work that we did to
establish the technical feasibility of risked-based
performance indicators as a potential enhancement to
the ROP.
We're seeking a letter from the ACRS
addressing whether you see this work as potential
benefit to the reactor oversight process, whether you
think our technical approach is feasible, and whether
you think we should continue to expand and/or add the
proposed alternative approaches to the Phase 1 report.
We issued the Phase 1 report in January. You have had
it for a few months now, and we're going to look
forward to see what comments you have from that.
Now, Tom Boyce, from NRR, who works in the
Inspection Program Branch, is here to go over the NRR
view of the interrelationship between the RBPIs and
the reactor oversight process.
MR. BOYCE: Thank you, Steve. As stated,
I'm Tom Boyce. I'm the Inspection Program Branch of
NRR. You heard about the Reactor Oversight Program
yesterday. I'm a member of the Branch who is
responsible for that oversight process, and we would
be the people who would be the users of the risked-
based PIs.
I wanted to start just by talking about
some of the environment surrounding the risk-based PIs
and the direction we're going. In the Commission PRA
policy statement and in their strategic plan, the
Commission articulated its intent to move in a more
risk-informed direction, and we think these risk-based
PIs are clearly a step in that direction. We also
wanted to point out that the current reactor oversight
process is a significant step in that direction. We
think it's much more risk-informed, objective,
understandable, and predictable than the previous
oversight process that was in place.
We also wanted to point out that industry
and the NRC have been responsive to larger movements,
advances in information technology, and the collection
of data is improving, the transmission of data is
improving through the use of the Internet and personal
computers. And the PRA models, specifically the SPAR
models under development by the NRC and the PRA models
that licensees are using, have continued to improve.
And so against that backdrop, it's more ripe for risk-
based PIs than we've had at any time in the past.
Next slide.
DR. POWERS: May I ask a question? The
industry, when it does risk assessments, it gets a
certification from an industry group for the PRA that
it uses.
MR. BOYCE: The question is do they get a
certification?
DR. POWERS: Well, I believe they do.
MR. BOYCE: Okay.
DR. POWERS: And what I'm asking is what
is the equivalent for the SPAR models?
MR. MAYS: Let me answer that, Dana. The
SPAR models, the Ref 3 models that we're using for
this program, we have instituted a process by which
they get reviewed by the contractor and by us as
they're being done. They're reviewed internally by
NRC personnel when we get them. And we also have a
process for doing on-site reviews where we go to the
plants and look at the as-built, as-designed plant and
what they've done in their PRAs to identify if there
are any shortcomings that we've had in that.
In addition, we've been using the SPAR
models for several years now in the Accident Sequence
Precursor Program. So whenever we evaluate the risk
significance of an event or condition at a plant using
those models, we would send those out formally to the
licensees to review, and we were getting feedback and
comment on those during that process as well.
DR. POWERS: So you really don't have what
I would call an independent review. And I'd invite
Dr. Wallis to comment on his experience with people
saying, "Gee, we've used a code for several years, so
it must be right."
MR. MAYS: Well, that wasn't the statement
I made, but I was saying we have had the opportunity
to get feedback from licensees about the validity of
models we've used through the ASP Program. That's not
a complete review, but it is more than nothing. And
we are going to every site for the models that we're
developing to have those reviewed by the folks on-
site. So I think we have a pretty substantial process
for being able to do that.
CHAIRMAN APOSTOLAKIS: Wouldn't it be a
good idea, after the Agency agrees to some form of an
ASME standard, to apply the standard to SPAR?
MR. MAYS: If we were to come up with a
standard, I think that would be appropriate.
DR. WALLIS: I hate to use the word
"independent" that was used this morning, but wouldn't
it useful to have also some independent check on these
things from -- I don't know where it would come from,
but just between you and the licensees, I'm suggesting
someone else might who was not so tied up in the
process be able to contribute to improving or
detecting --
CHAIRMAN APOSTOLAKIS: At this point,
having the licensees review them may be good enough.
DR. POWERS: How can you possibly say
that? I don't think that would inspire public
confidence if I were a member of the public.
CHAIRMAN APOSTOLAKIS: Still, the program
is under development. The numbers are not being used
in any real way by the Agency, right? And they have
30 SPAR models. They plan to have, what, 70 more or
40 more? I mean it's really important at this point
to make sure that their SPAR model for a particular
plant is not off the mark. So by having the licensee
review it, you get that assurance. But they are not
really taking any regulatory actions yet, because
eventually when they have the totality of the 70 SPAR
models, then they will have to think about how to have
maybe an independent review panel or somebody.
DR. POWERS: Well, I can see this now.
You're going to bring a review panel and say, "Here,
review 70 models."
CHAIRMAN APOSTOLAKIS: So you plan to have
established a review panel that will be reviewing them
as they are produced?
DR. POWERS: I don't know. That sounds
like a good idea to me.
CHAIRMAN APOSTOLAKIS: And the other point
is that let's not overexaggerate the value of these
review panels. I mean I can't imagine that they will
do a review like Sandia did for Indian Point and Zion
PRAs. I mean these panels will probably look at the
overall approach, how did you do common cause
failures, how did you do data analysis. But otherwise
it's a huge job. I mean it's huge now. It's huge in
the future.
DR. POWERS: And it seems to me that the
pattern for the review has been set by the Agency in
the kind of reviews that is applied to the codes
that's developed for severe accident analysis.
CHAIRMAN APOSTOLAKIS: Can you elaborate
on that?
DR. POWERS: They do a fairly detailed
review. The panel actually exercised the models.
They have a process set out, begins with are the
intentions of the model. They do a top-down, then a
bottom-up examination. They publish a report that has
their complaints about the -- their comments on the
models and includes the response from the model
developers.
CHAIRMAN APOSTOLAKIS: But here you're not
talking about a model that will be used to produce
results under different conditions. Here you are
talking about PRA for a particular unit. So, you
know, that approach will have to adapted to this
particular problem.
DR. POWERS: Well, I think it's adapted to
every curve, but it seems like a pretty good approach
to me.
DR. SHACK: And it will be independently
checked by the PRA that the licensee has. I mean,
obviously, I think if the licensee is getting
different results, then you're certainly going to hear
about it.
DR. POWERS: What I would worry about is
if you've overlooked some vulnerability in the SPAR
model that the licensee has overlooked, and the item
that comes promptly to mind is induced station
blackout.
MR. MAYS: I guess the question I would
have is who would have that level of knowledge outside
of us or the plants to be able to conduct that kind of
an in-depth review?
DR. POWERS: Another plant.
DR. WALLIS: Maybe that's right. Someone
who's done it himself knows the ins and outs, knows
the traps --
CHAIRMAN APOSTOLAKIS: Do you think that
will contribute to public confidence to tell them that
San Onofre was reviewed by Diablo Canyon?
DR. POWERS: Well, I think if I was to
formulate a panel, I would probably draw from a cross
section of the community; that is, I would look to
somebody with experience from the nuclear industry
with a similar type plant, someone from academia,
maybe even sophomore at Dartmouth. Well, they seem to
be very knowledgeable individuals. And maybe somebody
from the PRA specialist community. Budd Boyack's not
a bad choice.
CHAIRMAN APOSTOLAKIS: I don't think that
when you say that you appreciate the magnitude of this
effort. I'm not against a review, but just to say,
"Have these models reviewed," I mean we can start by
reviewing SAPHIRE, for heaven's sake, and apply what
you said earlier about the severe accident goes to
SAPHIRE, which is the basis for the models. Let's do
that first and then we can have a panel and so on.
And then go to the individual SPAR models and make
sure that we have a practical approach. That's all
I'm saying. I mean just to ask, "Have you reviewed
your 30 SPAR models," it seems to me is a little bit
too much.
MR. BARANOWSKY: I'm Pat Baranowsky, Chief
of the Operating Experience Risk Analysis Branch, and
I'd be glad to meet with the Subcommittee or the full
Committee regarding SPAR models and whether there's
adequate review or not. But I would like to point out
that, as George Apostolakis said, these models have
been evolving over a number of years, and there's a
difference between SAPHIRE, which is the tool, if you
will, and the model, which is the logic that reflects
the way the plant's built. And as Steve said, the
logic has been modified and is currently being looked
at closely on each one of these models.
The assumptions that go into the models,
for instance, how do model this sequence or that
sequence and are they complete, are primarily based on
the insights that we've derived from the IPEs and PRAs
that are in existence and the accident sequence
analysis work that we've done over the last 20 years.
They're not meant to be models that uncover new
accident sequences that nobody ever heard of before
due to unique design or operational characteristics at
a plant that aren't manifested in operating
experience. That's supposed to be the purview of the
licensee and other types of design and operational
reviews.
So they have a different purpose, and that
is to say if there's a new sequence and a contributor
that is unknown, I don't know that we would use the
SPAR approach to try and find that kind of thing. It
reflects what we understand today, our best
understanding about what should be in risk models and
a simplified version of them.
CHAIRMAN APOSTOLAKIS: And even more
significant question, I think, is the level of detail
that goes into the SPAR models. And I think the staff
is still working on that in some sense anyway. You
started with very simple models. Then you went to the
next level. And I think as you use them for your
purposes here, you will probably realize that we may
do a little more here, a little more there. That's
why I'm kind of reluctant to jump into expert panels
and all that at this point. Although for SAPHIRE, I
really think we should have a review, because it's a
model, it's a tool, it's been out there for years now.
It's the official PRA tool of the Agency. I mean we
should have a serious peer review, and I think it can
be done for a tool but for 30 SPAR models --
DR. SHACK: Well, I mean the question is
if you're going to spend that kind of money, is this
the way that you would spend it? I mean there are
lots of things to spend money on.
CHAIRMAN APOSTOLAKIS: Exactly. Exactly.
MR. MAYS: Well, I know I kept control of
the meeting at that point, so --
(Laughter.)
DR. WALLIS: Well, let me suggest, George,
though, that I mean I think Dana's raised an important
issue. It may not be the envisioned expert panel is
the solution, but something to sort of ensure that the
integrity and completeness of these things would be
good. And I don't know what should be done, but --
CHAIRMAN APOSTOLAKIS: I did not object to
the essence of their argument. I just thought that it
was a little bit too soon to do that for the
individual SPAR models. Let's do it for the tool
first and then after you guys say, "Now we have the 70
models and this is what we're using them for," then it
seems to me some sort of a review, not necessarily --
DR. POWERS: It seems to me you're begging
to get into the situation of where we come back and
say, "Well, these models really aren't what you really
want, but since you've already built 70 of them, we
might as well let you go ahead and do this."
MR. MAYS: I think it's a little more --
we may not have communicated as well, either through
this document or through other briefings to you, the
depth of what's going on with the SPAR model
development and what's been happening over time. We
started out very early on in the Accident Sequence
Precursor Program with just simple event trees, no-
fault trees, fault probability numbers as an estimate
of risk significance of events. We moved to models
that had more detail in them in terms of the event
trees that were more up to date with our current
understanding of success criteria, as PRA evolved
through 1150 and other things. And we have
subsequently expanded that SPAR models and the Rev 3
down through fault trees to include support states, to
include uncertainties explicitly in the analysis.
So we've made -- we had an outside panel
in 1992, I believe, come in from all over the place,
and George was a member of that working group in
Annapolis where we said from people from industry and
from academia and from the Agency, "What kind of
models do we need? What characteristics do they have
to have?" And so this SPAR model development has gone
along that kind of a development path from the
beginning.
And we also have internally to the Agency
a SPAR models users group, which are the people who
have to use risk understanding in doing their
regulatory business, who are our users and our
customers who say, "These are the features we need.
These are the characteristics it has to have. We've
set out a standard in that group for how should we go
about reviewing these models." So I think we may have
a more substantial review process than is patently
clear from this information.
And we agree that the models have to have
a reasonable reflection of the risk characteristics of
the plants for the purpose of what we're using here.
Our external reviewers, including the industry, has
told us they want to get the SPAR models and have a
review of them, and we agree with that. And so I
think we're on the same wavelength with respect to
what needs to be done, and that is we should have SPAR
models that are a reasonable representation of the
plant. How specifically we go about doing them, I
would propose we save for another day.
CHAIRMAN APOSTOLAKIS: At least until the
ASME standard is approved. In fact, I hope that your
guys on that joint committee that's developing the
standard know that the Agency's models group is
subjected to that standard. That's always a good
check. So can we continue?
MR. BOYCE: All right. I'm on page 4.
The first bullet there talks about two Commission
papers that NRR wrote that laid out the basis for our
Revised Reactor Oversight Program in early 1999. And
as you heard yesterday, we used both performance
indicators and inspection findings to take regulatory
-- to have regulatory engagement with our licensees.
We ran a pilot program for six months in 1999, and we
reported to the Commission the results of that pilot
program in SECY-00-049.
And in the SECY paper, we said that while
the future success of the Oversight Program was not
predicated on the risk-based PI Program, that we
thought that risk-based PIs would potentially support
a couple of areas. And we said there are certain
enhancements to our current oversight process where we
thought risk-based PIs would help. Those are actually
articulated in the last bullet. They're the
reliability indicators, unavailability, shutdown and
fire and containment indicators. And we also thought
that plant-specific performance indicators would be
useful in the future.
In order to make this happen, NRR wrote a
user need letter to --
CHAIRMAN APOSTOLAKIS: Let me stop you
right there, because that's something that has been
bothering me for a long time. If you read -- I don't
know how anyone who reads Appendix F of the report of
the staff issue in January can say that we don't need
plant-specific performance indicators. And in fact
the evidence there is so compelling that it seems to
me that the current reactor oversight process, the
revised process, risk-informed, has to immediately
start looking again at the thresholds.
All I have to do is look at the tables
that these ladies and gentlemen prepared, and I see
things that if I use the industry variability curve as
it is being used now, according to Appendix H of the
revised oversight process, to get into the red for
transient initiators, if I observe for three years,
collect data for three years, I will need 646
transients. For loss of feedwater, to get into the
red, I will need 355. To get into yellow, I will need
36. I don't know which utility or whether this Agency
would tolerate 36 losses of feedwater in three years
before it said, "Oh, now you're in yellow. We have to
do something about it."
It's clear to me, and the mathematics
shows it, that the thresholds we have now are no good.
They're too generic. If I were running the reactor
oversight process as it is now and I looked at this,
I would make it my number priority to revisit the
thresholds. Now, tell me why I'm wrong.
MR. BOYCE: Well, it's important to
remember that the purpose of the performance
indicators is to help us establish the right threshold
for regulatory engagement. I mean they're not
definitive unto themselves. Just because you have a
performance indicator does not mean that you should
immediately shut down the plant. It means you should
do further investigation to look at the causes.
DR. POWERS: It seems to me that he's
asking the opposite question. Would you really
tolerate 36 losses of feedwater in three years and not
engage the licensee?
MR. BOYCE: Well, I mean, actually, I was
trying to be supportive of the risk-based PI effort.
It sounds like you're suggesting that risk-based PIs
did not give you the correct indication.
CHAIRMAN APOSTOLAKIS: No. I was going
the other way. The risk-based PIs are giving you an
indication -- I don't know now; we have to review it
more and so on -- but they are raising a flag that the
thresholds you are using now are way off the mark,
because they're generic. And for losses of heat sink,
just as another example, to go to white, which is the
very first level of alert, right, I will have to have
19.5 losses in three years.
DR. WALLIS: Which heat sink is that?
CHAIRMAN APOSTOLAKIS: What?
DR. WALLIS: Which heat sink is that?
CHAIRMAN APOSTOLAKIS: The ultimate heat
sink.
PARTICIPANT: Condenser heat sink.
MR. MAYS: Let me help a little bit with
that, George. One of the realities of looking at this
from a risk perspective is that there are certain
elements, whether they be initiating events or whether
they be reliability, availability of particular
equipment, that have relatively lower risk importance,
and therefore in order to get to the pre-determined
thresholds that we have, you have to have a lot of
events.
And the other thing that's important to
recognize is that all of those thresholds in the
current ROP, as well as the thresholds that were in
the initial draft Phase I report that you had from us,
were based on having one variable out of all the
variables in the risk analysis change enough to get to
that threshold, while everything else at the plant
remained at its baseline performance. And what you
see is that for some elements the relative importance
of that particular element is such that if everything
else stays at baseline, you really have to change that
a lot to equate to that level of performance. That
tells you something about relative risk.
It also tells you that since risk is
really a multivariate function that you have a
possibility of sometimes having thresholds that seem
counterintuitive, because when you see the threshold
the thought that, "Oh, and everything else had to stay
at the same value in order to reach the threshold,
which was the basis for that calculation," isn't
really obvious to people.
And I think it's pretty clear that I would
expect that before you got to 16 losses of heat sink
in three years or 15 or ten, that the kinds of
conditions that would be necessary to make that happen
would also manifest themselves in other areas of the
plant performance. And if we have a process that
samples those other areas, you're going to see
multiple areas starting to degrade, and that's what
will get our attention rather than relying on the fact
that loss of heat sink is the only thing that's going
to change at this plant.
DR. POWERS: So what you're saying is
we've defined the parameter incorrectly.
MR. MAYS: I'm saying the current reactor
oversight process and the initial pieces that were in
the risked-based performance indicator report were
based on a concept which was we'll have a broad sample
of performance, we'll see how bad each one of those
individual pieces would have to get if everything else
was nominal. That was the basic philosophy. And an
implication of that philosophy is that the threshold
set by that might seem counterintuitive because in
real life there's more likelihood that you will see
multiple things go wrong in that case than just one go
really severely wrong.
CHAIRMAN APOSTOLAKIS: No, but -- no, no,
no. I think Dana touched on the real issue here. I
think either what Dana said is right, we defined them
wrong, or the criteria that were used to derive these
numbers and in the reactor oversight process were not
the same, and in fact they were not, because you are
using CDF changes, whereas they are using the generic
plant-to-plant variability curve for each event.
MR. MAYS: Only for the green-to-white
interface.
MR. BOYCE: For the white-to-yellow we
used limited SPAR models.
CHAIRMAN APOSTOLAKIS: I know, I know, but
for the green-to-white there was a difference.
MR. MAYS: Correct.
CHAIRMAN APOSTOLAKIS: And this other
thing that you mentioned, I don't know. I mean we are
looking at individual indicators. I don't remember
anybody making a presentation here that we are looking
at the combination.
DR. KRESS: That's what the integrated
performance indicator is supposed to do, isn't it?
MR. BOYCE: Right. That was in --
CHAIRMAN APOSTOLAKIS: Yes, but the
indicators themselves were developed on an individual
basis.
DR. KRESS: Yes. But they're going to
integrate.
CHAIRMAN APOSTOLAKIS: Either these
numbers make sense or they don't. We can't produce
different results under different studies and then
say, "Well, but the other results were okay too." It
seems to me that you make a very good case in Appendix
F that these things have to be plant-specific. You
say that clearly when it comes to unavailability. The
observability over diesel generators on reliability
varied greatly across the industry, from 2.5 tenths to
the minus four for BWR Plant 3 to 2.9 tenths to the
minus two.
Similarly, for RCIC unavailability, and
there you say, "Weak examination of data for other
systems revealed similar variation among units.
Therefore, we decided that only site-specific data
were appropriate for estimating the variability of
outage data at the plant." Now, if I read this and I
was running the reactor oversight process, wouldn't I
worry? Wouldn't I say, "Am I doing the right thing?"
MR. BOYCE: Yes.
MR. MAYS: George, let me help a little
bit on that too.
MR. BOYCE: Yes, I agree with you. In
fact, that's what we said. We thought that plant-
specific PIs were the way to go. I mean we said that.
CHAIRMAN APOSTOLAKIS: And what I'm saying
is there is a higher degree of urgency to this than
just saying, "We'll wait until Mays is done and then
take the results."
MR. BOYCE: There are other problems.
CHAIRMAN APOSTOLAKIS: Because this tells
me that -- well, this gentleman has been trying to
talk for a while now.
MR. HOUGHTON: I'm sorry. Tom Houghton,
NEI. Good afternoon. I thought I heard you say --
and perhaps I was wrong -- I thought I heard you say
that the current program has a very high number of
loss of heat removal scrams for the green/white
indicator. The indicator is two. You can have two in
three years is all you can have for the green/white.
CHAIRMAN APOSTOLAKIS: I understand.
MR. HOUGHTON: It's not a higher number
than that.
CHAIRMAN APOSTOLAKIS: No, no, no. The
numbers are quoted from here. I didn't mean that.
MR. HOUGHTON: Okay.
CHAIRMAN APOSTOLAKIS: But what I'm saying
-- let me emphasize what I'm saying. I'm not prepared
to claim that the numbers we're using now are no good.
No, actually I am.
(Laughter.)
But the numbers we're using now -- no, no.
I think I should rephrase this. I'm not prepared to
say that. What I'm saying is that there is sufficient
evidence from the analysis that is presented in
Appendix F of this report to convince me that we
really need plant-specific indicators, plant-specific
thresholds, and that we should make a much more
careful study, do much more careful study of the
observation time and the actual thresholds, of course,
using methods similar to Appendix F to make sure that
we have covered these uncertainties, which are
aleatory and epistemic because now you are really
dealing with the real world, and increased public
confidence or at least my confidence that what we're
doing is really rational.
So I guess the reason I -- I guess -- I
don't guess. The reason why I'm raising this is
because I think it's of a certain urgency for the
existing revised reactor oversight process. It's not
something that can wait until you guys are done. You
guys means research.
MR. MAYS: There are two issues that kind
of got woven up here together. One of them had to do
with the fact that you can have some fairly high
numbers for certain -- to get to certain thresholds,
notably yellow and red, that seem to be
counterintuitive because the idea is if you had
anywhere near that number of events, something else
would have -- we would have been doing them. And I
agree that that's a separate thing, and it has to do
with the nature of having single variate analysis in
a multivariate picture and the relative risk
importance.
The second point you made was about the
plant-specific nature. Now, one of the things we had
in the discussion here on verification and validation
is we went back and looked at how the plant-specific
data and information we had would compare with the
similar kinds of indicators and information in the
current reactor oversight process. So to give you a
little more -- maybe a little feeling of a little more
ease, we didn't find substantial differences in the
overall assessment of things between the risked-based
performance indicators and the Reactor Oversight
Program.
CHAIRMAN APOSTOLAKIS: And why do you say
that?
MR. MAYS: There will be -- in the
verification and validation section we talk about
that.
CHAIRMAN APOSTOLAKIS: Chapter 5 of the
main report.
MR. MAYS: What we did find was that there
were differences. Sometimes they were -- the risked-
based performance indicators indicated that
performance was worse than indicated in a similar
version of the reactor oversight process, and
sometimes they indicated they were better. And when
we get to the section where we discuss the alternate
ways of looking at RBPIs, in light of the comment of
how many we had, you'll see that the more integrated
approach that we took in this alternative section
helps to address both of those issues.
CHAIRMAN APOSTOLAKIS: You know what the
integrated approach is? Take the PRA of the plant,
look at the initiating event number they have, look at
the unavailability of the system, because they have
already done it. And say, "For this number I don't
want this deviation, I don't want that deviation," and
then you have the integrated view. You don't have to
do anything; the PRA had done it for you.
DR. KRESS: Your threshold would be delta
CDF --
CHAIRMAN APOSTOLAKIS: Exactly.
MR. MAYS: That's exactly what we --
CHAIRMAN APOSTOLAKIS: And also it will be
--
MR. MAYS: That's exactly what we did in
the alternate approach here, George, is we used the
entire model, and depending on whether we were looking
at the cornerstone level or whether we were looking at
a functional level on systems or a response to --
CHAIRMAN APOSTOLAKIS: That's not what you
do in Appendix F.
MR. MAYS: That's not in Appendix F.
That's the alternative stuff that we presented at the
Subcommittee last month. The stuff we're going to
present --
CHAIRMAN APOSTOLAKIS: The ultimate result
of all this is take the PRA, which comes back to my
favorite subject of objectives. See, as you read the
-- I'll ask you questions. I'm just talking to NRR
because they are here, but your turn will come.
But the objectives, the objectives are
extremely important, because you're playing there with
prior distributions. Appendix F was written by a
statistician, I think. He says, "Well, this number
doesn't make sense, so I'll use another prior."
You'll use another prior because the numbers don't
make sense? Perhaps you should be shot first.
MR. MAYS: Well, actually, that's not what
we did, but that's --
CHAIRMAN APOSTOLAKIS: That's what it
says. I can only go by what it says.
MR. MAYS: Well, actually, that's a
different characterization than I would put on it.
CHAIRMAN APOSTOLAKIS: My point is plant-
specific PRA, plant-specific thresholds make much more
sense than anything else, and it's your work to date
-- I appreciate your valiant efforts to defend your
colleagues -- but your work to date makes that urgent,
in my view.
DR. KRESS: And why have thresholds on
individual performance indicators?
CHAIRMAN APOSTOLAKIS: Well, they went to
trains, which is very good. We'll come to that if we
ever come to that. I mean they did some good stuff
there.
DR. KRESS: That would help, but why not
integrate it all at once?
CHAIRMAN APOSTOLAKIS: At some point.
DR. KRESS: It says suppose you're using
the PRA and plant-specific.
CHAIRMAN APOSTOLAKIS: Well, there are two
competing --
DR. KRESS: Call Bob Christie and say,
"Let's say the performance indicator on delta CDF."
PARTICIPANT: Is Christie here?
CHAIRMAN APOSTOLAKIS: No, no, no, no, no.
There are two competing --
MR. MAYS: He was, but he got scared and
left.
(Laughter.)
CHAIRMAN APOSTOLAKIS: -- elements here:
One is to be as high as you can, as you say, to go the
Christie way, and the other counter argument is that
you want something you can observe. So you have to go
-- that pulls you down, the other thing pulls you up,
and you have to --
DR. KRESS: No, no. But you're observing
the things that go into the PRA to make the delta CDF
calculation.
CHAIRMAN APOSTOLAKIS: Yes. And that's
what these guys are doing. And then they come back
and they tell you --
DR. KRESS: Yes, but don't put the
threshold on those, because they're determined, just
like he said, as if all of them say the same except
that one. Just look at all of them and integrate the
total change and see the effect on delta CDF and put
a threshold there, rather than have individual colors
for each PI.
CHAIRMAN APOSTOLAKIS: In an ideal world,
that's the way it should be done. You are asking the
Agency to take a gigantic step away from micromanaging
all the way out, and they will never do that. So
let's hope that they will go to the trains that these
guys are offering now, and then maybe later --
because, remember, we're going to discuss option two
a little later.
MR. BOYCE: I'm on page 5 now.
(Laughter.)
Actually, I mean, we're challenged as to
why we just don't do it immediately. And that giant
step forward is, I mean, really what we're facing
here. And we think that there's certain key
implementation issues that need to be looked at before
we go and take that giant leap forward or if we take
that giant leap forward.
And the ones that we've already discussed,
data quality and availability, SPAR model development,
and V&V. The V&V that I'm referring to was -- it's
not enough that we developed the SPAR models, we need
some way to gain what we were looking at was
acceptance by the licensees and the public, that the
SPAR models were going to give you a reasonable
answer. And we weren't saying a perfect answer, that
we modeled all possible events and all possible
scenarios; we were just saying a reasonable answer
with which we could regulate. So I think we had
identified these issues. They're in Section 5 of the
Phase I report. And I won't go into more of that.
I did want to make one more comment on
data quality and availability. The reliability data
is coming from a database that is called EPIX. It's
run -- that database, I think, is collected by INPO,
and it's the successor to NPRDS. And it was in
response an AEOD initiative for a reliability data
rulemaking, and industry said they would stand up EPIX
and populate it in lieu of that data rule. And that
was about 1997 time frame.
And industry has in fact followed through
on that effort, but it's still a voluntary initiative.
We don't have a requirement. There's no rulemaking
that says anybody needs to submit data. Even the
current reactor oversight process is still voluntary
submission of data. And we haven't taken a close look
at the EPIX database to say that there is 100 percent
participation in submission of data. We haven't said
that there is consistency in terms of submission of
that data. And we haven't done verification of that
data.
CHAIRMAN APOSTOLAKIS: Where would you get
your data? The current process, where does it get its
data?
MR. BOYCE: The reactor oversight process
is submitted directly from licensees to the NRC on a
voluntary basis.
CHAIRMAN APOSTOLAKIS: And why can't I do
that with risked-based performance indicators?
Remember, I am not advocating generic numbers, so I
don't need to have assurance of the whole of industry
in submitting data. I will do it on a plant-specific
basis.
MR. BOYCE: It does go back to acceptance.
I mean industry -- we worked very closely with
industry in order to get where we are today on the
current reactor oversight process. Industry has
already publicly stated that if we add -- I think
we're looking at an additional 30 performance
indicators, that they may not accept that on a
voluntary basis, because it's a huge additional
burden, and it opens up the potential that if you have
more performance indicators, you'll have more
opportunities across thresholds, you'll get more
regulatory attention. And they want to understand is
it really warranted? And we've heard that -- I think
you heard that at the Subcommittee meeting, and we've
heard that at public meetings.
And so we are working through these sorts
of issues, and that's implementation. And it's got to
be acceptable to all parties in order for this to work
correctly. They own the data, they need to help with
the models and make sure they're right, and it's got
to be a cooperative effort.
CHAIRMAN APOSTOLAKIS: Okay. Now, again,
for me that's a non-issue, and let me tell you why.
This is a plant-specific issue, and this Agency has
already done similar things on a plant-specific basis.
But the Maintenance Rule, I didn't hear anybody
complain about data at that time. You asked the
licensee, "Tell us what the threshold should be," and
there is a rule out there, and we're using it. Why
can't we do the same for the oversight process? "Mr.
Licensee, tell us in the integrated model, for
initiating for this and that, what would be the
thresholds?" And, of course, we look at them, we
study them, we create an Appendix F, blah, blah, and
then eventually we agree. We've done it for the
Maintenance Rule. What's so difficult with this?
MR. BOYCE: I guess you need to weigh the
costs and benefits. When you go to the Office of OMB
and we need to justify that the benefits would exceed
the costs.
CHAIRMAN APOSTOLAKIS: Okay.
MR. BOYCE: I mean that's one bureaucratic
hurdle.
CHAIRMAN APOSTOLAKIS: I understand that,
but at the same time this is hailed as a -- the
revised process is hailed as the major regulatory
change of the last 20 years. But I don't want to
elaborate the point too much.
There is one other major issue that I
think has not been addressed, neither by this project
nor by the revised oversight process. And because it
has not been addressed, we see a lot of problems here
and reaction from NEI. It seems to me that somebody
should study the tradeoffs between using a performance
indicator and baseline inspection. The way we appear
to be handling this is we are looking at the
performance indicators. Now these guys come up with
a total of 30 or so. The industry says immediately,
"Wait a minute now. How many are we going to have?"
Because the industry doesn't see on the same piece of
paper we're going to have these indicators, and we
will relax the Baseline Inspection Program in these
areas, because these areas are covered by the
indicators. As long as you don't see that tradeoff,
you will have these objections all the time.
So it seems to me that's a high-level
issue of equal importance as the previous one, but I
think both, maybe this project and most importantly
the people who run the oversight process, they should
address, because otherwise we'll have this perennial
problem. We have one transit indicator. Now you want
to make them four, I think, or some three or four.
Why? What kind of tradeoff is that? You're just
increasing the burden.
MR. BOYCE: I think philosophically we
agree with you. We would like to say that our revised
reactor oversight process was in fact a significant
step in that direction. When we took a look at going
from our Core Inspection Program to our Baseline
Inspection Program, we did exactly that sort of
approach, conceptually. We took the best data that we
had available at the time, and we said this is the
sort of PIs that we can get insights on a specific
area of plant performance, and we don't need to do
additional inspection in that area. I think you know
that -- I mean that effort was limited, but we're
pragmatists here. We're getting to that point, and we
can't expect perfection on the first try.
The risk-based PI report, as you also
know, laid out a systematic approach to here are the
accident sequences, here's the data you can collect
for performance indicators, here's the data you can
collect on an industry-wide level, and here's the gaps
that could be covered by inspection. And you brought
that up at the Subcommittee. We think that sort of
approach has got merit. We would like to see the
effort move to be more mature and gain greater
acceptance before we say, "Okay, let's charge
forward."
But in the meantime we have done a
separate effort where we're taking the significant
risk insights from various studies such as the
Initiating Events study, and research has provided
that to our inspectors and is providing those sorts of
insights to the Inspection Program Branch, and we're
attempting to incorporate those significant insights
into our current inspection procedures. It's not
perfect, but at least it's a step in the sort of
direction that you're alluding to.
CHAIRMAN APOSTOLAKIS: Now, Steve, I
understand it's not part of your charge to look at
these tradeoffs. You're just looking at the
feasibility of having certain indicators, right?
MR. MAYS: That's correct. We were
looking at what could be technically feasible using,
basically, off-the-shelf and readily available models,
tools, and data. And I think we should point out that
the Reactor Oversight Program has, as an integral part
of it, a change process where proposals to change the
indicators and the reactor oversight process can go
through. And that process involves meetings with
internal and external stakeholders, understanding of
what the implications of the information is, and an
opportunity to look at what the potential costs or
benefits are as part of the reactor oversight process
change process. We've only gone through a couple of
different things in the oversight process from that
standpoint, but I do think we have a mechanism for
doing that.
So I believe what we raised in the report
was based on our understanding of the models, methods,
and data and where this would potentially fit in the
oversight process. We said these are what we think
are the key implementation issues. And from our
discussions with internal and external stakeholders,
we've got pretty good agreement that those really are
the issues and that the process for dealing with those
issues is through the ROP change process.
CHAIRMAN APOSTOLAKIS: Maybe there is a
process, but I think the process does not emphasize
enough that within the process we are doing these --
we are making these tradeoffs between baseline
inspection and performance indicator in a systematic
way. Because otherwise, if everything is so good, why
is industry complaining that you are trying here to
increase the burden? Surely, they must know what the
process is all about.
But I think we're running out of time
here, so can you tell us what the real message you
want to send us is by summarizing your --
MR. BOYCE: I think that NRR is cautiously
supportive of the Risk-Based PI Program. We would
like to try and engage industry further to resolve
their comments on burden using the technical merits of
this product and perhaps taking a look at our
inspection practices to see if there's some solution
to those. And we'd like to try and keep moving
forward with this effort. We've endorsed it in a user
need letter, and we'd like to see the results.
I think that right now the comment period
on the Phase I report expires on the 14th of May, and
we're going to take a look at the comments that we get
and try and deal with them. And I think the schedule
for issuing this Phase I report is November time
frame. So we hope to address some of those issues
between now and then.
MR. LEITCH: I have a question about the
unplanned power change indicator that's in the ROP
now. And my question is not so much about the
definition, and I understand that may be up for
reconsideration, the precise definition of that. But
that kind of information, unplanned power change,
seems to me to be a valuable indicator, and I
understand that it doesn't really have any linkage to
risk. In other words, the risk-based -- that kind of
an indicator would not be in a Risk-Based PI Program.
And my question, basically, is if we go to
risk-based, is the thought that we have to be all
risk-based? In other words, would an indicator such
as that necessarily fall by the wayside?
MR. BOYCE: That goes back to the earlier
question I think we had on the thresholds for certain
of the indicators and why we have particular
indicators. When you get to the pragmatics of
regulating, you end up doing some things that are not,
say, fully consistent with risk techniques, like the
scram indicator. Scrams you can tolerate, I don't
know, 25 on a plant before you get past ten to the
minus six CDF. And yet we have found by comparing the
scram indicator to what used to be our definition of
problems plants -- the watch list and near-watch list
type of plants -- there was a fairly good correlation
between plants that had a high number of scrams and
plants we thought were problem plants.
And so in terms of regulatory engagement,
we found the scram indicator to be a very useful
indicator. So I can't prejudge a decision as to where
we would be, but we think we would probably continue
that scram indicator for that reason. And we think
that risk-based PIs could be an enhancement to our
current set of indicators, perhaps replacements for
many, but we would retain certain ones because they
other insights beyond pure risk.
MR. LEITCH: And the power changes made
could very well be one of those?
MR. BOYCE: It could be. I don't want to
get ahead of the problem, but it could be. All right.
I'll turn it over to Steve on page 6.
MR. MAYS: In light of the fact that we
now have about 40 minutes left for the section we
expected to take between five and ten minutes in the
initial phase, I think we may need to address an
abbreviated version even of what we have here. If
it's suitable to you, George, I would like to skip
down to the sections that you asked at the
Subcommittee that we specifically go to, which means
I will skip over the information about the potential
benefits and our development process. And I want to
go first to the table, which is on your page 8 and
give you a flavor of what we had from the draft Phase
I report and then move into the specifics of what we
had in those areas you asked us to spend more time on.
This table shows what's in the existing
Reactor Oversight Program as PIs and what areas
through our development and work we've determined as
proposed risked-based performance indicators. We went
over in greater detail the derivation of these in the
Phase I report with the Subcommittee, so we've only
put a summary of what that information is here. This
shows that the RBPIs cover more and often different
aspects of the impacts of performance on plant-
specific risk. And we'll show you some more specific
results and calculations in the V&V discussion.
You'll note that there are a couple of asterisks on
this chart that indicate potential performance
indicators that we either didn't have all the models,
data or capability to put together PIs right now,
although we think they might be something we could od
in the future.
CHAIRMAN APOSTOLAKIS: Now what you're
saying with this table, Steve, the way I understand
from the discussion so far, is that, yes, the Risked-
Based Performance Indicators Program identify more
potential indicators for mitigating systems, for
example. But you are not necessarily advocating that
these be adopted. You are saying these are feasible.
And it's another decision whether, you know, we want
to use all of them, what to do with the baseline
inspection, and so on.
MR. MAYS: That's correct.
CHAIRMAN APOSTOLAKIS: That's the way I
see it.
MR. MAYS: That's correct.
CHAIRMAN APOSTOLAKIS: Okay.
MR. MAYS: A notable thing also that we
want to bring your attention to is we had been asked
by NRR in their user need letter, as I mentioned
before, to see what we could do to come up with
indicators for shutdown, fire, and containment areas.
And we're going to talk about what we came up with for
shutdown. We were unable to produce performance
indicators for fire and containment because of either
lack of models or lack of available data.
We have three things we need to develop a
risked-based performance indicator for potential use.
The first one is a model that reasonable reflects the
risk, and the key word there is reasonable; not
perfect but reasonable. The second one is we have to
have baseline performance data to put into the model
so that we can vary that through sensitivity analysis
to see where the threshold should be set. And the
third thing we need is an ongoing source of data to
compare that performance to the thresholds.
In the case of fire and containments, we
were lacking in both models and data. In the case of
shutdown, we were able to find models and a baseline
performance and information to potentially use the
PIs. But also in the shutdown, we're not currently
gathering the data right now, but it's something we
believe is potentially able to be done relatively
easily. So we've gone ahead with the shutdown
performance indicators to discuss those.
CHAIRMAN APOSTOLAKIS: So you're not
necessarily saying that a shutdown PRA is better than
the fire PRA.
MR. MAYS: Correct. I'm not saying that.
CHAIRMAN APOSTOLAKIS: I think you have a
question coming from somewhere there, no?
DR. POWERS: Can I ask a question about
your Mark I containment spread?
MR. MAYS: Sure.
DR. POWERS: Correct me if I'm wrong, but
I believe that containment spread of Mark I is
connected also to the low-pressure injection system.
MR. MAYS: That's correct.
DR. POWERS: And most of the Mark I
containments have blanked out the containment spread;
it's non-operational.
MR. MAYS: I'm not --
DR. POWERS: It requires a manual change
to make it active.
MR. MAYS: Not that I'm aware of.
MR. HAMZEHEE: I don't think we noticed
that in our work.
DR. POWERS: I could be wrong about that.
MR. MAYS: Not that I'm aware of.
DR. POWERS: I don't think I'm wrong but
I could be.
MR. MAYS: I believe they're manually
initiated, but I don't believe they're -- I don't
think they have an automatic set point where they come
on, but I believe that they are still capable and
functional in the systems.
MR. LEITCH: They're operated from the
control room. It requires manual actuation from the
control room.
MR. MAYS: In the area of shutdown for
performance indicators, the Subcommittee asked us to
spend a little time on that. The process we used here
is a different approach slightly from what we did with
the other types of indicators that you've seen, either
in the ROP or in the other parts of the RBPI report in
that this indicator is more a measure of the impact of
configurations during a small period of time, the
outage, as opposed to an accumulation of performance
data over time, such as the reliability of a pump or
the frequency of an event that you would track over
time and history and be able to trend.
This has been linked more towards a SDP
type analysis of conditions than the standard
classical indicator definition, and we recognize that
that's the case.
Let's go to the next page here. The key
in this process was the acknowledgment that there are
certain necessary combinations of decay heat, reactor
coolant system inventory, and equipment availability
the utility must go through in order to conduct a
refueling outage. So we wanted to be able to take
into account that that was something that was a
necessary part of operations. It had some risk
associated with it. And if we were going to make
performance indicators associated with shutdown
operations, we had to allow that particular portion of
the risk to be there without penalty.
So the baseline risk was taken into
consideration. We looked at shutdown PRAs. We looked
at information about plants and how long they were
spending in various conditions in shutdown. And we
took that indication in the baseline information
that's on these tables for BWR and PWR.
Then we looked and said how much time
would somebody spend in categories of high, low,
medium or early reduced-inventory vented conditions
that would result in accumulation of risk in addition
to that baseline. And we set the thresholds according
to that to be consistent with the ROP thresholds of
ten to the minus four, ten to the minus five, and ten
to the minus six delta CDF associated with being in
performance areas outside the norm.
DR. KRESS: If the containment is
compromised during that same period, why should you
use those same deltas as your criteria? Shouldn't you
have a more stringent delta?
MR. MAYS: The issue of containment was
one where our problem is model availability to be able
to assess what the risk implications and set
thresholds are with respect to that. We're basically
going off of core damage frequency here, because
that's what we have the readily available models to
do.
DR. KRESS: But I would have thought you
might have gone a little more severe in the thresholds
for those.
MR. MAYS: The problem we faced there was
--
DR. KRESS: Maybe five or ten.
MR. MAYS: What?
DR. KRESS: Maybe five or ten.
MR. MAYS: Maybe. The problem there is it
was, again, what factor do you use and what's your
basis for saying that that particular factor has an
implication to public risk. And we just were not
capable of doing that in this particular analysis. I
don't disagree, because we said in the report that
having containment models for both at-power and
shutdown conditions would give us the ability to
determine what the impacts were on those, which we're
not able to do now.
So what we have here is baseline
information. And then on the next two slides what we
have is examples of configurations associated with
specific times, decay heats, and RCS conditions that
a plant might be in during a shutdown outage.
CHAIRMAN APOSTOLAKIS: So your indicator
here is the time the plant spends in that state?
MR. MAYS: That's correct. So what we
would do, for example, is you'll have examples on this
table where if you have a diesel generator out under
a certain set of conditions, the table will tell you
whether that's a low, medium, high or a nothing in
terms of how much you need to accumulate. So you
would accumulate all the time you spent in those
conditions under the low, add them all up and see if
that exceeded the threshold. You do a similar thing
for medium, a similar thing for high.
Now, there is one special case we have
here, which is called the early reduced-inventory
vented condition, which in order to do shutdowns
plants are often having to go into mid-loop, install
nozzle dams, do other kinds of things to conduct their
outages. Early on in the regulatory business, there
was a shutdown rulemaking effort that was underway.
There was an agreement made that there would be a
process by which the industry would put together a set
of standards for dealing, for how they would conduct
outages under those conditions.
So this indicator that we've proposed here
recognizes that condition and says, "If you are
conducting early reduced-inventory vented conditions
in accordance with the, I believe it was NUMARC 9106
guidance for shutdown configuration control, that we
would set our thresholds assuming that you had those
configurations met. If you're not in those
configurations in accordance with that document, you
would automatically transfer into the high category
under this scheme, which is a more severe and more
limiting setup.
So we're trying to give appropriate credit
for the baseline of what you have to do to get into a
shutdown and refueling. And then indicate if you've
done performance issues that exceed that, what their
potential risk significance is. And so we also have
another slide here which gives the BWR corresponding
conditions for that.
CHAIRMAN APOSTOLAKIS: Now these times are
the cumulative times over a period.
MR. MAYS: The cumulative times over the
refueling outage. So, for example, if you're a plant
operating state 4, hot shutdown with the RCS boundary
in tact, and you had a diesel generator out of service
for a certain time, that would be a low in this chart.
So you'd add up that time. And any other low times
that you were in during that outage would all be
counted together, and you compare that to the
thresholds on the previous page to see whether you had
exceeded the threshold or not. And if you're in an
area of operation where it's a blank cell, you can be
in that as much time as you want.
And, again, the industry commented during
our public meeting that we had the week after the ACRS
Subcommittee that they believed this tool was probably
more appropriate to use as a significance
determination process type of tool rather than a
performance indicator type tool. In other words, you
would use this tool to determine after the fact, if a
plant was in a certain outage condition, whether that
outage condition was really important or not.
CHAIRMAN APOSTOLAKIS: But it seems to me
that in order to go to the SDP, some sort of deviation
from something has to be observed. What is that
something in this case? If you don't have an
indicator and a threshold, why would you even enter
the SDP?
MR. MAYS: Well, the issue there would be
is this somebody's had discovered as part of their
outage, for example, that they had had equipment out
of service, like two diesel generators, when they
weren't planning on it originally. And you would go
back into something like this process and say, "Well,
what was the risk associated with being in that
condition for however long you were in it? What were
the RCS conditions and the decay heat conditions when
you were in that?" And you would make an assessment
based on this kind of an approach.
CHAIRMAN APOSTOLAKIS: I guess I don't
understand how you would decide to make the
assessment. Don't you have to deviate from something?
MR. MAYS: I agree you do --
CHAIRMAN APOSTOLAKIS: The SDP says -- I
mean the examples we heard yesterday were they forgot
to do a test. They're supposed to do a test; they
didn't do it.
MR. MAYS: Yes. Right.
CHAIRMAN APOSTOLAKIS: So that's sort of
a violation of some sort.
MR. MAYS: Right.
CHAIRMAN APOSTOLAKIS: So now you enter
the SDP or in another instance what did they do?
There was something else. But if it's something
they're supposed to do and they didn't do it, then I
go to SDP. If I don't have an indicator here, what is
that something that will make me go to the SDP?
MR. MAYS: I'm not aware of what that
would be.
MR. BARANOWSKY: This is Pat Baranowsky
again. Are you done, Dr. Kress? Am I interrupting?
DR. KRESS: Go ahead.
MR. BARANOWSKY: What I was going to say
is that remember the industry is committed to
following certain guidelines during shutdown. And one
of the things we do in inspections and verify that
they've followed those things. So as part of the
inspection they might verify that they were operating
in accordance with those guidelines, which could then
be fed into this model, if you will, to assess the
findings associated with that.
CHAIRMAN APOSTOLAKIS: So the point is
that one way or another you have to have some sort of
--
MR. BARANOWSKY: Yes. There has to be a
way to get in there, but I believe there is a way.
Maybe Tom Houghton could help me.
MR. HOUGHTON: Yes. You have to have a
performance issue, meaning either you have some event
or occurrence or you have some violation or behavior
which is viewed as suspicious in some way. If you're
not following a procedure, you're committing a
violation, and that procedure's significance, that
violation could be assessed using this process. Is
that helpful to you?
CHAIRMAN APOSTOLAKIS: That could be,
could be.
MR. HOUGHTON: Yes. I mean if you viewed
-- if you looked and there was a tech spec violation
in terms of having RHR capability, you could use this
process to determine what the risk impact of that was
and put it in perspective.
CHAIRMAN APOSTOLAKIS: It seems to me,
though, the industry should be arguing the other way,
because this already allows you some, quote,
"violation" without anything happening. Now you're
saying, no, I will have the procedures. If I deviate
a little bit, I will have to go through the whole
process, which doesn't make sense to me. Because this
already has built into it what's allowed. So I don't
have to do anything else.
MR. HOUGHTON: Well, I think it's a little
different than what's allowed, because, for instance,
there's not a limit on mid-loop operation. However,
if you look at the thresholds built into this, one
might find oneself crossing a threshold when you're
doing the perfectly right thing, which is if you're
having a problem, not to rush through to keep the
hours under two hours. And in fact the most difficult
time -- the most risky time is going in or coming out
of the mid-loop. So here I am. I'm approaching, the
clock's ticking off. I've about reached the
threshold. Two more hours I go into a yellow
threshold when I really should be stopping all work
and saying, "Let's find out what's wrong. Let's plan
and do it correctly." So that's part of the concern
about the --
CHAIRMAN APOSTOLAKIS: You are really
discouraging people from doing the cautious thing.
MR. HOUGHTON: It may or may not, and we
need to look at that more carefully.
CHAIRMAN APOSTOLAKIS: No, that's a very
valid point, I think. Which brings me to my other
favorite topic. This implies that what really
controls the risk here is the time of -- the duration.
I don't like that. Because that means that no matter
how high the risk is during that time, as long as the
exposure is short, we're okay.
DR. KRESS: Well, wait a minute, George.
CHAIRMAN APOSTOLAKIS: I know.
(Laughter.)
DR. KRESS: And you're a PRA guy.
CHAIRMAN APOSTOLAKIS: I know. You and I
have disagreed about this in the past. I don't see
why we can't disagree today.
DR. KRESS: Yes, okay.
DR. POWERS: Well, I guess the point I
would appreciate a little advice on is the summation
of hours. I mean if I enter a medium configuration
for two hours and then I come out of it, go along and
I find I have to go back in to it, why should I sum
that previous two hours? I escaped scott-free there.
Why shouldn't it be the continuous period that I'm in
there that gets evaluated?
MR. MAYS: Well, I think the answer is
that this -- if you're in for two hours and you come
out and you go back in for two hours, we're measuring
the accumulation of risk that you've incurred over
this outage. So what we're doing is saying over this
outage the accumulation of risk you have incurred by
being in these states which have relatively high risk
significance is what we want to know. We don't want
to -- you know, the idea then, if you --
DR. POWERS: I think I understand what
you're doing.
MR. MAYS: If you didn't have that
philosophy, then you could be in the high risk thing
for up to one hour before you get to threshold, back
out, come back in for a few minutes, go back up to it
again, and you would just never be there. And, in
effect, you would have been there the whole time.
DR. KRESS: Well, what bothers me about
that is -- I think it's a reasonable thing, but what
bothers me is how do you add high and low and medium
together?
MR. MAYS: Well, that's the thing we
haven't done here, we haven't done here.
DR. KRESS: Yes, I know. By the same
concept, it has to be done some way. So that's the
one that bothers me about it.
CHAIRMAN APOSTOLAKIS: Why do you have to
add high and low?
DR. KRESS: Because they represent the
cumulative risk.
MR. MAYS: That would be the cumulative
impact of the entire thing.
DR. KRESS: You can't just add the times.
CHAIRMAN APOSTOLAKIS: That's what I'm
saying, but why would you have to add them?
DR. SHACK: Well, he says he's interested
--
DR. KRESS: He's accumulating risk.
You've got to accumulate off of this.
CHAIRMAN APOSTOLAKIS: But I thought it
was cumulative for each category.
MR. MAYS: It is cumulative for each
category.
CHAIRMAN APOSTOLAKIS: Not in total.
DR. KRESS: Yes, but that doesn't make
sense.
MR. MAYS: Dr. Kress is raising the
question as if I am in the white for my low and the
white for my medium and the white for my high, what's
the net total effect?
DR. KRESS: Are you not in red overall?
MR. MAYS: And I haven't gone to the
further step of accumulating that all together,
although that could be done.
CHAIRMAN APOSTOLAKIS: Isn't that an issue
for the Action Matrix?
MR. MAYS: That's the way we set it up to
do it here, but that's another thing where we could,
as we're doing the alternate approach, we could
potentially accumulate them all together as well.
MR. HAMZEHEE: We have the same thing for
at-power situations. We don't have an accumulative
impact measurement right now except the Action Matrix.
So you have the same situation.
DR. KRESS: Yes, you would, absolutely.
CHAIRMAN APOSTOLAKIS: When it doubt, give
it to the Panel.
MR. MAYS: Okay. The next thing we wanted
to talk about was the work associated with how much
risk coverage do we have with these RBPIs and what's
the verification and validation that we've done? What
I want to do here is indicate that we have gone back
and looked at this from two different standpoints:
One from kind of a false vessel approach, one from a
risk achievement worth kind of approach.
And I'd like to put up the next slide,
which shows one of the comments that was made earlier
about how do you use risked-based performance
indicators versus risk-informed baseline inspection?
So what we did and what's in the report for all the 23
plants that were in the Phase I report is we went back
and we went through the IPE database, which was
compiled after all the IPEs were put together, as to
what the dominant sequences were at the various
plants.
And what we have in this graphic display
is we have a box around all of the areas that are part
of the dominant sequences in the IPE database where we
either have a risked-based performance indicator, we
either have an industry trending information or we
have an initiating event indicator.
And what you can see fairly quickly just
from looking at this is there aren't very many
dominant sequences for which we don't have some
multiple way of looking at what the performance of the
plant has been with respect to dominant sequences.
The other thing that it also tells you is
the areas in the dominant sequences for which we don't
either a mitigating system indicator or an initiating
event indicator or an industry trend are areas that we
should be covering in a risk-informed baseline
program.
So to answer you earlier question,
although it's not on this particular chart, you could
potentially go into this and say, "Okay. If I've got
these things covered by indicators, what are the
things I should have in my Risk-Informed Baseline
Inspection Program? I think that's one of the
valuable things that this particular program has done
is to make that more clear from a risk perspective
what those particular areas should be.
CHAIRMAN APOSTOLAKIS: And I still have
the issue, though, that you raised, which is, is it
really fair -- maybe you didn't put it in the same
words -- but is it really fair or reasonable to take,
say, the first box there, TRX, okay, number 8,
sequence number -- no, eight is no good. Tell me what
the sequence means. I can start with a TRX and then
I have the HPCT? Is that what that means?
MR. MAYS: This was a sequence where you
had a transient and you had failure of the automatic
depressurization --
CHAIRMAN APOSTOLAKIS: Oh, okay.
MR. MAYS: -- and failure of DC power.
CHAIRMAN APOSTOLAKIS: And for these I can
or cannot have --
MR. MAYS: I don't have risked-based
performance indicators for those. So those will be
areas that should be covered for that particular
function through the Risk-Informed Inspection Program.
CHAIRMAN APOSTOLAKIS: The baseline.
MR. MAYS: Right.
CHAIRMAN APOSTOLAKIS: So you are
addressing that issue now here, the tradeoffs. Very
good. But let's look at number 23 where I have the
same transient, but now you're telling us with the
boxes that I can have indicators for the two
mitigating systems, right, RCIC and HPCT.
MR. MAYS: That's correct. And the reason
this one was --
CHAIRMAN APOSTOLAKIS: Now, wait a minute.
Let me finish my thought.
MR. MAYS: Okay.
CHAIRMAN APOSTOLAKIS: So now when I set
my indicator here, my thresholds, I should take into
account, I think, in some way the fact -- I mean is it
reasonable to set the threshold in such a way that TRX
alone, its frequency, should trigger a ten to the
minus five or six change in CDF? I thought you were
arguing earlier that doesn't make sense. You
shouldn't do it one at a time.
MR. MAYS: What we did was we had -- when
we, in the Phase I report, looked at, for example, the
HPCI train reliability, we said if the HPCI train
reliability changes and everything else stays the same
for all the sequences, what would be the change in CDF
associated with that? So it wasn't just associated
with TRX; it was associated with all the sequences for
which HPCI would be affected. However, it assumes
that RCIC, the transient frequency, the LOCA
frequency, the diesel generator reliability are all at
their nominal values.
CHAIRMAN APOSTOLAKIS: So I thought you
meant something else then. But I think using the
plant-specific PRA, I can work with these things and
define the indicators at an appropriate level so that
I take advantage of these indications that I have now.
I'm not prepared myself now to tell you how to do
that, but I think that's a good thought.
In other words, on the one hand, as we
said earlier, the PIs should be as high as possible on
the PRA where the CDF is at the top, okay? And I will
try to do that as much as I can with the sequence. On
the other hand, I have this issue of having to observe
some data, which pulls me down, okay?
DR. KRESS: You're going to have a really
tough time there, George, because what these PIs are
is a sample --
CHAIRMAN APOSTOLAKIS: That's correct.
DR. KRESS: -- of things that are part of
the PRA. And you're sampling a limited -- it's a
limited sample, and you're going to look at the
degradation of all of them, and some of them may have
improved, actually. But what you're going to try to
do is now infer from that what the total plant change
has been on all the things that affect the PRA
results. That algorithm doesn't exist, and that's the
problem right here. And I don't think you can set
individual thresholds on these things without that
algorithm, and that's my problem.
CHAIRMAN APOSTOLAKIS: Okay. But maybe
there is another way out of this.
DR. KRESS: The other way out of it is to
use the Bob Christie -- here is here now.
CHAIRMAN APOSTOLAKIS: No. Christie is
only one element of this.
DR. KRESS: And set the threshold on delta
CDF itself.
CHAIRMAN APOSTOLAKIS: No, no, no. But
there is another way of doing this. You remember that
this Committee has asked the staff to explain how the
Action Matrix was developed and what does it mean --
why two reds make this and one yellow and one white.
DR. KRESS: Yes. That impacts --
CHAIRMAN APOSTOLAKIS: We can use this
table now --
DR. KRESS: That impacts on that.
CHAIRMAN APOSTOLAKIS: -- to scrutinize
the Action Matrix --
DR. KRESS: You're right. That would --
CHAIRMAN APOSTOLAKIS: -- rather than
worrying about the thresholds for individual events,
which have the problems we mentioned.
DR. KRESS: But once again, in order to do
that, you have to have this missing algorithm that I
talked about that says the total effect on the whole
PRA, due to the changes, which are variable, variable
changes, and going in different directions, you have
to have some sort of algorithm to convert that.
CHAIRMAN APOSTOLAKIS: I think Steve and
his colleagues can do some sensitivity studies for us
--
DR. KRESS: They might, they might.
CHAIRMAN APOSTOLAKIS: -- by taking tables
like this --
DR. KRESS: They haven't done this yet.
CHAIRMAN APOSTOLAKIS: Well, because they
are overwhelmed, but they can do it. They can do it.
They can do these calculations, and you never know.
Maybe you'll find that two whites usually lead to the
same changes --
DR. KRESS: It not just a matter of doing
some calculations that are sensitivity. It's a
missing algorithm that's a correlation. It's a
correlational algorithm between these things that's
missing. It's not just a matter of doing some
calculations.
DR. POWERS: But, Tom, it's not missing.
It's maybe implausible to create?
DR. KRESS: Pardon?
DR. POWERS: It may be impossible to
create.
DR. KRESS: Maybe. That's my point.
CHAIRMAN APOSTOLAKIS: It could be. It
could be.
DR. KRESS: And so you have to do
something in its stead. And I don't know what that
something is, but you have to make some reasonable
assumptions or reasonable approximations that are
maybe bounding or maybe a little more conservative
than you might want.
CHAIRMAN APOSTOLAKIS: But I can take it
the other way. What if I were doing something
negative? If I take Table 4-2(a) and pick two whites
or a white and a yellow from sequence 5 and sequence
20 and I calculate those delta CDF, assuming
everything else is the same, and I find it's X. Then
I take another white and another yellow and I find
that the new delta CDF is 20 times X. Then I know I
have a problem with the Action Matrix.
DR. KRESS: Well, that's something --
CHAIRMAN APOSTOLAKIS: That's a negative.
DR. KRESS: That doesn't tell you how to
deal with it.
CHAIRMAN APOSTOLAKIS: No. But it tells
me I --
DR. KRESS: It tells you you have a
problem.
CHAIRMAN APOSTOLAKIS: Which I don't know
right now.
DR. KRESS: But I already know you have a
problem.
CHAIRMAN APOSTOLAKIS: I don't know that
I have a problem, because these guys come in here and
say it's a professional judgment; this makes sense.
But this will be definite proof that you have a
problem.
DR. KRESS: Well, that would be
worthwhile.
CHAIRMAN APOSTOLAKIS: And then Steve will
come back and justify it.
DR. KRESS: Then I would say, "I told you
so."
MR. MAYS: As soon as you sign the check,
George.
(Laughter.)
CHAIRMAN APOSTOLAKIS: Now, Steve, we're
really running out of time, and I trust that you can
summarize your presentation. Still got the letter?
MR. MAYS: Yes, we do. Let me go to move
down to -- I will go with two things.
CHAIRMAN APOSTOLAKIS: This is a wonderful
table, by the way.
MR. MAYS: Thank you.
DR. KRESS: Yes, that's a good table.
CHAIRMAN APOSTOLAKIS: It really is. See,
again, I can't resist this. Why didn't we do all this
work before revised the oversight process?
MR. MAYS: Actually, we were putting
together a program, as you're aware of. From the
beginning, we came down in 1995 and spoke to the ACRS
about our plan for risk-based analysis reactor
operating experience, and we laid out a matrix at that
time that said here's the stuff we're trying to get
data on, on a plant-specific basis and across systems
and components and things to say this is the
information we would use to be able to understand risk
implications of operating experience. So we've been
working on this since 1994 and 1995 time frame to get
the basic methods, models, data, and information
together to be able to do this kind of thing.
Now, the Reactor Oversight Program
development and the crisis that came about in the
summer of 1998, I guess, came here and that helped to
provide an impetus for doing an oversight program that
was more along the lines of what we were working on
here. And we're just continuing to try to push that
envelope a little bit more as we get more data, more
capability, and more information.
Because, remember, the thing we're trying
to do here is go to progress, not perfection. We
don't want to end up in the old source term problem
where you have a source term that had a generation
coming from a need and then subsequently you might
have 20 years of research to get more technically
capable and competent understanding of the source
term, but you couldn't change it until you got one
that everybody thought was more perfect. So we're
trying to avoid that problem here. We recognize that
there are places where this doesn't do everything you
might ever want to do. But we believe it's --
CHAIRMAN APOSTOLAKIS: You're not implying
that the Committee does not appreciate the distinction
between progress and perfection.
MR. MAYS: No, I'm just saying that we
have to make sure we keep that in mind as we go
forward.
CHAIRMAN APOSTOLAKIS: The Committee does
keep that in mind, just as the Committee understands
what engineering approximation is.
MR. MAYS: Let's go to -- I want to go to
the alternate approach.
CHAIRMAN APOSTOLAKIS: I think you should
highlight some of the good stuff you have here and
tell us what you are trying to do.
MR. MAYS: Okay. I want to go to the
alternate approach thing, because we've bumped into it
a few times, and I want to talk about that a little
bit.
One of the things we got a lot of comments
on was the excessive number and increase in the PIs
implicated by potentially adopting these. And the
major limitation that drove us to the number of PIs
that we've done was a philosophy that says that you
are going to set thresholds at the basis of where you
were collecting data. That's the way it had been done
in the past. That's the way it was in the reactor
oversight process. And we were making our first
attempt at risked-based performance indicators using
that.
What we subsequently decided to do was to
go back and re-look at that and see if we could come
up with a different concept that would reduce the
overall number of indicators but still keep the
fidelity towards risk that we were having in the RBPI
process on a plant-specific basis.
So what we did was we said -- let's go to
this Figure 1 now. If we break core damage frequency
down into two major groups, the initiators and the
mitigation, you can subsequently break those down into
some general categories, such as transients, LOCAs,
and special initiators for the initiating events. And
you can break mitigation systems down, generally, into
functions like reactivity control, heat removal, feed
and bleed, recirculation, which are the kind of
general terms people talk about when they make
functional event trees or talk about risk assessments.
So let's go to Figure 3 now. What we did
was we said let's reevaluate the concept a little bit.
So what we would do is we'd take the same inputs that
we were having for individual risked-based performance
indicators in the Phase I report, and we said let's
put them into a more complicated, a higher level
functional model, and then compare the sequence
changes in core damage frequency that we would get by
exercising that model.
So we did that. This is work we've done
since the Phase I draft report was published. And
what we came up with was three potential hierarchies
that we could do these indicators for. One of them
was at the cornerstone level. So we would say -- we
would have one indicator for initiating events and
mitigating systems that would represent the overall
impact of all of the changes for the data that we were
gathering for the individual indicators. So we would
have an indicator that said for mitigating systems,
whatever the changes were in reliability, whatever the
changes were in availability for all the systems, we'd
integrate them together through the risk model and say
what was the net change in core damage frequency
associated with that performance.
Now, the advantage there is we now have
the integration you were talking about earlier. Maybe
one system's unavailability went up, and maybe another
one went down. Maybe certain performances went
differently. But we would now have an integrated
approach to doing that. And we would have an
indicator at the cornerstone level of the reactor
oversight process.
CHAIRMAN APOSTOLAKIS: And, again, an
alternative to that is not to worry so much about the
indicator, where to put the indicator, to keep the
indicators at the lower level, but have the Action
Matrix take care of these things. In other words, as
you enter the Action Matrix, if you have a change in
a mitigating function that I can measure its impact on
the CDF, then I react differently than if I had just
something else. So there is a combination there that
it's not just where I put the indicators.
MR. MAYS: The other thing we looked at,
doing the same approach, was to -- we looked at
putting together functional-level indicators, and we
chose two groups to try this out on. One group was by
initiators. So we would, for example, say four
transient initiators, what is the impact of all the
different variations in the mitigating systems on
those sequences associated with transients. And then
we'd have another indicator for those sequences
associated with loss of off-site power. And we would
have another indicator for those sequences associated
with loss of coolant accidents. So we found that we
could go back through the models and put an indicator
where we would have three to five indicators per plant
that were more rolled up and more integrated at a
functional level, although less integrated than at the
cornerstone.
And then the last level was down at the
component or train level where we had already done
work in the risked-based performance indicators. And
at the Subcommittee, one of the things that was
brought up was, "Well, why don't you just do them at
all these indicators? Why don't you have maybe an
official indicator at the cornerstone level and you
have functional or component level indicators so that
you can understand when you have a non-green
performance what specifically it was about it that was
non-green and what was actually causing that condition
to be done?" That's a potential possibility that we
could do.
We were looking for some advice based on
that concept in the stuff we showed at the -- that's
in the package here as well that we showed to the
Subcommittee as to whether you thought that was
something we ought to pursue, that ought to be in the
Phase I report or something we ought to take more time
and maybe put in Phase II of the RBPI development. So
we're looking for some input on that.
But we think we have models that if we
take this data, we can evaluate risk performance on a
plant-specific basis at whichever level we choose to.
I think that's the thing you should be taking away
from this. And the question of what's the right level
to do is something that would have to be negotiated
with the industry, the other external stakeholders,
and the public to say what makes the most sense as an
improvement on the existing reactor oversight process.
And there's pluses and minuses for each of them, which
we discuss in some of the other slides here.
So having done that, we also had a meeting
-- I want to go back to this industry one -- we also
had a meeting with the public the week after we had
the Subcommittee meeting, and this is a summary of
some of the issues that they thought were important.
I think we also presented the alternate approach at
that meeting, and these are the issues that the
industry folks raised during that meeting. I think
these are primarily issues associated with how do we
implement this stuff and what are the implications to
plants in terms of the regulatory responses if we were
to implement a process like this.
I think the oversight process/change
process is supposed to be able to be the place where
we evaluate and make assessments on that. The thing
I'm looking for from the ACRS in terms of a letter, we
want to know what your feelings are about whether this
looks like it's a potential benefit to the reactor
oversight process that should be pursued.
DR. POWERS: Steve, let me ask you this
question: Suppose we did something like this and
suppose I'm a member of the public and I say, "Gee,
how do these guys have three fire barrier penetration
seals out of commission? It sounds pretty hazardous
to me, but they tell me it's a green finding." That's
what you will tell me that it's a green finding. And
I say, "I wonder how in the world did they arrive at
that conclusion that it was a green finding?" Am I
going to be able to figure out how you got to that
being a green or am I going to have to take that on
faith?
MR. MAYS: I think with respect to what
we've done with risked-based performance indicators,
we will have the capability out there in the public
domain for somebody to duplicate our analysis and our
work. I mean not every single member of the public
will be able to do that, but I mean we'll have the
information out there so that people will be able to
do that.
The case you're representing would be from
the significance determination process. I'm not as
familiar with the specifics of the fire SDP, but my
understanding is the logic and the framework for if
you have this condition, we characterize it this way
and that causes a result to come out would be
available and open to the public.
But I think what you're raising is a
larger question. And the larger question is, how, as
an agency, do we communicate risk importance to the
public and in what context do we do that? That is a
significant challenge that we faced for a long time.
I agree it's something that we can improve on.
I'm not sure what exactly that form should
be, but I agree we're going to have problems in that
area with any new oversight process that we've come up
with. And people need to be able to have some sense
of feeling of what does the green mean, what does the
white mean, and how do I know what the implication of
that is to me? Now I believe the Oversight Program
tried to do that in SECY-99-007 and in the NUREG that
they issued, which was the summary of that, but I'm
just not in a position to really say much more than
that.
DR. POWERS: It's a very thorny problem,
and I choose fire protection, because fire is one of
those hazards that nuclear power plants face that's
very palpable to any individual. I mean you just know
fire is a bad idea, and you kind of know what it's
going to do. And so when you see failures in the fire
protection system, some of those are very familiar.
They're unlike pressurizers or high-pressure injection
systems. You have many of them in your own house or
your own business that you work at.
And so you see failures of these things.
You say, "Gee, that ought to be significant. I would
do something about that in my own business if I saw
those fire penetration seals failing." And it's not
the licensee is not doing something about it. It's
that the regulator doesn't feel like he needs to do
anything about, because he finds it a green finding.
But that' not very easily communicated to an
individual who has been cautioned to worry about
nuclear power plants.
I mean it came up today in the meeting
with the Commission. I think it's an area that we
can't continue to say, "Gee, that's a problem we're
going to have to address one of these days." We've
got to address it. And it seems like you have the
vehicle for doing it.
CHAIRMAN APOSTOLAKIS: I just don't think
that's a problem.
MR. MAYS: Well, I think we have a --
CHAIRMAN APOSTOLAKIS: Why don't you tell
the public whatever that means that green means that,
look, this is a major industrial facility. It has
40,000 components, it has 800 people working on it.
Little things happen here and there. By design and
regulations and so on we have allowed for these, and
in this particular case our analysis shows that it has
an insignificant impact. What's wrong with that?
MR. MAYS: Well, I think you're touching
on one of how might one go about doing that, and I
think my interpretation of Dana's question is do we
have an agency process for making sure that that kind
of communication takes place in a consistent way so
that people have an understanding of that? That's the
age-old question that Chauncey Starr raised years ago
in his "Perceptions of Risk" Paper.
And I think Dana's correct, every person
in their house can say, "Oh, I know fire's a bad
thing." I had a fire in my kitchen once. But nobody
understands what the issue about the availability of
the high-pressure core spray pump, because they don't
have any high-pressure core spray pumps. Maybe they
can make an analogy to their sump pump in their house
or something, I don't know.
But I think risk communication is an
important feature that we have to be able to do as an
agency in order to meet our strategic goals for public
confidence. I just don't think --
CHAIRMAN APOSTOLAKIS: The reactor
oversight process, I thought it was very good. Have
you guys seen this?
MR. MAYS: Yes.
CHAIRMAN APOSTOLAKIS: No, I know you
have. Have you gentlemen seen it?
MR. MAYS: Thanks a lot, George.
(Laughter.)
MR. BOYCE: If I could use that as a segue
a minute. We have to face this issue in the reactor
oversight process today, how do you communicate SDP
results in a coherent manner that's understandable?
And what you have to look at -- it primarily comes
down to the web page really.
DR. POWERS: Even to very technically
sophisticated people, how do you communicate the SDP
results?
MR. BOYCE: That's exactly right. And we
have that problem. And the primary vehicle, actually,
turns out to be the web page for everybody. And
everybody includes intervenor groups, casual members
of the public who are browsing from America Online,
licensees, staff members --
CHAIRMAN APOSTOLAKIS: Do you have any
data that showed you -- give you some idea of how many
members of the public actually do this?
MR. BOYCE: Actually, yes. If you go onto
-- in fact, you can access it from the internal web
yourself. If you go onto NRC's home page, there's a
spot there that says, "Web Statistics." And it will
tell you -- it's actually pretty good. It's a
contractor program --
CHAIRMAN APOSTOLAKIS: What does it tell
you?
MR. BOYCE: -- that collects data on I
guess it's the domain names that have accessed the
pages, the entrance page, the exit page, the number of
hits on a page, and that sort of thing.
CHAIRMAN APOSTOLAKIS: But that doesn't
tell you that these people were public.
MR. BOYCE: Well, what you end up doing is
you find out that they come from aol.com, and you find
out they come from nrc.gov, and you find out that they
come from dot-org. And, so you can get a rough idea
of the usage.
CHAIRMAN APOSTOLAKIS: Oh, you know that.
Okay. So there are some data.
MR. BOYCE: Yes, from the domain names.
CHAIRMAN APOSTOLAKIS: So there is a
significant number of hits from --
MR. BOYCE: From the members of the
general public.
CHAIRMAN APOSTOLAKIS: -- a basis where we
might suspect there is public involved?
MR. BOYCE: Well, yes. As a matter of
fact, one of the -- it's interesting that whenever we
issue a press release, the number of hits spikes on
our web pages.
CHAIRMAN APOSTOLAKIS: Whenever you do
what?
MR. BOYCE: Whenever we issue a press
release.
CHAIRMAN APOSTOLAKIS: Is that right?
MR. BOYCE: The number of hits spikes.
And it comes from places like America Online. The
geographical --
CHAIRMAN APOSTOLAKIS: But it could be
inside NRC?
MR. BOYCE: It may very well could be.
CHAIRMAN APOSTOLAKIS: I mean those guys
are professionals. I don't count them as public.
DR. WALLIS: Well, the press releases is
attractive, because it might be understandable. I
think a hit doesn't mean that the person who hit
understood what he read.
MR. BOYCE: Correct.
DR. WALLIS: That's the problem I think
you have.
MR. BOYCE: Correct. And trying to bring
it back to where we are, the web pages is our primary
vehicle for communication right now. And what we have
tried to put on it is this colorized scheme to make it
easier to understand. And we put all our inspection
reports by cornerstone on the web page so that you
start off with a color, and if you have a white color
or yellow color, you can click on the color and you
get down to the next level of detail. The next level
of detail would be perhaps an NRC assessment letter
saying, "We've reviewed your performance over the
previous year, and this is our assessment." If you
want to know about a specific topic, like an
inspection finding, you click on that color. It will
take you down to the inspection report, which talks
about the NRC's view of that.
We're getting to the point where we're
putting our, what we call, SDP letters on the web so
that all the information and how we characterize it
will be there. I'm not going to tell you it's
perfect, but it's what we're doing today. We've
gotten additional -- we had a public communication
session as part of our lessons learned workshop at the
end of March, and we got a lot of feedback that we
needed to do better in this regard.
So we're at the forefront telling you what
we're doing. We can't solve the world's problems, but
here we are.
CHAIRMAN APOSTOLAKIS: Steve, is there
anything else that you think you should point out to
the Committee?
MR. MAYS: No, I think the key thing that
I want you to come away with is that we have the
ability, using readily available data and models, to
be able to estimate plant-specific performance impacts
on risk in several areas that are broader, more
comprehensive, and can be integrated, using the
alternate approaches we're proposing here, to give us
indication of performance at various levels. And if
this is something the Committee thinks we should go
forward with, we would appreciate hearing about it.
If there are aspects of how we're doing it you'd like
us to do different, we'd like to hear about that too.
I think realistically it's going to take
a considerable amount of time to meet with the
external folks, go through process, because this is
going to be primarily voluntary process to do. And
we're going to have to show people what we have,
examine the stuff in a bigger picture than just what
the technical stuff is. But I want the implications
of what it will mean to you. But that's specifically
what the Reactor Oversight Process Change Program and
procedure is designed to do.
CHAIRMAN APOSTOLAKIS: You said that your
so-called alternative approaches are described where?
MR. MAYS: They're only the presentation
we made to you on the Subcommittee and the stuff
that's in this particular thing. They are in the
report, because we got these comments after the draft
report was put out, and we went to be proactive rather
than just sitting on our hands until the comments came
in that says, "That's too many PIs." We said, "Well,
what other things, since we know that issue, can we go
work on now?" And what we've done is we said, "There
are some things that we could do that can solve some
of the problems we've had in other areas."
Because one of the things we found, for
example, in the ROP comparisons in this, when did the
integrated look, we found that sometimes we would
have, on an individual PI basis, a green and a white.
And when you get to the integrated, it comes out
green, because the green had improved so much and it
was on the same sequence as the white, it basically
counteracted it. And on the other hand, we found
cases where we had green and green indicators, and you
put them in the integrated indicator and they come out
white, because they were green, but they were both
getting worse at the same time. So even though one
individual didn't cross over an individual threshold
together, they would have crossed the threshold. I
think that's an important -- from my risk perspective,
that's an important piece of information to have.
CHAIRMAN APOSTOLAKIS: It's very
important.
MR. MAYS: And we also had -- in the ROP
comparison stuff, we had examples where the ROP would
indicate one color, and we would see worse and other
cases where the ROP would say worse, and we would see
green. And we were able to go back and look at each
one of those specific cases and look at them from the
standpoint of what's making this true and that face
validity test, which we used in the slides, we were
able to come to a reasonable conclusion from a risk
perspective of why that really was true.
For example, we were using a plant-
specific threshold instead of a generic threshold.
For example, we weren't averaging diverse trains; we
were using individual trains. So those were all the
kinds of things we found that I think tell me, anyway,
we can do a better job of understanding risk
performance with this process than the current ROP.
And, again, progress not perfection. That's not
saying it's broken and dead and is wrong. We're
saying what we have here is potentially better.
CHAIRMAN APOSTOLAKIS: What you're saying
-- I'll give you an example for me to understand it
better. A particular indicator of the plant may
formerly be yellow but because the utility is aware of
it and they're doing something else better, the
overall impact may be zero, right?
MR. MAYS: Well, the impact may be white,
it may be green, it may be still yellow, I don't know.
What I'm saying is without an integrated model you
can't tell.
CHAIRMAN APOSTOLAKIS: And you have the
tools to investigate.
MR. MAYS: I think we have the tools to
investigate that.
CHAIRMAN APOSTOLAKIS: Speaking of tools,
Steve, do you also have tools to test the hypothesis
that if human performance and the safety culture of
the plant deteriorates, then we will see the impact on
the equipment decline in performance?
MR. MAYS: We have the tools to determine
when we see degradations in the performance of the
equipment, whether or not the factors causing that
were related to Corrective Action Program or other
things. We don't have tools to directly measure
Corrective Action Program and then posit what the risk
impact would be. So if you were to look at public
risk and make yourself a hierarchy, here's public
risk, and then somewhere below public risk is core
damage risk, and somewhere below that is system or
train level performance, and somewhere below that is
component performance. I think what you see is that
the safety culture is somewhere below that in terms of
being how leading you want to get from public risk
down to the least level of detail that you might be
able to do.
I don't have metrics to link safety
culture measures that --
CHAIRMAN APOSTOLAKIS: But do you have
tools?
MR. MAYS: I have tools to be able to see
when I see a performance degradation at the lower
levels of risk to be able to go back and examine
whether the fundamental causes of that were safety
culture, corrective action or other problems.
CHAIRMAN APOSTOLAKIS: So maybe that's
something different. Maybe it has to do with root
cause analysis.
MR. MAYS: Correct.
MR. HAMZEHEE: If the impact is on the
equipment performance.
MR. MAYS: Right, if the impact is on --
CHAIRMAN APOSTOLAKIS: Don't give me
cryptic statements, Hossein.
MR. MAYS: If the impact were to be --
CHAIRMAN APOSTOLAKIS: What else could it
be?
MR. MAYS: Well, for example, on the
ability of the operators to respond to an accident.
So we don't have data on being able to make sure that
you initiate slick within five or ten minutes after an
accident. So we don't have that kind of data either.
CHAIRMAN APOSTOLAKIS: Okay.
MR. BOYCE: The Allegation Program does
compile statistics at an industry level.
CHAIRMAN APOSTOLAKIS: But we are not
using those to confirm this hypothesis.
MR. BOYCE: Correct. In terms of tools,
it's not a tool, but that's at least the best
indicators we have for a safety conscious work
environment.
DR. WALLIS: George, you never confirm my
hypothesis; you just disprove it.
CHAIRMAN APOSTOLAKIS: Yes, yes. I stand
corrected. Thank you, gentlemen. This was a very
lively session; appreciate it. Are you happier today?
MR. BOYCE: I was able to respond better
to your questions today, which does make me happier.
CHAIRMAN APOSTOLAKIS: Okay. Thank you
very much. Now we will hear from Mr. Houghton of NEI.
MR. HOUGHTON: Good afternoon. My name is
Tom Houghton. I am the Project Manager for the
reactor oversight process at NEI.
DR. KRESS: This is your first test to see
if you can turn that on.
MR. HOUGHTON: First test is -- okay.
Well, I think I have four slides here, and I've tried
to summarize a lot of points on these. We do support
movement towards risked-based performance indicators
with some caveats. And the caveats, a very important
one, depends upon the ability to integrate what's
going on across the different aspects of regulatory
space. And by that I'm particularly talking in the
mitigating area to the dichotomy between design basis
technical specifications and risked-based performance
indicators.
And it plays a big role, because the
inspectors inspect to the design basis, and if we're
trying to move towards risked-based performance
indicators, we're shifting the focus of this
performance indicator. The performance indicator's
purpose is not to measure risk. performance
indicator's purpose is to help the NRC manage its
resources and determine where to put its inspection
resources. So the inspectors are aiming at design
basis, i.e. the automatic function would not have
worked. And the risk-based indicator allows operator
recovery, because the mission time is seven days, and
there is seven days to restore the function.
We have a big dichotomy here. And we're
seeing that already between the Maintenance Rule and
the tech specs. And we'll see it even more in the
risked-based performance indicators unless we address
this problem up front with a plan that solves the
problem so we're not having people going in different
directions. And that really is a key issue in going
forward with risked-based performance indicators.
Second point I put on here is the PIs and
the inspection findings, their aim is to tell the
inspectors how much additional inspection to do beyond
the baseline. And, therefore, the indicators need to
provide that value at the same time not adding
additional burden and to help us all focus on what's
risk-important. Now, I said complement inspection
activity, I didn't say reduce.
DR. WALLIS: This refrain about avoiding
unnecessary burden, there's always a complementary
side. When additional burden is appropriate it should
be there.
MR. HOUGHTON: Absolutely, absolutely.
Now, I agree with you, but to do that, one needs to
look and say, "Okay, the current number of hours in
the baseline inspection is actually slightly higher
than it was."
DR. WALLIS: You really should say the
regulatory burden should be appropriate.
MR. HOUGHTON: Yes, yes. It should be
appropriate. To do that, one needs to ensure that
additional reporting falls under 50.9 and has to be
accurate to very fine levels is appropriate for the
amount of effort people are going to have to put into
that. And we do have inspectors that have gone down
and looked at 15 minutes of availability time, of the
time that was written in the log, as opposed to
something else. And it can cause a lot of inspection
effort by the NRC and by the licensees unless we're
careful. And by adding additional indicators, we add
to that area. So that we would say, let's add more
indicators, but let's have a tradeoff here. And if
there is no tradeoff, then there's no advantage to
doing it other than to gather more information to what
purpose.
This 0609 Manual Chapter is the chapter
that tells NRC how to proceed with interpretations of
performance indicators, and we've had about 256
questions over the year on interpretation of
indicators, mostly in the mitigation unavailability
area. But it tells them what process to go through.
And I think although research is -- as I understand
it, research's duty here is to look at the technical
feasibility, but we're looking ahead to see if these
indicators are practicable to be used, okay?
So we're looking at those aspects, okay,
easy to understand. We would wonder an indicator
which rolled up, either to a cornerstone or at a
higher level and how difficult that would be for
someone to be able to readily understand. I mean you
may not know what a high-pressure injection system is,
but you know it's a system. If you're talking about
the cornerstone of initiating events, that's an
abstraction.
I think I covered the other points there,
but the 608 is important.
DR. POWERS: Could you explain the title
of the slide? The title has me confused.
MR. HOUGHTON: Oh. I'm glad you asked
that, because I should have discussed that. The
purpose of the performance indicators and the
inspection findings is to help determine where
management should put resources. And we basically
have three stakeholders. We've got the Regulatory
Commission, which needs to assign resources, we've got
the industry, and we've got the public. And our
feeling is is that you have to -- these indicators
have to meet the needs of all three stakeholders in
this process. So you can't have extremely
sophisticated indicators, you can't have indicators
that are hard to collect accurately, and you've got to
have indicators that are actionable by the Commission
and by the people that are living with them. That's
my point.
DR. POWERS: I understand better now.
First I thought we were talking about producing
electricity.
MR. HOUGHTON: I'll hold it at the ROP
level.
Some comments on the draft PIs themselves,
and I think this is partly understanding and working
through. But the thresholds need to be set at
practical levels for action. That may vary from what
a very strict risk study tells you. So if I were to
look, for example, at the loss of heat removal
threshold for one of the plants in the study, you'd be
allowed 0.7 reactor scrams with the loss of heat
removal in a three-year period. That means your
threshold is less than one. That means there is no
threshold. There are some difficulties in going
strictly by a risk-based threshold system. It needs
to be modified to be practicable.
Another example for you is the general
transient green/white threshold, as I looked through
the plants that were reviewed, varied. One plant
would have a threshold of 1.2 general transient scrams
per year; another one would have 8.2. Now if I'm a
plant manager and I have two scrams in a year and I
get an extra inspection and I get a mark of a white,
and my neighbor has eight scrams in a year, and he is
considered in the green band, that doesn't make sense.
It just doesn't make sense. It has to be an indicator
which is --
CHAIRMAN APOSTOLAKIS: No, but -- well, in
all fairness, if you're running a plant where the
threshold is two, there must be some serious reasons
why.
MR. HOUGHTON: No, I think it -- and I
defer to the risk experts, but the threshold is based
on a ten to the minus six delta CDF.
CHAIRMAN APOSTOLAKIS: Yes, but that is
converted for your plant to a threshold of two, which
means you don't have enough mitigating capability,
right?
MR. HOUGHTON: But this PI won't get at
that problem.
CHAIRMAN APOSTOLAKIS: So you should pay
the price. That's the way I see it.
MR. HOUGHTON: But the PI won't get at
that problem.
CHAIRMAN APOSTOLAKIS: No, the threshold
gets at the problem.
DR. POWERS: For your first example, you
just want that to be one per four years, is that all?
MR. HOUGHTON: Well, right now we've
combined the loss of heat sink and the loss of -- the
current indicator combines the loss of heat sink and
loss of feedwater, and the green/white threshold is
actually two. And there are several plants that have
tripped that threshold, and they've done extensive
cause analysis for the situation. But that isn't a --
I'm pointing out that there are practicalities that
need to be -- you can't blindly use a risk-based
approach.
CHAIRMAN APOSTOLAKIS: And what I'm saying
is that, you know, of course you should be practical,
but at the same time, there is a reason behind this.
And maybe a plant that is very well defended can
afford to have maybe a couple more transients a year.
Whereas another one that is not may be should not.
DR. POWERS: Wait until you see the kind
of performance indicators we were talking about
earlier that are very much more complex. I mean then
you'll have some real problems with that practicality.
MR. HOUGHTON: But I would think --
CHAIRMAN APOSTOLAKIS: No, I appreciate
the point, but I want you to appreciate mine.
MR. HOUGHTON: Yes, sir; I do. But I
would say is that the venue for the discussion of
whether you have a robust enough mitigating system is
not the ROP, because the ROP is looking at your
performance under the current rules and regulations
and activities you're supposed to do.
CHAIRMAN APOSTOLAKIS: Anyway, you're
talking about the draft PIs as given by these guys
today?
MR. HOUGHTON: Yes, yes.
CHAIRMAN APOSTOLAKIS: Okay.
MR. HOUGHTON: And we went through these
discussions, actually, when we set the thresholds in
the ROP, because we did have differences such as
these, and what happened was is there was an
accommodation of, "Well, let's use three scrams per
year. Even though one plant is 2.1 and another is 7,
we'll use three, because what we're really trying to
do is look at are you maintaining and operating in an
effective manner."
The mitigating systems, the most important
issue, as I said to you several weeks ago, is the
unavailability definition. And as I just said, it
gets into issues of design basis versus risk basis.
It gets into credit for operator action. It gets into
cascading of support systems and whether we do that or
not. And it gets into the reliability indicator in
place of demand fault exposure. And we're very -- we
support very much working towards, moving towards a
more risk-based approach in this area, because we
think that's appropriate, and it's more in line with
the Maintenance Rule, and it can help to avoid this
problem. I talked about it, having two or three
different targets that you're aiming at.
MR. MAYS: Tom?
MR. HOUGHTON: Yes.
MR. MAYS: If I might, the issues on
design and licensing basis for unavailability and
whether or not operator action is credited and role
support systems and the fault exposure times were all
issues that in the draft RBPI were done in the
direction to which you're concerned that we should be
moving.
MR. HOUGHTON: Yes, I agree, and that's
what I was trying to say. We see that as moving
positively. However, the tech spec issue is looming
out there.
The component class PIs, we feel that
better covered by the SDP and by the extent of
condition in root cause analysis rather than having
separate PIs. We don't feel there's a -- that there
would be less inspection coming about through having
PIs in those areas.
Now on the shutdown PI, I think there was
some discussion of the level of our concern in terms
of the amount of time and basing it on time when we
think that you could have negative consequences of
people trying to rush out of conditions. And it's not
really appropriate to have indicators like this. We
want to hear more about it, but when you look at some
of the thresholds in that table, you'll find that
they're very unforgiving, and you can move from being
green to yellow in just two or three hours. And when
you're trying to be careful you don't want to put
yourself in that situation. It's not clear that
that's a good idea. We do think that it could be very
helpful in the Phase II once you know you have a
problem to see what sort of the risk level was.
Adding PIs requires examination of the
Action Matrix. A comment here: I heard the talk
about rolling up PIs to a higher level and then
putting that in the Action Matrix. But the Action
Matrix includes both the inspection findings and the
PIs. And the Action Matrix really is more of a logic
table to tell you if you have two or more -- if you
have single white, be it a PI or an inspection
finding, the NRC is going to look at your root cause
and look at your corrective action. And it might
require up to 40 hours of additional inspection.
That's what that means.
The second column tells you that you have
two or three white indicators in a particular
cornerstone, whether that's physical security or
emergency planning or the barriers or mitigation. And
what it's saying is, "We're not so sure you're
handling things right, and we're going to come and
look at your ability to do root cause and look at your
ability to integrate this problem across different
systems."
The next column, a yellow or degraded
cornerstone says, "You have a more systemic problem
and we're going to increase the inspection level
higher." The next column probably has you in a
diagnostic, like Indian Point. There's a very
interesting inspection report that showed Indian Point
if it had been under the new system for the year prior
to the steam generator tube rupture. And it shows you
that the Action Matrix in the system would have shown
a steady degradation and the need for more inspection
earlier on at Indian Point. I commend that to you to
see how that worked, because that's an actual case
study. It wasn't applying to them at the time.
So we see the Action Matrix as not being
a risk meter at a certain level, but we see it as
indications of problems across distinct areas. And as
they increase, the Agency needs to take a closer and
closer look at the problem. So we would really feel
that aggregating these PIs you still have to compare
it with inspection findings, so you're not really
integrating risk with the inspection findings. And we
really think that the PIs at the level they are are
actionable indicators.
DR. WALLIS: I understand that, but
remember the public looking in, this is really a risk
meter, and the public is not interested in a
management tool. It's interested in how a state
complies.
MR. HOUGHTON: Well, but they're
interested, I think, in how does the NRC judge the
plant. And you now can click on the Action Matrix,
and you can see the 79 plants that are in the licensee
control band, the 16 or 18 that are in the next one,
the three or four that are in the next one, and
finally Indian Point on the site. And it also tells
you why they've changed from column to column. So it
--
DR. WALLIS: I don't really care about
that. If I was a member of the public, I probably
look at it say, "Well, I want a good feeling that
these things are safe enough. Here's a measure I've
got." So it's going to be used in some way as a risk
meter whether you like it or not.
MR. HOUGHTON: Agreed, but I'm not sure
that -- it's not clear to me that a risk number would
tell someone more than being told that there are
systemic problems across different areas. That's my
opinion.
MR. SIEBER: It seems to me that that's a
two-edge sword. For example, if you could predict the
declining performance at Indian Point and then begin
doing diagnostics and additional inspections, that
probably would not have prevented the tube rupture.
MR. HOUGHTON: Right.
MR. SIEBER: Okay. So now the Agency is
called into question. You knew this Plant was going
downhill, yet you weren't able to prevent this event,
even though the two are not associated. And I think
you have to be careful about that, because a lot of
these events are random events.
MR. HOUGHTON: Well, and that's very true
is that things are going to happen that are not going
to be caught by inspection, and they're not going to
be caught by performance indicators.
MR. SIEBER: Yes, there's another effect
that occurs that I've seen happen in plants is you go
in with a diagnostic team that lasts three, four, five
weeks and has five to ten people on it. That really
disrupts the operation of that plant. And I think
that plant is more vulnerable during that time when an
inspection is going on from a risk standpoint.
MR. HOUGHTON: They certainly find more
things.
MR. SIEBER: They certainly do, and it
ties up management, and it ties up your engineering
staff, it ties up your licensing people, and it has to
be done, but it's a cross-cutting issue.
MR. HOUGHTON: Although they might not
have -- and I think the system, the way it works now,
does attack cross-cutting issues, because it says, "Do
I have a problem across different areas?"
MR. SIEBER: Right.
MR. HOUGHTON: Which says, "Does my
maintenance force have a problem with the Corrective
Action Program? Does my training organization have a
problem with operations experience from other plants?"
So that it does give you a feeling of whether there
are problems across different aspects of the
organization, which rolling up, to me, doesn't quite
give --
MR. SIEBER: Thank you.
MR. HOUGHTON: Other questions for me?
Appreciate the opportunity to talk to you.
CHAIRMAN APOSTOLAKIS: Thank you very
much; appreciate it.
Now, we will not need transcription after
this point. And tomorrow afternoon, actually, we'll
see you again at 1:30 when we discuss the general
design criteria. Because in the morning there is no
need for transcription.
(Whereupon, at 3:32 p.m., the NRC Advisory
Committee Meeting was concluded.)
Page Last Reviewed/Updated Monday, August 15, 2016