Plant Operations - May 9, 2001
Official Transcript of Proceedings
NUCLEAR REGULATORY COMMISSION
Title: Advisory Committee on Reactor Safeguards
Plant Operations Subcommittee
Docket Number: (not applicable)
Location: Rockville, Maryland
Date: Wednesday, May 9, 2001
Work Order No.: NRC-203 Pages 1-257
NEAL R. GROSS AND CO., INC.
Court Reporters and Transcribers
1323 Rhode Island Avenue, N.W.
Washington, D.C. 20005
(202) 234-4433 UNITED STATES OF AMERICA
NUCLEAR REGULATORY COMMISSION
+ + + + +
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS
PLANT OPERATIONS SUBCOMMITTEE
+ + + + +
MAY 9, 2001
+ + + + +
+ + + + +
The Subcommittee met at the Nuclear Regulatory
Commission, Two White Flint North, Room T233, 11545
Rockville Pike, at 8:30 a.m., John D. Sieber,
JOHN D. SIEBER, CHAIRMAN
GEORGE E. APOSTOLAKIS, MEMBER
MARIO V. BONACA, MEMBER
THOMAS S. KRESS, MEMBER
GRAHAM M. LEITCH, MEMBER
WILLIAM J. SHACK, MEMBER
ROBERT E. UHRIG, MEMBER
GRAHAM M. WALLIS, MEMBER
MAGGALEAN W. WESTON, STAFF ENGINEER STAFF PRESENT:
DAVID ALLSOPP, IIPB/DIPM
TOM BOYCE, NRR/DIPM/IIPB
EUGENE COBEY, NRR/DIPM
DOUG COE, NRR
A. EL-BANIONI, NRR/DIPM/SPSB
JOHN HANNON, SPLB/DSSA
DON HICKMAN, IIPB/DIPM
J.S. HYSLOP, NRR/SPSB
MICHAEL JOHNSON, IIPB/DIPM
PETER KOLTAY, NRC DIPM
ALAN MADISON, NRR/IIPB
GARETT PARRY, NRR/DSSA
PHIL QUALLS, NRR/DSSA/SPLB
MARK SALLEY, NRR/DSSA/SPLB
MARK STORIUM, NRR/DIPM
STEVEN STEIN, NRR/DIPM
JOHN THOMPSON, NRR/DIPM
LEON WHITNEY, NRR/DIPM/IIPB
PETER WILSON SPSB/DSSA
SEE-MENG WONG, NRR/ADIP
Agenda Item Page
Introductory Remarks . . . . . . . . . . . . . . . 4
NRC Staff Presentation
Introduction . . . . . . . . . . . . . . . . 5
Significance Determination . . . . . . . . .28
NRC Staff Presentation (cont.)
Performance Indicators . . . . . . . . . . 203
Cross-cutting Issues . . . . . . . . . . . 234
General Discussion . . . . . . . . . . . . . . . 244
Adjournment. . . . . . . . . . . . . . . . . . . 257
CHAIRMAN SIEBER: The meeting will now
come to order. This is a meeting of the ACRS
Subcommittee on Plant Operations. I am John Sieber,
Chairman of the Subcommittee.
ACRS members in attendance are Dr. George
Apostolakis, Dr. Mario Bonaca, Dr. Peter Ford, Dr.
Thomas Kress, Mr. Graham Leitch, Dr. William Shack,
and Dr. Robert Uhrig.
The purpose of this meeting is to discuss
the reactor oversight process, which today will
include the significance determination process and
performance indicators. The action matrix will be
discussed at our next meeting in July.
We had our last subcommittee meeting with
the staff on oversight processes on December 6th of
last year. Maggalean W. Weston is the cognizant ACRS
Staff Engineer for this meeting.
The rules for participation in today's
meeting have been announced as part of the notice of
this meeting published in The Federal Register on
April 16th, 2001.
A transcript of the meeting is being kept
and will be made available as stated in The Federal
It is requested that speakers first
identify themselves and speak with sufficient clarity
and volume so that they can be readily heard. I also
request that all speakers please use the microphones
to aid the court reporter.
We have received no written comments from
members of the public regarding today's meeting. I
think we should now proceed with the meeting. Mr.
Mike Johnson of NRR will introduce the topic and the
MR. JOHNSON: Good morning. Thank you.
I am -- my name is Michael Johnson from the Inspection
Program Branch. I'm joined at the table by Doug Coe,
who is also from the Inspection Program Branch. He is
the Chief of the Inspection Program Section.
And as was indicated, we have a variety of
topics to talk about this afternoon -- I'm sorry, this
morning, and spilling over into this afternoon. And
there will be a bunch of additional participants,
including representatives from the Plant Systems
I've got John Hannon, who is the Branch
Chief of the Plant Systems Branch; J.S. Hyslop and
Mark Salley, who will be talking about specific issues
of interest to the ACRS and the significance
determination process, and other participants.
So, you'll see participants cycle in and
out for efficiency purposes throughout the
presentation this morning and, again, into this
As was indicated, today's briefing really
does focus on the SDP and the performance indicators.
This is, again, a continuation in a series of
presentations that we've had, the last one being in
December where we specifically talked about issues
relating to the SDP and performance indicators.
We appreciate the opportunity to talk to
-- talk to ACRS on an ongoing basis on these and other
issues. We have, in the past, benefited from these
And, in fact, in preparing for today's
presentation, I read over the transcript from our last
meeting, and we talked about many of the issues, I
think, that are on ACRS's mind with respect to the
ROP. And we'll continue dialogue on those very issues
So, it's been a fruitful -- a fruitful
exchange for us. And we know that this fits into your
schedule -- this meeting today fits into your
schedule, along with a presentation, I guess, in July
-- Mag, is that correct --
MR. JOHNSON: Yes.
MR. JOHNSON: -- to talk about the action
matrix and getting ready for September's session with
the committee in preparation for your letter to the
Commission on the ROP. And so, we're happy, again, to
be in front of the ACRS to talk about these various
In preparing for today's presentation, we
provided background materials. One of the primary
background materials are SECY-99-007, or excerpts from
SECY-99-007, that provide a lot of the basic
information for the concept of the ROP. And we've
talked about many of those issues many times with the
In addition, Inspection Manual Chapter
0609, which is our manual chapter that talks about the
significance determination process, was provided. And
we'll spend, again, a good portion of what we do today
talking about and responding to questions on the
significance determination process.
We were able to work closely with Mag, I
think, to understand what the issues were that you
wanted us to cover. Hopefully, we've been able to
factor those into our presentation. And I know to the
extent we haven't been able to, you won't be shy in
getting us to address the issues that you care about.
Today -- next slide, Doug -- we're going
to really focus on four, specific things. First of
all, I want to just say a few words about the initial
implementation status, and that is, that overall
result of the ROP to date, to bring you up to speed
with respect to where we are.
Then, after that, we will get directly,
again, into the significance determination process.
We've got a series of examples that we want to go
through with you to help you better understand the
significance determination process and how it is being
And in addition to some examples in the
reactor safety area that Doug is going to talk about,
we specifically will cover some fire protection -- the
area of fire protection, the fire protection SDP, and
an example in that area, again to help the ACRS better
understand how we're implementing the significance
Following that, we have a topic on
performance indicators, again to respond to your
questions on performance indicators and the issues,
again to continue the dialogue on issues that we've
talked about with respect to performance indicators
and respond to your questions.
And finally, we want to wrap up with some
-- we call them selective issues, but they really are
overall topics, if you will, that don't relate
necessarily to the individual topics that we would
have hit in getting there; so, again, an agenda, I
think, that responds to the questions that we know
you're interested in.
Next slide. Let me just say a couple of
words about the overall results. We are -- we have
wrapped up the first year -- or, I should say, are
wrapping up the first year of implementation of the
Last week, as a matter of fact, Regions 2
and Regions 3 -- Region 3 conducted their end-of-cycle
reviews. The end-of-cycle is the review that happens
at the end of the assessment year in which the regions
look at what has gone on in that year with respect to
the performance indicator results, the trip
thresholds, and the inspection findings of the trip
thresholds in terms of looking at, again, what actions
the Agency took in accordance with the action matrix
and getting ready for issuance of the annual
assessment letter that provides to the licensees and
to other external stakeholders the results of the
oversight process for that particular year.
Following the end-of-cycle, there will be
an Agency action review meeting. And just to put this
-- the Agency action review meeting in context, think
of the Agency action review meeting as a revamped
senior management meeting.
Again, it is the meeting of senior
managers somewhat different from the previous process
in that this meeting really is an opportunity for
senior managers to review and provide an affirmation,
if you will, of the results of the ROP with respect to
plants that ended up with significant performance
problems, and that means, for us, plants that ended up
in the action matrix in the multiple repetitive
degradative cornerstone. So, these are plants that
have really had some performance issues.
But secondly, in the Agency action review
meeting, we talk about industry trends, and what have
industry -- what has industry trends told us about
whether we've been able to maintain safety.
And finally, we look at self-assessment
results -- the results of self-assessment for the
first year of implementation, and what are the lessons
that we've learned, and what is the feedback that
we've gotten from stakeholders, and what changes do we
need to make to the process based on those? So,
that's where we are in the process.
MR. LEITCH: Could you say again what you
call that meeting?
MR. JOHNSON: That is called the Agency
action review meeting, the AARM, the Agency action
MR. LEITCH: Thank you.
MR. JOHNSON: We believe that we've
substantially exercised the ROP during the first year
of implementation. If you were to look at the action
matrix in terms of where plants fall in the various
columns of the action matrix, we had a number of
plants, a majority of plants, in the licensing
We had plants that ended up in in the
regulatory response band. We ended up having plants
that were at degradative cornerstones. That is a
further degradation of performance. And in fact, we
had a plant -- a plant IPS, Indian Point 2, that ended
up in the multiple repetitive degradative cornerstone.
That has enabled us, because of those
cross thresholds, be able to exercise all of the
supplemental inspection procedures. We've been able
to do all of -- to do our event follow-up procedures.
I almost said "all of our event follow-up procedures,"
but I stopped myself, Doug, because we didn't have an
IIT, thank goodness.
But we've got -- we've had a wide range of
performance, and therefore, a number of opportunities
to exercise many aspects of the ROP. And we think
that's been a good thing.
We've made several significant changes
based on lessons learned to date where we found what
we believe were flaws that needed to be corrected that
we couldn't wait on.
But our intent in going into the first
year of initial implementation was to try to maintain
the process stable, if you will. And so, we held off
making wholesale changes until the end of the year
where we could do a more considered self-assessment on
what changes we needed to make.
And you'll see those changes talked about
-- being talked about in a Commission paper at the end
of the year. And again, this will be talked about at
the Agency action review meeting, and we'll brief the
Commission on those results in July. l
DR. APOSTOLAKIS: But you'll talk about
them today as well?
MR. JOHNSON: I would suggest that we talk
about them maybe in the meeting in July. We'll be
closer -- we'll have a better opportunity to have done
the roll-up of self-assessment activities.
We'll be closer to the Commission
briefing, and we can give you a better idea of what
we'll be telling the Commission.
Finally, we believe that -- and we'll --
I'll just remind you that we've talked all along about
establishing some objectives for the ROP. And you're
well aware of those because you helped us form those.
We wanted this process to be more
objective, for example, to be more understandable and
predictable. And we think that the process has, in
fact -- is more objective, and more understandable,
and more predictable, and the other attributes that
we're measuring with respect to the fundamental
objectives of the process.
And we base that on some of the data that
we've collected with respect to the matrix. We base
that on the feedback that we've gotten from internal
stakeholders, and the feedback that we've gotten from
We do continue, again, to collect data on
the ROP. We have a set of matrix, if you will, with
criteria associated with those matrix in some cases to
enable us to draw some objective conclusions with
respect to how well the ROP is meeting its intended
And we'll continue to collect that data
and make decisions based on the effectiveness of the
ROP and to indicate -- implement changes based on what
that tells us as we go forward.
So, those are the overall results of the
first year of implementation. And again, we think
we've made a fair amount of progress with respect to
implementing the ROP.
DR. APOSTOLAKIS: Now, it says there on
the fourth bullet, "successful demonstration." I
wonder what the measures of success were. I mean,
what -- what could have happened that would have you
made you declare it unsuccessful?
MR. JOHNSON: We have -- that's a good
question. We have -- but it's not one, George, I
think you want me to answer today because that would
take us -- and I think the question goes to the self-
assessment process and the matrix that we've
established, and measure the goals of the ROP.
We have established those matrix. For
example, with respect to the process being
predictable, we measure things like did we -- did we
implement the procedures in accordance with the
criteria established for them?
So, there are various criteria we've
established, various matrix to identify each of the
various goals. And what I would suggest, again, is
that in that briefing that we do in the next meeting
in July, that we come back and talk to you a little
bit about what those self-assessment measures have
told us about the various --
DR. APOSTOLAKIS: But the ultimate goal of
this is to make decisions. So, without getting into
details, have you made any decisions using this
process that would have been different if the old one
had been followed?
MR. JOHNSON: Have we --
DR. APOSTOLAKIS: I mean, is the new
process leading to more rational decisions, or better
decisions, or decisions that make the stakeholders
better -- I mean, happier?
MR. JOHNSON: Yeah, we --
DR. APOSTOLAKIS: Isn't that the ultimate
MR. JOHNSON: Yeah. In general, we have
a -- we have a good sense of comfort with respect to
the ROP in its overall ability to achieve the
objectives that we set out for it.
So now, all I'm suggesting is I can't --
I can't show you the matrix that enabled us to get
there. In fact, we're still evaluating those matrix
because, again, the year just ended.
But yeah, we believe that the process is
-- has been more objective, is more understandable.
We've gotten specific feedback that says that the
process is more understandable.
The external stakeholders tell us the
process is more understandable. The internal
stakeholders tell us the process is more
So, yeah, we believe that the process,
again, at a high level, achieves its objectives. Now,
I've got to caveat that -- and that's why I want to
have this conversation again in July -- with several
First of all, it is early. We're still
analyzing the data. Secondly, for some of the matrix,
because the matrix are new, it's hard to make a call
on things like -- one of the things that we're going
to measure, for example, with respect to measuring
whether the program meets the NRC's performance
objectives, is does the program increase or enhance
Well, that's a tough measure. We've
gotten some ways that we're going to try to measure
that. We've gotten some early bench-marking results,
if you will. But it will take a year, or a couple of
years maybe, before we can have some strong
conclusions with respect to whether it does that.
So, again, what I'd like to do is to come
back to you and talk about the self-assessment process
a little bit and the results.
CHAIRMAN SIEBER: I think one of the
aspects that licensees look at, which I think is
important and you ought to evaluated, is whether the
licensees perceive the process as being fair.
You know, there were some times, years
ago, that perhaps some enforcement action was
interpreted as not as fair as it could have been.
And it seems to me, with the structure
that you've developed here, that the chances and the
opportunities to be fair are much enhanced over what
they have been in the past. But I think you ought to
look at that process.
And I guess another question that I have,
which is really a follow-on to George's question, is
are you making more decisions or less decisions, given
the state of the industry, with the new process, as
opposed to what you would have done under the old
MR. JOHNSON: Okay.
CHAIRMAN SIEBER: Go ahead.
MR. JOHNSON: We might -- we might
actually get into -- give you a better sense as to
whether we're making more decisions or taking more
actions as we go through -- as you see the SDP
exercise, for example. We'll tell you how we come out
And we'll try to -- we'll try to give you
a sense for what -- how the -- how the old program
might have dealt with those issues, to the extent
we're able to.
But in general, we've established, in this
ROP, what is a -- what is called a licensee response
band. And that means that we've come to the
recognition and the realization that there is a level
of performance and there is a level of performance
degradation at a very, very low level that really
falls within the responsibility of the licensee to
So, and that's different from the old
process. In the old process, we would have engaged,
perhaps, on issues that fell within that level.
CHAIRMAN SIEBER: That's right.
MR. JOHNSON: Under this process, we set
aside those that are in the licensee response band.
So, intuitively, the answer is that we make -- there
are fewer interactions.
DR. FORD: Mike, I have an even more basic
question. I'm new to this, and I'm trying to learn.
Can you tell me, in two sentences or three sentences,
basically what this is all about? Are you trying to
be proactive? Are you trying to reduce bureaucracy?
What are you trying to do?
MR. JOHNSON: Certainly, I'll try in two
sentences, and, Doug, kick me if I get much beyond two
minutes. The revised reactor oversight process, the
reactive oversight process, grew out of an effort that
we took on really early 1998, late 1997, out of some
concerns that the Commission had really with respect
to how we were assessing the performance of plants and
deciding what actions we were going to take.
And at that time, we had a number of
processes, a number of different processes, in place.
The Commission was concerned about subjective they
were. The Commission had a very strong sense that --
that subjectivity shouldn't be central to our
assessment process; that we ought to be objective as
For example, the Commission was concerned
about the fact that we had -- we could be -- we could
sense conflicting and sometimes overlapping messages
through our various assessment processes.
And so, the Commission directed the Staff,
or we got permission from the Commission, to do an
integrated review of our overall assessment processes
and to develop a replacement.
Around the mid-1998 time frame, we were
talking to ACRS. We were talking to external
stakeholders about that process. And we got feedback
on that activity. And the nature of that feedback was
still very critical, not just on where we were, but
with where we were trying to go with respect that
That caused us to step back, to take a
fresh look, and that fresh look became what is the
reactor oversight process.
And in essence, what this reactor
oversight process is, is it's a -- it's a process that
starts with -- it's a hierarchical process that starts
with the notion that there's a mission. It identifies
strategic performance areas that have to be satisfied
in order for the Agency to achieve its mission.
And then, we went and identified
individual cornerstones within those strategic
performance areas, the cornerstones being the key,
essential information that if we're able to satisfy
ourselves with respect to the performance plans, we
can have confidence that our overall mission is being
And so, that's what the reactor -- how the
reactor oversight process is structured. Now, within
each of the cornerstones, we have performance
indicators that -- that is, objective things that we
can measure about the performance of the plant, that
give us information about the performance of the
We also have inspections because we
recognize that performance indicators don't -- cannot
possibly tell us everything that we need to know with
respect to the individual cornerstones.
And we take those inputs from the
performance indicators and from the inspections and we
apply for thresholds to decide whether we ought to
take, as the regulators, some increased regulatory
action in accordance with an action matrix, a
structured matrix that enables us to meter out, if you
will, what our response ought to be based on the
performance of the plants.
And we take actions based on, again, the
performance of the plant. So, that -- that's what we
-- that's what we're about with respect to the ROP.
Now, today, we're going to talk about
performance indicators, so you'll get a better sense
of what that -- how they work. We're also going to
talk about the significance determination process.
It turns out, when you do inspections,
you've got to have a way, in this objective process,
to be able to look at the findings from inspections to
decide whether they're significant and warrant us
taking some increased action, if you will, or whether
they're minor, minor in nature.
And that's what the significance
determination process does. So, you'll get a sense
for how that works also today.
MR. LEITCH: Mike, I would just say, my
perception is, too, that -- and just to amplify what
you said, is that this was an effort to make the
regulatory process more predictable, and to give
licensees an early warning of regulatory issues.
I think, in the -- in the late 90's, my
perception was that it seemed to be -- the regulatory
process seemed to be very brittle in the sense that a
plant would be going along, apparently in good
condition from a regulatory viewpoint.
And then, all of a sudden, a situation
would occur, either an operating event or some
inspection would discover some particular flaw. And
then, once that opened up, it seemed like it rapidly
spread to the plant being effectively in a regulatory
shut-down sometimes initiated by the licensee, but, in
effect, a regulatory shut-down.
So, I think the effort here -- correct me
if I'm wrong, Mike -- but my perception is the effort
here is to try to -- is to temper those actions, make
them more predictable, and anticipate declining
regulatory performance and take action before it gets
all the way to "The sky is falling; we've got to shut
this plant down."
DR. FORD: Since the utilities are
stakeholders in this, are they part of the team?
MR. JOHNSON: We have had a number of
opportunities -- provide routine opportunities, as a
matter of fact, for stakeholders, external
stakeholders, to interact with us.
And that began back in -- back in 1998, as
a matter of fact. It was sort of the watershed
workshop that cast the structure for this. The
framework of ROP was an external meeting where we had
stakeholders; we had industry; we had the Union of
Concerned Scientists; we had -- we had everyone that
would show up involved in helping us develop and get
alignment on how that process out to be laid out. And
that continues today.
DR. FORD: But they're not part of this --
these results? They weren't part of the team that
came up with these results so far?
MR. JOHNSON: They -- well, we -- how can
I explain this? I'm trying to be very brief, and not
-- and not take a lot of Doug's time. We have --
we've had a number -- as we implement the process,
there are a number of opportunities for external
stakeholders to remain involved.
For example, when Don Hickman talks about
performance indicators a little bit later on, we're
going to talk about, for example, the fact that some
of the performance indicators caused -- that is a
reporting criteria that caused licensees to raise
questions that require some interpretation.
Well, in resolving those questions, those
scruply-asked questions we call them, we have a
monthly -- about a monthly meeting with the NRC and
the industry, attended by NEI. And it's a public
meeting where we take on those individual issues and
work to agreement on the decisions with respect to who
we should interpret the criteria or whether, in fact,
we ought to change those reporting criteria to address
That's an example of sort of the ongoing
interchange -- exchange that we have with external
stakeholders in implementing the process.
And so, they are -- the industry is. I
mean, when we talk about the results of the process,
we're going to tell you -- in July, we're going to
tell you how we've implemented the process from an
But we're also going to tell you how we
think that process has impacted the performance of the
industry. So, it's hard to separate the two.
DR. BONACA: Yeah, just one comment I
would like to make; it was simply there is an
impression almost that there was no significant
determination prior to this system.
There was, and the significance was based
on the degree of compliance. And today, the
significance is an elimination process based on risk.
That's really the big shift there, okay?
So, compliance, alone, is not anymore
material. I mean, typically -- I mean, if you had a
finding, nobody very much looked at, you know, is it
significant from a safety standpoint?
It was, you know, how far are you from
compliance within the acceptable regulation? And that
really was the basis for determination of
DR. APOSTOLAKIS: This Agency has been
accused of, in some past instances -- it's
overreactive. Would this process help us not to
overreact in the future?
MR. JOHNSON: Doug, do you want to take
that? In fact --
DR. APOSTOLAKIS: What did you say, Doug?
MR. COE: Yes, you're exactly right. It
helps us not to overreact, and it helps us not to
under-react. We want, as Mike indicated earlier, a
consistent and more predictable process. And I think
that your points were right on.
I think that the prior process, although
there was an attempt to be thoughtful and to be
consistent, it was more subjective. And over time,
there were differences that arose as to how we reacted
to various things, either under-react or overreact.
And so, this was the essence of the
concern that ended up where we are today.
DR. APOSTOLAKIS: Okay.
CHAIRMAN SIEBER: And in fact, that gets
back to the statement that I made earlier about the
perceived fairness of it all and -- which, to me, is
a very important aspect of what it is you're doing
MR. COE: And people say fairness is
predictability and understandability --
CHAIRMAN SIEBER: Predictability and --
MR. COE: -- transparency --
CHAIRMAN SIEBER: -- consistency.
MR. COE: -- and consistency, yes.
CHAIRMAN SIEBER: Right.
MR. JOHNSON: Now, you might -- you might
remember Chairman Jackson's -- one of Chairman
Jackson's favorite words was "scrutability". And we
think this process goes a long ways towards helping us
be very clear about what the issues are, what are
determination of those -- the significance of those
issues is, and how we got to where we end up with
respect to what actions we ought to take.
So, we think the process -- and again,
that goes back to one of the key things that we're
measuring about the process, what we think the process
Okay, that's what I was going to talk
about under "overall results". Now, Doug is going to
start the SDP discussion.
MR. COE: Thank you. Just building on
what we just have talked about, the SDP is necessary
to characterize the significance of inspection
findings as one of two inputs to the action matrix;
the other being the performance indicators.
And the scale that was -- we tried to
achieve with the SDP is intended to be consistent with
the scale, the threshold scale, that is used for the
performance indicators and when -- and when we take
certain responses, based on those performance
It started with the application of risk
insight and risk thinking from a reactor safety
standpoint. But as you'll note, we have seven
cornerstones, some of which, in the safeguards area or
the occupational or public radiation health area, of
the emergency preparedness area, may not have a direct
link to a core damage frequency risk matrix.
And in those cases, we still have an SDP
because we still need an SDP to characterize
inspection finding significance so that it can feed
the assessment process.
And we try, in those cases, to make a
consistent parallel with the risk matrix of the
reactor safety SDP in order that the response that the
Agency would give to particular inspection findings is
consistent across cornerstones.
That's a more subjective judgement, and
it's one that we're -- you know, as we get more
experience, we continue to refine.
Today, we're going to talk about the
reactor safety SDP because we understood that this was
your primary interest. And so, what I'm going to show
you are -- actually, I've got four examples, two of
no-color findings -- and I'll explain what conditions
arise, or what circumstances arise, that we would not
colorize a finding -- one green finding, and one non-
These are all real examples. In fact, for
three out of the four, I basically cruised our website
and plucked those three examples, the first three that
you'll see, right out of our website. And I've
referenced the inspection report numbers if you care
to look further.
The fourth one, the non-green finding, is
also a real example, but it hasn't been published yet.
So, I've sanitized in terms of its -- the description
of what -- what the plant was and so forth. But it
was, in fact, a real example.
So, we'll get on with the first example.
The no-color finding category are findings which don't
affect the cornerstone, or which have extenuating
circumstances. These are the two primary categories
of no-color findings.
The decisions on whether to colorize the
finding or not is made prior to entry into the SDP.
It's -- the guidance that governs that is Manual
Chapter 0610*, and there's a series of questions that
I'm going to try -- I'm going to show you,
kind of at a high level, how those questions result in
a no-color finding.
The first example that I've got here was
an inspection procedure that asked the inspectors to
look at licensee LERs. And the finding that was
reported in an LER was the missed surveillance test
for the control room oxygen detector.
Now, the guidance in 0610* is that if it
-- if a finding does not affect the cornerstone and,
therefore, cannot be process by an SDP, then it is
documented as a no-color finding.
In the example that we have here, the
cornerstones in the reactor safety area are initiating
events, mitigating systems, barriers, and emergency
preparedness. And that's under the reactor safety
So, looking at that particular finding,
the lack of a -- or the failure to do a surveillance
test for a control room oxygen monitor, when looked at
from the standpoint of does it -- does it actually
affect the cornerstone, and would it -- would it be
possible to evaluate that finding through the SDP
process using delta-core damage frequency, or delta-
large early release frequency as the matrix.
The answer would be no, and that's what
came out of the 0610* lodging.
DR. KRESS: Doug, what's the purpose of
that oxygen monitor?
MR. COE: The purpose of the oxygen
monitor, I would presume -- and I can't say for sure,
but based on my general understanding -- is that
oxygen monitors are there in case the control room is
enclosed, becomes enclosed, sealed, through, you know,
control room isolation functions.
And therefore, then there's a --
DR. KRESS: A chance of depleting --
MR. COE: -- monitoring process -- a
monitoring instrument that, then, would tell the
operators that they were getting dangerously low
DR. APOSTOLAKIS: But then, it seems to me
that it would be under the broad category of reactor
safety, would it not?
DR. KRESS: That's what I was wondering.
MR. COE: You could say it could be under
the broad category of reactor safety. The next
question you could ask is how would you characterize
it was significant?
If you're looking at delta core damage
frequency or delta large early release frequency,
there's really no -- there's no connection there.
DR. APOSTOLAKIS: Well, I would say that
its contribution to the facility is really negligent.
I mean, that's probably a more accurate statement.
And one does not need to do an analysis to
MR. COE: I wouldn't necessarily disagree.
But what we're trying to do is come up with guidance
that does produce the right results. And in fact, in
very early stages of the development of this process,
the criteria that we're discussing here on how to
color -- how to choose not to color an inspection
finding didn't exist.
And there was a -- there was a thought
being given at that time that there were -- if you
couldn't meet the threshold for greater significance,
it would be a green finding. And we would basically
have a lot of green findings, okay?
Now, somewhere along the way, in the
development of this process, it was decided that
having a no-color category would be useful, I think
initially because of the extenuating circumstances
that we're going to talk about in a minute.
DR. APOSTOLAKIS: But in this particular
case, if we would go back to the previous -- slide
five -- no, this is six, yeah --
MR. COE: Yes.
DR. APOSTOLAKIS: -- the definition says,
"findings which do not affect the cornerstone or which
have extenuating circumstances." It seems to me
saying that the finding does not affect the
cornerstone is too strong.
I mean, has a negligible impact; I think
that's more accurate. Maybe that's what you mean by
"does not affect." And when you are elaborating on
it, you actually -- that's what you said, that, you
know, calculating that was really a waste of resources
for this particular case.
MR. COE: Yes.
DR. APOSTOLAKIS: And the end result is
known in advance. It's going to be very, very small.
But it does fall under the cornerstone of reactor
DR. SHACK: But what is the question the
man asked to make that decision? Does he say, "Does
this affect -- is this going to affect the initiating
events?" What are the actual questions he asks
himself so he comes up with that answer?
MR. COE: Well, those are questions that
are articulated in 0610*, Appendix B. And for reactor
safety cornerstones, not including emergency planning,
they include the following questions: Could the issue
cause or increase the frequency of an initiating
event? That's the first question.
The second question is, could the issue
credibly affect the operability, availability,
reliability, or function of a system or train in a
There's four questions. The third
question is, could the issue affect the integrity of
fuel cladding, the reactor coolant system, reactor
containment, or control room envelope, the integrity
of those things?
And four, does the performance of the
issue involve degraded conditions that could
concurrently influence any mitigation equipment and an
In other words, could you -- could you
affect the likelihood of an initiating event at the
very same time with the -- with the same issue that
you would degrade a mitigating function?
So, those are the questions that are
asked. And I don't disagree that --
DR. APOSTOLAKIS: Doug, let me -- I think
communication and using the right words are very
important whenever you do things like this. I mean,
we found that out and PRAs and so on.
Instead of saying that we will not do --
I mean, we screened things out in the PRA repeatedly,
and nobody objects, okay? And nobody has come back
and said, "Gee, you know, you really missed it, after
25 years of experience."
Instead of saying we're not going to do
it, maybe a better way of saying it is that a crude
evaluation shows, or a conservative evaluation shows,
that the CDF is negligent.
That sends the message that you have
thought about it; you have evaluated it. You have not
made the decision in advance not to evaluate it; which
I think you are evaluating in some sense, you just
don't want to spend too much time on it because, you
know, professional judgement evidence shows it's not
going to make any difference.
So, I think sending the message in a
different way is probably better.
MR. COE: We can take that comment because
it gets at a discussion, a dialogue, that has occurred
since this guidance was formulated. And there are
persons on the staff who feel much the same way you
I would say that there are other examples,
and perhaps this isn't the best example, but, for
instance, a finding which involves a missed
surveillance test, which then, the surveillance test
is subsequently performed and found to be acceptable.
Now, was there an impact on the
cornerstone? Was the cornerstone functioned -- were
any of the characteristics or attributes in the
cornerstone for mitigating systems affected?
Well, the answer would be no, not at all
in that case. So, maybe that's a better example of a
finding which doesn't affect a cornerstone. And to
try to define that threshold, does it or doesn't it,
is somewhat subjective, I would have to say.
DR. APOSTOLAKIS: I understand. And I
think -- you know, I think you understand the spirit
of my comment. But I have another question. If the
finding does not affect the cornerstone, and you guys
have declared that these are the things you care
about, why bother?
MR. JOHNSON: I'm sorry, why --
DR. APOSTOLAKIS: Why bother to look at it
MR. COE: Typically --
DR. APOSTOLAKIS: You know that --
MR. COE: Typically they are violations,
that -- for example -- oh, I don't know; we've got
violations of, like I say, missing surveillance tests
or of other administrative regulatory requirements
that can't really be processed through the SDP.
And in fact, one of the reasons why the
no-color finding category came into being in the first
place was to assess whether or not these findings that
could not be processed through the SDP warrants a
significant determination process of their own.
And this question has come up in a number
of areas, such as -- and most particularly, I think,
in the cross-cutting areas, human performance issues,
where mistakes are made, or errors are made, in the
cross-cutting areas of performance -- problem
identification and resolution.
So, you know, it's a broad category of
things that we find cannot really -- don't really --
don't really comport with an SDP that's been created.
And we really can't make a link to core damage
frequency changes or delta alert changes.
So, we're left with this set of findings
that may be regulatory issues, may be regulatory
violations, that we're not sure what to do with. So,
we put them in this category.
DR. APOSTOLAKIS: I thought we were trying
to get away from that.
MR. JOHNSON: We are, George. Let me give
you an example -- let me give you another example that
perhaps adds to the examples that Doug has given that
were very good.
One of the things that you'll recognize
that we ought to care about, as a regulator, that may
not have a direct impact on the cornerstone, as we've
been able to measure in terms in terms of the results
of an inspection finding, is something, for example,
that would impact the regulatory process or our
ability to -- to effectively regulate the performance
of the licensee.
For example, let's suppose -- and this is
a scenario that we've come -- we've had some concerns
with respect to performance indicators. And that
would be, for example, a situation where a licensee
inaccurately reported a performance indicator, you
know, or let's -- the example where there were some
willfulness, a violation that was willful in nature.
And maybe that would have a -- maybe there
would be an element of that that would have an impact
on the plant, that you could run through an SDP that
would have an impact on the cornerstone.
But the willful nature, or the inaccurate
reporting, or you know, those kinds of issues are also
issues that, again, when you look at the questions and
the things in 0610*, the excerpt that we just handed
you, are not things that you necessarily get through
an SDP on, but things that we ought to care about as
the regulator. Those are other examples.
DR. APOSTOLAKIS: But you care about them
because they're supporting things that need to be
MR. JOHNSON: Absolutely.
DR. APOSTOLAKIS: -- because they affect
MR. JOHNSON: Or they eventually --
DR. APOSTOLAKIS: Could affect -- could
MR. JOHNSON: Could affect the
DR. APOSTOLAKIS: Good, good.
DR. SHACK: Let me come back -- I didn't
like the answer to the surveillance test one because
this one -- you know, this one, if I answer these four
questions, if the thing failed the surveillance test,
I think I would have answered the four questions in
the same way.
You know, when it comes to you, a
surveillance test, and you say, "Okay, it was an
important component; I missed the surveillance test.
But when I did test it, it was okay, and it had no
impact," that's an answer I don't think I like because
that tells me I got lucky.
You know, if I can't -- you know, it seems
to me these things should be hypothesized. You know,
if I missed a surveillance and if the surveillance
test was negative, then I could still answer these as
If I missed a surveillance test and it
happened -- you know, the thing that has always
bothered me about these things is everything is going
to be green until something really happens.
You know, yeah, it's no problem if you
miss a surveillance test as long the thing is working
well. You know, I either find out the thing is not
working when I need it or in a surveillance test.
MR. JOHNSON: Well, and I know Doug has
got a perfect answer for this, but let me just cut in
with my less than perfect answer, and then he can
You know, when we say -- when things make
it through the findings threshold; that is, they are
more than minor, we're not -- we're talking about, in
every case, something that we want the licensee to do
No-color findings are not -- for example,
a missed surveillance test or, you know, anything that
we documented in the inspection report as a finding or
as a green finding, you know, a finding on very low
risk significance, are all issues that the licensee
needs to correct.
It's not that we're setting them aside,
that they can -- that they can have the option of
doing nothing with. They've got to put them in their
corrective action program, and we look at their
corrective action program as part of our periodic PI&R
-- PI&R, problem identification and resolution
inspection procedure, to make sure that they're doing
something with those issues.
So, we're not -- we're not setting them
aside, but they are clearly less significant than in
a situation where you would have had, say, a missed
surveillance test found -- that a surveillance test
was found that there was a problem with that
And that component, when you go back and
you look and see, there was a condition that was
brought on by some issue that happened a long time
ago. And so, you can really look at how long that
particular situation existed.
MR. LEITCH: So, why do you take --
DR. APOSTOLAKIS: So, why --
MR. LEITCH: Excuse me. I was just going
to say, could you take me through the line of
reasoning that would apply? Here's the licensee that
missed one surveillance test, and this is the only one
he has missed in a year, versus another licensee that
has missed ten surveillance tests in a year.
And every one of those goes through the
analysis, and every one is no-color. Is there some
kind of a trending? How do you -- how do you deal
with that, or would you like to deal with that?
MR. JOHNSON: The way we do that is -- and
Steve Stein, is he wondering around the audience?
Make sure your ears perk up for this. The way we do
that is, if we have a finding, and that finding is
more than minor -- I'm sorry, a finding is more than
minor, and we're documenting it in the inspection
If there is some cross-cutting element of
that finding, we document that in the inspection
report. And when I say "cross-cutting," I mean things
like -- the cross-cutting issues are things that are
-- have impact on whether the licensee has a good PI&R
system, problem identification and resolution system.
If they're human performance in nature, if
they're human performance things that are going on,
that are, again, cross-cutting, and if there are
safety conscience work environment issues that are
going on -- different from safety culture -- safety
conscience work environment -- by that, we mean is
there something that is indicative of there being a
chilling effect, if you will, a hesitancy on the part
of the plant staff to raise issues. Those are cross-
Well, if we have a finding in an
inspection report, and there is this cross-cutting
nature to it -- performance, problem identification,
safety conscience work environment, those get
documented in the inspection report.
And as a part of our problems
identification and resolution inspection, we -- today,
on an annual basis -- and we're changing the period as
to that a little bit, and making some other changes
that we think improve that inspection.
But we look at those issues, the
collection of those kinds of issues, to see if that
tells us that the licensee has what we call a
substantial -- a trend, an average trend with respect
to substantial problems in this cross-cutting area.
And we document those in the assessment
letter. We talk about those with licensees to get
licensees to get them resolved.
CHAIRMAN SIEBER: Does that mean that
you're actually doing a bean count as you go through
the period of missed surveillance and other kinds of
things that licensees do that cause a non-cited
violation because you can determine whether a cross-
cutting issue is there or not?
MR. JOHNSON: I wouldn't say -- I wouldn't
use the word "bean count". In fact, the Commission
was very careful with us to give us -- the Commission
told us to be very careful with respect to how we --
how we treat issues that are green.
The Commission was concerned that we would
take a collection of -- we would count greens, things
that have a very low risk significance --
CHAIRMAN SIEBER: Right.
MR. JOHNSON: -- and we would somehow
amalgamate them, if you will --
CHAIRMAN SIEBER: Right.
MR. JOHNSON: -- and roll them up into
something and make a big splash.
CHAIRMAN SIEBER: Now, that's the old
MR. JOHNSON: Right, that was the old
system. But we think it's very --
CHAIRMAN SIEBER: And you can always write
a finding against your QA program.
MR. JOHNSON: Exactly.
DR. BONACA: Let me ask you a question
more specific. Go to the next slide, if you could.
Look at disposition of finding, "confirmed entry into
the licensee corrective action program." I mean, we
come back to this, as we came back before.
Here is what -- are you abandoning the
issue once it's in a corrective action program, or are
you looking for how timely they're going to address
the issue, and whether or not this is a repeat issue?
I mean, these are two fundamental elements
of the corrective action program. And that answers a
lot of questions. If you say, yeah, we're going to
count it, and we keep an eye on that, then I am
comfortable with this.
MR. JOHNSON: That's exactly what we're
DR. BONACA: Okay.
MR. JOHNSON: We're saying that we --
DR. BONACA: So --
MR. JOHNSON: In this PI&R inspection,
that's exactly what we do; we go look at what is in
the corrective action system. We ask ourselves, you
know, is the licensee dealing with issues? Are there
issues there that are significant that the licensee
hasn't dealt with; you know, are there -- are there
Steve, do you want to -- now is a good
time for you to jump in.
MR. STEIN: Steve Stein, Inspection
Program Branch; I just wanted to clarify one point on
that previous example. What made it a no-color
finding was that -- was the equipment, was the control
room oxygen monitor, not the fact that it was a missed
The missed surveillance on a mitigating
system, on an injection valve, or pump, or on the EDG,
would fall within a cornerstone and would go through
And all the conditions associated with
that could make that more than of very low
significance. So, that's the point I wanted to make,
that what made that no-color was the equipment, not
the fact that a surveillance was missed.
MR. JOHNSON: All right --
DR. APOSTOLAKIS: Which seems to me
supports what I said earlier, that you really need a
conservative analysis as to be in your mind, and
dismiss it, which I think is perfectly all right. I
mean, that's how we do all these things.
MR. JOHNSON: Yes, because --
DR. APOSTOLAKIS: If we have a problem
somewhere, and we analyze it, and we find out the
delta CDF is delta less -- or smaller than something,
then that's a green. So, green is good.
MR. JOHNSON: No, no. Green is not good.
Green issues are still issues that we think the
licensee needs to do something with.
DR. APOSTOLAKIS: So, if the number of
scrams is smaller than the number you specified, which
is good, you give the guy a green, don't you?
MR. JOHNSON: With respect to -- okay, we
were talking about inspection issues. Now, green with
respect to the performance indicators means that that
performance is in the expected range, sort of this
nominal range, of licensee performance.
DR. APOSTOLAKIS: Right.
MR. JOHNSON: So, in that case with
respect to scrams, yes, scrams --
DR. APOSTOLAKIS: But green cannot mean --
I mean, is one of them light green and the other is
MR. JOHNSON: They're green.
DR. SHACK: Well, but still, no color
does, in fact, highlight the fact that it's green. I
mean, there's a difference between no color and green.
DR. APOSTOLAKIS: Exactly, that was my
question. Why bother? Why not declare this a green?
MR. JOHNSON: We --
DR. SHACK: What's the difference?
MR. COE: That's a dialogue that has
occurred on an ongoing basis within the Staff.
MR. JOHNSON: And in fact, this was an
issue that we talked about. We recently had an
external lessons learned workshop to roll up the
results of the first year of implementation and to
talk with the industry and other external
And this issue of no-color findings was
one that we talked about, and the kinds of concerns
when something like -- you know, you've got a system
that uses colors. We can understand the significance
But here are these findings that you don't
assign a color to because you say they're outside of
the cornerstones. And what's the significance of
There seem like there are a lot of those,
perhaps. We should really do something with no-color
findings. And in fact, we went into that workshop
with a proposal that we were going to turn those no-
color findings into greens.
And what do you think the industry's
response was? The industry said, "Don't make those
things all greens. We care about greens just like we
care about everything else."
And so -- and so, the dialogue continues
with respect to how to treat these issues.
DR. APOSTOLAKIS: Okay, so let's put in a
different way then. Why don't we go to the
performance indicator for scrams; and if you are below
the appropriate threshold, that is a no-color finding
instead of green?
MR. COE: That's not a finding.
DR. APOSTOLAKIS: It's the same rationale.
MR. COE: You see, that's not a finding,
George. That's -- you've got performance indicators,
which are just data collection, and then you've got
findings which are actual -- some kind of deficiency
The licensee's performance was deficient
in some respect, and that was the source of the
DR. APOSTOLAKIS: So, "finding," you're
using it in the regulatory sense?
MR. COE: That's right.
DR. APOSTOLAKIS: Again -- I don't know,
this no-color business is not very comforting.
MR. JOHNSON: We agree, we agree. But
again, having said that, there are issues -- you can
get yourself to the point where you can find issues
that when you try to treat them through the SDP, you
But when you ask yourself the group three
questions in that excerpt that we handed you, if they
still are things that we ought to be concerned about
as an Agency, then those are things that are no-color
DR. BONACA: Actually, I think, you know,
in part is the issue that -- if you look at the
specific of this, you know, for it to be significant,
you would have to have a significant event: a
release, a problem with the control room, the need for
oxygen in it, and then find that you don't have it
because you didn't monitor it right.
So, the risk, in itself, is minute. And
yet, there are thousands of activities where
compliance is important because without compliance,
you don't have the assurance that in case yo have that
kind of residual event that happens, you can deal with
And I think that somehow we have to still
deal with this thousands and thousands of compliance
issues. And so, I'm trying to understand how -- I'm
not disagreeing at all with you, George.
I'm only saying that they still are
significant individually because if you don't maintain
some level of significance applied to them, well, you
would have literally collapse of this commitments; I
mean, particularly --
DR. APOSTOLAKIS: Well, I think there are
DR. BONACA: -- what people are going to
say, "Well, likelihood for it to happen is so remote,
why should I" -- you know?
DR. APOSTOLAKIS: It seems there are two
issues, Mario, that have been raised here. One is the
consistency of the approach, the self-consistency.
DR. BONACA: Sure.
DR. APOSTOLAKIS: In other words, when we
do something with the PIs or the significance
determination process and we declare something to be
no-color or color of this, we have to be self-
The second is I appreciate that, you know,
we want to have a higher degree of confidence by doing
-- having these additional requirements. But I
thought over the last four or five years, we were
trying to move towards a risk-performance based
system, which is different in spirit.
So, I don't know how -- I mean, it seems
that we are still worried about things that are --
admittedly, the risk is insignificant.
DR. SHACK: But I think their four
screening questions are very good. You know, they
seem to me to, you know, have answers that are
scrutable and, you know, do a first cut at coming up
with those questions in a way that an inspector, I
think, has a chance of dealing with them.
DR. BONACA: Absolutely, and look at the
disposition of this; it goes into the corrective
action program. I mean, it simply says "Do it." And
the only activity is they want to monitoring if the
corrective action program works. So --
DR. APOSTOLAKIS: If I look at the action
matrix -- which we're not going to discuss today --
but if I look at it, I think if I have a green finding
someplace, do I ask them to do something?
MR. JOHNSON: You're not --
DR. APOSTOLAKIS: No, no, it has to be a
licensee corrective action. When do I involve the
corrective action program and ask them to do
something? It has to be white?
MR. COE: The action -- well, no.
MR. JOHNSON: The problem identification
and resolution inspection that I talked about happens
as part of the baseline, happens for every plant,
When the action matrix gets invoked is
when thresholds are crossed. So, if you had a white
issue, then you'll see -- you'll see that you change
DR. APOSTOLAKIS: Yes, so, nothing green;
you don't do anything when it's green?
MR. JOHNSON: Right, but we'll talk --
we'll go through the action matrix.
MR. COE: Remember, a licensee that has
findings and performance indicators in the green range
is considered in a licensee response band. That's the
characterization we've given it from an assessment
point of view.
So, what is it, 80 percent of the plants,
or whatever it is, are in the licensee response band,
and we expect them to deal with the lower level
issues, the ones that we don't feel a need to engage
And so, their corrective action program is
expected to correct those lower level issues before
they become manifested in larger issues. And their
motivation to do that, of course, is to -- is to
continue to be treated in the licensee response band;
that is, not get extra NRC attention, and inspection
effort, and activity.
DR. APOSTOLAKIS: Now --
MR. JOHNSON: Before we leave, can I also
just add to you --
DR. APOSTOLAKIS: We're not going to
MR. JOHNSON: -- add that when we talked
about this issue, you know, we really thought the
stakeholders who would be most concerned with no-color
findings would be members of the public.
But in fact, Dave Lochbaum, who was -- who
was there, didn't really share our view. He didn't --
he wasn't all that concerned about no-color findings,
to be all that honest.
And maybe it's because when we started off
the year of initial implementation, we had -- we had
a number of no-color findings. But that has gradually
decreased as we were able to get out guidance with
respect to these screening questions.
And so, the numbers really are -- and I
don't want to leave you with the impression that there
are a lot of these things going around. There truly
And I think this may be a concern that we
were more worried about than either the industry or
others, like stakeholders.
DR. APOSTOLAKIS: You see, Mike, one of
the -- I am concerned on self-consistency. You have
an example later where an inspection finding led to
MR. JOHNSON: We have one, yes.
DR. APOSTOLAKIS: Okay. Now, inspection
finding, by definition, is -- Doug just told us is
some sort of violation somewhere. You forgot
something; you did something incorrectly.
MR. COE: It could be a violation or it
could also be some kind of deficient performance that
was not a violation. That's fundamentally it, but
which contributed to an increase in risk, for example.
DR. APOSTOLAKIS: And you declare that as
a green. On the other hand, when it comes to
performance indicators, green means expected
performance, you just told us. Isn't there an
MR. COE: Yes. Actually, in that respect,
there is. Both the -- well, the performance
indicators include performance that we expect to occur
as well as, in some cases, that which we don't expect.
For example, unavailability performance indicators in
the reactor safety cornerstone have a component of
unavailable that occurs due to normal, routine
maintenance, which is acceptable, as long as it's
performed under the maintenance rule guidance.
And then, there might be additional time,
exposure time, of unavailability of equipment that's
due to some kind of deficiency that is, then, added to
that performance indicator. So, that's a particular
performance indicator where you've got a combination
of poor performance contributions to that indicator,
as well as acceptable performance.
But fundamentally, you know, you're right.
An inspection finding is always associated with some
performance issue; a PI may not be.
DR. APOSTOLAKIS: But ultimately, the
inspection findings feed into the action matrix too.
MR. COE: Yes.
DR. APOSTOLAKIS: So, now, it seems that
the green means something different for those two, and
we have different questions for the whites and the
MR. COE: Well --
DR. APOSTOLAKIS: -- which presumably will
mean something different too.
MR. JOHNSON: But George, with respect to
the action that we take as an Agency, there's really
no difference. If you have -- if you have a
collection of only these kinds of findings that we've
been talking about, they end up in the licensee's
corrective action system. The licensee takes actions
to address them.
We periodically go out and look, and
that's it. The licensee -- and the licensee -- the
performance is in the licensee response band, and so
our actions are to do the baseline inspection.
If a plant has a scram or, let's say, two
scrams in 7,000 critical hours, and the threshold is
there scrams for 7,000 critical hours, once again,
from a regulatory perspective, we're not doing
anything. The licensee is in the licensee response
Now, if a licensee doesn't get
increasingly concerned as they get close to that
threshold, we think that's a problem. But again,
we're not going to engage because the licensee -- the
plant is in the licensee response band.
It's only when they trip that threshold
that that is that deviation from nominal performance
to the white for those performance indicators, or we
have an SDP result that is white in any of the -- in
any of the SDPs.
It's that -- it's that that gets us to
increased action, based on the action matrix. So, I
-- now, they --
DR. APOSTOLAKIS: I guess --
MR. JOHNSON: -- they come together in a
way that is consistent.
DR. APOSTOLAKIS: Well, we're going to
have another subcommittee meeting to discuss the
action matrix, so maybe a lot of these questions will
come back when we do.
DR. SHACK: We're beating this to death.
You just cut my --
DR. APOSTOLAKIS: It's already comatose.
DR. SHACK: I would have thought you'd had
zillions of no-color findings. But what's really
happening is the inspector is not going out and
zinging them for things that -- I mean, he's providing
another level of screening before he even asks the
four questions, rather than playing gotcha.
Because I can't believe an inspector
couldn't go through a plant and just keep writing up
everything under the sun if he had a quota of
citations that he had to fill.
MR. COE: That's right. And the NRC has
enforcement guidance on what constitutes minor
violations. And we've actually tried to incorporate
on that and expand on it a little bit in this process.
So, the type of thinking that you're --
that you're thinking about now, that the inspector
does, is actually -- we've tried to formalize this in
this guidance. I'm just not discussing it right now.
But it's -- it is there.
DR. SHACK: So there really is even
another level --
MR. COE: Yes.
MR. JOHNSON: Yeah. In fact, they are --
we haven't talked about them, but they're the group
one questions. If you look at the hand-out, there are
some group one questions that really help the
inspector try to distinguish what is truly minor.
And I guess we don't -- I don't want to
take Doug off and have him go through those. But yes,
there is screening even before that.
MR. COE: But the overall objective is to
get inspectors to be sensitive to the things that are
potentially risk-significant, the things that are
potentially significant in the other cornerstones.
And we've given them the -- the yard stick
of the SDP is to help to define that in a much better,
clearer way than we have in the past and, you know,
towards the goals of objectivity and consistency.
So, let me pursue this now because the
next example I've got on no-color findings is actually
in the other category, which is, perhaps, maybe, a
little bit more clear, less subject to, you know,
dialogue and debate.
That is that that kind of a finding, which
as extenuating circumstances -- and we define
"extenuating circumstances," in our guidance. But
principally, it's issues that may involve willfulness
or issues in which the regulatory process is impeded
because certain information which was required to come
to us did not.
Okay, and in this particular inspection
finding, the licensee submitted an application for
operator license -- I believe that -- yes, an operator
license application was submitted. And it was -- it
incorrectly stated that certain training had been
So, we were about to act on a license
application to give an exam to an operator, and the
information that we had was incorrect. The operator
had not received the training that the license
application stated that he had.
Okay, per our guidance, this is a finding
that potentially impacts our ability to perform our
function, since we have been given information that's
And in that case, if the impact of that
can -- you know, does not affect the cornerstone; and
in this case, it clearly did not because we caught
this before the license -- before the operator was
examined and put on-shift, then it's a no-color
So, again, it's exactly as before. We
confirm that the licensee entered that into their
corrective action program, and then we treat it as a
DR. UHRIG: Was this just an accident, or
was it an error, or was this deliberate?
MR. COE: Well, I can tell -- I can
certainly say that it -- our assessment was that it
was not deliberate, okay? Because if it was, it would
have been captured in a different -- in a different
In fact, willfulness, in many cases I
think what you would expect to see, not just as a non-
cited violation. We would probably examine it for
enforcement action above the non-cited level, as a
severity four or a three, or higher. So --
DR. UHRIG: Usually, most of these things
are errors somewhere along the line.
MR. COE: Yes, yes. But when we -- when
they -- when we find them, we have to have a process
to deal with them --
DR. UHRIG: Yes.
MR. COE: -- and to deal with them in an
appropriate way. And again, I think that if the -- if
the process is set up to disposition lower
significance items, or findings, it allows the
inspectors to do more -- to spend more effort, on
areas that are potentially of greater significance.
And that's the intent.
MR. LEITCH: Does it enter into your
decision at all whether the item has been already
entered into the licensee's corrective action program.
Like say, for example, this issue here, say in --
before it comes to your attention, say the licensee
has reviewed the matter and said, "Oops, we found a
glitch in our training program. This fellow didn't
really get this training," and they put it into their
corrective action program, would that -- would that,
in any way, affect this? Might this then --
MR. COE: Yes, actually --
MR. LEITCH: -- drop off the -- drop off
the -- and not even be considered a finding?
MR. COE: Our guidance does not provide
for inspectors to, what we call, mind the licensee's
corrective action programs, except in one specific
case, and that's our periodic problem identification
and resolution inspection procedure.
MR. LEITCH: Right, right, yes.
MR. COE: And there, we send in a team of
inspectors on a periodic basis to do just that. And
we look at the corrective action program, the items
that are in there, in a -- you know, we try to look at
them in a risk-informed way or, you know, looking for
the items of greatest significance and looking for
trends and patterns and that sort of thing.
But the findings that come out of that
have to be linked to the SDP in terms of their
significance. In other words, we -- again like Mike
said earlier, we're not allowed to go in there and
aggregate things and then make them -- if they're all
green issues, or would be green issues if we put them
through our SDP, we could not make a bigger deal out
of that than the most significant of those findings
DR. UHRIG: Suppose that this had happened
before, had been put in the corrective action program,
and it failed, and now it's showing up again. How do
you -- what is the impact of this having happened
MR. COE: Well, if this is a repeat --
DR. UHRIG: Yes
MR. COE: -- that you're talking about, a
repeat kind of condition, the philosophy that we're
operating under is that licensees should be correcting
these things at a lower level; and that if they don't,
if this continues to repeat, and if the source of the
continuation of this repeating problem is a more
fundamental issue associated with their management
controls or what-not, that we would expect ultimately
that we would have inspection findings and/or
performance indicators that would cross the threshold
from green -- from the licensee response band -- into
white in which we would engage.
DR. UHRIG: But it would be in the
corrective action program, not here?
MR. COE: Yes, it would -- if their
corrective action program was not functioning, we
would expect, over time, to see these kinds of issues
manifested as higher significance issues.
If the licensee was doing a good job, and
maybe, you know -- they're managing at a lower level,
clearly. They're trying to keep the -- you know, the
problems at a low level.
And the real question is, for us as an
Agency is, has the threshold -- I mean, we're going to
allow -- and I think somebody said earlier, we're just
waiting for things to happen.
Well, we're not really because what we're
doing is we're trying to define a threshold so that
when things happen of a certain significance, we
engage the licensee. And the intent is, is that we
engage the licensee at a level before a significant
impact to public health and safety occurs.
So, when a licensee issue comes up that's
greater than green, it goes into the white region for
instance, then we engage at a certain level. We
expect that that was -- is still not a significant
impact on public health and safety.
And we are going to engage at that level.
We consider that an early engagement as the problems
have now departed from the licensee response band, and
now they're in the regulatory response band. So,
we're going to get involved.
MR. JOHNSON: Good, good. I just wanted
to add one thing to make suer that we leave you with
the right impression. If, for example, an inspector
comes across an issue, they do not not make an issue
an issue because the licensee already found it.
If it's an issue, we set aside whether the
licensee found. We look at that issue and treat that
issue in our process.
Now, when we go out and we do our
supplemental inspection, in the case where issues have
crossed thresholds, then is when we would recognize
what the licensee has done with respect to finding the
issue, and correcting the issue, and so on, and so
But there's no -- there's no provision --
well, inspectors do not -- you won't see one of the
questions in the screening questions that is, has the
licensee already found it, or is it already in the
corrective action program? That's not the -- we don't
want licensees -- we don't want inspectors thinking
about those kinds of things.
But again, what Doug has said is true; we
don't want inspectors also -- we also don't want
inspectors living in the licensee's corrective action
program where they're simply, Doug's words, minding
the corrective action program, looking through them
for issues that we can bring up and document as our
own inspection reports.
MR. LEITCH: But the fact that you have
few and declining numbers of non-white inspection
findings would seem to indicate that that's happening
anyway, right? The licensee must have 10 or 15 of
these a day, issues entering the corrective action
CHAIRMAN SIEBER: That's about right.
MR. LEITCH: And many of those could be --
could somehow be a non-white inspection finding. So,
there must be a de facto going on, a kind of -- these
many low level issues are just not even surfacing as
non-white -- non-color issues.
MR. JOHNSON: Right, right. Yes, and
that's what we mean when we say we don't want
inspectors to mind the corrective action program.
We think it's healthy for licensees to find their own
issues, to put them in their corrective action
And we don't -- we don't want a program
that discourages that by raising those -- pulling
those issues out, raising them, documenting them, you
know, just for the sake of getting greens on the
CHAIRMAN SIEBER: On the other hand, even
though an issue may be licensee-identified, if it is
truly risk-important, you would still have enforcement
action regarding that. For example, the failure of
all emergency diesel generators to start or load, even
though the licensee may have discovered that and
corrected it, it still is a matter for enforcement --
MR. JOHNSON: It still matters --
CHAIRMAN SIEBER: -- is that not correct?
MR. JOHNSON: Exactly, it still matters in
the reactor oversight process. It's still something
that we would take action on if they cross thresholds,
including -- including, perhaps, enforcement.
CHAIRMAN SIEBER: Right, okay, thank you.
MR. COE: Right. At the moment, there's
-- it doesn't matter who finds it or whether it's
self-revealed. We'll assess its significance, and
we'll utilize the action matrix accordingly.
CHAIRMAN SIEBER: Right.
MR. COE: Okay, that's the last example on
no-color. The next example is a green inspection
finding. Again, this is under the reactor safety or
the mitigation cornerstone -- mitigation systems
In this case, during the conduct of an
inspection procedure that was looking at surveillance
testing, inspectors identified that an RHR system
bypass valve had been temporarily modified to be in
its full-open position.
However, the licensee hadn't done any
evaluation following that modification as to assure
that the technical specification flow requirements
were being satisfied.
However, the subsequent evaluations that
the licensee performed showed that the system flow did
meet its surveillance test requirements.
Okay, now, this differs somewhat because
-- from the previous examples because there was a
definite impact, is considered to be a definite
impact, on the cornerstone; in other words, a safety
function or a function of a component was affected.
The flow was reduced. There was an
impact, a physical change, a difference. And it was an
adverse difference. It was the flow is less, okay?
When screened through 0610* questions that
we discussed briefly before, this conclusion is drawn:
that it did affect the mitigating systems cornerstone
and, therefore, its disposition would be, again, just
as before, to confirm that the issue had been entered
in the licensee's corrective action program.
In addition, since it did affect the
cornerstone, we would proceed to do a phase one SDP
analysis. And the question that the phase one SDP
asks in this particular instance is whether or not the
system function had been affected, but whether -- not
only had -- did the system function -- was the system
function affected, but was operability and design
That's one of the questions. In fact,
that's the first question that is asked of an issue in
the mitigating systems cornerstone and when it enters
the -- this SDP.
And in fact, if that answer is yes, that
operability and function were maintained, the issue
screens, at that point, as green. And the licensee is
expected to correct that, or the conditions, the
underlying conditions, which caused that.
But we would not engage further with any
further inspection of its root causes or, you know, we
would leave that up to the licensee.
MR. LEITCH: Would a notice of violation
have been issued in this case?
MR. COE: That's a good question. I don't
know the answer to that. I'd have to go back to the
inspection report. I didn't -- that didn't jump out
MR. LEITCH: Just another similar
question: is a non-color synonymous with a non-cited
CHAIRMAN SIEBER: No.
MR. LEITCH: Are those two categories --
MR. COE: No. A green -- a violation
which is given green significance is going to
normally, in almost all cases, be given a non-cited
MR. JOHNSON: Yes, that's true.
MR. LEITCH: Yes, okay. But a no-color is
always a non-cited?
MR. COE: Well, if a no-color finding
arises because of willfulness or an issue which
significantly impedes regulatory process such that we
would consider it -- you know, we may consider for a
severity level enforcement action up to, and
including, civil penalties.
MR. LEITCH: But it would still be a no-
MR. COE: But it would still be a no-
MR. LEITCH: I see, okay.
MR. JOHNSON: Yes, we're sort of a little
squeamish on answering the questions, your specific
questions, only because -- and I'm looking at Steve.
Steve is not even looking at me now because he knows
that we're having an ongoing dialogue with the Office
of Enforcement on this issue.
And in fact, one of the things I intend to
do when we come in July is bring the Office of
Enforcement along with us because I think we ought to
be able to talk about -- to address your questions
about what enforcement also comes out.
But to be quite honest, with respect to
what is a no-color finding, Doug is exactly right.
There are issues -- there are issues that receive
And the Office of Enforcement doesn't
consider those to be no-color findings. And I don't
want to make this -- I don't want to make this overly
convoluted, but let me just say that we have to --
we're still working with how we -- with this whole
topic of no-color findings and how we eventually end
up with respect to what is a no-color finding.
When we set up the action matrix early on,
we intended that there would be two kinds of things.
There would be things that you could run through the
SDP that would receive a color, and that color was
synonymous with where they would fit, that we could
run through the action matrix and come up with a
It was also the recognition that severity
levels would still apply for things that received
traditional enforcement that were -- that is, things
that were outside of -- things that could impede the
For example, those would receive --
willful violations, those would receive traditional
enforcement. So, you could have a situation where you
have a finding that -- you'd have a collection of
things: things that receive colors, things that
received a severity level, right?
And so, it's not as -- and so, when you
ask a question, do all no-color findings -- are all
no-color findings NCVs, well, that really depends on
how you define a no-color finding with respect to how
you treat these traditional enforcement items under
that definition of no-color findings.
In July, we'll have our act together
because we will have -- we will have closed the loop
on the dialogue with respect to no-color findings, and
we'll be able to answer that.
So, if you can hold whatever questions you
have for that --
MR. COE: All right, the next example is
the white inspection finding. In this case, an oil
leak was identified by an inspector on an emergency
feedwater pump. This is in a pressurized water
reactor which had, over a period of time, forced the
operators to make daily oil additions in order to even
maintain visibility in the oiler.
And when the residents had questioned
this, the ultimate answer, the licensee determined
that there were some bolts on the bearing housing that
had been loose. The oil had been leaking into a drain
sump, so it hadn't been puddling on the floor.
So, it was basically, you know, not
gathering the kind of attention perhaps that it
should, other than the fact that the operators were
adding oil every day.
Ultimately, it was found that if the pump
had been called upon to operate, it would have only
run for a few hours. And then, the bearing oil would
have gone away, and the pump bearing would have --
would have become irreparably damaged, and that that
condition occurred for 39 days.
Once again, the 0610* documentation
threshold was met because it did affect a mitigating
system cornerstone. The pump was actually inoperable,
unavailable, for that 39-day period of time.
Now, again, it would have operated for a
few hours, but from a -- from the standpoint of a
significance determination, the going in assumption
was that it would not meet its mission time. It would
not have satisfied its safety function in the long-
Phase one SDP, another one of the
questions that the phase one SDP asks is whether or
not an actual loss of function of a single train has
occurred. And the threshold that's put on that is if
it's greater than its tech spec allowed outage time.
And that's not necessary a risk-informed
threshold, but it is a threshold that has -- we have
historically used, even in the Accident Sequence
And we have borrowed it and continued its
use in this program.
DR. APOSTOLAKIS: Are these questions in
this Appendix B?
MR. COE: No, these particular questions
-- in phase one, now we have left 0610*, which is
essentially the documentation threshold and the
discussion of no-color findings. And now, we've
entered 0610, Appendix A, which addresses the
significance determination for reactor safety
And in that document, you would find a --
in fact, you're going to see it in just a moment -- a
worksheet that lists these questions for phase one --
for the phase one SDP process.
And this question -- my only point here is
that's the question which kicks this issue into a
phase two. In other words, if there was an actual
loss of safety function and it was greater than tech
spec allowed outage time, and therefore, it is
deserving of further attention, further analysis; not
that it wouldn't come out potentially green upon
further analysis, but we can't say that for sure right
now. And it needs to be looked at further.
Again, in an overall sense, we have a
graduated approach here. Phase one is a relatively
broad screening process that allows an inspector to
assess that something is not -- does not need further
analysis or review from a risk standpoint, that they
can call it green.
They don't have to accept that, by the
way. The guidance that we have put out is that if an
inspector wants to exercise a phase two process,
they're more than welcome to do that. And in fact, we
encourage it, even if they think it's a green, because
that starts them and continues them into the process
of gaining risk insights into that particular issue
and, in fact, into that plant in general.
But in this particular example, that's the
screening question that's relevant, and it forces the
next phase of SDP analysis.
Now, the next phase of SDP analysis is
phase two. And I've got a more detailed set of
documents here that we can look at. But the high-
level picture of the phase two analysis is the
following. The worksheets are utilized that are
We've created inspection notebooks, risk-
informed inspection notebooks, which contain a series
of worksheets. And the purpose is to identify what
this impact has had on the dominant accident
DR. APOSTOLAKIS: Now, why dominant?
MR. COE: Because that's what drives the
risk. That's what drives the significance. In other
words, the question that we're asking is which -- for
this particular degradation, this particular
deficiency, what effect has that had on the sequences,
the accident sequences?
Which sequences were affected by that
particular degradation, and how much remaining
mitigation capability was left to mitigate those
DR. APOSTOLAKIS: But if you have a
problem somewhere that affects a system that does not
appear, say, in the top five sequences, but affects,
you know, the following seven, what happens then? I
MR. COE: Okay, it's a good question.
Your question sort of infers that we're only listing
the dominant accident sequences for review. In fact,
the sequences are written at a very high level. Any
one sequence is essentially a functional sequence.
The sequence that's represented here,
which was the one that came up the highest, when you
look at it as far as what's changed, is a transient
with a loss of power conversion system, essentially
loss of the main condenser and the turbine, followed
by a loss of all the other emergency feedwater
And therefore, you have a loss of function
of emergency feedwater, followed by a loss of primary
feed and bleed. Now, that's the sequence --
DR. APOSTOLAKIS: How can you lose feed
MR. COE: Well, you didn't in this case.
In this case, the only thing that was affecting that
sequence was EFW, the EFW pump.
DR. APOSTOLAKIS: You mean you lose the
capability to feed and bleed?
MR. COE: This is simply a functional
accident sequence. The end result of this sequence
occurring is core damage, okay?
DR. APOSTOLAKIS: Right.
MR. COE: So, what we're doing is we've
got a whole bunch of sequences listed in the
worksheets. And the idea is which of those sequences
was affected by a -- in other words, what changed?
Which sequence baseline value for that core damage
frequency risk contribution has changed?
Well, this one changed because this one
used to have two motor-driven pumps and a turbine-
driven pump. It now, for a period of 39 days --
DR. APOSTOLAKIS: Has one.
MR. COE: -- there is only one motor-
driven pump and one turbine-driven pump. So, that's
the change that has occurred, okay?
DR. APOSTOLAKIS: Right.
MR. COE: So, this element has changed;
the other two have not. And they retain their
original baseline assumptions on frequency of this
event occurring and likelihood or probability of this
loss of function occurring as well.
This essentially is providing defense in-
depth. And the remaining mitigation capability that
remains here is providing us a defense in-depth to
sustain reactor safety --
DR. APOSTOLAKIS: So, do you have --
MR. COE: -- even when you had this
DR. APOSTOLAKIS: Do you have those sheets
here and --
MR. COE: Yes.
DR. APOSTOLAKIS: -- the information that
the inspector has?
MR. COE: Yes.
MR. JOHNSON: Yeah, I was going to tell
you that Doug has -- you are actually going to go
through those sheets, aren't you?
MR. COE: We'll go through them in as much
detail as you wish.
CHAIRMAN SIEBER: That's going to take a
MR. COE: This is just the high level
treatment. I'm giving you the answer. And then, as
you have interest --
DR. APOSTOLAKIS: Wait a minute --
MR. COE: -- we'll go through the details
of how we get there.
DR. APOSTOLAKIS: So, all these, pages 16,
17, 18 --
MR. COE: Yes.
DR. APOSTOLAKIS: -- 19 -- do you want to
get to that right now or --
CHAIRMAN SIEBER: No, I think that what we
-- what our best bet is to continue and finish with
the overall explanation as to what went on. And then,
we can take a break and come back and --
DR. APOSTOLAKIS: Okay, good, good.
CHAIRMAN SIEBER: -- dive into the
MR. COE: Okay, the -- what comes out of
the analysis essentially, for this particular
sequence, is that you have to define what the
likelihood of the initiating event is.
And actually, it's -- we have a table, and
I'll show it to you. And we've characterized bands of
likelihood, which are essentially probabilities, with
the letter characters that represent those bands.
In this case, this is the highest
frequency. In other words, this -- what this
represents is, is that the initiating event frequency
is greater than one in ten years, and the exposure
time is greater than 30 days.
So, the likelihood of this event occurring
within the 39-day period of time is characterized as
"A". We'll get into what that -- a little more detail
of what that means later.
In addition, the mitigating system credit
that I was talking about earlier, there were two
remaining motor-driven -- I'm sorry, there was one
remaining motor-driven emergency feedwater pump left.
So, the mitigation credit for that
function here is two, which is a representation of
10-2 likelihood that that remaining motor-driven pump
would not be available, plus one, which is
representing the turbine-driven emergency feedwater
pump availability of -- in this case "1" represents
10-1 likelihood that it would not be available.
So, those are added, two plus one here.
And then loss of feed and bleed, normally we give feed
and bleed about a 10-2 credit for unavailability. And
in this case, that's represented by this "2".
So, you add these up and you get "5". You
don't assume that you can recover that damaged pump
should it have been called upon to function. And so,
you are left with a credit -- a mitigation system
credit of five.
When combined with the likelihood rating
of "A," you enter another table and you end up with a
significance result of white, which is a
representation of a change in core damage frequency
because that pump, that one pump, is degraded for
those 39 days. The change is between E-6 and E-5 per
DR. APOSTOLAKIS: So, the credits are
really the exponents?
MR. COE: Yes, that's essentially the
negative log rhythm of the unavailability figure --
DR. APOSTOLAKIS: Good.
MR. COE: -- the baseline unavailability
CHAIRMAN SIEBER: And I guess that's on
page 23, right, how you get -- you know, the --
MR. COE: Page 23?
MR. JOHNSON: Yeah, we'll get -- we'll get
MR. COE: Yes, page 23 is the table which
actually produces the final result --
CHAIRMAN SIEBER: Right.
MR. COE: -- for that sequence, okay?
Now, phase three -- you know, it's acknowledged, and
it was acknowledged right at the very start, that
phase two is a crude process.
Its value is that it's in the hands of the
inspector, who is the closest person to the actual
plant design, plant operation, and can be the best
person suited to identify if any of the assumptions
that are being used in this level of analysis are
incorrect. And that's --
DR. APOSTOLAKIS: What is -- I'm sorry.
What is the CDF of this particular plant? Do you know
the baseline CDF?
MR. COE: I think it's about 3E-5 or 4E-5
per year based on their IPE. And I don't know if
that's been updated.
DR. APOSTOLAKIS: So, this is one-tenth of
MR. COE: Pardon?
DR. APOSTOLAKIS: The change was one-tenth
of that, right?
MR. COE: It's in that range, yes. It's
in that range of 10-6 to 10-5 per year.
DR. APOSTOLAKIS: Right.
MR. COE: Right. The phase three process
was an acknowledgement -- the need for it was an
acknowledgement that there would be occasions -- yes?
DR. APOSTOLAKIS: Is yellow or white
worse? Which one is worse?
MR. COE: Yellow is worse.
DR. APOSTOLAKIS: Yellow is worse?
MR. COE: Yes, by an order of magnitude.
MR. JOHNSON: And red is worse even.
DR. APOSTOLAKIS: Yeah, I knew that.
MR. COE: So, phase three was essentially
an acknowledgement that risk analysts would have to
probably get involved at some point to either confirm
that a phase two analysis was correct, that the
assumptions were appropriate, and that the analysis
was producing an answer that was defensible, or it may
be that the SDP, itself, the phase two worksheets --
there are certain cases where the SDP worksheets will
For example, if a component is not
considered totally unavailable, but a deficiency has
reduced its reliability, the phase two worksheets
The only way to assess that is through the
use of adjusting the -- you know, making some
judgement about the change in reliability, and then
processing that through a computer-based model.
That's the only way to do it.
So, there are occasions when phase three,
we anticipate, would be needed and, for the most part,
because the worksheets have been so delayed in coming
out. For the first year of operation -- of
implementation for ROP, we've done a lot of phase
And we're hoping to relieve the burden
somewhat on the risk analysts that have been doing
these analyses by the implementation of these
CHAIRMAN SIEBER: I presume the region
analyst is doing the phase three evaluation and the
licensee is doing it with their own PRA at the same
time. And the chance of the answers being exactly the
same are probably nothing.
MR. COE: There's always differences.
CHAIRMAN SIEBER: So, how do you resolve
MR. COE: Well, we talk about them and
understand what the differences are. In many cases,
the licensee is making assumptions that, you know, at
least initially, we are not necessarily willing to
So, the process allows for us to come
forward with a preliminary analysis and put it on the
table and then, in fact, offer the licensee the
opportunity to come back in a formal way, through a
formal, docketed process, a public meeting, to come
back and give us further information, upon which then
we can make a final decision.
We don't have to come to consensus with a
licensee in order to produce an SDP result. But we do
offer the licensee the opportunity to provide us with
as much information as they feel is pertinent so that
we can make a well-informed decision.
CHAIRMAN SIEBER: So, this would be a
matter for an enforcement conference if one were to
MR. COE: We call it a regulatory
CHAIRMAN SIEBER: Right, okay.
MR. COE: And the purpose is primarily to
gather the information that we need to make our final
CHAIRMAN SIEBER: Okay, thank you.
MR. COE: Right. I would point out that,
in many cases -- and this is an ongoing issue for us.
And how we do this better is to account for the
external initiating event contributions.
As you're aware, the level of detail and
sophistication and, in fact, the complexity -- because
of the complexity of external initiating events is --
we don't have as much information as we have for the
And so, we do the best we can with the
information we have. But in some cases, it has to be
a fairly qualitative judgement as to whether or not
there's a contribution that would affect the analysis
In this particular case, an enforcement
action was issued as a notice of violation, or will
be. Again, this hasn't come to completion yet. But
the expectation that this kind of a violation would be
a cited violation, okay?
Earlier, we were talking about the non-
cited violations at the lower-significance levels.
When an issue of -- when a finding reaches the white
or above significance level, then NCVs are not an
option under our program.
We cite the violation. We require the
licensee to respond in writing on the docket.
CHAIRMAN SIEBER: Now, when you issue a
cited violation -- and let's not use this as the
example, but just in general -- you still have the
levels under -- like you had under the old enforcement
policy? How do you determine what level you're in?
MR. JOHNSON: Actually, when -- for an
issue like this that you've been able to run through
the SDP, we've set aside those levels.
CHAIRMAN SIEBER: Okay.
MR. JOHNSON: The significance is given by
the color of the finding. And because it is an issue
that's associated with a violation, it is a cited
So, with respect to the ROP and things you
can run through the SDP, in general, we don't have
severity levels. In general, you don't have civil
And I'm saying "in general," because if
you had a situation where there was some actual
consequence, it is possible to have a color, to also
-- but then again, run it through and assign a
severity level and, in fact, issue a civil penalty.
But in general, for most violations, we're
talking about a color and a citation if it's a
CHAIRMAN SIEBER: That's enough, right?
MR. JOHNSON: Yes.
CHAIRMAN SIEBER: Okay, thank you.
MR. JOHNSON: Okay, that's actually --
that's the high level treatment of SDP for the reactor
safety. I would also point out that you're going to
hear a little more detailed treatment of a fire
protection example a little bit later.
The fire protection example has an entire,
separate appendix that they work through. But then,
the outcome of that is an input to these kinds of
worksheets, these SDP worksheets that we've been
So that there's some initial work that has
to be done in order to process a degraded fire barrier
issue, that type of issue. And we'll get to that a
little bit later.
Now, I don't know if you wanted to take a
break now, because what I would be prepared to do is
to go into some more detail on this particular white
finding issue -- example.
CHAIRMAN SIEBER: I think right now would
be an appropriate time to take a break, at least for
me. And let's reconvene at 10:30.
(Whereupon, the proceedings went off the
record at 10:10 a.m. and resumed at 10:30 a.m.)
CHAIRMAN SIEBER: I think we can resume at
this time. Mike?
MR. COE: Okay, I'd just like to walk
through some of the details of the white inspection
finding example that I showed you a moment ago. We'll
go into whatever amount of detail that you care to.
The first part of an SDP in the reactor
safety arena is a clear documentation of the
condition. Factually and for the purposes of
establishing exactly what the impact was on plant
risk, we have to not include hypothetical situations,
such as single-failure criteria.
So, we basically ask the inspectors to
document factually what the condition is. We also ask
them to think about the systems that were affected,
the trains that were affected, the licensing basis or
the design basis of the system.
That's sometimes not necessarily the whole
story because it might have risk-significant functions
that go beyond the design basis of the system. But at
least as a matter of completeness, we wanted to ensure
that that function was articulated. And that's done
here on this sheet.
Maintenance rule category is important
when we ask some of the questions in the next -- in
the next part of the phase one process. In the time
that the identified existed, or is assumed to have
existed, is important, again, from the standpoint of
the final risk determinations because it's one of the
So, this is just a more complete
description of the identified finding that I
illustrated earlier at a high level.
So upon documentation that you saw there,
the inspector is given a worksheet that looks like
this to help them identify if, in fact -- or which
cornerstone was affected.
At this point, the decision had been made
already that a cornerstone was affected. And based on
the earlier questions, it's anticipated that it would
be the mitigation systems cornerstone.
But this worksheet lays out all three
cornerstones in the -- or three of the cornerstones in
the reactor safety strategic area, and asks the
inspector to clearly identify what, exactly, is the
cornerstone of concern.
In some cases, it might be a combination
of cornerstones. It might be -- a single issue might
affect both an initiating event frequency, as well as
a mitigation system cornerstone. But this is just to
lay out the assumptions.
And I would point out that the
documentation expectations for the SDPs in this area,
and in other areas across the board, across all
cornerstones, are expected to be clear and
reproducible, such that an individual member, a
knowledgeable member, of the public could take the
inspection report and the description of the
significance basis and take our guidance documents,
such as Manual Chapter 0609, the SDP process, and
arrive at the same conclusion, so that it's
reconstructible -- the basis is reconstructible.
Okay, so that's -- here again, the
mitigation cornerstone is the one that we're -- it
should be this -- this next sheet here should be the
next one in your package. I hope it is.
CHAIRMAN SIEBER: No.
MR. COE: It's not? Well, go to the next
page, then. We might have reversed these two.
CHAIRMAN SIEBER: Page 18.
MR. COE: Okay, yes, page 18 is this sheet
here. And it's actually the next one after the one I
just showed. Here, the inspector is given a series of
questions to determine whether or not the issue can be
screened as green at this point and processed
accordingly, or whether it needs a further treatment
which may, or may not, result in a higher level of
significance, but at least warrants the further
treatment in phase two.
In the case of mitigating systems --
mitigation systems, the inspector asks these series of
questions. If the issue had impacted the initiating
event cornerstone or the containment barrier, part of
the barrier cornerstone, then you would ask -- he
would ask these questions.
In this particular example, the feedwater
-- the emergency feedwater pump example, the first
question was, is the finding a design qualification
deficiency that does not affect operability? The
answer is no, it did affect operability.
And so, you go to the next question: is
the finding an actual loss of safety function for a
system? Well, that was defined in the first page,
what the system was.
The system, in this case that was
affected, was the emergency feedwater, and the whole
system had not been lost. So, the answer to that
question is no.
The third question is, does the finding
represent actual loss of the safety function of a
single train of a system for greater than a no-tech
And this gets to the criteria that I
indicated earlier, causes this to be answered yes.
And therefore, we go to a phase two analysis. In
other words, there is a need to look at this issue in
further detail to assess its significance.
The other page that's labeled 17 is not --
is not used at this point in the process because it
deals with external initiating events. And it's a set
of screening questions that would be used at the end
of this process.
If this process established a level of
significance above green, then we would come back and
we would look at these questions, and determine
whether or not, either qualitatively or
quantitatively, depending on the information we had
available, whether or not external initiating events
were a contributor to this significance.
But we'll -- we can come back to that
after we finish the internal event treatment.
Now, each plant -- basically, now we have
established the assumptions that we're going to be
using in utilizing the various worksheets in the risk-
informed, plant-specific inspection notebooks, which
include these phase two worksheets.
This table here represents, for this
particular plant, the initiating event frequencies for
all of the initiating events that would be subject to
consideration by this SDP.
And this table basically requires the
inspector to -- and there's another table that will
help in just a moment. But this table allows the
inspector to determine what the initiating event
likelihood is for the period of time, the exposure
time, for the degraded condition.
So, in this case, 39 days is this column
here, this first column. And as we will find out --
I have already -- like I said, I've already given you
But as we will find out, the initiating
event that prompts the sequence of greatest interest
here is one that has a -- starts with a reactor trip
with a loss of power conversion. So, that's this
initiating event here.
The assumption -- the going-in assumption
of its frequency is in this range. And since it's a
30 -- greater than 30-day period, that's why it
resulted in this estimated likelihood rating of "A,"
which represents, once again the likelihood that that
event will occur within that period of time. Okay,
and we can come back to that as need be.
Now, how does the inspector know which
sequences he needs to examine in order to assess which
have been affected by this particular problem?
Well, the affected system; that is, the
system that has the problem, is emergency feedwater.
And this table, which again is specific to each plant,
each plant's design, and is incorporated into that
plant-specific notebook -- this would indicate that
the EFW system, which is composed of two motor-driven
pumps and one turbine-driven pump, appears as a
mitigation function in all its initiating event
sequences, with the exception of medium LOCA, large
LOCA and loss of instrument air.
So, in other words, the inspector's next
task is to pull all of the worksheets out -- and we'll
be going through a couple of those in a moment -- that
-- except for these, okay? Because in all of those
other worksheets, EFW appears as a mitigation
And so, it has now been affected. So,
therefore, the likelihood of those sequences occurring
has been affected. And then, ultimately, we're going
to seek to answer the question, how much have they
been affected, okay?
The next -- the next two pages are just a
continuation of that table.
CHAIRMAN SIEBER: Now, these are plant-
specific; is that not correct?
MR. COE: What I've gone through so far is
all in the plant-specific, risk inform notebook.
CHAIRMAN SIEBER: Okay.
MR. COE: One is generated. There's about
70 or so such notebooks that are either developed or
in the final, the very final, stages of development.
And they cover all 103 operating reactors.
CHAIRMAN SIEBER: All right. This table
here is a generic table. It appears in our overall
guidance document for the reactor safety SDP in Manual
Chapter 0609, Appendix A.
And what will happen here is that for
those sequences that the inspector has identified as
having been impacted, each sequence will be given an
initiating event likelihood based on the particular
initiating event for that sequence.
And then, each sequence will be judged in
terms of the remaining mitigation capability that
remains given that this one aspect of its mitigation
-- of the mitigation function is unavailable.
And depending on how much mitigation
function capability remains -- again, this is
remaining mitigation capability -- and these are just
examples. They're not a complete set of examples;
they're just examples -- then, for that sequence,
we'll generate a color.
And the color will reflect the delta core
damage frequency change associated with that sequence
on an order of magnitude basis. Okay, so we'll come
back to this table as well.
In fact, I would suggest you maybe put a
marker in that because we'll want to refer to that as
we go through the analysis.
This is the first worksheet that -- and,
in fact, is the one that had the white sequence in it
that I identified as dominating this particular
The worksheet is -- the way it's laid out
is the first row up here, the first line, carries
forward information that has already been determined
from the previous tables.
This comes from that first table, and it
indicates that this particular initiating event
frequency is found in row one of that table, and that
the exposure time assumption was that it was -- that
-- from that table was greater than 30 days, and that
that result was "A," as we saw, okay?
Now, this next section of the table
defines what these mitigation functions are that
appear in these accident sequences. Again, these are
high-level functional accident sequences.
So, for each function, such as EFW, EFW
will be described in terms of the plant components
that are available to mitigate that -- to provide that
function in order to mitigate that initiating event.
And this is describing the full,
creditable mitigation capability for each of these
safety functions. So, this is as much credit as you
can take for that safety function for that particular
In the case, in the specific case, of
emergency feedwater, again emergency feedwater, in
this plant, can be achieved -- the safety function can
be achieved -- with any one of two motor-driven EFW
trains and that the two of them together, therefore,
comprise a one, multi-train system amount of credit.
And there's a numerical value associated
with that credit, and we'll talk about that in just a
second -- in a minute.
In addition, the turbine-driven EFW train,
there's one. And it is a full, 100 percent capable
train. So, one of one of that train is also capable
of providing the full function.
Now, in addition, there's -- there needs
to be steam relief on the secondary side through
either one out of two ADVs -- that's atmospheric dump
valves -- or one out of 20 safety valves, okay?
And that's -- that's a necessary -- that's
there basically to -- for completeness. It doesn't
really factor into the credit because you have so many
What you're really limited by is the ways
of putting feedwater into those steam generators.
Okay, now, looking down here at these
three sequences that are listed, we see that EFW
appears in all three sequences. Therefore, all three
sequences have be affected by this degradation, by
So, treated individually, we say how much
mitigation capability remains for each affected
sequence? In the case of EFW, we have one motor-
driven EFW pump remains, and one train, one electro-
mechanical train, is given a credit of two, that it
represents 10-2 unavailability.
Okay, a turbine-driven emergency feed
pump, because it's a -- what we call an automatic,
steam-driven train, we have only given it a credit of
one, 10-1, unavailability.
So, there's one in ten chances that that
turbine-driven EFW pump would not function upon
demand. But there's one in 100 chances that the
motor-driven EFW pump would not function upon demand.
And that's based on our generic insights
in terms of the differences between electro-mechanical
train reliability and steam-driven train reliability.
So, that's the amount of credit we're willing to give
in this particular, rough analysis.
The other function that's associated that,
if it were to fail along with these other failures or
events, would lead to core damage is high pressure
The high pressure recirculation function
is achieved through -- you know, for -- and I said
"high pressure," so it's the high pressure function --
is achieved through one out of two high-pressure
injection trains, which, in this case, there's a note
here indicating that there are actually three pumps
for two trains -- taking suction from one out of two
low pressure injection trains. And all of this
requires operator action.
Now, in order to assess the value or the
credit that's given to that function -- and remember,
that function has not been impacted by this
deficiency. So, we're going to give full credit.
You'll notice that, for the EFW, we only
gave the credit that was remaining. The fact that the
other motor-driven pump was degraded or unavailable is
reflected by the fact that it does not appear as
credit for mitigation capability.
In the case of high pressure recirc, there
has been no impact on that function. And therefore,
full capability is creditable. How much credit is
In this case, operator action essentially
is the most restrictive feature because if operator
action doesn't occur, the function will not be met.
And there is guidance in our 0609 document
that describes how that should be treated. In this
case, we give credit of three, which represents 10-3
likelihood that operators will not successfully
implement high pressure recirculation.
If you sum those numbers up -- oh, and
this column right here, failure -- recovery of the
failed train, in each of these cases of an identified,
actual impact, the question often arises, can the
issue or can the degradation be recovered by an
For example, if this had not been a
bearing oil problem; if it had been a switch left in
the wrong position and an operator in the control
room, based on indications, could identify that and
recover from that deficiency, then credit for that
recovery might be warranted.
We give credit under -- only under certain
meeting -- if the action would meet certain specific
criteria, which are actually listed on this worksheet
in the next -- in the next slide.
But in this particular case, we are not
giving any operator recovery credit because our going-
in assumption is that the bearing will fail, and it
will fail catastrophically.
And therefore, there's not time available
to recover from that. So, the function is completely
Now, the sum of these, if our math is
correct, should be six. And that value of six is the
mitigation credit for that sequence. And if you go
back to the table, page 23, that we had up there just
a moment ago, that sequence had an initiating event
likelihood and a remaining mitigation capability of
six, which is this column.
So, that puts it in green. But notice
that green is right next to a white, so we're high
green, okay? I'm not going to use dark green or
But clearly, we're up there -- we're less
-- we're about an order of magnitude away from a
white. And that may be significant later, so -- and
we'll talk about that in a minute.
The second sequence is treated the same
way. In this case, early inventory high-pressure
injection is satisfied by a multi-train system up here
indicating one out of two high-pressure injection
trains -- again, there's three pumps, but there's two
trains. And that injects from a borated water storage
One multi-train system is given a credit
of three, and that represents the combination of the
individual components, the individual train
reliabilities, plus the added factor of a possible
common-cause problem mode.
So, at a high level -- in other words, if
it was -- if it was two independent trains, totally
independent, each train would have a credit of two.
And you could add those together if they were two
independent and diverse trains to get four for our
total mitigation credit.
But if the two trains are identical and
they're part of a multi-train system, then you can't
-- you can't just add those up without accounting for
the potential for common-cause failures.
And when you add that in, that drops you
back to about a 10-3 in a rough sense. So, that's the
basis for that. So, that also, then, is given a
credit of three, which differs in basis from what we
gave up here, because up here, it was based on
operator action. Down here, it was just simply based
on the electro-mechanical train unavailabilities.
Again, it produces a remaining mitigation
capability rating of three -- of six, excuse me. And
we're still dealing with the same initiating events,
so we're still dealing with "A".
And if you go back to the table, it's
identical to what we had there before; again, green,
and I've added here next to white. "NTW" stands for
next to white.
Now, the third sequence is the one of
interest, and this is the one that I represented
earlier on the high-level slide.
The EFW credit is the same as before. In
this case, the feed and bleed credit -- if we look at
the feed and bleed function up here -- or in this
case, it's defined as primary bleed because it really
is based upon the availability -- not only operator
action, but the availability of bleed sources.
Similar to what I described up here for
high-pressure recirc, the things that drives this
credit here is operator action.
And for feed and bleed, primary feed and
bleed, we allow a credit of about 10-2 likelihood of
not succeeding. And so, that credit of two is what is
represented here; and when summed, gives five total,
which, again, if you go back to the table that I
showed earlier, would get you into a white range.
DR. APOSTOLAKIS: So, if the operator
credit is one, what would happen to this finding? It
would be what color?
MR. COE: Well, if you -- if you assume
that for feed and bleed, that operators were only --
you were only comfortable, for whatever reason, giving
operators credit of 10-1 likelihood of --
CHAIRMAN SIEBER: It would be red.
MR. COE: -- of failure, you would be in
the next -- you would go to the next color up, right?
Because this would be one; this total would be four.
DR. APOSTOLAKIS: So, it would be yellow.
MR. COE: White would go to yellow.
DR. APOSTOLAKIS: Now, that -- you know,
the operator actions of this kind, you know, it is
very difficult to quantify the probability because
it's not the probability of failure. It's the
probability of failure to decide to do it, hesitation,
So, I mean, what would that mean now?
This is independent of the -- of the violation or the
MR. COE: This is independent of the
violation, but it's assumption that is used as part of
the significance determination.
DR. APOSTOLAKIS: But your action, though,
from the enforcement matrix, would be different; I
mean, white versus yellow, right? And the full plant
will be penalized depending on assumptions that have
been somewhere else.
MR. COE: That's right.
DR. BONACA: For example, in this case,
you do have -- in the simulator training, they are
begin trained to enter into the --
DR. APOSTOLAKIS: It's not a matter of
doing it correctly. It's a matter of deciding to --
DR. BONACA: Absolutely, and that is
always absurd -- that is always absurd how fast they
get into it. I mean, that sequence -- so, there is
information available on-site.
DR. APOSTOLAKIS: No, but you remember the
Davis Bessie thing where --
DR. BONACA: I understand that, but --
DR. APOSTOLAKIS: -- economic consequences
DR. BONACA: Oh, yeah.
DR. APOSTOLAKIS: So, I don't know that
the 10-2 is on solid ground. I mean --
DR. BONACA: No, I'm with you. I mean,
there is information -- at least in later years, I
know that there is a of emphasis on are they doing it
MR. COE: And I would point out that, on
the next page, which is a continuation of this table,
there's a note that we've added that indicates that
based on the license -- this particular licensee's
IPE, the human error probability for this function
that they used, in their own analysis, was 3.4 E-2,
So, we're not far off. But the main point
that I want to make here, I think, is that all of --
you know, your point is exactly right. The objective
here is to come up with what was the impact on risk?
And we use a core damage frequency risk
matrix as the means of getting to that answer. And
it's all based on a lot of assumptions. And what
we're trying to do here is to bring these assumptions
out into the open so that they can be examined.
And again, the inspectors, very often, are
the persons who are in the best position to point to
an assumption and say, "I know that's not true. I
know that's not right."
We want the analysis to represent the
truth, as best we know it. And in order to get those
assumptions out on the table, we're using this kind of
process. And we're asking the inspectors to go
through the same kind of thinking process that would
prompt them to think well, have I seen any problems in
the simulator with feed and bleed?
Is there any evidence that there's a
problem in this area that I can't -- that this
assumption should really be one instead of two, and
that maybe the color should be yellow instead of
DR. FORD: I'm a material scientist, and
therefore, used to making a hypothesis and examining
it with fact. This is a very logical approach, but is
there any way of going aback and double-checking it
against experience, actual, factual experience?
MR. COE: In terms of do we have a
DR. FORD: Yes.
MR. COE: -- of operator performance?
DR. FORD: Yes. I mean --
DR. APOSTOLAKIS: I think you are talking
about the whole approach, and not just operations, are
DR. FORD: Exactly. I mean --
DR. APOSTOLAKIS: The whole SDP?
DR. FORD: For instance, your whole --
this whole table is based on input in item one.
You're putting a "1" in that top, left estimated
MR. COE: Yes.
DR. FORD: What happens if it was a
different frequency basically you've got a time-
dependent degradation or whatever it might be?
MR. COE: Sure.
DR. FORD: How would that -- is there
anything that you can double-check these estimates,
reasonable although they may be, against observations?
MR. COE: Well, actually, yes. The basis
for the -- for instance, the initiating event
frequencies that we're using comes from a study that
was completed a couple of years ago by -- it started
out by AEOD, and then became research.
But these assumptions -- you know, the
order of magnitude that we chose to use for these
various initiating events actually came out of an
initiating event study, which is published in a new
It represents the best insight that this
Agency has, based on operating experience that has
been gathered over the years as to what we expect the
generic frequency of those events to occur at.
Now, it's important that the inspector
understands that these are assumptions. And when
they're applied to their specific plant, that specific
plant's experience may differ. But the assumption of
where -- where we're starting out assuming that that
initiating -- that frequency is, or what the
mitigating system reliability is, is starting out with
a generic value.
And those assumptions are exposed through
this process and thereby, allow the inspector to
challenge them if, based on their own knowledge and
understanding of that plant, they feel that they
DR. FORD: So, when you say you have 70 of
these documents going out, which cover all 103
operating stations, they may well change, depending on
the history of that particular plant?
MR. COE: Well, we --
DR. FORD: Bad water chemistry control or
whatever it might be?
MR. COE: Well, I think that -- yes, well,
what you're saying is that the plants will change over
time. They may modify the plant. The reliability of
certain equipment may change over time based on issues
What we've tried to do here is establish
these starting values at a more or less conservative
DR. FORD: Okay.
MR. COE: We think we've got a more or
less conservative set of assumptions here for most
things. And we're continuing to monitor the process
to identify, you know, areas where something might
come up and we might identify that our assumptions may
not be as conservative as we expected.
And so -- but in general, we think that if
this process renders -- and I don't mean that --
whenever you do risk analysis, you really shouldn't be
using conservative assumptions, right?
You should be trying for the best, most
reasonable assumptions possible because conservative
assumptions can often, you know, cause the results to
skew and may obscure other things that you're
And so, we're not trying to be over-
conservative in our assumptions, but the numbers we're
using are based on systems studies, for example, such
as that are generated by research now: ox-feed water,
diesel generator systems.
They've gathered information from LERs and
other databases, and they've done statistical
analyses. And the numbers that we're using, such as
a credit of 10-2 for a single, electro-mechanical
train, a credit of 10-1 for a automatic steam-driven
train, come from -- or at least are checked against --
the results of those studies.
DR. FORD: Okay.
MR. LEITCH: Now, could you go back to
slide 24 for just a second?
MR. COE: Sure.
MR. LEITCH: I have a question about how
you get the "2" in the column there that's labeled
"remaining mitigation capability" --
MR. COE: Yes.
MR. LEITCH: -- "2 motor-driven emergency
feed pump". Where does that "2" come from? I guess
my question is, is this all pre-printed on this sheet,
or is this the result of this specific --
MR. COE: No. Actually, it's a good
question. There is a -- and I apologize; there is a
separate table that's in 0609 that defines the credit
given to a multi-train system. In fact, it defines a
MR. LEITCH: Okay.
MR. COE: And the credit that's given to
a single train -- or in cases where operator action
comes into play, the credit is actually given right
here because operator action credit will change from
sequence to sequence, you know?
So, we don't -- we don't try to define
that in a table. We put it right up here.
MR. LEITCH: Okay.
DR. APOSTOLAKIS: So, Doug, you said
earlier -- but let's confirm it once again -- all of
these tables are plant-specific?
MR. COE: Yes, the tables that I'm
representing here are plant-specific.
DR. APOSTOLAKIS: And the numbers?
MR. COE: Yes.
DR. APOSTOLAKIS: The credits?
MR. COE: Well, the numbers are -- start
out to be generic, such as a credit of two for a
single train, and one for an automatic steam-driven
train. And the frequency of initiating events started
out to be what was represented in the new reg study
that research provided.
As we've gone through the process of
asking licensees for comment, they may have provided
us with some additional information upon which we can
make a decision that we should alter that initiating
event frequency, or that we should alter that
mitigating system function, or that we should alter
that operator reliability, HEP value.
DR. APOSTOLAKIS: Have many licensees
actually asked you to make these more plant-specific
by submitting such requests?
MR. COE: Licensees typically gave us a
lot of information that they felt was more accurate
for their plant. You know, I think in almost every
case, every licensee gave us feedback that they felt
was better reflective of their plant.
Now, we didn't accept that carte-blanche,
obviously. And in fact, there is some advantage to
sort of staying with some more generic assumptions as
Now remember, I said there was a phase
three process too. If our phase two tool is a little
bit over-conservative, we're willing to accept that
because it's expected that if the phase two results
are challenged by the licensee because they have a
better analysis, and typically they will, then we'll
get into a more detailed level of analysis that would
-- would, then, have to account for some of the more
specific differences that the licensee was using
relative to our assumptions.
DR. APOSTOLAKIS: But the determination of
the color is not on a generic basis, correct?
MR. COE: The determination of the color
comes directly from this analysis and these
worksheets, based on the plant-specific assumptions.
DR. APOSTOLAKIS: No, but I mean you have
a matrix somewhere that tells you that a five, right,
is a white? That was --
CHAIRMAN SIEBER: Page 23.
DR. APOSTOLAKIS: Yeah, yeah.
MR. COE: Well, relative to that
particular initiating event likelihood.
DR. APOSTOLAKIS: Right, but it's -- this
is not plant specific.
MR. COE: This table right here is not
plant-specific, that's correct.
DR. APOSTOLAKIS: It is not plant-
MR. COE: Yes, that's correct.
DR. APOSTOLAKIS: It appears to me it
should be; I mean, because a plant-specific nature is
already in the -- is it possible that the same number
at one plant should be a green and another should be
yellow? Does that make sense?
MR. COE: It could make sense if the
plants' designs for the green plant had more
mitigation capability than the one that had the yellow
plant -- or, I mean, the yellow finding.
DR. APOSTOLAKIS: No, but I said the same
number. If you had more mitigation ability, the
number would not be the same.
MR. COE: Well, if the number.
DR. APOSTOLAKIS: You wouldn't get the
MR. COE: If the number was the same, the
color would be the same. The color is representing
the band, an order of magnitude wide, and that doesn't
change. That is a -- that is a threshold, a set of
thresholds, that is consistent with the PIs, and is
DR. APOSTOLAKIS: But --
DR. SHACK: But this is really sort of
giving you an exponent on CDF. So, I mean, it really
goes back to 1174. And so, it is the same for all
DR. APOSTOLAKIS: No, but 1174 uses a
fundamental input, the baseline -- so, no, I'm not
saying that it should be. It just occurred to me that
the decision on the color is generic, but the input
into the matrix is plant-specific.
And I'm wondering whether this is
consistent -- self-consistent. I mean, but I hadn't
thought about it.
MR. COE: Well, you raised the point about
baseline CDF. And our metric here, remember, is the
change in CDF. We're not referencing these colors to
any baseline, any particular baseline CDF. They are
referenced only to the change --
DR. APOSTOLAKIS: Right.
MR. COE: -- delta core damage frequency
and delta LERF.
DR. APOSTOLAKIS: But even in 1174, when
the baseline CDF is greater than 10-4, we drop the
MR. COE: Yes.
DR. APOSTOLAKIS: -- by another magnitude.
MR. COE: Right. For permanent changes to
the plant --
DR. APOSTOLAKIS: Permanent changes.
MR. COE: -- that may be appropriate.
DR. APOSTOLAKIS: Yes.
MR. COE: These are performance
deficiencies that have resulted in temporary
DR. APOSTOLAKIS: Now, do you see a time
in the future where all this will be computerized?
MR. COE: Good question. Maybe.
DR. APOSTOLAKIS: Maybe you see a time, or
maybe there will be?
MR. COE: It is possible. My thoughts
are, it is possible that this is an intermediate step
along the way to the use of -- the employment of more
sophisticated risk tools by field inspectors.
The challenge that we face today, and we
have faced over the past few years when we've tried to
risk-inform our processes, even before ROP, is that
inspectors -- we were not able to give inspectors
sufficient training to allow them to utilize the
computer-based tools effectively that had been
developed, all right?
We established the SRA Program, the Senior
Reactor Analyst Program, in 1995 in order to begin to
get at that need. And it took almost two years of
training before the SRAs were really, fully qualified.
This is a way of accommodating the needs
of ROP while, at the same time, in a very
complimentary way, giving inspectors a better risk-
informed perspective of their particular plant, and of
the risk -- of the probabalistic framework that is, in
many cases, not something that they had to deal with
day-to-day in the past.
They deal with a very deterministic
framework of compliance-oriented and the -- the
decisions as to what was important and what was not
were based on their own experience and the various
pressures that were brought to bear by their own
management, but the licensee, and by the public or
So, what we've tried to do here, and as
we've said repeatedly, is to come up with a more
predictable and objective tool. And this is -- the
risk metric is the way we've chosen to do that.
But the inspectors have a challenge of
understanding better the assumptions and the
limitations of the risk tools that we employ. And so,
this is -- this is the way of accomplishing that.
MR. JOHNSON: And I would only add that my
-- the way I respond to your question, George, is to
say that we think -- we think there is something that
is valuable with having inspectors, at this stage,
work through these sheets.
And in the future, for efficiency purposes
or for accuracy purposes, it might make sense to
computerize it. But today, we think this -- we get --
we get maximum benefit in terms of enabling inspectors
to understand not just what the significance is, but
working through why it's significant.
DR. APOSTOLAKIS: I fully agree with you
that we can view this as a training period where
people really understand what PRA is all about. But
at the same time, as you know, the Office of Research
is putting all the IPEs into a -- so far, they're
calling is SPAR?
MR. COE: Yes.
DR. APOSTOLAKIS: So, after we have a SPAR
model for each unit, maybe that would -- and that will
not happen tomorrow, so --
MR. JOHNSON: Right.
MR. COE: No. In fact, you know, we're
struggling -- I know Research is struggling with the
level of effort and the amount of resources that they
can devote to completing the work on the more
sophisticated -- the next revision to those computer-
But even once they're completed -- you
know, even once they're written, an important aspect
of that is to go out and check them against licensee
analysis results --
DR. APOSTOLAKIS: Sure.
MR. COE: -- and to make visits to the
site to ensure that the assumptions that those models
have in them are accurate.
And then, there's the question of ongoing
maintenance of those models, and how much effort we're
willing to put into that.
And then, there's a whole argument that
says, well, maybe the licensees ought to just provide
their own models for our use. And there's ongoing
discussions at high levels regarding that.
So, how it all plays out in the end, I
don't know. I hold out that there's a possibility
that inspectors will become risk-informed enough to be
able to use the tools if they exist, the computer-
But right now, I think the agents -- not
only the inspectors, but the management, the decision-
makers in the Agency, need to have a process that
forces the revelation of these assumptions as they
make these decisions so that we can legitimately claim
that we have a risk-informed process.
Because if all the assumptions are buried
into computer models somewhere, and we're making the
decisions based on the results coming out of a
computer relative to some standard or some threshold,
I'm not sure that I can call that risk-informed, okay?
MR. LEITCH: I think I may be getting
mixed up a little bit between core damage frequency
and change in core damage frequency. I guess my
question basically on the next slide, 24 I guess it
is. On that last example, could there be a scenario
where normal operations gets a green?
DR. APOSTOLAKIS: Gets what?
MR. LEITCH: Gives a green. In other
words, you're running along normally with three --
with two -- you just had it there a minute ago.
MR. COE: Yes, slide 24, right?
MR. LEITCH: Slide 24, yeah. It's
DR. APOSTOLAKIS: Slide 24 is up there.
MR. COE: Oh, thank you.
MR. LEITCH: Say you had both motor-driven
pumps and a turbine-driven pump, and you assume, say,
one for feed and bleed. Does that give you a green
indicator in normal operations?
MR. COE: Well, first of all, you only
look at these if they've been changed. So, if a
baseline contribution of a particular sequence -- the
baseline contribution of all full mitigation
capability is potentially white, okay?
I don't think that happens very often, but
it's possible, right?
MR. LEITCH: Yes.
MR. COE: Because white represents a
single, functional sequence that contributes anywhere
from 10-6 to 10-5. You know, and most core -- most
plant baseline CDFs are between 10-5 to 10-4.
But the point is, is that you only look at
this if there has been a change.
MR. LEITCH: Okay.
MR. COE: Now, the theory --
MR. LEITCH: But if you just --
MR. COE: Philosophically, what happens
is, if you're only looking at the sequences that have
changed, if we were to look at the core damage
frequency with the change, we would add all the
baseline sequences, the ones that didn't -- weren't
MR. LEITCH: Okay.
MR. COE: And then, when we subtract off
the baseline, all of those go away. All those
sequences go -- the contribution to all those
sequences goes away. So, all we're left with is the
one that changed.
MR. LEITCH: That changed, yes, yes.
MR. COE: So, that's -- theoretically,
that's how we can call this delta CDF.
DR. APOSTOLAKIS: So, this is really CDF-
oriented not LEFT?
MR. COE: Well, we haven't talked about
LERF yet. But the LERF -- the significance standard
for LEFT is essentially one order of magnitude more
sensitive than for delta CDF.
DR. APOSTOLAKIS: But you do have the
tables and everything?
MR. COE: Right. We just -- this issue
didn't impact that, so we're not talking about that
DR. APOSTOLAKIS: Yes.
MR. COE: Okay, so this is only one of
several worksheets. Now, I mentioned that,
essentially, the guidance for this example was that
all the worksheets had to be looked at, with the
exception of a few LOCA worksheets.
DR. APOSTOLAKIS: I wonder if you have --
I mean, one of your cornerstones is emergency planning
MR. COE: Yes.
DR. APOSTOLAKIS: -- which is beyond LERF?
MR. COE: Yes.
DR. APOSTOLAKIS: So, how would you go
MR. COE: We have a separate significance
determination process specifically designed to address
findings coming out of emergency preparedness
DR. APOSTOLAKIS: So, you're using level
MR. COE: No, it's more -- the logic of
that SDP is more related to the nature of the
deficiency that caused the problem.
DR. APOSTOLAKIS: It's not risk --
CHAIRMAN SIEBER: It's determined risk.
MR. COE: Correct. It's really -- you
can't really claim it to be risk-informed, although
what we've tried to achieve with all of these other
cornerstones that you can't link directly to core
damage frequency or delta LERF metrics, is a
The Agency's response is commensurate with
the type of deficiency that has occurred. And it is,
I think in the formulation of those SDPs, somewhat
But what we're trying to achieve is the
same goal, the same level of consistency.
CHAIRMAN SIEBER: Okay.
MR. COE: Okay?
CHAIRMAN SIEBER: And you have the same
situation in physical security, right?
MR. COE: Yes, right. In fact, you know,
as you're probably aware, we made an attempt early-on
to incorporate risk-informed processes -- a risk-
informed SDP process into the physical security SDP,
particularly -- specifically for assessing the
significance of force-on-force exercises.
And that proved to be unworkable. And I
-- you know, I was involved in trying to make it work,
and I, and others, were just simply not successful.
You know, there's too many differences from a -- you
know, when you're talking about risk in terms of
sabotage events and the level of intent that -- you
know, and all of the variations that can occur in
terms of recoverability of things under fairly
It just wasn't workable, so we're
redefining that now.
CHAIRMAN SIEBER: Okay.
MR. COE: Okay. The other sequences that
I thought I would show you -- I haven't gone through
all of them here. But the other ones that came out,
not white, but rather there was another sequence that
came out green, you know, right next to white, was a
loss of off-site power sequence.
In this case, I just wanted to point out
that the loss of off-site power initiating that
frequency is in a different row in table one. It's in
Exposure time, of course, is the same.
It's greater than 30 days. But the result now, if you
look on that table, is "B," which represents a less
frequent initiating event.
Now, that means that you don't have to
have quite as many mitigating -- quite as much
mitigation capability on the average for that
initiating event as you would for the one in the
higher frequency category.
And that -- that affects, you know, in a
probabalistic way, what the outcome of the
So, in this case, we look at EFW again.
These have been affected, and the choices that are
made here, in terms of remaining mitigation
capability, are exactly what we've described before.
In fact, even for this particular
sequence, the feed and bleed is also the same. And in
fact, that sequence, other than the initiating event,
is exactly the same as the one that got us the white.
Okay, in this case, because the LOOP
frequency is less than that transient without loss of
PCS frequency, this -- this value of five, instead of
getting us a white because we drop down one on the --
on this table here -- we're in the -- the LOOP is a
"B" likelihood here. And we come over here to five,
and we're green next to white, okay?
Now, you probably realize already that a
real PRA sums all of these contributions up. And what
we're dealing with here is sequence-by-sequence. And
we're saying the most dominant sequence, you know, is
the one that drives the color.
But in fact, we acknowledge and recognize
that an accumulation of sequences at lower levels may
sum up to something greater than the threshold that we
have set for green and white.
And the way we accommodate that in a phase
two level in the courser treatment that we give a
phase two, is to establish a summing rule. And the
summing rule says that if you have more than two of
these sequences that are green next to white, then you
should call that a white, okay?
Now, that is a somewhat arbitrary choice,
but we thought it was a reasonable one, at least to
start with. And that's not to say that if you even
had two greens next to white that that wouldn't, or
shouldn't, prompt maybe a more thorough analysis,
which is, you know, often easy to do with either our
own tools or utilizing the licensee's analysis, okay?
So, all I'm saying is that we recognize
that that's a limitation of this particular phase two
level of detail, and we've tried to account for that.
And that, if nothing else, gives inspectors -- you
know, reminds inspectors that that's really what's
going on here, that there's a potential for
aggregating, or summing I should -- you know, summing
these lower level issues to something that was of
Okay, that actually completes the
documentation that I had to show you and the example,
unless there's any other questions about what we've
MR. JOHNSON: Now, we're at a point in the
presentation where we'd like to get into the fire
protection SDP if that's possible. Does that fit?
CHAIRMAN SIEBER: That fits with me.
MR. JOHNSON: Okay. J.S., Mark, guys,
would you come up and join us?
MR. HYSLOP: Hey, Doug, I think you've got
MR. COE: They're up here, yes.
MR. JOHNSON: Would you like me to flip
for you or --
MR. HYSLOP: Hi, my name J.S. Hyslop, and
I was the co-developer of the fire protection SDP,
which was developed over a year ago. And Pat Madden
and I -- I'm in PRA -- Pat Madden, a fire protection
engineer, also developed this. We did it together.
And Pat has since moved on. And now, Mark
Salley, beside me, is now responsible for the fire
portion of the fire SDP.
This first slide indicates that I'm going
to give an overview in the presentation. Basically,
it's just the general remarks that are going to be
From that point on, we're going to get
into an example with a specific application of the
fire SDP on a set of fire protection findings we had.
And so, in that -- don't move on yet,
Mark, I'm going to talk about the identification of
the findings and -- clear identification of the
We're going to talk about the fire
scenario, and there -- a realistic fire scenario where
we, of course, have to take into consideration the
figuration of the room as well as the findings
And then, we're going to apply the SDP to
estimate a color and talk about the basis for the
degradation, as well as the failure probabilities
used. Go ahead.
I want to make some general remarks to
just give you some insight into what we're doing with
the process, as well as some information about it.
We're using techniques and data generally
accepted by the fire risk community. What do I mean
by the "techniques"? Well, the technique involved
consideration of a fire ignition frequency, the
defense in-depth elements, barrier suppression,
etcetera, and mitigating systems.
We put all those together, using the
appropriate probabilities, to get a change in core
damage frequency. The date --
DR. APOSTOLAKIS: J.S., most utilities,
the way I understand it, use the screening bounding
methods like five in their IPEEEs. Would that --
wouldn't that make it very difficult to calculate
delta CDF in this context?
MR. HYSLOP: Well, what we're doing is
we're trying to look at realistic scenarios. So,
we're actually using the emission frequency associated
with the scenarios. And we have tools to evaluate the
damage done by the fire, quantitative tools that we're
And you know, we're trying to estimate the
damage -- as a result, we try to estimate the damage
as reasonable as possible.
DR. APOSTOLAKIS: So, what you're saying
is that you are going to be use the IPEEE to some
MR. HYSLOP: We're using a lot of
information out of the IPEEE, but we reserve the right
to develop scenarios ourselves to disagree with those
in the IPEEE because, you know, that's what our
They go out in the field. They look at
the fire sources, and they make independent judgements
themselves about the damage done. And I'm going to be
talking -- my next point is that it's an evolving
And I'm going to give you some information
that -- about some of the things we're working on to
improve that right now.
So, it's an evolving process. We just
released the second version of the SDP. It's my
understanding that that was distributed to you by IPB.
And there, we've got some clarifications
and just -- on identifying and evaluating realistic
fire scenarios, as well as guidance to assist the
inspectors to determine the degradation level
associated with the weakness of our inspection
We also have -- it's an evolving process,
so we have future plans. First of all, we have a tool
development to assist the fire protection inspectors
to evaluate the effectiveness of manual actions
specific to fire.
You know, you may have manual actions
specific to fire because of evacuation of the control
room, or you may have manual actions specific to fire
because of heavy smoke being in the vicinity.
This project was -- we got a lot of help
from Research on this. Nathan Siu was using the data
available to him through the Fire Research Plan.
So, they provided the foundation. NOR has
since looked at that and made some modifications to
it. You know, the next step, of course, is to
document this, and to go around with industry, get
their feedback, as well as the other stakeholders,
because that's the way the reactor oversight process
Another tool that we have under
development, really by Mark Salley, is the development
of a quantitative tool to estimate the fire damage as
the result of a fire ignition source.
And Plant Systems is working on that now,
and developing templates and a guide to use for the
And so, the next step is, what are we
doing with our stakeholders? Are we telling them
about this? And yes, we are. There was a fire
protection information forum held, I don't know, a
while ago I attended. And I talked to them about
these -- about these plans that we have for the fire
And then, there was the reactor oversight
workshop, which was held a month or two ago, or so.
And there, we had a fire protection break-out session
where Mark and I attended, as well as some other fire
protection people. Some SRAs came and, you know,
And we talked about, again, what we're
doing. And we've been talking to industry the whole
time, in response to your comment earlier. Throughout
this development process, early on, before we even
used it, we had many meeting with industry: with NEI,
small meetings with industry where we had a couple
hundred people there -- here. And we ran --
DR. APOSTOLAKIS: What do you call a large
MR. HYSLOP: Well, I guess I would say --
DR. APOSTOLAKIS: More than a thousand?
MR. HYSLOP: No, no, no. There were a
hundred people there. There were a hundred people.
That's a large meeting for me. I come from a small
CHAIRMAN SIEBER: A thousand would be
DR. APOSTOLAKIS: It's a medium.
MR. HYSLOP: Anyhow, the last point is, we
have a state-of-the-art research plan going on ten
floors -- being managed ten floors up, and Nathan Siu.
And he's doing work on suppression. He's doing work
on fire barriers.
And I've told him that I'm interested in
the insights that he gains from his program because
this is an evolving process. And likewise, I'm
interested in your comments. Again, it's an evolving
Next. So, we're going to get into the
example right off the bat. You know, as I said, this
is based on fact.
We had an inspection, and the inspection
identified several findings. The first finding is the
suppression system, was a CO2 system. I'm just going
to tell you about these briefly.
Mark Salley, later in the presentation, is
going to get into these in more detail and tell you
his basis for our choosing a level of degradation
associated with these findings, okay?
So, just briefly, the fixed suppression
system wouldn't maintain the minimum concentration for
the fire hazard. There's a minimum concentration
required, and it was lower than that.
Also, there was a barrier problem. The
electrical raceway fire barrier system protecting
redundant trains didn't meet the one-hour rating. It
was substantially less.
DR. APOSTOLAKIS: So, this was the same
MR. HYSLOP: This is the same plant. This
is the same room.
DR. APOSTOLAKIS: Oh, okay.
MR. HYSLOP: Okay? This is one hour. So,
you've got -- and I'm going to talk about the
configuration, but you've got one room; you've got
fire barriers in that room that are degraded.
And of course, you know the regulations:
when you've got a one-hour barrier, you've got a fixed
suppression system also in tandem. And that fixed
suppression system responsible for protecting that
barrier was also degraded, okay? Now --
CHAIRMAN SIEBER: Let me ask a question.
MR. HYSLOP: Yes.
CHAIRMAN SIEBER: Could you imagine a case
where the lack of functionality of the suppression
system would cause the degradation of the fire
barrier, and therefore, you get basically two issues
out of one defect?
MR. HYSLOP: Well, we look at these --
I'll let Mark Salley answer that more fully. But we
look at these synergistically in the analysis. We say
that these two compound the problem in the analysis.
And you'll see later in the slide how we do that. Do
you want to respond, Mark?
MR. SALLEY: Yes, if I understand your
question properly, are you saying the suppression
system would degrade the barrier?
CHAIRMAN SIEBER: Right, or the lack of
functionality of the suppression system. For example,
here, was the fact that the fire barrier did not meet
the one-hour rating independent of the fact that the
suppression didn't maintain the concentration?
MR. SALLEY: That's an interesting
question, and you take it all the way back to the
barrier qualification, in and of itself. If you
remember the whole thermal lag and the fire barrier
issues, another one that came down the road was kao
CHAIRMAN SIEBER: Right.
MR. SALLEY: -- which was a ceramic,
CHAIRMAN SIEBER: Right.
MR. SALLEY: And there, the hose drain at
the end of the fire exposure would be very important
to have got its qualification that the hose stream
wouldn't remove it. So, that should have been looked
at, at a lower level in designing the system.
CHAIRMAN SIEBER: Okay. So, what you're
saying is you do look at things in a synergistic
MR. HYSLOP: Yes, that's one of the
strengths of this method.
CHAIRMAN SIEBER: All right.
MR. HYSLOP: And then, the last -- the
last thing we have to consider is the time of this
degradation. If you remember in Doug's presentation,
the time affects the change in CDF.
A lesser time -- since it's an annualized
change in CDF, a lesser time has a less effect than a
long time, okay? And we find that these findings
existed greater than 30 days, and that's the largest
There, you have no reduction in CDF for
the time, and they existed simultaneously. And that
was determined during the inspection. Go ahead.
CHAIRMAN SIEBER: Another question: when
you talk about the fire barrier, it could they used
deficient material, or it could be the fire barrier is
defective, like there's a hole in it.
In this case, which was it? And in
general, do you treat them the same way, either
deficient material versus a breach in the system?
MR. SALLEY: When you get into the actual
evaluation, they would start falling in the same
matrix of the degradation --
CHAIRMAN SIEBER: Okay.
MR. SALLEY: -- as to how degraded they
CHAIRMAN SIEBER: All right.
DR. APOSTOLAKIS: J.S., I didn't
understand this argument about the 30 days. You say
that was greater than the maximum, therefore --
MR. HYSLOP: Yes, there are three time
ranges in the SDP: zero to three days, three to 30,
and greater than 30. And the greater than 30,
essentially, you assume you've got 300 or some days'
degradation. It's a factor of one that's used in
So, you don't get a reduction in your core
damage frequency if you're greater than 30 days, where
you would get a reduction of ten if you're three to
30, and a reduction of 100 if you're zero to three.
DR. APOSTOLAKIS: Okay.
MR. HYSLOP: Next slide. Now, I was
planning to jump right into the phase two, but I'll
talk about the phase one a little bit, although I
really don't want to spend much time on it because
it's not as important for this application.
Essentially, we recognize we have
significant degradations in defense in-depth. We
haven't talked about them, but you'll see that.
And this fire barrier and automatic
suppression protect essential equipment, equipment
that's on those sequences, and loss of would have a
big effect. That, alone, will put you into the phase
two process, so now I'm going to talk about the phase
The phase two, one of the earliest things
we have to do is ask the following question: can we
have a realistic fire scenario? You know, we've got
a -- we've got a degradation defense in depth. Do we
have a fire scenario that's going to challenge that?
And so, when you do that, you know, you
have a knowledge of the degradations, and you, of
course, need to have an idea of the configuration of
the room. And I'm going to talk a little bit about
This room was a 4160-vote essential switch
gear room, so you had your safety-related switch gear.
It was divided into three sections by partial-height
marinite walls. These walls went nearly to the
ceiling, but not all the way.
And so, you've got three sections, okay?
So, in each one of those sections, you had an
electrical train -- electrical bus of switch gear
where you needed two buses to support one mechanical
train. That's the way the plant was set up.
Now, if you had a fire in one of the far
regions, then we still had too much of a train. So,
you had a mechanical train of equipment. You really
got into trouble if you had a fire in the center one
because in the center, you had cables crossing over
from each of those electrical trains, over the end of
the center switch gear, okay?
CHAIRMAN SIEBER: And they went over the
MR. HYSLOP: And they went -- yeah --
CHAIRMAN SIEBER: Okay.
MR. HYSLOP: -- over the wall; over the
CHAIRMAN SIEBER: Where the plume would
MR. HYSLOP: Right, right. So, you know,
you've got a -- you've got an ignition source over the
end. A fire starts there, develops a plume,
potentially does damage. Mark is going to talk more
about this. So, do you want to go ahead, Mark?
MR. SALLEY: Sure, this is a good time to
pick it up. I'm Mark Salley from the Plant Systems
Branch. Pat Madden originally had started this. I
helped him a little bit. And Pat moved on, so I've
been picking up a lot of the fire protection with J.S.
from here on out.
George made an important comment earlier
about how this comes together. If you look back, the
IPEEEs, Generic Letter 8820, Supplement 4, there is a
starting point, especially for the people who used the
And they said, "Hey, look, we've done a
lot of work with Appendix R. So, from that Appendix
R starting point, we'll take this snapshot in time,
and we'll do this IPEEE."
From that IPEEE, the next progression is
where J.S. and I are pretty much going. So, I think
you can see, as we're moving along, that one bit of
information is building on the previous one.
To just giver a little summary here of
what J.S. is talking about, we have our three vital
switch gear, 4160, the vital buses, the three fire
DR. APOSTOLAKIS: Is there a reason why we
don't have this?
MR. SALLEY: Oh, I'm sorry. This was just
an extra. I thought I'd give you --
MR. HYSLOP: We just made this one.
MR. SALLEY: -- a real quick -- a little
DR. APOSTOLAKIS: How about a picture
being worth a thousand words and all that?
CHAIRMAN SIEBER: We'll pick up a copy.
We'll get you a copy. He can make it available to all
MR. SALLEY: The fire barrier separator of
the marinite walls that J.S. spoke about, the area of
concern is where the cables from the three merged over
the center unit here, okay?
Now, in the Appendix R-type strategy for
compliance, the requirement would say, okay, there's
a number of ways to do this. This licensee chose to
put one-hour fire-wrap, fire barriers, on those
And the room is -- has a full, automatic
suppression system; in this case, a manual CO2 system.
So, that was his method of compliance. As the
inspectors looked at it --
CHAIRMAN SIEBER: I'm not sure how an
automatic suppression system is a manual CO2 system.
MR. HYSLOP: We're getting --
CHAIRMAN SIEBER: It sounds like it's
MR. HYSLOP: We're going to get into that
on the next slide.
MR. SALLEY: Yes, this licensee did have
a manual here --
MR. HYSLOP: Sorry about that.
MR. SALLEY: -- yeah, with this; you're
CHAIRMAN SIEBER: But these are original
design problems with the construction of this room,
MR. SALLEY: Right. This is unique to
this licensee and --
CHAIRMAN SIEBER: So, this has existed
MR. SALLEY: Yes.
CHAIRMAN SIEBER: Okay.
MR. SALLEY: When the inspectors were
looking during their inspection, they found -- they
inspected the hardware in the plant. They, first off,
review the fire barriers.
In reviewing the fire barriers, what they
determined was that they really weren't one-hour rated
barriers as they --
CHAIRMAN SIEBER: Well, the walls weren't
because they weren't full height.
MR. SALLEY: Well, the --
CHAIRMAN SIEBER: And the wrap probably
had some other defect.
MR. SALLEY: The wrap is the concern here.
DR. APOSTOLAKIS: Wait a minute; what wrap
are we talking about?
CHAIRMAN SIEBER: The way the cables are
DR. APOSTOLAKIS: Oh, the cables. So, the
wall -- the barrier goes, what, not all the way to the
MR. SALLEY: Right.
MR. HYSLOP: Not quite.
DR. APOSTOLAKIS: So, what, a couple of
feet or --
CHAIRMAN SIEBER: So, it's really not a
DR. APOSTOLAKIS: What?
CHAIRMAN SIEBER: It's really not a
barrier, the way that drawing shows it.
DR. APOSTOLAKIS: Well, I mean, for some
events, it is.
MR. HYSLOP: My understanding was it was
quite higher than the switch gear, and it was a couple
of feet from the ceiling.
MR. SALLEY: This, of course, was probably
back-fit to Appendix R, and it was a unique
consideration where they put the non-combustible
marinite in to try to get some compartmentation
between the three pieces of equipment from their
DR. APOSTOLAKIS: So, what did you draw
MR. SALLEY: The area of concern is where
the cables --
CHAIRMAN SIEBER: Is the middle.
DR. APOSTOLAKIS: Right.
MR. SALLEY: -- from the three units came
together at a common point. Now, the licensee's
strategy for compliance would be, okay, from where
we've passed into this area, we need to install one-
hour fire wrap on those cables, so they can survive a
fire in this center unit.
The inspector is looking --
MS. WESTON: I think we need to say this
was probably -- was this an exemption --
DR. APOSTOLAKIS: No, you have to come to
MS. WESTON: I think --
DR. APOSTOLAKIS: Speak with sufficient
clarity and volume.
MS. WESTON: Your name?
MR. WHITNEY: Speaking of clarity, was
let's explain whether or not this was an approved
exemption or not, that this doesn't meet the letter of
Appendix R or does. Can you explain that, please?
MR. HYSLOP: That was Leon Whitney.
You've got to use yoru name when you --
MR. WHITNEY: Leon Whitney, Inspection --
MR. SALLEY: This is an actual
configuration, as I said earlier, for a plant to do
their Appendix R compliance come up with a strategy,
or this is an exemption for the barriers.
For example, in Generic Letter 8610, we
provide a guidance which the licensee would use here.
The requirement was still to have this one-hour fire
wrap in this area, which the licensee claimed they
As the inspectors looked into the detailed
testing of the fire barriers, they determined that it
really wasn't one-hour. In reality, it was probably
10 or 15 minutes of fire endurance from this barrier.
So, that would get them -- to enter the
SDP, there's a design requirement. They don't meet
that design requirement, and that would be the start
In addition to that, they looked at the
CO2 system --
DR. APOSTOLAKIS: Which plant is this?
CHAIRMAN SIEBER: If it hasn't been issued
yet, we shouldn't --
MR. HYSLOP: It was over -- it's over a
year ago. I guess it is; I don't know.
CHAIRMAN SIEBER: Let me point out that if
the inspection report hasn't been issued, then we
should not use the name here on the record, okay?
MR. HYSLOP: It should be. I don't know.
I don't keep up with that.
CHAIRMAN SIEBER So, if you don't know for
sure, don't tell us.
MR. HYSLOP: We don't know.
MR. SALLEY: We don't know for sure, but
it's a real plant and this is a real case.
MR. HYSLOP: It's old.
CHAIRMAN SIEBER: Let's move on.
DR. APOSTOLAKIS: Was this identified as
a critical area?
MR. HYSLOP: I can't remember. You know,
I can't remember, to answer your question.
DR. APOSTOLAKIS: Well, that's a good
question, I think, to investigate because that would
be a good test --
CHAIRMAN SIEBER: Well --
DR. APOSTOLAKIS: -- of the IPEEE.
CHAIRMAN SIEBER: Yes. On the other hand,
it depends on what the deficiencies are in the wrap.
Was the material bad? Was the installation bad, or
was there not enough of it?
DR. APOSTOLAKIS: No, but the IPEEE, we
don't look at deficiencies. I mean, this is a
critical area because all the cables come together.
CHAIRMAN SIEBER: On the other hand, if
you met the regulations, then five would give you an
DR. APOSTOLAKIS: Well, then, I'm curious
to know what five is.
CHAIRMAN SIEBER: Okay, right, and you're
degraded from five's answer at this point.
MR. SALLEY: Right, that's an important
point, George, because five had screening tools with
it. And one of the criteria, for example, was if you
have a suppression system and you meet the NFPA
standard, then you take credit for it.
As the inspection here looks deeper into
it and reviews that suppression system design, they
say, "Hey, wait a minute; for a licensing basis, you
don't meet your suppression system requirements."
CHAIRMAN SIEBER: Right.
MR. SALLEY: So, that could actually feed
back into the five analysis. But the five was a
snapshot in time. John?
MR. HANNON: Mark, this is John Hannon.
I would think it would be also important to recognize
that part of the inspection program itself has us
looking into the areas of most risk significance.
And we would draw from the five analysis
if that was what had supported the IPEEE. So, that
would have been an initiating cause to get us to look
at this room in the first place.
MR. HYSLOP: That's a good point, you
know? That's one of the things that they do, yeah.
MR. SALLEY: Getting back to our scenario,
we have the deficiency in the fire barrier, the cable
wrap. Looking further into the suppression system,
for the mechanics of the CO2 system to extinguish a
fire -- now, the hazard here would be the cables. The
cables would be a deep-seeded fire.
If we took the minimum NFPA 12 for the CO2
system design, it would tell us that you need a 50
percent concentration, and you need to hold that for
20 minutes with a deep-seeded fire -- basically, by
suffocation, removing the oxygen leg of the fire
As they looked into the testing of the
system, what they found was the -- I'm jumping ahead
here -- what they found was the concentration was a
If everybody has a visual, I'd like to get
back to the -- that was an extra that I shouldn't have
MS. WESTON: Now, let me copy it.
CHAIRMAN SIEBER: Well, somebody -- you
need to be here.
MR. SALLEY: I just thought if we couldn't
get a good visual --
DR. APOSTOLAKIS: You've been trained well
CHAIRMAN SIEBER: Yes, I have.
DR. APOSTOLAKIS: -- these proceedings.
MR. SALLEY: Okay, there's one error in
the slides. We have one double-printed. This example
phase two will come later. So, please pass over that.
MR. HYSLOP: So, skip the one slide and
move to this one, please.
DR. APOSTOLAKIS: Degradations.
MR. SALLEY: Degradations. Now, the first
degradation is the suppression system. Now, the
comment was made about this area does have a
deviation. And yes, for a manual actuation, it
wouldn't be equivalent to an automatic actuation. So,
you would begin the degradation right there that well,
hey, this is going to require some human to find the
release box, release the system in the event that they
do have a fire.
Looking at the system further, the CO2
concentration -- like I said, the minimum would have
been for 50 percent, held 20 minutes, to extinguish
the design basis fire, which would be a deep-seeded
fire. They didn't meet that. They had a 46 percent
And the third thing we discussed was the
degradation to the fire barrier. How bad are we
degraded? This is a good question that the inspectors
get into routinely.
For example, if I had a 49 percent
concentration for 20 minutes, and the Code said the
minimum was 50, you know, we start splitting hairs for
the one percent of CO2.
Then, you can get into things like well,
gee, where are the cable trays? Are they in the top
of the room where the CO2 is heavier and it's going to
And we can get into a lot of technical
arguments through the SDP as we move on. But the fire
barrier, being approximately ten minutes, where we
originally had required an hour, is a pretty good
That's definitely a moderate to high
degradation for the fire barrier.
CHAIRMAN SIEBER: Was that due to damage
MR. SALLEY: I believe design in this
CHAIRMAN SIEBER: All right.
MR. SALLEY: And you can see the test that
they had indicated the barrier's rating was 10 to 15
DR. APOSTOLAKIS: I guess I don't
understand it. What tests are these? I mean, tests
that had been done in the past?
MR. SALLEY: Yes. The original tests --
DR. APOSTOLAKIS: And the licensee had
access to them, and they misinterpreted them or what?
MR. SALLEY: This all ties back to the
whole '90/'92 era of Thermalag as to just what is a
rated electrical raceway fire barrier system. This
isn't Thermalag; this is a different vendor.
So, we're seeing that same experience with
different vendors going back and looking at the
original qualification testing.
DR. APOSTOLAKIS: Was the licensee aware
of this fact, that, you know, based on the tests, the
rating was about 15 minutes?
MR. SALLEY: I believe the inspection, in
this case --
DR. APOSTOLAKIS: Does that come back to
what Doug was saying earlier about willful --
MR. SALLEY: I wouldn't say it's willful.
I would say, how hard do you -- do you look? I mean,
we operated a lot of years under the Thermalag where
we were under the impression that it was good until we
started really looking.
You know, just what did you test this to?
And just what was your configuration like in the test
compared to the plan? We got into all the details and
the rigor of the engineering --
CHAIRMAN SIEBER: But I think this one is
different than that. I think the Thermalag was
difficult to interpret the test results. And in fact,
I think there was a finding that some of those test
results were not accurate.
On the other hand, when somebody designs
a barrier system, you use the test results from a test
of the material and then calculate how much of it you
need based on the conditions you have in the room.
So, it, more than likely, is an error in
the application of the specific material to the
configuration, as opposed to misinterpreting the test
for a false statement, so to speak.
MR. SALLEY: We've -- we've --
CHAIRMAN SIEBER: That's the way I would
MR. SALLEY: Yes, we've got great
understandings into just how electrical raceway fire
barriers work. And that's a whole discussion --
CHAIRMAN SIEBER: That's right.
MR. SALLEY: -- that we've had numerous
times about the physics behind the barriers. But this
is all of that. The --
DR. SHACK: This is a latent design error,
is our best guess?
MR. SALLEY: That's an excellent way --
CHAIRMAN SIEBER: That's a good way to put
MR. SALLEY: -- excellent way to capture
it. The third thing, and getting back to the defense
in-depth of this, is the fire brigade. On this site,
they had a very good fire brigade. So, we would
expect the fire brigade to perform well within their
DR. SHACK: Just, again, what are the
questions the inspector asks himself to decide that
this is moderate degradation for the auto suppression
system and moderate to high degradation? How does he
pick those values?
MR. SALLEY: That's a very good question.
In the guidance that we provide in the Appendix,
there's numerous suppression systems. Not all
suppression systems are going to be created equal.
Let me just give you some examples here.
If you're dealing with yoru gaseous systems, your C2
and your halon systems, those are suppression systems.
That means that when they go off, they will put the
fire out. They are a pass/fail type of thing.
It's the little-bit-pregnant argument. I
mean, the system works or it doesn't.
A suppression system, a sprinkler system,
by its design, its original intent was to control the
fire. You know, it could limit it into an area until
manual suppression could come in and extinguish it.
So, you have those two schools of thought
in the fire suppression system design. Now, you get
into degradation. Let's take a sprinkler system. Say
a head had to be 12 inches from the ceiling, and, for
some reason, they installed them 15, 18 inches below.
Are they Code compliant? No.
Is it a degradation? Well, yes. Why is it
a degradation? Because the fire is going to have to
get a little bigger and a little hotter for the heat
layer to build on to actuate that sprinkler system.
Will the system go off? Well, it will
eventually go off. You may have a little more fire
damage, but it should be creditable.
With a gaseous system, it's not quite that
easy. In this case, the numbers are very close, so
we'd want to call that a moderate degradation.
Let's take that same CO2 system and say
the inspector found a check valve that was installed
backwards. Now, the system will get called upon, and
no agent would come on. So now, you definitely have
a high degradation.
Say he looked at the CO2 refrigeration
system and the tanks were empty; it's clearly a high
degradation. So, there is judgement calls. There is
engineering experience by the inspector as to what
category to pick. And usually, there's discussions
I'll give you an example of one I had --
a halon system in the past where it didn't make
concentration. The original design was for a surface
fire like you'd find in a flammable liquid.
You know, the argument the licensee put up
was, "Well, hey, we designed for a surface fire. We
really didn't anticipate a deep-seeded fire."
The only problem was the fire hazard was
the cable spreading room where all the fires were
deep-seeded and there was no surface fire to think of.
So, that's the kind of dialogue you'll exchange with
the licensee to get your degradations.
CHAIRMAN SIEBER: Now, generally speaking,
during the construction of the plant, or sometimes
during hot functionals, all of these systems are
tested, and the gaseous systems are tested, for
concentration. Is that not the fact?
MR. SALLEY: That's an interesting point.
Yes, they are tested. And sometimes, we are finding
systems, when we go back, and the inspectors are very
-- doing a very rigorous, thorough look at is just how
did your concentrations look and pulling the old strip
charges from the original design.
And we're fine, and some of it just quite
didn't make it, and maybe someone justified off, and
that's under question now.
CHAIRMAN SIEBER: Okay
MR. SALLEY: For example, if you missed by
a little bit, they said, "Oh, my problem was from
leaks over here, and I sealed those leaks. And I know
CHAIRMAN SIEBER: Or my calibration was
bad, and it deserves a correction.
MR. SALLEY: Right, and so those are
debatable things that still occur today, and they
CHAIRMAN SIEBER: But the reverse check
valve would have been found there, the fact that you
may not have enough suppressing agent in a tank that
would cause your system concentration not to be
MR. SALLEY: Right.
CHAIRMAN SIEBER: Okay, thank you.
MR. SALLEY: There has also bee some work
-- in 1986, we had a big study with Sandia on just how
much agent does it take to extinguish a fire? You
know, the National Fire Codes look at a broad band.
And deep-seeded to them is a cable fire in
a nuclear power plant; it's also a bale of cotton in
some other applications. They're all deep-seeded
fires by their definition.
We tried to refine it more to our hazards,
which were the cables. So, yes, we know the numbers
can be a little lower, and we have that guidance
CHAIRMAN SIEBER: Okay, thank you.
MR. SALLEY: So, that's the three key
points here of the defense in-depth in this specific
scenario. Knowing that and having assigned a rating
factor with that, we get back to the analysis portion,
which is where J.S. will pick it up, to how this all
comes together now to define some level of risk.
MR. HYSLOP: Yes. I just wanted to say
the documentation on those degradation levels are in
the public domain now. So, you can access that and
look at them for more explanation.
As I said before, in a fire risk analysis,
you're looking at the frequency of the fire, your
defense in-depth elements, and your mitigating
This first term, FMF, fire mitigation
frequency, really just deals with the frequency of the
fire and the defense in-depth that's left. Of course,
a fire which -- where the suppression system fail,
where your barriers should fail if challenged, you
know, these are fires that we're really worried about.
So, that's why we developed the FMF.
Now, the ignition frequency of the 4160
vital switch gear cabinet, we said that was the
cabinet in the center bay. So, it was an ignition
frequency associated with that cabinet that we're
concerned with for this analysis. And the IPEEE had
I'll give you numbers on the next slide;
I just want to talk generally right now.
The next terms, the automatic suppression
and manual suppression -- really, we had a manual
fixed suppression system here. So, "AS" was really a
manual suppression. We just didn't think about that
when we were writing the guidance.
But we take into account that it's manual
in the degradation rating, as Mark said. Manual
suppression, that's typically the fire brigade and any
type of early response that people -- that operators
would have to put it out.
DR. APOSTOLAKIS: But how -- I mean, these
things are not really modeled in the fire PRA. So, I
don't know how you can --
MR. HYSLOP: I'd like to get to that on my
DR. APOSTOLAKIS: Okay.
MR. HYSLOP: I'm going to talk about that.
Let me just talk about it generally, George, and then
DR. APOSTOLAKIS: Okay, no, no, that's
MR. HYSLOP: -- can get into the details.
DR. APOSTOLAKIS: That's fine.
MR. HYSLOP: And so, for the suppression
system, the manual suppression -- the manually
operated, fixed suppression system, which is "AS" and
the fire barrier, we had degradations. And we're
going to use those numbers in this equation.
Now, the fire brigade, everything was --
everything was great there. And so, we didn't have
any degradations, so we'll use a lesser failure
probability associated with it.
And we have this term, "CC". It's really
kind of like a common cause dependency term. There,
we recognize that, for some cases, if you have a
sprinkler system and you have a fire brigade, those
common delivery systems can introduce common cause
failures; fire -- your fire water pumps, you know,
it's a pressure for each one of those.
So, we recognize that there is an
additional failure mode in there, and we've taken it
into account. For this particular case, it was a
gaseous system, so it wasn't an issue.
Now, I've got the numbers on this page,
and then I'll explain them to you on the next page.
All I want to say is these are the numbers that we
attribute for the various degradations.
And the fire mitigation frequency
essentially says we have a factor, a 10-5, leading
into the mitigating systems. So, you know, we don't
have to have a lot of mitigating systems to derive us
to a green here. If we have none, then we're in white
Let's move to the next slide. And what I
want to say is, these numbers really are coined as
exponents of ten. Remember Doug had the 1, 2, 3, or
4, or whatever; well, you know, these are exponents of
10. So, "-3" is 10-3.
DR. APOSTOLAKIS: You have not included
transient fuels, have you? This is just --
MR. HYSLOP: There weren't -- to my
knowledge, there weren't any transients found during
DR. APOSTOLAKIS: But if you want to have
a frequency of fire --
MR. HYSLOP: You're talking about having
a probability of transient --
DR. APOSTOLAKIS: Yeah.
MR. HYSLOP: -- fuels, even though -- we
haven't included that That's something that we're
going to include in the next version.
We're going to be providing, in the next
version -- this is another thing of the evolution --
a whole set of ignition frequencies for inspectors to
use when the plant doesn't have them because some
IPEEEs didn't go to this level of detail.
They said, "We've got a room. We've got
suppression, and we've got some severity factors."
So, they never got into this.
DR. APOSTOLAKIS: Right.
MR. HYSLOP: So, that's going to be in the
next stage of this tool.
DR. APOSTOLAKIS: Okay.
DR. FORD: So, where do these numbers come
MR. HYSLOP: Turn to the next slide, and
I'll tell you.
DR. FORD: Oh, okay, you will tell us now,
MR. HYSLOP: okay, this table provides the
origin of these numbers. The top column -- the top
row of this table identifies the defense in-depth
elements. And I checked that we just had a one-hour
barrier, an automatic suppression -- or really, a
manually initiated one -- and a fire brigade for this
DR. APOSTOLAKIS: So, this is not just for
MR. HYSLOP: No.
DR. APOSTOLAKIS: This is something from
MR. HYSLOP: This is -- this is generic.
And I'll talk you about, you know, the source of
these, George, and how -- but let me get there. And
so, the first column talks about the level of
degradation, and we have three levels of degradation
in this technique -- you know, you might say two
levels, the moderate and the high.
The normal operating status when we --
when rate something, we find it meets Code typically.
But we still have some sort of failure probability
associated with those.
So, if we start talking about these, you
know the first question is, where did these numbers
come from? Is there any reference for these numbers?
Did I have to develop them? You know, what's the
And if we start for the three-hour barrier
for the normal operating state, we had -- new Reg 1150
developed these sandia during their preparation for
the study, I guess.
And in this particular study, these said
that a wall, three-hour barrier, fire-rated three-hour
barrier, had a one in a 1,000 chance of failing. That
was the base probability associated with it.
Now, if you had additional -- if you had
dampers or doors in that wall, they collected data to
support the unavailability of the door -- the doors in
that wall. And that's what would drive the failure
probability for the normal operating state.
CHAIRMAN SIEBER: Does that mean the door
is sometimes left open, or blocked, or does it mean
the door is really a three-hour barrier.
MR. HYSLOP: It means that if the door is
left open or blocked.
CHAIRMAN SIEBER: All right.
MR. HYSLOP: That was -- that's my
understanding. Let me tell you that there wasn't a
lot of documentation in the new regs on these. And I
told you that we we're working with the Office of
Research. We've asked the Office of Research if they
have any more insight to give us on these failure
probabilities in this table, we're very interested.
It's evolving and it's a state-of-the-art
CHAIRMAN SIEBER: Now, the plants usually
keep track of missing fire barriers and blocked doors
and things like that as part of their fire protection
monitoring system. So, there's a source of plant-
specific data for that that could be used, I guess, if
the licensee wanted to contest what you were doing?
MR. HYSLOP: Yes, the -- as Doug said, you
know, we have a phase three process. I'm talking
about the phase two. The licensee in any -- in any
and all of this study has the opportunity to present
additional information to refine the results.
Here, we've tried to provide generic data
so that we can get through the process.
CHAIRMAN SIEBER: Okay.
MR. HYSLOP: Now, if you go to a three-
hour barrier, one that has a high degradation, they
are -- the zero means that we're not giving any credit
for it. And the plant system documentation would
support minimal credit for this particular high-level
degradation of a three-hour barrier.
DR. APOSTOLAKIS: So, the inspector is --
I think Bill asked that question -- is provided with
information or guidance, how to declare something as
moderate or high?
MR. HYSLOP: Yes.
MR. SALLEY: Yes, if I could jump in,
J.S.? We have another example of a case we're working
right now. In an area, it was required, for their
original Appendix R compliance, to have a three-hour
box built around a number of cables that had
penetrated into an area.
They didn't need -- they needed to rely on
this A-train, we'll call it, inside the box for a fire
in the B-train area. So, by regulation, it was always
required to have a three-hour enclosure around it.
And the licensee was moving along,
thinking it was pretty good. The inspector went out;
the inspector was looking at it and said, "That box up
there," -- they said "Yeah, three-hour barrier for
He said, "Great, can I see the test
reports for it and the design basis?" The said,
"Sure." So, they started digging through it. When
they got into it deep, they really didn't have a test.
It sounded like a good idea to take these
non-combustible boards and assemble them here. And
they've existed that way since the mid-80's and have
taken the three-hour credit for it.
Now, we got into a discussion with them,
and what kind of a credit could we assign to this?
Well, we have no testing. We just know that it was a
CHAIRMAN SIEBER: It's zero.
MR. SALLEY: Right. Will the bolts fail,
and the box fall off, even if it's non-combustible?
They've covered it with phlomastic, which is a limited
So, to enter into this, it entered in as
a high degradation, zero. The licensee then, because
we got the zero, started working through it, built a
mock-up of this at a laboratory and tested it, and
found that it got approximately one hour.
So, we, in the analysis, further refining
it, went from the high degradation to, here, a
moderate because they did have some creditability to
that box after having tested it.
Once again, it was a good inspection
finding to go and look at that.
CHAIRMAN SIEBER: So, they had to conduct
a special test to even come up with that?
MR. SALLEY: Yes.
CHAIRMAN SIEBER: Okay. Now, let me ask
another question. And again, referring to "door" on
there, do I interpret that to mean that any three-hour
door in the plant that's expected to be open for 30
days a year?
MR. SALLEY: The door thing, I just want
to -- I understood a little different J.S., if I could
expound upon that. We give you three levels there.
We give you a -2, a -2.5 and a -3.
CHAIRMAN SIEBER: Okay.
MR. SALLEY: Not all three-hour fire
barriers are the same. For example, if I wanted a
perfect fire wall, I'd have 12 inches of concrete
poured, solid pour, no penetrations, no doors, no
nothing. I'd have a lot of confidence, and history
has proven that, that that's a pretty good three-hour
However, in a power plant, if I introduce
a door, well the door doesn't test the same as a fire
wall. The door criteria is much more lax, just by the
nature of the door. I mean, you have gaps.
If you have some flaming remote on the
other side, it won't perform as good as the wall, but
it's still a recent --
CHAIRMAN SIEBER: It's still a three-hour
MR. SALLEY: -- it's still a three-hour
barrier. You need to have that in there. If this
wall had numerous penetrations, I'd need to look at
those penetrations? And do I have tests? Do I have
designs? Do I have a comfortable feeling with all the
So, that -2 to -3 gives the inspector some
room to customize it for his application in
determining the --
CHAIRMAN SIEBER: It sounds to me a little
MR. SALLEY: Engineering judgement.
CHAIRMAN SIEBER: That's another way to
phrase it. Let's move on.
MR. HYSLOP: Okay, so basically, the
moderate degradation is a twist of a value between the
high and the normal operating state. And we all -- to
get to the values, you know, we used, we looked at the
one-hour barrier, the fixed suppression system and
The one-hour barrier in the normal
operating state was taken to be approximately equal to
a moderate degradation of a three-hour, in that, you
know, a moderate degradation of a three-hour is
somewhere between two and one hours. And so, that's
what we expect for a normal operating state for a one-
And then, the logic is similar for the
moderate and high degradations of the one-hour, as was
for the three-hour, the basis for the choices.
Now, if we talk about the fixed
suppression system, there, the normal operating state
of that is also taken from many studies. I know it's
in five, and I know it's in the PAR Implementation
Guide, the basis for this number.
So, there, that's judged as a normal
operating state. And again, for an automatic
suppression with -- where we have some degradation
which drives us to conclude that there's minimal
credit, we give it zero.
Now, the last one we talk about is the
fire brigade. And really, it's a manual suppression.
That's really what that is, is a manual suppression
there. Let's be quite frank, because if you look at
the fire brigade, you notice, for a high degradation,
we give credit.
And that's because there are fire watches,
there are operators going around a plant, and there's
data found in the PAR Implementation Guide that
supports that these people do put out some fires
before they get bad. So, we have some credit there.
The -1 there is -- it's often scenario-
important. But for cases where the IPEEEs looked at
lots of fire sources creating severe fires, .1 was
typically used in those analyses to support that. And
that was really the origin of the number here.
Let me see if I have any other comments.
So essentially, you know, I guess to sum up, some of
these normal operating states are supported by
industry, or NRC, or both, and guides. And the other
values were kind of deduced from common sense, good
judgement. Go ahead.
So, now, I'm going to move into the
reactor safety portion of this because we've
identified the fire mitigation frequency. This is the
-- these are the fires where -- which have the
opportunity to get big. Our suppression system hasn't
worked, and so we have some elements of our defense
in-depth that are going to fail to control this fire.
So, what I'm going to -- let's move to the
DR. FORD: Excuse me. The only data in
this whole page 37 is this Sandia new Reg 1150, the
previous one. The only hard data that you have --
MR. HYSLOP: Well, we have --
DR. FORD: -- 10-3 is --
MR. HYSLOP: No. Well, that was adopted
by industry also. So, that's a number generally
accepted in the PAR community. I think that was
derived in the 1150 studies. I don't know if industry
did any additional work before accepting that.
This is one of the things I've identified
to Nathan Siu, of the fire research plan, that we're
interested in having additional information on
because, you know, we recognize that this is one of
the -- one of the area that -- you know, there's a
limited information on fire on which to make our
DR. FORD: But that's my point; the only
referenceable data is that 10-3?
MR. HYSLOP: Well, there -- no, there --
no, that's referenceable also in either the five or
the PAR Implementation Guide, both of which are
DR. FORD: Okay.
CHAIRMAN SIEBER: I would like to suggest
that we're going through the basic principles right
now of how this worked, but you do have a specific
MR. HYSLOP: Okay.
CHAIRMAN SIEBER: And maybe we can do that
after lunch so that we can get through with the
general explanation and let us know.
MR. HYSLOP: Actually, we're doing the
example, but we're almost finished with it. So, I
CHAIRMAN SIEBER: Well, it looks like a
lot of sheets.
MR. HYSLOP: Well, that's okay. I can do
those in five -- in five minutes, and that's what I
intend to do. I'd like to -- if you don't -- whatever
you want to do.
CHAIRMAN SIEBER: I think this is a good
place, then, to stop before we get into all this
MR. HYSLOP: Okay.
CHAIRMAN SIEBER: And even though it's
extremely interesting --
MR. HYSLOP: Okay.
CHAIRMAN SIEBER: And why don't we recess
for lunch and come back at one o'clock? And we'll
finish this up then.
(Whereupon, the proceedings went off the
record at 12:08 p.m. and resumed at 1:02 p.m.)
CHAIRMAN SIEBER: We'll come back to order
and continue with Fire Protection SDP.
MR. HYSLOP: What we've done, just to
remind you, was to calculate a fire mitigation
frequency which was the frequency of the fires for
concern were those that aren't extinguished or
controlled by suppression and those which challenge
What I'm going to do is move on to an
evaluation which involves the reactor safety
worksheets. Doug Coe, in his presentation, talked to
you about an application of those sheets and I'm going
to talk to you about a different one.
Let's move to the next slide.
MR. HYSLOP: The next slide is the
worksheet for a small LOCA and the reason we're using
the small LOCA is because it's fire induced. As you
recall, we lost all the electrical trains. Losing
those electrical safety trains means that we lose our
component cooling water and our charging system and
losing both of those induces a small LOCA or RCP-Seal
LOCA. And this is consistent with the assumptions
used in the reactor safety process.
So now we have to say how significant is
that fire induced small LOCA? And if you look at the
sequences which are here on the left most column,
you'll see that first of all one sequence which leads
to core damage is a small LOCA and the loss of all
high pressure injection. In this scenario, we're
looking at the charging pumps and I think the SI
trains injected a slightly lower pressure, but you
could have some depressurization, therefore some
mitigating capability if they were available.
But upon losing all these electrical
trains and mechanical trains you, in essence, lose
your high pressure capability. So the reason I said
we could get through this quickly is because we really
give no credit for the mitigating capability in this
particular scenario. So our fire mitigation frequency
which had no reduction because of the length of time
that these degradations existed, essentially serve to
characterize the increase in core damage frequency
fully. And for this example, we get a white and you
would go through the same tables as Doug did. I've
just short-cutted it.
MR. HYSLOP: So if you go back to that
earlier slide it says that the resulting evaluation is
Now what would happen if we had felt that
that fixed suppression system wasn't worthy of any
credit at all? If you remember, it was immoderate,
based on the observations that the inspectors made.
Then it would be in a yellow territory. As you've
talked before, the yellow provides a different
response than the action matrix as does the white. So
we would have geared up a little more for this one.
If we repaired the fire barrier, for
instance, then we would have been pretty much at a
green/white threshold in that case and depending on
exactly where we were, we would have gone -- we may
have gone with a white for that because this is a
conservative approach and then allowed the licensee to
come in with a refine analysis to support his work.
So that's it for my presentation and
MR. COE: If I may add one thing, it's
important to make the point that the SDP process has
not removed the requirement for the staff to make
judgment and as you've seen here with the fire
protection as well as the earlier presentation that I
did, the judgments are now more rigorous, more
disciplined by this framework that we've chosen to
use, but in essence, there are still judgments and
they occur at the assumption level or at the basic
input level for these SDPs and the logic that then
processes those assumptions to a final result is clear
and is apparent to all of our stakeholders and is then
the subject of dialogue and discussion. So I do want
to make the point that we have not extracted judgment
from this process.
MR. SHACK: What's the feedback you get
from the inspection people about whether they feel
they can make these judgments?
MR. SALLEY: Can I take that one? In the
fire sense, let me pick that up and explain one other
thing that's kind of important if we go back to our
example. Now remember, this process is new. It's
evolving. J.S. told you that. We're getting better.
We're refining, we're doing, we're learning. If you
think back to your question in this case here with
judgment and such, there's one question that we just
kind of glossed over and it was done this way in the
actual example because it was so early on and it was
looked at and that is what's the fire potential, okay?
If I could argue from a licensee's standpoint and say
well, okay, yeah, it's a 10 minute fire barrier, but
when I go through the dynamics of combustion here, I
get a 6 minute fire, worse case. So that gives us
room to argue and move around within the evaluation.
One of the things that we're currently
moving on and this is the way we're seeing them now is
what's the realistic fire threat. As the newer SDPs
are coming in the findings, that seems to be one of
the up front questions. Could I have had, do I have
the chemistry there to give me the credible fire to
challenge these degraded barriers' suppression
systems. And that's where we're moving with the
effort now. With the inspectors, one of the things
that is -- if I for example say a gallon of
combustible liquid, in each one of our minds we
picture the fire that could be. It could be in a
kerosene lamp and you've got a hurricane lamp or you
could spill it all at once and get a big burn. How do
we make those judgments and that's what we're working
on. We have a quarterly workshop with the inspectors
to review the cases and J.S. goes through the cases
that we've been through in the last quarter and we're
starting to introduce some of these new tools and
methods on how to do the fire scenario development.
That's what the process is currently today. That's
what we're working on and going through the
development methods is where we start all getting the
same judgment and the same experiences, learning from
the different ones.
CHAIRMAN SIEBER: Now that may change the
significance of a given set of circumstances. It
doesn't change the fact that you would still be in
violation of Appendix R which is deterministic to the
violation whether it's cited or noncited or whatever
color it is, it still exists.
MR. HYSLOP: And as we've said, any and
all of those still go into the corrective action
program. They need to be fixed.
CHAIRMAN SIEBER: On the other hand, it
seems to me that the development of risk-based fire
analysis is not too far along. If I look at NFPA 805,
it discusses that to a great extent, but it seems to
me that that is in addition to the deterministic
requirements of Appendix R or branch technical
position 9.5.1 or the guidelines or whatever class of
plant that you're into. And until such time as you
risk-inform Appendix R, if you ever do it, where it
tells you, you don't need a one hour, three hour fire
barrier, you need a 20 minute or a 60 minute or a 90
minute fire barrier based on fire scenarios and risk
probabilities, it sort of puts us into an enforcement
juxtaposition into what the regulations tell us to do,
it seems to me.
Can you comment on that at all?
MR. SALLEY: The risk-informed performance
based approach, I see the SDP portion of this is
moving in the right direction and being fairly
valuable. For example, in the past, if you were to
just find the CO2 system, didn't meet its design
concentrations and the fire barriers didn't either --
CHAIRMAN SIEBER: It would be a Level 4.
MR. SALLEY: Right. At some point in
there someone would say well, how bad was it and some
engineer would walk out there and say well, you know,
we've got this switch gear and from what I've seen a
switch gear fire -- and it would be an opinion. A
pure opinion. It's going to be real bad or it's not.
Here, we at least are starting to
framework and say okay, from a risk standpoint. How
bad would it have been? What would the possible
outcomes be and we have a nice structured framework to
make a better determination. So I see it as a real
CHAIRMAN SIEBER: Now the example you
describe here is a Phase 2 analysis under the SDP.
What circumstances would cause you to do a more
rigorous analysis and if so, how would you do it?
There's a step beyond this, right, as far
as the degree or rigor?
MR. SALLEY: Right. And I guess we
haven't seen a whole bunch of Phase 3s, but one of the
areas that I've seen them go to is we go into the fire
CHAIRMAN SIEBER: Using what tools?
MR. SALLEY: That depends. You know,
C-FAST is a common piece of software put out that
people like to use and make approximations with. So
you would start seeing the fire modeling come into
more -- but also you would see in a Phase 3 from my
experience J.S., and please correct me, but the issue
of fire frequency, okay, people wouldn't want to say
what's the fire frequency of the room or what's the
fire frequency of that specific piece of --
CHAIRMAN SIEBER: Equipment.
MR. SALLEY: Equipment. And you'll see
that things -- the fire frequency can change orders of
magnitude, you can change colors. I like the colors
like -- you guys want to keep them green. I want lime
green and dark British racing green.
CHAIRMAN SIEBER: If you change by a
factor of 10, you change colors all together. You go
from a green to a white to a yellow to a red, right?
MR. SALLEY: You'd see more rigorous fire
dynamics development. You'd see more rigorous on the
fire frequency of the specific component rather than
average or an area and I think between a Phase 2 and
a 3 you would see things like the licensee taking it
serious and going to perform a test to see what
grading does that barrier really have. The NRC has
given us zero and we can't argue with their zero. So
they'd go out and try to get some hard number for it.
MR. HANNON: This is John Hannon. I'd
also add that there's an effort, we have underway now
at the NRR staff to look at the fire events database
to update that and that might provide more current
information. Think of using the SDP as far as fire
event frequencies, initiation frequencies.
CHAIRMAN SIEBER: Yes. Now is the
methodology you would use to do a Phase 3 analysis in
a fire protection area proceduralized or documented or
is this whatever you decide you want to do kind of
MR. HYSLOP: I haven't done any Phase 3s
associated with this. I've looked at a couple of
utility ones. We currently need better Phase 3
guidance and that's one of the things that we've asked
the Office of Research to provide us as a result of
In general, your technique is the same,
given the things that you're considering: frequency,
defense in depth and mitigating systems. I suppose if
someone could develop distributions they could think
of something other than the mean and incorporate that.
I don't know of anyone who has done that.
So I really don't have a very good answer
to your questions.
MR. COE: When you're talking about Phase
3 guidance, you're talking really about what kind of
standards exist in the general field or practice of
probabilistic risk assessment.
CHAIRMAN SIEBER: That's true.
MR. COE: And you may be aware that ASME
is working on some standards that the NRC is
participating on that committee with and they should
be coming out with a set pretty soon.
CHAIRMAN SIEBER: Well, they actually have
published a standard, but that's for regular PRAs, you
know, the very comprehensive ones. And it doesn't
seem to me, as I recall that standards that it tells
you specifically what models to use, what assumptions
to make, where you get your data from, how you derive
all these quantities that go in there. The fact that
it doesn't even describe initiation frequency, defense
in depth, mitigating systems or any of that. It's
sort of in the eye of the beholder at this point,
MR. COE: Exactly, and I think that the
process that we've devised here is one that helps the
decision makers of this Agency that are about to make
a risk informed decision, better understand the
assumptions that went into it. And I don't know that
that would change necessarily whether you're doing a
Phase 2 analysis or a Phase 3.
A decision made on the basis of a Phase 3
analysis should be just as understood in terms of the
influential assumptions that were used as a Phase 2.
CHAIRMAN SIEBER: I would think one reason
why you would go to a Phase 3 is because your Phase 2
analysis was challenged and that being the case, then
why not challenge the Phase 3 analysis?
MR. COE: In any case, what this does is
foster better discussion and a more focused discussion
between the staff and the licensee, typically. I've
seen this play out because anytime an issue is
characterized as greater than green, it comes to a
panel at headquarters. And the panel, subject to the
panel, is whether or not we are applying the SDP
process consistently and inevitably the discussion
gets down to the level of confidence that the staff
has and the assumptions that are most influential to
the result. And then when we discuss this with a
licensee, again, it focuses our discuss on those
assumptions which are most influential to the result.
And I think it's a more efficient way of processing,
of communicating with, both internal to the staff as
well as external.
MR. JOHNSON: But we do hear your question
and it's a good question.
CHAIRMAN SIEBER: Yes. I guess the other
thing that I'm thinking of is there really aren't a
lot of fires in power plants if we ignore waste basket
fires in some outbuilding some place. On the other
hand, there is talking about mining for noncompliance,
there's a lot of opportunities just due to the
complexity of the regulations to find design
deficiencies and testing deficiencies and so forth.
I mean you could really make a living doing that.
So I see the potential for enforcement
actions, noncompliances, noncited violations, what
have you, being always there.
MR. COE: We hope our inspectors are
sensitive to and looking for the most significant of
those because I think anybody could agree that as
large and complex a facility as these are, there will
definitely be some level of deficiencies that exist
all the time and the licensee should be identifying
and correcting those and our interest would be in
identifying those that are of greatest significance to
the public health and safety.
CHAIRMAN SIEBER: And that's what this
process is intended to do.
MR. COE: Is to focus our efforts as
CHAIRMAN SIEBER: Okay. I think that
clarifies that for me. Why don't we move on.
MR. JOHNSON: Okay. All right, Don will
you come up?
We're continuing through the presentation.
If you look in your packages, we're going to shift
gears now and talk about performance indicators and
Don is here and we hope to be joined by Garrett Perry
shortly to talk about a number of issues with respect
to performance indicators.
The first topic that I wanted to cover was
to talk about thresholds in a very general sense, just
to refresh your memory with respect to what we
intended to do with thresholds, not just performance
indicator thresholds, but thresholds in the ROP.
We're then going to talk about the process for
developing thresholds and I think there was some
interest in having us look at mitigating system, an
example of how we set those thresholds, so we're going
to do that, right, Don?
MR. HICKMAN: Yes.
MR. JOHNSON: And then last, but not
least, we're going to talk about PI reporting so you
understand a little bit of the mechanics of how we get
this PI data to the Agency.
Just by way of providing some explanation
or some reminder, if you will, about what we were
trying to achieve with thresholds in the ROP, again
and I made this point earlier, when we set out to do
the ROP we had the notion, in fact, industry very much
wanted us to recognize that there needed to be some
licensee response band. We weren't going to be able
to achieve zero defect. That was an unreasonable
expectation. There, in fact, needed to be some area
with which the licensees could operate their plants
and have problems, but that wouldn't warrant
necessarily an increase response on the part of the
regulator beyond what we do with respect to doing sort
of a baseline level of inspection at all plants to
make sure we have the necessary information along with
performance indicator information to begin to get an
indication about the performance of plants.
So there was this notion of a licensee
response band. Well, in order to make that work we
set up a series of thresholds and those thresholds
really serve as trigger points, if you will, for us to
take increased regulatory response.
Again, the greater the degradation, the
more thresholds, the more significant the threshold
trip, the greater the regulatory response and we'll
talk about the regulatory response when we talk about
the action matrix in July.
I do want to make the point that the
thresholds aren't intended to be predictive. And in
fact, we don't even like to use words like leading.
And in earlier presentations for the ACRS and in
multiple presentations, earlier presentations
throughout the development of the ROP, we have
typically gotten the question, are the thresholds
leading, are performance indicators leading and every
time we try to come back with a response that goes
very much like we don't guarantee, we don't believe
that it's appropriate for us to say that we can
predict or present an occurrence of an event. We
can't predict necessarily that at Plant A, whose
performance is at X level today is going to be at Y
level in a year from now. That's not what we set out
to do when we set the thresholds.
What we set out to do when we set the
thresholds was be able to trigger ourselves early
enough in a way that would enable us to take timely
action because what we don't want to have happen, we
don't have to have plants go into that unacceptable
column of th action matrix. We're talking about that
far right column of the action matrix where we've lost
confidence in their ability to maintain the design of
the plant. And we've got some words, some high levels
words that were taken from the order, from things like
-- like words we wrote in the Millstone order, for
example, where the Agency has lost confidence in the
ability of the plant to -- the licensee to operate
that plant safely. And so the thresholds are intended
to allow us to trigger, to respond in time to interact
before a plant would go into that column.
So we talk about timely, we talk about
thresholds as enabling us to take timely action where
we see these performance declines happening. And
thus, that's what we were trying to do with respect to
I guess I just wanted to pause for a
second and let us talk about thresholds before we go
further because I know there was and has been,
continue to be questions about what we were intended
to do with respect to the thresholds.
MR. BONACA: I don't want to belabor it,
but it's hard to believe that you can take timely
action if you don't have some leading indications that
you can work on. That's my comment.
You're saying on the one hand you don't
intend to have leading indicators. I can accept that.
Then on the other hand you say you want to be able to
have indicators that will give you the opportunity to
have timely action which means take action before
things happen. So that in and of itself implies you
expect them to be leading. So I don't know where
you're going with the two statements.
MR. JOHNSON: And it's sort of timely --
that's a fair comment. It's sort of -- is it timely
or is it leading to what and this is kind of the
discussion that we have.
One of the difficulties with the current
thresholds in some people's minds is that with respect
to the low level issues that they see at a plant, you
can get into -- some people firmly believe that they
can in terms of thing that you begin to see
indications, low level indications of human
performance, low level indications with respect to the
way licensees find problems or treat those problems,
that those provide an early indication, if you will
and if the licensee doesn't fix those, they're going
to end up with a problem.
And I guess I'm trying for a shift in
mindset. The old process used to have us look at
those issues and react to those issues. We often drew
conclusions based on a predominance of those kinds of
things, extrapolated them to say hey, if you don't fix
these things, licensee, you're going to end up on the
last list and the problem with that is that we
predicted about twice the number of plants that
actually ended up on the last list based on an
approach like that because what actually happens is
that at a very low level, unless you actually see
thresholds, unless you actually get to a point where
thresholds are being crossed, much of what you see to
cause you alarm because you never know whether what
you're seeing is a tip of the iceberg or it is, in
fact, what it is and there's not much beyond it.
And so again, the rigor of the thresholds
is to try to say if there are performance problems, we
want to have the threshold set low enough so that we
can trigger response as those performance problems
begin to occur, but again, if you have problems that
don't even reach that threshold, we're going to --
those fall in the licensee response band.
That's the balance I'm trying to strike
when I draw the line between what is timely. The
notion of being predictive, I mean we've had, you'll
remember maybe a couple years ago or three years ago
or so in response to a direction that we got from the
Commission, then I think the EOD looked at financial
indicators and the notion at that time was that
financial indicators would be an example of something,
a type of indicator that would be predictive. And
that the ACRS, at the strong urging of the ACRS, among
other stakeholders, we backed away from that approach
because again, what you would seize upon in terms of
being predictive could give you bad results, you could
end up seizing on something and thinking that you were
getting a valid prediction and in fact, you weren't
getting a valid prediction at all.
So again, the emphasis on the thresholds
was to allow us to recognize performance problems and
begin to interact the action matrix providers with
greater responses early on because again what we don't
want to happen is we don't want to have a plant where
tomorrow we decide for ourselves that that plant is
unsafe. We want to have had an opportunity to engage
and we think that engagement has to happen though
through results, performance issues that reflect
themselves and especially as they cross thresholds to
the SDP or performance indicator issues that cross
thresholds that we've set up.
MR. HICKMAN: If I could add to that just
a bit. The old AEOD performance indicators were
sometimes criticized and we ourselves also wanted to
try to make them predictive, leading. I know,
criticized for the fact that they were. That is very
difficult to do because you have to look at programs
that will ultimately reflect in performance at the
What those programs operate through people
and you never can predict how people will react to
programmatic weaknesses. Instead of trying to make
them predictive, what we always said we were trying to
do was to try to make them as responsive as possible,
as quick reacting to changes in performance at the
plant so that we could identify that as early as
In fact, we did some comparisons of the
trends of PIs against Agency actions, senior
management, meeting actions and things like that.
Putting on the watch list and those kinds of things.
And that was kind of rather informative.
But we want to be as reactive as possible,
particularly for this program because one of the
premises of the program is that if there's a risk
significant problem at a plant it will eventually turn
up in performance at the plant. If it doesn't do
that, then we say it's not particularly risk
important, if it doesn't reflect in some kind of a
performance at the plant. So we're looking for those
kind of performance problems to show up and we want to
identify them as soon as possible so we can step in
after crossing the first threshold into the white band
and try to take some action to prevent them going
further. That's the whole premise of the program.
MR. KRESS: Since George is not here, I'll
try to articulate a couple of questions that I
anticipate he might have asked about this slide. One
of them would be looking at the second bullet and the
third, delta CDF due to some change in these
performance indicators are likely to be plant
specific. How do you know that these are the numbers
that would be generic? How do you arrive at a generic
number for what is like to be plant specific? That's
The other question is what's the rationale
for choosing the 95 percentile for the first
threshold? Why is that a good number to use?
MR. JOHNSON: Okay, I'm sorry, was there
a third question?
MR. KRESS: Those two right now.
MR. JOHNSON: We actually were going to
get to those. Don was going to talk through the
actual process for developing thresholds and when we
get joined by Garrett Perry and I know Don's been
anxiously watching the door for Garrett to come in,
Garrett was involved in the original setting of the
thresholds. We'll talk about those issues.
MR. HICKMAN: Yes, we'll get into both of
those. We'll start with the first bullet.
The green-white threshold, the concept was
to identify plants with performance as an outlier to
the industry. We didn't go into this development with
the concept in mind of 95 percent or two standard
deviations or anything like that. When I show you
this slide, I think you'll see that it's very obvious
where the thresholds should be set and maybe I guess
we should go into that one right now.
MR. HICKMAN: This is an example of what
we did. This happens to be the safety system
unavailability of the aux. feedwater system.
Remember, now that we did this in the fall of 1998 and
so we took the most current full three years of data
that we had, that was 95 to 97 and this was -- we did
all this in concurrence with -- in agreement with the
industry, represented by NEI. We said that we would
take those three years and make them our baseline. So
we collected this data over that period for the best
data we could get for each of these PIs.
In this case, for our safety system
unavailability indicator we used the same definitions
that WANO had been using for many years. So they had
been collecting this data on a quarter by quarter
basis, taking 12 quarters and summing them up,
calculating the indicator.
We had three years worth of that. We had
12 quarters worth of that data. We took every plant,
in this case all of the PWRs, there's 71 on here. We
took the highest value during those three years and we
plotted it and that's what you see. It's the worse
case value, the highest unavailability of that system
for each of those plants.
MR. KRESS: Now, if I were going to draw
a line as a threshold through that, I would have
dropped down to the next level, instead of the one you
have because there's, to me, it looks like two modes,
two mode distribution. I would make the line right in
between the two modes. I don't see a real rationale
for the line you have up there.
MR. HICKMAN: That's set at 2 percent
which is the current threshold. If we had dropped it
down to the next line, that's 1.5 percent. I guess
you could argue about that. We looked also at the
number of plants, two things we looked at. One was
that there was a clean break. You didn't want to have
a plant, two plants slightly, very small difference
apart, but on opposite sides of the threshold.
So we looked for a gap. And as you point
out, it could have gone either place.
MR. KRESS: Yes.
MR. HICKMAN: We then also looked at the
number of plants. And this is not a hard and fast
rule. It wasn't like 95 percent was a hard and fast
number. It was of that order. And so we captured
five plants setting it at 2 percent, out of 71 in a
three-year time period. If we had dropped it down we
would have gotten 13 plants.
MR. KRESS: I don't understand why you
MR. HICKMAN: This one is probably a
little more controversial than some of the others.
Most of them were very clear where the threshold ought
to be. This one we could argue about whether it's 1.5
percent or 2 percent. You're right.
We felt that 5 plants was better perhaps
than the 13.
MR. UHRIG: You say this is just PWRs?
There's a hundred and some odd plants there, unless
I'm not reading it --
MR. HICKMAN: Well, the numbering system
is kind of strange. These are the graphs that we got
from NEI. They provided this data. And the numbering
isn't quite right. But if you count the bars, it's
MR. UHRIG: Okay.
MR. HICKMAN: If all the plants are there,
then there would be that number, but they're not all
there. It's confusing.
MR. BONACA: Again, this is not plant
specific at all. What I mean is that it doesn't
recognize the --
MR. KRESS: That was the other thing --
MR. BONACA: -- importance, the importance
of the unavailability to the specific plant.
MR. KRESS: Right.
MR. HICKMAN: That's correct.
MR. BONACA: Okay, so it doesn't recognize
MR. KRESS: It may be that that plant that
shoots up there has always been there and it didn't
MR. BONACA: Maybe there is another system
CHAIRMAN SIEBER: It might have five
MR. HICKMAN: We recognize that. We have
had many discussions about this. There's actually
four indicators per plant on the safety system
unavailability and we're undertaking a major effort to
kind of overhaul this. And of course, as George keeps
reminding us, we're aiming towards the plant specific
PIs, the plant specific thresholds. That's the goal.
MR. BONACA: This is a good effort there.
MR. HICKMAN: It's going to go from the
MR. HICKMAN: Let me go back to this slide
again. Now Garrett can talk better about this because
he did this and I'm not a PRA person, but basically
what he did was to take some generic vendor models
that we had. He used the old SPAR models, not the new
rev. 3 models, but the old one. And there were just
a limited set of those, I think about a dozen or so
and those were essentially vendor types of models for
the various configurations of the vendors for
Westinghouse two loops, three loops, four loops,
He then ran this parameter, varied the
parameter that we were monitoring to get a change in
CDF of 10-5 for the white/yellow threshold. And he
did that for each of the models and if you look in
Appendix H of attachment 2 to SECY 99007, that's
Garrett's appendix where he describes how he set these
thresholds and there are tables in there and it will
show for various plants representative of each of
these models what the numbers were. And essentially
what he did was to take the most conservative number,
the smallest number.
MR. KRESS: That's how he got around the
plant specific part of that.
MR. HICKMAN: Right. So to make sure that
essentially every plant was covered. If you read it
carefully, you'll see there's a few holes in there and
there's still work to be done on the thresholds, but
that was the basic approach.
The same thing was done for the yellow/red
threshold, but adjusting the parameter to get a delta
CDF of 10-4.
MR. KRESS: That's using the old SPAR
MR. HICKMAN: Yes. Right. Right now we
MR. KRESS: It's kind of group plants by
MR. HICKMAN: They're just vendor models.
MR. KRESS: There's one that's treated as
one type of plant?
MR. HICKMAN: Right, they're pretty
generic vendor models, but there's a particular plant,
I guess that it gets modeled after and they're listed
in the tables in Appendix H.
MR. KRESS: That represents these plants?
MR. HICKMAN: Yes.
CHAIRMAN SIEBER: I take it you couldn't
do that, use that technique for the green-white
threshold because almost all plants would be white
MR. HICKMAN: With green-white it would be
CHAIRMAN SIEBER: You would have -- all
you'd have to do is have one failure and you would be
white, a CDF at 10-6, right?
MR. HICKMAN: But as Garrett points out in
Appendix H, this method worked well because you'll see
that there is still quite a gap between the
green-white threshold and the white-yellow threshold.
So by going by outliers from industry norm, we think
we have a pretty good threshold. It gives us a decent
green band for the licensees to operate in and it
gives us a white band for us to react and to try to
prevent further degradation of performance. So it did
work out pretty well.
MR. KRESS: This is a one time fixed
event, threshold and it won't be adjusted later?
MR. HICKMAN: We set the thresholds this
way prior to the pilot program. At the completion of
the pilot program we looked again and we did make some
adjustments. Actually, it wasn't based on the pilot
data because we only had 13 plants, but when we got
the initial input from the entire industry, giving
their historical data, that's what we looked at and we
did make some adjustments based on that.
In some of the safety system
unavailability indicators, in the security equipment
performance index indicator an din the occupational
radiation exposure indicators. Also, in safety system
functional failures and scrams of loss of normal heat.
MR. SHACK: I also suspect that the finer
you make that delta CDF the more the plant specificity
makes a real importance, that is, if you did that at
1 times 10-6, you really would almost have to do it on
a plant specific basis. By the time you get to 10-4,
you're probably not terribly sensitive to --
MR. KRESS: I think you're exactly right.
MR. SHACK: Minor variations. So there's
a certain rationale to doing it that way.
MR. COE: That's a good point. I would
also point out that some licensees, because these
thresholds or these thresholds for unavailability in
this case are generic, may find that their own
maintenance rule, performance criteria for the same
piece of equipment allows much greater unavailability
for certain components that are being monitored by
these PIs and this is a source of concern to them,
that they're being held to this generic standard
whereas their own plant design, their unique features
of their plant design would allow a more
unavailability to accrue for that particular component
before they got to that risk threshold.
MR. JOHNSON: Yes, if you remember where
we were, as Don points out in 1998, we really were
trying to make progress, given the tools that we could
seize upon quickly, given the PIs that we could seize
upon quickly. We did create some new PIs and in fact,
we did end up trying to set thresholds for those and
then trying to benchmark those thresholds and make
adjustments to those thresholds in the pilot program.
And we recognize, as we go forward, that we'll need to
continue to work on and to refine the performance
indicators and the performance indicator thresholds.
We have a process that -- and we talked about this a
little bit at the last briefing, that as a formal
change process for changing PIs or changing thresholds
and it's a deliberate process that has us look and
pilot and benchmark before we make decisions about
changes. But again, I think we agree with the ACRS
that our thrust for the major improvement with respect
to PIs is in trying to, to the extent that we're able
to, do something with respect to being more plant
CHAIRMAN SIEBER: There's some slight
difference in the wording of the second and third
bullet. Is that just editorial or is there some
meaning you're trying to convey there that I'm
MR. HICKMAN: Garrett wrote that. I
really don't know.
MR. SHACK: The rule about parallel
CHAIRMAN SIEBER: Come to the right place,
MR. HICKMAN: If there are no more
questions on that, there was apparently a desire to
see how the process works, how we collect the PIs and
CHAIRMAN SIEBER: Okay.
MR. HICKMAN: I don't appear to have a
transparency for that. You will have it in your
CHAIRMAN SIEBER: 43. PI Reporting.
MR. HICKMAN: Yes, PI Reporting. I used
here an example, again, from the safety system
unavailability indicator. The PI is defined in the
guidance document, NEI 99-02. And I've shown that
It's the sum of the unavailable hours to
plan the unplanned and the fault exposure hours.
MR. UHRIG: What do you mean by fault
MR. HICKMAN: Fault exposure hours are the
hours that a train was in a failed state, but was
MR. UHRIG: Before you caught it?
MR. HICKMAN: He didn't know it was failed
until some time later.
MR. UHRIG: How do you know when that is?
MR. HICKMAN: Well, if -- let's say you
ran a surveillance test and it failed, but you could
trace that back to some maintenance that was done some
time prior to that test and if you could show that
that's what caused a failure then you would count that
amount of time.
MR. UHRIG: Okay, what about where there
are two surveillances, one, it passed, one, it failed?
MR. HICKMAN: If you had no way of knowing
when the failure occurred --
MR. UHRIG: Then you've gone all the way
back to the other one?
MR. HICKMAN: What you do is you use half
MR. UHRIG: Half the integral.
MR. HICKMAN: The standard statistical
technique, assume it's a uniform probability of
failure, divide by 2. It's good for large sample
sizes which we don't really have, but that's the way
it's typically done.
MR. UHRIG: All right.
MR. HICKMAN: And that's an issue that's
been a problem in this program for quite a while.
There's a lot of serious discussion about the use of
T/2. We have had a number, about three failures of 18
month surveillance tests, which meant licensees had to
count 9 months of unavailable hours which is -- and
then that sticks with you for three years, basically.
CHAIRMAN SIEBER: But that's been the fact
for a long time, you know. I remember that from 20 or
30 years ago.
MR. HICKMAN: That's pretty standard,
pretty standard technique.
So we do that. That is how we calculate
a train unavailability, per train.
MR. KRESS: The hours train is required,
is that to differentiate shut down conditions when you
don't need it?
MR. HICKMAN: Well, ideally it should.
What we're doing right now and what WANO does is to
simply lump them together. Ideally we would have
separate indicates for power operation and shut down
conditions, but right now we just lump them together.
MR. KRESS: This is just the number of
hours over which you determine the unavailability
MR. HICKMAN: Yes.
MR. KRESS: So it's the code of thermal
MR. HICKMAN: And what's used there is the
hours that the train is required per tech specs which
means if you're shut down and tech specs only require
one EDG, you can take the others out and do whatever
you want to with them and not have to count the hours.
Now the other thing that INPO did, INPO
actually developed the indicators in the late 1990s
and WANO started using them in 1995. INPO developed
them in the early 1990s. They did some tests
collecting actual data and then looking at easier ways
to calculate unavailability that would be less of a
burden on licensees with regard to the data they have
to submit. And they found that by taking these -- the
train unavailabilities of a system and averaging them
together, they came up with a system unavailability
that tracked pretty well with the real thing. The
numbers weren't the same, but they tended to go in the
same directions. So this is what they used and it's
what we are now using. It's not right. Ideally, you
would want to know when both trains were out at the
same time. You'd have to have the timing information,
but rather than collect all of that, they said this is
close enough and it suits our purposes and that's what
they were using and so that's what we're using. We
recognize the weaknesses.
MR. KRESS: But that other information is
probably available, just harder -- more work to get
MR. HICKMAN: Yes. And as you know, the
Office of Research is developing the risk-based
indicators and they're trying to get information like
that into the EPIX system so they can calculate
What the licensee actually submits to us
then is four numbers for each train, the planned
unavailability, the unplanned unavailability, the
fault exposure hours and the hours the train is
They send that to us in an e-mail with an
attached file that is actually a delimited text file.
That comes into our system here and it's automatically
dumped into a spreadsheet and each of those numbers is
put in the right bin. It's all automated. That
spreadsheet then calculates the values. That's been
thoroughly checked. All through the pilot program we
checked that to make sure it works properly.
So really the processing is hands off. We
do nothing with it until it's all in this spreadsheet.
We then take that spreadsheet. We send the data back,
first of all. We send the delimited text files back
to the licensee to say this is what we got. Is this
what you sent? That's the confirmation process that
takes about a week.
Then once they've confirmed that the data
we've received is accurate, then we review it. We
give the regions a chance to look at it and within a
week then we put it out on the external web.
Actually, at the end of the first week you
put it on the internal web for the regions to see and
a week later then we put it out on the external web.
And that's really all there is to the data
Are there any questions?
CHAIRMAN SIEBER: So basically, the way
you're using performance indicators differs from the
way plants use it. Plants use it as a predictive
measure and they collect sometimes as many as 250
different performance indicators saying that if you
have backlogs building up and so forth, that that's an
indication that your maintenance program, your
corrective action program or what have you is
declining and so they use that to redirect resources.
What you're doing is calculating and reporting changes
in risk, in effect, which is more or less real time.
If unavailability goes up, then the risk changes for
a given plant. And if reactor trips go up, the risk
from ATLAS is changed and so on down the line. So
there is a different concept between the way utilities
use performance indicators and the way you folks are.
And I think you have to do it your way so that it
matches the regulatory system. You don't want to be
in the business of managing the plant the way a plant
manager would do it.
So I think that's appropriate, what you're
MR. HICKMAN: That's exactly true. There
are a number of good indicators that will work if
people don't know you're tracking them and that's good
for plant managers to be looking at those kinds of
things like backlog. For us to take them and put them
on the web would not be good.
CHAIRMAN SIEBER: Well, you don't have
regulations that speak to backlogs.
MR. HICKMAN: Right. In fact, the backlog
will go away instantly if we start --
CHAIRMAN SIEBER: All you have to do is
sit down and do some homework.
MR. HICKMAN: That's right.
MR. JOHNSON: This whole shift in the
process with respect to our use of performance
indicators was really dramatic from what we had done
prior to the oversight process and to be quite honest,
we were a little bit surprised at the industry's
willingness to go forward with some of the performance
indicators. By that, what I mean is we've got
thresholds on scrams for 7,000 critical hours and
there's no regulatory requirement that says that a
plant shouldn't have four scrams per 7,000 critical
CHAIRMAN SIEBER: Or 10.
MR. JOHNSON: Or 10.
CHAIRMAN SIEBER: Except it does change
MR. JOHNSON: So what we did, what we set
out to do and what we were able to accomplish is that
we chose a set of indicators that we believe is
indicative, now they're not perfect, they're not as
risk-informed in some cases we would like them to be,
but they give us insights along with inspections into
issues that begin to emerge at a plant at a level
where we as a regulator ought to engage as opposed to
where licensee management ought to be doing its
CHAIRMAN SIEBER: Okay.
MR. LEITCH: Has this definition been the
one you've used here in this whole one year, initial
one year period?
MR. HICKMAN: Yes.
MR. LEITCH: Have any of the other
definitions changed during the one year period like
scrams and if so, how did they change?
CHAIRMAN SIEBER: Yes, they did.
MR. HICKMAN: You may be aware that we
just finished a pilot program for replacement scram
indicator. Are you aware of that?
There were a few people in the industry
who were concerned about unintended consequences,
unintended influences on operators from counting
manual scrams, so the industry -- industry
representatives working within NEI developed an
alternate indicator to replace that one and we just
finished a pilot program. The intention is to count
exactly the same thing and that was automatic and
manual scrams, but without ever using the word scram
in the definition, so it's kind of a funny thing. But
we are looking at that now. We have criteria to
evaluate that against and we'll use that to serve as
MR. LEITCH: That's one thing that
confused me. In the pilot program you counted both
manual and automatic scrams, just like the initial one
year program, it's just the matter that they didn't
call them scrams?
MR. HICKMAN: Right, that was the intent.
Whether we did that or not is still yet to be
determined. We're looking at the data now. We just
got it in, final, couple weeks ago.
CHAIRMAN SIEBER: Well, that particular
argument goes back about 10 years because the industry
made the same arguments to INPO that says you're going
to inhibit the operator from manually tripping the
reactor and the INPO philosophy is to trip it manually
before some automatic system takes you out which
lessons the transient on the plant in a lot of cases
and so I guess I wonder whether counting manual scrams
is really the right thing to do, even though from the
standpoint to causing an initiating event by the twist
of a wrist does change the risk of the plant because
it causes a lot of other things to happen.
Is there something on either side of that
question as to whether you count it or you don't count
MR. HICKMAN: As you know, the AEOD PIs
agreed to use the same definition as INPO.
CHAIRMAN SIEBER: Right.
MR. HICKMAN: When they started those in
CHAIRMAN SIEBER: Right.
MR. HICKMAN: And we only counted
automatic scrams while critical for that reason. But
there were people here who were concerned that
operators might try to beat the PI by manually
scramming it, so we monitored that we never really saw
any signs of that. Manual scrams have remained
relatively constant around 40 per year, up until the
last couple of years. Some as high as maybe 55, some
down to about 29 or so, but roughly averaging around
40. They're down a little bit now, down into the low
30 range, but of course, the automatic scrams have
come way down from several hundred, down to about 50.
CHAIRMAN SIEBER: Right.
MR. HICKMAN: But from the very beginning
and working with NEI on this, we never really doubted
whether we needed to count manual scrams because the
conditions in the plant that require a scram are the
same and whether the operator manually scrams it or it
takes an automatic scram, whatever has gone wrong with
the plant that required that scram is what we want to
CHAIRMAN SIEBER: Yes, but the technical
challenge to the plant is typically less because you
haven't reached the set point or the limiting safety
MR. HICKMAN: That's true. As the
operator scrams it, he may prevent other automatic
actions by not reaching --
CHAIRMAN SIEBER: And less than the
excursions that the plant goes through during a
MR. LEITCH: But this new definition,
revised definition, not using the word scram is
separate pilot program. That is, the initial one year
period, nothing has changed during that period?
MR. HICKMAN: No. We're still using the
same indicator that we started with in the pilot
program and it says the indicator counts all automatic
and manual scrams while critical.
MR. LEITCH: Was there a change or is
there a change being contemplated with regard to
unplanned power changes?
MR. HICKMAN: Yes. We're getting ready to
try a pilot program on a replacement for that.
Actually, there are two proposals, one from the NRC
and one from NEI that we'll pilot. The concern there
is that we had a 72-hour rule, basically it said if
the time between the identification of a problem and
beginning to insert negative reactivity is greater
than 72 hours, then it doesn't count. This was
something that was of concern to NEI and the industry
that we shouldn't count power reductions that are
planned. It was never the staff's intention. We
never worried about whether it was planned or not. We
used a definition that's in the monthly operating
report and there, the distinction was not planned
versus unplanned. It was forced versus schedule which
is not exactly the same thing.
CHAIRMAN SIEBER: That's right.
MR. HICKMAN: And what we captured in the
monthly operating report was whether they had to come
down at the first available opportunity to fix it, or
whether they could ride through that and continue on.
At that time, when the monthly operating
report was put into place, the first available
opportunity was considered to be the next week. So
that was the criterion. But what's happened is with
the 72-hour rule, that provides an incentive for
licensees to --
CHAIRMAN SIEBER: Struggle along.
MR. HICKMAN: And ride it out. And in
fact, we had a licensee who was very forward with us
and he told us, I can't afford another power change.
I'm going to ride it out and he did that a couple of
In defense of the licensee, he didn't do
it when he thought it was a safety problem, so even
though it was going to cause him a problem, he did
shut down and he did count it, but when he thought he
could get away with it, he didn't do it.
CHAIRMAN SIEBER: Well, that's one of the
problems with performance indicators across the board.
People know what the thresholds are and what the goals
are and they will manage the plant to meet those
expectations. And that's not always in the plant's
MR. JOHNSON: That's right.
CHAIRMAN SIEBER: And so that should be an
important factor when you folks are divining what kind
of performance measures you're going to use, because
you might as well face it, people do manage to those
MR. JOHNSON: Absolutely.
MR. HICKMAN: That's true and this is a
particular problem in the initiating events
cornerstone and the mitigating systems cornerstone.
CHAIRMAN SIEBER: That's right.
MR. HICKMAN: We've had a number of
successes in the program in the emergency preparedness
cornerstone and in the physical protection
cornerstone. If we could make all of the indicators
like those in the EP cornerstone, they provide the
incentive for the licensee to do the right thing, that
is, we've got a drill exercise performance indicator
and a drill participation indicator. And if he's
having problems with either one of those, the answer
is to run more drills and get more people in the
drills. And we've had great success. We've had
people who were not paying attention to whether there
were people who were actually getting trained or not
on a regular basis and when we started the PI, they
realized that and they responded and they brought
their PIs down to within the green band. And that's
good, if everybody stayed within their green band,
that would be good.
The same thing happened in the security
equipment performance index. We had a couple of
licensees who had very bad problems with their
security equipment and had just never gotten
management attention and as soon as the PI came out
and the manager looked at that, he said what's this
all about and he immediately fixed the problem.
CHAIRMAN SIEBER: Yes, but there was a
practice among licensees in security to say that if I
put a watchman in place or a response officer in place
of the defective piece of security equipment, that
compensating measure was equivalent to having that
piece of equipment in the service, so they would sit
down and calculate it's going to cost me $25,000 to
fix a TV camera, how long can I keep a watchman there
to watch that zone? And will I, in effect, make out
economically by doing that? Okay, so what you've done
there is change the economic balance of supply/demand
situation for the management.
MR. HICKMAN: Sooner or later though
they'd have to fix it, but I mean at some point the
cost of the guard is going to exceed the cost to fix
CHAIRMAN SIEBER: That's true. It all
depends on whether you have capital money or operating
money to spend.
MR. HICKMAN: That's true.
CHAIRMAN SIEBER: Some day I'll have a
meeting to explain the power plant economics, but some
plants don't have capital money. You know, they just
don't have a rate base, so they don't want to spend
MR. SHACK: Has anybody objected to any of
these PIs as a backfit?
MR. JOHNSON: Not to my knowledge, no.
MR. COE: There has been some discussion
at high levels regarding the earlier question, the
earlier point that was made is that these aren't based
on regulatory requirements and therefore there's a
question out there about de facto regulation. But I
think that those haven't been, there hasn't been a
unified chorus of individuals out there that are
complaining about that. I'm speculating, but I think
it's primarily because they see greater net benefit,
you know, the disadvantages as they perceive them are
offset by the benefits of the program. So they're
willing to work with us and continue to evolve the
program to what they hope would be better in the
MR. JOHNSON: I actually think we could be
more positive. There was concern early on about
whether we needed a regulatory requirement to collect
these, regulation to collect these performance
indicators and NEI said God forbid, don't do that.
And we said okay, we'll have this voluntary PI program
and if you guys don't give us PIs, we'll go do
baseline inspection to get the insights. Well, we've
not had licensees not give us performance indicators
because they don't buy the program.
Now having said that, we work very closely
with the industry and other stakeholders and the
public meeting to refine the reporting criteria, to
make sure they're reasonable and understood. So it's
been a lot of work for us to be able to implement this
voluntary aspect of the ROP. But there's not a course
MR. SHACK: I hear that plants collect 200
PIs and whenever the risk-based PIs are mentioned, oh
my God, the burden is incredible, can't believe it and
it just somehow seems like a mismatch here. Again,
maybe there's a difference between collecting the data
for your own purposes and swearing to the NRC that
this data is accurate and I'm ready to go to jail if
MR. JOHNSON: Yes and those are some of
the issues. In fact, the last time I sat in on the
risk-based performance indicator talk that you all
were given by Research and that is what licensees tell
us. I think what we heard from licensees of late is
we've got this new oversight process. We've got PIs
associated with that process. Why don't we live with
that for awhile and why don't we consider very
carefully adding additional performance indicators
that could result in additional burden. So there is
definitely that theme that we're getting.
And again, when we go to collect
performance indicators, I sort of am remembering now
how that last risk-based performance indicator
briefing went and some of the issues that came up that
were discussed and I think we have an IOU, as a matter
of fact, to the ACRS that came out of that briefing,
but again, remember, the performance indicators
provide a valuable piece of information. Now the
performance indicator program is a voluntary program.
It turns out there are OMB clearance requirements,
requirements with respect to collecting data from more
than nine licensees. So if we go to do that, we've
got to make the case about burden and about benefit.
And so we're -- we think we are appropriately cautious
with respect to adding new PIs to make sure that they
give us the benefit that we need, but at a cost that
That was some of the sense that we
discussed last time. You're right. You do hear the
industry say hey, don't give us a whole bunch more of
performance indicators when what we have is okay for
CHAIRMAN SIEBER: I think the other
problem that comes up sometimes is the fact that if
NRC comes out and says I want this performance
indicator and I'd like you to send it to me, but my
definition is different than WANO's, then the licensee
sees that as a whole new indicator because they have
to engage somebody to produce it every month for you.
I think on the other hand, the industry appears to
prefer risk informed and performance based regulation
to deterministic regulation and if it adopts that kind
of preference, they have to cooperate and I think
that's what you're seeing.
MR. HICKMAN: And you hit on one of their
big concerns and that is if they have to calculate
unavailability one way for WANO and another way for
the maintenance rule and another way for us --
CHAIRMAN SIEBER: That's right.
MR. HICKMAN: That's a burden.
CHAIRMAN SIEBER: It's confusing too,
because it's usually the same person who's doing all
the calculations and to keep all that stuff straight
for a whole bunch of different indicators is
MR. HICKMAN: Especially if you're going
to be held to 50.9 requirements for sending it to us.
CHAIRMAN SIEBER: That's right.
MR. HICKMAN: The other aspect of that --
I just lost it. Oh, the other aspect to the more
indicators is in their view it's just more ways to go
white and why do we need more ways to go white if
we've got 18 already that work.
MR. JOHNSON: Okay, that captures the
discussion we plan to have on performance indicators
although I do note that Garrett is in the room.
CHAIRMAN SIEBER: Too late. Unless one of
the Members has a question that they would like to
direct to Garrett.
MR. JOHNSON: Okay, the last section that
we want to cover and we've just got a couple of slides
is there were some selected issues. Two of the SECY
issues I think we've already talked about, and that is
we talked about thresholds and the threshold for green
to some extent. Hopefully, you're satisfied and we
don't need to talk about fire protection any more,
because the fire protection people are no longer in
the room and I can't even spell fire protection.
CHAIRMAN SIEBER: Well, I'm the chairman,
but the one who asked the question isn't here. So
I'll take it upon myself the duty to go over it with
MR. JOHNSON: Okay, the last topic that we
wanted to talk about was the topic of cross-cutting
issues because we know there has been some interest
with respect to this topic and for that Jeff Jacobsen
is going to talk very briefly about cross-cutting
MR. JACOBSEN: Okay. I guess where we
left this, just a little brief history as
cross-cutting issues is something that has come up
throughout our engagement with the public and internal
stakeholders with regard to how cross-cutting issues
are treated in the new oversight process. And cross-
cutting issues we defined originally as three issues:
human performance, safety conscious work environment
and problem identification and resolution. So when we
talk about cross-cutting issues, those are the three
things we're talking about.
The fundamental assumption when we
designed the framework for the revised oversight
process was that these cross-cutting issues would show
up either in the performance indicators or in the
baseline inspections, in a sufficient time frame to
allow us to engage before a real safety issue arose.
We consciously did not design a program to
specifically go after human performance, for instance,
because we thought that if human performance was weak,
it would show up in one of the performance indicators,
reactor trips or unavailability if it was maintenance
related to human performance, etcetera.
With regard to safety conscious work
environment, a similar analogy was thought that
weaknesses in that area where people are afraid to
bring problems up or there's retribution, our
experience has been that those facilities performance
has suffered as a result of that and we would see it.
CHAIRMAN SIEBER: You would also see that
as allegations, would you not?
MR. JACOBSEN: Right, which we also
monitor kind of outside of the performance indicators
and baseline inspection, but it is part of our overall
We do, however, have a significant portion
of our inspection program that's directed at problem
identification and resolution because we believe that
is a very important part of the process, so we look at
that. We were looking at it annually. We recently
made a decision to change that to a once every two
year inspection. So we do look at that.
CHAIRMAN SIEBER: How do you determine
whether the licensee for any given plant has set a low
enough threshold for formally identifying problems?
MR. JACOBSEN: Our experience has been
that each licensee's program is somewhat unique.
CHAIRMAN SIEBER: That's right.
MR. JACOBSEN: We don't have a go-no go,
per se, for what's a low enough threshold. What we
would use would be if we, for instance, in our other
inspections identify problems that we think are
significant, that the licensee didn't get into their
corrective action program for whatever reason, we
would infer that they do not -- they either don't have
a low enough threshold or they aren't looking in the
If we're finding stuff or other external
organizations are finding issues, and the licensee
isn't finding them, then that's either a threshold
question or a question that they just aren't looking
in the right areas.
CHAIRMAN SIEBER: Well, how do you weave
that into the regulatory system? I mean you could
determine that through observation and inspection, but
how do you bring that --
MR. JACOBSEN: How do we act on it?
CHAIRMAN SIEBER: Well, how do you relate
that to the requirements of the regulations?
MR. JACOBSEN: Well, Appendix B has -- is
really the appropriate regulation.
CHAIRMAN SIEBER: You can cite anybody for
anything through Appendix B.
MR. JACOBSEN: Right, well, most things.
There are some areas that Appendix B isn't applicable
and that has actually come up in this process,
emergency preparedness, for instance.
CHAIRMAN SIEBER: Right.
MR. JACOBSEN: Appendix B is not
The way we deal with it is if we were to
have an inspection finding that turned out to be a
significant finding and if we found out the root cause
of that finding was related to threshold issue or
improper evaluation of a previous issue, we would deal
with that in that manner.
CHAIRMAN SIEBER: Okay.
MR. JACOBSEN: It would be on a for-cause
basis for the most part.
MR. LEITCH: What's the basis for moving
that inspection module from annually to semi-annually?
MR. JACOBSEN: That was a very general
statement of what we're doing. In addition to
changing the frequency, we've done some other things.
We've beefed it up a little bit so although we're
going to do it less frequently, we're going to add
some resources to it because we think that the look
every two years in a deeper way is more effective than
doing it annually in not as deep a way.
The basis for it in our experience,
licensees' programs such as this will not change
significantly on a one year basis. We've seen
declines in corrective action programs, trends, but we
believe that a frequency of every two years will be
sufficient to pick that up and if we went and did an
inspection at a facility and found they had a good,
corrective action program one year, it would be highly
unlikely, in our opinion, that it would decline
significantly in one year. It's more of a cultural --
it's almost analogous to plant culture. And that's
something that you know takes a long time to turn
around. It also pretty much takes some time to go
down. So that's -- we're also adding some additional
requirements where we're going to instead of doing a
team inspection, we're going to look at some limited
samples throughout the two years on a per inspector
basis. So every so often, one of the inspectors is
going to pick something in the corrective action
program and do an in-depth inspection of that one
item. And then every two years the thought would be
that you would integrate all those insights that you
got throughout the year, as well as the insights you
got while you're doing the team inspection into a more
broad assessment of the corrective program.
MR. LEITCH: Okay.
MR. JACOBSEN: And the last thing we're
going to is an we'll talk about this a little more
when we get to the action matrix discussion next time,
in July, the other element that we're adding is is
we're beefing up the role of this PI&R inspection or
I guess I should say if a plant would end up in the
action matrix in the degraded cornerstone column, we
would, in fact, consider -- the regions would consider
doing a problem identification and resolution
inspection. We think that provides a better
opportunity for licensees, for the NRC to look at the
performance of the licensee and the performance of the
PI&R program in a specific event where they've crossed
some thresholds. So we think, in balance, even though
we say we're going from a single year to a biennial
frequency, we've done some other things of PI&R that
we really believe make it more, a much more effective
MR. FORD: Just for information, what does
move out of the licensee response band, they don't
correspond with it? What does it mean?
MR. JACOBSEN: The second item?
MR. FORD: Yes.
MR. JACOBSEN: Okay, I'll go into that.
Our experience with the first year of implementation
of the revised oversight process has pretty much
supported the first assumption and what we mean by
that is plants that we've looked at and we have
concerns about in the cross-cutting areas and
primarily they've been in the problem identification
and resolution area. For instance, if we did our
annual team inspection and we had a lot of green
findings, we haven't had any white findings or greater
as a result of the corrective action inspections.
We've had very few white inspection findings overall.
But in the PI&R area we haven't had any. But we've
had a lot of green ones and if you look at the plants
where there's been a lot of green findings and where
the inspection team came away with concerns about the
adequacy of the program, in all cases those plants
have moved out of the first column, that licensee
response column of the action matrix, either to a
degraded cornerstone column or a regulatory response
column which has allowed us to engage further and to
look in a more programmatic sense at the corrective
A good example of that is Kewaunee where
we had concerns with their performance during our
problem identification and resolution inspection.
They had a yellow performance indicator and when we
went out and did that, we identified broader concerns
with the corrective action program as well. As a
result, they totally revamped their corrective action
So these four facilities are examples of
facilities where we had concerns after doing the
baseline inspection and they also -- we had
opportunities to look further as a result of our
The contrary to that is we have not
identified any plants where we have significant
concerns in the cross-cutting areas that have not
moved out of the licensee response. So it's been a
very close tie between the performance and actually
crossing the thresholds that allow us to engage
The third bullet, no significant
precursors caused by cross-cutting issues, well, in
fact, the definition of significant precursors, I
believe, is an event that's defined as having a 1 in
1,000th greater chance of leading to a reactor
accident. There haven't been any of those period.
Really, if you were to look at the
fundamental assumption and the basis of the ROP is we
would be concerned if we had, for instance, one of
these significant precursors and found out they were
caused by a cross-cutting issue and we didn't have an
opportunity to go after it and prevent it. That
The way we're going to deal with that is
kind of on the next page. We're going to look at
things at a threshold actually lower than significant
precursors. We're going to look at ASP events and
inspection findings that come out yellow and red and
we're going to look and see in those instances whether
cross-cutting issues were one of the root causes that
caused the event or the inspection finding to occur.
And if so, would our program have at least given us
the opportunity to identify those type of issues.
So I guess the bottom line is we believe
our fundamental premise of the ROP with regard to
cross-cutting issues still appears to be true.
However, we still have some on-going actions to
continually challenge that and ensure that, in fact,
we are focusing our resources in the right direction,
as we do with all areas. It's not limited to
That's pretty much all I wanted to go
MR. JOHNSON: Very good.
CHAIRMAN SIEBER: Thank you very much,
I'd like to take a few minutes to ask if
any Members have any comments that they'd like to make
based on what we've heard today?
MR. UHRIG: I just have a question. This
was handed out. I don't know if you handed it out or
this came from somebody else.
MS. WESTON: I passed it out and my only
question is what are the titles of the codes.
MR. UHRIG: Among other things.
CHAIRMAN SIEBER: Okay, there are the
MR. UHRIG: The other question had to do
with there's a number after, for instance, white 3.
Does that mean three findings?
MS. WESTON: That's the inspection summary
findings for the first quarter.
MR. UHRIG: That would be third quarter,
3 would mean third quarter?
MS. WESTON: That's what he's looking at.
This is the first quarter. This is all the first
quarter. These are cornerstones.
MR. UHRIG: What does the 3 mean?
MS. WESTON: I don't know.
CHAIRMAN SIEBER: Since we're still on the
record, maybe we could have people speak into the
MR. JOHNSON: What you're looking at is
one of the web page printouts and we've got a number
of these employees to summarize the results for all of
the plants, in addition to be able to pull up any
individual plants, these are the performance
indicators and the inspection results. So Don is
going to try to answer the question.
MR. HICKMAN: What you see here is for
each plant and each cornerstone, you see the
inspection finding results. What they show there, the
ones with the numbers, the color is the color of the
highest, the most significant one and the number is
the total number. It doesn't necessarily mean there
are three whites in that block, but it means there are
three and the highest one is a white.
MR. UHRIG: Okay. So I take it where
there's no number, there's only one finding?
MR. HICKMAN: Yes.
MR. UHRIG: For example, most of the
greens are that way?
MR. HICKMAN: Yes.
MR. UHRIG: Okay.
CHAIRMAN SIEBER: The other question was there
are a large number of no findings. That means simply
that this is the first quarter and during that first
quarter there was no evaluation?
MR. UHRIG: No.
MR. HICKMAN: They had -- they conducted
inspections and had no findings.
CHAIRMAN SIEBER: None at all.
MR. JACOBSEN: They may or may not have
done an inspection in that area. In either case,
there were no findings.
MR. BONACA: And green means simply --
CHAIRMAN SIEBER: That there was a
MR. BONACA: Yes, but for example, the
initiators, the first category, a green would mean
simply that it was --
CHAIRMAN SIEBER: Well, it means there was
a finding which means there's a deficiency, but it's
within the licensee's prerogative and control to fix
MR. JOHNSON: Exactly.
CHAIRMAN SIEBER: Without additional
MR. UHRIG: Notice in some cases, sister
plants for instance here, Peachbottom 2 and 3 had
whites both -- is that a common failure? Is it a site
failure? Is it just the individual plants happen to
come out that way?
MR. HICKMAN: It depends.
MR. UHRIG: Same with quad cities.
MR. JOHNSON: Every unit has -- the ROP is
specific for the unit with respect to the performance
indicators and the inspection findings. And so it's
MR. UHRIG: Is that cornerstone emergency
MS. WESTON: Occupational radiation
safety. It's the sixth column.
MR. UHRIG: Internal rad.
MR. JACOBSEN: And in some cases the
finding can affect both units. In other cases, it may
be two separate independent white findings of a unit.
MR. HICKMAN: For the cornerstones and
site white programs like EP and occupational radiation
safety and security, they both get account.
MR. BONACA: The question I have is you
pointed out that there is a correlation between plants
that has a problem and the effectiveness of the
corrective action program and I always believed that.
But a question I have do you have any specific set of
indicators on the corrective action program being used
or is it so, is it again considered subjective by a
licensee, a judgment he may express on that?
I'm going to the fact that more and more
we are speaking about objective evidence and when I
look at this data, I mean I can interpret it and it
tells me something. But I still believe that the
corrective action program tells me much more than
anyone of these boxes. That's my personal belief, if
I could get into it. And so the question I have is
when you do the inspection, since there is no
quantitative assessment that is translated into a
caller, do you use some specific indicators and are
they agreed to by the licensees?
MR. JACOBSEN: I'll answer it a couple of
ways. First of all, we have some indicator and that
is if we have findings we do run those findings
through the SDP so we have either so many green issues
or so many white issues. That's a very crude
MR. BONACA: Okay.
MR. JACOBSEN: The second, I guess, answer
to that is every licensee has their own set of
indicators that they're using to measure their
programs. The problem is that eery one of these
programs is different and every licensee has a
different set of indicators with different thresholds.
The third answer is we understand it would
be a big improvement if we could develop some more
objective ways of assessing these corrective action
programs. Because our assessment right now is largely
qualitative and not quantitative. So we do have a
task group that we're working towards and it may not
be performance indicators as we think of them today,
but we are looking at developing a more objective way
of assessing the corrective action programs. And if
we were to come up with indicators, we would have to
get industry to buy in. It gets back to the question
that we raised, how much burden do we want to add for
what gain? We might have to develop site-specific
thresholds, for instance, and then you have to
validate the indicators.
MR. BONACA: But typically, you do have
some -- like threshold level, is it low or high? And
you have some way of -- agreed to by the industry. I
mean I've seen, I can go from one site to the next and
I've been there looking at corrective action programs
and I can see they all speak the same language, pretty
much, because there is a lot of shared information
today. The other one is categorization. Okay, what
do you lump into category 1, 2, 3? Do you have the
right percents distributed there? What is the time of
response? I'm just pointing out that maybe, by now,
there is more consistency among the programs than not.
MR. JACOBSEN: Well, they're becoming more
consistent and we're -- and the industry has done some
work in this area and INPO has some inspections that
they do. Nobody has been able to come up with any
joint performance indicators.
MR. BONACA: True.
MR. JACOBSEN: But we're looking and we're
going to continue to look at that and the types of
things you mentioned are good. WE do have our
procedure broken down into areas that we look at. We
look at threshold and we look at prioritization and we
specifically have attributes that we look at in each
of those areas, but to take those and quantify them is
a whole -- I know two plants, one that has 10,000
items they put in their corrective action program a
year, and one that has 1,000 and they both may work
real well. It's just how those two programs are
managed. It's very hard to say to somebody you need
to have so many thousand items in your corrective
action program or your threshold is not low enough.
You don't want to do that. You have to be real
MR. JOHNSON: John, did you have anything
you wanted to add? I know you like to talk on this
MR. COE: Only that your comment is a very
good one and it's been one that I know that I've been
thinking about a lot for several years, because the
process of these inspections is as Jeff indicated,
very qualitative. At one point, as -- in my previous
existence as an analyst, I actually went out and tried
to do some more quantitative look at corrective action
programs by taking the current open issues and gauging
them according to their functional impact and then
also gauging them in accordance with their risk
importance, and then essentially combining those two
elements for that each item to come up with kind of a
composite list of those issues which were both
functionally, had functional impact associated with it
and had risk significance. And that might be one way
of assessing whether or not the licensee is applying
the correct priorities, okay, and investing the right
level of resources, if they're grading their resources
in a manner which makes sense from a risk standpoint.
In addition, there's a question out there
that could be asked about what about the accumulation
of lower level issues that in the risk kind of sense
combined together, synergistically, to provide a
greater risk impact than each one looked at
individually. So these are the kinds of issues and
the kinds of questions you're raising are very good
ones and they're ones we've been thinking about.
MR. BONACA: Well, the reason why I raise this
is also some, for example, some licensees are more
aggressive because they have been having problems and
they tend to do more cause analysis. Others, who
believe they are very good or they believe that, they
tend to say we do too much and so they go now to
apparent cause in many more cases because there is
some kind of complacent setting. You'd be surprised
how the first type of individual finds more things.
Therefore, you tend to say he has more problems. And
the other one doesn't find that much because he does
all the apparent causes and very few causal
evaluations and on the surface he has less problems.
And so you tend to think the other guy is better off.
I've seen these cases and compared them and you are
surprised on how you can get truly the wrong
conclusion. And so that is the point I was making.
Maybe there are some indicators that can be determined
to help in that process because I think it's such an
MR. JOHNSON: Very good. I think it's a
good point. The last thing I would point out with
respect to that is that we raised this issue with the
industry, continue to raise this issue with the
industry and the last time we raised it with the
industry, you might appreciate that the industry
doesn't feel like we need to do more with respect to
performance indicators particularly in this area.
MR. BONACA: Somehow I'm not surprised.
But they don't mind that you are looking
into it, right?
They can't do anything about it. That's
the fundamental area of inspection.
MR. JOHNSON: That's right.
CHAIRMAN SIEBER: I guess I'd like to ask
if any other Members have questions or comments that
they would like to make at this time?
MR. SHACK: I guess just the one I'd make
is it seems to me like such a fundamental area, that
is one you wouldn't want to back off on the
inspections and that's always the price that industry
is looking for. Yeah, we'll give you a PI if you back
up. But this certainly seems like about the last
inspection you want to back off on.
MR. JOHNSON: Yes, absolutely, and that's
why I was careful to say we don't believe that we're
backing off on PI&R. We think that what we're putting
in place is a more effective PI&R and that's really
the focus of our changes in that particular area,
although I think there is a net decrease of 25 hours
a year or something.
MR. JACOBSEN: Yes, it about 5 or 6
percent. That's on paper anyway. What actually gets
implemented is --
CHAIRMAN SIEBER: Actually, it seems to me
that the number of modules in their rigor has
increased under this new program from what it was
before which has pluses and minuses and the pluses, of
course is more directed inspection and the minuses,
that there's less abilities for the region and
individual site inspectors to use their discretion to
respond to special situations in the plant. And I
guess that as you gain more experience in the
inspection process, you'll be able to judge whether
the balance that you now have is appropriate compared
to something more akin to the past practice which
seemed to have more flexibility in it than the current
MR. JACOBSEN: Actually, the change we're
making in PI&R responds to that very comment, that one
of the regions felt strongly about having this part of
the program where we could look at things in a more
real time basis than they thought. So rather than
doing it as a team once a year, we're going to pick
these things throughout the year. So that actually
responds to that flexibility question. So we are
looking at that and making changes as appropriate.
CHAIRMAN SIEBER: Right. Okay. Any other
questions or comments?
MR. LEITCH: I'm still just perhaps a
little confused about the expectations for the
predictive nature of this reactor oversight program.
If you had a hypothetical plant that was
running along with basically green performance
indicators and no color inspection findings, and then
it's had a track record of that for several months, a
year and then you come through some self-reviewing
event, you find that the plant has a lot of problems
and winds up in a regulatory shutdown, would you be
disappointed with the reactor oversight program or
would you say well, this is not a predictive program,
we had no way of knowing that?
I'm still groping for what the expectation
MR. JOHNSON: Yes. If we saw a plant --
we as an Agency, we constantly look at these
situations and we do a lot of hand wringing and soul
searching and we try to make decisions about whether
the process, the performance that results is a process
failure. And if I saw a plant that was in the
licensee response band that ended up in the degraded
cornerstone corner, that doesn't mean that we've had
a programmatic failure.
Now in our self-assessment matrix we look,
we will look, we continue to look at jumps in plant
performance across multiple columns of the action
matrix to see if there was something that should have
been in the process that was not in the process. But
the process hasn't failed because again, we haven't
built a process that we guarantee predicts that kind
If you tell me, if you're painting a
picture of a plant that was in the licensee response
band today, that tomorrow we have to issue an order to
remain shut down, that is their performance is
unacceptable, then yeah, I think we have to really
step up to the plate and talk about whether we need to
do something drastic with respect to the program.
MR. LEITCH: I mean admittedly, I have not
seen such a thing. I'm not saying such a thing
exists. I just don't understand your expectations.
CHAIRMAN SIEBER: Any other questions?
Since there are none I would like to comment to you,
Mike, and to all the speakers today that I think you
have been very responsive to the questions that we
asked. I thought your presentations were well
prepared. And I think that you're on the right track,
but you've only been in this business for a short time
and I'm sure you're still in the learning process and
as time goes on you, for sure, will make some
adjustments in what you're doing today, but it just
seems to me this is a step forward and I want to thank
you for putting in the time and effort to give us good
presentations and well thought out responses.
So with that I think we can conclude,
unless anyone else any comments or statements to make.
We can conclude with today's meeting and again, thank
you very much.
MR. JOHNSON: Thank you very much.
(Whereupon, at 2:40 p.m., the meeting was
Page Last Reviewed/Updated Tuesday, August 16, 2016